Non-finite complementation

Non-finite Complementation

LANGUAGE AND COMPUTERS:STUDIES IN PRACTICAL LINGUISTICS

No 65

edited by Christian Mair

Charles F. Meyer Nelleke Oostdijk

Non-finite ComplementationA usage-based study of infinitive

and -ing clauses in English

Thomas Egan

Amsterdam - New York, NY 2008

Cover design: Pier Post

Online access is included in print subscriptions:see www.rodopi.nl

The paper on which this book is printed meets the requirements of"ISO 9706:1994, Information and documentation - Paper for documents - Requirements for permanence".

ISBN: 978-90-420-2359-8©Editions Rodopi B.V., Amsterdam - New York, NY 2008Printed in The Netherlands

Contents

Acknowledgements ix

Preface xi

1 The purpose and scope of the study

1.1 Introduction 1 1.2 Why a usage-based study? 2 1.3 The form of a usage-based study 7 1.4 Schemas and Constructions 10 1.5 The scope of the study 12

1.5.1 The constructions investigated 12 1.5.2 The Corpus 15

1.6 The content and organisation of the chapters 17

2 Classification of the constructions

2.1 Introduction 19 2.2 Classification according to the semantics of the matrix verbs 19

2.2.1 Matrix verbs in different-subject constructions 20 2.2.2 Matrix verbs in same-subject constructions 26 2.2.3 Different-subject and same-subject constructions compared 30

2.3 Classification in terms of TAM relations between matrix verb and complement situation 32

3 Earlier Studies

3.1 Introduction 45 3.2 Specific versus general 45 3.3 Factive versus non-factive 48 3.4 Non-fulfilment versus fulfilment 51 3.5 Referring versus non-referring 55 3.6 Future versus actual 57 3.7 Future versus interior 61 3.8 Less immediate versus more immediate 70 3.9 Holistic construal versus immediate scope 75 3.10 Conceptual distance versus conceptual overlap 78 3.11 Summary 85

vi

4 Complement Types and Complementisers

4.1 Introduction 89 4.2 The bare infinitive 91 4.3 The to infinitive 93

4.3.1 The to infinitive in complement constructions 94 4.3.2 The to infinitive as subject and adverbial 99 4.3.3 Weaknesses of the targeted alternative approach 103

4.4 The -ing form 107 4.4.1 The -ing form in complement constructions 109 4.4.1.1 Comparison with the progressive 109 4.4.1.2 -ing complements encoding activities 111 4.4.1.3 -ing complements encoding accomplishments 114 4.4.1.4 -ing complements encoding states 116 4.4.1.5 -ing complements encoding achievements 118 4.4.1.6 Prototypical characteristics of the -ing complement 121 4.4.1.7 The schematic meaning of the -ing form 128 4.4.2 The -ing form as subject and adverbial 132 4.4.3 The -ing complement form: a summary 134

4.5 The to -ing form 135 4.6 The four complement forms compared 141

5 Constructions in contrast: Same-time constructions

5.1 Introduction 145 5.2 Bare infinitive and -ing complements with Perception verbs 145 5.3 Same-time bare infinitive and -ing versus Judgement to infinitive 151 5.4 Same-time -ing versus Forward-looking to infinitive 156 5.5 Same-time -ing versus General to infinitive 161

5.5.1 Positive Attitude constructions 162 5.5.2 Negative Attitude constructions 171 5.5.3 Comparative Attitude constructions 177 5.5.4 Summary of Attitude constructions 183 5.5.5 Aspect constructions 184

5.6 Same-time constructions: a summary 196

6 Constructions in contrast: Forward-looking constructions

6.1 Introduction 199 6.2 To infinitives versus bare infinitives 200

6.2.1 Communication constructions 200 6.2.2 Constructions containing help 204 6.2.3 Constructions containing let, allow and permit 214 6.2.3.1 Types of subject and TAM relations 215 6.2.3.2 Force dynamics 217

vii

6.2.3.2.1 Tokens encoding permission 220 6.2.3.2.2 Tokens encoding prohibition 228 6.2.3.2.3 Why ‘allow S2 to infinitive’ but ‘let S2 bare infinitive’? 232 6.2.3.3 Let and allow: a summary 236 6.2.4 Constructions containing dare and need 237 6.2.5 To infinitives versus bare infinitives: a summary 243

6.3 To infinitive versus -ing complements 243 6.3.1 Mental Process constructions 243 6.3.2 Attitude constructions 249 6.3.3 Aspect constructions 254 6.3.4 To infinitive versus -ing clauses: a summary 267

6.4 Causation constructions 267 6.5 Forward-looking constructions: a summary 278

7 Constructions in contrast: Backward-looking constructions

7.1 Introduction 281 7.2 Mental Process constructions 281 7.3 Attitude constructions 286 7.4 Communication constructions 288 7.5 Aspect constructions 291 7.6 Backward-looking constructions: a summary 303

8 Summary and Conclusions 305 References 311 Appendices Appendix 1: The Matrix verbs 318 Appendix 2: Tables of Matrix verbs 405 Appendix 3: List of technical terms 425 Index 429

Acknowledgements

A version of Chapter 2 was published in 2006 by Presses Universitaires de Franche-Comté as “The classification of non-finite constructions in English”, in Constructions Verbales et Production de Sens, edited by Daniel Lebaud, Catherine Paulin and Katja Ploog. A version of section 4.4.1 was published in 2007 by Peter Lang as “The prototypical meaning of the -ing complement form”, in Corpora and ICT in Language Studies edited by Jacek Walinski, Krzysztof Kredens and Stanislaw Gozdz-Roszkowski. I would like to thank both these publishing houses for permission to reprint extracts from these papers.

Preface

This book is a completely rewritten version of a dissertation presented to the University of Oslo in 2002. Although there is not much left of the original text, thanks are first of all due to Johan Elsness, who supervised the work on which the book is based. Not only did he fulfil his supervising duties to the letter, but he made many helpful suggestions, some of which I have only now taken on board as I reworked my way through my data. The committee for the dissertation comprised Stig Johansson, Christian Mair and Anne-Marie Simon-Vandenbergen. Thanks to all three for a lively discussion at the formal defence and for subsequent encouragement in my revision of the text for publication. I am very grateful to my employers, Hedmark University College, for providing me with sufficient leave of absence to bring this project to fruition. Invaluable aid was provided along the way by the IT staff at the college, who helped me out whenever my very limited technical knowledge left me high and dry, and to the library staff, who got hold of all the books and articles I needed. Eric van Broekhuizen at Rodopi also deserves thanks for his comprehensive (and comprehensible!) advice on technical matters. I have been working, on and off, on this project for about ten years. I am grateful to all the colleagues, too numerous to mention, who have given me positive feedback on work in progress. Over the last few years I have been the happy recipient of help in the form of books and articles sent to me by scholars in various countries. To Hendrik de Smet, Teresa Fanego, Elly Van Gelderen, Geneviève Girard, Solveig Granath, Günther Rohdenburg, Juhani Rudanko and Michael G. Wherrity, a very warm vote of thanks. A special thanks must go to Ágoston Tóth, who developed the Sisyphus Concordancer, which I used in my research. While on the subject of helpful colleagues, pride of place must go to Dominique Boulonnais, who read through the whole penultimate version of the manuscript, and Anne-Line Graedler, who took on the task of proof-reading the final version. Any faults that remain are a result of my being incapable of leaving well enough alone. Finally I must say thanks to my family; to my parents for their support and encouragement when I first started studying the English language thirty-five years ago, and to my children for theirs when I returned to the study of language some twenty years later. Above all, I must thank Synnøve Skytterholm, without whom the book would not have been written. With good grace she has put up with periods of both physical and, more frequently, mental absence on my part. To use one of my favourite constructions, I can assure her that I intend being more present in the future! Thomas Egan Hamar, November 2007

Chapter 1

The purpose and scope of the study

1.1 Introduction

This book presents the results of a usage-based study of infinitive and -ing complement clauses in present-day British English. It is usage-based in two senses. In the first place it is written in the spirit of Langacker’s “dynamic usage-based model” (Langacker 1999: 91-145). In the second place the data for the study have been taken from actual usage data in the British National Corpus (hereafter BNC). The study thus combines the methods of corpus linguistics and the theoretical stance of cognitive linguistics. Teresa Fanego has recently asked the question “Is Cognitive Grammar a usage-based model? Towards a realistic account of English sentential complements” (Fanego 2004). As a matter of fact, recent years have witnessed many studies that combine a corpus-based methodology and cognitive/functional theoretical approaches, including book length studies such as Gries (2003) and Mukherjee (2005). The present volume is intended to be a contribution to the enterprise of combining corpus and cognitive approaches. As such it addresses directly the area of English sentential complementation referred to by Fanego, to whose question I will return in the final chapter. The object of this study is to further illuminate the distribution and meaning of infinitive and -ing complement clauses, i.e. to cast further light on the contexts in which these forms are used, and the possible reasons for their employment in these contexts. In the space of the last twenty-five years major contributions to our understanding of complement structures in general have been made by such scholars as Givón (1984, 1990, 1995), Langacker (1987, 1991, 1999) and Wierzbicka (1988, 2006), to name but three, and to our understanding of complementation in English by Dirven (1989), Dixon (1991, 2005), Duffley (1992, 1995, 1999, 2000, 2003, 2004, 2006), Fanego (1992, 1996, 1997), Mair (1990, 1995, 2002, 2003), Rohdenburg (1996, 1999, 2003), Rudanko (1984, 1989, 1996, 1998, 2000, 2003a) and Verspoor (1997, 1998, 2000), to mention just some of those who have engaged most actively with these constructions. It is thanks to all of these scholars that the verb “illuminate” in the opening sentence of this paragraph can be qualified by the adverb “further”. However, the very fact that I embarked upon the present study indicates that I believe that there is room for another contribution in this area. I will explain, in the next section, why I believe this to be so. Before doing so, however, some words may be in order about complement clauses in general, and in particular about how I have drawn the distinction between them and what are normally called ‘adjuncts’.

The purpose and scope of the study 2

The distinction between complements and adjuncts is usually seen to depend upon the closeness of the relationship between clauses and (matrix) verbs, on whether or not “their form is determined by the lexical verb” (Greenbaum et al: 1996: 76), or whether “they complete the meanings of the governing verbs with which they share their subjects” (Rudanko 1989: 10). The position which I adopt is similar to that of Mair (1990), who writes:

If a constituent is semantically integrated into the predicate verb phrase, if it fills a structural slot provided for by the ‘valency’ of the predicate verb or if – to use alternative technical terminology – it is ‘subcategorized for’ by the predicate verb, then this constituent is likely to be an object or a complement of the verb. If, on the other hand, a constituent qualifies a clause as a whole by adding information about when, where, how or why an event took place, it is considered to be an adverbial adjunct of the clause. This distinction is useful as long as one bears in mind that ‘a simple binary typology, of complements and non-complements, will cover only part of the facts’ (Matthews 1981: 123). The prototypical ‘object’ or ‘complement’ and the prototypical ‘adverbial’ are not precisely defined and mutually exclusive grammatical categories but opposite end-points in a continuum of forms. (Mair 1990: 200-201)

In this book I use the term ‘complement’ for both clausal ‘objects’ and non-optional adverbials. This means, for instance, that I consider obligatory adverbial -ing clauses after verbs like ‘sit’ or ‘lie’ to be complements (unlike Rudanko (1989: 10) and Duffley (1999: 305), to name but two). From the constructional point of view, a clausal complement is regarded as an obligatory component of a construction. In practice, there is no doubt in most cases as to whether a clause is obligatory or not (whether or not it fills a slot that must be filled). In cases of uncertainty I have followed the example of Mair (1990: 161) in giving suspect utterances the benefit of the doubt and including them in the investigation.

1.2 Why a usage-based study?

There are two main reasons why the present study is usage-based. The first has to do with the very nature of language. As Langacker puts it: “In a usage-based model ‘substantial importance is given to the actual use of the linguistic system and a speaker’s knowledge of this use; the grammar is held to be responsible for a speaker’s knowledge of the full range of linguistic conventions, regardless of whether these conventions can be subsumed under more general statements’” (Langacker 1999: 91, citing Langacker 1987: 494). Furthermore, “In [Cognitive Grammar], a language is described as a structured inventory of conventional linguistic units. The units (cognitive “routines”) comprising a speaker’s linguistic knowledge are limited (by a restriction called the content requirement) to semantic, phonological, and symbolic structures which are either directly

Why a usage-based study? 3

manifested as parts of actual expressions, or else emerge from structures by the processes of abstraction (schematization) and characterization” (Langacker 1999: 98). The present study aims to describe the structures of non-finite complement clauses in English as they are manifested in the actual experience of speakers and writers of present-day British English. It also attempts to describe the relevant “structured inventory of conventional linguistic units” at various levels of schematicity, of which more in the next section. The second reason why the present study is usage-based is practical. Quite extensive reading of the literature on non-finite clauses left me somewhat uneasy about many of the examples on which grammarians base their arguments. It is often difficult to judge to what extent the sort of made-up sentences frequently employed by linguists reproduce the patterns of actual language use. Readers of the literature will often find themselves confronted with examples whose grammaticality must be taken on trust, in the sense that no evidence is offered for their actual occurrence among the products of the users of the language. One reason for making use of made-up examples may be the attraction of minimal pairs as heuristic devices. Consider, in this respect, the following extract from Jackendoff (1990):

(5) a. Sue believes that Joe likes her.

b. Sue believes Joe to like her.

[…] the […] two sentences are virtually synonymous, so it seems clear that these two syntactic frames use (essentially) the same lexical item believe. (Jackendoff 1990: 20)

The problem with Jackendoff’s assertion of synonymy is that one of the two frames in question does not seem to actually occur in (at least British) English. When the subject of the complement predicate in a construction encoding a judgement is an Experiencer, the complement situation is always encoded in a finite clause. In other words Jackendoff’s (5)a. is grammatical, his (5)b. is not. Minimal pairs are very useful for pointing out subtle differences in meaning, and it is very seldom that one encounters such pairs in natura. However, when constructing them, one must take great care that one’s sentences are based upon naturally occurring utterances. Otherwise, one may end up expending energy on the discussion of totally artificial examples. Many linguists of a cognitive bent also rely on made-up examples and, in doing so, risk basing their conclusions on very insecure foundations. Consider the following passage from Hamawand (2002).

A to-infinitive complement denotes the notion of subsequence by symbolising a distance in time between the main and complement events. The to shows that the complement event is a less direct or immediate


event, and thereby makes it subsequent to the main event. […] As illustration, let us have a look at an example. In They like to negotiate/negotiating a deal, the to-infinitive profiles one or a series of distinct occurrences of the process of negotiating. It denotes a bounded event, which is complete by inference. The to-infinitive is seen as prospective, namely the main clause subject is not yet engaged in the complement event and consequently not definite about the chances of making a deal. (Hamawand 2002: 68)

In this passage Hamawand bases his argument on a construction that is not found in present-day English. Today the non-modalised ‘like to’ construction, unlike its modalised counterpart, the ‘would like to’ construction, cannot be used with future reference in assertive utterances. It is only so used with negated matrix verbs and in if-suggestions (see section 5.5.1, p. 168 for details). One thing is the problem created by artificial data. Another is the reliance by modern scholars on outdated genuine data used by their predecessors. Take the following extract from Jørgensen (1990), writing about the ‘remember to have’ construction:

Schibsbye, and not a few other grammarians, fail to mention the important fact that while the present infinitive is absolutely distinct from the gerund in function, the perfect infinitive will, when used – and that occurs pretty often – as a past-time form have the same ‘retrospective’ function as the gerund:

She remembered to have seen the bats flying low over a burnished pool at sunset…. (D.H. Lawrence, The Trespasser…1932: 69)

He spoke familiarly of many names that I remembered to have seen in the newspapers. (John Buchan, The Thirty-Nine Steps… 1954: 10)

She remembered to have heard that the last Merefield… had died in extreme old age. (Patricia Wentworth, The Listening Eye… 1960: 160)

This common usage is attested by Otto Jespersen. (Jørgensen (1990: 147)

As Fanego points out, writing of earlier forms of English, “remember takes a perfect infinitive that explicitly signals that the reference is to past time. This type of construction is no longer acceptable today, but was in common use in Modern English […] and can even be found in texts dating back to the middle of the twentieth century” (Fanego 1996: 75). The data in the BNC may be taken as support for this judgement of Fanego’s. There are just two instantiations of the


‘remember to have’ construction in the corpus. The most recent of them occurs in an extract from a letter written by John Stuart Mill. To round off this brief look at some examples in the literature, let us consider a construction that actually does occur, albeit infrequently, in present-day British English, the ‘see x to infinitive’ construction. There is a small handful of coherent (non-anacoluthic) occurrences of this construction in the BNC. As many as three of these are actually from Duffley (1992), who has in part relied for his examples on Bolinger (1974) and Jespersen (1940). Bolinger, in turn, has taken examples from Jespersen, and one of his examples, I saw John to be obnoxious, is repeated as recently as in Horie (2000). It would appear that some of these examples live a life of their own, begetting heirs in the scholarly literature down through the decades. Examples like these abound in the literature on non-finite clauses. I shall cite no more of them, as the point I am trying to make should by now be clear. It is that the ‘sentences’ upon which some grammarians base their arguments must be approached with caution. Moreover, even where such made-up sentences do bear a resemblance to actually occurring examples there is some doubt as to their exact theoretical status. Utterances exist in space and time. In them various linguemes, to adopt the term used by Croft (2000) to signify entrenched units of varying degrees of complexity, are combined to produce more complex structures, which may take the form of full sentences. However, at the level of Grammar, either that of a particular speaker, or that conventionalised in a particular speech community, there are very few units indeed of clausal complexity. This is not to say that there are no fully entrenched units of clausal length. There are, but for the most part these are saws, proverbs, conventionalised performatives, and the like. Some typical examples from the BNC are:

(1) ‘And every cloud has a silver lining.’ (FU7 347) (2) ‘I pronounce you man and wife.’ (H7P 702) (3) I mean to say.’ (ACE 1134)1 On the whole, however, the grammar, whether of the individual speaker or of the community, contains entrenched units either simpler (such as words and phrases) or more schematic (such as constructions) than full sentences. What, then, are the sentences that some scholars spend so much time discussing? They are types of utterances, which may be based upon genuine utterances or founded on the whims of the author. If they have any claim to existence, it is in the domain of

1 Apart from italicising and underlining to highlight certain strings of words, all tokens are cited exactly as they appear in the World (2000) Edition of the BNC. Thus (3) contains a closing, but not an opening, quotation mark. One could perhaps have omitted this final quotation mark. On the other hand, including it shows that the token is from a written, not an oral, text, and that it represents the final clause in a conversational turn consisting of more than one clause.


scientific theory, in what Popper (1972) calls the third world, distinguishing it from the first world, which is the concrete world all around us, and the second world, which consists of mental representations. Within the model of a grammarian, they correspond to the sentences we find in utterances in the ‘real world’, which they may resemble to various degrees. The more one reads of the literature on complement clauses the greater is the impression one gets that this resemblance is in some cases tenuous indeed. Given that there is a certain amount of false data out there in the literature on complement clauses in English, the question arises as to where we are to look for more accurate data. In answer I can do no better than to quote a passage from McGregor (1997):

[...] it is only through investigations of language in use that we can come to any reasonable understanding of the meanings of grammatical constructions. Meanings are not directly or immediately available to the consciousness or to the introspective examinations of native speakers of a language. It is necessary to amass and study a body of contextualized instances of use in order to come to any adequate understanding of the inherent meaning(s) of any construction (or morpheme, for that matter): the inherent meaning is what remains after contextual meanings have been extracted. Use must be studied in order to come to an understanding of the system that lies behind it. In turn, understanding of the system behind use inputs to the understanding of use [...] (McGregor 1997: 388-389)

The basic goal of the present study was, by examining what McGregor calls ‘a body of contextualized instances’, to provide a sounder basis for making judgements about the actual occurrence and distribution of some forms of non-finite complementation in English. Basically, I was trying to answer three questions:

• What infinitive and -ing complement constructions occur in present-day British English?

• How often do these occur in present-day British English? • Why do certain verbs ‘take’ certain forms of complement?

The answers to the first two questions may be found in the tables of Appendices 1 and 2. The third question is dealt with in Chapters 5 to 7, with particular emphasis on matrix verbs that occur in two or more constructions. The term “body of contextualised instances”, used by McGregor, might serve as a definition of corpus. The present study is corpus-based. It is by no means the first such study of non-finite clauses in English. It does, however, differ from previous studies in two important respects. In the first place, it is the first study with any pretensions to comprehensiveness to deal with both infinitive and -ing complement clauses. Of major corpus-based studies, Freed (1979)


confines her attention to Aspect matrix verbs, Andersson (1985) to different-subject constructions and Mair (1990) and Duffley (1992) to infinitive complements. In the second place, there is a difference of scale. Freed’s corpus consisted of nine novels, Andersson’s of two million words of fiction and academic prose, and Duffley used the Brown University Corpus and the Lancaster/Oslo/Bergen Corpus, which together amount to some two million words. Mair based his investigation on the Survey of English Usage, which at the time contained fewer than 900,000 words. In the present study I made use of the 100,000,000 word BNC. Not only did this give me a much better chance of hitting upon instances of infrequent constructions; it also allowed me to draw firmer conclusions about the relative frequency of various types of constructions than would be possible using a smaller corpus. I hope that the evidence presented in this study concerning the relative frequency of various construction types will be of use to scholars, irrespective of whether or not they are in sympathy with the proposals made to explain the motivation behind the employment of the various forms. These questions of ‘how?’ and ‘why?’ as opposed to ‘what?’ are addressed in the next two sections.

1.3 The form of a usage-based study

In the previous section I stated three goals of this study. Conflated, the first two involved the ascertaining of the relative distribution of various forms of infinitive and -ing complement clauses in present-day British English. The third goal was to attempt to discover the motivation behind this distribution, if indeed any such might be detected. In other words, why do infinitive constructions occur with certain matrix verbs, while other verbs take -ing complements? And why do some matrix verbs take both types? In attempting to answer these questions I proceeded in the same manner as the language learner, by abstracting recurring patterns from the data available in the corpus, in the manner of Figure 1.1. Needless to say, this manner of proceeding was not meant to imply the rejection of all the answers to these questions which have been tendered by previous scholars. It was just a question of keeping an open mind and working upwards from the utterance level. If I happened to arrive at the same answers as previous scholars, from a different starting point, and by a different route, this would only serve to reinforce the validity of their conclusions.


8

Figure 1.1: Illustrating how the meaning of linguemes, both lexemes, in this case ‘ask’, and constructions of varying degrees of schematicity (the ‘ASK to’ construction, the ‘matrix verb to’ construction) may be extracted in a usage-based study

… infinitive…

……(bare) infinitive…

..to infinitive …to NP..

…Effort Verb…

….Matrix Verb to…

…Matrix Verb NP to…..

…Effort Verb -ing…

…Effort Verb to…

…Communication Verb to…

…Attitude Verb to…

…ASK…

…ASK to… …OFFER to…

…AGREE to…

…Ask NP to…

…Ask to… …Beg to… …Request to…

….asks to…. …asked to…

…may ask to…

…has asked to…

… -ing … …to…

The form of a usage-based study 9

Figure 1.1 is intended as a partial illustration of the way in which generalisations may be made on the basis of raw data consisting of utterances containing the ‘ask to’ construction, comprising the bottom line of the figure, and comparable constructions. Strictly speaking, it may be taken to illustrate an inductive process, given that all of the arrows point upwards, starting with the actual utterance level and ending up with the maximally schematic level of constructional morphemes. In actual fact, however, both the grammarian and the language learner also utilise abductive-analogical mechanisms. (See Givón 1995: 59 for the role of such mechanisms in empirical studies.) When faced with a ‘new’ construction, like for example ‘implore to’, the learner/grammarian is likely to place it, provisionally, among the ‘ASK to’ constructions, on the basis of analogy with other, more familiar, constructions to which it seems to bear a resemblance in form and meaning, and to work his or her way ‘downwards’ from there. Figure 1.1 is not only intended to depict a method of analysis. It may also serve as a partial illustration of one possible network model of the constructions in question. (Note, by the way, that the heavy type around certain boxes in Figure 1.1 is not meant to indicate the ascription of prototypicality to the item in question, as is the case with Langacker’s notation for networks, but merely to illustrate the path followed in ‘climbing upwards’ through the various levels of the figure.) That is, not only may the learner learn the constructions in this fashion, they may also be stored in his or her grammar in much the same sort of network. To the extent that this is so the arrows between the boxes in the figure should be double-headed (the language-user wishing to ask someone if they may do something has several suitable, partially synonymous, constructions at his or her disposal). The bottom line in the bottom-up method of analysis illustrated in Figure 1.1 consists of full utterances. There is no room in the figure for these to be reproduced in full, but each occurrence of ‘ask to’ at this level comes complete with a range of arguments and adjuncts. The subject of the matrix verb at this level is a pronoun or a full nominal, or an understood second person in the case of imperatives. At the second level ‘Ask to’ in the heavy type box represents a construction: the ‘subject’ at this level is the argument role of ‘Asker’. This is the level at which lexicogrammatical information is stored. As we climb upwards through the figure, the matrix predicates become progressively schematic for more and more constructions, with the subjects becoming equally generalised. Thus, at level 3, the subject of ‘ASK to’ is an ‘ASKER’ (who may be a ‘requester’ or even a ‘beggar’). At level 4 the subject is a ‘Communicator’, at level 5 an ‘Actor’. How many such argument roles should be allowed for in a grammar is a matter of some dispute among linguists. Dowty, for instance, maintains that “we should not postulate a thematic role type that is limited to one or two verbs (or a small set of near-synonyms), but should rather expect each role type to be applicable to a reasonable range of verb meanings” (Dowty 1991: 556). Dixon (1991: 10) is much more liberal in this respect, maintaining the need for some forty or fifty semantic roles. In fact, the solution chosen depends on one’s


10

attitude to the integrity of the constructions postulated. If one thinks that a construction such as ‘ASK to’ on the third level of Figure 1.1 makes for a feasible unit of analysis or a unit in a mental network - if, in other words, the various ‘ASK to’ constructions have enough in common to justify their being subject to general statements - then it is a question of mere terminology whether we denominate the ‘subject’ of this construction type as the ‘ASKER’ in the manner of Dixon, or the ‘AGENT of ASK to’ in the approach recommended by Dowty. I, myself, find it more convenient to adopt different labels for the role types at the various levels illustrated in Figure 1.1, for the very reason that there are material differences between the degree of schematicity of the various predicates at the respective levels. It seems natural, therefore, to assume that the arguments of these predicates also differ in schematicity. The highest level in Figure 1.1 differs from those below it in that items on this level no longer consist of construction-types but of grammatical morphemes or morpheme-types. These are basic closed-class items, which may be contrasted with the open-class items represented by the verb ‘ASK’ on level 3. In the next section I will take a first brief look at the status of these elements and how they are combined to produce real-life utterances.

1.4 Schemas and Constructions

All linguists, as far as I am aware, would admit the existence of matrix verbs as meaningful lexemes. They would also acknowledge the existence of what are often called complementisers, in English that, to and -ing. All are in agreement that both these types of element are entrenched in the grammar, whether it be that of the individual or that conventionalised in the speech community. Grammarians differ, however, on two important points:

• whether complementisers are to be regarded as meaningful or meaningless elements (so-called ‘markers’), and

• whether the combination of verb + complementiser in a construction is to be regarded as an entrenched unit of the grammar in its own right.

Those who regard complementisers as meaningless do not, as a rule, accord constructions unit status. If to is meaningless, then combining it with ‘ask’ is not likely to result in a form-meaning combination worthy of a separate entry in the grammar. Those, on the other hand, who regard complementisers as meaningful elements are not, thereby, necessarily committed to acknowledging the status of constructions as entrenched units. It could, in theory, be the case that ‘ask’ and to, for instance, were both entrenched units which were combined anew every time somebody wished to request permission to do something. This would, of course, be tantamount to learners climbing all the rungs of the ladder in Figure 1.1 and then pulling the ladder up after themselves! In practice most linguists who accept that complementisers are meaningful regard this theoretical possibility as highly

Schemas and constructions 11

unlikely and accept the existence as meaningful units of (at least some) constructions. I think the evidence in favour of the meaningfulness of constructions is overwhelming. It follows that I also accept that complementisers are meaningful elements. Arguments for their meaningfulness have been advanced by scholars such as Bolinger (1968, 1974), Wierzbicka (1988), Dirven (1989), Dixon (1991), Langacker (1990, 1991, 1999), Duffley (1992, 1995, 1999, 2000, 2003) and Verspoor (1997, 1998, 2000). There is no need to rehash their arguments here. The fact that complementisers are meaningful does not, however, mean that their meaning is at all obvious. It is not. Existing, as they do, at so many removes (so many rungs of the ‘ladder’ in Figure 1.1) from the arena of meaning-in-context, the level of the utterance, their meaning must needs be highly schematic. Scholars such as Dirven (1989: 115) and Verspoor (1997: 418) have raised the question of to what extent it is possible to extract a coherent abstract schema for any particular complementiser. In this study I started by assuming this to be possible and embarked, therefore, upon the quest of discovering just how these abstract schemas should be defined. To put it simply, one of my aims was to fill the boxes in Figure 1.2 with functional, in addition to formal, content.

infinitive -ing

bare infinitive construction

(bare) -ing construction

to to infinitive

construction

to -ing construction

Figure 1.2: The principal forms to be explained in the study

To sum up then, the study aimed

• to establish, with a greater degree of certainty than has hitherto been the case, the relative distribution of the various constructions containing infinitive and -ing complement clauses in present-day British English,

• to elucidate the meanings of the various constructions in question, paying particular attention to those constructions containing the same matrix verb, and

• to establish the meaning of the complementisers to and -ing. In the next section I will describe the types of construction that were investigated.


12

1.5 The scope of the study

1.5.1 The constructions investigated

In a usage-based study there is no gainsaying the fact that quantity matters. Moreover, unless one wishes to presume too much on the indulgence of one’s readers, one is forced to choose between comprehensiveness of coverage and depth of analysis. One can write a lot about a little or a little about a lot. Something will always be lost in the trade-off between comprehensiveness and depth. When dealing with non-finite complement clauses, one can restrict the selection of matrix verbs and/or the types of complement to be dealt with. I chose to do both, but whereas my selection of complement clause types has been quite sharply curtailed, I have aimed at a measure of comprehensiveness in the matter of matrix verbs. The clausal complements that I chose to focus upon are post-verbal infinitive complements, both bare and preceded by to, and -ing complements. A small number of the latter may also be preceded by to. Neither of the lists of matrix verbs chosen as a starting point for this study were designed with these to -ing constructions in mind. Some of them did, however, turn up in the course of my investigations. They are discussed in section 4.5. (See Rudanko 1996: 43-68 for a discussion of more of them.) Wh-clauses, exemplified by (4), were excluded on the grounds that something had to be and they constitute an easily-defined class of their own, which display minimal alternation between complement types, always taking infinitives, either bare or to. Also excluded were the complements of verbs like seem and appear in what are often termed subject-to-subject raising constructions. (5) is a typical example of one of these constructions. (4) You must also consider where to put heaters and radiators, and whether

you still need a fireplace as a focal point in the room. (HGW 358) (5) She seemed to enjoy it’. (CS1 390) The exclusion of constructions like (5) was not due to any theoretical bias, merely to considerations of economy. Apart from this one type of matrix verb, an attempt was made to be as comprehensive as possible by including all the appropriate verbs listed in Visser (1973) which were found by Rudanko (1989) to be in current use. Rudanko’s list was complemented/supplemented by those in Quirk et al. (1985) Ch.16, sections 36 to 42, 50 to 53 and 63. The focus of this study is on constructions containing active voice matrix verbs and infinitive and -ing complements. There are two main types of constructions containing active-voice matrix verbs in English, same-subject (equi-subject, ‘SS’) constructions, in which the subject of the complement clause predicate is identical to the subject of the matrix verb, and different-subject (‘DS’) constructions, in which the subject of the complement predicate is profiled as distinct from the subject of the matrix verb. The two subjects may be

The constructions investigated 13

referentially identical (as in She told herself to…) but they are profiled as two distinct entities. SS and DS constructions are distinguished in the tables in Appendix 2, but are discussed together in the body of the text, where appropriate. (6) – (8) exemplify DS constructions. (6) I was relieved to see a dozen or so people form up in front of me. (AEO 1945) (7) We can anticipate good things happening. (ADE 85) (8) Data General considers full relational databases to be overkill and says

that run-time versions are enough. (CP5 105) In DS constructions the subject of the matrix verb (which I will refer to as S1) is distinct from the subject of the complement predicate (referred to as S2). DS constructions are in some ways more semantically transparent than their SS counterparts. On the whole, in the case of DS constructions, what you see is what you get! This is because the explicit syntactic encoding of two highly profiled figures (two trajectors in Langacker’s terminology) leads to the backgrounding of the role of the speaker.2 DS constructions rarely encode the stance of the speaker to what is being communicated. This is sometimes the case with SS constructions, especially the sort containing matrix verbs like condescend which I have termed Applied Attitude. (See Chapter 2 for an explanation of all the classifications in the study.) On the other hand, DS constructions may well signal the attitude of S1 to the complement situation. In Langacker’s terms (see 1991: 215 & 442), one may say that the matrix sentence in a DS construction is objectively construed by the speaker and addressee. The complement situation may, however, be subjectively construed by S1. I have chosen to lump together all sorts of DS constructions, including those called monotransitive, ditransitive and complex-transitive (by, for example, Quirk et al. 1985 and Mair 1990), what Wierzbicka (1988) calls ‘wanting’ to and ‘opinion’ to and what Dixon (1991) calls Modal (FOR) TO and Judgement TO constructions. These distinctions overlap to some extent. The matrix verb want, for instance, is monotransitive according to the system of Quirk et al. (1985) and Mair (1990), a ‘wanting’ to predicate according to Wierzbicka (1988) and a Modal (FOR) TO predicate according to Dixon (1991). Expect, on the other hand, is complex-transitive according to Quirk et al. (1985) and Mair (1990), an ‘opinion’ to predicate according to Wierzbicka (1988) and a Judgement TO predicate according to Dixon (1991). The fact that I did not initially distinguish between these various classes does not mean that I reject the validity of these distinctions. It was merely a consequence of working upwards from the utterance level. SS constructions may be exemplified by (9) – (11).

2 For an explanation of trajector and other technical terms, see the list of terms in Appendix 3.


14

(9) Although he helped make development a causal science his theories were largely wrong; even so, his preformationist ideas, in which nuclear determinants were unequally distributed during cleavage did stimulate the key experiments. (ASL 410)

(10) I used to enjoy going out. (CH2 11241) (11) He would not see me there; he condescended to see me at my hotel. (GIA 2158) Constructions like (9) – (11) are called same-subject constructions for the obvious reason that the same element functions as subject of the matrix verb and the complement predicate. In (9) he is the subject of both helped and make, in (10) I is the subject of both enjoy and going out (and used, for that matter) and, in (11), he is the subject of condescended and see. It is constructions like (11) which are somewhat opaque in that they encode the stance of the speaker in the manner referred to above. Grammarians of diverse theoretical persuasions, such as Farkas (1988), Mair (1990), Dixon (1991) and Achard (1998), have posited the existence of a separate subject for the complement predicate in SS constructions, which subject is invisible to the naked eye, but which is said to be “missing”, “omitted”, “understood” or “implied” and to be identical in reference to the subject of the matrix verb. I prefer to reserve terms such as these for the understood subject of the complement predicate in utterances like Love does not involve giving fancy parcels tied up with red bows (AYK 901). No further mention will by made of this type of construction. In this study, I adopt an approach similar to that of Matthews (1981), who analyses I want to do it in the following manner:

Subject Predicator Object

I want to do it

Subject Predicator

Figure 1.3: Illustrating the grammatical relations of I want to do it: from Matthews (1981: 80)

My preference for Matthews’ approach is influenced by what I take to be a common-sense disinclination to apply the label ‘omitted’ to elements which are demonstrably all present and accounted for, if not in the exact position one might wish to find them. This approach is, however, very much a minority one among linguists who have written on non-finite clauses. Nevertheless, I do not think anything of moment in my discussion of these constructions hinges upon one’s accepting it and rejecting the ‘understood subject’ approach. I hope, therefore, that the discussion of same-subject constructions will prove of interest to all those

The constructions investigated 15

who are interested in these, irrespective of how they may define the sameness of the subject(s).

1.5.2 The Corpus

The present study is based on data in the British National Corpus. The advantages of using the BNC for a usage-based study such as the present one are obvious. In the first place there is the question of its sheer size. Non-finite complement clauses vary greatly in their incidence. If one wishes to register occurrences of less common constructions, one obviously has a better chance of doing so using a corpus as large as the BNC. In addition the BNC contains material from a wide variety of written and spoken genres and registers. Finally, it is both widely available and widely used, so any conclusions reached can be compared to those of other investigations.3 Starting out with some 310 matrix verbs, it would, in theory, have been possible to extrapolate all of the tokens of these verbs in constructions containing infinitive and -ing complements from the corpus. However, there are two factors which render this task an insurmountable one. Both are related to the size of the corpus. In the first place some matrix verbs, such as have, get and make, occur so frequently as to make manual sorting impracticable (there are as many as 20 matrix verbs represented by more than 50,000 occurrences in the BNC). And, in the second place, strings ending in *ing are so very common that the BNC’s custom-built automatic search engine, SARA, will not search for them. Because of these two factors I was forced to construct a restricted sub-set of the data in the corpus as a whole. The next question was how the content of this subset was to be determined. One possibility was to restrict the study to one genre or register, to ‘spoken dialogue’, for instance. But I did not want to risk losing sight of constructions that occur mainly in written genres (although, according to Greenbaum et al. 1996: 85, it is only in the case of to infinitives among non-finite clause types that one finds significant differences between writing and speech). In the end I decided to restrict my investigation to a certain number of randomly chosen examples of utterances containing each matrix verb, and to examine this subset of utterances for tokens containing infinitive and -ing complement clauses. From the BNC I downloaded sets of tokens that varied in size from 100 to 2,000, before ending up with 1,000 as a number which seemed to give a reasonable guarantee of accuracy in the sense that a selection of

3 Here I must enter a caveat with respect to numbers. While almost all citations in the book are from the World Edition (2000) of the BNC, all total numbers are based on the first edition of the corpus. Improved tagging in the later versions means that the totals for each verb returned for these is likely to be different. In all but a handful of cases the difference is less than 5%.


16

different random samples of tokens of the same matrix verb yielded very similar results with regard to the incidence of the various types of complement clause. No appreciable increase in accuracy was achieved by increasing the sample to 1,500 or 2,000 tokens. This is of course not strictly true in the case of very rare constructions, especially when these contain very common matrix verbs. In some of these cases, where I suspected that my material might be unrepresentative, I went back to the BNC as a whole and made an additional search for these particular items. For verbs represented by less than 1,000 tokens in the BNC, I downloaded all available occurrences. Having proceeded along these lines, I was left with some 310 separate data bases, each consisting of up to 1,000 utterances containing one and the same verb, a certain number of tokens of which presumably contained complement clauses. (This number actually varied from 799 in the case of attempt to none whatsoever in a handful of cases. The latter are listed in Appendix 1 together with the rest of the matrix verbs.) Having ascertained the total number of each type of non-finite clause occurring with each particular verb, it was then possible to project totals for the BNC as a whole. The projected totals for the six most common constructions ranged from some 35,000 in the case of DS -ing to some 300,000, in the case of SS to. The constructions are listed as a) to f) and exemplified in (12) – (17), with the projected totals illustrated in Figure 1.4. Constructions containing passive matrix verbs with to infinitive complements are included in the figure for comparison, although these are not discussed in the text.

a) Same Subject + to infinitive b) Same Subject + -ing c) Different Subject + bare infinitive d) Different Subject + to infinitive e) Different Subject + -ing f) Passive Matrix Verb + to infinitive

(12) Dams helped to reduce river’s flow. (K56 578) (13) She had never imagined wearing clothes like this, or that they could make

such a difference. (AT7 1101) (14) But once I'd done it, I swore I'd help others do it.’ (CCW 2112) (15) But the kids are full of the future and we help them to plan. (CBC 348) (16) He could not imagine her living anywhere else but in Wychwood Forest.

(HHC 446) (17) She was discovered to be a nurse by day and an escort girl by night.

(K4W 4249)

The corpus 17

0

100000

200000

300000

400000

SS to

SS -ing

DS bare

DS to

DS -ing

PMV to

Figure 1.4: Projected totals for tokens of the most common infinitive and -ing

complement constructions, including Passive Matrix Verb (PMV) to infinitive constructions

1.6 The content and organisation of the chapters

The remaining chapters are organised as follows. Chapter 2 describes two systems of classification of the relevant constructions which occur in the corpus. Chapter 3 contains an overview of the conclusions reached in a selection of earlier studies on non-finite complementation in English. The reason for postponing this review until after the presentation of the classification in Chapter 2 is to enable us to employ the terminology introduced in the latter. Chapter 4 builds upon the various analyses in Chapter 3 and proposes an interpretation of the meaning of the complement forms that differs in some respects from all of these. Chapters 5 - 7 are devoted to constructions containing two or more forms of complement either with the same matrix verb or with near-synonymous matrix verbs such as allow and let. In Chapter 5 constructions containing same-time complement situations, such as ‘like -ing’ and ‘try -ing’ are contrasted with to infinitive constructions containing the same matrix verbs. Chapter 6 is the longest in the book. It deals with the domain in which we find the greatest variety of complement types, which is that of the future vis-à-vis the time of the matrix verb. In this chapter we contrast pairs of constructions containing such verbs as bid, fear, help and intend. Chapter 7 contrasts constructions which contain -ing complements encoding situations located prior to the time of the matrix verb, containing verbs like remember, regret and cease with their to infinitive counterparts. Finally Chapter 8 contains a summary of the discussions in the other chapters.

Chapter 2

Classification of the constructions

2.1 Introduction

The constructions in this study are classified along two parameters, one of which refers to the semantics of the matrix verb, the other to the Tense-Aspect-Modality (TAM) relationship pertaining between the matrix verb and the situation in the complement clause. Thus (18) is described as an instantiation of a Forward-looking Communication construction, and (19) as an instantiation of a Backward-looking Aspect construction. (18) He expressly forbade men to call other leaders ‘Father’. (C8L 958) (19) He finished scribbling and held the Filofax sheet up so we could see it.

(HWL 2228) (18) is described as a Forward-looking construction because the actions of calling are profiled as taking place, if at all, in the future vis-à-vis the forbidding. It is described as a Communication construction because forbid here encodes an act of communication. Similarly (19) is described as Backward-looking because the activity of scribbling precedes its conclusion, and as an Aspect construction because the matrix verb encodes the cessation of this activity. The reason for introducing the classification employed at this point, before considering the approaches taken in earlier studies, is to facilitate reference to the various types of constructions in the following chapters. The classification according to the semantics of the matrix verbs will be described in section 2.2 and the classification according to the TAM relationships in section 2.3.

2.2 Classification according to the semantics of the matrix verbs

As pointed out in section 1.5, complement constructions may be divided into two main types, according to the numbers of subjects they contain. In the present work these are referred to as Different-subject, or DS constructions and Same-subject, or SS constructions. Some matrix verbs occur in both types of constructions, others in only one. Matrix verbs occurring in DS constructions are listed in section 2.2.1, in SS constructions in section 2.2.2. The two will be compared in section 2.2.3.


20

2.2.1 Matrix verbs in different-subject constructions

Different-subject constructions contain two subjects, one of which encodes the trajector of the matrix verb, to borrow Langacker’s (1987) terminology, the other the trajector of the complement clause predicator. The first of these will be referred to as S1, the second as S2. In some constructions S2 is also a landmark of the matrix verb.4 This is the case, for example, in (20). In other constructions it is not S2, but the complement clause as a whole that is profiled as the landmark of the matrix verb. (21) is an example of such a construction. (20) My father taught me to shoot, I can't imagine why he didn’t teach me to

drive'. (H8X 2660) (21) Certainly, but I should hate you to forget that he has scored more runs in

Test cricket than any other Englishman. (K3X 873) The distinction between these two types of complement clause is by no means always clear-cut, as is pointed out by Quirk et al. (1985: 1216). Mair (1990: 100) makes the point that: “As there does not seem to be a single syntactic testing criterion which is privileged in the sense that it matches the semantics of matrix verbs more closely than the others, it is the meaning of V + NP + to-inf. structures which, ultimately, provides the only rationale for classifying them as monotransitive, ditransitive or complex transitive.” The semantics in question are postulated by Achard (1998) to be related to the notion of ‘salience’. He writes:

The relation profiled by the main verb can be viewed as the kind of contact established with the complement scene by some conceptualizer. Contact is generally established with a particular entity of the scene, which will be called “salient”. In the case of [two French complement constructions], the infinitival process as a whole and the main participant in that process (the logical subject of the infinitive) are in direct competition for initial salience, i.e. the point at which conceptual contact is established with the complement scene. (Achard 1998: 86-7)

The categorisation of matrix verbs in the various different-subject constructions in this book resembles Achard’s in its emphasis on salience. It is rooted in Langacker’s notion of the interpretation of the profiled situation in terms of a “stage model”. Discussing Perception constructions, Langacker writes: “the role of the perceiver is in many ways analogous to that of someone watching a play [...] the stage model idealizes a fundamental aspect of our moment-to-moment experience: the observation of external events, each comprising the interaction of

4 For an explanation of trajector and landmark, see the list of technical terms in Appendix 3.

Matrix verbs in different-subject constructions

21

participants within a setting” (1991: 284). In categorising the constructions, I take it that the complement clause situation is profiled as unfolding upon a stage and then, ignoring the role of the speaker, enquire as to the nature of the role played by S1 (the matrix verb subject). Is S1 profiled as a mere member of the audience, as it were, as in Figure 2.1, or does he or she play some role in guiding the action on the stage, such as that of prompter or director, for example, as in Figure 2.2? The figures are simplified representations of actual speech situations. They are similar in this respect to those of Achard, who writes: “For ease of exposition, the figures in this chapter only represent the speaker’s conceptualisation (the objective scene…), and the viewing organization within that scene. The speaker herself, as well as the main conceptualizing relation are not represented.” (Achard 1998: 188) In both figures MV stands for ‘Matrix verb’, CCV for ‘Complement clause verb’.

observes influences

Figure 2.1: S1 as ‘Observer’: does not Figure 2.2: S1 as ‘Influencer’: play an active role in the plays an active role in the on-stage situation. on-stage situation.

Both Figure 2.1 and Figure 2.2 illustrate constructions with to infinitive complements. The form of the complement is not material to the point here, which is that, in constructions like that illustrated in Figure 2.1, S1 is primarily an observer of the onstage situation; he or she may merely observe it (Perception constructions), think about it (Mental Process constructions) or form an attitude towards it (Attitude constructions), but does not influence its unfolding. In

S2 CCV to S2

to

CCV

S1

MV MV

S1

22 Classification of the constructions

constructions like that illustrated in Figure 2.2, on the other hand, S1 actually mounts the stage, so to speak, either to issue advice or instruction to S2 (Communication constructions), to facilitate the realisation of the complement clause situation (Enablement constructions) or even to enforce its realisation (Causation constructions). A minority of Communication constructions are of the type illustrated in Figure 2.1. These are those in which the addressee of the communication is a fellow member of the audience, as it were, rather than S2. ‘Describe S2 -ing’ is a case in point. Although Figures 2.1 and 2.2 are based on Langacker’s notion of the on-stage model, they reflect differences between two types of construction that have been noted by linguists of various schools. Dixon, for instance, refers to them as “Secondary-B” (my Figure 2.1) and “Secondary-C” (my Figure 2.2) and distinguishes between them in the following manner:

Secondary-C verbs are like Secondary-B verbs in introducing just one role (the subject of the secondary verb) in addition to the roles of the verb in the complement clause. They differ semantically in that Secondary-B verbs simply describe the subject’s attitude towards some event or state (John wants/expects Mary to propose the toast) whereas the subject of a Secondary-C verb plays a role in bringing about the event or state (John forced/permitted/helped Mary to propose the toast). (Dixon 1991: 192-3)

All of the DS constructions were divided into six categories. Below is a concise semantic description of these categories. Each definition is accompanied by an illustrative example, some of which have been truncated.

Perception constructions: S1 merely registers the action on stage. E.g. You noticed me watching you, I shouldn’t wonder. (HGV 4587)

Mental Process constructions: S1 registers the action onstage, thinks about it and may form a judgement about its ontological status. E.g. Most scientists believe the infill to be lava…... (GW6 803)

Attitude constructions: S1 registers the action on stage and formulates an attitude towards it. E.g. I dread mine reaching their teens. (BLW 1071)

Communication constructions: S1 is not on stage as such, but may prompt the onstage S2 from the wings, as it were, or may comment on the onstage situation to another member of the audience. E.g. The master commands you to stay. (HGE 788)


23

Enablement constructions: S1 either sets the stage for S2’s realisation of the complement clause situation or assumes the role of a minor character in assisting S2 to realise it. E.g. Outside he allowed her to examine the bird. (FPF 1467)

Causation constructions: S1 directs the realisation of the complement clause situation by S2, who has no independent say in the matter. E.g. He halted, forcing the rest of the field to bunch up behind him. (HP0 2042)

Two points should be made at the outset about these categories. The first is that they are not to be considered watertight. Some matrix verbs take complements in several categories. Some matrix verbs straddle several of them. The second point is that some verbs do not seem to fit very comfortably into any of them. I will illustrate both of these problems, starting with matrix verbs that belong to two categories. (22) ‘feel’ ( Perception): She watched the way the wind filled it and felt the

boat dig deeper into the water, moving under her like some wild creature straining against a leash. (H7W 1931)

(23) ‘feel’ (Mental Process): Women often feel that they are not at risk because they do not feel themselves to be associated with any of the so called ‘high risk groups’ and, although we are encouraged to have safer sex, AIDS does not really seem to be an issue for us. (CJ9 650)

(24) ‘tip’ (Communication): B Wilson He was also the only person who tipped Linfield to win the championship last season but perhaps you forgot about that. (K2D 1748)

(25) ‘tip’ (Enablement): Either you tip the dustmen to take it away or you take it to the tip yourself. (A5N 76)

Note that, in the case of feel, the same matrix verb occurs with a different complement form in (22) and (23). In (24) and (25) the forms of the complement are identical, although the meaning is different. The fact that both matrix verbs display the same degree of polysemy with nominal objects lends support to the decision to ascribe them to different categories when they function as matrix verbs. One thing is the question of how to classify constructions that appear to have a clear foot in more than one category: quite another is the problem posed by those that appear to find no natural place in the system of classification. One such matrix verb is require, exemplified in (26) and (27).


24

(26) We will require the gas and electricity companies to invest in insulation and other energy-saving measures. (AMA 607)

(27) The deal involved a transfer fee plus a complicated sponsorship package, which required Keegan to advertise and do promotional work for Scottish and Newcastle Breweries. (A6Y 749)

‘Require S2 to infinitive’ constructions stipulate the imposition by S1 of an obligation on the part of S2 to realise the complement clause situation. One might choose to regard the imposition of this obligation as a form of causation and classify it, therefore, as a Causation construction. On the other hand, one might choose to focus on the act of imposition and classify it as a peripheral member of the Communication category, or on the indirection of the causation and classify it as an Enablement construction. Although in some doubt, I have chosen to include it both among the Enablement and the Causation constructions, dividing the tokens among these two on the basis of the degree of directness involved in the causing action. The term ‘peripheral member’ employed in the previous paragraph, raises the question of the internal structure of the categories, a matter which time and space preclude our exploring in detail here. Suffice it to say that they are to be viewed as prototype-based categories rather than classical criterial-attribute categories. (See Lakoff 1987 for a detailed discussion of the distinction between the two types.) In other words each class contains both central and more peripheral members. Take the six classes of different-subject constructions comprising Perception verbs, Mental Process verbs, Attitude verbs, Communication verbs, Enablement verbs and Causation verbs. Each of these contains central, prototypical members, such as see, think, like, tell, enable and cause and more peripheral members such as catch, hazard, trust, mime, seek and avoid. The six categories of different-subject constructions are all traditional ones. Some of them may be labelled differently by various scholars, and the borders between them may be drawn somewhat differently, but, on the whole, there should be no problems involved in comparing the discussions of the various constructions in this study with other studies. Table 2.1 contains a comparison of the categories with those of some other grammarians, chosen more or less at random. Achard’s classification for French is included to show that the classes are by no means language specific.


25

Table 2.1: The present classification of different-subject constructions contrasted with various other classifications

Present study Palmer 1988 Dixon 1991 Verspoor 2000 Achard 1998 Perception Perception Attention Perception Perception Mental Process Futurity/

Report Thinking/ Wanting

Mental Process/ Judgement

Propositional Attitude

Attitude Attitude Liking Attitude Emotional Reaction

Communication Futurity/ Report

Speaking/ Relating

Communication Declaration

Enablement Causation Letting/ Helping/ Deciding

Manipulative Causative

Causation Causation Making Causative Causative Table 2.2 contains a list of all the matrix verbs in the study belonging to each category. The lists are not, of course, meant to be exhaustive for each category, but they are comprehensive enough to be representative. Eight of the verbs, fancy, feel, find, forbid, miss, perceive, require and tip, are listed twice as they are used in two distinct (although related) senses. They are written in italics in the table.

Table 2.2: Semantic classification of matrix verbs taking different-subject infinitive and -ing clauses. Verbs listed twice are in italics.

Semantic Type Matrix Verbs

Perception catch, discover, feel, find, hear, miss, notice, observe,

overhear, perceive, see, smell, spot, spy, taste, watch Mental Process adjudge, anticipate, assume, believe, conceive, consider,

contemplate, envisage, expect, fancy, feel, find, foresee, forget, guess, hazard, imagine, intend, judge, know, perceive, plan, presume, recall, reckon, recollect, remember, suppose, take, think, understand

Attitude abhor, abide, adore, bear, begrudge, brook, crave, deserve, desire, detest, dislike, dread, endure, fancy, fear, hate, justify, like, love, mind, miss, prefer, regret, relish, resent, stand, trust, want, wish

Communication acknowledge, advise, announce, ask, beg, beseech, bid, challenge, claim, command, convince, counsel, dare, declare, defy, demand, describe, direct, discourage, encourage, enjoin, entreat, exhort, forbid, implore, incite,


26

instruct, invite, mention, mime, order, own, persuade, plead, pledge, press, proclaim, profess, pronounce, propose, recommend, remind, report, request, summon, suggest, tell, tip, urge, volunteer

Enablement afford, allow, appoint, arrange, assign, assist, authorise, bribe, choose, contract, design, elect, employ, empower, enable, engage, entitle, equip, fit, forbid, help, involve, let, mean, name, need, pay, permit, require, risk, save, seek, suffer, teach, tip, tolerate, use, vote

Causation avoid, cause, compel, condemn, constrain, decide, delay, detail, determine, drive, force, get, have, impel, induce, inspire, keep, lead, leave, make, oblige, postpone, preclude, prevent, prompt, push, require, set, start, stop, trouble

2.2.2 Matrix verbs in same-subject constructions

Same-subject constructions contain just one subject. This subject encodes both the trajector of the matrix verb and the trajector of the complement clause predicate. Thus, in (28), she is said to be both the subject of contrives and the subject of have, and in (29) He is both the subject of vowed and of challenge. (28) Effortlessly, she contrives to have a violin class which clashes head-on

with hockey. (H9Y 724) (29) He vowed to continuously challenge the government. (CH2 10254) Many grammarians have posited the existence of a separate subject for the complement predicate in constructions such as those in (28) and (29), which subject is said to be “missing”, “omitted”, “understood” or “implied” and to be identical in reference to the subject of the matrix verb. As mentioned in section 1.5, I think this sort of approach adds an unnecessary complication to the analysis of same-subject constructions. The same-subject constructions were divided into six categories on semantic grounds. A glance at the semantics of the matrix verb is often, though not always, sufficient to indicate the class to which any construction should be assigned. Below is a concise description of these categories. Sp stands for the speaker, by which is meant the producer of an utterance, whether it be spoken or written. S stands for the subject.

Effort constructions:

S expends energy of some sort in order to realise some sort of situation. E.g. Vietnam would strive to achieve an early comprehensive political solution to the war in Cambodia. (HL8 81)

Matrix verbs in same-subject constructions

27

Mental Process constructions:

S thinks about being (engaging) in some sort of situation. E.g. For a wild moment she contemplated locking the door against him.... (JYE 3415)

Attitude constructions:

S formulates an attitude towards some sort of situation which may possibly involve him or her. E.g. I hate going to the cinema. (KP5 1743)

Communication constructions:

S describes, raises the question of, or expresses an attitude concerning, his or her possible participation in some sort of situation. E.g. I agreed to assist. (FPN 1210)

Aspect constructions:

S commences, continues or discontinues his or her participation in some sort of situation. E.g. They began to ascend the staircase. (G0E 3461)

Applied Attitude constructions:

Either Sp or S formulates an attitude about some situation (definitely) involving the latter. E.g. Nevertheless she condescended to reply. (CB5 2468)

The above definitions resemble those employed for different-subject constructions in that the categories are not meant to be considered watertight. Their composition is in part dictated by the need to divide up the corpus data into manageable chunks for further analysis, and this process leads inevitably to a certain blurring of boundaries, with some potential for inter-category leakage. Be that as it may, five of these six categories are traditional ones. The first four all contain what Givón (1995: 118) calls “non-implicative modality verbs”, since they do not normally imply that the situation in the complement clause actually takes place. (There are some exceptions such as the Effort constructions ‘manage to infinitive’ and ‘contrive to infinitive’.) The category I call ‘Aspect’ contains those matrix verbs often called aspectuals or aspectualisers, which normally entail the actual occurrence (or non-occurrence) of the complement situation. Finally, the category which I have chosen, for want of a better term in the literature, to label ‘Applied Attitude’ contains constructions that partake, to some extent, of the characteristics of both Attitude constructions and of Aspect constructions. They resemble the latter in that they indicate clearly whether or not the complement situation was realised; they resemble the former in that they always encode an attitude of some sort, either on the part of the speaker or the subject, to this same situation. Consider, in this respect, (30) – (32). (30) As Mary admitted she liked to see the videos first alone - she also may

have found the situation rather threatening. (HNW 766) (31) Eliot continued to see the danger of London reduced to a waste land where

(with a momentary glance at Gerontion's coughing goat) debris lies. (A6B 1725)


28

(32) He would not see me there; he condescended to see me at my hotel. (G1A 2158)

In (30), an instantiation of an Attitude construction, the addressee knows that Mary has a preference for unaccompanied video watching, whenever called upon to view one. Although she could not have acquired such a preference without some experience of the activity in question, the actual utterance does not refer to any particular exercise of the viewing faculty. In the Aspect construction in (31), on the other hand, the addressee is left in no doubt that the state of danger perception is one actually experienced by Eliot. Again the complement situation is encoded as habitual, but this time there is no hint as to whether either Eliot himself or the speaker approves or disapproves of this penchant of his. (32) resembles (31) in that the addressee understands that the action of ‘seeing’ was definitely carried out. However, that is not all he or she is given to understand. It is also evident that the speaker considers the subject to have acted in conflict with some sort of innate sense of snobbery in coming to meet him. It is this combination of attitude with action, of the expression of an attitude applied to a definite action, that I have labelled ‘Applied Attitude’. Table 2.3 contains a list of matrix verbs belonging to each of the six categories. We have already seen how some matrix verbs taking different-subject complements had to be assigned to two or more separate categories, while others straddle two categories. The same point applies to some few same-subject matrix verbs. Any verbs listed twice in Table 2.3 are in italics.

Table 2.3: Semantic classification of matrix verbs taking same-subject infinitive and -ing clauses, with verbs listed twice in italics

Semantic Type Matrix Verb

Effort aim, assay, attempt, battle, bid, conspire, contrive,

endeavour, essay, fail, help, labour, learn, manage, push, seek, strain, strive, struggle, sweat, try

Mental Process anticipate, consider, contemplate, decide, determine, envisage, expect, foresee, forget, imagine, intend, judge, know, mean, overlook, plan, plot, purpose, recall, reckon, recollect, remember, resolve, scheme, think, trust

Attitude abhor, abide, ache, adore, aspire, bear, burn, care, chafe, crave, delight, desire, detest, die, dislike, dread, enjoy, face, fancy, fear, fret, hanker, hate, hope, like, loathe, long, look, love, mind, pant, pine, prefer, regret, relish, stand, stomach, thirst, value, want, wish, yearn

Communication acknowledge, admit, agree, apply, ask, beg, beseech, claim, confess, consent, contract, crave, decline, demand, deny, describe, discuss, engage, entreat, implore, mention, offer,

Matrix verbs in same-subject constructions

29

plead, pledge, pray, press, proclaim, profess, promise, propose, refuse, renounce, report, request, swear, threaten, undertake, volunteer, vote, vow, whine

Aspect begin, cease, commence, complete, continue, discontinue, fall, finish, get, grow, keep, lie, proceed, quit, recommence, remain, repeat, resume, set, sit, start, stay, stop, take, use

Applied Attitude

affect, afford, ape, arrange, assume, avoid, bear, begrudge, bother, choose, condescend, dare, defer, defy, deign, delay, deserve, disdain, elect, endure, escape, eschew, evade, face, fall, feign, forbear, get, grudge, hasten, help, hesitate, invite, justify, make, mime, miss, move, need, neglect, omit, pay, postpone, presume, pretend, require, resent, resist, risk, scorn, scruple, serve, shirk, shun, speed, stand, suffice, tolerate, trouble, undertake, venture

Like their different-subject counterparts, the six categories in Table 2.3 are to be considered prototype-based, rather than classical (categorial-attribute), in nature. Each of them contains central, prototypical members and more peripheral members. As one would expect of categories structured around prototypes, some matrix verbs appear to fit more comfortably into these classes than is the case for others. For instance, verbs like start, love and forget are clearly Aspect, Attitude and Mental Process, respectively. Not all verbs admit of such ready classification, however. There are also verbs that are not difficult to categorise but which quite clearly belong to two classes. For instance bear is clearly an Attitude verb when followed by a to infinitive, as in (33), but what about when followed by -ing, as in (34)? (33) She could never bear to watch her brothers fight, least of all when one of

them decided to give Patsy a pasting. (EEW 1818) (34) The tune can, of course, always bear strengthening, and this can be done

either by giving it to the two trumpets in unison, and the alto and tenor parts to the tenor trombones, or by giving the alto and tenor parts to the 2nd trumpet and 1st trombone respectively and doubling the tune in the lower octave on the 2nd trombone. (GVS 1041)

In (33) it is clear that the subject is encoded as possessing an attitude to fraternal combat. In (34) it is the speaker, if anyone, who is encoded as possessing an opinion about the melodic structure of the piece of music in question. Equally difficult to categorise in terms of these six classes is the verb stand followed by to and meaning ‘be in a position to’. When followed by -ing it is invariably attitudinal, as in (35). The same is occasionally the case with to infinitive clauses, as in (36). But what are we to make of sentences like (37) and (38), which would appear to represent the dominant sense of stand when followed by a to infinitive clause?


30

(35) ‘I don't think I could stand being a nurse because I would get too involved, so I hope one day to join the fire brigade,’ she said. (CF9 154)

(36) ‘You know how I can’t stand to be fussed over.’ (FPK 495) (37) Pupils with special needs also stand to benefit from other developments in

mainstream education. (CJG 77) (38) I’m the one who stands to lose the licence, pal. (HGN 3091) The verb that stand in (37) and (38) most resembles is risk, which only occurs with -ing clauses, and which also defies easy categorisation in terms of my six categories. One possibility would be to establish a residual category to contain verbs like risk and some senses of verbs like bear and stand. In the end I decided to classify all three as Applied Attitude verbs where the content of the ‘attitude’ in the case of stand is limited to epistemic predication.

2.2.3 Different-subject and same-subject constructions compared

Both different- and same-subject constructions were divided, in the previous two sections, into six classes. Three of these bear the same label, to wit the Mental Process, Attitude and Communication classes. The Perception, Enablement and Causation classes require the presence of two separate (or at least separately encoded) participants, a Perceiver, Enabler and Causer, and a Perceived, Enabled and Causee, and thus preclude same-subject constructions. Effort matrix verbs are, in a sense, the single trajector counterpart of the Enabled and Causation categories, differing from them in that the originator and executor in the case of Effort constructions is one and the same participant. I should, however, add that only two matrix verbs, help and push, occur in both Effort and Enablement/Causation constructions in my material. In the case of Aspect constructions, the semantic content of the matrix verbs has been bleached to the extent that the semantic role of the subject is determined by, or co-varies with (to adopt a less deterministic terminology), the complement clause predicate rather than the matrix verb itself. There is thus no room for an independent role to be played by a second subject. Finally, Applied Attitude constructions resemble to a certain extent Aspect constructions in that they stipulate whether or not a certain action is carried out by the subject, at the same time as they encode the attitude of either that same subject or the speaker to the complement situation. Table 2.4 contains details of the forms of complement clauses that occur in my material in the various classes.

Different-subject and same-subject constructions compared 31

Table 2.4: Forms of complement occurring with various type of matrix verb in both different-subject and same-subject constructions

Complement form to infinitive bare infinitive -ing to -ing Mental Process DS/SS DS/SS Attitude DS/SS DS/SS SS Communication DS/SS DS DS/SS SS Perception DS DS DS Effort SS SS SS Enablement DS DS DS Causation DS DS DS Aspect SS SS SS Applied Attitude

SS SS SS

We can see from Table 2.4 that Mental Process, Attitude and Communication matrix verbs may occur in both same-subject and different-subject constructions. However, only a minority of the matrix verbs in each class actually do occur in both environments. The relevant totals are given in Table 2.5.

Table 2.5: Matrix verbs occurring in both different-subject and same-subject constructions compared to those only occurring in one or other of these types, with horizontal percentages

Matrix verbs only in SS only in DS in both SS & DS Totals Mental Process 9 23% 14 35% 17 43% 40 Attitude 22 44% 8 16% 20 40% 50 Communication 25 32% 36 47% 16 21% 77

At this stage it seems pertinent to ask how informative Table 2.4 actually is? How useful would the generalisations it contains be to the language learner, for example? I do not mean to suggest that the language learner would be constructing hypotheses downwards from the maximally schematic level, nor that he or she would be doing so in the absence of actual language data, but, nevertheless, at a certain stage mental connections must be made, networks must begin to take shape in the language learner’s internal grammar. I do not think it necessary to go into detail here about the deficiencies of Table 2.4 in this respect. Suffice it to say that it seems impossible to draw any worthwhile inferences with respect to the form of a complement given the information contained in the first


32

column. In the next section we will consider the question of whether a classification in terms of TAM relations is more informative in this respect.

2.3 Classification in terms of TAM relations between matrix verb and complement situation

The second mode of classification is based on the relationship between the matrix verb and the complement situation and, in particular, on whether the latter is profiled as occurring at a particular time vis-à-vis the time of the matrix verb. Thus while the classification in the previous section is verb-based, the one in the present section takes into account the constructions as a whole. Even the most cursory glance at corpus data will reveal that some complement situations are profiled as occurring at the same time as the situation encoded by the matrix verb, some as occurring prior to that time, and some after it. The three possibilities are illustrated by (39) – (41) respectively. (39) We could see the ship to Hamburg coming closer. (FPU 1963) (40) I recall her pacing the sitting-room while I am doing my homework,

pausing every so often to stand at one of the windows and look down into the busy street below. (HD7 1331)

(41) I’ve asked the restaurant to bring you round a takeaway. (JXT 1200) Constructions like the one instantiated in (39) will be described as Same-time, constructions like (40) as Backward-looking. The terms should be both self-explanatory and uncontroversial. Both types imply the occurrence of the situation in the complement clause, which may either be once-off, as in (39) or serial/recurrent, as in (40). Constructions like (41) will be described as Forward-looking. This term, unlike the other two, calls for some comment. What exactly is meant when we say that the situation in (41) occurs after the time of the matrix verb? In the very nature of things we cannot be as certain of the occurrence of a situation in this domain as is the case for situations in the present and past. Unlike the other two types, Forward-looking matrix verbs do not necessarily entail the occurrence (or non-occurrence) of their complement situations. Our world knowledge would indicate that the expectations of the subject in (41) with respect to the delivery of the food ordered are likely to be satisfied, but we also know that there is no cast-iron guarantee that this will be the case. In some cases, indeed, such as (42) for example, the expectations of the subject are explicitly shown to be unfulfilled. (42) We expected people to just come and go, but most didn’t - they had a great get together.’ (GX2 302) It is this element of uncertainty regarding the realisation of the complement situation that has led many scholars to label the to infinitive as encoding potentiality (see Chapter 3 for a discussion of some of these approaches). What

Classification according to TAM relations 33

exactly is meant by ‘potentiality’ in this connection? It means, I think, that the situation in the complement clause is a plausible candidate for realisation in what is termed ‘projected reality’ in Langacker’s dynamic evolutionary model, illustrated in somewhat simplified form in Figure 2.3. Past Present Future

Figure 2.3: An illustration of the dynamic evolutionary model, based on Langacker (1991: 277)

In Figure 2.3 the area enclosed by the two horizontal lines represents reality, past, present and future. Note that the present is not illustrated as merely momentary, but as having a certain (short) extension in time. The fact that the lines are dotted rather than solid in the domain of the future is meant to illustrate the existence of a lower degree of certainty with respect to the actual occurrence of situations in this domain. The representation of the future in the figure is a simplification of Langacker’s model, in that he distinguishes between two future domains, which he refers to as potential reality and projected reality (Langacker 1991: 277-278). The difference between the two relates to the degree of certainty with which we can predict future situations on the basis of present facts and past experiences. In my analysis I conflate these two domains to one, which I refer to as the projected future. My reason for so doing is grounded in problems in applying the binary distinction between potential and projected reality to the various Forward-looking constructions in the corpus. The distinction is of definite value in some cases. For instance, as we shall see in section 6.3, the contrast between some Forward-looking to infinitive and -ing constructions may profitably be analysed in terms of degree of likelihood of the complement situation coming about. Nevertheless, if we confine our attention to Forward-looking to infinitive

non-reality

reality

non-reality


34

constructions, we find a cline of likelihood with respect to the realisation of their complement situations ranging from rather unlikely to very likely indeed. The degree of likelihood varies according to the power of the subject to bring about the realisation of the complement situation, something which is very difficult to factor out of the analysis. Given these difficulties I decided to stick with the category I call projected future for all Forward-looking constructions. It refers to the domain of situations after the time of the matrix verb as the subject or speaker expects these to evolve. The three classes Same-time, Backward-looking and Forward-looking do not exhaust the inventory of constructions. There are three more types, illustrated here by (43) – (45).

(43) ‘In the afternoon my wife likes to go out and look around car boot sales.

(K3X 908) (44) He imagined opening the drawing-room door. (H8N 626) (45) I had considered my body to be lumpy, untidy, anomalous and entirely

unsuited to the person within. (CEE 908) There is no consensus in the literature about how to interpret the construction in (43). I will therefore discuss it at somewhat greater length than the other five types. As I see it, constructions such as (43) profile situations as likely to occur at more or less regular intervals. One might go so far as to say that they are realised by the subject on all suitable occasions. They are instances of what Langacker (1999) calls general validity predications. He writes of these that

The situation they describe may hold for either a bounded or an unbounded span of time, i.e. their validity has a temporal scope. An indefinite, potentially open-ended set of instances of the basic event type can occur within that scope. General validity predications do not however profile these instances, but rather the higher-order relationship (of genericity/habituality) that they constitute or manifest. (Langacker 1999: 249-250)

Although Langacker himself does not mention to infinitive complements in his discussion of general validity predications, his characterisation of these could well have been drawn with constructions like the one in (43) in mind, so neatly does it encapsulate the sense of the predication in the example. According to Langacker's definition, an indefinite, potentially open-ended set of instances of the basic event type can occur within a span of time. Although the predication encodes a higher-order relationship rather than any specific occurrences within the bounded or unbounded time span in question, utterances may well contain an indication of the circumstances under which such actuation might be expected to take place. The likely realisation of a situation may be triggered by what Langacker (1997: 206) refers to as enabling conditions. According to Langacker such conditions “constitute part of the conceptual basis for the characterization of general validity predications. [...] Enabling conditions are not homogeneous in


nature or amenable to any simple characterization (e.g. strict causality): they can be implicit or explicit, general or specific, salient or non-salient, etc.” (1997: 206). Enabling conditions may be encoded by temporal adverbials, as in Langacker's examples, cited here as (46) – (48). (46) A cat stalks a bird whenever it gets a chance. (47) A cat usually stalks a bird whenever it gets a chance. (48) A cat occasionally stalks a bird when the opportunity presents itself. (Langacker 1997: 206: my italics) The situations in (46) – (48) are encoded as likely to occur at irregular intervals. In particular the situation may be encoded as likely to occur whenever a suitable opportunity presents itself. We find similar encoding options in general validity predications encoded in to infinitive constructions. Examples (49) – (51), for instance, all resemble (48) in containing the adverbial 'occasionally'. (49) Edie was not dismayed and continued to exhibit occasionally in mixed

exhibitions at the Leicester Galleries and at the London Group. (F9U 529) (50) Stuart, just for something to say, occasionally used to go 'Adams, Aitken,

Apted, Bell, Bellamy...’ (EDJ 313) (51) Martin occasionally, and fancifully, likes to picture himself as a daredevil

cuester, a flashing Jimmy White to my grinding Cliff Thorburn. (ECU 1513) Situations may also be encoded in to infinitive complement constructions as likely to occur at more regular intervals, as in (52) – (54). (52) We do strongly recommend that you continue to weigh yourself regularly.

(AD0 1996) (53) She used to get tight regularly. (B34 213) (54) If you like to drink alcohol regularly, you will probably find this is the

most pleasant week of the diet, provided you don't experience a reaction to it. (FEX 1762)

Situations may even be encoded as likely to occur invariably, as in (55) – (56) (55) He always liked to tell me things about my lover that my lover kept from

me. (FAT 2835) (56) Is there a certain time when you always love to sit down with a relaxing

drink and something to eat? (AD0 1563) What do I mean when I say that events may be encoded as likely to occur invariably? I mean that they are likely to be realised on all suitable occasions. In (55), for instance, the exchange of information may be presumed to take place


36

during intimate conversations between the two parties concerned. And in (56) the speaker questions the addressee about the possible existence of this sort of occasion. The notion of suitable occasions will be elaborated upon below. Enabling conditions, multifaceted as they are, may be signalled by a multiplicity of adverbial types. Some more examples are contained in (57) – (61), all with the matrix verb like.

(57) I like to get out on the road when I can or go to the studio. (CK5 1286) (58) He liked to finish it each morning before he fetched the new day’s papers.

(AC4 2555) (59) Retrievers are particularly keen to play with this type of toy, while all

puppies like to have something to chew, especially when they are teething. (CJE 220)

(60) I like to use an Ibanez Tube Screamer quite a bit; I have other pedals like EQ, delay and compressor, but I hardly use them.’ (C9L 2365)

(61) With clothes, most of us like to believe that our tastes are idiosyncratic, which suggests that we will often go out of our way - and pay through the nose - to buy something a bit different. (BNC 727)

The adverbial in (57), “when I can”, is similar to the one in Langacker's example, cited as (47) above, “whenever it gets a chance”. They both imply the realisation of the complement situation on all suitable occasions, at the same time implying that such occasions are not likely to arise all that often. The occasion of likely realisation referred to in (58), “each morning”, is, of course, of much more regular occurrence. Whereas both (57) and (58) encode habitual predications, (59) is generic. The occasions of likely realisation in this case are “when they are teething”, a period of inevitable occurrence in the life of small dogs. The adverbial in (60), “quite a bit”, differs from those in (57) – (59) in encoding frequency rather than ‘time when’. It implies that the complement situation is realised at frequent intervals. Finally, the adverbial in (61), “With clothes”, is an adjunct of respect rather than a time adverbial. However, it may be paraphrased as ‘when it comes to (thinking about) clothes’. In this sense it may thus be understood to refer to the occasion of likely realisation of the complement situation. In examples (57) – (61) the enabling conditions, or likely occasions of realisation, are all encoded explicitly. There are also many cases in which the enabling conditions are implicit. In such cases we must make use of our world knowledge to decipher them. (62) – (64) are cases in point. (62) Will you continue to get your prescriptions, even if the medicines are

expensive? (G2P 295) (63) Does your dog like to sing to the soaps? (A17 135) (64) Harry loves to talk about his gardening exploits. (ACY 851)


Our world knowledge informs us that we are likely to get prescriptions filled when we run out of medication, that the dog referred to in (63) must be in the same room as a television set tuned to a soap opera for the question of his singing to arise, and that Harry in (64) is likely to launch forth on the topic of gardening whenever a willing ear can be found. To put it in a nutshell, the complement situations in these examples are likely to be realised whenever a suitable opportunity arises. Moreover, what is to be understood by 'a suitable opportunity' can vary greatly, without this variety causing us any problems in the practical interpretation of utterances. As Langacker points out with respect to enabling conditions in general, these “are complex, multifaceted, and not reducible to any single notion (e.g. causation). Yet neither are they mysterious or inherently problematic. In referring to enabling conditions we are essentially only recognizing that an event type's encyclopedic characterization subsumes more than just its profile, and that general knowledge has a ubiquitous role in the construal of expressions” (1997: 207). In the case of positive attitude constructions such as non-modalised ‘like to’, ‘love to’ and ‘adore to’ (the modalised versions of these all encode a complement situation located in the projected future), the complement situation is encoded as likely to be realised by the matrix verb subject on all suitable occasions within the relevant time span. There are also some negative attitude constructions containing to infinitive complements that encode general validity predications. (Positive attitude constructions are discussed in detail in section 5.5.1, negative ones in section 5.5.2.) Whereas the positive constructions encode situations which the subject would strive to realise on all appropriate occasions, the negative ones encode situations the subject would strive to avoid on all occasions of their possible realisation. (65) – (69) exemplify to infinitive complements with hate, loathe, can't bear and can't stand. (65) She hated to use the word ‘common’, only common people called other

people ‘common’; her mother taught her that. (FRC 1228) (66) He was so soft-hearted, he hated anyone to cry. (FSF 86) (67) ‘I have broken up with my boyfriend and I can’t bear to let any man touch

me. (CBF 294) (68) I can’t stand to see people being cruel to animals it absolutely appals me,

it does really, I feel like to take an axe to them (KCY 1467) (69) ‘You loathe to be categorized. (ECU 3509) In (65) the subject tried to avoid using the word ‘common’. Similarly, in (66), the matrix verb subject was of such a disposition that he tried to avoid occasions of people crying. On no occasion will the subject in (67) consider allowing a man to touch her. Similarly the subject in (68) will do his or her utmost to avoid passively observing cruelty to animals. It is not so easy for the subject in (69) to avoid realisations of the complement situation. Whether he is characterised or not is up to others to decide. Nevertheless, should an occasion of characterisation arise he will invariably react to it with detestation. Note that it is possible that the


38

complement clause situation in utterances such as these may never be realised on what Langacker refers to as the actual as opposed to the structural plane (see Langacker 1997: 202-205). As Langacker puts it: “a general validity predication does not necessarily imply the occurrence of any event instance” (1997: 198). A generic sense for some to infinitive constructions is also posited by Freed (1979), who writes: “In general, the to V form of a sentential complement, whether occurring with start, begin, continue, or cease, carries with it a generic reading. Once again a generic reading of an event suggests a REPETITION (or a series) of the event in question, occurring at DIFFERENT MOMENTS throughout an unspecified stretch of time. In all cases the temporal nature of these is intermittent…” (1979: 152) If one were to adopt Freed’s analysis, one would have to classify non-finite complements along three axes rather than two. The ‘start to infinitive’ construction would be classified as ‘Forward-looking Generic Aspect’, the ‘continue to infinitive’ construction as ‘Same-time Generic Aspect’, etc. When one examines the tokens in the corpus, however, one finds little evidence for the putative generic nature of to infinitive complements after begin and start (see Figure 6.10 on p. 264). I therefore categorised these two simply as Forward-looking: ‘cease to infinitive’ and ‘continue to infinitive’ are labelled General, which is the term I have adopted for general validity predications. General constructions encode situations located on a structural or virtual plane. They are not however situated in a mental space exclusive to any one individual, unlike the two constructions instantiated in (44) and (45). (44) encodes a situation as unfolding before the mind’s eye of the subject, with no intimation as to whether this situation has ever been, or ever will be, realised. Examples (70) – (72) are similar to (44) in this respect. (70) I envisaged myself staggering down, a dog under each arm, and roared

them to heel. (AS7 1829) (71) A JUDGE is considering recommending deportation of a man convicted of

deception. (K55 1406) (72) Jane contemplated designing a jumper based on her favourite sport of

canoeing, but found it too ambitious. (CA2 1084)

(70) – (72) resemble (44) in that they all encode an act of ‘thinking about’ something on the part of the subject without this something’s being located either in the definite past or present or in the projected future. If someone is considering or contemplating doing something, they are merely entertaining a mental image of so doing, without any element of commitment to the realisation of the situation in the complement clause. This sort of construction may be distinguished in this way from other Mental Process constructions such as ‘remember -ing’, which locate the complement situation clearly in the time before the matrix verb, or ‘expect to infinitive’ which locate it clearly in the future. They thus form a class of their own. It is a small class, containing only seven matrix verbs and ten constructions. These will be referred to as Contemplation constructions.


Constructions like (45) resemble Contemplation constructions in that they too encode a mental act. In their case, however, this act involves the formulation of a hypothesis about some situation or some participant in a situation which may be located within the dynamic evolutionary model. The hypothesis in question may be formed by the subject, as in (45), (73) and (74), or by the speaker, as in (75) and (76).

(73) Amnesty international considers Moncef Triki to be a prisoner of

conscience. (CFG 361) (74) He assumed the pigtail to be her own, but it wasn’t. (FSP 569) (75) How can people in Scotland presume to call themselves ethnic Scots and

even more presumptuously assume ‘English immigration’ might be a passing phenomenon. (K5M 5235)

(76) She wrote that note while pretending to look in her handbag. (HR7 3533) In different-subject constructions like (73) and (74) the hypothesis, which is entertained by the subject, is encoded explicitly. In same-subject constructions like (75) and (76) the hypothesis, which is entertained by the speaker, is less explicit. Thus in (75) the hypothesis is that people who call themselves ethnic Scots are exhibiting a measure of presumption, in (76) that the subject did not actually carry out the action of looking. Both these construction types will be referred to as Judgement constructions. We have now distinguished six types of construction on the basis of the relationship between the matrix verb and the complement situation. These are defined and exemplified below.

Same-time constructions:

Some situation is profiled as occurring simultaneously with the matrix verb. E.g. I can taste blood running down the back of my throat...... (A74 3123)

Backward-looking constructions:

Some situation is profiled as occurring before the time of the matrix verb. E.g. Adam stopped walking and stared at her. (HA6 1924)

Forward-looking constructions:

Some situation is profiled as expected to occur after the time of the matrix verb. E.g. He then requested Lucy and Jean to come into the kitchen. (HHB 4158)

General constructions:

Some situation is profiled as likely to occur on a more or less regular basis. E.g....God’s own country, as the Canadians delight to call it. (A0P 217)


40

Judgement constructions:

Some situation is hypothesised to be true. E.g. Theodora judged her to be in her early forties.... (HA2 802)

Contemplation constructions:

Some situation is profiled as occurring in some domain. E.g. For a wild moment she contemplated locking the door against him... (JYE 3415)

Just as is the case with the semantic classification of matrix verbs in the previous section, one encounters occasional problems in classifying constructions in terms of these six categories. The cases of ‘cease to infinitive’ and ‘continue to infinitive’, which could plausibly be classed either as General, or as Backward-looking and Same-time respectively, have already been mentioned. One might also wonder whether Judgement Applied Attitude constructions, such as (75) and (76), might not be better classified as General and Same-time constructions. Does not the subject in (76) pretend at the same time as she looks? The answer, of course, is that she does not, because she does not actually look at all! In other words, only one action is encoded in this construction, and that is an action designed to give a false impression to onlookers. In this respect ‘pretend to infinitive’ resembles ‘seem to infinitive’ and ‘appear to infinitive’, which are both pure (i.e. maximally transparent) Judgement constructions. We encounter a different sort of problem in the case of some constructions containing Attitude matrix verbs, especially matrix verbs that are almost always used non-assertively, like bear and mind. It is not always easy to distinguish between Same-time and Forward-looking readings of some constructions. Compare (77) and (78). (77) Do you mind eating a bit later on? (KE3 6549) (78) Well yeah you don’t mind giving anybody a lift. (KCT 8678)

In (77) it is clear that the subject is being asked about his or her present attitude to a future situation. In (78) the attitude, in this case one of ‘not minding’, seems to refer to a same-time situation. As the majority of the occurrences in the BNC resemble (77) rather than (78), ‘mind -ing’ has been categorised as a Same-time construction (as has ‘bear -ing’). A final problem is posed by what one may term ‘exceptions’, where a matrix verb is used with an unusual form of complement. For instance, the downloaded samples contained 476 tokens of ‘allow S2 to infinitive’ and 147 tokens of ‘cause S2 to infinitive’. They also contained one token of ‘allow S2 bare infinitive’ and one token of ‘cause S2 bare infinitive’. Should one therefore categorise allow and cause as verbs that may occur with both types of infinitive? They are certainly not equally likely to occur with both. Mukherjee comments on these sorts of language data in the following passage:


Generally speaking, corpora contain a vast amount of performance data, i.e. the linguistic behaviour of many native speakers in many different communicative situations. However, corpus-based models of grammar should not attempt to explain all corpus data in their entirety, because the data will always include instances of, say, unacceptable language, clear mistakes and intended ungrammaticality. It stands to reason that performances of such and similarly dubious kinds should not be included in a corpus-based grammar of language use. (Mukherjee 2005: 87)

I would be reluctant to label the tokens containing allow and cause as unacceptable or clear mistakes (they are included in Appendix 1 so readers can make up their own minds on this point). They seem to function perfectly well in context. Nevertheless, they are clearly idiolectal and, as such, examples of the sort of performances which Mukherjee maintains should be excluded from a grammar of language use. It may be worth pointing out, by the way, that should one decide to include the constructions containing allow and cause with bare infinitive complements in the system of classification proposed in the present section, this would have no consequences for the system as such, given the existence of Forward-looking bare infinitive constructions with such matrix verbs as let and make. Indeed one may speculate that the very existence of such constructions in the language may facilitate the idiolectal production of the bare form of the infinitive after allow and cause. Be that as it may, grammars have a tendency to leak, as was pointed out by Sapir (1921), and there would appear to be no reason why complement constructions should be less prone to leakage than other form-function combinations. One can never completely exclude the possibility of a construction being employed in certain circumstances to encode a non-canonical function. Moreover, in cases where a form-function combination has changed in the recent past, it would not be surprising to encounter occasional vestiges of earlier usage. For instance, some general validity predications were encoded by -ing clauses as recently as Late Modern English. The BNC itself contains a few tokens of ‘prefer -ing’ being used to encode general validity predications, a function for which the form was generally employed two hundred years ago (see section 5.5.3). My material as a whole, however, did not contain sufficient tokens of -ing complements employed in this manner to warrant positing for the present-day language as a whole a class of General constructions containing an -ing complement. Apart from these problematic cases, the constructions slot comfortably into one or other of the six categories. Moreover, unlike the classes which were defined in terms of the semantics of the matrix verb, most of these classes contain both same-subject and different-subject constructions, as may be seen in Table 2.6.


42

Table 2.6: Forms of complement occurring with various types of matrix verb in both DS and SS constructions according to the relationship between the matrix verb and the complement situation

Complement forms to infinitive bare infinitive -ing to -ing

1. Same-time DS DS/SS 2. Backward-looking DS/SS 3. Forward-looking DS/SS DS/SS DS/SS SS 4. General DS/SS 5. Judgement DS/SS SS 6. Contemplation DS/SS

The classification in Table 2.6 has several obvious advantages compared to the one in Table 2.4. In the first place we can see at a glance that it allows for 24 possible combinations of form and function, of which only 11 are realised, in contrast with Table 2.4, which allows for 36 possible combinations of which as many as 27 are realised. More importantly, Table 2.6 highlights similarities between the functions of different- and same-subject constructions which are camouflaged in Table 2.4. Moreover, Table 2.6 shows the (almost) complete predictability of the form of non-finite complement in the case of three of the six categories, Backward-looking, General and Contemplation. In addition, it predicts that non-finite Judgement constructions always contain the complementiser to, followed by either an infinitive or an -ing form, and that Same-time constructions never contain the complementiser to, irrespective of whether they contain an infinitive or an -ing form. In short, it is only the form of non-finite Forward-looking constructions that remains totally unpredictable given the information contained in the table. The fact that the classification in Table 2.6 is superior to that in Table 2.4 should not be taken to imply that the former is of no value. It is certainly true that we can make substantive generalisations about the complementation patterns of the class of Judgement constructions as a whole, to take one of the classes in Table 2.6, in a way that is not possible for the class of Mental Process constructions as a whole. However, at a slightly more fine-grained level, it may also be possible to make substantive generalisations about the complementation patterns of the class of Mental Process Judgement constructions that are not valid for the class of Communication Judgement constructions. At an even more fine-grained level we can make generalisations about the ‘consider S2 to infinitive’ construction that are not valid for the ‘suppose S2 to infinitive’ construction. The claim I am making here is merely that at the coarser-grained levels of Tables 2.4 and 2.6 the system of classification employed in the latter is more useful than that employed in the former to the


linguist (or language learner) seeking general patterns in the system of English non-finite complementation. Table 2.6 does not tell us anything about the relative frequency of the various construction types. The number of construction types in each category is indicated in Table 2.7. The numbers for same- and different-subject constructions are amalgamated.

Table 2.7: Number of constructions in the present study according to the relationship between the matrix verb and the complement situation

Complement forms to

infinitive bare

infinitive -ing to ing Totals

1. Same-time 8 73 81 2. Backward-looking 29 29 3. Forward-looking 220 12 67 3 302 4. General 19 19 5. Judgement 48 3 51 6. Contemplation 10 10 Totals 287 20 179 6 492

All constructions in the present study are classified using both systems of classification, according to the semantics of the matrix verb, and according to the relationship between that verb and the complement situation. The constructions are listed alphabetically in Appendix 1, together with an indication of their frequency and an example of each. Appendix 2 contains tables with actual and projected totals for the number of tokens of each same-subject and different-subject construction, sorted according to the types of matrix verb. Having now established a frame of reference that allows us to distinguish the various sorts of constructions, we are in a position to embark upon a consideration of the conclusions reached in a selection of earlier studies of non-finite complementation. This forms the topic of the next chapter.

Chapter 3

Earlier Studies

3.1 Introduction

Countless scholars have engaged with the challenges posed by infinitive and -ing constructions in English, so many indeed that one cannot hope to mention more than a small proportion of them in an introduction of this sort. I have chosen to concentrate on the work of some scholars who have been particularly influential in the field and to present a selection of their main arguments thematically, rather than chronologically. Thus, in section 3.2 I look at the work of some grammarians who have described the distinction between the constructions in terms of whether they profile a situation as specific or general, in 3.3 at some work that discusses them in terms of whether the matrix verbs imply the occurrence or non-occurrence of the complement situation, and in 3.4 at the related notions of fulfilment and non-fulfilment. Section 3.5 will be devoted to the work of Conrad who sees the distinction in terms of referring and non-referring predicates and 3.6 to the work of Wierzbicka who sees the to infinitive form as encoding futurity and the -ing form actuality. Section 3.7 presents the views of Duffley who describes the distinction between the two forms in terms of futurity versus interiority, and 3.8 to the contributions of Verspoor who discusses the distinction in terms of various degrees of immediacy, and more or less direct causality. Section 3.9 contains a short summary of the views of Langacker and 3.10 a presentation of the opinions of Smith and Escobedo. Finally section 3.11 contains a summary and an overview in tabular form of the main points discussed.

3.2 Specific versus general

Wood (1956) is one of the earliest attempts to explain the use of infinitive and -ing clauses with various sorts of matrix verb, in particular with those verbs that collate with both forms of complement.5 He nails his colours to the mast at the outset by stating that “in a few cases the difference in meaning [between minimal pairs of sentences containing the two forms] is negligible, but difference there is;

5 Duffley (2003: 325) points to some earlier treatments by Sweet (1891), Jespersen (1940), Poutsma (1904) and Kruisinga (1931).

Earlier Studies

46

and in other cases it is of much more consequence” (Wood 1956: 11). As for the exact nature of the difference between the two constructions, Wood takes the to infinitive clause to always imply the presence of an agent, while the -ing form is said by him to encode situations in general. He states:

…. where the infinitive, although it does not specify an agent, usually implies one, the gerund represents the activity as it were in vacuo, without reference to any agent or occasion. When we say To lie is wrong we are thinking of you, me or anyone else telling lies; that is to say we think of it in connection with a subject, though not with any particular subject, and with various, though unspecified occasions. But when we say Lying is wrong we are thinking of the activity or practice in a universal sense, as a vice having an existence independent of the individual who succumbs to it. (Wood 1956: 11)

Why does Wood posit the presence of an understood agent as subject of the to infinitive form? I would suggest that To lie is wrong evokes the idea of choice, of the alternative of telling the truth, and it is against the implicit background of this alternative that the action of choosing to lie is likely to excite opprobrium. The exercise of choice presupposes the presence of a chooser. Lying is wrong, on the other hand, takes the action of lying as a given in the particular context, and merely states that this action is to be condemned. Having stated the basic distinction between the two forms as he sees them, Wood goes on to discuss various distinctions of usage between the two forms, making in the process many acute observations concerning these. The distinction between the general and specific is introduced in the following passage:

The gerund denotes something more general, the infinitive something more specific. Contrast I don’t like writing to the papers (expression of a general dislike) and I don’t like to write to the papers (on a particular occasion or subject). (Wood 1956: 13)

This ascription of a more general perspective to the -ing construction is one that is shared by other scholars, among them Dirven (1989). Couched, as it is, in subjective terms, it is not especially amenable to invalidation, but Duffley (2003), in a corpus study of infinitive and -ing clauses as subjects, writes that “One obvious observation revealed by the study of the corpus is that there is no significant difference between the gerund and the infinitive as to their capacity of expressing particularity or generality” (2003: 335). Another problem with this formulation is that one may readily propose alternative interpretations of Wood’s examples. One might, for example paraphrase the sentence containing -ing as “I write to the papers and this practice gives me no (simultaneous) pleasure”, in other words as a Same-time rather than a General construction. (For a definition of these categories see section 2.3: for a detailed discussion of the distinction with respect to verbs such as like see section 5.5.1, p. 162.) As for Wood’s to infinitive example, it could either encode a specific Forward-looking predication, the sense

Specific versus general 47

ascribed to it by Wood, or it could encode a general validity predication. Of 107 non-modalised tokens of ‘like to’ in my material, all 83 positive polarity tokens encode general validity predications. Among the 24 negative polarity tokens, 10 encode general validity predications, the other 14 specific predications. All 14 are Forward-looking. (79) is an example of the General type, (80) of the Forward-looking type. (79) ‘First of all because Italians don’t like to go abroad! (CBV 1360) (80) Gloria seemed so pleased that Dot didn’t like to say how constricting the

coat felt round the neck. (AC5 376) Of 20 tokens of positive polarity ‘like -ing’, all are Same-time, either once-off, as in (81), or serial as in (82). The 25 negated tokens are also all Same-time, either once-off, as in (83) or serial, as in (84). (81) A letter from your Aunt Emily told us how much she likes having you and

how accomplished you are becoming. (H8X 1203) (82) WE KNOW you like shopping at factory shops. (K54 3184) (83) I don’t like her being there, and I'm lonely on my own. (B0W 1028) (84) ‘I don’t like being got at. Sam’s always getting at us.’ (AT4 123) To be fair to Wood, he recognised that the to infinitive could also be used in a general sense. In the first passage quoted above he refers to its being used on “various, though unspecified occasions” (Wood 1956: 11). And immediately after the last quoted passage, he goes on to write:

It is true that a general sense may sometimes attach to the infinitive, but usually as a series, a succession or a recurrence of a specified fact or situation. As a child he loved to watch the trains go by suggests his watching them on a number of different occasions as the opportunity occurred; he loved watching the trains go by merges all these occasions into one and abstracts the characteristic common to them all. (Wood 1956:13)

In this passage Wood equates general predications, which I take to be higher-order predications (see section 2.3) with serial/recurrent predications, which I see as instantiating repeatedly actuated situations. However, when he uses the phrase “as the opportunity occurred”, Wood puts his finger on one of the prototypical characteristics of General to infinitive constructions, which is that they encode situations which are likely to be realised whenever a suitable occasion arises. He does not, however, pursue this train of thought. Interestingly, this same minimal pair of Wood’s is employed by Dirven (1989) in his preliminary characterisation of the two forms. He writes:

In both sentences the speaker presupposes the truth of the complement, namely that someone watched the trains go by. But in the first sentence

Earlier Studies

48

with the to-infinitive it is each single occurrence of the process of watching and consequently also the series of individual occurrences. In this respect, the to-infinitive is comparable to a countable noun which may denote one or several instances of a given entity or event. In the second sentence with the gerund, it is no longer the individual occurrences of watching the trains, but rather the unspecified and unbounded duration of some phenomenon, here watching trains. In this respect, the gerund is comparable to an uncountable noun or mass noun: not one single occasion or series of single occasions is meant here, but some vague extension of the process of watching. (Dirven 1989: 115)

Duffley is critical of Dirven’s interpretation of the two sentences. He writes: “This analogy appears to have a rather tenuous basis in reality. Although in [the sentence containing the to infinitive complement] one does feel an impression of separate occurrences, this would seem to be linked to the impression of the speaker trying to watch trains go by whenever he could rather than to any inherent singularizing value attached to the infinitive. As for the claim that the gerund-participle denotes the unbounded and unspecified duration or vague extension of a process, it is entirely without foundation” Duffley (2006: 75). For my part, I would agree with Dirven’s assertion in this passage that the to infinitive complement form resembles in some respects a countable noun, with the proviso that it cannot be used to encode one as opposed to several “instances of a given event”: to infinitive constructions used in this way necessarily imply a sample greater than one. As for the characterisation of the -ing form as encoding “some vague extension of the process of watching”, I would suggest that we can drop the qualifier “vague”. As I see it, the example in question instantiates a Same-time construction. As such it indicates that the watching and the pleasure derived from it were coterminous. In defence of Dirven’s characterisations, it should be stated that he, himself, considers these to be exploratory and provisional in character. Indeed, he states that “we simply know too little about complementisers at present to venture upon a decisive, general characterisation of the cognitive distinctions they embody” (1989: 116). Suffice it to say at this point that we know one cognitive distinction that they do not embody, which is that the to infinitive profiles a situation as more specific than the -ing form.

3.3 Factive versus non-factive

A very influential paper which viewed the distinctions between the infinitive and -ing forms in logical terms was “Fact”, first given by the Kiparskys in 1967 and first published in 1970. This paper contains two specific contentions about the distribution of infinitive and -ing complement clauses. The first states that “Gerunds can be objects of factive predicates, but not freely of non-factive predicates” (Kiparsky and Kiparsky 1971: 347), and the second that “Only non-factive predicates allow the accusative and infinitive construction” (1971: 348). Given that factive predicates must, by definition, encode events perceived as

Factive versus non-factive 49

either same-time or in the past, the contention that gerunds can be objects of factive predicates tallies with the evidence of the present study, which shows that if a Same-time or a Backward-looking predicate takes a non-finite complement, this will be in the -ing form, regardless of whether the matrix verb in question is factive or not. As for the assertion that gerunds cannot freely be objects of non-factive predicates, this is also true to the extent that there are some forms of construction, such as Judgement constructions, which never contain an -ing complement. On the other hand there is no doubt that the -ing form occurs in many non-factive constructions. Indeed, out of a total of 179 constructions in this study containing an -ing form of complement, as many as 77 refer to complement situations located either in the future, vis-à-vis the time of the matrix verb, as in (85), or simply in the mind of the subject, as in (86). (85) The longer they had delayed acting, the more difficult it had become for

them to act. (CDS 260) (86) Apparently, the blank white space on that page matched the expression on

my face as I envisaged programming the Alpha 2.0 unaided. (C9M 1201)

In fact, of all four non-finite complement forms in the present study, the -ing form occurs with the greatest variety of matrix verb types. The Kiparskys also claim that only non-factive predicates allow the accusative and infinitive construction. They use the term “accusative and infinitive construction” to refer to what I term Judgement constructions, exemplified by (73) and (74) in Chapter 2, repeated here for convenience as (87) and (88).

(87) Amnesty international considers Moncef Triki to be a prisoner of

conscience. (CFG 361) (88) He assumed the pigtail to be her own, but it wasn’t. (FSP 569) That Judgement constructions are non-factive is, of course, true. The question is whether we can make anything of this fact in our exploration of the meaning of the complement forms. The Kiparskys do not pursue the point. So much for matrix verbs which clearly indicate the factivity or otherwise of the complement situation: the Kiparskys were, of course, aware that not all matrix verbs do so. They write:

[…] we find verbs which occur indifferently with factive and non-factive complements, e.g. anticipate, acknowledge, suspect, report, remember, emphasize, announce, admit, deduce. Such verbs have no specification in the lexicon as to whether their complements are factive. On a deeper level, their semantic representations include no specifications as to whether their complement sentences represent presuppositions by the speaker or not. (Kiparsky and Kiparsky 1971: 360)

Earlier Studies

50

The Kiparskys appear to argue that, when used with an -ing clause, such verbs may be factive, when used with a to infinitive clause, they are always non-factive. There is no doubt that the distinction is true of General and Same-time Attitude constructions, at least: he likes to watch trains is non-factive, he likes watching trains factive. The distinction does not hold for General and Same-time Aspect constructions, like he continued to watch trains and he continued watching trains. Nor does it hold for Forward-looking constructions, all of which are non-factive, regardless of the form of the complement clause. Despite these drawbacks to their analysis, the Kiparskys did shed light on certain areas of non-finite complementation. More importantly, perhaps, they served as a source of inspiration for others to seek semantic explanations for the differences in distribution between various types of clausal complement. According to the Kiparskys, the choice of complement form is influenced by the speaker’s presuppositions about the truth value of the proposition expressed. Other approaches which are based on the notions of presupposition and consequence, or of implicativity and non-implicativity, are those of Karttunen (1971), Freed (1979) and Rudanko (1984 & 1989). What these approaches have in common is an emphasis on the modality of the predication in the complement clause (which modality is to some extent dictated by the matrix verb). Karttunen proposed a distinction between what he called implicative and non-implicative matrix verbs. In the case of the former, which includes verbs such as manage and remember, a non-negated matrix verb is said to presuppose the realisation of the complement situation, while the negation of the matrix verb implies that the complement situation is not realised. In the case of the non-implicative verbs, such as agree and hope, there are no implications as to the realisation of the complement situation with either assertive or negated matrix verbs. Karttunen points to some important distinctions between these two types of predicates, both of which are forward-looking, in the sense that the complement situation, if it is realised at all, occurs in the future vis-à-vis the matrix verb. It is unfortunate that he confines his attention to one form of complement, the to infinitive, as the distinction he draws may well be relevant to constructions containing either an infinitive or an -ing form of complement with the same matrix verb. To take just one example, he includes try among the non-implicative predicates, although the ‘try -ing’ construction, unlike its ‘try to’ counterpart, is clearly implicative. I have one final, perhaps minor, caveat with Karttunen’s analysis, and this concerns his use of the term ‘obligation’. He writes: “Remember as the main verb presupposes that the subject was obligated to carry out the act described in the complement. It also presupposes a basic willingness on the part of the subject to carry out his obligation” (Karttunen 1971: 351). In actual fact, I think the notion of ‘obligation’ is more suited to describe peripheral uses of ‘remember to’, such as the one in (89) than more prototypical uses such as in (90) – (91).

Factive versus non-factive 51

(89) No matter what you drink, remember to do your bit for the environment by using the bottle and can banks which are now available in most Sainsbury’s car parks. (BN7 1389)

(90) But at least, thought Charles, I remembered to pull up my socks. (ASE 2111) (91) so er we must remember to buy him a card and something. (KD8 6309)

There is no doubt that in (89), which we may term ‘hortatory remember’, the speaker is trying to impose an obligation on the addressee. This is a construction much favoured by advertisers of all sorts. More common, though, are examples such as (90) and (91), which are more suitably described in terms of ‘prior intention’. Rudanko is another scholar who discusses non-finite constructions in terms of intention. Rudanko (1989) contains a classification of over 220 same-subject constructions, 200 of which are categorised in terms of four criteria, related to the notions of desideration, intention, endeavour and communication. The remainder are classed as exceptions. One point made by Rudanko relates to a tendency for situations viewed negatively by the speaker or subject to be encoded in -ing clauses: “it is being suggested here that Equi is characteristically governed by verbs expressing volition, and further, that infinitival Equi constructions typically co-occur with verbs of positive volition while the -ing pattern, lacking to and the historical associations of to, displays a tendency to favor verbs expressing negative volition. The former tendency is more marked than the latter” (Rudanko 1989: 150). This tendency was confirmed by the data from the BNC. Of 24 matrix verbs employed in Same-time Attitude constructions, only four, adore, enjoy, like and love, are positive in connotation, and 56% of the tokens of ‘like -ing’ are negated. All of the approaches mentioned in this brief discussion of factivity, implicativity and intention have made important contributions to our understanding of some types of non-finite complement constructions. Their main weaknesses lie in the number of constructions for which they fail to account. For instance, exceptions classed as “other” by Rudanko, include constructions containing such common matrix verbs as miss, discuss, remember and pretend. We turn next to the related notions of fulfilment and non-fulfilment.

3.4 Non-fulfilment versus fulfilment

The contrast between non-factive and factive predicates corresponds, to a certain extent, to the contrasts between non-fulfilment/fulfilment proposed by Kempson and Quirk (1971) and between potentiality and performance posited by Quirk et al. (1985). These distinctions resemble the non-implicative/implicative distinction to the extent that they account for the distribution of a greater number of constructions than does the factive/non-factive distinction. For instance, all Same-time and Backward-looking -ing constructions imply the fulfilment/performance

Earlier Studies

52

of the situation in the complement clause when the matrix verb is affirmative. To be characterised as factive, this implication would have to be maintained under negation, something which is not the case for many constructions (‘try -ing’ for example). There are, however, a number of problems involved in describing the -ing form in terms of fulfilment or performance. Some of these are recognised by Quirk et al., who write that in the case of some matrix verbs,

[…] the difference is more subtle, and may be overruled or neutralized by the meaning of the verb of the main clause. For example, the negative meaning of avoid and escape cancels out the sense of ‘performance’ in He escaped/avoided being branded a traitor. (Quirk et al. 1985: 1191)

‘escape -ing’ and ‘avoid -ing’ may both be classified as Forward-looking constructions, since the eventuality of being branded would be expected to come about in the projected future in the normal course of events. All such constructions normally encode situations the eventual realisation of which is in some doubt. In other words one does not have to think in terms of the negativity of the matrix verb overriding the meaning of the complement form in order to explain the non-realisation of the complement situation. The very fact that the two constructions referred to by Quirk et al. are Forward-looking is sufficient to exclude the ‘performance’ reading. Quirk et al. are not alone in ascribing the sense of ‘potentiality’ to the to infinitive construction. One of the first to propose this characterisation was Bolinger who asserted that “there is a properly semantic contrast between the nominalizations carried by -ing and those carried by the infinitive. It is a contrast between two aspects: reification versus hypothesis or potentiality” (1968: 124). We have seen in Chapter 2 that the -ing form is used in four different contexts, in Backward-looking, Same-time, Forward-looking and Contemplation construct-ions. The term reified is obviously suitable for the first two of these. Is it equally apt for Forward-looking constructions, such as (92) and Contemplation ones, such as (93)? (92) Bolger intends running St Jovite over a more suitable 10 furlongs in the

Derrinstown Stud Derby Trial at Leopardstown before tackling the Epsom Classic. (AKE 1619)

(93) Forget taking the train for an evening at the theatre or the weekend family excursion into the country. (K5D 8224)

(92) clearly refers to an event situated in the projected future, in that, all being well, it will come about in due course. (93) refers to situations that are profiled as ‘out of time’, as it were: they are purely hypothetical. One may suitably paraphrase (93) as “Forget the idea of taking….”. May examples such as these be properly described as reified? Bolinger might answer that in these sorts of contexts we are dealing with what he terms metaphorical rather than literal reification.

Non-fulfilment versus fulfilment 53

Verbs such as want, wish, hope, expect, command, etc. apply to unrealized possibilities; the complement that goes normally with them is the infinitive. Verbs such as enjoy, visualize, detest, understand, deny, approve, etc. usually apply to actualities or to possibilities conceived as actualities – literally or metaphorically they reify the action; their normal complements are -ings. (Bolinger 1968: 127)

This explanation in terms of metaphorical reification accounts for examples (92) and (93). In (92) the subject, Bolger, is profiled as regarding the horse’s participation in the race as something that will unfold in the future. In (93) the subject is enjoined to consider the two activities. If he is to do so they must unfold before his mind’s eye. The question is whether it would not be simpler to characterise the -ing form in terms of profiling a situation as extended or unfolding in some unspecified domain, in the manner of Dixon, discussed below. When the domain in question is then specified as same-time or in the past, it would follow that the extended or unfolding situation is literally reified. Bolinger characterises the to infinitive in terms of hypothesis and potentiality. We have seen in section 2.3 that the to infinitive is used in three different ways, in Judgement, General and Forward-looking constructions, exemplified here by (94) – (96). (94) He assumed the pigtail to be her own, but it wasn’t. (FSP 569) (95) Whatever our Customers want to say, we like them to speak to the staff at

Shell stations first. (BNX 999) (96) No, I simply wish them to be removed from this place. (G17 1536) There is no doubt that the complement in the Judgement construction in (94) may properly be described as hypothetical. It is even open to invalidation. But does this characterisation aid us in distinguishing it from the Contemplation -ing construction, exemplified in (93)? Is the putative connection of the ponytail to the head of the girl not seen as real in the mind of the ‘assumer’ in (94) and, if this is the case, may we not describe the predication in terms of metaphorical reification? The distinction between the hypothetical on the one hand, and the metaphorically reified on the other, does not seem to provide a very solid basis on which to distinguish these constructions. Turning to the General and Forward-looking senses in (95) and (96), I would agree that the to infinitive always encodes a situation which is profiled as a potential candidate for realisation. However, I think that one might go a step further and describe it, at least in some cases, as encoding a likely candidate for realisation. In the following passage, Dixon (1991), another advocate of the ‘potentiality’ characterisation, teeters on the brink of postulating probability rather than mere potentiality as a salient characteristic of the form.

Earlier Studies

54

A (FOR) TO complement, as in (30b) [I remembered to lock the door (but then Mary took the key and pushed it down a grating, so I couldn’t)], refers to the involvement in an activity of the subject of the complement clause (which is here coreferential with the subject of the main clause). The unmarked situation is that he would if at all possible become so involved, i.e. on hearing just I remembered to lock the door one would infer that the speaker did lock it. The subject would only not become involved if something outside his control intervened, as in the parenthesis added on to (30b). (Dixon 1991: 222)

This formulation of Dixon’s is similar to the one adopted in section 2.3 for Forward-looking constructions, and which was based on Langacker’s notions of potential and projected reality in his dynamic evolutionary model (Langacker 1991: 278). In a Forward-looking construction, such as (96), there is no doubt that the subject will realise the complement situation if its realisation is up to him or her. And in a General construction, such as the one in (95), the subject will realise the complement situation on all suitable occasions. Thus, while I agree with the assertion of grammarians such as Quirk et al. and Dixon that the to infinitive form encodes potentiality, I think that it may be possible to be more precise as to the exact nature of this potentiality. We will return to this question in section 4.3. Dixon’s analysis of the to infinitive in terms of potentiality is similar to that of Bolinger and Kempson & Quirk. He differs, however, from them in his characterisation of the -ing form as encoding “an activity or state as extended in time” (1991: 218). How does this characterisation fit the four types of -ing construction described in section 2.3, illustrated here in (97) – (100)? (97) John adored being with his grandchildren. (B34 771) (Same-time Attitude) (98) Some described seeing a couple having tea in a restaurant with an elderly

parent. (BLW 715) (Backward-looking Communication) (99) In about 1854 he commenced making Portland cement as well as lime.

(H09 940) (Forward-looking Aspect) (100) He imagined opening the drawing-room door. (H8N 626) (Contemplation Mental Process) In (97) and (99) the being and making are certainly profiled as a state and activity lasting a certain period of real time. They may thus be appropriately described as extended. In (98) the real-time process of seeing need not have lasted for very long, but it may still be taken to refer to more than a punctual registration of the activity. As for (100), it may also be taken to refer to an activity that is extended, not in real time, but in the mind of the subject. If one were to alter Dixon’s formulation “an activity or state as extended in time” to “an activity or state extended in some domain, mental or physical”, it would appear to account for all

Non-fulfilment versus fulfilment 55

four uses of the -ing complement form. We will return to this definition in section 4.4.

3.5 Referring versus non-referring

We have already noted Dirven’s (1989) assertion that the to infinitive complement form resembles in some respects a countable noun. Another scholar who compares infinitive and -ing forms to nominals is Conrad, who maintains that “the distinction between referring and non-referring NP’s is decisive for the choice of the gerund and the infinitive as subject and object, the gerund being used as a referring NP and the infinitive as a non-referring NP” (1982: 68). This distinction is almost the opposite of the ones proposed by Wood and Dirven. Conrad devotes more space to the infinitive and -ing form as subject than as complement. Of the -ing form as subject, he writes:

[…] if the surface subject of a sentence consists of a gerund, alone or expanded by other linguistic material such as complements or adverbials, this subject NP can always be interpreted as a referring NP. That is to say, the gerund always refers to one locatable instance, or several locatable instances, of actions, processes, states &c. (Conrad 1982: 92)

Conrad’s attribution of a referring function to the -ing form is compatible with Dixon’s description of it as encoding an activity or state as extended in time. As was pointed out above this definition does not account for the Contemplation constructions, but may be emended to do so. One might even say that the notion of reference follows from the notion of extension. If some situation is profiled as having extension in some domain, then it must necessarily exist in that domain. The opposite is, however, not the case. A situation may be profiled as existing, without necessarily being seen as extended. Indeed, Conrad’s idea of a referring predicate would seem to be equally applicable to the bare infinitive as to the -ing form. Consider, for instance, (101).

(101) She saw Fand drop her spear and sink to the ground as light as a feather,

eyes closed and pale hair streaming. (F99 1987) There is no doubt that the complement in (101) refers to a concrete incident of spear-dropping. The characterisation proposed by Conrad for the -ing form is therefore just as appropriate for the bare infinitive. This does not, of course, mean that he is incorrect in positing a referring function for the -ing form. There is, however, reason to suspect that this characteristic may be epiphenomenal in nature. As for Conrad’s characterisation of the to infinitive as non-referring, he states that this entails that “the infinitive does not refer to individuated, locatable occurrences of the action, process &c. denoted by the infinitive, but the use of the infinitive is compatible with locatable occurrence” (1982: 118). So much for what

Earlier Studies

56

it is not: Conrad’s positive characterisation of the to infinitive is in terms of dispositions. He defines disposition as:

a conditioned probability, which is not necessarily characterized by any actually existing, observable state of affairs or features of a state of affairs, though it may be […] Speaking generally, one is justified in saying that a person (P) has a certain disposition (D), if it is predictable that in a specified kind of situation (S) P will show a specified kind of behaviour (B), if there are circumstances appropriate for this kind of behaviour (C). (Conrad 1982: 83)

Let us see how this characterisation may be applied to the three senses of the to infinitive described in Chapter 2. Recall that General constructions were defined in section 2.3 as constructions in which the speaker profiles some situation as likely to occur on a more or less regular basis. Conrad’s suggestion that they will occur whenever circumstances are appropriate seems to capture the essence of this sort of to infinitive construction. To what extent is this sort of construal appropriate to the other two sorts of to infinitive construction? When we give voice to a judgement about somebody or something, we may be said to be expressing a hypothesis about a disposition in Conrad’s sense. So we may also conclude that his approach is applicable to the use of the form in Judgement constructions. However, when it comes to Forward-looking constructions, Conrad’s proposals run into problems. Let us confine our attention, for the moment, to Forward-looking Aspect constructions. Of these he writes: “When the gerund is used after these verbs it refers to one or more concrete, individual instances of the action denoted by the gerund. The infinitive has the negative characteristic that it does not refer to concrete, individual instances of the action. It is non-referring and it usually mentions dispositions” (Conrad 1982: 146). Consider examples (102) – (105) in the light of this distinction. (102) A little boy came out of the woods opposite and began skiing down the

slope towards the road. (B0U 993) (103) He came out from refreshments, and we were all bored to death, waiting to

go round again, flings his cape round his shoulder, took his helmet off and started dancing down the road, throwing flower petals about. (B24 830)

(104) They began to ascend the staircase, holding Jimmy between them. (G0E 3461) (105) He started to haul the punt up the rollers. (BMR 1617) All four predications in these examples refer equally to concrete individual instances of actions, to borrow Conrad’s words. His proposal does not seem to be able to distinguish between the two forms in this case. Indeed, the thesis that the infinitive does not refer to individuated occurrences leads Conrad to some rather strained interpretations of the ‘begin to infinitive’ construction. For instance, he states of sentences such as To fill in some of the time he found some College stationary and began to write and She began to rummage in an untidy desk that

Referring versus non-referring 57

“it is probably the author’s intention to say something about concrete actions performed in the situations described, but he does not refer to these actions” (1982: 157). Is it really possible to speak about something in this way without referring to it? In fact Conrad, himself, recognises that there are problems applying his characterisations to Forward-looking constructions. He writes: “in practice, the difference between unique and non-unique reference is not so sharp in sentences with future reference. The reason is, of course, the notorious difficulties involved in identifying intensional and non-existing phenomena” (Conrad 1982: 56). To sum up, Conrad’s characterisation of -ing complements as referring predicates is more appropriate for Backward-looking and Same-time constructions than Forward-looking and Contemplation ones. Similarly, his characterisation of to infinitive complements as non-referring works well for General and Judgement constructions, but not for Forward-looking constructions. In the next two sections we will consider some approaches that display the opposite tendency, focusing as they do on the Forward-looking sense of the to infinitive.

3.6 Future versus actual

Wierzbicka (1988) distinguishes between two main types of to infinitive complement, which she refers to as ‘to of wanting’ and ‘to of opinion’. (Dixon 1991 makes a similar distinction between what he calls ‘Modal FOR (TO)’ complements and ‘Judgement TO’ complements). The latter correspond roughly to what are termed Judgement constructions in the present study, the former cover both Forward-looking and General constructions. Wierzbicka writes: “Roughly speaking, TO complement clauses imply ‘wanting’ if (1) the main verb is a verb of wanting, or (2) the main verb is (or can be interpreted as) a verb of intentional action, or (3) there is no main verb (as in Hamlet’s question)” (Wierzbicka 1988: 29), and that “In most types of TO complements which have been discussed here there is also a clear future orientation (‘this will happen’), and there are reasons to think that this feature, too, should perhaps be regarded as part of the semantic invariant of all TO complement constructions” (1988: 165). There are two main drawbacks associated with positing futurity as a central component in the meaning of to infinitive complements. In the first place, a glance at the evidence of Table 2.7 (p. 43) shows that, while some 77% of complement constructions containing the to infinitive are Forward-looking, there are two other types of construction containing this form which do not lend themselves equally naturally to an interpretation in terms of futurity. In the second place, the same table shows that all four types of complement dealt with in this book may encode events situated in the projected future. Some 73% of the total number of such constructions contain the to infinitive form of complement, but a sizable minority do not do so. Therefore, although the majority of to infinitive complements occur in constructions which connote an element of futurity, this semantic component

Earlier Studies

58

appears singularly ill-suited for distinguishing between these and other forms of complement. It should be pointed out that Wierzbicka herself expresses reservations regarding the applicability of the notion of futurity to Judgement constructions containing to infinitive forms. She writes:

It is less clear whether a component referring to the future would also be justified for the other type of TO opinion complement which we have considered:

I know (believe) Mary to be a Mormon.

But for these sentences, too, we have postulated a kind of disclaimer:

I don’t say: people know (believe) this

I say: I know (believe) this

And this disclaimer, if spelled out more precisely, probably contains a reference to the future. (Wierzbicka 1988:166)

In order to defend the ascription of the notion ‘futurity’ to these Judgement to constructions one has, first of all, to accept the existence of the implied ‘disclaimer’. One has, in other words, to accept that a sentence like I believe Mary to be a Mormon is contrastive. Not only that, one has to accept that the contrast resides in the subject rather than the complement predication. That is, the implicit contrast, according to Wierzbicka, takes the form I (and not other people) believe Mary to be a Mormon rather than I believe Mary to be a Mormon (not a Muslim). Now, if a construction encodes an implicit contrast, one would expect this to be made explicit on occasion. A search of the tokens downloaded from the BNC, however, did not throw up a single example of a Judgement construction in which the subject’s opinion is contrasted explicitly with that of other potential ‘judges’. On the other hand there were several examples, including (94), repeated here as (106), and (107), in which the complement situation was explicitly contrasted with an alternative.

(106) He assumed the pigtail to be her own, but it wasn’t. (FSP 569) (107) Maggie woke up imagining it to be morning, only to discover it was still

dark. (AN7 847)

Applying Wierzbicka’s approach to (107), for example, one would have to paraphrase it as Maggie (though not other people) imagined it to be morning, rather than Maggie imagined it to be morning, not night. That the latter interpretation is the correct one is shown by the clause only to discover it was still dark. Thus, even if one accepts that Judgement to infinitive constructions are

Future versus actual 59

contrastive, the contrast does not necessarily refer to the person of the subject. And, if it does not apply to the subject, Wierzbicka’s argument for ascribing an element of futurity to Judgment constructions collapses. Another of Wierzbicka’s contentions about Judgement constructions, to wit that “TO complements are compatible with the element ‘know’, but only in the subjective first person mode of ‘I know’…., by contrast, THAT complements introduce an ‘objective’, impersonal, ‘one can know’ perspective” (Wierzbicka 1988: 165) has been evaluated by Noël (2003) in the light of corpus evidence. He found that only 20% of tokens of ‘know (x) to be’ had a first-person singular subject, while 58% had a third-person subject. The corresponding figures for ‘know that’ were 19% and 53%. These figures disprove Wierzbicka’s assertion that “Sentences such as ?John knows Mary to be a Mormon are much less natural than .... I know Mary to be a Mormon” (Wierzbicka 1988: 51). Noël also shows that Wierzbicka’s contention that the ‘find x to infinitive’ construction implies personal and experiential knowledge and is therefore not compatible with predications such as you’ll find her to be Mexican wilts in the light of corpus evidence (see Noël 2003: 353-354). The value of both Wierzbicka’s and Noël’s analyses, however, is weakened somewhat by their failure to distinguish between two Judgement ‘find S2 to infinitive’ constructions, a Mental Process construction meaning ‘S1 judges S2 to infinitive’ and a Perception construction meaning ‘S1 discovers S2 to infinitive’. The two are exemplified in (108) and (109), respectively. (108) I found them (almost universally) to be gross without openness, and

cunning without refinement. (B0R 168) (109) The exhilaration soon faded upon checking his fuel state, which he found

to be low. (CGL 1645)

Having demonstrated that Wierzbicka’s characterisation of to infinitive constructions, while it may be appropriate to Forward-looking constructions, is not equally compatible with Judgement ones, we can now turn to her views on -ing constructions. Of these she writes:

The ING complement doesn’t require that something should be actually done, but if it refers to an action at all then it requires that this action be seen as occurring at the same time to which the main verb refers. For most verbs, this will mean that the action referred to in the complement will be seen as real (or at least as real as the event referred to by the main clause). But for those mental verbs which involve imagination, the reality of the action will not be implied. On the other hand, it appears that ‘sameness of time reference’ is always required. (Wierzbicka 1988: 72).

In this passage Wierzbicka uses the phrase most verbs. If we turn again to Table 2.7 (p. 43), however, we see that Same-time constructions account for only 41% of -ing construction types. According to the data in Figure 4.3, p. 123, they only account for 50% of -ing tokens.

Earlier Studies

60

Wierzbicka goes on to distinguish between two kinds of -ing complement, what she calls “temporal ones” and “non-temporal ones”. She maintains that “the non-temporal ones refer to facts or possibilities, the temporal ones refer to events, processes and actions, that is to ‘things’ happening at some particular time” (Wierzbicka 1988: 162). In the terms of the present study, Backward-looking constructions would be considered non-temporal, referring to facts, and both Forward-looking and Contemplation constructions would also be considered non-temporal, referring to possibilities. Only Same-time constructions would be temporal. Wierzbicka devotes most of her attention to this sort of construction. She remarks that they are closely related in meaning to progressive finite clauses.

…. just as a participial ING clause, e.g.

She sat talking to him.

refers to something happening at the same time as something else, so does an ING complement clause, such as

She enjoyed talking to him.

This explains why one cannot say, for example:

*She wanted talking to him.

The future orientation of want is incompatible with the simultaneous orientation of ING. (Wierzbicka 1988:162-3)

It is possible to suggest another (more plausible?) explanation for the infelicity of the same-subject ‘want -ing’ construction. One might say that it only makes sense to speak of wanting something, if this something is relatively uncertain of realisation. When used in Forward-looking constructions the -ing form tends to imply that the complement situation is almost certain to be realised in the projected future. In the ‘intend -ing’ construction, for instance, the subject is portrayed as being very confident indeed of his or her ability to bring about the situation encoded by -ing. (See section 6.3.1 for a comparison of ‘intend to infinitive’ and ‘intend -ing’.) If it is up to oneself to decide the outcome, it makes little sense to speak of one’s ‘wanting’ it. Note, however, that one cannot be absolutely sure of one’s ability to command the obedience of others to one’s wishes. It is this lack of absolute certainty that sanctions the employment of want in the ‘want S2 -ing’ construction, as in (110).

(110) On Saturday morning I want them thinking, I don’t want them going into

their shells. (K5A 3415)

Future versus actual 61

To return to Wierzbicka’s comparison of the -ing complement form to the progressive, this seems to be perfectly apt for Same-time constructions. That is, the utterance She enjoyed/liked/hated talking to him refers to an experience of talking rather than the fact of having talked. She was talking to him, which she enjoyed/liked/hated makes a much better paraphrase than She talked to him, which she enjoyed/liked/hated. This does not necessarily hold for Backward-looking, Forward-looking and Contemplation -ing constructions. I remember talking to him may be paraphrased I remember that I was talking or I remember that I talked. (The possible lack of a mid-interval perspective in Backward-looking, Forward-looking and Contemplation constructions is discussed in section 4.4.1.) Summing up Wierzbicka’s characterisations of non-finite complement forms, one may conclude that her interpretation of the to infinitive form fits Forward-looking constructions, but that it suffers when extended to the other forms of to infinitive complement. As Noël puts it:

The fundamental mistake made by semantic extremists of Wierzbicka’s (1988) and Verspoor’s (1990) ilk seems to be that they are overextending a meaning that can be argued to be connected with one kind of patterning of a certain form (e.g. the occurrence of a to-infinitive after verbs like want and order, associated with “futurity” by Wierzbicka (1988) and with “causality” by Verspoor [1990]) and which may well be its original patterning, to all its patternings. (Noël 2003: 369)

Similarly, Wierzbicka’s characterisations of the -ing form is well-suited to Same-time complements, but much less so to the other three forms. Verspoor’s analysis in terms of causality is mentioned by Noël in the above quotation. We will return to her work below. But first we must consider the approach of Duffley, perhaps the most consistent and persistent advocate of an interpretation of the to infinitive in terms of futurity.

3.7 Future versus interior

In his book The English Infinitive (1992), Duffley conducted an in-depth corpus-based (though not quantitative) investigation of bare and to infinitive clauses. Since then he has published on the -ing form (1995), and on both infinitive and -ing as complements of Aspect verbs (1999) and, with Tremblay, on Effort verbs (1994). He has also co-authored, with Joubert, an article on both forms of complement with Mental Process verbs (1999). He has written on the contrast between the two forms with transitive matrix verbs in general (2000) and on the two forms as subject (2003). Duffley (2006) contains revised versions of some of these papers. Throughout these various works, his view of the meaning of the to infinitive form has not altered appreciably. In Duffley’s opinion, it consists of a combination of a form of the preposition to and the infinitive proper. He writes:

Earlier Studies

62

The postulate put forward in this study is that the potential meaning of to before the infinitive is more abstract than that found in the spatial use of the preposition, and can be stated as follows: the possibility of a movement from a point in time conceived as a before-position to another point in time which marks the end-point of the movement and which represents an after-position with respect to the first […]. The potential meaning of to as described above fits in with that of the bare infinitive in the following very simple way: the latter evokes that which defines the end-point of the movement denoted by to. (Duffley 1992: 16-17)

According to Duffley there is thus always some element of futurity (vis-à-vis the matrix verb) involved in the use of the to infinitive. Duffley’s most recent work is couched in cognitive terms but he wrote originally within the framework of a “psychomechanical” theory. Another linguist who shares the psychomechanical approach, Bailey, published a very similar characterisation of the to infinitive complement form in the same year as Duffley: “In the to V construction, to, conserving its original sense of ‘movement towards’, indicates that the event referred to by the verbs is seen as potential or ‘in the offing’. The point of view from which the event is considered is situated before the event” (Bailey 1992: 186). We saw in the previous section that the ascription of an element of futurity to to infinitive complements, while it makes sense in the case of Forward-looking constructions, is more difficult to accommodate to the two other sorts of to infinitive constructions. Nor does it seem very well suited to distinguish to infintive from -ing complements, a substantial minority of which also occur in Forward-looking constructions. The tasks facing an author such as Wierzbicka or Duffley who assigns such importance to futurity are not however limited to demonstrating how this can be seen as a distinguishing factor. More important is how they apply this notion to the two types of to infinitive complement construction which are not Forward-looking. We will therefore look at how Duffley’s approach can be reconciled to both Judgement and General constructions, starting with the former, exemplified by (111) – (113).

(111) Too late Yartek discovers one key to be the fake one. (F9Y 530) (112) Artemis then saw her father standing by his favourite window, in the

company of a tall dark-haired woman, whom Artemis found to be staring back at her. (EEW 809)

(113) We cannot be one hundred per cent sure that, just because we have observed the sun to set each day on many occasions, the sun will set every day. (FBE 301)

(111) – (113) are Judgement Perception constructions: they involve an act of judgement as well as one of perception on the part of the subject. According to Duffley’s analysis, the latter must always precede the former, the element of futurity lying in the act of judgment always succeeding the registration of the sense impressions. He writes: “An examination of the actual meaning expressed

Future versus interior 63

by the to infinitive after verbs of perception in the active voice shows therefore a necessary before/after relationship between the event of the verb of perception (which has shifted to evoke the notion of inference) and the event of the infinitive (which denotes the conclusion reached by means of the inference)” (Duffley 1992: 35). While this sort of explanation may seem plausible in the case of occurrences like (111) and (112), one may question whether it is equally well-suited to account for the complement of observe in (113). In particular, it seems counter-intuitive to divide the action of seeing the sun setting into two components, one of observing the actual phenomenon and another of concluding that what one is observing is an event of the sun setting. If it is difficult to detect the element of futurity in the case of at least some tokens of Judgement Perception constructions, it is even more difficult to do so in the case of the much more numerous Judgement Mental Process constructions. Consider Duffley’s explanation of the ‘know S2 to infinitive’ construction.

The common denominator of this type of use is [….] viewing knowledge of the referent denoted by the direct object of know as a condition for attributing what the infinitive denotes to this referent. Since a condition has a logical priority with respect to what it allows, there is a before/after relation between what know predicates and what the infinitive does, which calls for the use of to. (Duffley 1992: 50)

These remarks may well be true of the ‘know S2 to infinitive’ construction, as exemplified in (114). However, Duffley’s point here would seem to be equally applicable to Backward-looking Mental process constructions, as in (115) – (116).

(114) The does were retained in hessian sacks and I have known some to bite a

hole in the sack and escape. (BNY 1731) (115) I remember Jean coming home really starry-eyed about you. (EFG 396) (116) I recollect one girl putting on dark glasses as a disguise in the hope that

she would be luckier than Oliver Twist. (AMC 217)

Just as the subject in (114) has to be able to identify a number of does in order to attribute the action of ‘biting holes’ to them, the subject in (115) has to be able to identify Jean in order to predicate the action of coming home, and the subject in (116) has to be able to identify the girl in question in order to predicate her assumption of disguise. (115) and (116) are instances of an after/before rather than a before/after relation between the matrix verb and the complement predicate. In other words it does not necessarily follow from the fact that “a condition has a logical priority with respect to what it allows” that there is a before/after relation between the matrix verb and the infinitive. This is not to say that a before/after relation may not be a property of the ‘know S2 to infinitive’ construction, but, unsuited as it is to distinguish between Judgement and other Mental Process constructions, it is unlikely to be a core property of the form.

Earlier Studies

64

So much for Duffley’s conception of futurity as it applies to Judgement constructions: how does it account for General constructions? Surprisingly enough, perhaps, given the problems they have posed to other linguists wrestling with non-finite complementation, Duffley does not seem to have paid much attention to these. General attitude constructions, as in (117), are hardly mentioned by him at all. Duffley (1999) does, however, devote some attention to ‘continue to infinitive’, which is analysed in the present work as a General construction. It is exemplified by (118) – (120).

(117) I’m German, so I love to eat meat, but in moderate amounts. (C9P 962) (118) THE BURMESE government has released a disgraced former intelligence

chief, but is continuing to arrest civilians in a growing ‘terror campaign’, according to diplomats in Rangoon. (A4H 93)

(119) Women from both sides of the Atlantic continued to push back the frontiers of travel throughout the twentieth century. (ASJ 376)

(120) Maybe he would continue to ride pointlessly around the forest of Haling Heart until he died in the saddle, and maybe even then he would still ride pointlessly around. (GW2 783)

Duffley (1999: 318) points out that the ‘continue to infinitive’ construction is much more frequent than its -ing counterpart. His figures of 108 for the former and 16 for the latter in the Brown and LOB corpora combined, correspond well to mine of 437 and 40 for my random selection of 1,000 tokens of continue in the BNC. Duffley remarks that “This predominance of the to plus infinitive construction indicates that the notion of movement towards the realization of an event is highly salient in the idea of continuation” (1999: 318). To this one might object that the realisation of an event is always highly salient in any construction containing the verb continue. Indeed, it is just as salient in the case of the -ing complement form, which is actually used more frequently than the to infinitive with certain types of complement predicates to encode specific once-off situations (see Table 5.8, p. 194). The ‘continue to’ construction, on the other hand, is normally used to encode general validity predications, as in (118) – (120) (see Table 5.5, p. 185). These, by definition, encode the likely realisation of situations on separate occasions. Duffley writes in connection with this point: “Usage does not however support Freed’s (1979: 94) claim that continue to evokes ‘a start-and-stop progression of events (all identical in nature) requiring an interruption between occurrences of the event’ and so carries with it an implication of actual interruption” (Duffley 1999: 319). Freed may have gone too far in viewing this tendency as a defining characteristic of ‘continue to’. Nevertheless, corpus evidence indicates that there are so many tokens of this type that Duffley’s statement about usage is too one-sided. (See section 5.5.5 for a comparison of the two continue constructions.) While Duffley attempts to subsume all uses of the to infinitive form in complements under the futurity umbrella, he distinguishes between two main uses of the -ing form, depending on whether or not the immediate context provides a


point permitting (the speaker and addressee) an internal perspective on the situation encoded. He writes

The potential meaning of the -ing [….] would therefore simply provide for a representation of the interiority of its event. If the latter is referred to a support situated at a particular instant within [the former], this produces the impression of imperfectivity characteristic of the “progressive form” and of the adjectival uses of -ing. […] If however the relation of the event to the support is left virtual, there is no one point singled out within the event as being occupied by the support and therefore no interior division of the event into an accomplished part situated before this point and an unaccomplished part located after it: the latter’s interiority is consequently evoked as a homogenous whole made up of all the positions between the beginning and the end of the event. (Duffley 1995: 9-10)

Duffley’s interpretation of the progressive is in line with the standard view in English grammars which, at least since the time of Leech’s Meaning and the English Verb (1971), has characterised the form in terms of incompleteness and limited duration. Huddleston and Pullum (2002), for instance, include imperfectivity and limited duration among the six features they list for the progressive form, although they describe limited duration as one of two strong implicatures and not part of the core meaning of the form. The other three core features posited by Huddleston and Pullum are ongoingness, duration and dynamism. The other strong implicature is mid-interval. All six of these features are compatible with Duffley’s interpretation of the progressive. The question I would like to raise here, and to which I will return in greater detail in section 4.4, is how many of them are features of the -ing form in other uses, such as the gerund. In particular I would ask whether the -ing form in gerunds exhibits an element of ongoingness or dynamism. Let us consider this question as it relates to the four types of complement construction in which we find the -ing form, starting with Same-time constructions. Examples (121) – (123) all contain Same-time Attitude predications. Are the complement situations portrayed as homogenous wholes, as Duffley would have it, or as dynamic, ongoing situations?

(121) Like you, I enjoy running my own business. (JY9 1963) (122) I wish I had some bread - I like feeding the ducks. (A74 2399) (123) Yes but you can still revolt against something you like doing. (KGN 889)

Common to (121) – (123) is the simultaneous nature of the pleasure afforded by participation in the activities in the complement clause. As Wierzbicka (1988: 60) puts it: “If we enjoy doing something we enjoy it at the very time when we are doing it”. In each case the situation prompting the pleasure felt is contemporaneous with the feeling of that pleasure. In other words the construction does not encode pleasures of recollection or anticipation. In (121)

Earlier Studies

66

the subject is running his or her own business and in (122) the subject feels pleasure whenever her or she feeds ducks. As Dixon writes about ‘liking’ verbs in general: “the LIKING verb may relate to the Experiencer’s feelings about some activity as it unfolds; ING is then the appropriate complement choice” (Dixon 1991: 260). Dixon’s use of the verb unfold implies that the activities in question are not viewed as homogeneous wholes but as ongoing activities. It is difficult, if not impossible, to come up with a test that will distinguish between these two interpretations of -ing. One can only appeal to the reader’s intuitions. Does (121) mean something like I like the fact that I run my own business or I like the experience of running my own business? If we view the former as more likely, we may be inclined to conclude that the complement in (121) is to be construed as a homogeneous whole. If, on the other hand, we think the second is the more likely interpretation, we may be inclined to conclude that the -ing form has more in common with the progressive than Duffley, for one, admits. Two forms of Same-time construction that are often contrasted are -ing and bare infinitive Perception constructions. The former are exemplified by (124) – (126).

(124) Third, as we monitored our representative sample of schools over a three

year period, we could perceive some of them undergoing a process of structural transition, shifting away from the simple two-tier model towards departmentalism and/or a matrix structure. (G1F 872)

(125) She could hear Penry moving about upstairs as he made up the other bed. (JYC 1462)

(126) Karen watched the steward opening the bottle at the sideboard. (ECK 1581)

Again one must appeal to the reader’s intuitions with respect to the nature of the profiling of the complement situation in these examples. Is it at all realistic to perceive the various stages of the process of structural transition in (124) as constituting a homogeneous whole? And do we not apprehend the movements of the bed-maker in (125) as involving various different locations? To put it simply, do we not picture him moving from one side of the bed to the other? As I see it, both of these examples encode activities profiled as consisting of various differentiated constituent parts that succeed one another during the period of time indicated by the matrix verb. These differentiated parts make up a whole which is imperfective in the sense that the person engaged in making the bed, for instance, may be said to be doing so whether they are tucking in the sheets or turning back the counterpane. In (126) the matrix verb ‘watch’ itself denotes a durative situation, so it is natural to interpret the act of opening the bottle as consisting of differentiated actions, observed one after the other. It is, moreover, instructive in this connection to refer to a claim of Quirk et al., who state of their example:

“Tim watched Bill mend/mending the lamp.


The bare infinitive, having nonprogressive meaning, implies that Bill did the whole job while Tim was watching; the -ing clause, with progressive meaning, has no such implication.” (Quirk et al. 1985: 1206)

Quirk et al. would appear to be quite correct in ascribing progressive meaning to the -ing clause in this sort of example. Bill’s repairing the lamp is an activity of a certain duration, dynamic in character and not completed. In other words Bill is in the middle of mending the lamp when Jim is watching him. It would appear that the -ing complement shares most of the characteristics of the progressive in at least some Same-time constructions. Let us see whether this is also the case for Backward-looking -ing constructions, as in (40), repeated here as (127), and (128).

(127) I recall her pacing the sitting-room while I am doing my homework,


(128) Elizabeth toiled for hours on end and her children remember her machine going non-stop. (B34 404)

Both (127) and (128) encode a complement situation that has occurred before the time of the matrix verb. In both cases this situation has been experienced by the subject. There seems to be no doubt that in each case we are presented with a situation which the subject recalls to his or her mind’s eye. The question is whether this situation is recalled as a homogeneous whole or as unfolding in some domain. The occurrence of adverbials such as while I am doing my homework in (127) and non-stop in (128) indicate that the situations in question are reviewed by the subjects as unfolding over a certain time interval. Moreover, the reference to the pauses in the activity in (127) indicates that it is not to be interpreted as unbroken or homogeneous. I conclude that at least some Backward-looking -ing constructions share some of the characteristics of the progressive construction. (129) – (130) are both Forward-looking -ing constructions.

(129) Our players are beginning to dread playing at home. (CEP 3629) (130) I started losing my hair at twenty nine. (HVE 157)

Does (129) refer to the fact that the team will have to play every second match at home, or to the whole drawn out 90 minutes long experience, with all its various bits and pieces? And is the activity of hair-loss in (130) an ongoing activity, encompassing the loss of a strand here, a lock there, or is it to be viewed as a homogenous whole? In cases like these, which will be discussed in greater detail in section 4.4.1, I think a case can be made for a progressive-like reading. There remains one type of -ing complement clause, the one we have labelled ‘Contemplation’. It is exemplified by (131) – (133).

Earlier Studies

68

(131) It is pleasant to contemplate the Fabii learning Greek, while the Greeks were admiring Roman fides. (H0K 227)

(132) When two hours had passed and he hadn't returned, Lyn imagined him finding his way back to Bale's and by this act leading her there in search of him. (FU2 1059)

(133) Each night as he lay in bed he would imagine himself catching the train from the local station - such reveries usually assumed the initial achievement of getting outside the camp - or stalking the frontier guards among the mountains with Switzerland a few hundred yards away across the snow ... (B0U 1693)

These constructions encode the present contemplation by the subject of the unfolding of some situation before the mind’s eye. This situation may actually have been realised in the past, as in (131), or possibly occur in the future, as in (132), but it is not situated as such within the confines of the dynamic evolutionary model (see p. 33). Note that the complement situation in (133) is impossible for the subject to realise. What concerns us here is the internal structure of the complement situations, whether or not they are to be interpreted as homogeneous wholes. Given that the subjects in (131) may actually have carried out the activity in question, one may ask which of the following sentences makes the best paraphrase:

(134) It is pleasant to contemplate the fact that the Fabii learned Greek, while

the Greeks were admiring Roman fides. (135) It is pleasant to contemplate the fact that the Fabii were learning Greek, while the Greeks were admiring Roman fides. I would suggest that of these two paraphrases (134) portrays the activity of learning as a homogeneous whole, but that it is (135) that best captures the sense of (131). In other words the -ing complement form in (131) carries with it an element of progressive meaning. The same point may be made about (133). The subject plays out the various steps of escaping in his mind. Having established that there are at least some cases in which the -ing complement form partakes more of the nature of the progressive than is asserted by Duffley, we may ask why he should have made the claims that he does in this respect. In Duffley (1999) he explicitly criticises some scholars who have pointed to similarities between the progressive and the -ing form in the following manner:

The -ing form is often treated as if its semantic content in all of its uses was the same as that found in the progressive construction (cf. Wierzbicka 1988: 69, Langacker 1991: 445). Palmer (1974: 171) and Freed (1979: 72-73) have been quoted above as applying this analysis to aspectual verbs, calling the -ing an imperfectivizing operator. This definition of the -ing’s semantic value does not cover gerundive complement uses such as (20)


below, however, where the complement is already realized at the point in time corresponding to the matrix verb’s event:

(20) I remember filing that article somewhere in the top drawer of my filing cabinet.

Here the -ing’s event is necessarily evoked as a whole, i.e. perfectively and not imperfectively [...] In a use such as (20) [...] the interiority of the -ing’s event is simply taken as a whole, as the totality of positions contained between its beginning and end-points. (Duffley 1999: 303-304)

There is no doubt that the action of filing in Duffley's example is encoded as having been completed. Nor is there any doubt that the predicate ‘to file a particular article’ is inherently perfective, being an accomplishment in Vendler's (1967) terms. The question is whether adding the -ing form to the verb introduces an element of duration which serves to render the predication imperfective. To put it simply, does filing that article encode a process that lasts through time and, if it does so, may a subject engaged in carrying it out be said, at any given point, to be engaged in doing so? In concrete terms, when opening the filing cabinet, may he or she be described as filing that article, when rummaging among other files, may he or she be described as filing that article, etc.? There are, at least, two ways of interpreting Duffley’s example. Let us imagine two contexts for it:

(136) I remember filing that article somewhere in the top drawer of my filing

cabinet. It must be still there. (137) I remember filing that article somewhere in the top drawer of my filing

cabinet. The drawer was absolutely full, and I had to squeeze it in at the back.

In (136) it is most natural to view the action of filing as an undifferentiated whole, in (137) as an action consisting of various parts. The question here is whether, in the concrete (imaginary) speech situation, the speaker and addressee factor out the dynamic ‘progressive’ element in utterances like (136) or factor it in utterances like (137) and the genuine utterances (127) and (128). Or is the ‘remember -ing’ construction polysemous with two main senses, one holistic, the other ongoing? This sort of question will be discussed in detail in the next chapter. Suffice it to say at this point that Duffley’s characterisation of the -ing form as profiling a situation as a homogeneous whole is not equally convincing for all tokens of the form. This presentation of Duffley’s work on non-finite complementation does not to do justice to the breadth and depth of his work on either the infinitive or the -ing form. The focus, necessarily, has been on aspects of his approach which I find problematic, not on the many points with which I find myself in agreement. We turn next to the work of Verspoor, another prolific writer on non-finite complementation.

Earlier Studies 70

3.8 Less immediate versus more immediate

Whereas Duffley’s work has moved from a psychomechanical approach in Duffley (1992) to a more cognitive one in Duffley (2006), Verspoor’s has developed from a position where, influenced by Searle, she emphasised what she saw as logical relationships between a matrix verb and its complement (see Beukema & Verspoor 1991:153), to a more cognitive approach in Verspoor (1997), (1998) and (2000). However, in adopting a more explicit cognitive stance, Verspoor does not discard her interest in intentionality and causal relationships. Rather she develops her viewpoint to include notions of subjective and objective construal (Verspoor 1997: 419-20) and “general conceptual links [that] are experientially motivated” (Verspoor 2000: 207). This development in her thinking may be aptly illustrated by the following two quotations:

Verspoor [1990] argues that an immediate direct causal relationship is expressed by -ing complements and non-immediate direct causal relationships by to-infinitive complements. (Beukema & Verspoor 1991: 254)

My claim is that the English language, in the case of non-grounded events or states of affairs, codes […] more subtle differences in construal. When an event or state of affairs is accessed and construed as experienced directly (as denoted by the main verb), it is linguistically realized as a predicate adjunct, plain infinitive, or -ing complement clause. When the event or state of affairs is accessed or construed as experienced indirectly, it is linguistically realised as a to infinitive complement clause. (Verspoor 2000: 208)

Verspoor follows Goldberg (1995) in adopting a construction approach, in which both the matrix verb and the complement type contribute to the meaning of the construction as a whole. She underlines, moreover, that it is not primarily the semantic category of the matrix verb that determines the form of complement. As she puts it: “What is pertinent is the relationship between the matrix subject and what is stated in the complement. Of course, the lexical content of a matrix verb (semantic category) has something to do with establishing this relationship. But by using different complement structures, we can express differences, mainly to do with direct involvement, in this relationship” (Verspoor 1998: 509-510). For evidence that Verspoor is correct in emphasising this relationship rather than the semantics of the matrix verb per se, compare the data in Tables 2.4 and 2.6 in the previous chapter (pp. 31 and 42). In the following we will consider the nature of the relationships between the matrix verb and the complement which are postulated by Verspoor for to infinitive and -ing constructions. We start with her characterisation of the meaning of the former.

We may conclude that the to infinitive always expresses a moving towards an action, but not being there yet, just projected towards it […]. A to

Less immediate versus more immediate 71

infinitive expresses some sort of directness. With an epistemic verb it expresses that an opinion is directly based on some personal or experiential knowledge and the to expresses that X is moving towards a categorial state Y. (Verspoor 1998: 524-5)

This characterisation of the to infinitive form would seem well-suited to its use in Forward-looking constructions, with the proviso that in at least some constructions, such as ‘manage to infinitive’, the goal is profiled as being actually reached. As for General constructions, Verspoor explicitly excludes General Attitude constructions from consideration in her 1998 article on the grounds that “verbs of emotion (e.g. enjoy, regret, etc.) also usually take -ing complements” (1998: 509). If we distinguish between Forward-looking and other Attitude verbs, as Verspoor does, we will find that the most common matrix verbs occur in constructions with both to infinitive and -ing complements, with the General to infinitive constructions outnumbering the Same-time -ing constructions in most cases (see Tables 3 and 9 in Appendix 2 for relevant data). Surely these are the sorts of constructions one should be focusing on if one wishes to distinguish between the contributions made by the matrix verb and the complement form to the meaning of the complement construction? Be that as it may, the only discussion of General Attitude constructions I have come across in Verspoor’s writings is a brief reference in Verspoor 2000 to the difference between He loves to swim and He loves swimming. Of these she writes: “The to infinitive occurs after verbs that have in their meaning next to ‘feeling’ an aspect of ‘want’ towards a potential event. The -ing construction implies that the subject has directly experienced such an event and that the subject attributed (and still attributes) to the event a particular value” (2000: 221). It would appear from this characterisation that Verspoor is of the opinion that matrix verbs that occur in both General and Same-time Attitude constructions have a semantic component of ‘wanting’ that is factored out in conjunction with the -ing form of complement. This may be true in the case of Same-time constructions. It is, in any case, difficult to prove or disprove. But it is certainly not the case that all matrix verbs that connote wanting are incompatible with -ing complements, as may be seen in the Forward-looking examples (110), repeated here as (138), and (139).

(138) On Saturday morning I want them thinking, I don’t want them going into

their shells. (K5A 3415) (139) He didn’t fancy asking the street prostitutes for directions out of

Whitechapel. (HTY 8)

Verspoor’s characterisation of to infinitive complements does not seem entirely felicitous in the case of General constructions. She devotes a lot more attention to Judgement constructions. Her approach to these, however, has been criticised by Noël (2003) for lack of attention to usage data. Having surveyed some of her statements about these constructions, and shown them to be incompatible with the data in the BNC, he concludes:

Earlier Studies

72

It appears, therefore, that Verspoor’s (1990) semantic explications of complement selection are built on an incomplete picture of the distribution of these complements […] A characterization of the meaning of complement types in terms of different kinds of causation seems therefore to have no empirical foundation; if to-infinitives can express both immediate and non-immediate causation, and that-clauses both immediate causation and the absence of causation, then this casts serious doubt on the relevance of such a parameter for complement selection. (Noël 2003: 363)

There is little point in repeating here all the corpus evidence cited by Noël in his criticism of Verspoor. Instead I will point to another instance where such evidence might have led her to another interpretation of the constructions she is examining. This is the ‘pretend to infinitive’ construction, which she exemplifies by:

(140) John, walking around as if his feet were touching hot sand, pretended to be

on the beach. (Verspoor 1998: 511)

She remarks of this construction: “Especially the verb pretend, as it is an action verb, makes this causal relation [between the action and the desired outcome] clear. The to infinitive expresses that the complement subject is ‘acting out’ the action expressed in the complement […] We may conclude that the subject, figuratively speaking, moves towards the state of being expressed, is not there yet, but is projected towards it” (Verspoor 1998: 511). Although pretend may be an “action verb” – in the present study it is categorised as a Judgement Applied Attitude verb – being on the beach is not an action but a state. And is it correct to say, even figuratively speaking, that the subject moves towards the state of being expressed? Does not the ‘pretend to infinitive’ construction actually preclude such a movement? Verspoor reaches her conclusions regarding the form on the basis of just one example, an example, moreover, that is dissimilar in one important respect to the vast majority of occurrences of ‘pretend to infinitive’ in the BNC. Over 95% of the 443 tokens of ‘pretend to infinitive’ which I have examined resemble (141) – (143) in that an action or state is simulated with a view to deceiving a third party. Some 3% of the tokens, one of them reproduced here as (144), resemble Verspoor’s in that no deceit is intended, while the remainder mean ‘aspire to’. (141) ‘You looked so beautiful, pretending to be asleep in the passenger seat –’

(JXU 4464) (142) Instead he went over to one of the shelves and pretended to be searching for a book. (JXU 2390) (143) The legend of the fox that traps birds by pretending to be dead has been

depicted for centuries. ((BLX 1404)


(144) They are always pretending to be grown-ups playing soldiers, playing shop. (A7C 579)

In the majority of instances of the construction, the subject feigns an action or state with a view to deceiving the judgement of some onlooker. This aspect of the construction could not be missed by anyone examining usage data, but is overlooked by Verspoor. To sum up this discussion of Verspoor’s approach to the to infinitive complement form, her interpretation of it in terms of lack of immediacy and indirect causality seems more appropriate to Forward-looking than General and Judgement constructions. We now turn to her interpretation of -ing complements, which she characterises as follows:

I will argue that an -ing complement denotes a bounded event perceived or conceived from such a close range that the boundaries of the event are not within the subject’s perceptual scope. Such a close range vantage point is inherent in direct intentional action, direct perception, and directly experienced emotions. Each of these states and processes form lower level sub-schemas [….]. The broader generalization that can be extracted from these is that -ing prototypically symbolizes that there is a direct causal relationship between the event denoted by the complement structure and P1. (Verspoor 1997: 436)

How well does this characterisation fit the various senses of the -ing constructions that we have distinguished (by “P1” Verspoor means the matrix verb subject)? Situations involving direct perception are encoded in Same-time constructions, as are at least some directly experienced emotions. (145) – (146) exemplify these two types of construction.

(145) He observed three youths loitering nearby, who disappeared from view as

he arrived. (GXJ 2745) (146) ‘I’m not getting stuffy,’ retorted Edward, ‘but I do dislike outsiders

criticising something they don’t understand. (ASE 1231) In so far as a visual stimulus may be said to cause an experience of seeing, and the action of criticism a feeling of dislike, the second part of Verspoor's characterisation would appear to hold. As for the boundaries of the complement situations in these two examples, the left-hand boundary of the activity of loitering is not perceived by the subject in (145), nor is the right-hand boundary necessarily in view at the time of observation, although it is referred to in the relative clause. In (146) the subject does not necessarily have a full overview of the critical viewpoints voiced by outsiders. Verspoor’s characterisation of the -ing form would seem thus far to be well-adapted to at least some Same-time constructions.

Earlier Studies

74

Examples (145) and (146) contain complement situations in the form of activities. According to Verspoor, the -ing form cannot be used for states. She writes: “It cannot denote a state because the main function of -ing is to ‘homogenise’ a process that inherently involves a change. A process that involves change or occurs in bounded episodes is an event. Therefore, the -ing always denotes events” (Verspoor 1997: 449). Consider the Same-time complements in (147) – (149) in the light of this statement.

(147) Apart from being sick and having to keep going to the loo I loved being

pregnant. (FU1 1941) (148) Going to continue being cold you sort some gloves out. (KBJ 534) (149) I smoked for twenty seven years, and erm I’m getting on to a year now

since I stopped smoking and I feel a lot calmer, I hated being a victim to cigarettes! (FLM 104)

The complement situations in (147) – (149) are undoubtedly states. This does not exclude the possibility of their being bounded. The state of pregnancy is inherently bounded, as are periods of cold weather, and the state of nicotine addiction may also be subject to an end. Note that the fact that these states are bounded does not invalidate Verspoor’s primary contention, which is that the boundaries of the complement situation (which situation may take the form of a state as well as an event) are not within the subject’s perceptual scope. In (147) it was during her pregnancy, not at either extreme, that the subject is profiled as experiencing joy! Verspoor analyses what I termed Backward-looking, Forward-looking and Contemplation constructions in Chapter 2 as all involving an element of simultaneity, i.e. as variants of Same-time constructions. Like Wierzbicka she emphasises the fact that an act of remembrance involves a simultaneous mental action replay of the act remembered rather than the fact that the remembered event or state is anterior to the time of the matrix verb. She writes:

[The] notion of direct experience related to purely mental events such as remember or imagine may sound unlikely. However, an event does not have to take place in the real world to be perceived. Our folk knowledge tells us that mental events, either in specific real-life detail or abstract schematic forms thereof, do occur in the mind. Past events can be recalled and pictured, hypothetical events can be construed from past events (cf. Edelman 1991) and imagined, and possible real-life future events can be imagined and evaluated in the mind. In other words, it is not unlikely that in our folk model remembrance or imagination is seen as a mental state or process causing a mental event to occur in my mind, simultaneously to the act of remembrance or imagination. (Verspoor 1997: 446)

This argumentation seems intuitively appealing, at least with respect to Backward-looking and Contemplation Mental Process constructions. However, in


neither of these domains does the -ing complement form compete with a non-finite alternative. We have to turn to Forward-looking complements in order to test the usefulness of these arguments to distinguish -ing from to infinitives. Can they help us to distinguish between ‘intend to infinitive’ and ‘intend -ing’ for example, as illustrated in (150) and (151)?

(150) The Commanders cannot actually see what Lord Raglan intends to refer

to. (CBR 1524) (151) In the meantime, the solicitor intended administering the estate on behalf

of the family. (HHC 2205)

Unfortunately, Verspoor denies the very existence of constructions like the one in (151). She writes: “[-ing forms] may not occur with verbs that express that some form of prior intention is involved (refuse)” (1997: 436). Both constructions in (150) and (151) are Forward-looking Mental Process constructions. And the difference in meaning between them may be related to Verspoor’s primary distinction between less immediate and more immediate apprehension, less direct and more direct causation. As we shall see in Chapter 6, the ‘intend -ing’ construction may only be used when it is within the remit of the subject to bring about the situation in the complement clause, when he or she can cause it directly, to speak in Verspoor’s terms. The ‘intend to infinitive’ construction may also be used in such circumstances, but it is the only one of the two that can be used when it is clear from the context that the realisation of the complement situation is precluded, as in (150). We have now seen some areas in which Verspoor’s characterisations of both to infinitive and -ing complements have proved compatible with usage data, as well as various areas in which her theories are difficult to reconcile with corpus evidence. The existence of the latter should not, however, blind us to the insights in her various studies. In particular the emphasis she places on the ability of the speaker or subject to construe a situation in different ways is a valuable counterweight to more truth-conditional approaches. In her emphasis on construal, Verspoor draws on the approach of Langacker, to whose own writings we now turn.

3.9 Holistic construal versus immediate scope

Langacker has touched on the problems posed by complement clauses at various times. The following extract is from Langacker (1990):

[…] the process designated by the verb stem functions as the base for the infinitival or participial predication overall. The semantic value of the derivational morphology (to, -ing, -ed) resides in the effect it has on the process introduced by the stem: each derivational morpheme profiles a schematically characterized atemporal relation, and imposes its atemporal

Earlier Studies

76

profile on the processual base provided by the stem. In brief, the morphemes deriving infinitives and participles have the semantic effect of suspending the sequential scanning of the verb stem, thereby converting the processual predication of the stem into an atemporal relation. Where these morphemes differ is in the additional effect they have on the processual notion that functions as their base. I analyze the infinitival to as having no additional effect whatever: in the first person to leave or Jack wants to leave, the infinitive to leave profiles the same sequence of relational configurations as the verb stem leave, but construes them by means of summary scanning as a single gestalt. (Langacker 1990: 82)

In Langacker (1991), he modifies his view of to somewhat, stating:

At the very least, Wierzbicka has persuaded me that there is more to the meaning of the infinitival to than I have previously claimed […] that it merely suspends sequential characterization valid for all class members, while recognizing that its prototypical value further incorporates some notion of futurity. (Langacker 1991: 446)

Langacker’s position concerning to remains unaltered in Langacker (1999): he writes “to imposes a holistic (atemporal) construal on the envisaged event, and probably also places it in the future with respect to a temporal reference point (Wierzbicka 1988, ch.1), but it is not force-dynamic and does not focus on the evolutionary momentum of reality” (Langacker 1999: 339). The contention that to imposes a holistic (atemporal) construal on the envisaged event is widely accepted by scholars sympathetic to a Cognitive Grammar approach to complementation. The contention that it involves an element of futurity, at least prototypically, is another matter. It is this contention that will be the focus of attention in the following. The idea that futurity is a semantic component of the to infinitive form has already been criticised in sections 3.6 and 3.7 in connection with the discussion of Wierzbicka’s and Duffley’s theories, where it was pointed out that while the majority of to infinitive complement clauses occur in Forward-looking constructions, a substantial minority do not. To this one might add the further point that, with the exception of just three matrix verbs, all of the matrix verbs in these Forward-looking constructions themselves contain a semantic component of futurity. The three verbs in question are remember, forget and endure, the first two of which are also listed as exceptions by Rudanko in his classification (Rudanko 1989: 23). Thus in the case of 217 of 220 Forward-looking constructions containing the to infinitive, the putative connotation of futurity by the complement form is redundant. Further evidence of this redundancy is the fact that some 14 Forward-looking matrix verbs, among them begin, start and fear, take both to infinitive and -ing forms of complement. In these cases it is difficult to interpret tokens containing -ing complements as being in any way less future-oriented than their to infinitive counterparts. Moreover there are also some 14 future-oriented matrix verbs, among them recommence, postpone and fancy,

Holistic construal versus immediate scope 77

which do not seem ever to occur with to infinitive complements in Forward-looking constructions. These matrix verbs are just as future-oriented as those that only occur with to infinitive complements. Now the existence of massive redundancy is not, in itself, an argument against attributing futurity to the to infinitive form. Langacker has, himself, written of redundancy that

[…] it should be apparent that all grammatical constructions involve a certain amount of semantic overlap, and that full overlap – where the meaning of one element is fully subsumed by that of another – is just an expected limiting case. Redundancy must therefore be distinguished from meaninglessness. (Langacker 1991: 187)

However, the problem in the case of the to infinitive is not just that the proposed semantic component of futurity is redundant in almost all Forward-looking constructions, but that it must be factored out in 67 General and Judgement constructions. Take the case of ‘like to infinitive’, a General Attitude construction, exemplified by (152), and ‘would like to infinitive’, a Forward-looking Attitude construction, exemplified by (153).

(152) We don't want the problems on our own doorstep cos when the chips are

down, we like to go home, and we like to feel safe and if somebody's entered your premises as an intruder then all of a sudden your security been taken away. (KNF 320)

(153) After my exam, I would like to go for a course at the Outward Bound school where I can learn many new things. (HRV 347)

(152) clearly instantiates a General construction in the sense in which the term is used in this study, in that the action of ‘going home’ is construed as likely to occur at irregular intervals in time, to be specific whenever the chips are down. (153) instantiates a Forward-looking construction, in that the action of ‘going for a course’ is construed as something that is located by the subject in the realm of the projected future. Do we wish to posit a meaning component for to go in the one that is absent in the other? More generally, do we wish to posit as a prototypical characteristic of a form a feature that a) is often omitted, and b) is redundant when present? I would think that at the very least the burden of proof ought to rest with those who would answer ‘yes’ to these questions. In other words, while accepting Langacker’s view of the to infinitive form as an element that imposes its atemporal profile on the processual base provided by the stem, I would be more reluctant to follow him in accepting Wierzbicka’s assertion that it prototypically connotes futurity. Langacker describes the -ing complement form as follows:

[...] the -ing predication has several effects. […] besides the suspension of sequential scanning, it imposes on the base process a restricted immediate scope of predication, confines the profile to the component states within

Earlier Studies

78

this immediate scope, and construes these states as homogeneous. (Langacker 1990: 82-83)

In his discussion of the form in Langacker (1991) he again refers to Wierzbicka (1988), this time to her contention that the -ing form always encodes some kind of temporal overlap. He points out that there are many apparent exceptions to this characterisation of -ing, and concludes “one may still consider it unsettled whether all uses of the complementiser -ing involve temporal overlap. The issue, however, is merely a matter of whether that value is universal to the category or only prototypical” (Langacker 1991: 445). As was pointed out in the discussion of Wierzbicka above, not only are there many apparent exceptions to her interpretation of -ing in terms of temporal overlap, the majority of -ing construction types (and half of the tokens) are exceptions. Again we are faced with the question of whether we should posit as prototypical a characteristic that is only present in one type of construction, in this case Same-time -ing constructions, but not Forward-looking, Backward-looking and Contemplation constructions. Duffley has argued persuasively that the -ing form does not connote temporality in any way (see Duffley 1995: 5). In Langacker’s terms it imposes a restricted immediate scope of predication. When it is combined with a Same-time predicate, just like when it is combined with the be of the progressive, the connotation of temporal overlap is a consequence of the superimposed profiles of the predications. There would seem to be no need to posit this sort of overlap in the case of the other types of -ing complement. While I accept Langacker’s description of the -ing form in terms of restricted immediate scope of predication, I am more reluctant to accept that the constituent parts of the predication are always to be viewed as homogeneous. Arguments against this view were advanced in the discussion in section 3.7 of Duffley's interpretation of the -ing form, and I will not repeat them here. I would just like to draw attention to what Langacker says about punctual verbs and the progressive. He writes: “Under normal circumstances, an ‘internal perspective’ hardly seems possible of the point-like events they designate, and indeed, in the progressive they generally receive a repetitive construal” (Langacker 1991: 209). I would like to suggest that it may be possible to apply this notion of repetitive construal to the -ing form in various complement constructions. We will return to the internal structure of -ing complements in section 4.4. We turn now to the last study to be reviewed in this chapter, that of Smith and Escobedo (2001).

3.10 Conceptual distance versus conceptual overlap

Smith and Escobedo (2001) is an attempt to categorise same-subject to infinitive and -ing complement constructions from a cognitive point of view. The authors also look at two types of to infinitive purpose adverbials. The latter will not be mentioned in the following as they fall outside the scope of the present study. In Smith and Escobedo’s opinion both to infinitive and -ing construction types have

Conceptual distance versus conceptual overlap 79

a variety of related meanings. From these one can abstract more schematic meanings for the complement forms. They write:

Our findings suggest that principles of iconicity play a role in each kind of complement construction, which is semantically distinct and schematic for several distinct subschemas with related but separate senses. Assuming that the meanings of the matrix predicates overlap semantically in some way with the meaning of the morpho-syntactic elements in the complement types they prefer, we show that there are semantic criteria for why some matrix predicates prefer one type of complement or the other, but not both, and why others allow either kind of complement. In particular, the meanings of matrix predicates preferring to-infinitival complements usually evoke some aspect of the path-goal sense associated with prepositional to, whereas the meanings of matrix predicates preferring -ing complements usually evoke some kind of conceptual overlap between the matrix and subordinate processes which is conveyed by the meaning of -ing. (Smith & Escobedo 2001: 550)

We will consider Smith and Escobedo’s categorisation of the two complement types in turn, starting with the to infinitive, which they discuss in relation to the path-goal schema. The complement constructions which are our present focus of interest are divided by them into three groups, which they list as ‘3’, ‘4(1)’ and ‘4(2)’. Their Class 3 includes many of the constructions labelled Forward-looking in my classification, containing matrix verbs such as want, intend, attempt, hope and try. They note of this sense of the to infinitive that it “…lacks a spatial path sense, but occurs with matrix predicates which […] strongly evoke future intentions and volitionality” (Smith and Escobedo 2001: 553). After a short presentation of the constructions they conclude:

As noted by others, infinitival to often evokes the idea of futurity. This notion is derivable from the source-path-goal schema because, in the physical domain, a goal clearly lies at the endpoint of the path and is reached only after the path is traversed from the starting point. Reaching the goal thus occurs in the future relative to the time when movement along the path begins. The future-oriented sense of to thus evokes the conceptual transfer of the source-path-goal schema from the concrete physical domain to the temporal domain. To is partially motivated with the verbs in this set because its future sense overlaps semantically with the forward-looking meanings of the matrix verbs. (Smith and Escobedo 2001: 554)

The key word, to my mind, in this passage is evoked, in the first sentence. I would agree that the path-goal sense of the to infinitive is naturally interpreted as evoking futurity when the form is combined with a forward-looking matrix verb. However, Smith and Escobedo go a step further. They attribute a future-oriented sense to the to infinitive form itself. I have argued above, with respect to the

Earlier Studies

80

views of Wierzbicka, Duffley and Langacker, that this abstraction of a future meaning component from Forward-looking to infinitive constructions is going a step too far. There is no need to rehash these arguments here. I would, however, underline that the notion that the to infinitive form evokes the idea of futurity in Forward-looking constructions seems a perfectly sound one. The constructions in Smith and Escobedo’s Class 4(1) contain such matrix verbs as forget, fail and fear. They suggest that these predicates “evoke an unrealized (or unattempted) goal” (p. 555). In what domain is this goal situated? I would suggest that it is properly situated in what I have termed the ‘projected future’, that is the domain containing states and events as we would expect them to evolve given our present knowledge and past experience. These constructions encode the disappointment of the subject’s or speaker’s expectations with respect to the future, but they may still be regarded as forward-looking, and are classified as such in this book. Smith and Escobedo appeal to Langacker’s notion of subjectivity in their discussion of Forward-looking -ing constructions such as ‘dread -ing’ and ‘postpone -ing’. This notion may be applied equally well to constructions like ‘fear to infinitive’ and ‘fail to infinitive’. If one accepts that the constructions in Smith and Escobedo’s Class 4(1) are subjectively forward-looking, and groups them with the constructions in their Class 3, one ends up with a class that corresponds perfectly with the class of Forward-looking to infinitive constructions as defined in the present work. Smith and Escobedo’s Class 4(2) includes verbs such as appear, claim, pretend and profess, thus corresponding to our class of Judgement constructions. Of these they write:

[…] the to-complements of pretense predicates are not construed as goals at all, since they do not entail or imply any notion of contact (or even lack of contact) between the matrix subject and the states or processes construed as objects of to. Pretense predicates convey information about the relative certainty of a situation, and evoke that their complement processes are not objectively accessed (i.e. contacted mentally or physically) by the matrix subjects. Thus, even if I seem or appear to do something, I do not actually do it. Why are these predicates semantically compatible with to?

We propose that to-complements are motivated with these predicates because the objects of to in this subschema are construed not only holistically, but also as conceptually distant. (Smith & Escobedo 2001: 555)

Smith and Escobedo do not provide any arguments in support of their claim that the complements in question “do not entail or imply any notion of contact (or even lack of contact) between the matrix subject and the states or processes construed as objects of to.” As I interpret them they do imply such contact, but at the same time they imply that the existence of this contact may not be universally


acknowledged. Take the so-called subject-raising predicates ‘seem’ and ‘appear’. Smith and Escobedo are surely mistaken when they write of these that “even if I seem or appear to do something, I do not actually do it”. As examples (154) – (157) show, one may actually, though not necessarily, do it.

(154) A couple of hundred people came, and all of them seemed to stay

overnight. (A2S 166) (155) Personally, however, Diderot seems to have been genial and agreeable.

(AJV 801) (156) I seem to remember they put it in mono and did a limited edition in stereo

which, for those days, was a fairly advanced piece of thinking. (AB5 740) (157) Why does theorizing seem to be so much more prestigious than empirical

research? (B25 54) While there is a clear implication in (154) that not all of the guests stayed the night, there is no implication (155) that Diderot was not a genial character, nor in (156) that the subject does not actually remember. Readers will no doubt have their own views on the implications of (157)! To sum up, ‘seem to infinitive’ and ‘appear to infinitive’ are maximally subjective Judgement constructions. They do not carry the implication that the judgement in question is wrong, merely that it is hedged. Neither ‘seem to infinitive’ nor ‘appear to infinitive’ are included in the present study, but constructions like ‘claim to infinitive’, ‘pretend to infinitive’ and ‘profess to infinitive’ are. These all connote a degree of scepticism on the part of the speaker concerning the complement situation. Smith and Escobedo are of the opinion that these complement situations are not (objectively) contacted mentally by the matrix subjects. They offer no argumentation in support of this view. Might one not equally well claim that the establishment of mental contact is exactly what one does when one lays claim to the truth of a proposition? And may not such a proposition be suitably described as the goal of one’s thinking? I will return to the question of the meaning of the to infinitive form in Judgement constructions in the next chapter. Suffice it to say, at this point, that I do not share Smith and Escobedo’s pessimism with respect to accommodating them to the general path-goal schema. Of the three types of to infinitive complement construction listed in Chapter 2, we have as yet made no mention of General constructions. This is because Smith and Escobedo do not include them in their classification, and this despite the fact that they explicitly contrast to infinitive and -ing complements with verbs such as love, hate and prefer in their discussion of matrix verbs that occur with both types of complement (p. 561). Needless to say, their comparison of these pairs of constructions suffers from their failure to recognise the general nature of these to infinitive constructions. Smith and Escobedo sum up their discussion of to infinitive complements by saying:

Earlier Studies

82

We argue that all senses of infinitival to are meaningful because they evoke that the subordinate process is conceptually distant in some way from the state of affairs designated by the matrix predicates. Conceptual distance may be realized in spatial terms as a real path; more abstractly with future goals evoking purpose, intention, or volitionality; with unrealized goals; in terms of pretense or unreality, where the to-marked processes are construed as non-overlapping with (conceptually distant from) the matrix processes; or finally in terms of holistic construal relative to some kind of judgment or opinion scale. (Smith and Escobedo: 2001:556)

This summary appears to account for both Forward-looking and Judgement to infinitive constructions. General constructions are, as we have seen, omitted from their classification. Their class 4(2) corresponds to our Judgement constructions. As mentioned above, if we conflate their Class 3 and their Class 4(1), we find that the borderlines of the resulting category coincide with those of our Forward-looking category. We turn now to Smith and Escobedo’s classification of -ing constructions. They introduce these as follows:

Many authors have noticed that the -ing suffix functions as an imperfectivizing morpheme or operator that evokes a progressive sense whereby the verbal action is viewed as on-going, in process, or viewed internally to the process […]. By highlighting this internal perspective, -ing complements often involve “temporal overlap with the main-clause process” (Langacker 1992:305). We now show that temporal overlap is one type of a more general conceptual overlap that can be motivated with a number of matrix verbs selecting -ing complements. (Smith & Escobedo 2001: 556)

Smith and Escobedo divide -ing complement constructions into four classes, each of which may be compared to one of the four categories delineated in Chapter 2. Thus their first class, containing matrix verbs such as dislike, enjoy and keep, corresponds to our class of Same-time constructions. Their second class, containing verbs such as complete, forget and regret, corresponds to our Backward-looking -ing constructions. Their third class contains verbs like consider, contemplate and imagine, in other words what we have termed Contemplation constructions, and their fourth class verbs like avoid, dread and postpone, corresponding to our Forward-looking -ing constructions. Not only do Smith and Escobedo operate with the same number of classes as are posited in the present study, but the composition of these classes is also substantially the same. There are a couple of exceptions, including four verbs which, according to the criteria stated by Smith and Escobedo themselves at the outset (in footnote 4, p. 549) should have been excluded from their study. The verbs in question are advocate and recommend in their Class 1, and propose and suggest in their Class 3. All four of these verbs are Forward-looking and, in the case of each, the


understood subject of the complement predicate is not identical to the subject of the matrix verb. The other exception concerns the verbs anticipate and discuss which Smith and Escobedo group with the Contemplation predicates, to use my terminology. I classify both ‘anticipate -ing’ and ‘discuss -ing’ as Forward-looking constructions. In Contemplation constructions the complement situation (the object of contemplation) may be situated in any temporal or non-temporal domain. This is not the case for ‘anticipate -ing’ and ‘discuss -ing’. Corpus evidence shows that they are always used in connection with subsequent events and states. While futurity is a semantic component of the verb anticipate, this is not the case for discuss. Nevertheless, the ‘discuss -ing’ construction always seems to encode a conversation about a possible future course of action. Apart from these two exceptions, Smith and Escobedo’s classification corresponds very well with the one employed in this book. We will now proceed to look at their interpretation of the semantics of the four types of construction. Smith and Escobedo maintain that all -ing complement constructions exhibit some degree of conceptual overlap. In the case of Same-time constructions there is, as they point out, actual temporal overlap, thus amply satisfying their criterion. Their Class 2 comprises our Backward-looking constructions. Of these they state that they “generally evoke prior rather than actual overlap between the matrix and subordinate processes” (p. 557). That the subordinate process is prior to the matrix verb is obvious in these cases. That there is an element of overlap is not equally obvious. Nor do Smith and Escobedo support it with any arguments. As such it is difficult either to confirm or refute. Smith and Escobedo’s third class consists of Contemplation constructions. They maintain that the employment of -ing complements in these is motivated “because they evoke some kind of hypothetical or imagined conceptual overlap between the main and subordinate clauses” (p. 557). It seems obvious that such a conceptual overlap is indeed a property of the matrix verbs in question, at least in the rather vacuous sense that when one thinks about something, one thinks about it while one is thinking about it, but whether this can be said to motivate the choice of -ing as complement is another assertion that is unsupported in Smith and Escobedo’s paper. Smith and Escobedo’s fourth and final class of -ing complements are Forward-looking constructions. Of these they write:

Finally, we come to a set of matrix predicates taking only -ing complements which appears to pose an intractable problem for a semantically-based account of complementation, because their complements do not appear to evoke any kind of overlap whatsoever with the matrix processes (whether temporal, prior, or hypothetical). This set includes such predicates as avoid, delay, dread, escape, postpone, put off, and resist. The sense of the verb miss meaning “almost carry something out” also belongs to this group. Dirven refers to these as verbs of “near reality” which imply that “an activity was more or less due to occur, but was just not realized” (1989: 127). (Smith & Escobedo 2001: 558)

Earlier Studies

84

Implicit in this passage is the contention that Forward-looking -ing complements can only be accounted for semantically in terms of some sort of conceptual overlap. This is a view that Smith and Escobedo share with some, but by no means all, scholars. If, on the other hand, one does not posit conceptual overlap as a defining characteristic of the -ing form, the “problem” of accounting for these constructions is considerably simplified. Indeed Dirven’s characterisation, cited by Smith and Escobedo, seems to me to capture the essence of these Forward-looking -ing constructions, with the exception of ‘dread -ing’ which is deserving of some notice. It is worth pausing a moment to consider the ‘dread -ing’ construction for two reasons. The first is that Smith and Escobedo are incorrect in supposing that the verb dread does not occur with a same-subject to infinitive complement. It does, although the semantic range of such complements is restricted to mental process verbs such as think and imagine. More importantly, ‘dread -ing’ differs from the constructions listed by Smith and Escobedo in that the complement situation is normally not to be interpreted as something “just not realized”, to borrow Dirven’s phrase, but rather as something the realisation of which is well-nigh unavoidable. Consider examples (158) – (159).

(158) Our players are beginning to dread playing at home. (CEP 3629) (159) A lot of American guys dread coming to Europe, but it’s not an excuse.’

(K3H 762)

There are 25 instances of ‘dread -ing’ among my 1,000 downloaded tokens of dread. Some of them describe situations in the complement clause that may or may not happen, but in no case is their happening a matter for the subject to decide. More common are situations which are portrayed as unavoidable. Take (158) for example: given that every second match is normally played at home, the situation dreaded by the players is bound to arise. Similarly, in (159) these American guys are scheduled to come to Europe, whether they like it or not, and they anticipate this relocation with acute displeasure. It seems safe to conclude that the ‘dread -ing’ construction implies that the situation in the complement clause is very likely indeed to come about in the projected future. The very same point may be made in the case of other matrix verbs like delay, defer and postpone, as is evidenced by (160) – (162).

(160) Present value: if you delay paying a bill, you can earn interest on the

money in the meantime. (FT8 2503) (161) Tired by her journey from the north of England and the heat, she had

decided to defer unpacking until the next day, standing her cases temporarily to one side of the curtain that divided her cabin from the small galley. (HHA 31)

(162) He postponed implementing a scheme until more groups, including industry and business, were consulted. (A9N 345)


Each of these examples is typical of the way in which the construction in question in instantiated, and in each of them the complement situation is construed as almost certain to occur, albeit at a later date than first expected. Expected by whom, one might ask? ‘By the speaker’ is the answer – in other words what we are faced with in these constructions is subjectification. Smith and Escobedo are fully aware of this. They write: “selecting a verb such as avoid or resist (as opposed to a pretense verb) allows a subjective construal on the speaker’s part that conceptual overlap between the matrix subject and the complement process is obligated in some way, even if it is not objectively realized” (p. 558). Note again the emphasis that Smith and Escobedo place on conceptual overlap. Such an overlap must undoubtedly exist in the mind of the speaker, but this does not necessarily make it a motivating factor in the choice of the -ing form of complement. Having sorted the various types of to infinitive and -ing clauses into various categories, Smith and Escobedo compare a handful of pairs of constructions containing the same matrix verb. They open this discussion by stating that “Matrix verbs allowing either kind of complement are polysemous” (p. 559). Why should this be so? Consider the pair of constructions ‘fear to infinitive’ and ‘fear -ing’, or the pair ‘intend to infinitive' and ‘intend -ing’. Do we really want to say that fear has two different senses in the first pair, intend two different senses in the second? ‘Fear to infinitive ’ differs slightly in meaning from ‘fear -ing’, and ‘intend to infinitive’ differs slightly in meaning from ‘intend -ing’, but is there any reason why the differences should be attributed to the verbs rather than to the complement forms, or the constructions themselves? In dealing with constructions of any sort, the scholar must perforce abstract from various instantiations. In this situation there is always a danger of abstracting too much, of taking what are actually characteristics of the constructions as gestalts to be characteristics of one or other of their constituents. It would seem that on occasion Smith and Escobedo fall victim to the temptation to over-generalise. They are by no means the only students of infinitive and -ing forms to have done so. In the next and final section of this chapter, we will sum up the contributions of the various scholars whose approaches have been discussed.

3.11 Summary

This chapter contains a brief résumé and discussion of some of the work done by various scholars who have concerned themselves with non-finite complement clauses. Their conclusions are summarised in Table 3.1. Two points about this table should be made at once. The first is that the keywords in it cannot hope to do justice to the intricacy of some of the arguments in the various studies: as a rule these are intended by their authors to explain only the prototypical uses of the complement types. The second is that some of the scholars referred to have been concerned with the general meaning of the to infinitive and the -ing form and not just their meaning when they function as complements.

Earlier Studies

86

Table 3.1: Various opinions on the semantics of to infinitive and -ing clauses

Authors to infinitive -ing Wood (1956) specific general Dirven (1989) a given occurrence mere occurrence as such Kiparsky & Kiparsky (1971) non-factive factive Kempson & Quirk (1971) non-fulfilment fulfilment Quirk et al. (1985) potentiality performance Dixon (1991) potentiality activity extended in time Conrad (1982) non-referring referring Wierzbicka (1988) vague futurity vague simultaneity Duffley (1992 + +) future interior Verspoor (1997 + +) less immediate more immediate Langacker (1990 + +) holistic construal immediate scope Smith & Escobedo (2001) conceptual distance conceptual overlap

Table 3.1 shows at a glance that there is a good deal of disagreement among scholars as to the meaning of both forms. There appear to be four main schools of thought with respect to the meaning of the to infinitive. Conflating and simplifying, we can say that these four approaches define the main meaning of the form in terms of specificity, futurity, potentiality and distance. The assertion that the form’s main function is to encode specificity is compatible with its occurrences in Forward-looking and Judgement constructions, but not in General constructions. On the other hand the notion of futurity is obviously applicable to Forward-looking constructions, but is more difficult to reconcile to Judgement and General predications. The notions of potentiality and distance may be related to one another and, indeed, to the path-goal schema. A goal that has yet to be reached is a potential candidate for realisation. The notion of potentiality is well-suited to both Forward-looking and General constructions, one of which encodes a situation as a likely candidate for realisation in the projected future, the other as a likely candidate for realisation whenever a suitable occasion should arise. It is perhaps less suitable for characterising Judgement constructions, which are more naturally interpreted in terms of possibility than potentiality. Finally, the notion of distance seems to be applicable to all three types of construction. It is more abstract than the other characterisations. One may, however, need to operate at this degree of abstraction in order to posit a schema that will cover all three construction types. If there is considerable disagreement among grammarians as to the sense of the to infinitive, there is even less agreement as to the import of the -ing form. There are at least five schools of thought among the scholars in Table 3.1. Let us label these general, occurring, overlapping, ongoing and immediate. While it may be possible to interpret Same-time constructions as general predications, especially if one by ‘general’ means serial or recurrent (see the discussion of

Summary 87

Wood and Dirven in 3.2 above), it is much more difficult to do so in the case of Backward-looking and Forward-looking constructions. One might perhaps say that generality is merely a prototypical characteristic of the -ing form that is overridden in the Forward-looking -ing construction, but it is highly unlikely that a characteristic that is overridden in the majority of cases is actually prototypical. The notion of occurring (factivity, fulfilment, performance and referring) covers Same-time and Backward-looking constructions. It is less felicitous in the case of Forward-looking constructions, though there is no doubt, as we shall see in section 6.3, that Forward-looking -ing constructions encode a situation as more likely of realisation in the projected future than do Forward-looking to infinitive constructions. The notion of occurring is unsuitable for Contemplation -ing constructions, unless one extends it to mean ‘occurring in the mind’, in which case one encounters problems distinguishing between these and Judgement to infinitive constructions. Smith and Escobedo’s idea of conceptual overlapping is quite similar to Wierzbicka’s idea of vague simultaneity. In both cases the notion can be stretched to account for at least three construction types (the exception being Forward-looking constructions), but in doing so we bleed it of most of its conceptual content. Just as was the case with the notion of occurring, the idea of overlapping must be extended to refer to overlapping in the mind of the speaker or subject. We thus end up with a characterisation which is neither particularly substantive nor easily verifiable. The notions of immediacy and ongoingness may be related to one another. Dixon, Duffley, Langacker and Verspoor all view the -ing form as complement as sharing some central characteristics with the -ing form as it occurs in the progressive, although they differ as to which of these properties they emphasise. Thus Dixon focuses on the extended nature of an imperfective process, Langacker and Verspoor on the fact that it is construed as unbounded (or at least that its boundaries are not in focus), and Duffley on the fact that one can view such a process from a point within its defocused boundaries. All of these approaches account for Same-time and Contemplation -ing constructions. They all, however, encounter some problems in dealing with Backward- and Forward-looking constructions. Section 4.4 contains a detailed comparison of the progressive and complement -ing constructions, in the course of which we will touch on the various common properties posited by Dixon, Duffley, Langacker and Verspoor. To sum up this review of the approaches taken in earlier studies, my general impression is that while all of these approaches account satisfactorily for some types of to infinitive and -ing clauses, few of them do so in all cases. In addition, some scholars display a tendency to extend an analysis that is perfectly tenable on a lower level to a higher level where it proves, on investigation, incapable of accounting for all of the instances it is intended to subsume. In the next chapter I will attempt to sketch schematic meanings for each of the complement forms that will account for their instantiation in all types of construction in which they occur. Needless to say, it would have been very difficult to embark on this endeavour if it had not been possible to build upon the

Earlier Studies

88

contributions of the various grammarians whose views have been presented in this chapter.

Chapter 4

Complement Types and Complementisers

4.1 Introduction

The survey of earlier studies in the previous chapter shows that there is no shortage of opinion as to the meanings of the to infinitive and -ing complement forms. The approach presented in the present chapter has been influenced by the contributions of many of the scholars whose work was surveyed in Chapter 3. In common with most of these, I subscribe to what Givón describes as “the non-autonomy postulate: that language (and grammar) can be neither described nor explained adequately as an autonomous system” (Givón 1995: xv). One must always bear in mind the blindingly obvious but frequently overlooked fact that language is used by people to communicate with one another. It follows that the grammarian must adopt a dual perspective on linguistic issues, an individual, cognitive perspective and a social, functional perspective. We must be aware, in studying utterances, both of the functions they are intended to serve, and of the functions they are actually seen to serve. As Stubbs puts it: “A cognitive, intentionalist theory captures some aspects of language behaviour: language is individual, intentional and creative. But this is only half the picture. Language is also social, partly unintentional and routine, based on social conventions which are not open to introspection” (Stubbs 1996: 48). In addition we also need to keep in mind the fact that social conventions are subject to change and evolution. This is particularly so in the case of non-finite complementation in English. Scholars such as Mair (2002) and Fanego (2004) have pointed out that the -ing form is still advancing in certain contexts at the expense of the to infinitive form. A purely synchronic study, such as the present one, is in danger of missing out on instances of variation and change in progress. Utterances are the product of thinking individuals, and these are normally free to construe a situation in the ‘real world’ in various ways. At the same time this freedom is limited by the need to provide enough information for the addressee to conjure up a similar picture to that intended by the speaker. As Verspoor puts it: “One and the same situation may be described in many different ways, making use of different ‘conventional units’, depending on the ‘construal relation’” (Verspoor 1997: 428). Conrad, writing before the explosion of interest in cognitive linguistics, makes much the same point, when pointing out that two of his examples “may be used about exactly the same situation, but they express two different interpretations of this situation” (Conrad 1982: 129). Differences in the construal of one and the same situation may lead to differences in the choice

Complement types and complementisers

90

of complement form. Equally importantly, the differences in construal that are encoded in different constructions may be totally irrelevant to the speaker and addressee in a concrete utterance context. The speaker always has to make use of one or other of the constructions which the language system places at his or her disposal. Occasionally, speakers may be tempted to bend or break the rules, particularly when enough conversational implicatures are present to ensure that their addressees can nevertheless access the intended meaning, but on the whole speakers must make do with a very restricted set of forms to fulfil the myriad communicative functions they wish to carry out. Consider the following two utterances: (163) This time we intend to continue our protest and will escalate action if

necessary. (KA5 414) (164) Manchester design company Lord & Bowes has developed its new look,

including an editorial column and news roundup, and we intend continuing with our lengthier informative features. (K9H 14)

As will be shown in section 6.3.1, the main difference between the ‘intend to infinitive’ and the ‘intend -ing’ constructions is that only the former can be used in contexts where the realisation of the situation in the complement clause is excluded. It also tends to be used in other contexts where its eventual realisation is in the lap of the gods, so to speak. In contexts where the subject has a degree of control over its realisation, as in (163) and (164), the speaker may choose to employ either of the two constructions. In such cases the semantic distinction between the two forms just does not matter, and one would expect the addressees in the actual contexts to filter out any unwarranted connotations. This means that the corpus investigator cannot expect to be able to abstract all semantic distinctions from all tokens. So, given that the English non-finite system of complementation is still evolving and that this evolution is reflected in synchronic variation; and given the fact that the meaning of the complement forms will not necessarily be equally apparent in all instantiations; can we make any worthwhile generalisations about the meaning of the complement types and complementisers? I think that there is enough regularity in the usage in the BNC to answer this question in the affirmative. Provided we remain aware of the possible shortcomings inherent in viewing things from a purely synchronic perspective, we may establish fairly robust characterisations of each of the complement types. This is the aim of the present chapter. In 4.2 there will be a short presentation of the bare infinitive, short since it is the one form about which there is little dissension in the literature. Section 4.3 deals with the much knottier question of the semantics of the to infinitive and section 4.4 with the equally knotty topic of the -ing form. Section 4.5 deals with the to -ing form and section 4.6 contains a summary of the discussion in the preceding four sections.

The bare infinitive 91

4.2 The bare infinitive

According to the data in Table 2.4 (p. 31), the bare infinitive is used with six of nine types of matrix verb. It is used in same-subject Effort and Applied Attitude constructions, and different-subject Perception, Communication, Enablement and Causation constructions. Apart from Perception constructions, in each case it only occurs with a small handful of matrix verbs. (165) – (170) illustrate the six types. (165) Mr. Bruce Hepburn, the 24-year-old former student who helped found the

Imperial Ventures company which commissioned the survey, said if companies did not improve their image ‘they will go out of business’. (A94 241) (Effort)

(166) But she didn’t dare ask him about Spiderglass and what he was doing there. (FP0 256) (Applied Attitude)

(167) Cecilia Darne, who lived round the corner, said she heard a bell toll once at about eight in the morning. (EDN 355) (Perception)

(168) Judy, an elegant woman in her early 40s with a peaceful demeanour, bids us sit down on the verandah. (CK5 2825) (Communication)

(169) Eventually Phil’s dad had let them borrow his garage. (FNW 1283) (Enablement)

(170) He looked surprised when everyone laughed, then joined in himself, with very hoarse, loud laughter which made everyone start laughing all over again. (G12 2052) (Causation)

Considered from the point of view of the other method of classification in Chapter 2, based on the TAM relationship between the matrix verb and the complement situation, the bare infinitive form only occurs in two of the six types. It occurs in Same-time and Forward-looking constructions, but never in Backward-looking, General, Judgement or Contemplation constructions. One may wonder why it never occurs in these four contexts. The answer appears to be related to its always encoding a situation that is located in temporal proximity to the time of the matrix verb. Either it is actually realised at the time of the matrix verb, as is the case with Same-time constructions such as in (167), or, if it located in time after the matrix verb, it follows hard upon it. As a rule the complement clause subject in different-subject constructions is not profiled as having an independent say in the realisation of the complement situation. This is the case in (170). In the case of the ‘bid infinitive’ construction, the vast majority of tokens in my material (50 of 59 to be precise) encode actions of sitting or rising, as in (168), or of trivial motion, such as entering a room. In these cases the complement situation is one which one would expect the complement subject to realise without further ado. The bare infinitive complement form differs from its to infinitive counterpart precisely in that it points to an almost certain, rather than to a merely possible, or perhaps probable, eventuality. The Same-time construction is limited to Perception matrix verbs and encodes situations perceived as a whole at the same time as they occur. Forward-looking bare


92

infinitive constructions, both different-subject and same-subject, encode situations incorporating force-dynamic relations which ensure the immediate realisation of the complement situation. The notion of immediacy is not compatible with General and Backward-looking situations. The notion of certainty of realisation is not compatible with Judgment and Contemplation situations. Immediacy and certainty of realisation are two things that all instances of bare infinitive complement constructions have in common. Another is the fact that in all cases the event or state in question is profiled as an undifferentiated whole. In (167) for example, the sound of the bell is not to be construed as an ongoing event. Nor is the action of sitting in (168) to be construed as consisting of various component actions, as it were. We can say the bare infinitive is used to profile a process as a whole, making no reference to the unfolding in time of the process in question. As Duffley puts it: “unlike the -ing form, the infinitive always produces a representation of an event seen as a whole” (Duffley 1992: 141). The semantic content of the infinitive will be taken to be identical (in English) to that of the verbal stem. This infinitive/stem is what Langacker describes as a complex atemporal relation which is scanned summarily rather than sequentially. The terms summary and sequential scanning refer to the way in which conceptualisers, be they producers or receivers of a clausally-encoded message, view on the one hand a process as unfolding in time and an atemporal relation as frozen in time. According to Langacker finite verb forms profile a process scanned sequentially, the speaker/writer tracing through conceptual time the unfolding of the process in question. Non-finite verb forms, on the other hand, are scanned summarily, the conceptualiser taking in the whole picture at a glance, so to speak. Figures 4.1(a) and (b), from Langacker (1990: 80) illustrate the difference between the two modes of scanning. 4.1(a) illustrates a finite form like ‘falls’, 4.1(b) a non-finite form like infinitive ‘fall’.

Figure 4.1(a): sequential scanning Figure 4.1(b): summary scanning

The bare infinitive represents, in Langacker’s (1991: 421) terms, an instantiation of a type. Not only is it not grounded, it is not even quantified, i.e. it does not exhibit aspectual features. (See Achard 1998: 48ff. for a description of

The bare infinitive 93

the infinitive in French in these terms.) If we compare a situation encoded by a finite verb phrase to a motion picture, the bare infinitive may be compared to a frozen picture: in the first case we can picture a person running from point a to point b, in the second we see them frozen in mid-flight (or, perhaps more correctly, end-flight), as it were. To conclude, we seem to be justified in assuming that the bare infinitive profiles a situation as complete, bounded and totally lacking in aspectual contours.

4.3 The to infinitive

The to infinitive form of complement occurs with all nine types of matrix verb listed in Table 2.4 (p. 31), and in three of the six types of construction listed in Table 2.6 (p. 42). Of the twenty seven theoretically possible combinations of the two categories, twelve are actually found in the BNC. Seven classes of matrix verb occur in Forward-looking constructions, two classes in General constructions and three in Judgement constructions. One example of each of these is listed in (171) – (182). Forward-looking constructions:

(171) The Arctic was everything I expected it to be, and much more in the way

of high seas than I had bargained for. (FBR 101) (Mental Process) (172) ‘Bob,’ said Tessa, when he had put the phone down again, ‘do you want

me to come to the funeral or not?’ (GI2 2341) (Attitude) (173) I urge you to seriously consider this offer and to ensure that you reply

before the closing date. (AYR 180) (Communication) (174) In the day centre, they help you to get into college, or to get work, things

like that. (FR5 1648) (Enablement) (175) One held a knife to his throat while forcing him to drive to nearby

Hackney. (CBF 6069) (Causation) (176) And that is the moment when you begin to decide what you are going to

do about going grey. (G2V 3406) (Aspect) (177) There didn’t seem to be any point in pushing it, so I made to leave. (HTL 1679) (Applied Attitude) General constructions: (178) He liked to finish it each morning before he fetched the new day’s papers. (AC4 2555) (Attitude) (179) And yet Darlington council continues to receive planning applications

from companies eager to break into the market. (K52 4669) (Aspect)


94

Judgement constructions: (180) Cuvier’s system was open-ended because he considered each species to be

a variation of the underlying type adapted to a particular way of life. (G0H 221) (Mental Process)

(181) These, when I passed them, I saw to be sitting or kneeling beside cradles, rocking them, or singing, or gazing intently. (ABL 734) (Perception)

(182) Upmarket readers generally scorn to read about somebody we’ve never heard of having it off with somebody else we’ve never heard of.

(G2E 2121) (Applied Attitude) The classification and labeling of examples (171) – (182) may not meet with the approval of all grammarians. However, irrespective of the system of classification which they espouse, grammarians are faced with the challenge of postulating a schematic meaning for the to infinitive form that will account for its semantic contribution to the meaning of each of these instantiation types. As we have seen in Chapter 3, there have been many attempts to do this, but while most of these account satisfactorily for some of the constructions exemplified in (171) – (182), none do so for all of them. Moreover, not only should an account of the meaning of the to infinitive form be able to explain its use in these constructions, it should also offer some explanation for its apparent non-occurrence in three classes of constructions, to wit Same-time, Backward-looking and Contemplation constructions. In addition the characterisation should be compatible with the function of the form in other syntactic roles, such as the roles of subject and adverbial. In section 4.3.1 I will attempt to sketch a schematic meaning for the to infinitive form that will account for its use in the three types of construction in which it is found to occur. Section 4.3.2 raises the question of whether this characterisation is compatible with the use of the form in subject position and as an adverbial. In 4.3.3 I will air, and attempt to refute, some possible objections to the characterisation proposed.

4.3.1 The to infinitive in complement constructions

In section 4.2. the bare infinitive was equated with the stem of the verbal form and said to profile a situation viewed as a whole. Scholars appear to agree that the to infinitive also profiles a situation viewed as a whole. Certainly none of the grammarians whose work was reviewed in Chapter 3 maintain that it refers to an unbounded situation. Moreover, although they differ widely as to the exact sense of the form, all of them are in agreement that the to infinitive incorporates some sort of meaning component over and above the wholeness of the encoded situation. It thus seems opportune to ask what exactly is the contribution of to to the to infinitive? Before attempting to answer this question, it ought to be said that not all grammarians agree that it does make such a substantive contribution. Some scholars, perhaps most influentially Chomsky in Syntactic Structures

The to infinitive in complement constructions 95

(1957: 100), have argued that to does not have any independent meaning, that it is merely a sort of ‘marker’ of the following infinitive. There are two main weaknesses to this approach. The first one is formal: we often find the infinitive without to and we occasionally find to before an -ing form. The second is functional: the only context in which the addition of to would serve to distinguish the infinitive from the formally identical simple present is in newspaper headlines where this addition would serve to change an assertion into a prediction. The following two utterances may serve to illustrate the point: Verdict: United to win (AJY 458) and United win 3-2 (A2E 517). Omitting to in the first of these would change the prediction to a report in the historic present. Adding to to the second would have the opposite effect. The influence wielded by copy-editors may be considerable but it hardly extends to prolonging the existence in the language of an otherwise meaningless form! One cannot simply ignore the fact that to has been around for a long time. If it were as devoid of meaning as Chomsky and others of like mind would have it, why is it proving so resilient in a historical perspective? It has only been completely grammaticalised in four contexts, in ‘gonna’, ‘gotta’, ‘hafta’ and ‘wanna’ (see Krug 2000 for details). In all other contexts it stubbornly maintains a formally independent existence. In the light of these arguments against the meaningless hypothesis for to, it would appear safest to concur in the opinion expressed by the majority of the grammarians whose work was reviewed in Chapter 3, that it is best viewed as a meaningful element. There is a broad consensus among scholars that the to of the infinitive is historically derived from the preposition to. Fischer writes: “It is generally acknowledged in the literature that the allative preposition to (or its equivalent in other Germanic languages) developed into an infinitival marker when it became combined with an infinitive” (2003: 451). Given its historical provenance, many scholars have assumed that to is a variant of the preposition and that it instantiates a ‘path towards goal’ schema. As Bailey puts it: “In the to V construction, to, conserving its original sense of ‘movement towards’, indicates that the event referred to by the verbs is seen as potential or ‘in the offing’. The point of view from which the event is considered is situated before the event” (Bailey 1992: 186). Duffley also compares infinitive and prepositional to: “the potential meaning of to before the infinitive is more abstract than that found in the spatial use of the preposition, and can be stated as follows: the possibility of a movement from a point in time conceived as a before-position to another point in time which marks the end-point of the movement and which represents an after-position with respect to the first” (Duffley 1992: 16). Achard makes much the same point about à in French. He says that its meaning “includes both temporal and spatial relations involving one point: either static location of the figure at a point, or a dynamic motion of the figure along the path to a specified point” (Achard 1998: 57). A ‘path towards goal’ schema incorporates two important elements in addition to that of motion. These are direction and distance. All prepositions are used to locate an element in relation to another element. Location in the case of path prepositions involves direction. As for distance, the use of to allows for the


96

possibility that the mover (trajector) has still not reached his or her goal. Indeed there is no guarantee that he or she will ever do so. As Verspoor puts it: “We may conclude that the to infinitive expresses that the subject, figuratively speaking, moves towards the state of being expressed, is not there yet, but is projected towards it” (1998: 511). Note that, whereas to itself may just mean ‘towards/in the direction of’, it can be combined with some matrix verbs, such as manage or remember, that show that the target in question has actually been reached. This contrast may be exemplified by (183) and (184).

(183) But at least, thought Charles, I remembered to pull up my socks. (ASE 2111) (184) I plan to be finished in 18 months or so. (BN1 242) What both these Forward-looking constructions have in common is the fact that the pulling up and the finishing were still only potential future occurrences at the precise moment of remembering and planning. Recall that potentiality is a characteristic of the to infinitive form proposed by Quirk et al. (1985) and Dixon (1991) among others. There is no doubt that this characterisation is compatible with the use of the form in (171) – (177) as well as (183) – (184). One may, however, be tempted to ask whether they have more in common than mere potentiality. The term potential implies the existence of possible alternative situations – if no such alternatives existed, one could speak in terms of certainty rather than potentiality. But the situation encoded by the to infinitive is not just any old alternative: it is the focused or targeted alternative, the alternative with the spotlight on it, so to speak. In many cases it is to be construed as a probable rather than a possible candidate for realisation. Thus, in contexts of expectation or wanting, as in (171) and (172), the alternative future scenarios of being and coming are focused upon in preference to other potential situations, such as not being and not coming. In (174) and (175), the outcomes of getting into college and driving to Hackney are profiled as the very likely consequences of helping and forcing. How may such a view of the to infinitive be reconciled with matrix verbs, such as fail and forget in (185) and (186) which explicitly exclude the realisation of the complement situation? (185) Beryl sends her apologies - she forgot to give me your message. (ANY 1101) (186) That was her all over: the tease that failed to deliver. (H9N 114)

Both (185) and (186) encode somewhat more than the non-realisation of a merely potential situation. They encode the non-realisation of an expected situation, of what, prior to their non-realisation, one might have described as a likely outcome. Given that all Forward-looking to infinitive constructions contain a complement situation encoded as likely of realisation, one may suggest the following characterisation of these complements:


Forward-looking to infinitive complements: a situation, viewed as a whole, is profiled as expected, but not certain, to occur at some remove in the future from the matrix verb.

One may propose a similar definition of General to infinitive complements, such as those in (178) – (179) and (187) – (189). (187) When we are hungry we love to eat bread. (HS7 263) (188) Bernstein and Reynolds prefer rent reviews to be conducted as

arbitrations, and that appears also to be the preference of the property community where the rent is substantial. (J6Y 239)

(189) The two men ceased for a time even to acknowledge one another in the street; and though they later resumed formal courtesies, close friendship was dead. (B0R 458)

As we saw in Chapter 2, constructions such as these encode general validity predications. To be more exact they profile situations as likely to be realised on all suitable occasions. These occasions may be more or less regular of occurrence. Indeed instantiations of General constructions often make concrete reference to the occasions on which one can expect the situation to be realised (or not realised in the case of negative constructions such as ‘loathe to infinitive’ or ‘cease to infinitive’). These references commonly take the form of adverbials, such as when I can in (178), every morning in (179), when we are hungry in (187) and in the street in (189). There is no need to multiply the number of examples. It should be clear by now that the definition below accounts satisfactorily for the meaning of these constructions.

General to infinitive complements: a situation, viewed as a whole, is profiled as likely to occur on a more or less regular basis.

The third type of construction containing to infinitive complements is the Judgement type. It is exemplified by (180) – (182) and (190) – (193). (190) The younger said in Arabic what I assumed to be, ‘Hey, have you seen

this?’ (AT3 1905) (191) A great adventure, a fitting enterprise for one who had known herself from

infancy to be set apart for some rare destiny, and one that she had thought herself to have pursued courageously, successfully, with a redeeming love that had rescued even the anguished, complex, hostile Aaron, and had saved him from his wilder flights. (FB0 1466)

(192) Trent had been correct in guessing their destination to be on the north fork of the Belpan River. (AMU 907)

(193) For a second he supposed them to be Prussian, then recognized the shape of the cloth-covered helmets. (CMP 417)


98

All Judgement constructions encode a conjecture on the part of the subject (or, in the case of some Applied Attitude constructions, the speaker) about the complement situation. This conjecture may be a firm one, as in the case of the ‘know S2 to infinitive’ construction instantiated in (191), or a good deal weaker, as in the case of the ‘guess S2 to infinitive’ construction in (192). Irrespective of the subject's degree of conviction, however, the complement situation is encoded as one of several alternatives. Moreover, it is the alternative which the subject considers most likely to be true. Thus in (190) it is clear that the subject may be mistaken in his assumption regarding the import of the utterance he hears, and in (192) his guesswork may well be faulty. In fact in (192) the subject's assumptions are correct, in (193) incorrect, without this contextual difference in any way affecting the meaning of the two constructions in question. In all Judgement constructions the focus of the subject (or speaker) is on one of several possible situations. The constructions may therefore be defined as follows:

Judgement to infinitive complements: a situation, viewed as a whole, is profiled as likely to be true.

Having characterised to infinitive complements in the three constructions types in which they occur, we must now inquire as to what the three definitions have in common. It is clear that while all three construction types share the common meaning core ‘the most likely of several alternatives’, the combination of matrix verbs and to infinitive complements leads to rather different interpretations of the complement situation in the three instances. Thus the sense of the Judgement construction will often be understood in an utterance, such as (193), to mean ‘the subject thinks this is the most likely complement situation'. The General construction, on the other hand, will often be used in a sense something like ‘on any given suitable occasion, the complement situation is likely to be realised’: (178) is an example of this sort of construction. Finally, the Forward-looking type, the most clearly compositional of the three, is often used in the sense ‘the complement situation is the most likely to be realised of several possible future scenarios’, as in (183). Having said that, there is no doubt that there are several elements of overlap between the three senses. Thus the Judgement sense encodes an assertion that something is true in general. To this extent it resembles the General sense. And the General sense points both backwards and forwards in time, in so far as it encodes something that is likely to occur on suitable occasions in the past, present and future. It thus overlaps to a certain extent with the Forward-looking sense. The main difference between the three senses is related to the domain in which the targeted alternative is predicated. The Forward-looking sense refers to a likely alternative in the projected future, the General sense to the likelihood of a situation occurring at intervals in time, and the Judgement sense to the likelihood of a particular alternative being true in the relevant domain. Abstracting from these three domains, we arrive at the following definition of to infinitive complements in general.


to infinitive complements: a situation, viewed as a whole, is profiled as the more/most likely of two or more alternatives in some specified domain.

This definition accounts for the occurrence of the form in all construction types. Another question is how we account for the fact that the to infinitive does not occur in three of the six categories. The reason why it is not used to encode Same-time or Backward-looking situations would appear to be that it always implies an element of doubt as to the realisation of the complement situation. There is no room for the admission of doubt as to what is happening or what has happened. (There is, of course, room for philosophical doubt, but English, at least, does not allow for encoding such doubts by means of non-finite complement clauses.) Nor is there any reason to imply doubt with respect to Contemplation situations. The question of their possible occurrence is simply not an issue, firmly located, as they are, in the realm of irrealis. The fact that the to infinitive does not occur in any of these three contexts is thus simply a consequence of its always encoding a likely, but not certain, eventuality. The to infinitive encodes a targeted alternative: there is no place for implicit alternatives in Same-time and Backward-looking constructions. In the next section we will examine whether this sense of targeted alternative may be said to be a property of the to infinitive as subject and adverbial.

4.3.2 The to infinitive as subject and adverbial

In the previous section I attempted to narrow down the meaning of the to infinitive form when used as a post-predicator complement. In actual fact, of course, to infinitive clauses occur in many other positions in English utterances. They occur both in clauses and phrases, as complements and as modifiers. There is no space here to do more than scratch the surface in relation to these other uses. It may, nevertheless, be of interest to try to ascertain to what extent the characterisation proposed in the previous section is applicable to the to infinitive in some other contexts. As this book is concerned with the clausal level, the focus in this section is restricted to some other uses of to infinitive clauses on this same level. We will first look at some instances in which it figures as a syntactic subject and then at some instances in which it is used as an (optional) adverbial. In both cases we will consider to what extent the characterisation in the previous section is compatible with these uses. A usage-based study such as the present one requires hard data, so a random sample was downloaded from the BNC of tokens of utterances containing initial ‘To’ followed immediately by the four common verbs live, love, win and write (the first two stative, the third an achievement and the fourth either an activity or an accomplishment). This yielded 40 instances of to infinitive clauses as subject. A random 500 tokens of to infinitive forms of the same four verbs were also downloaded. From these were extracted the occurrences in which the form is used adverbially. This yielded 50 tokens. 40 and 50 tokens may seem like


100

very small data bases on which to base any conclusions, but it should not require many tokens to illustrate the inappropriateness of the characterisation proposed in the previous section, should it prove less than felicitous. To infinitive clauses as subjects are of rare occurrence in all registers according to Biber et al. (1999: 722). Not only are they rare in themselves but they appear to collocate with a very limited range of predicators. More than two-thirds of the downloaded occurrences in my material collocate with a copular predicator, as in (194) and (195). The remainder, with one exception, cited as (200) below, all collocate with Enablement verbs like involve, require, demand or mean, as in (196) and (197).

(194) To write of male artisans is tautologous. (HXC 97) (195) ‘To love human beings is still the only thing worth living for; without this

love you really do not live.’ (B1F 1291) (196) To win a significant amount of new business would require a big cultural

change at the company. (BMB 1669) (197) To live life to the full involves awareness of the physical, emotional,

mental and spiritual self. (AYK 418)

Can the four to infinitive clauses in (194) – (197) be said to refer to ‘the most likely of several alternatives in some domain’? Do they, in other words, imply the existence of other possible alternatives? I think the answer to the second question, at least, is ‘yes’ in all four cases. In (194), for instance, the speaker clearly implies that the tautology involved in applying the epithet male to artisans is avoidable. In (195) to love human beings is presented as one of several theoretically possible goals. In (196) the subject clause also clearly instantiates one of several possible outcomes, although in this case it is hardly correct to state that it is the most likely one. Indeed, I think it would be more correct to say of (196) that the to infinitive clause encodes the highlighted of several possible alternatives, rather than the most likely alternative. It may well be the case that a maximally schematic definition of the to infinitive form should be phrased in terms of the focussed or targeted of several alternatives, and that when employed as a post-predicator complement this element of focussing leads to its being interpreted in terms of likelihood of realisation. 6 Three of the four tokens of to infinitive cited as (194) – (197) occur with a main verb in the present, the fourth a modalised main verb. These are the two

6 Note that Chuquet (1986), writing about the to infinitive as subject, also

maintains that it encodes the most likely, but not certain, of two alternatives. His exact formulation is as follows: “Nous avons déja avancé que To + Base verbale constituait par rapport à la notion le choix de p, c’est-à-dire (p/p’) [….] La représentation (p/p’) pour l’infinitif indique que p est privilégié sans que cela entraîne nécessairement la validation de p dans (p, p’).” (Chuquet 1986: 115)

The to infinitive as subject and adverbial 101

forms of main verb that collocate most frequently with this form of subject. However, four of the 40 tokens occur with a verb in the past tense. Of these four tokens with a predicator in the past, one is an instance of reported speech, in which the speaker in question presumably used the present at the time of speaking. The other three are cited as (198) – (200).

(198) To live in such a castle meant demands on the purse. (A68 1949) (199) To write about the events up there was hard, sometimes a real struggle, but

it had to be. (ECG 1468) (200) To love myself appeared to go against all that old upbringing, all that old

training. (B19 1070)

The question is whether the subjects in these three tokens can be said to encode the targeted of several possible alternatives. All three appear to encode specific rather than general predications. In (198) the understood subjects of live actually did live in the castle in question. In (199) the understood subject actually did write about the events in question. And in (200) the speaker actually did indulge in self-love. As all three situations occurred prior to the zero point of discourse, there seems, at first glance, little reason to employ an alternative-evoking form. In the case of (199) one could perhaps maintain that the employment of the to infinitive in preference to the -ing form is motivated by the wish to emphasise the element of struggle in the writing in question, by hinting that its successful completion was by no means guaranteed at the outset. In (198) the subjects had only recently moved into the castle, so one might perhaps read into the choice of complement form a wish to contrast the expenditure necessary to maintain a castle with that necessary for more normal housing. Finally, the extended context of (200), cited in (201), might prompt one to conclude that the speaker in question has been influenced by the form of complement in the preceding sentence.

(201) It was trying to love myself, the starting point of it all, that seemed to lead

me in so many strange directions. To love myself appeared to go against all that old upbringing, all that old training. (B19 1070)

One may find these arguments for the employment of a to infinitive clause in these cases more or less convincing. The main point is that it is possible, in certain circumstances, to employ the form to refer to situations in the past. Recall that the to infinitive complement never occurs in Backward-looking complement constructions. There would appear to be rather less form-function isomorphism with respect to the use of the form in subject position. We turn now to the to infinitive as adverbial. The 50 tokens of adverbials in the downloaded sample all encoded either the purpose or the result of the situation encoded by the main predicate. The former type is exemplified in (202), the latter in (203). That the distinction between the two is not always clear-cut may be seen in (204), which contrasts two possible interpretations.


102

(202) Patient works out in gym to win back his strength. (CFB 1085) (203) The 29 year-old Jersey professional, a former assistant to Tommy Horten

at Royal Jersey, shot a closing 66 to win by six shots. (G2C 1379) (204) The majority of young people who leave home to live rough are found not

to have chosen this course but, rather, to have had it thrust upon them by circumstances. (CRS 557)

The fact that all of the downloaded to infinitive adverbials are either purpose or result adverbials accords with the observation of Biber et al.’s that “to-clauses […] are almost always used to convey purpose […] The only other notable use of to-clauses is to show results” (1999: 828). The tokens in this rather small data sample were split almost evenly between the two types, but this is probably due to the fact that almost all occurrences of ‘to win’ were of the result type, (202) being, in fact, the only exception. The data did not throw up a single instance of ‘to love’ as an adverbial, presumably because of a degree of incompatibility between the semantic specifications of this verb and the notion of purposefulness. To what extent may purpose or result adverbials be characterised as encoding ‘the most likely (or the targeted) of several possible alternative situations in some domain’? In (202), for example, is to win back his strength one of several theoretically possible reasons that the patient in question might have for working out? The obvious answer is ‘yes’. What about outcome adverbials, such as to win by 6 shots in (203): does this encode one of several theoretically possible outcomes to shooting a closing round of 66? Again the answer is undoubtedly ‘yes’? May these situations then be described as the most likely of several alternatives? I think not. May they, on the other hand, be described as the targeted of several alternatives, or can one be more specific? On the basis of this small sample of occurrences, one might venture the following schematic characterisations of the two adverbial forms.

to infinitive purpose adverbial: a situation, viewed as a whole, is profiled as the most desired, by the main verb subject, of several theoretically possible situations in the projected future. to infinitive result adverbial: a situation, viewed as a whole, is profiled as the one actually realised of several theoretically possible situations in the projected future.

Both of these characterisations are compatible with a more schematic characterisation of the to infinitive form as encoding the targeted of several possible alternative situations. The following definition would, therefore, appear to account for all uses of the to infinitive in this and the preceding section.

to infinitive: a situation, viewed as a whole, is profiled as the targeted of several theoretically possible situations in some specified domain.

The to infinitive as subject and adverbial 103

We may posit a network like the one in Figure 4.2 for all clausal uses of the form. In Figure 4.2 we can see that, when used as a subject, the to infinitive form inherits the specifications of the maximally schematic sense without adding any semantic content to this sense. The other three uses all involve the addition of some semantic content, the nature of this addition varying in accordance with the general functions they are used to encode.

clausal to infinitive

Subject Complement Purpose Result

Figure 4.2: Illustration of network containing the most prominent clausal uses of the to infinitive form, as Subject, Complement, Purpose Adverbial and Adverbial of Result

4.3.3 Weaknesses of the targeted alternative approach

The explanation of the meaning of the to infinitive form in terms of a targeted alternative which was proposed in the preceding section presupposes the existence of other alternatives. Such alternatives may on occasion be explicit, as in (205) and (206), previously cited as (42) and (94), respectively. More often, they are latent or implicit, as in (207) and (208). One problem facing the proponent of a targeted alternative hypothesis is to demonstrate the existence of an element that is latent or implicit.

(205) We expected people to just come and go, but most didn’t - they had a

great get together.’ (GX2 302) (206) He assumed the pigtail to be her own, but it wasn’t. (FSP 569)

targeted of several possible situations

targeted of several possible situations

most likely of several possible situations

most desired of several possible situations

realised of several possible situations


104

(207) The Serbian forces’ westward drive into Bosnia fits into a map disclosed for the first time yesterday, which shows Serbian leaders intend to capture more than 70 per cent of Bosnia, leaving Croatian and Muslim populations isolated in small pockets of land. (AKR 527)

(208) If you have a problem and cannot hostess your group meeting as arranged, PLEASE contact the organiser who will help you, if at all possible, to find another venue nearby to avoid frustration and disappointment for others. (ANM 1645)

The alternatives in (205) and (206) are expressed in the utterances. In (207) and (208) the existence of alternatives is also obvious. In (207) the alternative is capturing less than 70% of Bosnia, in (208) of not finding another venue. The matrix verbs in (205) and (206) only occur with one sort of non-finite complement, the to infinitive form. However, they both also occur in finite complement constructions, as in (209) and (210), both of which also admit of explicit alternatives. (209) We expected that a number - perhaps an embarrassingly large number - of

students would decide that they simply did not like our courses - and we were prepared to smooth the path of transfer. In the event, there were a few who wished to leave, and a few others who decided that other initial choices had been mistaken, and who wished to join. (G0W 1101)

(210) It was a race which highlighted his incredible distance running ability, and the TV and newspaper press naturally assumed that Zarei was an athlete dedicated to winning and record breaking, but nothing could be further from the truth. (AR7 1251)

Given the evidence of (209) and (210), it is obvious that, while we can abstract a latent alternative meaning component from (205) and (206), we cannot attribute this component solely to the to infinitive complement form. This does not mean that the to infinitive form does not encode such a component. It may well be the case that the profiles of the matrix verbs and the complement form overlap, and that the presence of such a component in the semantic make-up of the matrix verb motivates the choice of the to infinitive complement form. Be that as it may, the proponent of the hypothesis that the notion of targeted alternative is part of the semantic profile of the to infinitive form obviously has to seek evidence elsewhere than in these sorts of constructions. The matrix verbs in (207) and (208), intend and help, differ from expect and assume in that they occur with non-finite complements other than the to infinitive. It is my contention that these other forms, -ing in the case of intend and the bare infinitive in the case of help would not be equally felicitous in the contexts of (207) and (208) for the very reason that they do not admit of alternatives. As was mentioned in section 4.1, the ‘intend –ing’ construction is restricted to contexts in which the eventual realisation of the complement situation is profiled as being within the remit of the subject. This is clearly not the

Weaknesses of the targeted alternative approach 105

case in (207). As for the ‘help S2 bare infinitive’ construction, it is never used in contexts which admit of the non-realisation of the complement situation. For this reason, it is the only one of the two used to encode direct face-to-face offers of help, as in (211) and (212).

(211) I’ll help you get started, Mummy’s gotta get on darling I’ve got tons to do.

(KP8 568) (212) Here, let me help you pick up the bedding and we’ll take it downstairs.

(HHC 2549)

In contexts like these the use of to would involve the subject’s hedging his or her offer of help with an intimation that complement situation might not be realisable. This would be somewhat paradoxical; at the very least, such a construction would be marked, compared to the bare infinitive one. As stated above, it seems reasonable to interpret the complement situations in (207) and (208) as encoding targeted alternatives. As such they imply the existence of latent alternatives. Now this implication cannot be a property of the matrix verbs in question as these also occur in non-alternative-evoking constructions. So it must either be a property of the complement form or of the composite construction. And given that the definition of the to infinitive in terms of targeted alternatives is compatible with its employment as subject and adverbial, as was demonstrated in section 4.3.2, there would seem to be no reason to ascribe this meaning component to the construction's gestalt rather than the complement form. Simon-Vandenbergen (2003) takes issue with the hypothesis that the to infinitive form evokes latent alternatives. She maintains that this hypothesis is difficult to prove, that it fails to account for the use of alternative complement forms with one and the same matrix verb, and that it necessitates ad hoc explanations of the nature of the alternatives implied by the various constructions incorporating the form. There is no doubt that a hypothesis that makes reference to implicit rather than explicit elements is not easy to disprove. However, the hypothesis could be proven false by identifying contexts containing the to infinitive form which excluded alternatives. It is a corollary of the targeted alternative hypothesis that to infinitive complement constructions cannot be used to encode situations that do not admit of any such alternative. In other words if the form were to be used in Same-time or Backward-looking situations (i.e. situations the existence of which can be posited with absolute certainty) the hypothesis would be considerably weakened. At first sight ‘continue to infinitive’ and ‘cease to infinitive’ might appear to be such exceptional constructions. They certainly seem very much like a Same-time and Backward-looking construction, respectively. However, both constructions can also be described as General constructions, as in Freed (1979: 152). Both constructions encode situations that occur at more or less regular intervals. In neither of them does a non-stative situation have to be in progress at the time of the matrix verb. In other words one can continue, or cease, to do something that one is not doing right now. When


106

viewed in this light both ‘continue to’ and ‘cease to’ may be viewed as instantiations of the General to infinitive schema, which is in turn an instantiation of the more abstract to infinitive schema. Apart from these two constructions there are none containing to infinitive complements for which one can plausibly argue a Same-time or Backward-looking interpretation. The second point raised by Simon-Vandenbergen is related to the contribution to the meaning of constructions made by the matrix verb, on the one hand, and the complement form on the other. There is no doubt that there may be various degrees of overlap between the semantic profiles of the matrix verb and the complement form. In some cases the overlapping can be so extensive as to entail a considerable degree of redundancy. This is the case with the matrix verb keep and the -ing complement form, both of which profile the complement situation as extended in time. Similarly, the matrix verb choose and the to infinitive complement form both profile the complement situation as one of several possible alternatives. The targeted alternative profile of the to infinitive form is most prominent when there is no element of overlapping between it and the semantic profile of the matrix verb. This is the case with Attitude matrix verbs, such as like. In this connection Simon-Vandenbergen asks whether a sentence like I like swimming may not also be said to evoke alternatives like I hate swimming or I like skating. To this one can only answer that there is no evidence that they do so. These constructions are normally used to encode the expression of enjoyment pure and simple in an activity engaged in by the subject at the time of his or her engaging in it. (213) – (215) are typical in this respect.

(213) I really like doing things together. ( A74 318 ) (214) I’m with you, for a start, and I happen to like being with you. (JYA 2489) (215) A letter from your Aunt Emily told us how much she likes having you and

how accomplished you are becoming. (H8X 1203) There may, of course, be contexts in which it would be natural to interpret utterances containing -ing complements as evoking alternatives, but there seems to be no reason to assert that such reference is an inherent property of the -ing complement form. There is some evidence that it is a property of the to infinitive. On the other hand, the fact that a construction has a certain semantic profile does not mean that all aspects of this profile are likely to be equally prominent in all contexts of use. We have already noted (on p. 75) in the case of the two intend constructions that in some contexts the semantic distinction between the two forms just does not matter, leaving it up to the addressees in the actual contexts to filter out any unwarranted connotations. The third objection raised by Simon-Vandenbergen to the latent alternative hypothesis is what she sees as the need for ad hoc explanations in certain contexts. She refers in particular to constructions containing the matrix verbs choose, prefer and deign (2003: 254-255). We have already noted that there may be varying degrees of semantic compatibility between the profiles of the matrix verb and the complement clause. In some cases, for example choose and the to

Weaknesses of the targeted alternative approach 107

infinitive complement, there is a considerable degree of overlapping. Both the verb and the complement form imply the existence of alternatives. The same overlapping exists in the case of prefer and the to infinitive complement. But prefer also occurs with an -ing complement. In the latter case there is no overlapping of the semantic profiles. Every construction at the ‘matrix verb (lexeme) + complement form’ level has a semantic profile composed of the profiles of its two components. The meaning of the complement form is more abstract than that of the matrix verb. What Simon-Vandenbergen refers to as ad hoc explanations are characterisations of the meaning of constructions at this level. The sort of latent alternative implied by ‘want to infinitive’, which is one of several possible future scenarios, is different to the sort of latent alternative implied by ‘like to infinitive’, which refers to a recurrent scenario preferred by the subject to possible alternatives. The meaning of the complement form is the same. The meaning of the matrix verb differs. The meaning of the complex form ‘matrix verb + complement form’ is also necessarily different. Given the great diversity in meaning of matrix verbs it is not surprising that the sort of alternative implied, explicit or implicit, should vary from construction to construction. This does not make the explanations ad hoc, just different. In this section I have listed, and attempted to refute, some of the objections that have been raised to the targeted alternative hypothesis. Chapters 5, 6 and 7 contain numerous comparisons of constructions containing the to infinitive form with constructions containing other forms of non-finite complement, either with the same matrix verb, such as like, fear and dread, or with near-synonymous pairs of verbs, such as let/allow, and make/force. In the course of these comparisons, reference will be continually made to the targeted alternative hypothesis, and readers will be provided with hundreds of examples on which to base their own evaluation of it.

4.4 The -ing form

Like the to infinitive, the -ing complement form occurs with all nine types of matrix verb listed in Table 2.4 (p. 31). It also occurs with four of the six types of construction listed in Table 2.6 (p. 42). Combining the two categorisations, it occurs in as many as eighteen different construction types, making it the most versatile form of non-finite complement. It occurs in Same-time constructions with six different types of matrix verb, in Backward-looking constructions with four different types, and in Forward-looking constructions with seven. It also occurs in the Contemplation construction type, which is always Mental Process. One example of each of the constructions is listed in (216) – (233).

Same-time constructions:

(216) ‘I adored her being there in a place I equate with hell.’ (K32 1033) (Attitude)


108

(217) In spite of her love of reading Eva denies being an egghead. (H7E 417) (Communication)

(218) You’re going to a friend’s for a barbecue on a hot summer’s evening, and you’re walking up the drive and you can smell the barbecue sizzling away. (JK8 17) (Perception)

(219) I tried working myself into the ground, but I could be totally exhausted and still remember. (H8F 3155) (Effort)

(220) The two men continued walking, trying to fathom the mystery of the fog-bound train. (CE9 279) (Aspect)

(221) Romany mimed them going about the boat, looking here and looking there. (FEM 1961) (Applied Attitude)

Backward-looking constructions:

(222) The warden at the time remembers seeing someone like Eila, but they could find no record of her name. (K1X 710) (Mental Process) (223) Now she regretted stripping the clothes from his body. (HTM 2278)

(Attitude) (224) Richard Baylis, defending, said his client admitted smoking heroin in the

past but had told him he would not be going near it again. (C88 527) (Communication)

(225) Tessa stopped crying and sighed a deep, uncontrollable sigh like a yawn. (G12 1551) (Aspect)

Forward-looking constructions:

(226) Then we envisage starting a housegroup on the estate, and then a monthly family service, and in the end a church. (C8L 70) (Mental Process)

(227) She says he must have been released early, and she’s dreading him coming to see her and demanding his parents’ address. (CKF 2137) (Attitude)

(228) The give-us-our trough back campaign is threatening to organise a protest march every week but Councillor Smith says he won’t even discuss moving the trough until he receives a full apology from his accusers. (K1H 525) (Communication)

(229) The Data Protection Act forbids us releasing names and addresses of members without their permission. (EEL 486) (Enablement)

(230) Yes, it’s prevention, you’ve got to prevent the crimes being committed, not just chase after the people and hang and flog them once you catch them. (JJG 666) (Causation)

(231) Then they began following my daughter and she was forced to leave the country as well. ( EVF 955) (Aspect)

(232) Joanne had delayed telling anyone until four days before she went into labour. (FU1 2231) (Applied Attitude)

The -ing form: introduction 109

Contemplation construction:

(233) She had never imagined wearing clothes like this, or that they could make such a difference. (AT7 1101) (Mental Process)

Not only is the -ing complement form extremely versatile with respect to the types of matrix verb with which it occurs, it is also very versatile when it comes to the type of complement situation it encodes. It may be used to encode activities, as in (218) and (231), accomplishments, as in (219) and (228), achievements, as in (222), and states, as in (216) and (217). The fact that it can be used to encode achievements and states distinguishes it from the progressive, which may only be used with activities and accomplishments in unmarked contexts. The nature of the relationship between the -ing of the progressive and the -ing of the gerund-participial, to adopt the composite term used by Huddleston and Pullum (2002: 1187), is just one of the puzzles facing the grammarian grappling with the semantics of the -ing form. He or she also has to explain its meaning in the four basic construction types in which it occurs and offer an explanation as to why it does not occur in the other two, the General and Judgement types. Finally, the explanation of its semantics should be compatible with its employment as subject and adverbial. These questions will be addressed in turn in the following sections.

4.4.1 The -ing form in complement constructions

4.4.1.1 Comparison with the progressive

In the discussion of Duffley’s approach to the meaning of the -ing form in section 3.7, we referred to Huddleston and Pullum’s (2002: 163) characterisation of the progressive in terms of six properties, ongoingness, imperfectivity, interiority, duration, dynamism and limited duration. Some of the scholars whose work was discussed in Chapter 3 posit one of more of these properties as central to the meaning of the gerund-participial -ing form. Dixon bases his characterisation on the notions of duration and ongoingness, Duffley his on the notion of interiority, and Langacker his on the notions of imperfectivity and interiority. There thus seems to be a degree of consensus that the -ing form as complement shares some meaning components with the progressive, although there is disagreement as to just which components these are. In the following we will consider to what extent each of the six properties which Huddleston (the sole author of the chapter in question in the Cambridge Grammar) proposes for the progressive may also be posited of the gerund-participial -ing form as represented in the corpus material. The properties proposed by Huddleston for the progressive are designed to account for its use with activities and accomplishments. They can also account for its use with achievements and states in marked contexts. We have already


110

noted the fact that the -ing complement form can be used with stative predicates, without these thereby losing their stative meaning. Thus, a felicitous paraphrase of (217) using a finite that complement would be Eva denies that she is an egghead, not Eva denies that she is being an egghead. In fact, not only may the -ing form be used with all of Vendler’s (1967) four classes of predicate, it may also be used with each of these in each of our four complement types. Table 4.1 contains a truncated example from the BNC of each of the sixteen possible combinations. ‘S-T’ stands for Same-time, ‘B-L’ and ‘F-L’ for Backward- and Forward-looking, and ‘Con.’ for Contemplation. One should note that as one moves from left to right and top to bottom in the table one may expect to encounter fewer and fewer corpus tokens of the various constructions. Nevertheless, one can certainly find tokens instantiating each sort of combination. We will refer to the examples in the table in the discussion below.

Table 4.1: Vendler’s four categories of activities, accomplishments, states, and achievements, as they are instantiated in Same-time, Forward-looking, Backward-looking and Contemplation -ing complement constructions

A.

Activities B. Accomplishments

C. States

D. Achievements

1. S-T

...You can smell the barbecue sizzling away...

... could hear doors opening and closing ...

I adored her being there...

I hadn’t noticed him clocking the fact ...

2. F-L

...they began following my daughter...

...and began screwing the inspection plate back into place ...

I anticipate this being very useful...

... have postponed making a decision...

3. B-L

...his client admitted smoking heroin in the past...

I recollect one girl putting on dark glasses ....

I recall her being near one of the windows...

Helen regretted giving her consent...

The -ing form: comparison with the progressive 111

A. Activities

B. Accomplishments

C. States

D. Achievements

4. Con.

...she had never imagined wearing clothes like this...

...he had contemplated writing a book of essays...

...so as to conceive them existing unperceived.

... he will consider taking a peerage.

In the next four sections we consider to what extent the sixteen construction types in Table 4.1 are compatible with Huddleston’s characterisation of the progressive. We begin with activities.

4.4.1.2 -ing complements encoding activities

Same-time activities such as 1A in Table 4.1, cited in full as (218), satisfy five of the six criteria posited by Huddleston and Pullum for the progressive. They encode an activity in progress at the time referred to by the matrix verb, and this activity is profiled as imperfective in that the barbecue may be said to be sizzling at any point during its extension. It is also profiled as mid-interval - some sizzling has taken place and there is more to come. It is dynamic, in that the sizzling consists of various heterogeneous components and it is durative in that it lasts for some time. The one point on which it does not conform to the progressive prototype is with respect to the latter’s connotation of limited duration. There is no sense in which the sizzling in (218) is to be construed as an activity limited in time. However, limited duration is not a core component of the meaning of the progressive, according to Huddleston, merely a strong implicature. Forward-looking -ing constructions such as 2A in Table 4.1 and (234) – (235) share four of the five characteristics exhibited by Same-time activity -ing constructions.

(234) Erm it’s certainly grown faster than we anticipated it growing and we are

not able to er we’re not able to carry out what we like to do with Neighbourhood Watch. (GY4 217)

(235) ‘We don’t want people smelling something funny.’ (C86 1780) These -ing complements all encode activities as ongoing, imperfective, durative and dynamic. They differ from Same-time constructions in that they do not normally adopt a mid-interval perspective on the unfolding activity. The subject in (234) does not anticipate it being in the middle of the growing process, as it were. Nor does the matrix verb subject in (235) adopt a mid-interval perspective on the smelling of the complement clause subjects. This lack of a mid-interval


112

perspective does not affect the imperfective nature of the activities concerned. They are profiled as ongoing unbounded activities that may be expected to unfold at some time after that of the matrix verb. An inchoative matrix verb like began in 2A encodes the complement situation as having a left-hand temporal boundary, but this is a property of the matrix verb rather than the complement form. One could presumably insert an adverbial such as ‘still’ into the complements (234) and (235). We may also note that mid-interval resembles limited duration in Huddleston and Pullum’s classification, in that it is described as a strong implicature rather than a core property of the progressive. Backward-looking activity -ing complements, like 3A in Table 4.1 and (236) – (239), are similar to both Same-time and Forward-looking constructions with respect to the four progressive characteristics shared by these two construction types.

(236) I recall her pacing the sitting-room while I am doing my homework,



(238) Interestingly, John Clare also describes old women singing and telling stories during the weeding and haymaking. Evidently Complaining Daphne, though a pastoral, has an actual connection with the life of agricultural labour in Northamptonshire. (AN4 1088)

(239) She stopped speaking abruptly and stared at Clare with a look of fixed terror in her eyes. (H8N 1632)

All these Backward-looking constructions encode activities profiled as ongoing, imperfective, durative and dynamic. The question is whether they imply a mid-interval perspective on the unfolding activity. (236) and (237) have already been discussed as (127) and (128) in section 3.7. In both cases the subject recalls to his or her mind an unfolding activity, and in both cases the subject appears to view this unfolding activity from the inside, as it were. Thus, in the case of (237), that her machine was going makes for a more felicitous paraphrase than that her machine went. In (238) the subject also seems to adopt a mid-interval perspective. The subject profiles the women as being in the middle of singing and telling stories. It is, however, difficult to say to what extent this impression is due to the matrix verb describe and how much to the complement form. Finally, it is a moot question whether or not the perspective in (239) should be described as mid-interval. The ongoing activity of speaking is construed from the point of its cessation, not a point prior to this. Note that the fact of its interruption does not imply that the activity itself is to be construed as having limited duration. We may conclude that activities in Backward-looking -ing constructions are ongoing, imperfective, durative and dynamic, and may or may not be mid-interval, depending on the matrix verb in question.

-ing complements encoding activities 113

The final category of activity -ing construction to be considered consists of Contemplation activities, as in 4A in Table 4.1 and (240) – (242).

(240) Can you imagine playing this when you’re just out of your head? (KCU 9252) (241) When you feel able to contemplate running again, turn to our Events

listings on page 69. (CB4 754) (242) When he invited formers through the Darlington and Stocton Times to

consider growing mushrooms as a way of diversifying he received well over 100 calls although he was only looking for ten. (K51 1384)

Again, all of these examples encode activities as ongoing, imperfective, durative and dynamic. None of them are profiled as of limited duration. This leaves the mid-interval property. As was the case with the Backwards-looking construction, the Contemplation tokens seem to vary with respect to whether they encode a mid-interval perspective on the complement situation, with verbs like imagine and fancy implying that the situation is viewed from a point within its unfolding. Verbs like consider and contemplate, on the other hand, do not carry this implication. Imagine and fancy are actually much more likely to occur with activities than are consider and contemplate. Over 40% of tokens of the ‘imagine -ing’ type in the BNC encode activities, as opposed to fewer than 10% of the ‘contemplate -ing’ type. We tend to imagine activities and states, to consider accomplishments and achievements. Be that as it may, with respect to the question of mid-interval perspective, we may conclude that Contemplation constructions resemble Backward-looking ones in that they may or may not encode such a perspective, depending on the matrix verb. We have now looked at complement situations of the activity type encoded by the -ing form in all four of our main classes of construction. The results of this discussion are presented in tabular form in Table 4.2.

Table 4.2: Progressive characteristics exhibited by activities in Same-time, Forward-looking, Backward-looking and Contemplation constructions

Activities S-T F-L B-L

Con

ongoing imperfective mid-interval durative dynamic limited duration

+ + + + + -

+ + - + + -

+ +

+/- + + -

+ +

+/- + + -


114

4.4.1.3 -ing complements encoding accomplishments

There are two sorts of predicates in Vendler’s (1967) classification that occur freely in the progressive. One consists of activities, the other of accomplishments. In the case of the latter the situation in progress is one which, unless interrupted, will come to an end of its own accord (under its own momentum, in force-dynamic terms). Same-time accomplishments, as in 1B in Table 4.1 and (243) – (245), resemble Same-time activities in that they share five of the six properties of the progressive, the exception being limited duration.

(243) She had a bad moment as she heard the alley door closing behind her, but

she was already on her way down. (GW0 34) (244) Meanwhile, she can help herself to control her panic attacks by slowing

down her breathing as soon as she feels an attack coming on. (A70 1108) (245) And business executives have disliked travelling to a special studio. (B73 1029) These examples are all ongoing, imperfective, durative and dynamic. They are also mid-interval. In (243), for instance, the alley door has started to close but is not yet closed. We saw in the previous section that the mid-interval property was not encoded by the -ing form with activities in Forward-looking constructions. It does not seem to be a property of such constructions with complement situations of the accomplishment type either. Consider in this respect 2B in Table 4.1 and (246) – (248),

(246) Buildings occupied by the PSP and by the Shiite militia Amal were also

repossessed and on July 27 Hezbollah began evacuating its largest base in Lebanon - the Sheikh Abdullah Barracks in Bekaa, 70 km east of Beirut. (HLM 2283)

(247) At about nine o’clock I start putting Mum to bed. (HJC 177) (248) Partnerships could henceforth be established between consenting adults so

that ‘two men could live permanently together without fearing prattling informers bringing down the criminal law upon them’. (ASK 298)

In none of these examples is the accomplishment viewed from a particular point within its unfolding. Nor is it profiled as being of limited duration. The other four properties are, however, exhibited by all three examples. Backward-looking accomplishment -ing complements, such as 3B in Table 4.1 and (249) – (251), differ from the six types already discussed in one important respect, relating to the property of limited duration.

(249) I will never forget Kevin McMonagle (Old Blue Eyes) singing Tom

Leonard’s brilliant version of ‘My Way’ (‘though some may mock/the macho talk/upon the Walk/of No Surrender/I’ve drank the rent/I’ve

-ing complements encoding accomplishments 115

clocked the wife/I’ve spewed my ring upon the fender’) or John Cobb as Alasdair’s sad flowerseller, or Siobhan Redmond’s Bo’ness hippy. (B38 1411)

(250) He also denies raping a 26-year-old woman in her home earlier the same day. (K5D 13420)

(251) Mark has recently finished writing a paper on the harmful effects of secrets in families. (HGU 728)

These examples all encode complement situations profiled as ongoing, imperfective, durative and dynamic. This much they have in common with the activity types and the other accomplishment types. They also share with some of these the lack of a mid-interval perspective. However, they differ from them all in encoding events as having limited duration. This follows from the fact that an accomplishment, unlike an activity, is inherently bounded (it is telic). One of the functions of the -ing form is to defocus such boundaries (see Langacker 2000: 230). However, in the case of Backward-looking accomplishments, located as they are in the past vis-à-vis the matrix verb, the boundaries are refocused. One might say that the boundaries resume their position on stage, as it were. Contemplation accomplishment -ing constructions such as 4B in Table 4.1 and (252) – (254) resemble Contemplation activities in most respects, the exception being the absence of a mid-interval perspective, occasionally present in the activity construction.

(252) Imagine scanning an image as a bitmap and then converting that to

PostScript. (G00 231) (253) Jane contemplated designing a jumper based on her favourite sport of

canoeing, but found it too ambitious. (CA2 1084) (254) Having just produced the film of Gore Vidal’s political play The Best

Man, Turman started to consider making the movie from Webb’s book. (C9U 263) While these tokens are undoubtedly imperfective and are profiled as unfolding before the mind’s eye, there is no one privileged mid-position from which we are invited to view this unfolding process. As in the case of complement situations of the activity type encoded by the -ing form, we sum up of our discussion of complements of the accomplishment type in tabular form in Table 4.3.


116

Table 4.3: Progressive characteristics exhibited by accomplishments in Same-time, Forward-looking, Backward-looking and Contemplation constructions

Accomplishments S-T F-L B-L

Con


+ + + + + -

+ + - + + -

+ + - + + +

+ + - + + -

4.4.1.4 -ing complements encoding states

Same-time -ing constructions encoding states, such as 1C in Table 4.1, cited in full as (216), and (255) – (257) share three of the progressive’s six properties. They are imperfective, mid-interval and durative.

(255) Apart from being sick and having to keep going to the loo I loved being

pregnant. (FU1 1941) (256) I don’t mind it being quiet’. (CEB 579) (257) Why had she never realised before now that Dana resented looking like

her sister? (H8J 1990) States are, by definition, imperfective and durative. One effect of using the progressive with a stative verb is to impose a limited duration construal on the situation in question. This is not the case with the -ing complement construction. For instance, in the case of (255), already discussed as (226) in Chapter 3, the state of being pregnant is undoubtedly inherently bounded, but its boundaries are not in focus in this construction. The subject merely states that she derived great pleasure from being in the middle of the condition of pregnancy, as it were. Forward-looking -ing constructions encoding states, such as 2C in Table 4.1 and (258) – (259) resemble Same-time statives in being imperfective and durative. They differ from them, however, with respect to the mid-interval property. (258) Referring to the rescheduling agreement signed with the club in March

1989 [see p. 37140], Falae said that ‘Nigeria does not foresee being in a position to maintain payments at contractual levels beyond the first quarter of this year’. (HL0 2811)

-ing complements encoding states 117

(259) The French goal-ace had feared being out of the game for two months after hobbling out of the 3-0 first leg defeat with a damaged hamstring. (CH3 4783)

The states in these examples are viewed as extended, without any particular time within their extension constituting a focal point. To put it simply, the subject in (259) fears the whole state of exclusion from the game of football, not just the being in the middle of such a state of exclusion. We may note that this lack of a mid-interval perspective is not peculiar to Forward-looking -ing complements encoding states. The same point holds for activities and accomplishments. Backward-looking -ing constructions encoding states, such as 3C in Table 4.1 and (260) – (261), are also imperfective and durative. (260) Oh, more more than that, because I was, I I can remember being in school

(JK1 532) (261) She might rue the night for many reasons, but deep down she could never

regret being part of something so beautiful. (HA9 2576) The complements in these constructions are not ongoing, dynamic or of limited duration. This leaves the mid-interval property. Do 3C, (260) and (261) profile the states in question from a point within their extension? As was the case with Backward-looking activity-encoding -ing constructions, the answer to this question seems to depend on the semantics of the matrix verb. In Mental Process constructions like 3C and (260), the perspective appears to be mid-interval, in the sense that the subjects situate themselves mentally within the period during which the state pertained. This does not seem to be the case with an Attitude matrix verb such as regret in (261). It is the whole state that is regretted (or ‘not regretted’ in this particular case). In allowing for both perspectives Backward-looking constructions encoding states resemble Backward-looking constructions encoding activities rather than accomplishments. The latter are never construed from a mid-interval perspective. The final type of stative -ing construction to be considered is the Contemplation type, exemplified by 4C in Table 4.1 and (262) – (263).

(262) Sir, did you ever consider being an English teacher? (KPX 231) (263) ‘I can’t imagine being like that, when I’m old’ (FB9 949)

Stative complements in Contemplation -ing constructions resemble the other three stative types in that they are imperfective and durative. They resemble Contemplation -ing constructions with activity complements in that they imply a mid-interval perspective with certain matrix verbs. As with activities, verbs like imagine imply such a perspective. In (263), for instance, the subject situates him- or herself within the extended state of being like that. This does not seem to be the case with a verb like consider. In (262) being a teacher is profiled as extended, but construed from a point outside its extension. We may therefore


118

conclude that the presence or absence of the mid-interval property depends on the semantics of the matrix verb. We should also bear in mind that the mid-interval property resembles limited duration in that it is a strong implicature rather than a core property of the progressive. Table 4.4 summarises the discussion of stative -ing complements.

Table 4.4: Progressive characteristics exhibited by states in Same-time, Forward-looking, Backward-looking and Contemplation constructions

States S-T F-L B-L

Con


- + + + - -

- + - + - -

- +

+/- + - -

- +

+/- + - -

4.4.1.5 -ing complements encoding achievements

Achievements resemble states in that they do no occur freely with the progressive. They are not durative, but punctual. If used with the progressive, as in ”We’ve played with a lot of spirit and commitment, but people defend that much better when they are winning” (K2D 3006), they resemble accomplishments, in that they connote a period before an inherent right-hand boundary. The question facing us now is how they are to be interpreted in -ing complements, starting with Same-time constructions, exemplified by 1D in Table 4.1 and (264) – (266).

(264) She disliked losing her temper against her will because it rarely achieved

useful results and symbolised impotence rather than strength: always better to channel anger than display it, she thought. (G15 392)

(265) Any criteria for defining ‘privileged information’, or what the ‘dangers’ are in revealing such matters as dispositions (of manpower) are so vaguely incorporated as to ensure that few will risk submitting an essay without approval, which might later be assessed as an ‘improper disclosure’.

(A0K 428) (266) Perhaps he’d always secretly resented giving his name to another man’s

child, she thought sadly. (H88 124) What is the effect of adding -ing to an inherently perfective root such as clock in 1D, lose in (264), submit in (265) and give his name in (266)? In the progressive

-ing complements encoding achievements 119

the total effect is the imposition of a (limited) durative construal. The -ing complement form also exhibits imperfectivity and durativity, although there is no implication that the latter is limited. Thus in (264) the losing which induces simultaneous dislike on the part of the subject is not presented as a single bounded event. It may occur again and again; and each time it occurs, it will prompt the same feelings of distaste. The potential for recurrence is obviously not present in a context such as that of 1D and (266). Nevertheless, it does seem to be the case that giving his name in (266) is also viewed imperfectively and as having a certain duration, just as it does in the progressive. It also seems to be the case that these constructions impose a mid-interval perspective on the situations in question. This is a function of the same-time nature of the construction. One dislikes or resents doing something as one is doing it. We may therefore conclude that Same-time achievements, like Same-time states, share four properties with Same-time activities and accomplishments. Like the latter they are imperfective, mid-interval, durative and of unlimited duration. Unlike them they are not ongoing and dynamic. Forward-looking achievements like 2D in Table 4.1 and (267) – (269) resemble Same-time achievements in being imperfective, durative and of unlimited duration. They differ from them in not adapting a mid-interval perspective.

(267) ‘I dread mine reaching their teens. (BLW 1071) (268) The most probable reason for this unjust behaviour by the Roman

Governor was that he feared some sort of riot breaking out in the capital city during the festival. (CEJ 2983)

(269) It is easy to foresee a DG very quickly falling flat on his face or stirring up more problems than he started with. (CBW 2651)

These utterances resemble achievements with the progressive in that they focus on a process leading up to a point in time. Thus whereas it only takes a birthday for a child to reach his or her teens, it is clear that the subject in (267) implies that the period before, and indeed after, this coming of age fills her with dismay. Similarly in (268) the focus is not restricted to the actual point of the riot’s breaking out. With respect to the question of whether or not they imply a mid-interval perspective, however, these constructions resemble the other three sorts of Forward-looking constructions in not stipulating any particular point within the complement situation from which the remainder is viewed. Backward-looking -ing constructions with achievements are exemplified by 3D in Table 4.1 and by (270) – (272).


120

(270) Eva recollects making a commitment to Christ when she became a junior soldier. (H7E 241)

(271) Give the board a good jolt and the l.e.d. should stop flashing for another minute. (C92 331)

(272) She regretted asking the question as soon as the words were out. (H8S 3006)

The first, and most basic, question to be addressed is whether the constructions in these examples encode the situation in question as perfective or imperfective. When Eve in (270) thinks back on her past is she focusing on the fact that she made a commitment or on the process of committing herself, at all points during which period she could be said to be in the process of making the commitment? Does the flashing in (271) refer to a once-off or recurring process? Does the subject in 3D think about the fact that she had given her consent, or does she replay in her mind the process of giving it? The answers to these questions are not given. Nor are there many such tokens in the corpus on which to base one’s conclusions. For my own part, I am inclined to come down on the side of imperfectivity and to say that these predicates admit of an element of duration. As is the case with Backward-looking accomplishments this element of duration is limited. This follows from the fact that the situations in question are telic and are located anterior to the time of the matrix verb. The sixteenth and last class of -ing complement construction to be considered is the Contemplation achievement class, illustrated by 4D in Table 4.1 and by (273) – (274).

(273) How could you even contemplate marrying a woman like that? (JXS 4011) (274) After spending hundreds of millions of dollars turning the division’s

manufacturing complex in Flint, Michigan, into a high-tech wonderland designed to beat the Japanese, GM’s bosses considered closing it down. (ABJ 2410)

In (273) and (274) the inherently punctual actions of marriage and closure are encoded as having a certain duration. That is, the whole predication is encoded as having a certain duration. It only takes a second to marry someone, but one can spend considerable amounts of time thinking about the prospect! The complement situations are imperfective, in so far as at any particular moment of contemplation of considering one may be said to be engaged in contemplating or considering. However, the situations are not viewed from any particular point within their temporal extension. They resemble closely Contemplation accomplishments in both these respects. Table 4.5 summarises the discussion of the four types of achievement constructions.

-ing complements encoding achievements 121

Table 4.5: Progressive characteristics exhibited by achievements in Same-time, Forward-looking, Backward-looking and Contemplation constructions

Achievements S-T F-L B-L

Con


- + + + - -

- + - + - -

- + - + - -

- + - + - -

4.4.1.6 Prototypical characteristics of the -ing complement

We have now looked at all four main types of -ing complement construction as they encode all four of Vendler’s types of predication. The results of our discussions were summarised in Tables 4.2 - 4.5. These are now conflated in Table 4.6, which mirrors Table 4.1 in the composition of its rows and columns. We can see that as one moves from top to bottom and left to right in Table 4.6 the various construction types share fewer and fewer characteristics with the Same-time activity (and accomplishment) -ing complements. These two types may therefore be said to be prototypical for -ing complements as a whole. We can see in Table 4.6 that Same-time activities and accomplishments share most characteristics with the other most frequent types. We can also see that Backward-looking achievements are the most peripheral members of the class of -ing complements. Only two of six features, imperfectivity and durativity are shared by all sixteen types of construction: these are schematic for -ing complements. Four other features - ongoingness, mid-interval, dynamism and unlimited duration - are shared by Same-time activities and accomplishments. The remaining thirteen types share three, four or five features with the prototype.


122

Table 4.6: Huddleston’s six characteristics of the progressive as exhibited by Vendler’s four categories of Activities, Accomplishments, States, and Achievements, as these are instantiated in Same-time, Forward-looking, Backward-looking and Contemplation -ing complement constructions

Activities Accomplish- ments

States Achieve- ments

S-T ongoing imperfective mid-interval durative dynamic limited duration

+ + + + + -

+ + + + + -

- + + + - -

- + + + - -

F-L ongoing imperfective mid-interval durative dynamic limited duration

+ + - + + -

+ + - + + -

- + - + - -

- + - + - -

B-L ongoing imperfective mid-interval durative dynamic limited duration

+ +

+/- + + -

+ + - + + +

- +

+/- + - -

- + - + - +

Con ongoing imperfective mid-interval durative dynamic limited duration

+ +

+/- + + -

+ + - + + -

- +

+/- + - -

- + - + - -

All tokens of -ing complements in the downloaded random samples were categorised according to the system of Tables 4.1 and 4.6. Totals for the corpus as a whole were projected for each category on the basis of these tokens. These totals were adjusted to the nearest hundred in order not to give an exaggerated

Prototypical characteristics of the -ing complement 123

impression of the degree of accuracy claimed for them. After all, they are projections, not exact figures. The results of these calculations are given in Table 4.7. Note that as one moves from top to bottom and from left to right in the table one encounters fewer tokens, with the exception of the class of Forward-looking and Backward-looking states. Same-time activities are the most frequent type of construction, Contemplation states and achievements the least frequent.

Table 4.7: Projected totals, to the nearest 100, for the BNC as a whole, of the sixteen combinations of matrix verb and -ing complement predicate

Activities Accomplishments States Achievements Totals S-T 28,000 6,000 3,000 1,900 38,900 F-L 17,500 6,100 900 1,000 25,500 B-L 6,700 2,100 1,500 700 11,000 Con 1,400 1,300 200 200 3,100 Totals 53,600 15,500 5,600 3,800 78,500

Comparing Tables 4.7 and 4.6 we see that, in the case of -ing complements, constructions encoding the prototypical sense of the form are used much more frequently than constructions encoding the most peripheral senses. The figures for tokens in Table 4.7 also mirrors quite closely the figures for types in Table 2.7 (p. 43), as may be seen in Figure 4.3, which contains details of the percentage of -ing types and tokens in my material.

0,0

10,0

20,0

30,0

40,0

50,0

60,0

Same-time Forward-looking Backward-looking

Contemplation

types

tokens

Figure 4.3: The percentages of types and tokens of four classes of -ing

complement construction.


124

Figure 4.4 illustrates the network of -ing complements with Same-time activities and accomplishments as the prototype. Accomplishments Activities States Achievements

Figure 4.4: The network of -ing complements according to their realisation in

Same-time, Backward-looking, Forward-looking and Contemplation -ing constructions with activity, accomplishment, state and achievement predicates, showing Same-time activities and accomplishments as the prototype. Double-headed arrows indicate the complete duplication of features. The arrow labelled a denotes the subtraction of the properties of ongoingness and dynamism. Arrow b denotes the subtraction of the property of mid-interval, and arrow c the addition of the property of limited duration.

The prototypical -ing complement, as it is realised in Same-time activities, as in (275), is very similar to the progressive, as in (276).

S-T S-T S-T S-T ongoing imperfective mid-interval durative dynamic

F-L Con.

F-L (B-L) (Con.)

F-L (B-L) (Con.)

F-L Con.

a

b

B-L B-L

c


(275) ‘But by the time I was eighteen, I realised that, as much as I loved studying the past, my greatest joy came from - well, I suppose you’d describe it as planning things and watching them grow. (JY7 3823)

(276) FIRST-TIME voter Alison of Fife Road, Darlington, is studying history and Psychology A-levels at Darlington College of Technology. (K55 7249) Both (275) and (276) are ongoing, imperfective, durative and dynamic. They also adopt a mid-interval perspective on the activities in question. The main difference between them is the limited duration implication in (276), in which Alison’s study of history is profiled as a temporary activity. Same-time accomplishments share all the properties of Same-time activities. (277) and (278) illustrate accomplishments in a Same-time construction and the progressive.

(277) And business executives have disliked travelling to a special studio. (B73 1029) (278) Detectives are travelling to Scotland with Denise’s parents, who are to

take care of little Gareth. (K41 591)

Again the main difference between the progressive and the Same-time complement is the implication in the case of the former that the travelling is soon to come to a close, to be specific at the point when the detectives reach Scotland (or some particular destination in Scotland). The travelling in (277) would also presumably come to a close whenever the special studio is reached, but this end-point is defocused in the complement construction. In fact, it is probably more correct to say that it is defocused by affixing the -ing ending to the accomplishment root, only to be refocused in the progressive by the grounding predication ‘be’. As may be seen in Figure 4.4, Same-time states and achievements are extensions of the activity prototype, the extension involving the loss of the two closely related components of ongoingness and dynamism. Thus, as we saw in section 4.4.1.4, these -ing complements are imperfective, durative and mid-interval and profile situations of unlimited duration. (279) and (280) instantiate states in the Same-time -ing construction and in the progressive.

(279) Jezrael hated being the only one there who wore a bodysuit but she would

have hated more the embarrassment of displaying herself with all her imperfections to the crude and lumpen miners. (FP0 2379)

(280) And if you happen to see him sitting in front of the television, dressed in a plain white vest rather than one of his talkative T-shirts, why assume that he is being a slob? (G2V 520)

Note that the difference between these two constructions is exactly analogous to the difference between Same-time and progressive activities, and Same-time and


126

progressive accomplishments, in that the progressive implies that the complement situation is of limited duration, an implication clearly lacking in the stative -ing complement. We find the same distinction in the case of Same-time and progressive achievements, as in (281) and (282).

(281) The nurses were well trained in dealing with rich patients who were used

to doing as they pleased and often disliked accepting the discipline of routine; they knew that the very old, the alcoholics, and the more than slightly batty patients (called ‘eccentric’) had to be carefully supervised. (FPB 2319)

(282) Where the consumer is accepting an offer by the creditor, the copy of the agreement must be given to him immediately (s63(1)). (J7D 351)

In (282) the right-hand temporal boundary of the process of acceptance is looming large, a point underlined by the urgency expressed in the temporal adverbial clause. In (281), on the other hand, there is no implication that the discipline in question is accepted once and for all. We have now traced the extension of the prototype -ing complement construction from Same-time activities to Same-time accomplishments, states and achievements. I will not examine in detail all the other extensions illustrated in Figure 4.4. Examples of each of the relevant categories are contained in sections 4.4.1.2 to 4.4.1.5. I will restrict my comments in the present context to the two processes of extension on the vertical axis of Figure 4.4, involving the loss of the properties of mid-interval and unlimited duration, respectively. Examples (283) – (287) all encode activities.

(283) The Scouse flicked over a page of his cowboy book and continued reading

as the three Commandos chorused, ‘You lying bastard’. (A61 1871) (284) The Captain is reading the paper, Corporal L is making the coffee and

Corporal M is half asleep. (A77 1414) (285) She stood there, leaning on the phone table, and started reading the names

incised into the wood. (EDN 1283) (286) We sold all our male records, stopped reading the newspaper and

watching the telly. (CF4 184) (287) Imagine reading some text. (HRD 1143)

The Same-time activity in (283) resembles the progressive one in (284) in adopting a mid-interval perspective on the activity in question. In (283) the Scouse is in the middle of reading, as is the Captain in (284). The extension of the prototype to the Forward-looking construction in (285) involves the loss of this mid-interval perspective. The subject in (285) is at the start of, rather than in the middle of, an ongoing imperfective situation. Backward-looking and Contemplation constructions are enclosed in brackets in Figure 4.4 since, as was explained in section 4.4.1.2, these may or may not denote a mid-interval perspective. Whether they do so or not depends on the matrix verb in question.


Thus the activity of reading in (286) is profiled from the point of its cessation. In (287), on the other hand, the perspective is mid-interval: it may be paraphrased Imagine being in the middle of reading some text. Activities are different in this respect from accomplishments which, as was pointed out in section 4.4.1.3, never allow of a mid-interval perspective in Backward-looking and Contemplation constructions. (288) and (289) exemplify Contemplation and Backward-looking accomplishments, respectively.

(288) Will busy managers contemplate reading such a large amount of material?

(CR4 932) (289) When Dr Mortimer had finished reading this strange story, he looked

across at Sherlock Holmes. (H7V 121)

In (288) there is no implication that the focus is on the managers’ being in the middle of reading all the material in question. Rather it is on the totality of the durative process of so doing. The same point may be made in relation to (289) which, in addition, exhibits the added property of limited duration. The reading of this strange story lasted through a definite limited period of time. Before leaving Figure 4.4 we may note that the relationship between the various constructions on the lower levels of the figure, those illustrating Forward-looking, Backward-looking and Contemplation constructions, mirrors that on the level of Same-time constructions. For example, Forward-looking -ing constructions encoding states and achievements, as in (290) and (291), bear the same relationship to Forward-looking activities, as in (292), as do Same-time states and achievements to Same-time activities.

(290) Mam was just about to be rehoused by the council, and didn’t fancy being

so far out of town. (H9D 2093) (291) ‘So - I gather you didn’t fancy joining my ‘harem”? (H8S 3255) (292) I don’t fancy playing tag along the road with those guys, even if you do, so

stop complaining.’ (JXV 354)

All three constructions in (290) – (292) encode situations as imperfective and durative. The Forward-looking activity in (292) has two properties in addition to these, ongoingness and dynamism. These properties are lacking in Forward-looking states and achievements, just as they are in Same-time -ing constructions encoding states and achievements. We have now delineated the semantics of the prototypical -ing complement construction and sketched some of the extensions from this prototype. It remains for us, in the next section, to consider the schematic meaning of the -ing form.


128

4.4.1.7 The schematic meaning of the -ing form

All sixteen combinations of the four -ing construction types and the four complement situation types described in sections 4.4.1.2. to 4.4.1.5 have two properties in common; they are all imperfective and durative. We therefore take these two properties to be schematic for the -ing complement type. The element of duration was said to be a defining characteristic of the -ing form by Dixon, who writes that “an ING complement refers to an activity or state as extended in time, perhaps noting the way in which it unfolds” (1991: 218). We concluded in section 3.4 (p. 54) that this definition could account for three of the four -ing complement types, the exception being Contemplation constructions, which do not locate situations in the time domain. In order to account for these, one could emend Dixon’s characterisation to read “an -ing complement encodes a situation as extended in some domain, either mental or physical”. In this definition the term situation has been substituted for activities and states because, as we have seen in 4.4.1.3 and 4.4.1.5, accomplishments and achievements may also occur in -ing complements. Our examination of the various construction types also allows us specify the sorts of contexts in which Dixon’s occasional property “perhaps noting the way in which it unfolds” pertains. This is in those cases which display the mid-interval property, in other words all Same-time constructions and some Backward-looking and Contemplation activities and states. Duration is one core characteristic of the -ing complement form. The other is imperfectivity. Langacker (1991: 91, 2000: 230) maintains that adding -ing to a perfective stem creates an imperfective relation that profiles a process as unfolding. No reference is made to its beginning or ending. He maintains, moreover, that all the component states making up this complex relation are viewed as homogeneous and scanned summarily, thus neutralising the distinction between perfective and imperfective processes. Thus, the addition of -ing is seen as neutralising the difference between a perfective verbal root and an imperfective one, as far as the profiled part of the process is concerned, the profile of both types being equally ‘flat’ or undifferentiated within the immediate scope of predication. Langacker is undoubtedly correct in maintaining that -ing has the effect of creating an imperfective relation profiling a process as unfolding. With respect to the proposed quality of internal homogeneity, however, it has been argued at several points in both this and the previous chapter that this is not a property of the -ing form as such. Recall the discussion in section 3.8 (pp. 66 – 67) of tokens (124) – (128). Two of these, (125) and (127) are repeated here as (293) – (294), (the latter was also cited as (236) on p. 112). (293) She could hear Penry moving about upstairs as he made up the other bed.

(JYC 1462) (294) I recall her pacing the sitting-room while I am doing my homework,


The schematic meaning of the -ing form 129

To repeat the questions raised in section 3.8, do we not apprehend the movements of the bed-maker in (293) as involving various different locations? To put it simply, do we not picture him moving from one side of the bed to the other? And do we not see the subject in (294) as walking forwards and backwards, forwards and backwards, and are not walking forwards and walking backwards different actions? Moreover, the reference to the occasional pauses in the pacing activity in (294) indicates that it is not to be interpreted as unbroken or homogeneous. We must conclude that, at least in some cases, the situation encoded by the -ing form may be heterogeneous. Langacker himself acknowledges that the internal states of situations encoded by the -ing form are not always completely homogeneous. However, he considers any variety in this respect to be of negligible importance compared to their similarity. He writes: “The component states are not identical in any strict sense, but their degree of divergence depends on the level of schematicity at which they are viewed. I propose that the participle focuses on the commonality of the profiled states as component members of the same base process, and portrays them as a homogeneous set on the basis of this abstract similarity” (Langacker 1991: 92). I would propose that the level of schematicity at which the components of the complement situation are viewed varies according to the complement type, with Same-time activities and accomplishments and some Backward-looking and Contemplation activities inviting a construal of the components as more heterogeneous. In other words those dynamic constructions which allow a mid-interval perspective also allow one to perceive the difference between the various components of the ongoing situation. Those constructions, whether dynamic or static, which exclude a mid-interval perspective, which encode situations as viewed from without, from a distance, do not allow one to perceive the heterogeneity of the various components. If some -ing complement types open for the detection of separate components within their extension, the question arises of how these are to be scanned. It is clear that those viewed from a distance may be scanned summarily, but the term ‘summary scanning’ is less appropriate to describe our method of construal of dynamic, mid-interval -ing constructions. Nor does ‘sequential scanning’ seem to be an appropriate label for the -ing in such contexts. Recall that the function of the -ing stem is to profile the central portion of a process, i.e. the process minus its starting and end points. The -ing form imposes this internal profile on the stem of any verb with which it combines. I suggest that the addition of -ing leads to the various component states being scanned not summarily nor sequentially but recursively, in a sort of loop. In section 4.2 we likened the bare infinitive to a frozen picture, contrasting it with the motion picture effect of a finite form. To extend the analogy, the -ing form of a verb may be likened to the action of running on the spot. It encodes a marking of time as it were, wherein one repeatedly goes through the motions of a process without actually kick-starting it. One can also employ the analogy of the circumference of a circle to illustrate this perpetuum mobile characteristic of the -ing form.


130

Figure 4.5: Illustrating the circular profile imposed on a process by the -ing suffix

If one looks at the circle in Figure 4.5 one can take in the whole picture at a glance. Any one point on the circumference of the circle is situated between two others and one of these is naturally viewed as being in front of, and the other behind, the point in question. When dealing with the space domain we may of course choose to scan the circumference of a circle in either a clock-wise or an anti-clockwise manner. However, in the time domain, there is an inherent order to the unfolding of processes, indicated in the figure by an arrow. Put simply, in a canonical ‘running’ situation, one runs forwards rather than backwards! Figure 4.5 illustrates the semantics of the general schema for the -ing form. It represents a process as lasting through time, as durative. It is also imperfective: at no point on the circumference of the circle is one no longer engaged in the unfolding of the process in question. The figure may also be taken to illustrate the -ing construction type with stative verbs. Activity verbs with -ing have the additional components of dynamicity and ongoingness. They are illustrated in Figure 4.6.

Figure 4.6: Illustrating the circular profile imposed on an activity by the -ing suffix

The components of an activity -ing complement are construed as heterogeneous in some construction types, the ones that allow of a mid-interval perspective. This heterogeneity is illustrated in Figure 4.6 by the changing form of the circumference. When viewed from a point on the circumference, in a Same-time activity, for instance, the various components are appraised to be different. This was the case with the woman pacing the floor in (294). When viewed from a distance, however, as is the case with Forward-looking activities, it is no longer

Activity -ing

Verb -ing

The schematic meaning of the -ing form 131

possible to make out the component parts and the whole activity appears to be homogeneous. Achievements resemble statives in that their component parts are always viewed as homogeneous. Langacker has proposed a recursive construal for punctual verbs in the progressive. He writes: “Under normal circumstances, an “internal perspective” hardly seems possible of the point-like events they designate, and indeed, in the progressive they generally receive a repetitive construal” (Langacker 1991: 209). Punctual verbs can easily be accounted for in the illustration in Figure 4.5 by marking a point on the circumference of the circle, as in Figure 4.7.

Figure 4.7: Illustrating the circular profile imposed on an achievement by the -ing suffix

In some cases, such as (272), which instantiates a Backward-looking construction with the Attitude matrix verb ‘regret’, it is natural to interpret the complement as involving just one turn of the wheel, as it were. In cases with Mental Process verbs, both Forward-looking, as in (269) and Backward-looking, as in (270), the subjects can replay the situation as often as they wish in their own minds. They are at liberty to scan the situation either summarily or recursively. Different matrix verbs invite different scanning strategies. The final type of complement situation to be illustrated consists of accomplishments. These may be illustrated as in Figure 4.8, with the same irregular circumference as activities and with a cross marking the end-point, as employed in Figure 4.7 for achievements.

Figure 4.8: Illustrating the circular profile imposed on an accomplishment by the -ing suffix

Accomplishment

Achievement


132

Accomplishments resemble activities in that their component parts may be profiled as heterogeneous in Same-time -ing complements. That is, when viewed from a point on the circumference of the circle in Figure 4.8, the differences between the component parts are apparent. When viewed from a distance, as is the case with all other types of accomplishment -ing construction, this difference is no longer perceptible. Whether the durative situation encoded by the -ing form is to be scanned summarily or recursively depends on the nature of the viewing relationship. The construction types that invite the construal of the complement situation from a mid-interval perspective allow us to distinguish between the wood and the trees so to speak. In the case of constructions in which the situation is construed from outside all we can see is the wood. The final question to be raised in this section is whether this variation in perspective should be incorporated in our schematic characterisation of the -ing complement form. In other words, is it to be regarded as a variable property of the form, or is it a function of the construction type (in the case of Backward-looking and Contemplation activities and states, a function both of the matrix verb type and the complement clause type). As there would seem to be no good reason for including variables in schematic characterisations, it makes most sense to regard recursive scanning, for example, as a property of a certain construction type. I therefore posit the following definition for the schematic sense of the -ing complement form:

The -ing form profiles a situation as extended in some specified domain. This definition also accounts for the -ing form as used in the progressive. In the next section we will enquire as to whether it is also appropriate for the -ing form when used as a subject or adverbial.

4.4.2 The -ing form as subject and adverbial

In section 4.3.2 we looked into the question of whether the characterisation proposed for the to infinitive complement form could also be applied to to infinitive subjects and adverbials. Now we will look into the same question with respect to the characterisation of the -ing form proposed in the previous section. We concentrate our enquiry on the same four verbs as in the case of the to infinitive, live, love, win and write (the first two stative, the third an achievement and the fourth either an activity or an accomplishment). A random sample of tokens with utterance-initial -ing forms of the four verbs was downloaded from the BNC. This yielded 65 instances of -ing clauses as subject. The -ing form as subject occurs with a greater variety of predicators than does the to infinitive. Some half of the 65 occurrences are with copular predicators, as opposed to over two thirds in the case of to infinitives. Most of the others belong to the Enablement or Causation categories, such as allow or make,

The-ing form as subject and adverbial 133

but there are also a handful of other types of predicator, such as blunt, take priority and come. Another feature of -ing as subject is that it occurs with predicators displaying a greater range of TAM relations. Among the 65 tokens of -ing as subject there are utterances in which the situation encoded in the -ing clause is located at the same time as, before, and after the time of the main verb, as well as occurrences in which it is not located in time at all. (295) – (298) illustrate all four possibilities.

(295) Sybil Myerson, of London’s Royal Free Hospital, who has studied

violence against GPs, said: ‘Living in fear of attack is a terrible stress and one which GPs could do without.’ (CEN 5527)

(296) Winning the French Open took him through the door, down the passage, into the street and out into the sunshine Fame. (A0V 1120)

(297) ‘Winning on Saturday is what is important - a glorious Cup tradition doesn’t pay bills but a Cup run this year will do.’ (CBG 5320)

(298) Winning a major prize in a competition is something people only dream of. (HP4 431)

In (295) the living is simultaneous with the stress, in (296) the winning came before the tennis player in question passed through the door, in (297) the winning is located in the projected future, and in (298) it is profiled as having a purely mental existence. In other words we find all four domains in which -ing clauses occur as complements represented among tokens of -ing clauses as subject. The ability of the -ing form as subject to occur in all sorts of contexts is no doubt related to its encoding a situation perceived as having a certain degree of independent existence. It is extended in some domain, mental or physical. Compared to the to infinitive form it has a greater degree of semantic independence, possessing, as it does, aspectual contours of its own. The second function which we will examine in relation to the -ing form is its occurrence in adverbials. 500 instances of infinitive and -ing forms of the same four verbs (apart from loving, which was only represented by 339 tokens) were downloaded from the BNC. From these were extracted the tokens, 190 in total, in which the -ing form was used adverbially. (299) – (301) illustrate various uses containing the form living.

(299) ‘Besides’, he went on, ignoring her interruption, ‘living in the midlands, I

enjoy short sea trips. (HA7 775) (300) No wonder you got headaches, living with a crime like that for all these

years. (FSF 265) (301) Anyway Malcolm quickly agreed to the terms on the lease and Steve and I

moved in, living there as well as using it as a rehearsal room. (A6E 1280)

May the -ing clauses in (299) – (301) be said to encode situations profiled as extended in some domain? And, if so, is it possible to specify the domain in question? In the case of all three utterances the answer to the first of these


134

questions is ‘yes’. The living, which is to be interpreted in terms of existence in (299) and residence in (300) and (301), is profiled as lasting over a certain period of time in each case. As for the second question, the answer must be ‘no’. Whereas the -ing clause in (299) encodes a situation that is unfolding at the same time as the main verb, in (300) the period of living in question precedes (and overlaps with) the suffering of headaches by the subject, and in (301) the period of residence obviously succeeds the action of moving encoded by the main predicate. Thus, yet again we see that it is impossible to pin down the -ing clause to any particular ontological domain. Nor, as is pointed out by Quirk et al. (1985), in their discussion of supplementive clauses, is it an easy matter to state the exact meaning of these clauses in many contexts.

The formal inexplicitness of supplementive clauses allows considerable flexibility in what we may wish them to convey. According to context, we may wish to imply temporal, conditional, causal, concessive, or circumstantial relationship. In short, the supplementive clause implies an accompanying circumstance to the situation described in the matrix clause. For the reader or hearer, the actual nature of the accompanying circumstance has to be inferred from the context. (Quirk et al. 1985: 1124)

The sort of flexibility referred to here by Quirk et al. may be detected to some extent in (299) – (301). Thus the -ing clauses in (299) and (300) clearly encode a causal relationship and in (301) a temporal (successive) one. In each case it is up to the addressee to construct a plausible relationship between the situation in the -ing clause and the situation in the main clause. There is no need to cite further examples of -ing clauses as adverbials. In all of the downloaded tokens, the -ing clause may be said to encode a situation profiled as extended in some domain. In other words the characterisation of the -ing form proposed for its use in complement constructions in section 4.4.1 has proved adequate to account for its use both as a subject and as an adverbial. This is in contrast with the situation for the to infinitive, the characterisation of which had to be adjusted (broadened) in order to account for these other uses. In the next section we sum up our discussion of the -ing complement form and inquire as to why it is (almost) never used in General and Judgement constructions.

4.4.3 The -ing complement form: a summary

We are now reaching the close of what has been an extended examination of the -ing complement form. The length of the examination was necessitated by the fact that the form can occur in so very many different contexts, with all sorts of matrix verbs and complement predicates and in four of our six basic construction types. The two construction types which are incompatible with the -ing form are Judgement and General constructions. It does not normally occur in General constructions for the simple reason that the aspectual contours of a situation

The -ing complement form: a summary 135

encoded by an -ing form are not compatible with the intermittently recurring nature of non-stative General complement situations.7 The -ing form profiles a situation as extended along the time axis. General constructions profile situations that are not extended as such but that are likely to recur at more or less regular intervals on the time axis. As for Judgement constructions, these always admit of an element of doubt as to the existence of the complement situation. They always point to an option chosen among several more-or-less latent alternatives. Situations encoded by -ing never imply any element of doubt. They always profile a situation as occurring, or non-occurring, in some specified domain. Apart from these two exceptions, -ing complements occur freely in both different-subject and same-subject constructions. Let us round off our discussion of the -ing complement form by revisiting, for a moment, some of the proposals for its meaning discussed in Chapter 3. We can see now that Wierzbicka’s idea of vague simultaneity and Smith and Escobedo’s idea of conceptual overlapping may suitably be applied to what we have shown to be the prototypical -ing complement type, comprising Same-time activities and achievements, as well as to other Same-time construction types. On the other hand, these characterisations are not applicable to half of the construction types in Tables 4.1 and 4.6. Dixon’s emphasis on the element of extension accounts, as we have mentioned on several occasions, for the majority of construction types. Langacker, who emphasises the imperfectivity of the situation encoded by -ing form, comes closest to accounting for the full range of meanings of the form. The main difference between the proposals made in this section and Langacker’s treatment is the extent to which we open for a construal of the internal portions of some -ing construction types as heterogeneous in nature.

7 The prohibition on the employment of -ing in General constructions was not as pronounced in Modern English as it is in present-day English. As mentioned in section 2.3, in the case of the matrix verb prefer one can still chance upon tokens where -ing complements are used in a general sense (see section 5.5.3).

4.5 The to -ing form

The fourth and final complement type to be considered in this chapter consists of to plus the -ing form. As was explained in section 1.5, the coverage of the to -ing construction in this book is less thorough than the coverage of the other three types. This is because neither of the lists of matrix verbs chosen as a starting point for the study were designed with to -ing constructions in mind. Nevertheless, some of these constructions did surface among the tokens extracted from the BNC. They occurred with three sorts of matrix verb and in two of the six basic construction types, Forward-looking constructions and Judgement constructions, as exemplified by (302) – (304).


136

Forward-looking constructions;

(302) You should aspire to being a specialist in the central nervous systems of business, which is what its accounts should depict. (AJ2 110) (Attitude)

(303) ‘Oh yeah,’ said Lydia gloomily and fell to plucking at the turf between her feet. (G0X 205) (Aspect)

Judgement construction:

(304) Chairman, I must confess to being erm, to being somewhat pessimistic

about this whole er this this whole problem of of the need to refurbish and the need to tackle numbers in our elderly persons homes. (J9M 133) (Communication)

Four of the six matrix verbs that were found to occur in to -ing constructions also occur in to infinitive constructions. These are admit, aspire, confess and profess. An attempt to dissect the meaning of the to -ing construction might profitably start by enquiring as to whether the various pairs of constructions containing these matrix verbs have any semantic components in common. Let us start by comparing the ‘profess to infinitive’ construction, exemplified in (305) and (306), with the ‘profess to -ing’ construction, exemplified in (307) and (308).

(305) Interviews with people who profess to enjoy classical music turn up all

sorts and conditions of appreciation. (CR7 452) (306) She professed to have been a cook in a house where I was once governess.

(EVG 2081) (307) At the Plough Inn in Hereford, everyone professed to enjoying a pint of

the local product: (K1T 3462) (308) She professes to having been a little sad when it all ended, but stressed that

all involved are still friends - to the point, indeed, where Trevor Hutchinson is now a lynchpin of her own band. (ED7 811)

(305) and (307) both contain the complement predicate enjoy. They differ, however, in that in (307) the imbibing of the local beer is to be understood as an accomplishment rather than a state, with the subject being interpreted as an Agent rather than an Experiencer. The -ing form adds a connotation of duration and imperfectivity to the situation: i.e. it is represented as extending through time. In (305), on the other hand, the enjoyment is profiled as a whole. The subjects are Experiencers rather than Agents. And, while (306) profiles a state as a whole, (308) encodes a state which is represented as lasting for a certain time (to be specific, at the time when it all ended). As for the contribution of to to the meaning of ‘profess to -ing’, it would appear to be similar to its meaning in

The to -ing form 137

‘profess to infinitive’, i.e. it signals out the targeted alternative which the subject claims to be true. The second Communication Judgement verb that can occur with both to infinitive and to -ing complements is confess, exemplified by (309) – (311).

(309) I must confess to be somewhat surprised to see the Labour Group wanting

to debate thisu this issue again after their lack lustre performance in Committee and Council on the previous debates on this subject and today’s been no different. (JJD 434)

(310) My Lord Mayor, I must confess to being quite surprised by the incredible naivety and complete lack of understanding that’s obviously rampant on the benches opposite with most of the items we’ve dealt with today.

(J9J 192) (311) Wilson was quick to see how provocative such a development would be to

the shipowners and refused to accept the post, though he confessed to have been ‘foolish enough to have allowed myself to be nominated and elected as honorary general manager, which meant that I was lending my name to an organization over which I had no control’. (FES 501)

There were 86 instances of ‘confess to’ among the 1,000 tokens containing confess downloaded from the BNC. Of these, 7 are followed by the infinitive and 79 by -ing. The to infinitive complement predicates are either be (3 tokens) or have (4 tokens), which are also the most common predicates in to -ing constructions, with 15 and 14 occurrences respectively. It is obvious that, of the two constructions, the one with the to infinitive complement is more restricted in its collocations. It cannot be used to refer to an ongoing activity, either in the past or the present, but must refer either to a state or the result of an action or process. It would thus be impossible to substitute it for the to -ing construction in (312) or (313).

(312) This week she finally confessed to carrying on behind her husband’s back.

(CBF 4816) (313) Further computer training: I have to confess to feeling a bit stumped by

this. (HX7 126) (314) In an age when sin and madness were so strongly associated, the only

reason for confessing to having been insane would be that complete recovery showed divine forgiveness of whatever sin had caused the madness in the first place; and this is the reason advanced by the two mediaeval authors who admit to having been mad, Margery Kempe and Thomas Hoccleve. (CFX 642)

The to infinitive complement is also more restricted in register, in that its use is typically confined to more formal contexts, either spoken or written. Nevertheless, despite these differences, the two constructions may be used to encode very similar situations, as in (309) and (310), and (311) and (314). The


138

fact that it may make little difference in some circumstances which of two alternative constructions a speaker chooses to employ is, as has been pointed out several times already, by no means unusual. It may, however, bear repeating that this fact should not be taken to mean that the two constructions are identical in meaning. The fact that only one of them can be used in certain cases, as is the case with (312) and (313), shows that there is, indeed, a difference between them. The to -ing complement form is some ten times as frequent as its to infinitive counterpart in the case of confess. The opposite is the case for aspire of which there are 143 tokens in the BNC with a to infinitive complement and only 14 of to -ing. The two constructions are exemplified by (315) and (316).

(315) Most Communists, despite their theoretical commitment to sexual

equality, looked askance at any woman who aspired to be more than a tractor driver or street-sweeper. (CCK 70)

(316) ‘If you ever aspire to being a gentleman,’ she said with quiet dignity, ‘you must take care of Katarina.’ (H94 3268)

The ‘aspire to -ing’ construction has not received a lot of attention. There are no examples of it in the OED. The Cobuild English Dictionary does list it, though without exemplification, but the Cambridge International Dictionary of English and Longman Dictionary of Contemporary English do not. Quirk et al. (1985: 1188) list ‘aspire to’ as a prepositional verb but only as one taking a nominal object. Rudanko (1998) does not mention it at all in his survey of matrix verbs that take both to infinitive and to -ing complements, while Rudanko (1996) notes that ‘aspire at -ing’ is obsolete but does not include ‘aspire’ in his discussion of verbs taking ‘to -ing’. Despite this relative paucity of previous notice there is no doubt, on the basis of the evidence in the BNC, that ‘aspire to -ing’ exists as a construction in its own right. The obvious question is how, if at all, it differs in meaning from the to infinitive construction. Table 4.8 contains figures for both aspire constructions detailing the extent to which the aspiration is asserted, the achievability of the goal and the incidence of the most prominent complement predicates. It will be immediately obvious from Table 4.8 that any assertion of complementarity in distribution cannot be based on crude distinctions such as those recorded in the table. Nor is it possible to posit a genre- or register-based explanation. Examples of ‘aspire to -ing’ are to be found both in works of fiction and non-fiction, in sources ranging from books to newspapers, from Hansard to unpublished mss. Any difference must be sought elsewhere.


Table 4.8: ‘Aspire to infinitive’ and ‘aspire to -ing’ compared with respect to the assertiveness of the matrix verb, the attainability of the complement situation and the extent to which they occur with complement predicators ‘be’ and ‘become’ (Note that only rows three and four are mutually exclusive.)

to infinitive

to -ing

purely assertive matrix verb

57 (39.9%) 8 (57,1%)

attainable SIT2 95 (66.4%) 9 (64,3%)

complement predicate be(ing)

43 (30,1%) 8 (57,1%)

complement predicate becom(ing)

8 (5.6%) 2 (14,3%)

Total 143 14

We begin our examination of the two constructions with utterances containing the most common complement clause verb in both constructions, be.

(317) This may be excellent fodder for people who are aspiring to be the next

winner of Mastermind. (B7G 1277) (318) ‘It is only the cream of our young manhood who can aspire to be one of

our glorious cosmonauts. (CDA 1551) (319) That’s not to say he aspired to being something of a genius himself, it’s

just that he felt a tremendous affinity with those people. (ART 176) (320) The Mesquite main strip aspires to being Las Vegas without anything like

the money or the reputation. (HH0 75) The ‘aspire to be’ construction is, in many cases, very similar in meaning to ‘hope to become’. Typical in this respect are (315) and (317) – (318). ‘hope to become’ would not suffice, however, as a paraphrase of ‘aspire to being’ in (319) – (320). Something like ‘pretend to the status of’ would be more appropriate. Contrast in this respect (317) where the complement situation, be the next winner, is clearly envisaged by the subject as the target of their future hopes, with (320), where the complement situation, being Las Vegas, is almost presumed by the subject to have already been realised. In both cases, the complement situation is encoded as the goal of the subject. If one aspires to be/being something, one will not necessarily attain it, but it is still very much the goal of one’s hopes. As for any semantic difference between the complement forms, this may be easier to spot with non-stative complement predicators, as in (321) – (324), than with the quintessentially imperfective ‘be’.


140

(321) But luckily for his successors, Montagus, Brudenells and, since 1790, Scotts, Montagu did not aspire to create an English Versailles.

(CKX 1020) (322) This is a similar fallacy to that of those in my profession who are

constantly aspiring to bring everybody up to the average (A69 707) (323) Most of us do not really aspire to developing huge muscles, enabling us to

lift great weights, but, by developing extra strength, the muscles are then working well within their capability for most everyday tasks. (EB1 1643)

(324) Her shops are called Shabby Chic and the most successful one to date is in New York, where ‘life is so hectic and claustrophobic that people aspire to making their home as comfortable and pleasant as they can.’ (CB8 2333)

All four complement situations in (321) – (324) are profiled as the goal, the targeted alternative, of their various subjects (or not, in the case of negated matrix verbs). They are also all accomplishments. The question is whether the right-hand temporal boundary is defocused in the to -ing complements, as it is in ordinary -ing complements. In other words, do the subjects in (324) aspire to achieving maximum comfort in their homes once and for all, or is this to be understood as an ongoing process? And is the development of huge muscles in (323) profiled as a closed or open-ended process? There is no test that will provide us with a clear answer to these questions. For my own part, my impression is that the to -ing complements exhibit the imperfective and ongoing nature posited of the -ing form in section 4.4, as well as the targeted alternative connotation posited of the to infinitive form in section 4.3. The final matrix verb to be considered in this section is admit, which is actually found in the BNC with three non-finite complement forms, as in (325) – (328).

(325) A man has admitted knocking down and killing a cyclist after drinking

more than five pints of lager. (K1H 2461) (326) Southall, who admitted speeding at an earlier hearing, was fined Pounds

700 and given six penalty points. (CBF 10890) (327) He admitted to have been a virgin at twenty-two. (ACS 828) (328) Another Toyota official in the USA referred to the project as ‘Yuki’s

dream’ and downplayed the company’s part in it, though he admitted to knowing few details. (BNV 1495)

Of the three constructions exemplified in (325) – (328), there is no doubt that admit strongly favours the -ing construction. Of 1,000 random tokens of admit downloaded from the BNC, (327) was the sole example of a to infinitive complement and (328) one of only two instance of to -ing. (In the light of these figures one may well wonder why the New Oxford Dictionary of English, which is based on the BNC, has one example of the latter construction and none of the much more common ‘admit -ing’.) ‘Admit -ing’ normally encodes a complement situation which is given in the discourse context. It is the situation of which the


subject stands accused. In (325) the subject has been accused of knocking down and killing a cyclist. In (326) the subject has been accused of speeding. In both cases the situation in question is profiled as having taken place in the past. They are classified as Backward-looking constructions. In (327), on the other hand, no one has accused the subject of having been a virgin at the age of 22: indeed, rather the opposite presumption prevailed in the universe of discourse, as the immediate context, cited here as (329), makes clear. (329) Or so the stories went, and excellent publicity they proved to be. But he

grew rich on no more than a promise. He himself was still quietly married to a girl he had been introduced to at high school. He admitted to have been a virgin at twenty-two. (ACS 828)

(329) is taken from an article about Hugh Hefner. In it the subject points to an alternative that is far from obvious to his addressees and claims that this alternative is the correct one. Similarly, in (328), the Toyota official might have been suspected of knowing rather more than a few details. In fact, he introduces a rather less obvious alternative and claims that this is true. Note that there is no suggestion of an alternative to the given complement situation in the case of the ‘admit -ing’ construction in (325) and (326). It is the element of appealing to the judgement of the addressee to accept that the complement situation is in conformity with the facts that has led to the classification of both ‘admit to’ and ‘admit to -ing’ as Judgement constructions. In this respect they both resemble ‘confess to (-ing)’. We have now looked at four matrix verbs that occur in to -ing constructions. In the case of each we have seen that the to -ing complement form shares some of the semantic characteristics of both the to infinitive and the -ing complement form. I therefore propose the following definition for the to -ing construction:

to -ing: a situation, viewed as extended, is profiled as the targeted of several theoretically possible situations in some specified domain.

In the next and final section of this chapter I will compare the definitions of the four complement forms arrived at in sections 4.2 to 4.5.

4.6 The four complement forms compared

In Chapter 1.4 the question was raised as to what extent it might be possible to posit abstract schemas for the four different non-finite complement types which together form the topic of this book. One of the aims of the study was to fill the boxes in Figure 1.2, repeated here as Figure 4.9, with functional content.


142

infinitive -ing

bare infinitive construction

(bare) -ing construction

to to infinitive construction

to -ing construction

Figure 4.9: The principal forms to be explained in the study The functional content in these boxes must be based on the definitions of the various constructions in the previous four sections. These are as follows: bare infinitive complements: a situation, viewed as a whole, is profiled as certain to occur in some specified domain. to infinitive complements: a situation, viewed as a whole, is profiled as the

more/most likely of two or more alternatives in some specified domain. -ing complements: a situation is profiled as extended in some specified domain. to -ing complements: a situation, viewed as extended, is profiled as the

more/most likely of two or more theoretically possible situations in some specified domain.

Characterisations of the four forms based on these definitions are contained in Figure 4.10. These characterisations are maximally schematic, in so far as they hold, to a greater or lesser extent, for all instantiations of these complement forms. There is also a duplication of the various properties by several constructions, both types of infinitive complement being said to portray a situation as a unitary whole, both types of -ing complement a situation as extended, and both types of to complement a targeted alternative. infinitive

-ing

- complement situation profiled as a unitary whole - (almost) certain to occur in some domain

- complement situation profiled as extended - existing in some domain, mental or physical

to

- complement situation profiled as a unitary whole - targeted alternative

- complement situation profiled as extended - targeted alternative

Figure 4.10: Maximally schematic characteristics of bare and to infinitive and -ing complement constructions

The four complement forms compared 143

The only slight discrepancy in this respect is that the bare infinitive construction is said to profile a situation as (almost) certain to occur in some domain and the -ing construction to profile a situation as existing in some domain, mental or physical. In fact the formulation with respect to the bare infinitive holds for all Same-time and Backward-looking -ing constructions. As for Forward-looking constructions, we have already mentioned that they usually profile a situation as much more likely of realisation than to infinitive complements with the same matrix verbs. We might even go so far as to say that they are almost certain to be realised. This leaves Contemplation constructions. These are restricted to the mental domain, in which they are always profiled as unfolding. If we drop the brackets around the adverb almost in Figure 4.10 we can apply this formulation to the -ing form. We can then rearrange the information in Figure 4.10 as in Figure 4.11. complement situation

profiled as a unitary whole

complement situation profiled as extended

almost certain to occur in some domain

bare infinitive

-ing

likely to occur in some domain

to infinitive

to -ing

Figure 4.11: Schematic characteristics of bare and to infinitive and -ing complement constructions

The rearrangement of the information in Figure 4.11 is designed to bring out more clearly the range of options which the arsenal of non-finite complement constructions places at the disposal of the language user. Both it and Figure 4.10 may, however, give the impression that we are dealing with more clear-cut distinctions than is, in fact, often the case. In particular they ignore the input of matrix verbs to the meaning of constructions. A matrix verb that itself encodes the idea of a targeted alternative, such as prefer, for instance, will be extremely compatible with the to infinitive complement form because of their overlapping profiles, but this does not mean that it is incompatible with the -ing form.8 Figure 4.11 lists the options available to the language user in any particular context of utterance. These enable the speaker to opt for the most

8 As a matter of fact the verb prefer when it entered the language occurred first with nominal and then non-finite -ing complements. It was only in the nineteenth century that it began to be used with to infinitive complements. (For details of this development see Egan: in press)


144

suitable construal of a situation. However, as has been mentioned several times already, there may well be situations which can be construed in any one of several different ways and where the variation in construal is of no practical importance to the speaker or addressee. In such cases the possibility of the existence of a latent alternative or the possible extended nature of the complement situation may be viewed as irrelevant by the speaker. Consider, for instance, a hypothetical situation in which a girl has proven reluctant to engage in conversation about her past. A man, whom we may call ‘John’, attempts to get her to do so. This attempt is successful. A third party, reporting on John’s success, may not consider the question of whether or not she continued talking for long to be pertinent to the point he or she wants to make. Nor may they be interested in highlighting the possibility that she may have chosen to remain silent on the topic. In this case it may make no difference to the speaker in question whether he or she chooses to say (330) John got her to talk about her childhood, or (331) John got her talking about her childhood. The essential point is that speakers have to make use of one of the options made available to them by the collective grammar of the speech community to which they belong. All of the options available to the speaker may well imply some element or other which he or she is not particularly interested in implying in any given situation. The speaker must, however, rely on the addressee to filter out these irrelevant implications. In other cases it may well make a difference to speakers which of the available options they choose. It is in cases like these that the distinctions posited in Tables 4.10 and 4.11 come into their own.

Chapter 5

Constructions in contrast: Same-time constructions

Spade took a cigar, trimmed the end of it and lighted it. Meanwhile the fat man pulled another green plush chair around to face Spade’s within convenient distance and placed a smoking-stand within easy reach of both chairs. Then he took his glass from the table, took a cigar from the box, and lowered himself into his chair. His bulbs stopped jouncing and settled into flabby rest. He sighed comfortably and said: ‘Now, sir, we’ll talk if you like, and I’ll tell you right out that I’m a man who likes talking to a man who likes to talk.’ Dashiel Hammett: The Maltese Falcon

5.1 Introduction

The focus in this and the following two chapters is on different forms of non-finite constructions containing either identical or near-synonymous matrix verbs. The latter category includes such pairs as try/attempt and let/allow. This chapter is devoted to Same-time constructions. Section 5.2 deals with bare infinitive and -ing complements with Perception verbs. Section 5.3 compares Same-time -ing and Judgement to infinitive complements with Perception and Applied Attitude verbs. Section 5.4 looks at Same-time -ing and Forward-looking to infinitive constructions with Effort and Attitude verbs, and section 5.5 at the contrast between Same-time -ing and General to infinitive constructions with Attitude and Aspect verbs. Finally, section 5.6 contains a summary of the various contrasts involving Same-time constructions.

5.2 Bare infinitive and -ing complements with Perception verbs

We begin our discussion of Same-time constructions with the well-recognised contrast between bare infinitive and -ing complements with Perception verbs. This contrast is often described in standard grammars in terms of progressiveness, with the -ing form said to exhibit progressive meaning. This is the stance taken by Quirk et al. (1985: 1206) and Huddleston and Pullum (2002: 1237). We have seen in section 4.4 that Same-time -ing complements only differ from the progressive with respect to one property, that of limited duration, so the standard description would seem to capture the essence of the contrast. Below we look more closely at the types of situations that are typically encoded by the two complement types.


146

All of the sixteen Perception matrix verbs contained in this study were found to occur with -ing complements. Half of them also occur with bare infinitive complements.9 These are feel, hear, notice, observe, overhear, see, spot and watch. The combined projected totals for the two types of complement with these eight verbs in the BNC (to the nearest 100) are 13,300 for the -ing complement and 9,700 for the bare infinitive complement. All eight verbs are related to one of three basic bodily senses, seeing, hearing and feeling, They are represented by the basic verbs see, hear, feel, and by semantic variants of these. Matrix verbs encoding the other two senses, of taste and smell, do not seem to occur with bare infinitive complements. If, as was maintained in Chapter 4, it is the case that bare infinitive complements profile a situation as a whole and -ing complements a situation as extended in some domain, the fact that taste does not appear to take bare infinitive complements is easily explained. It is due to the fact that objects of tasting are normally in transit through the mouth; they are, almost by definition, in progress. In the case of smell the absence of bare infinitive complements may well be related to the fact that the human sense of smell is not well-adapted to differentiating between ongoing and completed processes - smells have a tendency to linger on! There are four Perception matrix verbs in the study that encode a more general mode of perception, a mode not rooted in one of the specific bodily senses. These are catch, discover, find and miss. Semantically, these are more peripheral members of the category. The fact that two of them regularly take to infinitive complements, a form normally excluded by the other Perception verbs, may perhaps be taken as a reflection of this peripheral status. We will return to discover and find in the next section. For now we limit our attention to constructions encoding visual perception, exemplified by (332) – (334). (332) As we approach the quay we can see a small crowd waiting for our arrival.

(HH0 4345) (333) Trevor Sharpe, 16, saw his dad Roger fall to the ground 40 minutes into

the under-18s game. (CH2 4459) (334) For funerals, the Reverend Bottingley felt, were not a woman’s occasion,

they were too hard on the emotions, and on the feet and knees, and he did not like to see women weep in public. (AD1 28)

In (332) the complement clause clearly encodes an ongoing situation, the crowd being in the middle of waiting. In (333), on the other hand, the situation is profiled as complete. We could felicitously append ‘but they broke off their vigil’ to (332), but not ‘he interrupted his fall’ to (333). In other words, while the -ing

9 Table 1 in Appendix 2 contains real and projected totals for all Perception verbs.

Bare infinitive and -ing complements with Perception verbs 147

form after see in these utterances is imperfective, the bare infinitive is perfective. Langacker describes the contrast as follows:

Through time, then, the subject sees, as it occurs, the entire relationship profiled by the complement. The profiled relationship is necessarily limited to that portion of the subordinate process (V) which appears onstage within the temporal immediate scope: the full perfective process in the case of Ø + V [the bare infinitive], and for Ving a subpart that excludes the endpoints. (Langacker 1999: 232).

The contrast is described in very similar terms in Verspoor (1997: 444) and (2000: 214). A survey of all the tokens of the various visual perception constructions downloaded from the BNC reveals that these are the prototypical aspectual characteristics of the two forms in these constructions. Since it is perfective, the infinitive encodes completeness in the case of telic situations such as accomplishments and achievements, events, that is, with a well-defined right-hand temporal boundary. And all but two of the seventeen downloaded examples of ‘see S2 bare infinitive’ exhibit telicity. One of the two exceptions is cited as (334). In (334) there is no doubt that the subjects of the complement predicator definitely wept but there is no suggestion that their period of weeping would end when the matrix verb subject stopped observing them. It is much more common for activities such as that in (334) to be encoded by the -ing form after see, though not after watch, as we shall see. Watch differs from the other matrix verbs encoding visual perception in that it occurs more frequently with bare infinitive than -ing complements. Figure 5.1 compares the projected totals of see and watch with the two complement forms. Examples (335) – (339) illustrate the various uses of watch. (335) Karen watched the steward opening the bottle at the sideboard. (ECK 1581) (336) Early one morning, in the merry month of May, I stood in this dining room

watching a group of climbers from all over the world being introduced to The Great Tradition. (CCP 1346)

(337) When they were both soaked in the greasy oil, she watched him pour the remainder of the bottle over the rubber sheet. (FPX 2083)

(338) He watched her trot away down the road, red stirrup light shining on Midnight’s dark flank. (AT4 1315)

(339) I’d, I would rather watch Davie Cooper play football any day! (FL5 366)


148

0

2000

4000

6000

8000

10000

infinitive -ing

see

watch

Figure 5.1: Projected totals of bare infinitive and -ing complements of see and watch

The matrix verb watch itself denotes a durative situation. After it the -ing complement, as in (335) and (336), has its usual ongoing, mid-interval sense. The bare infinitive complement, on the other hand, although encoding completion in the case of telic predicates, such as the accomplishment in (337), would appear to be neutral as to completion in the case of atelic activities such as those in (338) and (339). The evidence of (338) and (339) appears to contradict the assertion of Dirven that “The difference between the bare infinitive and the present participle construction depends on the distinction between a whole or a completive and a non-completive, e.g. durative situation” (Dirven 1989: 123). We have already touched upon this distinction in section 3.7, where we cited Quirk et al. as stating that in the sentence Tim watched Bill mend the lamp, ”the bare infinitive, having nonprogressive meaning, implies that Bill did the whole job while Tim was watching” (Quirk et al. 1985: 106). The claim that the bare infinitive implies that the action was completed while the matrix clause subject was watching is, in fact, only true of telic complement clause predicates. In the case of atelic predicates, one could substitute -ing complements for their bare infinitive counterparts in sentences such as (338) – (339) with little change of meaning. What little change there is involves a clearer focus on the unfolding nature of the complement clause situation. The bare infinitive complement forms of these sentences entail their -ing counterparts. That is, in (337), if she watched him pour the remainder of the bottle, she must have watched him pouring it. Note that one could also substitute a bare infinitive form for the -ing complement in (338). In the case of the telic complement in (335), however, a sense of completion would be added by the substitution of the bare infinitive for -ing.

If one asks whether accomplishments like (337) or activities like (339) are more typical of the ‘watch S2 bare infinitive’ construction, the answer is ‘neither’. The 131 tokens of the construction among the 1,000 tokens of watch downloaded from the BNC contain 65 telic and 66 atelic complements. The contrast with the ‘watch S2 -ing construction’ is striking, in that as many as 55 of 59 tokens of the


latter encode activities. We noted above that a handful of the complements of see are atelic, but there is no doubt that watch stands out among the visual perception matrix verbs in occurring with a greater semantic range of bare infinitive complements. Figure 5.2 illustrates the comparative ratio of telic and atelic bare infinitive complements of watch and see.

0

10

20

30

40

50

60

70

80

90

100

see watch

telic

atelic

Figure 5.2: Ratio of telic to atelic bare infinitive complements of see and watch,

based on a sample of 17 of the former verb and 131 of the latter. Why should watch favour atelic bare infinitive complements much more than see? It differs from see, and from other visual perception matrix verbs like notice, spot and spy, in two main respects. In the first place it encodes a durative activity and in the second place its subject is agentive. The reason why the bare infinitive after watch is not always completive must be sought in a combination of these two features. The fact that the matrix verb subject is an Agent may contribute to his or her being accorded greater prominence in the composite situation, leading to the durative profile of his or her watching being superimposed on the neutral contours of the situation encoded by the bare infinitive. Whereas non-agentive matrix verbs of visual perception tend to be punctual and to favour -ing complements and atelic situations, watch, which is a verb of active perception, favours the bare infinitive form of complement. This is used to encode situations which may or may not be telic. The -ing form is, again, mostly used to encode atelic situations. The tendency for -ing complements to be employed to encode activities and bare infinitive complements to encode accomplishments is also true of the non-visual perception matrix verbs that occur in both bare infinitive and -ing constructions. (340) – (344) are typical utterances encoding aural perception.


150

(340) I’ve got some ear-plugs but they don’t work - I still hear him grinding, but don’t hear the alarm. (CFS 1161)

(341) It was while Cecilia was running upstairs for the third time to tell Tina she must get up, she really must or she would be late for school, that she heard the Cambridge School bell utter a single toll. (EDN 1343)

(342) What had really caused her, in the end, to escape from her life as an American Princess was overhearing her English maid talking with one of the senior parlourmaids. (HGE 2910)

(343) I overheard him say that he lost his desire, lost his drive and everything. (CBG 7918)

Both hear matrix verbs encode non-agentive perceptional experiences. A survey of the downloaded occurrences of both verbs reveals that their -ing complements normally encode atelic, and the bare infinitive complements telic, situations. In this respect the activity in (340) and the achievement in (341) are typical of the two constructions. In (340) the complement clause subject is engaged in the ongoing open-ended activity of grinding, while in (341) the complement clause encodes a single complete action of tolling the bell. The typicality of (340) and (341) is confirmed by the results of a forced-choice elicitation test described in Quirk (1995: 164). In this test the subjects had to insert either ‘slam’ or ‘slamming’ into the matrices “I heard the door … all night long” and “ I heard the door … just after midnight”. 59 of 64 subjects chose to insert ‘slamming’ in the former and ‘slam’ in the latter matrix. There is a greater ratio of -ing to bare infinitive complements in the case of overhear. This is presumably related to the fact that only situations of verbal communication are ‘overheard’. Situations involving ‘saying’, ‘telling’ or ‘mentioning’ may be either telic or atelic (depending principally on the nature of what is communicated), and thus may be felicitously used in both types of complement clause. Other conversation predicates, such as talk, discuss and speak are normally atelic and thus more likely to occur in -ing clauses. The point may be illustrated by (342) and (343). Feel resembles hear and overhear in that one can both feel something unfolding over a certain period of time and one can feel something happening more or less instantaneously. Both modes of perception are illustrated in (344) – (347). (344) As she stood in the middle of the crowd, she felt a hand touching her

bottom. (CDT 1118) (345) I could feel him rooting through my past. (CJA 2732) (346) He resolved not to look at the ground again until he felt his ’chute open.

(B3J 2533) (347) He felt the water in the pit rise up to meet him. (C86 1860) ‘Feel S2 -ing’ resembles the ‘hearing’ -ing constructions in that the complement clause predicate is almost always atelic. Thus, touching, in (344), is to be


interpreted in this context as an activity rather than an achievement, as ‘making continuous contact with’ rather than merely ‘glancing against’. The typical bare infinitive complement, on the other hand, is telic, either an achievement, as in (346), or an accomplishment, as in (347). To sum up, the non-visual perception constructions containing matrix verbs hear, overhear and feel resemble their visual counterparts in so far as -ing complements are employed to encode situations envisaged as extending through time and bare infinitive complements situations viewed as a unitary whole. In the next section we will consider some cases where to infinitive complements occur with Perception matrix verbs.

5.3 Same-time bare infinitive and -ing versus Judgement to infinitive

It happens on occasion that a Perception matrix verb is followed by a to infinitive complement. This does not occur very often in the case of sensory perception verbs - in the random downloadings of 1,000 tokens of these verbs that form the database for the present study, only observe occurred with a to infinitive complement. Nevertheless, as pointed out by Noël (2003: 360), there are some occurrences in the BNC of to infinitive complements with such matrix verbs as notice, see and feel. Example (348) is from my own database, examples (349) – (352) from Noël (2003). (348) We cannot be one hundred per cent sure that, just because we have

observed the sun to set each day on many occasions, the sun will set every day. (FBE 301)

(349) Other boys and girls were flitting hither and thither among the trees, singly, without a word or a sign of communication with one another. A few were older, many were younger, than I. Some like myself were looking about them. Others had found what they sought. These, when I passed them, I saw to be sitting or kneeling beside cradles, rocking them, or singing, or gazing intently. (ABL 734)

(350) Allow the patient to speak of what they notice to be wrong with themselves and try not to put words into the patient’s mouth. (B1R 262)

(351) My fingers ache and I feel my face to be beetroot-red. (G02 46) (352) As soon as he left the car he sensed the air to be damper and colder than in

London, musty with the scent of the distant North Sea, whose breakers were grinding the beach a few miles away at Felixstowe. (GUP 1193: omitted from World edition)

Bolinger (1974) distinguished between what he called perceptual and conceptual verbs: of these he wrote: “The bare infinitive is not used with verbs that are lexically conceptual [.....]. But the opposite - the to with verbs that are lexically perceptual - is quite normal. The perceptual verbs are being used in a conceptual sense, roughly equivalent to the same verbs with a that clause: I saw it


152

to be true = I saw that it was true”. (Bolinger 1974: 66). If this contention that the to infinitive complement clause is more or less equivalent to the finite that clause is correct, it would mean that the targeted alternative interpretation of the former was cancelled in these constructions. Such a cancellation requires an explanation. We return to this question below. Let us, however, begin by taking a closer look at these to infinitive constructions. We start with example (348), which we may contrast with the other non-finite constructions in (353) and (354).

(353) Eye-witnesses observed the Cherokee descend on final approach to a

height of thirty feet before the engine power was heard to increase as a go-around was initiated. (CAU 1474)

(354) He observed three youths loitering nearby, who disappeared from view as he arrived. (GXJ 2745)

The first point to be made about (348) and (353) – (354) is that all three utterances appear to be grammatically well-formed, despite Palmer’s (1965: 171) claim that observe only takes an -ing form. All 17 -ing complements in my material resemble (354) in so far as the complement situation profiles an extended activity or state. In (353) the bare infinitive complement predicate is an accomplishment, its inherent right-hand boundary reached by the aircraft at a height of thirty feet. It encodes the situation of descending as a whole, up to and including this right-hand boundary. In (348) the complement situation is also profiled as a unitary whole – the sun is not just seen to be going down, but to actually go down. Does it make sense to describe the setting of the sun as a targeted alternative, as the goal at the end of some path? In fact, the speaker in (348) invites such an explanation. In his or her opinion, it is just possible that a person observing the sun at the appropriate time of day will not see it set (he or she says We cannot be one hundred percent sure). In other words the speaker in this case views the action of setting as one of several alternative possibilities, albeit in this case the one whose realisation is signalled by the path-goal morpheme to. Do any other of the utterances listed as (349) – (352) invite this sort of contrastive inference? Both (350) and (352) do so explicitly, the former in the conjoined clause beginning and try not to, the latter in the comparative phrase than in London. This leaves examples (349) and (351). With respect to the former, I would venture the suggestion that utterances of the form ‘see x to be y’ imply an element of revised perception, perception after the mists have cleared, so to speak. A typical sort of usage is: “In the dark/from a distance, we imagined x to be y: we later saw it to be z”. (349) is of this type, as are (355) – (356). (355) As they approached, Reni rose from his seat at a table near the large

rectangular pool which was the centrepiece of what - as Huy now saw it to be - was an unconventionally asymmetrical garden. (H84 2027)

(356) In those days I think I must have regarded many of the people dotted about my landscape more as bushes than as the human beings I now see them to

Same-time bare infinitive and -ing versus Judgement to infinitive 153

have been: some dull and uninteresting like laurel bushes; some like lavender sweetly scented; some thorny, to be avoided; and then rose bushes, gooseberry bushes, fuchsia bushes - but all just part of the scenery, and I withdrawn from them into my dreams. (AC7 1083)

Both (355) and (356) contain the time adverbial now, indicating the newness of the perception. This same function is served by the punctual time adverbial when I passed them in (349). If we return to Bolinger’s contention that the to infinitive complement clause in these constructions is roughly equivalent to the finite that clause, we can now stipulate the nature of the equivalence in the following manner: “see x to be y” = “see that x is actually y”. If we paraphrase examples (348) – (352) and (355) – (356) using finite complements, we will find that adding the content disjunct ‘actually’ leads, in most cases, to the production of more felicitous paraphrases. Thus, in (349), “I saw that they were actually sitting…” conveys the sense of the original much better than the bare “I saw that they were sitting…”. Similarly, in (350), “what they notice is actually wrong” is more felicitous than “what they notice is wrong”. There is a simple reason why the paraphrases containing the disjunct ‘actually’ seem to work better. The reason is that ‘actually’ always points to a targeted alternative. In other words, finite complements containing ‘actually’ share an important part of their semantic profile with non-finite to infinitive complements. Note, by the way, that this explanation does not hold for example (351). In its case a paraphrase without actually would be just as felicitous as one containing it. The matrix verb feel also occurs in a Mental Process Judgement construction as in (357) – (358). (357) Women often feel that they are not at risk because they do not feel

themselves to be associated with any of the so called ‘high risk groups’ and, although we are encouraged to have safer sex, AIDS does not really seem to be an issue for us. (CJ9 650)

(358) Marcus, thus present occasionally at Jack’s house, where he evidently felt his visits to be a matter of duty, inevitably encountered the women, who treated him with suitable respect and awe, and Jack, who was nervously affable, and even on two occasions Gildas to whom he nodded politely. (APM 750)

(357) and (358) resemble the other four downloaded tokens of this construction in that they share the same complement predicator be. This is also the predicator in (351). It is possible that the existence of this other ‘feel S2 to be’ construction influenced the employment of this form in (351). Be that as it may, one can certainly accommodate (351) to our general schema for to infinitive clauses, in that being beetroot-red is just one of several alternatives for cheek-colour, and not the most obvious one at that!


154

There are two non-sensory Perception verbs that occur with both -ing and to infinitive, but not bare infinitive complements. These are discover and find, exemplified by (359) – (362). (359) They would has/e swum naked only Rufus’s father had discovered them

doing this and made a fuss out of all proportion to the offence, if offence it was. (CDB 1157)

(360) When she came back from Confession, uncomforted, she found her cell mate kicking the doctor who was trying to feed her, and at the same time yelling that he should take his hat off in the presence of a lady.

(FSP 2477) (361) Too late Yartek discovers one key to be the fake one. (F9Y 530) (362) The exhilaration soon faded upon checking his fuel state, which he found

to be low. (CGL 1645)

The -ing constructions denote the perception by the matrix verb subject of something which has been hidden from him or her. What is hidden is usually an ongoing process, as in (359) – (360). The reason why neither of these verbs, nor the closely related catch, take a bare infinitive complement is presumably due to their basic sense of coming upon something in progress precluding the perfective component inherent to the infinitive form. (361) and (362) involve an act of judgement as well as one of perception on the part of the matrix verb subject. They resemble the two -ing constructions in that they encode a sense of the subject’s coming upon something hidden. In their case, however, what is hidden is one of several alternatives. In (361) the implicit alternative is another key. In (362) the alternative is a greater amount of fuel. There is one more verb that occurs in Same-time -ing and Judgement to infinitive constructions. This is feign, exemplified by (363) – (367). (363) The Spurs striker is alleged to have feigned being head-butted by

Coventry’s Andy Pearce. (CH3 7434) (364) So we ended up staying the night at a hotel, The Majestic, the big one right

on the sea front As usual Malcolm feigned being borassic so she ended up shelling out for two rooms. (A6E 548)

(365) We feigned not to grasp the reference, though Durham City readers may understand it very well. (K4T 9056)

(366) The context of the Canterbury Tales is one in which Geoffrey Chaucer, a historical poet, feigns to speak through a fictitious mouthpiece, ‘Chaucer’ the pilgrim-narrator, who both, in turn, feign to speak through a further series of mouthpieces, the individual pilgrim-tellers. (HXS 917)

(367) Corbett rubbed his eyes wearily, feigning to be more exhausted than he really was. (BMN 1775)

In (363) and (364) feign is similar in meaning to mime. Both ‘feign -ing’ and ‘mime -ing’ encode situations in which the subject goes through the motions of

Same-time bare infinitive and -ing versus Judgement to infinitive 155

acting out the complement situation in order to convey to his or her audience the nature of the situation in question. In the ‘mime -ing’ constructions the observers are privy to the nature of the pretence. There is, in other words, no attempt to deceive. The subject goes through the motions of performing the action encoded in the complement clause, without possessing the appropriate props. The -ing clause profiles the unfolding in the mental domain of a situation mirroring that acted out in the physical domain, with it being left up to the observer to add the necessary props to complete the picture. ‘Feign -ing’ resembles ‘mime -ing’ except that in its case the subject does indeed intend to deceive the observers. ‘feign to infinitive’, on the other hand, resembles ‘pretend to infinitive’ in that it encodes a judgement by the speaker on the subject’s putative realisation of the complement situation. Both ‘pretend to infinitive’ and ‘feign to infinitive’ encode in their complement clauses the goals of the subject’s artifice, the to infinitive always signalling the situation which the subject hopes to delude his or her addressee into accepting as congruent with reality. This situation is the targeted alternative of the subject. The ‘feign -ing’ construction resembles the Judgement constructions in so far as it implies an attempt on the part of the subject to deceive. In (363), for example, the subject goes through the motions of the complement situation, acting it out with the intent of conveying a false impression to the observers. Nevertheless, in cases like this the -ing complement could not be replaced by a to infinitive complement. It is the actual playing out of the simulated situation that is profiled and the omission of this ongoing element would render the construction infelicitous. Turning to the ‘feign to infinitive’ complements, the situation in (365) could not be rendered by an -ing complement, the ‘grasping’ in question not being susceptible to extension through time. Nor would it be felicitous to substitute ‘speaking’ for to speak in (366). The situation referred to pertains at all times, throughout all the tales: being universal it is not compatible with the internal focus of the -ing form in Same-time constructions. Finally we may contrast (364) and (367), both of which contain copular complement clause predicates. Just as was the case with (365), the predication be exhausted is not naturally employed in an ongoing sense: thus, one may say ‘he’s just being borassic’ but hardly ‘he’s just being exhausted’. Although it is possible to tease out distinctions between the two feign constructions, it must nevertheless be admitted that they seem a lot closer in meaning than do the other pairs discussed in this section. This is due to the fact that the matrix verb feign itself implies the existence of an alternative to the situation in the complement clause. Perception verbs never imply in themselves the existence of alternatives. This makes it easier to discern the contribution of the to infinitive form to the meaning of Perception Judgement constructions. In the next two sections we will look at some much clearer contrasts involving Same-time -ing constructions, beginning with the contrast between Same-time -ing and Forward-looking to infinitive constructions.


156

5.4 Same-time -ing versus Forward-looking to infinitive

The two Effort verbs try and attempt may both be followed by -ing and to infinitive complement clauses. Both types of attempt construction are Forward-looking, as is the ‘try to infinitive’ construction. The ‘try -ing’ construction, on the other hand, is either Same-time or Contemplation (there is one Forward-looking exception in the database: this will be discussed below). For ease of comparison it seems best to discuss all these constructions together. And, as some 80% of the tokens of ‘try -ing’ are Same-time, it has been decided to include them in this chapter. ‘Attempt to infinitive’ and ‘try to infinitive’ are among the most frequent of all non-finite constructions and are both much more common than their -ing counterparts (see Tables 7 and 13 in Appendix 2 for projected totals). They also resemble one another closely in the matter of Tense, Aspect and Modality (TAM) relations except in one respect, ‘try to infinitive’ occurring more frequently in the progressive (19% of all tokens of ‘try to infinitive’ are either past or present progressive as opposed to 7% of tokens of ‘attempt to infinitive’). The two matrix verbs are also commonly perceived to be quite similar in meaning. Most dictionaries define one of them in terms of the other, and they all agree that the situation ‘attempted’ is likely to be difficult. One measure of difficulty is frequency of achievement and a close study of the downloaded tokens reveals that in 18% of all occurrences the goal of the subject was definitely not achieved in the case of ‘attempt to infinitive’, as opposed to 10% of all occurrences in the case of ‘try to infinitive’. (368) and (369) illustrate non-realisation of the complement situation in both constructions. (368) Between 1893 and 1896 Fridtjof Nansen attempted to reach the North

Pole by deliberately entombing his ship, Fram, in the ice near Ostrova Novo Sibirskiye (the New Siberian Islands). (A6T 1014)

(369) I’m very depressed and have tried to commit suicide several times. (CBA 1537) Another difference between the two constructions pertains to the person and animacy of the subject. As may be seen in Table 5.1 ‘try to infinitive’ occurs frequently with first- and second-person subjects and almost exclusively with animate subjects, in contrast with ‘attempt to infinitive’ on both counts. (370) and (371) are examples of the two constructions differing in terms of TAM relations and person and animacy of the subject. (370) The project therefore attempts to devise a more general descriptive

framework for the intonation of the languages investigated and to establish a general theory of intonation structure which takes account of the different role of intonation in languages of different types. (HJ1 1812)

Same-time -ing versus Forward-looking to infinitive 157

(371) I was trying to find the place we met her yesterday morning when I heard you shouting and the sounds of ambush so I rode to investigate.’

(H90 1437)

Table 5.1: Person and animacy of subjects of attempt and try in to infinitive constructions with horizontal percentages

1st person 2nd person 3rd person

animate 3rd person inanimate

Total

attempt

51 6.4%

16 2.0%

622 77.9%

109 13.7%

798

try 143 21.0%

64 9.4%

462 67.8%

12 1.8%

681

(370) contains an inanimate subject and is in the simple present rather than the progressive. It is a good example of the sort of context in which a speaker would be less likely to use ‘try to infinitive’. (371), on the other hand, encodes a situation where the choice of ‘try to infinitive’ is influenced by the subject’s encoding his or her own experience, and where this experience is profiled as ongoing. However, these are mere tendencies, rather than absolutes. In fact, both to infinitive constructions are mostly employed with third-person animate subjects. Turning now to -ing complements, the ‘try -ing’ construction proto-typically encodes the adoption of some means, the undertaking of some activity, in an effort to reach some goal. A distinction that is often made in the literature is between making an effort, and ‘putting to the test’. One scholar who formulated the distinction in these terms was Wood. He wrote “When try is followed by the gerund it means experiment or put to the test, as in Try eating more slowly (Advice to one suffering from indigestion after meals); but when it is followed by the infinitive it means attempt e.g. Try to eat more slowly (an instruction to a child who is gobbling its food)” (Wood 1956: 15). Dirven expresses the same distinction: “The verb try means either ‘to make an effort’, and then takes a to-infinitive, or else ‘to do an experiment’ and then takes a gerund, since in this meaning the verb try is a verb of mental activity” (Dirven 1989: 129). In a similar vein, Verspoor (1998) writes: “The difference between the -ing and to infinitive structures after verbs such as try might also be explained by the fact that the -ing complement expresses that try is an intention in action, whereas try when followed by a to infinitive structure implies (un)willingness and prior (non)intention” (Verspoor 1998: 523). All these explanations make the point that, unlike the ‘try to infinitive’ construction, whose understood goal is in the projected future, in the ‘try -ing’ construction the complement situation is itself simultaneous with the expenditure of effort encoded by the matrix verb. ‘Try -ing’ is, in other words, a Same-time construction. The ultimate goal of the


158

subject engaged in the action of trying may either be completely explicit, as in (372), implied in the context, as is ‘forgetting’ in (373), or left up to the hearers to deduce on the basis of the general context or their general world knowledge, as in (374). (372) Try drinking more slowly so you can skip some rounds. (K3A 188) (373) I tried working myself into the ground, but I could be totally exhausted

and still remember. (H8F 3155) (374) The company has tried going green at thirty of its stores, but it claims

customers can’t tell the difference between environmentally friendly produce and that which has been sprayed with chemicals. (KRT 4832)

In none of these examples does the -ing clause encode a targeted alternative in the projected future. It is rather a means or an instrument employed by the subjects to assist them in traversing the paths towards their expressed or unexpressed goals. This prototypical sense of ‘try -ing’ accounts for 80% of the examples downloaded from the BNC. There remain, however, five examples that cannot be accounted for using this schema. Four of these are rhetorical imperatives, of which (375) and (376) may serve as examples. (375) Try selling nappies and dogfood to someone who lives alone and never

intends to have a baby or a dog. (ABK 2384) (376) Try telling that to Niall - a practised afternoon riser back home - at

breakfast after he’s been assaulted by crack-of-dawn alarm calls. (CAE 1589)

At first sight it might appear that the selling of nappies in (375) and the telling of that in (376) are to be interpreted as the goals of efforts the subjects are directed to expend. But this would be to ignore the point that these activities are considered impossible by the speaker. (375) and (376) are, in fact, instantiations of Contemplation rather than Same-time or Forward-looking constructions. The understood goal of the effort is to ascertain the lack of feasibility of attaining it. In other words:

It is not every example of imperative ‘try -ing’ which has this sense: in (372) it has its prototypical sense. Imperative ‘try -ing’ must therefore be considered a polysemous construction, the exact meaning of which in any particular context has to be deduced by the hearer on the basis of the tone of the speaker, accompanied by a subjective judgement of the likelihood of the realisation of the complement situation.

try doing x = imagine doing x to see if it is ‘do-able’: it isn’t!


Among the downloaded tokens of ‘try -ing’ there was one in which it is used in the prototypical Forward-looking sense of ‘try to infinitive’. This is (377). There was also just one token of ‘attempt -ing’, cited as (378). (377) The city had tried ameliorating its appalling traffic congestion by

restricting entry to the city: cars whose registration plates ended in an even number were permitted to enter one day and those in an odd number the next. (BNS 371)

(378) And she couldn’t do anything about Castelfonte, she couldn’t even get them to attempt making it profitable - she never understood about the land, that wasn’t in her blood. (F9R 302)

There can be no doubt that restricting entry is the means employed by the subject in (377) to reach its goal of ameliorating its appalling traffic congestion. Why the speaker in this case should have chosen to employ the -ing rather than the standard to infinitive complement form must remain a mystery. Wherever a language contains two similar constructions (similar in this case in that they both contain the same matrix verb and a non-finite clausal complement) with distinct senses, it is always possible that some speakers will get their grammatical wires crossed, so to speak, thereby imbuing one construction with the sense of another, closely related construction. This would appear to have been the case here. Moreover, the fact that -ing clauses also occur in Forward-looking constructions with other classes of matrix verb may have facilitated this act of wire-crossing. As for (378), we might be tempted to set it aside as an example of idiosyncratic usage were it not for the fact that Duffley and Tremblay (1994) record seven instances of this construction in their material.10 There is no doubt that the -ing clause in (378) is profiled as the goal of the subject’s effort. Duffley and Tremblay maintain that, unlike the more polysemous try, the verb attempt always encodes the expenditure of effort, thus constraining all objects to be interpreted as goals, whatever their syntactic form. They write: “The quasi-synonymity of attempt to and attempt + – ing is due to the fact that the notion of effort aimed at attaining a goal is inherent in the meaning of attempt, whereas this is not the case with the verb try” (Duffley and Tremblay 1994: 574). This explanation seems perfectly plausible. It does not, however, help to explain the reasons for choosing an -ing complement in certain contexts. Rohdenburg (1995) points out that the -ing complement form was not uncommon after attempt in the eighteenth century. He writes: “In former centuries, the -ing construction was not yet as specialized semantically as it is today. Accordingly, the -ing form was used in many cases simply as a convenient means to avoid a clash of two successive to-infinitives. For instance, the use of -ing forms associated with attempt in the 18th century is largely due to this avoidance strategy” (Rohdenburg 1995: 381).

10 Conrad (1982: 175) mentions in a footnote that he has never come across the construction.


160

Duffley and Tremblay only cite one example of ‘attempt -ing’, but one may find it interesting that this example resembles (378) in that the matrix verb is itself a to infinitive. If this form of matrix verb is a motivating factor in the choice of the -ing complement form, it cannot be a very strong one, as to infinitive complements are very common in this position. (379) and (380) are just two of the dozens of tokens among those downloaded from the BNC. (379) As you talk, you’ll find she’ll begin to understand the patter, until

eventually she begins to attempt to put on her clothes all by herself. (G2T 585) (380) Subjects were asked to attempt to remember the 40 junctions for which

they had given ratings. (HPM 730)

It is impossible to discover just why the speaker in (378) chose what is very much a minority option. All we can say is that the choice may have been influenced both by the fact that the -ing form of complement is employed in many other Forward-looking constructions and by the fact that there is no Same-time ‘attempt -ing’ construction to motivate against its employment. There is a third Effort verb, help, that may be followed by a (bare or) to infinitive and an -ing clause. The two forms of infinitive complement will be contrasted in our discussion of Forward-looking bare and to infinitive constructions in section 6.2.2, p. 204. ‘Help -ing’, however, is a Same-time construction. In its case the matrix verb is always modified by the negated modal auxiliary cannot/could not, as in (381) and (382). An instance of the ‘help to infinitive’ construction with ‘can’ (it does not occur with ‘cannot’ in the BNC) is cited as (383), and of the ‘help but bare infinitive’ construction as (384). (381) Life as seen from the platform at Paddington is most varied and one

cannot help thinking that an official who spends most of his time there must gain a vast knowledge of human nature... (AR0 161)

(382) Absorbed as he was by the sight of a family in grief, a new phenomenon for him, he could not help observing, with some awe, the large number of military men present to testify to their friendship with the dead man. (CDN 918)

(383) This is a great worry to us and we believe we can help to prevent these losses by giving them something to distract them from smoking like liver, bacon, eggs, or steak and onions. (J56 988)

(384) As I eat I cannot help but notice the Frenchman attending to the pig; he has now cut the animal’s throat and has set alight to the straw, the flames and smoke engulfing the unfortunate pig. (A61 2230)

(381) and (382) are both Same-time Applied Attitude constructions, the ‘not helping’ and the thinking/observing happening at the exact same time. The ‘cannot help but bare infinitive’ construction in (384) is also a Same-time construction, and one very close in meaning to the ‘cannot help -ing’


construction, the only difference being the fact that the bare infinitive complement is profiled as a unitary whole. The ‘help to infinitive’ construction is Forward-looking, the prevention in (383) being the target of the help. The fact that this latter construction does not occur at all with cannot/could not may be related to the degree of entrenchment of ‘cannot/could not help’ with the Same-time -ing and ‘but bare infinitive’ constructions. This is, however, impossible to prove or disprove. There is one other verb, endure, that occurs in Same-time -ing and Forward-looking to infinitive constructions. They are both classified as Applied Attitude constructions. The Same-time construction, exemplified by (385) always entails the realisation of the complement situation. The Forward-looking construction, exemplified by (386), always implies its non-realisation. (385) ‘Just as you’re going to have to endure being separated from Jeff every

day - at least for a few hours, while we do our English lessons.’ (JXT 862) (386) But Carteret’s wife, who frequented health spas, could not endure to live

with him or he with her: there were no children. (CD2 1061)

There are three tokens of ‘endure to infinitive’ in the downloaded material, all modalised negatives, with the complement situation representing a goal that the subject does not wish to realise: it is a targeted alternative to be avoided at all costs. The -ing complements of endure are profiled as actually occurring in real time. In (385), for example, the condition of being separated is something the subject will have to put up with while it lasts. In (386) the condition of living with her spouse is a possible alternative in the projected future that the subject avoids at all costs. If one were to substitute ‘could not endure living’ for could not endure to live in (386), the meaning would be that the subject actually was living with him and found this situation unbearable. In this section we have seen that there are three matrix verbs that occur in both Same-time -ing and Forward-looking to infinitive constructions. These are try, help and endure. In each case the -ing constructions profile a situation as actually pertaining at the same time as the matrix verb. All the to infinitive constructions, on the other hand, point to a targeted alternative in the projected future, one which the subject is desirous of attaining in the try and help constructions, desirous of avoiding in the case of endure.

5.5 Same-time -ing versus General to infinitive

There are nine matrix verbs in the present study that occur in Same-time -ing and General to infinitive constructions. Eight of these are Attitude verbs, adore, bear, hate, like, loathe, love, prefer and stand. The ninth is the Aspect verb continue. Some of these constructions are very common indeed. Both ‘like to infinitive’ and ‘continue to infinitive’ figure among the ten most frequent to infinitive


162

complement constructions, while their two -ing counterparts are both among the top twelve (see Tables 13 and 14 in Appendix 2 for details). In section 5.5.1 we look at matrix verbs that encode a positive attitude on the part of the subject to the complement situation, in 5.5.2 at constructions that encode a negative attitude, and in 5.5.3 at the comparative matrix verb prefer. Section 5.5.4 is devoted to the two continue constructions.

5.5.1 Positive Attitude constructions

Three of the eight Attitude matrix verbs that occur in both Same-time and General constructions encode a positive attitude on the part of the subject. These are adore, like and love. As we have seen in Chapter 3 there is no consensus in the literature about how to specify the meaning of utterances that we have termed General constructions. Duffley (1992, 2006), for instance, ascribes to them a ‘future’ sense. Huddleston and Pullum (2002) maintain that they evoke the idea of (repeated) change. De Smet and Cuyckens (2005), writing about like and love, argue that these encode a network of meanings including habituality and enjoyment. Carter and McCarthy (2006) write:

Hate, like, love and prefer can be followed by either -ing or a to-infinitive. The difference in meaning is often not great, but -ing emphasises the action or event in itself, while the infinitive places the emphasis more on the results of the action or event. (Carter and McCarthy 2006: 515)

The hypothesis, advanced in section 2.3, that the to infinitive is used in these constructions to encode general validity predications, implies that, contrary to what is maintained by Carter and McCarthy, there is in fact a rather clear difference in meaning between the -ing and to infinitive constructions, although this difference may not be relevant in all contexts of actual usage. Let us begin by contrasting the two same-subject constructions containing like. ‘Like -ing’, exemplified by (213) – (215) on p. 106, repeated here as (387) – (389), portrays a situation where the subject gets an amount of pleasure out of the process encoded in the -ing clause as it is unfolding. (387) I really like doing things together. (A74 318) (388) I’m with you, for a start, and I happen to like being with you. (JYA 2489) (389) A letter from your Aunt Emily told us how much she likes having you and

how accomplished you are becoming. (H8X 1203)

In (387) – (389) the pleasure afforded by doing things together, being with you and having you is in each case simultaneous with the liking. In the case of both activities, as in (387), and states, as in (388), the perspective is mid-interval. Moreover, unlike the ‘like to infinitive’ construction which will be discussed below, there is no doubt in each case that the emotion expressed is one of pleasure, pure and simple.

Positive Attitude constructions 163

Some half of the matrix verbs in the ‘like -ing’ construction in the data are negated. One may wonder whether these sentences, exemplified by (390) – (391), share with their positive counterparts the element of simultaneity. (390) ‘I don’t like taking your money, Ruth,’ he protested. (CB5 875) (391) that’s why I don’t like going to the theatre with Derek, he, he doesn’t clap

or anything! (KST 4337)

The extended context of (390) shows that the subject is in the process of taking money and it is this money-taking that he simultaneously dislikes. Similarly, in (391) it is the actual action of going to the theatre that calls forth feelings of dislike. The mid-interval property of the Same-time construction does not, however, mean that the activity/state in the complement clause has to be unfolding at the exact same time as the liking. It is sufficient that it has happened in the past and that the subject is apprehensive of its happening again. Verspoor makes a similar point: “These constructions may imply that the situation is actually taking place […]. However, these constructions may also express a much more general attitude towards such a situation, but it does always imply that such a situation has taken place” (Verspoor 1997: 447). I think it preferable to reserve the term “general attitude” for the sort of higher-order general validity predications encoded by the to infinitive construction. As for these -ing constructions, I would be more inclined to say that they encode invariant particular reactions to recurring situations rather than general ones. In other words it is always a particular complement situation that prompts the attitude encoded by the matrix verb in the -ing construction. The main point is that the situation is profiled as one that may well happen again and that it is the actual unfolding of the activity/state that is disliked. Closely resembling ‘n’t like’ in this respect, are matrix verbs encoding a negative attitude on the part of the subject, such as dislike, detest and abhor. 68% of the downloaded tokens of ‘like to infinitive’ are modalised as in (392) and (393), with the vast majority of the modals being forms of would. (392) Ultimately, she would like to work in film, behind the camera. (CGB 786) (393) Are there any changes you would like to make now? (CCN 1787) In (392) the realisation of the wishes of the subject would lead to her working in the film industry. If this occurs, and, of course, nothing is certain where the future is concerned, it will be after the time of the utterance. As the Cambridge Grammar puts it, the ‘would like to-infinitive’ construction “projects into the future and resembles a verb of wanting” (Huddleston and Pullum 2002: 1242). It is thus a Forward-looking, rather than a General construction. Over 90% of the non-modalised ‘like to infinitive’ tokens encode general validity predications. The remainder, to which we will return at the end of this section, are Forward-looking. The general nature of the predication in the assertive non-modalised ‘like


164

to infinitive’ construction is clearly apparent in examples such as (51), repeated here as (394), (395) and (396). (394) Martin occasionally, and fancifully, likes to picture himself as a daredevil

cuester, a flashing Jimmy White to my grinding Cliff Thorburn. (ECU 1513) (395) Like many golfers, and professional golfers are particularly prone to this, I

liked to alter my clubs occasionally - to change the lie of a wedge by a degree or so, or thicken the grip of a putter, or change the swing weight of a driver. (CS4 1903)

(396) ‘I like to buy British, but the candle colours became uneven, delivery was unreliable and they were expensive. (BMB 1619)

What exactly is the meaning of the ‘like to infinitive’ construction as it is used in (394) – (396)? As a first generalisation we may suggest that if one likes to do something one normally does it whenever a suitable opportunity arises, because it affords one a measure of satisfaction. As Conrad writes of his example “Soames liked to talk during dinner”, it “says something in general about Soames’s views about talking during dinner, he would talk during dinner because he thought that it was a good idea to do so. It does not mention his reaction on any particular occasion where he talked during dinner, it does not even predict his reaction when he is talking during dinner” (Conrad 1982: 165). In utterances like (394) and (395) the complement situation is usually realised to the satisfaction of the subject: there is no question of the success or failure of the picturing and altering. The fact that the content of the complement clause may not always be realised does not lessen the degree of liking. In so far as the realisation of them lies within the remit of the subject we may normally take it for granted that realised they will be, whenever a suitable opportunity presents itself. However the possibility of non-realisation is not totally excluded. Thus in sentence (396) the speaker ends up not buying British. In each case the to infinitive clause encodes a targeted alternative, being the situation the subject would like to realise on all suitable occasions. While on the subject of the targeted alternative hypothesis, we may note that Wood employs the word “choice” in his description of the ‘like to infinitive’ construction. “When like and (do) not like take the gerund they suggest enjoyment or repugnance respectively […]. But with the infinitive it suggests desire, preference, or choice, and in the negative reluctance” (Wood 1956: 15). The possibility of choosing entails the existence of several alternatives. We may also note that the fact that the to infinitive form implies the existence of several alternatives serves to predict the results achieved by Kempson & Quirk (1971: 554, also referred to in Quirk 1995: 169), in their elicitation test in which their subjects were required to insert either ‘to get’ or ‘getting’ into the frames “I like ______ up as soon as the alarm rings” and “I like _______ up when the weather is warm”. In the experiment 43 of 50 respondents inserted the to infinitive form in the first sentence and the -ing form in the second. This tallies with the impression


that there may exist alternatives in the first case, but not the second - while some of us may be able to afford the luxury of spending an extra five minutes in bed on occasion, most of us cannot choose to hibernate through whole seasons! The definition of the ‘like to infinitive’ construction which was proposed above involves the realisation of the complement situation by the subject at every suitable opportunity, i.e. whenever what Langacker (1997: 205) refers to as enabling conditions are satisfied. The occurrence of such a suitable opportunity may be more or less frequent. It is often signalled by an adverbial such as occasionally in (394) and (395), or the underlined elements (57) – (61) in section 2.3, p. 36, and in (397) and (398). (397) And when the flat’s not in use for entertaining clients I like to make it

available to all my employees.’ (H8S 1333) (398) ‘I like to go to confession straight after I’ve been to the hairdresser,’ said

Lili. (G06 932) In examples (397) and (398) the likely occasions of realisation are encoded explicitly. There are also many cases in which these occasions are implicit. In such cases our rule of thumb as language users, our working hypothesis, is that they are likely to be realised on appropriate occasions. As was pointed out in section 2.3 we must make use of the context or of our world knowledge to identify the sort of occasion that is meant in any particular instance. (399) is a case in point. (399) Frau Schmidt liked to do things properly and was an incurable romantic.

(J19 808) Our interpretation of what constitute suitable occasions is naturally informed by our encyclopaedic knowledge of the event types in question. Even without access to the context, we know that the situation in (399) pertains whenever Frau Schmidt is called upon to do something. The point is that, unlike the construction with -ing complements exemplified in (387) – (389), none of these tokens encode the actual realisation of the complement situations. Rather, they all encode a higher order predication of the likelihood of their being realised in the appropriate circumstances. The ‘like to infinitive’ construction involves the exercise of a choice between two or more possible alternative situations and the motive governing this choice is the effort to achieve some sort of satisfaction. This tallies with Bladon’s (1968) assertion that ‘like to infinitive’ has a semantic component of “desire fulfilled” and Wierzbicka‘s (1988) assertion that it always contains an element of “wanting”. Bladon (1968: 212) actually maintains that the ‘like to infinitive’ construction may be used in one of two distinct senses, to encode either what he calls “desire fulfilled” or “actual occasional enjoyment”. He exemplifies these two senses by means of the following two sentences, which differ in intonation pattern: (p) She liked to have breakfast in \ bed and (q) She \ liked to have


166

breakfast in bed. According to Bladon, (p) encodes “desire fulfilled” and (q) “actual enjoyment”. According to the approach adopted in the present study both these sentences imply the existence of alternatives, the intonation focussing on the locus of the implicit alternative. In (p) the implicit alternative is ‘eating someplace else’, in (q) it is ‘didn’t like’. Unlike the ‘like -ing’ construction which always encodes (simultaneous) pleasure felt by the subject, the nature of the satisfaction aimed at by the subject in the ‘like to infinitive’ construction may vary greatly. It may be moral in nature: it may be the satisfaction afforded by a good conscience. It may be sensual. In the latter case, but not necessarily the former, it is likely that one would also like doing it. Sentences (400) – (404) contain some examples of the ‘like to infinitive’ construction which vary in the degree and type of satisfaction that the complement situation affords the subject. (400) Similarly, since most men like to be seen to do the paying it is a tactful

precaution if, at the start of the evening out, she gives him the theatre tickets ‘to take care of’... (HXG 357)

(401) Though it would certainly be in keeping with his media image if he were to use Seawitch for what the tabloids liked to call a ‘love nest’.

(H7W 2739) (402) The wild narcissi and beautiful peonies, which liked to grow among grass

or in the shade, hidden among the roots of the trees, were over. (G3B 646) (403) Choose long-lasting and highly water resistant products if they like to play

in or around water. (BN7 290) (404) ‘With Joan Armatrading, we did 2-3000 seaters, and that just worked

fantastically, especially as I like to communicate with the audience, and people these days aren’t used to that. (C9M 2205)

In (400) the subjects’ primary concern is with the preservation of their self-image. In (401) the use of a euphemism satisfies a real or apparent squeamishness on the part of the subject about calling a spade a spade. In (402), one of the few occurrences in the BNC with a non-animate subject, what is satisfied is the need for a suitable growth environment. In (403) and (404) the satisfaction is of the sort that affords the subjects pleasure and enjoyment. In contexts like (403) and (404) we may conclude that the subjects also ‘like -ing’. We will now proceed to contrast the two constructions. Before doing so, we cite some more typical examples of the ‘like to infinitive’ construction. (405) ‘Just as well you’re away before father comes in; he likes to gossip.’

(GW3 2503) (406) ‘I like to go round the park in Bognor and photograph squirrels’ (CH6 4783)


(407) Their parents like to leave toys by the graves, but on several occasions they’ve returned the next day to find them missing. (K1K 3968)

(408) Some people like to build a little pond-side ‘house’ to contain their filter - when others prefer to disguise it with plants and shrubs. (C95 3078)

We have already noted that the Same-time ‘like -ing’ construction only allows for the expression of one sort of satisfaction, that of pleasure felt (simultaneously) by the subject. Insofar as the satisfaction afforded by the exercise of choice with a ‘like to infinitive’ construction may involve the pleasure faculty, as in (403), (404), (405) and (406), we may presume that these entail the ‘like -ing’ construction: if one likes to gossip one also presumably likes gossiping. It is certainly difficult to imagine that the activities described in (405) and (406) would be indulged in for reasons other than pleasure. This does not mean that a ‘like -ing’ clause would have the same meaning as a ‘like to infinitive’ clause. To underline the difference, I will paraphrase (406) in (409) and then construct a parallel sentence in (410) with a paraphrase in (411). (409) Whenever I get the chance to do so, I choose to go round the park in

Bognor and to photograph squirrels there, because these activities afford me a measure of satisfaction (which happens to take the form of pleasure).

(410) I like going round in the park in Bognor photographing squirrels. (411) I sometimes go round the park in Bognor photographing squirrels.

Whenever I do so, I get pleasure from this activity. (409) and (411) are designed to bring out the difference in meaning between (406) and (410). However, the point has already been made, and it bears repeating, that this difference may be completely irrelevant to the speaker and addressee in the actual context of utterance. The two utterances share the meaning components ‘I photograph squirrels in the park. I like that.’ If achieving a mutual understanding of these components is the main concern of the discourse participants, then it may matter little to them which of the two alternatives the speaker chooses. At one end of the satisfaction spectrum are utterances like (403) – (406), in which the satisfaction takes the form of pleasure, pure and simple. At the other end there are many activities which we may choose to do and which give us satisfaction of one sort or another, but which could never be described as pleasurable. We may encode these in ‘like to infinitive’ clauses like (407) and (408) from which we cannot infer corresponding ‘like -ing’ clauses. It is difficult to imagine either of the activities in (407) and (408) as pleasurable in themselves. While there are many forms of satisfaction that do not necessarily imply pleasure, the opposite is not the case. The feeling of pleasure normally affords us satisfaction, and ‘like -ing’ clauses normally imply their ‘like to infinitive’ counterparts, as in (412) and (413).


168

(412) I wish I had some bread - I like feeding the ducks. (A74 2399) (413) People like feeling pity for people, it makes them feel lucky. (C86 2634) (414) Yes but you can still revolt against something you like doing. (KGN 889:

omitted from World edition) Examples of ‘like -ing’ that do not imply ‘like to infinitive’ are more difficult to come by, although easy enough to imagine. Indeed it is possible to imagine many things we might do that would give us pleasure but that run counter to the satisfaction of other attitudes we might have, eating fatty foods, for instance, or smoking cigarettes or driving above the speed limit. The fact remains that we do not seem to focus very much on expressing these pleasures, at least not using this construction, possibly because of the very conflict involved in exercising them: indeed, the one clear example of a sentence like this which I have come across in the BNC, (414), actually points to such a conflict. I will round off this discussion of like by returning to the use of the non-modalised to infinitive construction to encode specific predications in the projected future. Some 8% of the tokens in the random data sample are of this sort. There are two contexts in which ‘like to infinitive’ may be used in this way. They are with negated matrix verbs, as in (415) and in if-suggestions, as in (416). (415) I don’t like to say this but there’s a society in Glasgow buying tickets for

the children for pantomime. (J8B 1523) (416) And if you like to check yours now just okay. (KLW 1316) The use of the non-modalised ‘like to infinitive’ constructions to encode specific rather than general validity predications was much more common in Late Modern than in present-day English. There has been a significant decrease in the use of the form in if-clauses over the last two hundred years (see Egan: in press). Today if-clauses with non-modalised like used to encode suggestions are in direct competition with the modalised variety, as may be seen in (417) – (422) (417) So if you like to write that one down. (FMH 679) (418) Right so if you’d like to label that triangle. (FMJ 185) (419) If you like to look back at the cathedral office in an hour, I’ll see it’s ready

for you. (HA2 2181) (420) If you’d like to come back to my office we can discuss it there,’ he said,

looking at Fairham. (G01 966) (421) If you like to get on I’ll fetch my horse. (A0R 2245) (422) If you’d like to go into the next room, please. (F77 523) It is difficult to discern any substantive semantic or pragmatic differences between the pairs of tokens with modalised and non-modalised matrix verbs in (417) – (422). The development of these two forms over the last two hundred years appears to be an example of drift in the classic sense of Sapir (1921), with the ‘would like to’ form gradually, but inexorably, replacing its non-modalised


‘like to’ counterpart. The construction in (415) with the negated matrix verb is proving rather more resilient in present-day English. Its continued existence may well be bolstered by the existence of a similar use of the non-modalised ‘hate to’ construction, to be described in the next section. The other two positive attitude matrix verbs, adore and love, differ from like, as a glance at Table 9 in Appendix 2 will show, in being more likely to be followed by an -ing complement. It appears that the stronger the emotion, the more deeply felt the attitude expressed by the matrix verb, the greater the likelihood of its being followed by -ing. The two -ing constructions always signal that the subject gets a great deal of pleasure out of the process encoded in the -ing clause as it is unfolding. The process is mid-interval. (423) – (427) are typical examples of the two constructions. (423) She adored working on the million pound project which was predicted

would rival Paul Hogan’s Crocodile Dundee smashes. (ADR 1388) (424) I adore eating out and always choose something different from my normal

home-made menus. (C9Y 2107) (425) John adored being with his grandchildren. (B34 771) (426) She loved listening to the well-known tunes, and gave herself up to the

beauty of the music. (GW8 630) (427) She met me in London wearing a wide-brimmed hat, khaki shorts and a

loose white top; she says she loves showing off, feels her legs are her best feature and that short skirts suit her. (ADG 1181)

‘Love -ing’ and ‘adore -ing’ resemble ‘like -ing’ in that the activity indulged in or state experienced is a source of (in this case considerable) pleasure as it is unfolding. The person who adored working on the project was working on the project and the person who adores being with grandchildren is with them. Similarly in (426) the subject was listening to the music, something which was the source of simultaneous pleasure for her. Examples (424) and (427) are somewhat different as they refer to serial activities. Someone who adores eating out is not necessarily eating out at the exact moment they express their adoration. What is however entailed is that they have eaten out in the past and expect to do so in the future. Simultaneity, in this broad sense, is implied by both constructions, and the presumption is that the content of the complement clause is realised. One loves doing something while one is doing it, and in order to love it one must be in the middle of doing it, either in the middle of one particular instance as in (425) and (426) or, in the middle of a series, as in (424) and (427). ‘Love to infinitive’ and ‘adore to infinitive’ resemble ‘like to infinitive’ in that approximately two thirds of all matrix verbs are modalised (with a form of would), resulting in Forward-looking rather than General constructions. As for the non-modalised constructions, they differ from like in that they invariably encode general predications. Exemplified by (428) – (432), they resemble ‘like to infinitive’ in that they profile a tendency on the part of the subject to realise a situation whenever a suitable opportunity arises. This tendency is, however,


170

stronger, the subject being likely to grab the opportunity with both hands, as it were. (428) Absolutely Yes. I mean it’s lovely because he er he adores to get into does

Edmund and that means that he and George go about this lovely big er grounds that they’ve Yes. got and Edmund does lots of grass cutting and er tree cutting and all manner of things with Mm. George. (KC0 4993)

(429) When we are hungry we love to eat bread. (HS7 263) (430) But serious or lighthearted Central News always brought you the offbeat,

like the catfish who loved to be fed by hand. (K1B 817) (431) Benny loved to hear stories of the orphanage. (CCM 24)

It happens sometimes that the occasions for the realisation of the complement situation are specified, as in (429). More common is the intimation that they will be realised whenever possible. Thus in (428) Edmund got into the gardens whenever he could. The fact that ‘love to infinitive’ signals a feeling of pleasure rather than sundry sorts of satisfaction, would lead us to expect that these constructions would entail their ‘love -ing’ counterparts, that, for example, (431) entails (432). (432) Benny loved hearing stories of the orphanage. However, as we saw in the case of like, the fact that one construction entails the other does not mean that they are synonymous. The following two paraphrases of (431) and (432) are designed to bring out the difference in meaning between them. (433) Whenever Benny got the chance to listen to stories about the orphanage he

grabbed it eagerly because it gave him great pleasure. (434) Listening to stories about the orphanage gave Benny great pleasure. The fact that (431) entails (432) and almost, if not completely, vice-versa - (432) certainly very strongly implies (431) - means that the choice between the two may not be of any great consequence to the speaker. One factor which may influence his or her choice is whether the complement situation is completely new to the discourse. This is related to a tendency to make general validity predications about new situations on the one hand, and to adopt a same-time perspective on already mentioned situations on the other. Thus the situations in (423), (424) and (426) have already been mentioned. This is, however, not the case with (425) and (427). The complement situations are new to the discourse in the to infinitive constructions (428) and (430), but not (429) and (431). The frequency of exceptions in the corpus is by no means as great as one might get the impression of from these examples, which have been chosen to illustrate the possibility of both types of construction occurring in both types of discourse context. Nevertheless, the very existence of such exceptions shows that we are


here faced with tendencies rather than absolutes. In the end the choice in a concrete speech situation between saying b, and saying a and implying b, is a matter for the individual speaker to make. To sum up this discussion of Same-time -ing and General to infinitive constructions with positive attitude verbs, we may conclude that while there are clear differences in meaning between the two types of construction, these differences do not always have practical consequences for the language user in actual contexts of communication. This overlapping of possible contexts of use is more extensive in the case of the more powerful emotions encoded by love and adore than the more luke-warm like.

5.5.2 Negative Attitude constructions

The two matrix verbs hate and loathe both occur in to infinitive and -ing constructions. Hate is often conceived of as the semantic opposite of love. As far as taking non-finite complement clauses goes, however, it differs from love in three respects. In the first place hate is much more likely to occur with an -ing complement clause. Secondly, ‘hate to infinitive’ is much less likely to be modalised, and, thirdly, non-modalised ‘hate to infinitive’ resembles like more than love in so far as it can be used to encode a specific predication in the projected future. In fact it is so used more frequently than like. The exact figures for each of these three types of construction (based on the 1,000 downloaded tokens of each verb) are given in Table 5.2.

Table 5.2: Forms of non-finite complement with modalised and non-modalised love and hate with vertical percentages

love hate modalised + to infinitive (Forward-looking) 60 48% 33 17% non-modalised + to infinitive (Forward-looking) 0 26 14% non-modalised + to infinitive (General) 33 26% 43 23% -ing 32 26% 88 46% total no. of tokens 125 190

Why should there be such a difference between the complements of love and hate? It may well be the case that if we hate to do things we just avoid doing them, and feel no need to remark on our distaste for them. If, on the other hand, we are, or have been, in the middle of an activity we detest, we may well feel the need to broadcast the fact. (435) – (439) are typical tokens of ‘hate -ing’.


172

(435) I smoked for twenty seven years, and erm I’m getting on to a year now since I stopped smoking and I feel a lot calmer, I hated being a victim to cigarettes! (FLM 104)

(436) He hated leaving her there but he descended the ladder because there was nothing else he could do. (CN3 131)

(437) And I hated turning out even more since my marriage. (G35 441) (438) He hated butchering cattle. (J19 1188) (439) ‘Well, I hate dreaming about them too, so I don’t. (FRH 235) In (435) – (439) the situation in the complement clause is simultaneous with the matrix verb; the subject hates this situation as it is unfolding. Thus in (435) the subject hated the addiction while it lasted. The subject in these sentences does not appear to have any choice in whether the process in the complement clause unfolds or not. Indeed, this absence of any viable alternative is stated quite explicitly in (436). (439) is paradoxical in this respect: the ‘hate -ing’ sentence implies that the subject does dream about them, but then this proposition is denied in the so clause. The ‘hate to infinitive’ construction might have been an option for the speaker here, given the clear choice the subject appears to have had. But if he or she were to choose the to infinitive construction they would lose the rhetorical effect of the paradoxical about-face. The speaker implies a lack of choice and then exercises the very choice that was implied to be lacking. The ‘hate -ing’ construction resembles the positive attitude -ing constructions both in encoding a mid-interval perspective and in not implying absolute simultaneity between the situation in the complement clause and the matrix verb. It means something like: “Whenever I do x, I feel extreme distaste for it while I am doing it”. In (435) and (436) the subject is in the middle of a single state or action, in (437) and (438) in the middle of a series of activities. In (437), for instance, the subject has experience of having to turn out since his marriage and expects to have to do so again. Every time he does so he feels an intense dislike for the process. Loathe shares with hate this semantic component of extreme dislike. In addition the nature of the dislike has a sort of stomach-turning tinge to it: hate may on occasion be intellectual, loathe is always visceral. ‘Loathe -ing’ is exemplified by (440) – (442). (440) Frederick Bissett, Senior Scientific Officer, loathed being called Fred.

(CLD 482) (441) Otherwise it was like a wrinkled prune, lying beside its mother, who

loathed having her breasts pulled and sucked at for milk, which however came. (CD2 2002)

(442) I mentioned this to no one for two practical reasons: one was that I loathed swimming, and if I pretended that I was still menstruating, signing the little red book every four weeks, I should be able to evade an unpleasant experience for at least one week out of four; the other was that I feared further reprisals might be taken against me. (CEE 923)

Negative Attitude constructions 173

The ‘loathe -ing’ construction resembles the ‘hate -ing’ construction in that it encodes the actual realisation of a situation that evokes extreme distaste in the subject. In (440) it is the actual action of being referred to as Fred, an action that is profiled as actually occurring, that calls forth the simultaneous feeling of revulsion on the part of the subject. Similarly in (441) it is the actually occurring activity of suckling that is disliked by the mother. As in the case of the other Same-time constructions, strict simultaneity is not required. Thus in (442) the subject can loathe swimming even when she is not engaged in that activity. It is sufficient that she has swum, is apprehensive of doing so again, and that every time she does so she hates it. 62% of the non-modalised negative attitude to infinitive constructions resemble the majority of the positive ones in encoding general validity predications. However, whereas the positive ones profile situations which the subject would strive to realise on all appropriate occasions, the negative ones profile situations the subject would strive to avoid on all occasions. (443) – (446) exemplify to infinitive complements with both hate and loathe. (443) She hated to use the word ‘common’, only common people called other

people ‘common’; her mother taught her that. (FRC 1228) (444) He was so soft-hearted, he hated anyone to cry. (FSF 86) (445) Margaret Stanhope hated to disappoint her boss. It almost broke her heart

when she saw that darting look of imitation cross his face as she brought him bad news. It didn’t happen often - when it came to personally assisting, she had few equals - but, when it did, she felt ridiculously guilty. (BP7 787)

(446) ‘You loathe to be categorized. (ECU 3509) In (443) the subject tried to avoid using the word ‘common’. Similarly, in (444), an instantiation of a different-subject construction, the matrix verb subject was of such a disposition that he tried to avoid occasions of people crying. (445) actually stipulates the frequency of occurrence of the situation in question - It didn’t happen often. One complement predicate stands out by its frequency of occurrence in the ‘hate to infinitive’ construction. This is ‘see’ which accounts for 13 of 46 tokens of non-modalised ‘hate to infinitive’. (There are only two tokens of seeing among the 79 -ing complements.) It is exemplified by (447) – (449). (447) Also, I love animals and hate to see them suffer. (CKL 427) (448) Relatives hated to see a sheet over the faces of their loved ones. (JYB 427) (449) ‘And I hate to see you unhappy, and you won’t feel better if you stay in and frowst by the fire all day. (APU 2226)

(447) is clearly a general predication in the manner of (443) – (446). It happens that the subject occasionally sees animals suffer, and this experience he or she


174

would do anything to avoid. The general nature of the predication in (448), on the other hand, resides in the non-specific plurality of the subject. It is generic rather than habitual. That is, it does not encode the case of a single individual being repeatedly exposed to a distressful situation, but of a succession of individuals being exposed to a similar situation. Each individual is, presumably, exposed to it only once. Finally, (449) encodes the realisation of the complement situation on a single occasion. It is therefore very close in meaning to the Same-time -ing construction. What difference there is between the two may be due to the implication of the existence of possible alternatives in (449). I hate to see you unhappy implies that there is a possible alternative - that the person in question does not necessarily have to be unhappy. This implication is made explicit in the conjoined clause. I hate seeing you unhappy would merely assert the existence of the unhappy state and the fact that the subject finds the observance of this unhappiness disturbing. Let us try to further tease out this distinction with the aid of minimal pairs, each of which contains a genuine token containing ‘see’, and paraphrases of these. (450) Jannie instinctively hid her cigarette behind her back when the kitchen

door opened; John hated to see her smoking while she cooked. (G12 2) (451) He hated seeing wartime documentaries; they didn’t seem to affect Anna

in the same way - she seemed to thrill with excitement as the guns flashed and the tanks churned through ruined streets. (AE0 3505)

(452) John hated seeing her smoking while she cooked. (453) He hated to see wartime documentaries; - they didn’t seem to affect Anna

in the same way. (454) (= 450) If she were to smoke while she cooked, which she sometimes did,

John would dislike this intensely: so much so in fact that he might choose to look away or attempt to stop her.

(455) (= 451) Whenever he watched a wartime documentary, he felt extreme discomfort as he was watching it. Anna did not share this feeling of discomfort.

(456) (= 452) She smoked while she cooked, on one particular occasion or a series of occasions. Whenever she was doing this John experienced a severe sense of discomfort.

(457) (= 453) If he were to watch a wartime documentary, which he sometimes did, he would dislike the experience intensely: so much so in fact that he might choose not to do so.

The paraphrases, (454) to (457), are designed to bring out the differences between the pairs of sentences. It is clear from them that (450) entails (452). The same thing cannot be said of (451) and (453). In fact there is something odd about (453). The oddity is a result of the incompatibility of the element of choice inherent in the construction with the to infinitive and our knowledge that we ourselves are normally masters of what we wish to watch/see on television. Note


that the same problem does not arise in the case of (450). We cannot normally decide whether other adults are to be allowed to smoke or not. We may choose to look away, but this will not wipe out the impression we have already received should we have come upon a smoker unawares. We have seen in Table 5.2 that 26 tokens of non-modalised ‘hate to infinitive’ are Forward-looking rather than General. The complement predicates in the case of 19 of these tokens are verbs of communication, such as admit, ask, say and tell, sometimes prefixed by have to. (458) – (461) are typical examples of this construction. (458) Er actually, er much as I hate to admit it, I think you could be right. (KD5 5711) (459) He said, ‘I hate to have to ask, but could you let me have the twenty

pounds? (EV1 65) (460) I hate to say this guys but lets compare the Irish with the English squads

for tomorrow. (J1G 1509) (461) Well I hate to tell you but this is that’s the same smell that Dorothy’s

house has. (KCT 1159) 25 of the 26 tokens of Forward-looking non-modalised ‘hate to’ are first-person, as inn (458) – (461). 24 encode direct speech (one of the two exceptions is cited as (463)). In each case the speaker expresses a disinclination to carry out some course of action, which they then proceed to do. The 7 tokens containing non-communication predicates share this sense, with one exception. (462) – (463) represent the typical sense, (464) the exception. (462) ‘I hate to break your heart, but it’s time I was on my way.’ (JY9 834) (463) I hated to disappoint her, but I had to say I hadn’t. (H0D 1959) (464) She shrugged casually, hating to spoil the impression of intimacy between

herself and Piers by confessing that he had told her nothing. (H8H 2860) In (462) and (463) the speakers go on to carry out the actions for which they express hatred. In (464), on the other hand, the subject behaves as she does in order not to spoil the impression in question. In other words “spoil[ing] the impression of intimacy” is here encoded as a possible situation in the projected future the realisation of which the subject attempts to avoid. In this respect this one token resembles modalised ‘hate to’. There are two other negative attitude matrix verbs that take both to infinite and -ing complements. These are bear and stand. When functioning as attitude verbs both of these are modified by can or could (or, much more seldom, be able to) and occur in non-assertive contexts. As was the case with hate a considerable number of occurrences of ‘cannot bear to infinitive’ are Forward-looking, rather than General. There are, however, a minority of occurrences of the latter type. These are exemplified by (465) – (467). Attitude constructions containing bear and stand with -ing complements are exemplified by (468) – (469).


176

(465) ‘I have broken up with my boyfriend and I can’t bear to let any man touch me. (CBF 294)

(466) ‘Even the most hard-hearted father could hardly bear to have them out of his sight for long.’ (EFV 131)

(467) I can’t stand to see people being cruel to animals it absolutely appals me, it does really, I feel like to take an axe to them (KCY 1467)

(468) ‘Some people can’t bear being closed in,’ said Clarissa. (G1D 3054) (469) ‘I don’t think I could stand being a nurse because I would get too

involved, so I hope one day to join the fire brigade,’ she said. (CF9 154)

The complement clauses in (465) – (467) clearly encode General rather than Forward-looking predications. On no occasion will the subject in (465) consider allowing a man to touch her. Similarly the subjects in (466) and (467) will do their utmost to avoid letting their children out of their sight, or passively observing cruelty to animals, respectively. (468) and (469), on the other hand, encode actuated Same-time situations. In (468) it is the actual experience of being closed in that the subjects find unbearable. And in (469) it is the day-to-day experience of nursing that the subject would find difficult to bear. When the complement situation is stative, as in (466) – (469), the to infinitive constructions entail their -ing counterparts and vice-versa. Thus in (467) the person who cannot stand to see people being cruel to animals occasionally experiences this situation. When this happens we may conclude that he or she cannot stand seeing it. Similarly, in (468) the people who cannot bear being closed in, will presumably also go to any lengths to avoid this state. In that case it is true to say that they cannot stand to be closed in. In a sentence like (465), on the other hand, with a more agentive subject, the to infinitive construction does not entail its -ing counterpart. This is because an -ing clause would imply the actual realisation of the complement situation, something which we would expect a more agentive subject to be able to avoid. The fact that negative attitude to infinitive constructions may, on occasion, entail their -ing counterparts, and vice-versa, may perhaps facilitate the use of the one in place of the other. In section 5.4, p. 159, we came across one instance, (377), of the ‘try -ing’ construction being used in the prototypical Forward-looking sense of the ‘try to infinitive’ construction. A similar point may be made with respect to (470). (470) The last thing I want is to upset him or the family - I can’t bear hurting

people - but I’m harmless; you can tell him that!’ (H94 2638)

In (470) a Same-time construction seems to be used to encode a General predication, a same-time reading being unlikely given the agentive nature of the action of hurting people. If one hates hurting people, one does not have to do it! Examples such as (470) are extremely rare in the corpus and do not warrant our conducting a whole-scale revision of the classification presented in Chapter 2. In their cases we see a construction being used to serve a function not its own, as it


were. It is a bit like using an axe to hammer in a nail. We may occasionally do so, without thereby lessening the inherent chopping properties of the axe. The fact that someone may occasionally use a General construction to encode a Same-time predication does not take away from its basic general sense.

5.5.3 Comparative Attitude constructions

Prefer resembles like and hate rather than love in so far as a non-modalised matrix verb combined with a to infinitive complement may be used to encode a Forward-looking predication. (471) – (474) illustrate the four relevant types of predication. Details of their incidence are given in Table 5.3. (471) The look Vitor shot her said he would prefer to discuss the matter in

private. (JY9 2038) (472) I preferred to accept the fiver.’ (HWN 509) (473) One explanation is that cats prefer to eat small meals on frequent

occasions, rather than gorge on large, infrequent meals. (BMG 363) (474) Not surprisingly, the Emperor preferred rowing quietly on the lake to

scrambling up and down rocks, though sometimes both sovereigns would organize excursions, taking their guests in carriages on more decorous visits to the surrounding countryside. (ANR 635)

Table 5.3: Forms of non-finite complement with modalised and non-modalised prefer with vertical percentages

prefer modalised + to infinitive (Forward-looking) 132 28% non-modalised + to infinitive (Forward-looking) 14 3% non-modalised + to infinitive (General) 295 63% -ing 29 6% total no. of tokens 470

As is the case with positive and negative attitude verbs a general validity predication encoded by prefer, as in (473), normally entails its same-time serial counterpart. We saw in the previous two sections that both the positive and negative attitude verbs, when complemented by a to infinitive clause, imply the exercise of a choice on the part of the matrix verb subject. The targeted situation is preferred (or dispreferred) to one or more alternatives. In the case of prefer the semantics of the matrix verb itself implies such a choice with all types of complement, whether phrasal or clausal. This semantic component of implied choice in the matrix verb adds to the difficulty of distinguishing semantically between general constructions and same-time serial constructions, such as the one in (474). Indeed general validity predications were actually commonly encoded


178

by ‘prefer -ing’ in Late Modern English (see Egan: in press), and there are some relics of this construction in the BNC. (475) and (476), for instance, both contain general validity adverbs in generally and usually. (475) Corral generally prefers buying from commercial galleries rather than

auction houses and she often consults the artists about the works the Museum intends to buy. (EBS 707)

(476) Macho man usually prefers talking to his own sex because it’s safer, and he’s more interested in proving he’s as ‘male’ as his peer group than in entering a relationship which requires an ability to give, to love, to be tender. (H83 939)

(475) and (476) are the only two ‘prefer -ing’ tokens in the whole of the BNC to contain adverbs like generally, normally or usually. These adverbs, incidentally, do not occur in the BNC with quintessential same-time predicates like enjoy or dislike. There are over 20 tokens of these adverbs with ‘prefer to’, two of which are cited as (477) and (478). (477) They had noticed that females of both types generally prefer to mate with

a quadrimaculata male, if given a choice. (GU8 176) (478) The children could take early dinner, but they usually preferred to eat with

us in one of the dining rooms. (AMW 457) So rarely does one encounter tokens such as (475) and (476) in present-day English that I do not consider it necessary to posit a whole class of general validity predications encoded by the -ing complement form. This is especially so given the clear historical trend towards the assumption of the function of encoding general predications by to infinitive complements, also in the case of prefer. Another area in which prefer has come to resemble like, love and hate is in the encoding of predications in the projected future using a modalised form of the matrix verb, although, as may be seen in Table 5.3, one can still encounter tokens where such predications are encoded by a non-modalised form of prefer. (479) – (484) illustrate these two constructions. (479) ‘I shall be at Trebyan for a while.’ ‘I’ll drop you off there, sir.’ ‘No, I

prefer to walk.’ (GWB 2955) (480) ‘Do you want to go and sit down some place?’ Maggie asked. ‘No, I’d

prefer to keep walking.’ (AN7 2262) (481) ‘But I prefer to see you safely to your flat. (JXS 3848) (482) ‘Yes, but I’d prefer to come with you. (GUS 309) (483) ‘So, are you going to climb on board or do you prefer to stay here in the

water? (JXT 1643)

Comparative Attitude constructions 179

(484) ‘We’re fully booked for this evening, sir. Oh, just a moment ... I do have one seat left in the stalls. Or would you prefer to try your luck up in the Gallery?’ (KAT 113)

One may perhaps still discern a greater degree of diffidence on the part of the speaker in the first person expressions in (480) and (482) than in (479) and (481). The modalised variant may perhaps be interpreted as exerting less pressure on the addressee to respect the wishes of the speaker. On the other hand what was originally the more diffident form may be in the process of becoming the unmarked or default mode of expression, as is the case with ‘if you’d like to’ discussed in section 5.5.1, in which case the connotation of diffidence is likely to be reduced in time. Whatever about the tokens with a first person subject, the example with the second person subject in (483) seems rather more stilted, at least to my ears, than the one with the modal in (484). (483) is the only one of eleven tokens of ‘do you prefer to’ in the BNC as a whole to encode a forward-looking rather than a general validity predication (there are fifteen tokens of forward-looking ‘would you prefer to’). To conclude this historical digression we can state with some certainty that prefer seems to be moving in the same direction as like and love with situations in the projected future being typically encoded by modalised matrix verbs followed by to infinitive complements, and situations of general validity being encoded by non-modalised matrix verbs and to infinitive complements. We have already noted that the verb prefer itself implies the presence of a choice on the part of the subject. The choice to hand may either be encoded explicitly or implicitly. Both sorts are exemplified in (485) – (488). (485) and (487) exemplify the ‘prefer to infinitive’ constructions, (486) and (488) ‘prefer -ing’. (485) Its actions belied Bush’s declaration on Nov. 15 that he preferred to hold

this course than to do something that did not work. (HLD 485) (486) Did you prefer, prefer being a stevedore to driving the crane? (HSH 167) (487) These are the sort of boots I would prefer to use for autumn and winter

walking when a tough and sturdy boot is required. (A65 1970) (488) But although they live on the edge of Wentworth Golf Course, Lady

Haslam prefers gardening when at home. (A6L 1265)

(485) and (486) differ from (487) and (488) in that they make explicit mention of the dispreferred alternative. If we confine our attention for the moment to explicit acts of comparison where the discarded alternative is signalled by a preposition or the like and is mentioned after the preferred choice, we will see in Table 5.4 that explicit comparison is more typical of the ‘prefer -ing’ construction.


180

Table 5.4: ‘Prefer to infinitive’ and ‘prefer -ing’ contrasted with regard to incidence of explicit ‘post-prefer’ comparison and type of preposition/subordinator introducing dispreferred option

rather

than than to and

not to instead

of Total number of explicit comparisons

Total construct-ions

prefer to infinitive

32 9 1 2 2 46 10% 441

prefer -ing

1 0 9 0 0 10 34% 29

Discussing the two prefer constructions, Bladon writes: “The -ing nominal is obligatory when the second term of a preference is expressed” (Bladon 1968: 208). As the evidence of the raw figures in Table 5.4 attests, this is not true of present-day British English. For every -ing complement in cases such as this, there are more than four instances of a to infinitive complement. Eagleston arrests Bladon on this point, asserting that “Alternatives may be involved in both the to- and -ing patterns, but the alternatives are not of the same nature” (Eagleston 1972: 142). If two alternatives are mentioned, like is typically compared to like. We thus have two basic structures x and y, exemplified by (485) and (486). x. S prefers to A (rather than to B) y. S prefers A-ing (to B-ing)

There are, however, two exceptions in the corpus to this tendency, two instances in which a to infinitive clause is compared to an -ing clause. These are (489) and (490). (489) Instead of this, Morita spent the time taking clocks apart, listening to

Western classical music and preferring to study physics to putting himself into business. (FRL 1729)

(490) How most people prefer to be actively involved in sailing the boat rather than just sitting and admiring the view. (BNH 961)

In (489) and (490) the act of studying or state of being actively involved are profiled as whole situations of general validity. What of the acts of engaging in business or sitting around in the boat: how are these profiled? They certainly do not seem to be ongoing same-time activities, which is the prototypical sense of the -ing complement form. We have already referred on several occasions to the question of the degree of semantic compatibility between the profiles of the matrix verbs and complement clauses. In some cases, for example choose and the to infinitive complement, there is a considerable degree of overlapping. Both the


verb and the complement form imply the existence of alternatives. The same overlapping exists in the case of prefer and the to infinitive complement. In the case of ‘prefer -ing’, on the other hand, there is no overlapping of the semantic profiles. However, there is no doubt that the two profiles are compatible. If we say that prefer means something like ‘judge a to be more acceptable in some respect than b’, then the two constructions would mean roughly the following: ‘prefer to infinitive’: judge a, one of several alternative situations, construed as a whole, to be preferable to b. ‘prefer -ing’: judge a, a situation construed as extended in time, to be preferable to b. We may then combine the two definitions as follows: ‘prefer to infinitive rather than -ing’: judge a, one of several alternative situations, construed as a whole, to be preferable to b, a situation construed as extended in time. The choice of this unusual construction in (489) may be a result of the writer’s wishing to make the rejected option seem more ‘active’, more ‘alive’, as it were, than the option actually chosen. In (490), on the other hand, the adverb actively and the phrase ‘in sailing’ lend a sense of liveliness to the normally static to infinitive, making ‘sitting’ seem just as natural as ‘sit’. As may be seen in Table 5.4, 10% of ‘prefer to infinitive’ constructions contain explicit post-prefer comparisons. The corpus also contains instances of explicit ‘pre-prefer’ alternatives, as in (491) and (492), and utterances in which the alternative is not mentioned, at least not in the same sentence, as in (493) and (494). (491) But the IWC has yet to decide whether to continue or end the moratorium,

preferring to ask more questions of the scientific committee regarding the management procedure (see Nature 358 , 99; 1992). (FBP 361)

(492) Rather than call it ‘primitive’, I prefer to see it as direct, relating what happened on the Earth directly with events in the heavens. (BMT 1095)

(493) ‘Rob prefers to come out with the complete picture in his mind, the game-plan mapped out. (CEP 2722)

(494) His three children have all made their own way in the world, preferring not to join the family firm. (A6L 239)

These constructions all encode predications of general validity. In (492), for example, the preference expressed by the subject encodes a general way of looking at things. In (493) Rob has a standard approach to all games in which he takes part. In (491) and (492) a discarded alternative precedes the expression of preference. In each case the element mentioned first is accorded greater conceptual prominence than would otherwise be the case. It is spelt out clearly,


182

before being discarded. Eagleston (1972: 142) asserts that the discarded option in the ‘prefer to infinitive’ construction is not given serious consideration by the subject. I would have thought this unlikely in an utterance such as (491), for example. In most cases of ‘prefer to infinitive’ the discarded alternative is implicit in the sentence. It may, of course, have been mentioned explicitly or implied in a previous stretch of text. Here, for example, is the expanded context of (493). (495) ‘My game is more suitable to the new laws. I believe I am a more flexible

player who can sometimes play off the cuff and throw the opposition. ‘Rob prefers to come out with the complete picture in his mind, the game-plan mapped out. He is a great player but we are different types of fly-half.’ (CEP 2722)

Here the alternative to to come out with the complete picture in his mind is clearly to play off the cuff. In other cases the alternative is so obvious it does not have to be stated. This is especially true in cases where the infinitive is negated, as in (494). There are also cases, like the modalised (487), where it is possible to conceive of more than one alternative. To sum up, ‘prefer to infinitive’ encodes the expression of choice between two, or possibly more, alternatives. The discarded option(s) may be either explicit or implicit. The option chosen must, of course, be explicit. ‘Prefer -ing’, exemplified by (486), (488) and (496) – (499) also encodes the expression of a choice on the part of the subject. (496) After his liaison ended and he was alone again, he wrote simply and

forlornly, ‘I hated being alone so much that I preferred being with a bad whore to being alone.’ (CBN 1360)

(497) if you you said you preferred to work here, you preferred working here to other places? (FYL 83)

(498) Though my brothers too enjoyed shooting, they preferred hunting, and my mother, who had hunted in Ireland before she went to Abyssinia, now took it up again. (H0A 963)

(499) ‘Some women prefer seeing make [sic] gynaecologists because they find them more authoritative.’ (G2V 1972)

‘prefer -ing’ differs from ‘prefer to infinitive’ in that in its case the -ing clause always seems to involve a choice between two, and only two, alternatives, both of which are usually explicitly mentioned, either in the sentence itself or the immediate context. Eagleston maintains that, in the case of ‘prefer -ing’, “there is also the strong implication that both items are pleasing, that the subject is merely stating his order of priorities at a given point in time, that he is not being asked to make a choice between something that is enjoyable and something that is distasteful” (Eagleston 1972: 143). The evidence of the corpus supports this


assertion to a certain extent. The subject may, however, be faced with a choice between two distasteful situations, as is the case in (496). The -ing construction differs from the to infinitive one not only in implying the existence of two, and only two, alternatives. It is also more likely to mention the discarded alternative explicitly. There would also appear to be fewer instances of ellipsis in the discarded alternative: (497) is atypical in this respect. In both cited occurrences containing two contrasted -ing forms, (486) and (496), both -ing clauses are summoned up and evaluated from a mid-interval perspective: one compares two activities as they are unfolding (one in the physical, one in a mental, domain) and signals a clear choice between them. Even when the discarded option of the subject in a ‘prefer -ing’ construction is not explicitly mentioned, it is never difficult to recover from the context. Thus in (498) shooting is explicitly mentioned before the subject’s preferred activity, while in (488) the image of living on the edge of a golf-course evokes the idea of golfing as a natural past-time option which is then discarded in favour of gardening. Finally, in (499) the alternative sex to ‘make’ (sic) is presumably ‘femake’! It was noted above that (497) is unusual in that it contains ellipsis. Another interesting point about it is that it contains an example of both types of complement clause, both ‘prefer to work’ and ‘prefer working’. Here we may observe the contrast between the more general, soi-disant, sense of ‘prefer to infinitive’ and the more immediate, unfolding same-time sense of ‘prefer -ing’. The difference between the two prefer constructions is not as great as it is in the case of the positive and negative attitude constructions. Nevertheless, a difference there is, and it is the exact same difference we have noted in the case of other pairs of constructions containing the same matrix verb. On the one hand a process is viewed as a whole from a distance as the targeted option; on the other it is construed from within as it unfolds before our eyes.

5.5.4 Summary of Attitude constructions

Let us round off this comparison of General to infinitive and Same-time -ing Attitude constructions with a brief glance at how the distinction between the two has been drawn by previous scholars. Conrad contains a useful summary of some of these distinctions.

Many authors consider that after verbs such as like, hate, prefer the general rule is that the infinitive refers to a specific case while the gerund has a more general meaning (for instance, Poutsma, Zandvoort, Schibsbye, and Koziol and Hüttenbrenner). Thomson and Martinet state that the gerund refers to habits and past actions, while the infinitive refers to future and potential actions. To Bolinger the essential difference is that the gerund represents the action in a more vivid way. Kruisinga thinks that when these verbs take the infinitive they are verbs of volition and when they take the gerund they are more like verbs of sensation (for instance,


184

like = ‘have a pleasant sensation’). When they are used with the gerund they express a contemporaneous sensation, while the infinitive expresses a future action. Bladon’s main rule is that like with the infinitive means ‘want’ and with the gerund it means ‘enjoy’. (Conrad 1982: 17)

The assertion by Poutsma et al. that the -ing form is more general in meaning is not supported by the evidence of the corpus. Indeed, the opposite would seem to be the case. It is General to infinitive constructions that are normally used to refer to habits and not the -ing form. Kruisinga, however, seems to have got closer to the facts of usage with his distinction between verbs of volition and verbs of sensation, at least if one is willing to interpret volition in a rather broad sense as ‘displaying a general inclination towards’. One may, however, question whether it is necessary to ascribe multiple senses to the matrix verbs themselves, in the manner of both Kruisinga and Bladon, rather than ascribing differences in meaning to the various construction types in which they feature.

5.5.5 Aspect constructions

There are seven matrix verbs in this study that occur in Same-time Aspect constructions. These are continue, keep, lie, remain, repeat, sit and stay. There are three matrix verbs, cease, continue and use(d), that occur in General Aspect constructions. Continue is thus the only Aspect verb to occur both in a Same-time and a General construction. As may be seen in Table 5.5, the majority of to infinitive tokens encode general validity predications, the majority of -ing tokens specific predications, either once-off or serial. We will return below to the sense of the specific predications with ‘continue to’ and the two general predications with ‘continue -ing’. We will, however, start by looking at prototypical uses of the two constructions, exemplified by (500) – (504). (500) However, existing publications on France, Spain, Portugal, Germany and

Italy continued to sell modestly but steadily and it may be possible to commission further studies in due course. (HC5 313)

(501) The region has continued to develop projects to assist professionals working in religious broadcasting. (EB9 402)

(502) From the mint Canada continues to produce coins with an aviation theme. (J1B 627) (503) ‘You talk, I’ll listen,’ Marler replied and continued piling marmalade on

his buttered toast. (ARK 1102) (504) Yeah I genuinely look all I can say about this new album is if I’d have

continued recording for the last twenty odd years and had a sustained recording career like Cliff continued singing, this would have been the album that I would’ve ended up doing anyway. (KGH 905)

Aspect constructions 185

Table 5.5: General and specific predications encoded by ‘continue to infinitive’ and of ‘continue -ing’ with vertical percentages

General

predications Specific

predications Total

continue to infinitive 416 99.5% 21 35.6% 437

continue -ing 2 0.5% 38 64.4% 40

Total 418 59 477

(500) – (502) do not encode specific activities of selling, developing or producing, but rather the higher order predication of the likely realisation of these situations on appropriate occasions. (503), on the other hand, encodes the repeatedly actuated acts of marmalade piling on a single occasion. (504) encodes intermittent acts of recording and singing on a series of real, or in the case of the first counterfactual predication imaginary, occasions. As was the case with prefer, the to infinitive complements in these examples may seem very similar in meaning to the -ing complements. In the case of prefer we saw in section 5.5.3 that the similarity was in part due to the matrix verb’s encoding the idea of a latent alternative, thus overlapping in meaning with the to infinitive complement form. The opposite is true in the case of continue, which itself profiles situations from a mid-interval perspective, which is one of the properties of the -ing form in Same-time constructions (see Table 4.6, p.122). Freed refers to the difficulty of distinguishing between the two constructions in the following passage.

In general, the to V form of a sentential complement, whether occurring with start, begin, continue, or cease, carries with it a generic reading. Once again a generic reading of an event suggests a REPETITION (or a series) of the event in question, occurring at DIFFERENT MOMENTS throughout an unspecified stretch of time. In all cases the temporal nature of these is intermittent; the V-ing form, on the other hand, has a durative reading which here refers to the unspecified duration of a SINGLE EVENT. The main area of confusion when considering this distinction is the aspectual effect of the V-ing form on those verbs which name events that are series or accomplishments with plural noun-objects (which then become series). For these, the V-ing form creates an iterative reading, easily confused with a generic one. (Freed 1979: 152)

The Forward-looking Aspect matrix verbs start and begin will be discussed in section 6.3.3. One may however pre-empt this discussion by pointing out that it is difficult to find hard evidence in the BNC for Freed’s distinction between the complement forms in terms of genericness/seriality in the case of situations located in the future. It is much easier to verify in the case of continue, where the


186

-ing form of complement is favoured in my material by a margin of 9 to 4 to encode situations involving iteration on a single occasion, as in (503). The ‘continue -ing’ construction is very similar in its semantic make-up to the ‘keep -ing’ construction. Both Freed (1979) and Brinton (1988) maintain that continue and keep may both be used to encode both continuative and iterative situations. As Brinton puts it,

Both keep and continue occur freely with activities and denote the continuation of the nucleus of the situation […]: with iterative activities, they denote the repetition of the activity on one occasion when the V-ing complement is used, the repetition of the activity on different occasions when the to V complement is used. (Brinton 1988: 88)

The data in my study suggest that while the two -ing constructions may both occur freely with activities, they differ in that keep is less commonly used than continue to encode continuative as opposed to iterative situations. Only 19 of 115 downloaded tokens of ‘keep -ing’ (20%) were continuous, and 11 of these contained the complement predicate go. 15 of 40 tokens of ‘continue -ing’ (38%) were continuous, and none of them contained the predicate go (there are 69 tokens of ‘keep going’ in the BNC as a whole, compared to 4 of ‘continue going’). Nor do the data in the BNC lend support to the assertion that the -ing complement form tends to be used with continue to encode repetition on a single as opposed to different occasions. Consider in this respect (505) – (508). (505) She did not hear him and vigorously continued splashing. (H8A 3604) (506) Add the cocoa mixture and the egg yolks, in alternate spoonfuls, and

continue beating until smooth and creamy. (A7D 1546) (507) KATE couldn’t make up her mind all through dinner whether Ace was

playing a deep game with her or had decided that as she was available, so to speak, he intended to continue making love to her. (HGM 3135)

(508) Despite his retirement as chairman, Mr Jackson will continue attending the monthly history society meetings in Witham Hall. (K52 2131)

(505) and (506) resemble (503) in encoding repeated realisations of splashing and beating on a single occasion. (507) and (508) resemble (504) in encoding the repeated (serial) realisations of love-making and attendance on separate occasions. 9 of the 38 tokens of specific predications encoded by ‘continue -ing’ in my data encode iteration on a single occasion and 13 iteration on separate occasions. The remaining 16 encode continuative rather than iterative situations. As mentioned above both continue constructions always make at least implicit reference to some point after the initiation of a situation or a series of situations. The matrix verb itself may be said to adopt an internal perspective on a durative situation or a series of situations, looking both forwards and backwards, as it were. In the case of ‘continue to infinitive’ we are therefore faced with a General construction which partakes of one of the prototypical characteristics of the Same-time -ing construction, the mid-interval perspective.


This is bound to lead to a blurring of the distinction between the two sorts of construction. Given the fact that both constructions always make at least implicit reference to some internal point, and given that this point must perforce be located after the initiation of the situation, it follows, as Freed quite rightly points out, that the ‘activity’ must be presupposed to have, in fact, been initiated (see Freed 1979: 94). Sometimes the sentence itself makes direct reference to some point after the point of initiation, as in (509) and (510). The initiation of the situation may already have been mentioned elsewhere in the discourse. In the latter case the first mention of the situation may be at some remove from the predication of its continuance, as in (511). (509) These were already declining by 1400 and apparently continued to do so

throughout the fifteenth century. (HWG 75) (510) They had followed it every morning and would continue to follow it until

their sentences were up. (G01 32) (511) Then, as she and Silvia packed up their things after their final English

lesson, she told the other girl, ‘I’ll be leaving tomorrow, just as soon as your exam is over.’ As she spoke, she fiddled with the corner of her notepad. Silvia was horrified. ‘You can’t!’ she protested. ‘You’ll miss Guido’s test run! You’ll miss the celebrations!’

‘I know. But I have to go.’ ‘Why, for heaven’s sake? I thought you didn’t have to start work again for another three weeks?’

‘I don’t, but there are things I have to do at home.’ Nervously, Ronni continued to fiddle with her notepad. (JXT 490) In (509) the point beyond which the declining continues is 1400. In (510) it is the point in the past in which perfect had is anchored. In (511) it is the second mention of fiddling, an activity whose initiation was signalled by its first mention. It is not always the case that the initiation of the situation being continued is referred to explicitly in the discourse context. In fact, as may be seen in Table 5.6, more often than not the situation being continued is new to the discourse, in the sense that it is not already activated or active.

Table 5.6: Given and new situations encoded by ‘continue to infinitive’ and ‘continue -ing’ with horizontal percentages

Given situations New situations Total

General ‘continue to infinitive’ 113 27.2% 303 72.8% 416

Specific ‘continue to infinitive’ 18 85.7% 3 14.3% 21

Specific ‘continue -ing’ 13 34.2% 25 65.8% 38

Total 144 331 475


188

Specific predications with both to infinitive and -ing complements will be discussed below. (512) – (514) exemplify general validity predications in which the situations to be continued are not currently active in the discourse. (512) She continued to go out to dinner with academics, to receive the hard-

drinking architect. (FB0 2159) (513) The Glenavon fans were obviously unhappy with the way their team has

continued to play crap football. (HJ3 7577) (514) On the airliner scene, Air Bridge continue to acquire Electras, with a

further two former Canadian examples registered during August and September to supplement their expanding fleet of the type. (CLU 1326)

(512) is from a work of fiction. The subject’s propensity for dining with academics, though it may be familiar to readers of the novel, is not mentioned in the hundred immediately preceding sentences. (513) is from a letter to the editor, dealing primarily with the behaviour of Glenavon fans, not their football team. There is no previous reference to the latter in the letter. (514) is from a news report in a magazine. Interested readers may perhaps be presumed to be aware of the acquisitions in question, but they have not been previously mentioned in this report. Thus in none of these three utterances, which may be taken as representative of over 300 general validity predications, is the situation whose continuation is predicated active in the discourse. Nevertheless it may be inferred that the complement situation pertained before the point in question, even though we lack any previous reference to it. By employing the ‘continue to infinitive’ construction, the speaker guarantees that this is indeed the case. Like ‘continue to’, ‘continue -ing’ is often used to encode the continuation/iteration of a situation that is new to the discourse context. As was mentioned above, ‘continue -ing’ is either continuative or, if iterative, may refer to repetition on either a single occasion or a series of occasions. (515) – (517) further exemplify the ‘continue -ing’ construction. (515) This will allow it to continue being part of a household, rather than being

confined on its own in kennels. (CJE 229) (516) The two men continued walking, trying to fathom the mystery of the fog-

bound train. (CE9 279) (517) He continued nodding thoughtfully as Ybreska detailed the site. (CDA 1162) (515) and (516) both refer to continuative situations, a state and activity respectively. In both cases the situation continued has already been mentioned. In (515) one could substitute a to infinitive complement for the -ing complement with little change in meaning, the matrix verb itself sufficing to indicate the durative nature of the complement clause situation. ‘to walk’ would perhaps be less felicitous in (516), as it would open for the implication that walking together was something the two participants were in the habit of doing. However, since


this implication could easily be filtered out by the addressee in an actual discourse context, there is no reason to expect that ‘to walk’ would be strongly disfavoured. The complement in (517) is iterative, nodding being a punctual activity. This repetition takes place on a single occasion, as is the case with examples (503), (505) and (506). The action of nodding has already been mentioned in the discourse, unlike the actions of piling marmalade in (503) and splashing in (505). Note that in all these examples ‘continue -ing’ resembles ‘continue to infinitive’ in that the perspective of the speaker is located at some point after the beginning of the continuing or recurring situation in question. 38 of 40 tokens of ‘continue -ing’ in my data resemble (515) – (517) in encoding specific once-off or serial predications. The remaining two tokens, cited as (518) and (519), appear to encode general validity predications. (518) We continue taking the French ships, but they take none of ours. (ALU 807) (519) A variety of difficulties were foreseen about the ability of fundholders to

continue making savings. (EA2 900) Despite the fact that the object determiner is definite, (518) appears to encode a situation involving the capture of non-specific ships. One could perhaps argue that it encodes a same-time predication, involving serial iteration. If, however, one were to substitute ‘to take’ for taking, I would have no hesitation in describing the construction as general. Moreover, if one were to produce this utterance today it is likely that one would employ the to infinitive rather than the -ing form of complement. In fact, the token is actually from Late Modern English, from a letter written in 1758. It thus belongs to a stage in the language when there was less form-function isomorphism in the area of non-finite complementation than is the case today. Unlike (518), (519) is an example of contemporary usage. It is classified as General because of its generic subject, no specific action of saving being referred to. One might speculate that the employment of the -ing form is here influenced by infinitive form of continue. However, the to infinitive form of complement is actually almost twice as common as the -ing form after infinitive continue, as may be seen in Table 5.7, which contains details of the TAM relations of both continue constructions. According to the raw figures in Table 5.7 the to infinitive form of complement is actually more likely to be used than the -ing form after infinitive continue. Moreover, of the twenty-four relevant examples of ‘continue to infinitive’ in Table 5.7, only two of the matrix verbs are bare infinitive in form. We may contrast this with ‘begin to infinitive’, where seven of nine infinitive matrix verbs are bare.


190

Table 5.7: TAM of matrix verb continue in to infinitive and -ing constructions, with horizontal percentages, based on 437 tokens of ‘continue to infinitive’ and 40 tokens of ‘continue -ing’

Present Past Perfect Mod

al Non-finite

Imper-ative

simp prog simp prog pres past -ing inf.

continue to inf

4 1%

106 24%

5 1%

137 32%

0 16 4%

8 2%

115 27%

16 4%

24 6%

continue -ing

1 3%

5 13%

0 9 23%

0 1 3%

0 10 25%

0 14 35%

Why should the to infinitive complement behave differently after continue than after begin? I would suggest that the difference is related to a difference in the domain evoked by the path-goal schema instantiated by to. In the case of ‘begin to infinitive’ the targeted alternative is a complement situation located in the projected future. The ‘continue to infinitive’ construction, on the other hand, normally encodes a general validity predication. Let us look at some examples of the construction in which the matrix verb is also a to infinitive. (520) On Sept. 23 the Philippines asked the USA to continue to prevent Marcos’s

corpse from being returned to the Philippines [see p. 38398]. (HLB 1592) (521) MTM’s bankers have agreed to continue to provide funds for the deeply

troubled chemicals group until the end of the month. (AJH 268) In both (520) and (521) the first to infinitive, the complement of the Forward-looking matrix verbs ask and agree, points the way to a targeted alternative in the projected future. The prevention of the return of Marcos’s corpse, and the provision of funds for the chemical group, are both profiled as occurring after the asking and agreeing. The paths to the targeted alternatives are traversed through time (syntagmatically), the ‘continuation’ being subsequent to the asking and agreeing. The prevention and provision, on the other hand, are profiled from within as general targets. Rohdenburg (2003) has argued in favour of a horror aequi principle to explain the avoidance of successive pairs of to infinitives. This principle certainly seems to motivate the avoidance of such duplication in many contexts. Why then does it not do so in the present one? The answer, I think, may be related to the fact that the pairs of to infinitives serve different functions. One encodes a Forward-looking predication, the other a General one. It is because the two occurrences of the to infinitive encode paths in two distinct domains that there is no reduplication or redundancy involved. Although the evidence of Table 5.7 shows that we cannot conclude that the -ing form was chosen in (519) to avoid consecutive infinitives, such a wish may well have played some part in motivating the employment of the construction. Only 8% of all tokens of continue are in the infinitive, but as many


as 35% of the -ing complements follow infinitive continue. Be that as it may, (519) must still be classified as an exception, similar to (475) and (476), in as far as it instantiates the encoding of a general validity predication by means of an -ing clause. We now turn to the 21 tokens of ‘continue to’ that seem to encode not general but specific predications. (522) – (525) may serve as representative samples of this sort of construction. (522) The two men continued to stroll leisurely around the perimeter fence

which separated the living quarters and parade grounds of the air-base from the restricted areas of the hangars and control tower. (CDA 1881)

(523) But it was hard to scowl, she found, when her heart was suddenly skittering at the way the dark eyes were quite openly smiling as they continued to focus on her now lightly rising bosom. (JXT 481)

(524) The nurse continued to stare disapprovingly. (HJH 72) (525) The other one continued to burn. (G0P 2926) The situations encoded in these tokens are all realised on a single occasion. Thus the strolling in (522) and the focussing in (523), both of which, incidentally, are new to the discourse, are not encoded as likely to be realised on appropriate occasions. The same point may be made about the predications in (524) and (525), both of which are prefigured in the immediate context. How, then, should these predications be categorised? Are they Same-time predications? If so they are the only instances of such predications encoded in to infinitive clauses in my material, with the possible exception of (449) discussed in section 5.5.2 (p. 173). We saw in sections 5.5.1 and 5.5.2 that while the majority of non-modalised Attitude to infinitive constructions encode general validity predications, there are a minority that are Forward-looking. Could the predications in (522) – (525) be characterised in the same way? There is no doubt that the activities of strolling, focusing, staring and burning are predicated as taking place after the time of the matrix verbs. That is, if we know the exact time at which the continuance or iteration is predicated as occurring, we also know that the situation in the complement clause occurred after that time. It therefore does not pose any problem for the system of classification proposed in the present study to categorise these 21 tokens of specific ‘continue to’ as Forward-looking. Unlike most Forward-looking to infinitive constructions ‘continue to’ entails the realisation of its complement situation. It would thus resemble in this respect ‘begin to’, to be discussed in section 6.3.3 and causative constructions containing such verbs as force and get to be discussed in section 6.4, as well as the Effort verb manage and the Mental Process verb remember. However, classifying specific ‘continue to’ as Forward-looking raises the question of whether we should not also classify ‘continue -ing’ as a Forward-looking rather than a Same-time construction. To do so would involve placing ‘continue -ing’ in a different category to ‘keep -ing’ and ‘remain -ing’, something which may seem counterintuitive. On the other hand not to do so would seem to


192

involve making an unwarranted distinction between specific ‘continue to’ and ‘continue -ing’. Before making up our minds on the question, it may be worthwhile taking a closer look at the two constructions. We have seen that they resemble one another in that both of them always make implicit reference to some point after the initiation of a situation or a series of situations. Is there any difference in meaning between the two constructions when used in this way? Consider examples (526) and (527). (526) The Doctor didn’t seem tired and continued to run. (HTY 3278) (527) As Will turned and threw up his arms to head them off, Constance

continued running. (CEY 385) One scholar who has written of the contrast between the constructions is Verspoor, who distinguishes between ‘continue to infinitive’ and ‘continue -ing’ in the following manner: “if continue is followed by an -ing complement, it implies an uninterrupted action in process (intention in action), but if continue is followed by a to infinitive, it implies an interrupted action and a ‘prior intention’ of beginning again” (Verspoor 1998: 523). It must be admitted that it is very difficult to find evidence of such a “prior intention” in the examples in the corpus. There is certainly no implication of interrupted action in (526), for example, as one can see from the expanded context in (528). (528) ‘We can’t give up,’ the Doctor said. Blake needed to rest. He turned

around and slowed down, seeing no sign of the monsters. The Doctor didn’t seem tired and continued to run. ‘Don’t stop,’ he shouted, running on ahead. Blake sighed, and ran to catch up with his travelling companion. (HTY 3278)

Verspoor is correct in asserting that once-off -ing complements normally encode an uninterrupted “action in progress”. They rarely encode an interrupted action on a single occasion, although they can do, as shown by (529). The problem is that the to infinitive when used to encode specific predications, as in (522) – (525), does not normally imply the existence of interruptions either! Quite the opposite is the case in (530), for example. (529) Later, after leisurely baths, we would go down to the bar for our aperitifs

and continue reading or play cards until dinner, which was always excellent and ended with a savoury - an almost obsolete course, sadly. (CES 558)

(530) Fael-Inis continued to play without ceasing, and the music spun and shivered and soothed, and across the slanting features there was a look of the utmost concentration now, for it is only at the invitation of a Mortal that the Lad of the Skins can enter a house and do his terrible work.

(G10 196)


In (529) it seems natural to interpret the activity of reading as taking place on a single occasion but as having been interrupted by the subjects’ ablutions. The activity is still profiled from a mid-interval perspective as ongoing, despite the interruption. In (530) it is explicitly stated that the activity of playing is not interrupted. The to infinitive form was defined schematically in section 4.3 as encoding a targeted alternative. One may therefore consider it apposite to ask whether one can detect the presence of any latent alternatives in the specific ‘continue to infinitive’ construction. (The latent alternative of non-realisation on any particular occasion is implied, by definition, in all general validity predications.) If we confine our attention to the tokens already cited, we may note that such an alternative is explicitly mentioned in (528) where the alternative of ‘resting’ is strongly contrasted with that of ‘running’. Other tokens where an alternative is implied include (531) – (533). (531) If the diaphragm is torn (a wc may be difficult to flush), or the washer is

worn, it may prevent the ball valve from closing properly, and the cistern will continue to fill until it runs out of the overflow. (CG5 1684)

(532) She looked up into the haughty face but he said nothing at all. He merely continued to advance (HGK 3602)

(533) He seemed to regard the exchange purely as a joke, but Marise continued to glower beneath her perfectly drawn black brows, and Faye too looked irritated by Greg's words. (H9H 1324)

In (531) the implied alternative involves the ball valve’s closing properly. In (532) it involves the subject’s saying something instead of advancing. In (533) it is taking the exchange in question less seriously. In both (532) and (533) the contrast with the implied alternative is signalled by but. What of the ‘continue -ing’ tokens? Do not they too evoke latent alternatives? Given the semantic make-up of the matrix verb continue, the fact that anything predicated as continuing could presumably also cease to do so, it would not be surprising if they were found to do so. And it is, in fact, true of one of our tokens, (515), in which the alternative to “to continue being part of a household” is explicitly stated to be “being confined on its own in kennels”. However, (515) is not typical of the construction. None of the other tokens of ‘continue -ing’ which we have cited strike such a contrastive note. For instance in (503) there is no implication of an alternative to piling marmalade, in (516) no implication of an alternative to walking, and in (517) no implication of an alternative to nodding. Now the apprehension of latent semantic elements is very much a subjective matter, and thus difficult to corroborate scientifically, but there do exist some more objective correlatives, such as the presence of a conjunction like ‘but’ or a conjunct like ‘however’ or ‘nevertheless’. There are three occurrences of such contrastive markers among the 21 tokens of specific ‘continue to’ in my material. (532) and (533) are two examples. On the other hand there are also two such occurrences among the 40 tokens of ‘continue -ing’. These are cited as (534) and (535).


194

(534) Father-of-two Mr Evans, the Aintree recall man for the past seven years, was not at his home in Bryn, near Wigan, yesterday. He and his wife, Edna, were known to be distraught by the National furore. Nevertheless, Mr Evans has continued working at Aintree’s sister course at Haydock. (K4J 1611)

(535) The drawing suggests that there might come a time when the system does not recover at the bottom of its dive but continues heading downwards to collapse. (H0E 1333)

A close examination of all downloaded tokens of both constructions leaves one with the distinct impression that there are more contrastive utterances in the case of ‘continue to infinitive’ than ‘continue -ing’. However, yet again we are dealing with tendencies, rather than absolutes. Are there, then, no substantive differences between specific ‘continue to’ and ‘continue -ing’? If we confine our attention to once-off predications, which is the area in which both types seem to be employed in a similar fashion, we find that they are very similar in terms of animacy of subjects, 11 of 13 subjects being human in the case of ‘continue -ing’ and 19 of 21 in the case of ‘continue to’. They are also very similar with respect to the extent to which the complement situation is new or given in the discourse. The data in Table 5.6 show that 18 of 21 tokens of specific ‘continue to’ are given. The table shows that only 13 of 38 tokens of specific ‘continue -ing’ encode given complement situations, but 10 of these are actually to be found among the 13 tokens encoding once-off predications. One area, however, in which the two constructions do seem to differ is in their collocations with types of complement predicate. For instance as many as 4 of the 21 tokens of specific ‘continue to’ in my data contain the complement predicate stare. A search of the BNC reveals that the corpus as a whole contains 56 tokens of ‘continue to stare’ and a single token of ‘continue staring’. Table 5.8 contains details of three common complement predicate types in both sorts of construction, vision predicates, conversation predicates and physical motion predicates.

Table 5.8: Types of complement predicate in specific once-off predications encoded by ‘continue to’ and ‘continue -ing’ with vertical percentages

in random sample

stare, look, watch in BNC

talk, speak in BNC

run, walk in BNC

Specific once-off ‘continue to’ 21 62% 105 89% 19 45% 13 38%

Specific once-off ‘continue -ing’ 13 38% 13 11% 23 55% 21 62%

Total 34 118 42 34


The data in Table 5.8 are restricted to tokens encoding specific predications which are realised on a single occasion. In other words, general validity predications and serial predications have been omitted, as have any tokens containing multi-word verbs. The three types of predicate in questions occur with both types of construction. Moreover, if one adds together the figures from columns 3, 4 and 5 and compares the results to column 2, one finds no significant difference between the totals, indicating that the random sample on which we have been basing our discussion may be taken as representative for the corpus as a whole. Nor is there any significant difference between the figures for conversation and physical motion predicates. However, when we compare the figures for the vision predicates with the two other types, we find that the differences between them are statistically significant at the level of p = 0.001. In other words the speaker wishing to encode the continuation of an activity involving the visual faculty is much more likely to choose the to infinitive complement form than is the speaker who wishes to encode a situation involving a communicative or locomotive activity. Why should there be this difference in the form of complement chosen to encode the different types of bodily activities? Is there any difference between vision verbs on the one hand and conversation and motion verbs on the other that might motivate the difference? All three are activities. They are all imperfective, unbounded and internally homogeneous. At any sub-interval of staring or talking, the subject may be said to be fully engaged in the activity of staring or talking. Having said that, the notion of internal homogeneity may admit of nuancing. Talking involves moving the lips, walking involves moving the legs, but staring does not involve moving the eyes. In the real world, of course, one cannot continue to stare for very long without blinking. However, the need to occasionally blink is factored out of the meaning of stare, which is commonly defined in terms of looking “with eyes wide open” (Sinclair 1987) or “fixedly” (Pearsall and Hanks 1998). If we are more likely to conceive of visual activities as wholes and conversational and motion activities as ongoing, this may be a motivating factor in the choice of the to infinitive form to encode the former and the -ing form to encode the latter. Nevertheless, as is shown in Table 5.8, both complement forms may be used to encode all three types of activities, so once again we are faced with tendencies rather than hard and fast rules. The question was raised above as to whether specific ‘continue to’ should be categorised as a Forward-looking construction and, if so, whether ‘continue -ing’ should also be so categorised. Our discussion has shown a considerable degree of overlapping between the two constructions. Despite this I have chosen to retain my categorisation of ‘continue -ing’ as a Same-time predication, placing it in the same class as ‘keep -ing’ and ‘remain -ing’. As for specific ‘continue to’, I have chosen to label it as a Forward-looking construction, along the lines of non-modalised specific ‘like to’ and ‘hate to’, discussed in 5.5.1 and 5.5.2. Placing specific ‘continue to’ and ‘continue -ing’ in separate categories is not a particularly satisfying solution, given the undoubted similarities between the two constructions. The categories in question, however, are to be viewed as


196

prototypical in structure, allowing for a certain degree of overlap at their boundaries. Moreover, most constructions fit easily into one or other of the categories posited in Chapter 2. In Chapters 5, 6 and 7 I have chosen to concentrate on the more intractable constructions, one of the least tractable of which is specific ‘continue to’. To sum up this discussion of the two non-finite complement constructions containing continue, we may conclude that the distinction between the Same-time -ing construction and the General to infinitive construction is somewhat blurred in the case of both continuative and iterative situations. In both cases this blurring is due to the connotation of a mid-interval perspective by the matrix verb. This is one of the prototypical properties of the Same-time -ing complement form. There is thus a degree of overlapping between the semantic profiles of the matrix verb and the complement form in the case of ‘continue -ing’ that is lacking in the case of ‘continue to’. In the latter construction the matrix verb contributes its mid-interval perspective to a construction containing a general validity predication. We have also seen that a minority of ‘continue to’ constructions encode specific predications and that the latter are particularly difficult to distinguish from ‘continue -ing’ constructions.

5.6 Same-time constructions: a summary

In this chapter we have compared Same-time -ing constructions, which are prototypical for -ing complement constructions as a whole, to various other complement types in constructions sharing the same matrix verb. In section 5.2 we compared bare infinitive and -ing complements with Perception verbs and found that the common view of the difference between the two, that -ing complements are employed to encode situations envisaged as extending through time and bare infinitive complements situations viewed as a unitary whole, was confirmed by our examination of the corpus data. In section 5.3 we contrasted Same-time -ing and Judgement to infinitive constructions. We saw that the occurrence of the to infinitive complement form with Perception verbs generally invited a contrastive reading, that something seen to be x was contrasted, explicitly or implicitly, with another possible situation. In section 5.4 we looked at Same-time -ing and Forward-looking to infinitive constructions with the same matrix verb. We noted that, while the distinction between these two types is normally quite clear-cut, as one would expect given the fact that they encode situations in two different ontological domains, there were a couple of cases where speakers used what appeared to be the ‘wrong’ form, for instance example (377) in which 'try -ing' is used in a Forward-looking context. A small handful of similar exceptions surfaced in section 5.5, which dealt with the contrast between Same-time -ing and General to infinitive constructions. While the contrast between these two forms is again a clear one, we did encounter a couple of instances, such as (449), in which the to infinitive form is used to encode a same-time predication. However, exceptions like these are clearly just that - exceptions.

Same-time constructions: a summary 197

They are so rare as to render it unnecessary to revise the classification proposed in Chapter 2. One other problem encountered in section 5.5 is a blurring of the distinction between the two construction types when the matrix verb itself denotes a property which is also denoted by one of the constructions. We have seen that the verb prefer itself points to a targeted alternative, thus overlapping with a core property of the to infinitive complement construction. The verb continue, on the other hand, profiles a situation from a mid-interval perspective, thus overlapping with a core property of the Same-time -ing complement construction (though only a prototypical property of the -ing complement form). This sort of overlapping poses a particular challenge to the linguist attempting to tease out the meanings of the complement forms. We will meet a different sort of problem in the next chapter, in which the matrix verbs all point to situations in the projected future. These situations may, however, be encoded by various forms of complement construction. Our task, again, will be to work out the differences in meaning between them and to relate these differences to the characterisations of the various forms proposed in Chapter 4.

Chapter 6

Constructions in contrast: Forward-looking constructions

6.1 Introduction

There are 302 Forward-looking constructions in the present study. 220 of these contain the to infinitive complement form, 12 the bare infinitive, 67 the -ing form and the remaining three the to -ing form (see Table 2.7, p. 43). In no other domain do we encounter this proliferation of non-finite constructions. The speaker is often faced with the choice between two complement forms with one and the same matrix verb. Our tasks in this chapter are, then, to investigate the differences in meaning between the various constructions speakers may choose between, and to explain the motivation behind the choices they make. We will attempt to do this by examining pairs of constructions containing the same matrix verb. We will also look at pairs or groups of constructions which contain matrix verbs from the same semantic field, such as let and allow, or make and force. The fact that situations located by the speaker in the domain of the projected future are susceptible to a greater variety of construal options than situations in any other domain may be related to our lack of knowledge about how the future will come to evolve. It is, perhaps, worth pointing out that we also find an extensive range of options for encoding situations in the projected future in the case of simple finite clauses, such as those in (536) – (541).

(536) I leave on Sunday and I have my timetable. (AEA 284) (537) And, anyway, I need some sleep if I’m going to leave first thing in the

morning. (JXU 701) (538) I’m leaving shortly with two civilians for Athens. (CKC 1198) (539) I’m to leave in the morning, remember?’ (HGY 2328) (540) I’ll leave for France on Boxing Day.’ (HP0 477) (541) One afternoon in bed in her flat she said, ‘I will be leaving soon, you

know.’ (A0U 904)

The speaker who wishes to make use of a non-finite complement construction to encode a situation which he or she expects to come about in the projected future is never faced with more than two options in the case of one and the same matrix verb. There may, however, be up to three options within the same semantic field. We have already looked at the contrast between the to


200

infinitive and the to -ing form in section 4.5. We therefore restrict our attention in this chapter to the other three, much more common, forms. In section 6.2 we will compare to and bare infinitive constructions with Communication, Applied Attitude, Effort and Enablement matrix verbs. In section 6.3 we will compare to infinitive and -ing constructions with Mental Process, Attitude and Aspect verbs. Finally, in section 6.4 we will investigate Causation constructions in general, as the field of causation admits of both to and bare infinitive as well as -ing constructions.

6.2 To infinitives versus bare infinitives

Although there are only twelve Forward-looking bare infinitive constructions, some of these are very common indeed. For instance, ‘let S2 bare infinitive’ and ‘make S2 bare infinitive’ are both among the half-dozen most frequent non-finite constructions regardless of type. And although neither let nor make occurs in any other sort of non-finite, or indeed finite, complement construction, they may both be compared to near-synonyms that take other forms of complement, such as allow and permit in the case of let, and force and cause in the case of make. There are also some matrix verbs such as bid, dare, help and need which occur with both forms of complement. We will start our discussion of these constructions in 6.2.1 by looking at Communication constructions. Then in 6.2.2 we look at both Effort and Enablement constructions containing help. Section 6.2.3 is devoted to Enablement constructions containing let, allow and permit. Finally, 6.2.4 contains a comparison of to and bare infinitives with dare and need. Consideration of make is postponed until section 6.4 when we will consider Causation constructions with -ing complements as well as both sorts of infinitive.

6.2.1 Communication constructions

There are three Communication matrix verbs that occur in the corpus with both to infinitive and bare infinitive complements. These are beg, exemplified by (542) and (543), beseech, exemplified by (544) and (545), and bid, exemplified by (546) and (547).

(542) Master Hussey, he begs you also be in attendance.’ (HGG 58) (543) ‘Utterson,’ said a voice from inside the study, ‘I beg you to leave me

alone!’ (GV7 700) (544) I beseech you say not one word but ‘yes’ or ‘no’ till I have said all I have

to say. (FU4 1774) (545) ‘I beseech you to have trust, madam,’ persisted the cardinal. (CCD 1365) (546) She fetched her a tumbler of whisky, bade her drink, sleep, forget it.

(EV1 2375) (547) He bade her to rest and eat well and soon she would feel better. (ADS 719)

To infinitives versus bare infinitives: Communication constructions 201

As may be seen in Table 6.1, these three matrix verbs differ greatly in the proportion of bare infinitive to to infinitive complements with which they occur in the corpus.11 In the case of beg and beseech, the to infinitive complement form clearly predominates.

Table 6.1: Real and projected totals for Communication constructions containing matrix verbs beg, beseech and bid

Totals per 1000(x)/

real totals (y) Projected totals (x)/

real totals (y) Matrix verb

BNC x/y

to bare to bare beg 1945 x 178 2 346 4 beseech 64 y 12 2 12 2 bid 1342 x 12 59 16 79 Totals 374 83

Of the two examples containing ‘beg S2 bare infinitive’, one, cited as (542), is taken from a historical novel. The other is from a poem written in a non-standard dialect (the poem contains the line “Dat’s me sista yu beating upstairs” (F9M 1216), to give just one example of its non-standard syntax). Of the two examples containing ‘beseech S2 bare infinitive’, one, cited as (544), is taken from the text of the stage play Pamela, based upon, and containing dialogue lifted verbatim from Richardson’s eighteenth-century novel.12 The other is from a most curious work, entitled Warriors of Christendom, which, incidentally, also accounts for six of the examples of ‘bid S2 bare infinitive’. Warriors of Christendom is said by the BNC to be published in 1988. The actual text in the corpus however is Robert Southey’s 1808 translation into what can only be

11 The second column in the table, labelled BNC, contains the total number of tokens of the verbs in question in the first version of the BNC. The third column distinguishes between those verbs represented by over 1,000 tokens, which are labelled ‘x’, and those represented by under 1,000, labelled ‘y’. Columns 4 and 5 contain the number of occurrences of the various constructions in question in the downloaded samples, and columns 6 and 7 total numbers of tokens projected for the BNC as a whole on the basis of the figures in columns 2 and 4-5.

12 Richardson's original version is as follows: “I beseech you say not one word but Yes or No, to my questions, till I have said all I have to say”. It is quite possible to interpret the form say here as an imperative, rather than a bare infinitive. The classification of (544) among the tokens of the bare infinitive construction in the present study was dictated by the orthography.


202

described as early neo-Gothic of a Spanish text from 1637. It would appear that not only are ‘beg S2 bare infinitive’ and ‘beseech S2 bare infinitive’ somewhat archaic, but that they are perceived to be so by authors of historical fiction, who employ them to add a bogus sheen of authenticity to their works. There is no doubt that (543) and (545) are more representative of the way these two matrix verbs are employed in present-day British English. Bid, however, is a different kettle of fish. The bare infinitive construction may be a favourite with authors of historical fiction, but, even if one were to exclude examples from such sources from the data, the remainder would still comfortably exceed the total number of to infinitive examples. The ‘bid S2 bare infinitive’ construction would seem to be alive and well in present-day British English, at least in written registers, there being no occurrence of it in the spoken dialogue part of the BNC. But so, too, is the ‘bid S2 to infinitive’ construction. The question then is what, if any, substantive semantic differences there are between the two. There is no doubt that in contexts like those of (546) and (547) they appear very similar indeed. But are these occurrences, chosen because of their very similarity, typical of the two constructions? Given the low number of total occurrences of the to infinitive construction, it is difficult to be categorical with regard to it, but the fifty-nine occurrences of its bare infinitive counterpart do make for a basis on which to build more solid generalisations. And when we look at these occurrences we see that over twenty of the complement clause predicates encode actions of either sitting or rising, as in (548), and another thirty acts of motion, or lack thereof, such as coming, going or staying, as in (549). (548) Judy, an elegant woman in her early 40s with a peaceful demeanour, bids

us sit down in the verandah. (CK5 2825) (549) A gruff male voice bade them enter. (HH1 3841) In (548) and (549) it is clear that the matrix verb subject (SI) confidently expects the complement subject (S2) to realise the complement situation, which is both trivial and, no doubt, in their own interest. Upon hearing a knock on the door one normally bids the knocker enter. It is reasonable to interpret the knock as signalling a desire on their part to do so. There is no reason why there should be any suggestion of a possible alternative scenario being contemplated by the knocker, no suggestion that he or she may be less than willing to do the bidding of S1. The response of S2 is expected to be immediate and in line with the expressed wishes of S1. There may, of course, be various sorts of reasons for taking the compliance of S2 for granted. The reason for it may lie in the relative social status of S1 and S2, the former’s wish being the latter’s command, as it were, as in (550), or implying as much, as in (551).

(550) ‘Walk your horse forward till I bid you halt,’ ordered Owen. (K65 1521) (551) If you had not so stupidly bade me keep quiet, I should have done the trick

long since, and you would not have had to leave the place at all. (HGV 547)

To infinitives versus bare infinitives: Communication constructions 203

(552) He woke me at two this morning and gave me his best horse and bid me not to stop till I had overtaken you on the way. (FU4 959)

(553) Therefore I bid you to put away anxious thoughts about food and drink to keep you alive, and clothes to cover your body. (ALH 480)

(554) On these walls were none of the usual posters bidding young mothers to drink milk in pregnancy and bring their toddlers for a twice-yearly check-up. (A73 2038)

Turning to the ‘bid S2 to infinitive’ construction, exemplified by (547) and (552) – (554), it is noticeable that not one of the dozen downloaded occurrences contains a complement predicate of sitting or rising, which accounted for almost 40% of the bare infinitive examples. Nor do they encode other trivial day-to-day situations. There is only one predicate encoding motion, in example (552), which describes an action considerably less mundane than that of entering a room, for example. Goldberg, writing about the ‘Caused-Motion’ construction in English, mentions that some of the verbs in her study can also occur with infinitival complements when no direct causation is involved. She writes that “…expressions with infinitival complements do allow the theme to make a cognitive decision [….] Also, they do not presuppose that the theme will actually move along the specified path” (Goldberg 1995: 173). One might argue along these lines that the difference between the two bid constructions is that while both allow for S2 (the theme) to make a cognitive decision, it is only the variant containing to that allows for the possibility of S2’s moving along an alternative path. To what extent do the ‘bid S2 to infinitive’ tokens evoke the possibility of a latent alternative course of action? In (553), for instance, S1 would appear to think that there is a real chance that S2 will not accede to his request. The same point may be made in relation to (554), which implies that not all mothers can be expected to drink milk of their own accord during pregnancy. Duffley (1992) also makes the point that the bare infinitive construction presupposes the realisation by S2 of the complement situation. However, he sees the contrast between the two forms in terms of immediacy and futurity. He writes:

The contrast between the bare and to infinitive constructions with bid can therefore be stated in terms of whether compliance is taken for granted or not. When it can, no room is left for the non-realization of the infinitive event, and so the concurrent representation evoked by the bare infinitive is appropriate. When compliance is not taken for granted, to is used in order to ‘futurize’ the infinitive event’s actualization, i.e. to evoke it as something which the person receiving the request may or may not decide to do. (Duffley 1992: 82-3)

The approach in the present work differs from that of Duffley in that both constructions are taken to be Forward-looking, though one of them profiles the complement situation as being more certain of realisation in the (near) projected future than the other. However, there is no doubt that Duffley is correct in stating


204

that in the to infinitive construction the receiver of the request is normally a free agent with respect to carrying out the complement situation. I take the implication by S1 of a possible alternative to be at the root of the basic semantic difference between the two constructions. It follows from the fact that the to infinitive construction implies a possible alternative that it is used by an S1 who does not necessarily expect S2 to be at his or her beck and call. In contexts where S1 has good reason to expect compliance, the bare infinitive construction is favoured.

6.2.2 Constructions containing help

Mair writes of help that “this verb is a corpus linguist’s delight because its distribution in texts is so clearly influenced by stylistic, contextual, semantic and structural constraints, few of which are categorical in the sense that one variant is excluded in a specific environment” (Mair 1995: 261). Kjellmer (1985: 159) also discusses what he terms “the unusually wide constructional possibilities of help”, which he sees as underpinning its fluctuation between the two non-finite complement forms. The testimony of the corpus evidence in the BNC indicates that these scholars are correct: both bare and to infinitive complements occur with help in most contexts in both same-subject and different-subject constructions. The four constructions containing help have been the subject of several corpus-based studies. Previous to those by Kjellmer and Mair was a study in 1983 by Lind based on data from fifty modern British novels. Data from these three studies are given in Table 6.2. As Lind’s (1983) figures were criticised by Kjellmer (1985: 158) as being unrepresentative of present-day British English, it is interesting to observe how closely the figures from the BNC approximate to them. The data in Table 6.2 clearly show that the bare infinitive constructions are advancing at the expense of the to infinitive ones in both British and American English. There is no reason to suspect that this development is in danger of slowing down. Indeed, Mair (2002) predicts that help is on its way to becoming an auxiliary verb. A purely synchronic study, such as the present one, is obviously at something of a disadvantage when dealing with constructions that are undergoing change (and indeed expansion) to the extent that these are. Nevertheless, the focus of the present study is on non-finite complement constructions as these are represented in the BNC, and it is the state of play around 1990 that we will concentrate on in the case of the help constructions. That the distribution among our downloaded sample is representative of British English at the time is corroborated by Mair’s figures for FLOB.

To infinitives versus bare infinitives: help constructions 205

Table 6.2: Figures for ‘help to infinitive’ and ‘help S2 to infinitive’ compared to ‘help bare infinitive’ and ‘help S2 bare infinitive’ in various corpora, with horizontal percentages

help (S2) to help (S2) bare Total Lind (1983) 80 44% 102 56% 182 LOB: Kjellmer (1985) 94 78% 27 22% 121 Brown: Kjellmer (1985) 55 31% 125 69% 180 FLOB: Mair (2002) 77 39% 122 61% 199 FROWN: Mair (2002) 44 18% 203 82% 247 Present Study (BNC) 189 42% 259 58% 448

Bearing in mind the reservations expressed in the previous paragraph, we will now proceed to examine the two pairs of constructions to see if we can uncover any differences in the ways in which they are used. There are many cases in which help, like its close relation ‘assist’, is used to encode no more than the transference of some skill, but there are also a good many in which the matrix verb subject actually plays an active role (alongside of the complement clause subject in the different-subject constructions), in the effort to realise the complement situation. Some scholars, among them Dixon (1991: 230) and Duffley (1992: 25-29) would have it that the latter sort of situation is normally encoded using a bare infinitive form of complement. Mair (1995) agrees in principle with this assertion but points out that there are many exceptions. He goes on to write:

These and similar semantic constraints should probably not be considered narrowly, i.e. with the intention only of accounting for the distribution of the various patterns of complementation found with the verb help. Rather they are specific instantiations of a more general principle of iconicity in syntactic coding which could be roughly formulated as follows: direct acts of causation and assistance seem to allow a syntactically more reduced type of complement clause than indirect ones, in which assistance is rather like advice in which the enabling condition/cause and resultant state are easy to keep apart. (Mair 1995: 262-3)

Certainly both forms of complement may encode situations of hands-on assistance, as is evidenced by (555) – (556).

(555) All day, they helped Mother to unpack and arrange everything in the

rooms. (FRY 155) (556) Every night she and Ranald helped the old man move four hives up the

four miles of rough track to the heather moor. (APW 37) In (555), the matrix verb subject plays an active role in unpacking and arranging, and in (556) in moving the hives. This sort of usage is commoner in the case of


206

‘help S2 bare infinitive’, with 21 occurrences (15%) in my material than in the case of ‘help S2 to infinitive’, with 10 occurrences (12%). It is only animate matrix verb subjects that are likely to play such an active participatory role, and about half of all occurrences of ‘help S2 bare infinitive’ have such animate subjects, compared to only about a third of all occurrences of ‘help S2 to’. This difference is significant at the level of p= 0.05.13 Figures for number and person of matrix verb subjects for all four constructions are contained in Table 6.3, their TAM relations in Table 6.4.

Table 6.3: Person and animacy of subjects of help in same- and different-subject to and bare infinitive constructions with horizontal percentages

3rd person 1st person 2nd person animate inanimate

Total

help to 2 2% 0 26 25% 77 73% 105 help S2 to 5 6% 3 4% 22 26% 54 64% 84 help bare 4 3% 1 1% 34 28% 83 68% 122 help S2 bare 19 14% 4 3% 45 33% 69 50% 137

Table 6.4: TAM of matrix verb help in same- and different-subject to and bare infinitive constructions, with horizontal percentages, based on 105 tokens of ‘help to’, 84 tokens of ‘help S2 to’, 122 tokens of ‘help bare’ and 137 tokens of ‘help S2 bare’

Present Past Perfect Non-finite

Imper-ative sim prog sim prog pres past

Modal -ing inf

help to

1 1%

25 24%

3 3 %

26 25%

2 2%

6 6%

0 35 33%

4 4%

3 3%

help S2 to

2 2%

15 18%

2 2%

11 13%

1 1%

3 4%

1 1%

27 32%

7 8%

15 18%

help bare

0 4 3%

0 18 15%

0 4 3%

4 3%

21 17%

1 1%

70 57%

help S2 bare

1 1%

9 7%

1 1%

17 12%

0 1 1%

1 1%

30 22

10 7%

67 49%

13 Lind (1983) contains figures for animacy widely at variance with these. 60% of his tokens of ‘help S2 to infinitive’ are animate as are as many as 94% of his tokens of ‘help S2 bare infinitive’. While genre differences may account for some of this discrepancy and the thirty-year gap between the compilations of the two corpora for some more, the sheer extent of the gap in the case of ‘help S2 bare infinitive’ remains puzzling.


The figures in Table 6.3 illustrate two differences between the four help constructions. In the first place the ‘help S2 bare infinitive’ construction appears to be more likely to be used with first-person subjects than its to infinitive counterpart. Secondly, it is more likely to be used with animate matrix verb subjects in general. The raw figures show that it also outnumbers the to infinitive construction in the case of inanimate subjects, but this difference is much less marked. The figures in Table 6.4 reveal a third difference, to wit the incidence of the two constructions with non-finite matrix verbs. The question going abegging is whether these numerical differences can be of assistance in helping us (to) decipher a difference in the basic semantics of the two constructions. The first point to be made about the two pairs of help constructions is that there are many contexts in which the speaker may choose to employ either of them with little difference in meaning. This is most typically the case where there is no doubt that the complement situation is actually realised, as in (557) and (558).

(557) Isobel helped her friend to settle in Paris and find work; one day Maria

asked her if de Chavigny might be interested in buying her jewellery. (C85 2116)

(558) And so he leaned his weight on my shoulder and I helped him walk to his horse. (FR6 743)

The targeted alternative interpretation of the to infinitive presupposes the implication of alternative scenarios. However, such alternatives are likely to be backgrounded when the complement situation is profiled as realised. Still, not every complement situation is profiled as certain of realisation, as (208), repeated here as (559), and (560) will serve to illustrate.

(559) If you find you have a problem and cannot hostess your group meetings as

arranged, PLEASE contact the organiser who will help you, if at all possible, to find another venue nearby to avoid frustration and disappointment for others. (ANM 1645)

(560) And if we can get all of those five right it helps all of us within the group to hopefully get it right first time. (H48 806)

In these examples a degree of uncertainty regarding the eventual realisation of the complement situation is signalled by the adverbials if at all possible and hopefully. There are no instances of such adverbials in ‘help S2 bare infinitive’ constructions in my data. It may be the case that some principle such as Rohdenburg‘s Complexity principle (Rohdenburg 1995: 368) is at work here, the use of the to infinitive being conditioned by the extent of the complement predicate’s remove from the matrix verb. One should, however, point out that adverbials may also occur between the matrix verb and the complement predicate in the bare infinitive construction, but then in the form of process adjuncts, and not disjuncts (“to help you discreetly shape up” (K3P 506), for example). Mair


208

(2003b: 243) maintains that if the adverbial if at all possible were to be dropped from (559), one could also drop the to before the infinitive with little change in meaning. For my part I would maintain that if one were to move, rather than drop, the adverbial, one would still be most likely to use the to infinitive. The reason for this is that the adverbial opens for the non-realisation of the complement situation. I have come across no tokens of ‘help S2 bare infinitive’ which open for the possibility of non-realisation in this fashion. It seems that the bare infinitive construction after help just does not seem to open for the non-realisation of the complement situation. It may be instructive in this connection to compare (561) and (562), which both contain the conjunction ‘or’.

(561) In the day centre, they help you to get into college, or to get work, things

like that. (FR5 1648) (562) This tool can be developed to help nurses scrutinise their work or evaluate

quantitatively the effects of change within a ward on nurse activity. (EE8 627)

In (561) the conjoined to infinitive clauses encode alternative goals, in accordance with the prototypical sense of the complement form. The complement subject may either get into college or get work, but not both. In what may at first sight appear to be the very similar (562), or is not used to join two mutually exclusive situations. In fact, it could be replaced by ‘and’ with little change in meaning. The two conjoined complement situations are simply not presented as alternatives. Having established that the to infinitive construction is preferred when the speaker wishes to profile the object of help as possibly unattainable or as one of several realisable alternatives, let us return to the question raised above as to how the differences in the figures in Tables 6.3 and 6.4 may be related to differences in the semantics of the two constructions. To begin with Table 6.3, we note a considerable difference in the number of occurrences with first-person matrix verb subjects (S1s). As the number of occurrences with second-person animate S1s is quite similar, it is unlikely that the first-person difference is attributable to a difference in genre or register. Why then should speakers, in assuming the role of S1, prefer to encode their helping by using the bare infinitive construction? Let us consider some examples, (563) – (566).

(563) Then in came a basket from Fortnums, which I helped her to open. (CA6 1774) (564) So you want me to help you to produce a plan to solve the whole thing in

the next five or six hours. (AC4 2719) (565) I’ll help you get started, mummy’s gotta get on darling I’ve got tons to do!

(KP8 568) (566) Here, let me help you pick up the bedding and we’ll take it downstairs.

(HHC 2549)


Of five examples of first-person ‘help S2 to infinitive’ in my material, three resemble (563) in being narrative in nature, (564) being the only example taken from dialogue. In (564) the speaker clearly adopts a sceptical attitude as to the feasibility of realising the complement situation in the time span allotted - he or she opens for the possible alternative scenario of not managing to produce a plan. The ‘help S2 bare infinitive’ construction is also used in narrative contexts. In addition, it is the only one of the two used in the downloaded data to encode direct face-to-face offers of help, as in (565) and (566). In contexts like these the use of the to infinitive would involve the subject’s hedging his or her offer of help with an intimation that the complement situation might not be realisable. This would be somewhat paradoxical; at the very least, such a construction would be marked, compared to the bare infinitive one. At all events, the difference in the incidence of first-person subjects in the two constructions would appear to be attributable to the greater tendency of the bare infinitive construction to be employed in making direct offers of assistance. Examples (557) – (566) all contain different-subject constructions. A close examination of all the downloaded occurrences of the two same-subject constructions reveals that these are also used in two main senses, exemplified by (567) and (568) on the one hand, and by (569) and (570) on the other. (There are, of course, many examples situated on a cline between these two extremes.)

(567) Robinson contributed by helping to found the Central Statistical Office in

1941. (CR8 1457) (568) His wife’s grandfather helped found the ILP (Independent Labour Party)

in 1893. (A2J 171) (569) The thunder helped to camouflage the noise of my steps. (HG5 223) (570) Only one quick injection is necessary to help give you the protection you

may need from flu. (A0J 249)

The distinction between the two senses pertains to the domain of force dynamics. Put simply: A help (to) B = (i) A be partially responsible for wholly bringing about B, or (ii) A be wholly responsible for partially bringing about B (567) and (568) exemplify (i): in both cases a complete act of founding took place. (569) and (570) exemplify (ii): in (569) the act of camouflaging was incomplete: in (570) total protection is not conferred by the one injection. Thus the two constructions may both be used to encode two types of situation, which differ in respect of internal and external quantification. Again we see that both constructions may be felicitously employed in most contexts. For what it is worth, an intuition-based examination of the downloaded same-subject tokens revealed only 13 instances (5.7%) where I would, myself, be reluctant to use the alternative construction. Some examples of these are cited as (571) – (573).


210

(571) The Flex Toothbrush has a unique flexible neck which has been created to actually absorb any excessive pressure during brushing which can help to avoid gum damage. (CDJ 42)

(572) The tongue acts as a pump, helping to suck in small forms of water-life and squeeze them down the throat. (BLX 1820)

(573) The row over John Taylor helped lose the seat to the Liberal Democrats, but this time the Tories say their candidate will be chosen on ‘electability’. (K26 2252)

(571) is barely coherent. The subject of help, the antecedent of which, is apparently the purposive infinitive clause beginning to actually absorb... . It is the absorption of excessive pressure that presumably contributes to the avoidance of gum damage. Now the avoidance of gum damage represents a goal that is not likely to be reached once and for all; it remains a potential target. In (572) the sucking and squeezing are goals that are repeatedly sought and repeatedly attained, the ongoing nature of the process being underlined by the use of the -ing form of help (it is presumed that squeeze is also a target of to). In (573), on the other hand, the focus is exclusively on the result, on the loss of the seat in question. This sentence does not encode a goal-oriented activity, the outcome being in no sense the goal of the subject. Of course, in almost all utterances containing help the result is (at least partially) realised. As mentioned above, this fact may well serve to background the implication of latent alternative goals in tokens with the to infinitive, which might, in turn, be expected to lead to a levelling of the distinction between the two complement forms. We have yet to address the question of why the bare infinitive form of complement is preferred after infinitive forms of help. Table 6.4 shows a much greater tendency for the bare construction to be used after infinitive forms of help in both same- and different subject constructions. Table 6.5 contains details of the syntactic/semantic roles of the infinitive matrix verb help in different-subject to and bare infinitive constructions

Table 6.5: Syntactic/semantic roles of the infinitive matrix verb help in different-subject to and bare infinitive constructions, with vertical percentages

(to) help S2 to infinitive (to) help S2 bare

infinitive purpose adverbial 8 53% 36 54% complement of verb 3 20% 17 25% modifier of noun 3 20% 12 18% complement of adjective 1 7% 1 2% “headline” future 0 1 2% Total 15 67


The raw figures in Table 6.5 show that the speaker is some four times more likely to use the bare form in different-subject constructions when the matrix verb is in the infinitive. One may wonder whether the choice of complement form is influenced by the syntactic/semantic role played by the matrix verb itself. Table 6.5 indicates that this is unlikely to be the case. The ratio of ‘(to) help S2 to infinitive’ to ‘(to) help S2 bare infinitive’ only varies from about 1:4 to 1:5 in the first three rows. The fact that the bare infinitive construction is some four times more common with infinitive help than its to counterpart, combined with the fact that there is a good deal of overlap between the situations that may be felicitously encoded using either construction, suggests that the bare infinitive construction is the default choice after infinitive help in most cases. A closer examination of the downloaded tokens would seem to indicate that this default status is particularly prominent in two semantic contexts. The first of these is where the matrix verb subject plays an active part in the realisation of the complement situation, in the sense that he or she literally does some of the complement action.14 Examples (574) and (575) are of this type.

(574) Canada’s Transportation Safety Board sent a team of inspectors to help the

Saudi authorities investigate the crash. (HL9 2762) (575) POLICE are appealing for shopkeepers to help them trace an axe and a

sledgehammer thought to be the weapons used in the murder of Liverpool couple Jean Larkin and Garry Pettitt. (K3T 586)

In (574) the team of inspectors are actually going to do some of the investigating. Similarly, in (575) the shopkeepers are asked to play an active role in the tracing. There are no examples of this sort among the occurrences of ‘(to) help S2 to infinitive’ in my material. The insertion of to into these two utterances would lead to the possible implication of the subject’s providing indirect assistance. Wood states that: “It is never wrong to insert to; it can be omitted only when the “helper” does some of the work, or shares in the activity jointly with the person that is helped” Wood (1956: 107). If not actually “wrong” the insertion of to in

14 Duffley (1992) writes that “The consensus then is more or less that help + bare infinitive is used when the helper participates directly in the activity for which he is giving assistance, while help + to is preferred when the assistance is felt as mediate or indirect. Examples from our corpus confirm these observations but show that the line between mediate and immediate assistance is not always […] easy to draw.” (Duffley 1992: 25). I would concur with the latter statement. My impression is that this distinction, although it plays a role in different-subject constructions, does not do so in same-subject constructions, at least not to any significant extent. Lind (1983: 271) voices similar reservations about the distinction.


212

contexts like (574) and (575) would lead to a change in the possible implications of the utterances. The second area where the bare version is almost an automatic choice is where the complement clause subject plays a non-agentive role in the complement situation. This is the case in 31% of the occurrences of ‘(to) help S2 bare infinitive’ but only 7% of the occurrences of ‘to help S2 to infinitive’. Typical examples are (576) – (578).

(576) They took me down to the labour ward and offered me a sedative to help

me sleep. (G2T 762) (577) We have also arranged optional excursions to help you enjoy the best this

region can offer. (AMD 359) (578) In this chapter we provide a detailed summary as a practical aid, to help

the reader to see precisely what Nietzsche is saying. (H0N 955)

It is commonly accepted that to instantiates a path-goal schema. This lies behind our characterisation of to infinitive complements in terms of targeted alternatives. The fact that over 93% of complement clause subjects in the ‘(to) help S2 to infinitive’ construction are agentive may be related to the fact that Agents are typical movers, in the sense that they initiate their own motion. As such they may progress towards their target under their own steam, as it were. This is not the case with Experiencers. In (576) and (577) sleep and enjoyment come to the complement subject, not the other way round. The one exception to this tendency in my data is (578). It may well be that in this case the use of to is intended to underline the fact that an understanding of Nietzsche is not an easy goal to achieve, and that, even with the proffered help, the target is still some distance away. Turning to the same-subject constructions, the figures in Table 6.4 show that again the bare form of complement is overwhelmingly preferred after ‘to help’.

(579) CHANCELLOR Norman Lamont could still be quizzed by MPs on why

he did not declare an £18,000 donation from anonymous Tory benefactors to help to pay a legal bill. (CBF 12322)

(580) The aid package included US$36,000,000 to help pay off debts amounting to $48,000,000 resulting from the ambitious IPSECO power station project. (HL1 2563)

(581) ‘Failure to do so is failing to try to help save a future victim from this scourge of wanton killing.’ (K2W 842)

We may begin by noting the close similarity between (579) and (580). This shows that the choice of construction in not dictated by the nature of the complement situation. In addition, introspective judgements indicate that one could replace the bare version with the to version in most contexts, and vice-versa, with little alteration in meaning. Why should this be so? Let us contrast ‘to


help to pay’ and ‘to help pay’ by means of the illustrations in Figures 6.1 and 6.2. In these figures both constructions encode the successful achievement of a partial result (hence the shading of some of the rings symbolizing pay in both figures). In both of them the to before help is to be understood as Forward-looking (they are both purpose adverbials) and in both the path extends right up to the target of the complement situation. In Figure 6.2 the second to also encodes a path extending right up to the goal. Now in a situation where the path to a goal is traversed successfully it is perfectly possible to break up this journey into separate legs, but one may well be tempted to ask what is the advantage in so doing. The construction illustrated in Figure 6.2 incorporates an element of redundancy that English speakers apparently prefer to avoid in default or unmarked cases. This is even truer of cases like (581) which contain two paths before help.

Figure 6.1: Schematic illustration of ‘to help pay’

Figure 6.2: Schematic illustration of ‘to help to pay’

s

to

help pay

s

to to

pay

help


214

To sum up, there is no doubt that the corpus evidence points to a considerable degree of overlap between the contexts in which the two pairs of help constructions are used. We have seen that in the case of the two ‘help S2 to infinitive’ constructions, the target encoded by the complement predicate is usually reached by the subject. In other words, the complement situation is always encoded as realised, either partially or wholly. The bare infinitive construction makes no reference to a path at all, merely focussing on the (at least partially) successful result. The semantics of the two constructions are thus very similar and it should not therefore be surprising that in many instances speakers may choose to use either of them to encode one and the same situation. There do, however, appear to be some factors that influence the choice between the two forms. In present-day British English, at least, the choices of speakers seem to conform to the following tendencies:

• If the matrix verb subject is profiled as playing an active, agent-like role in

the carrying out of the complement activity, speakers tend to choose the bare infinitive construction.

• If there exists some doubt as to the eventual realisation of the complement situation, speakers normally choose the to infinitive construction.

• If the complement subject in a different-subject situation is non-agentive, speakers normally choose the bare infinitive construction.

6.2.3 Constructions containing let, allow and permit

Two of these three matrix verbs are very common indeed, as may be seen from the figures in Table 6.6. There are also many situations in which all three may be employed with no very discernible difference in meaning. Examples (582) – (587) have been chosen to illustrate the similarities between the three matrix verbs. (582) – (584) each contain an animate matrix verb subject and the complement clause predicate go, (585) – (587) an inanimate matrix verb subject and the predicate operate. (582) If that's what they want for me, Benny asked herself, why on God's earth

are they allowing me to go to university? (CCM 1797) (583) They permitted me to go down into Tara's Sorcery Chambers for this.

(G10 1235) (584) From the time that I was about ten, I used to look forward to Mr. Golding's

visits, for sometimes on never-to-be-forgotten occasions he would let me go down into the cellar with him. (B22 426)

(585) New York-based, Parallex Graphics Inc has introduced XVideo for Insoft Inc's Communique software conferencing system: XVideo, a real-time full-motion colour system, which allows users to operate in workgroup conferencing situations. (CSX 312)

Constructions with let, allow and permit: introduction 215

(586) High fields permit experiments to operate at high densities, and the higher the density the better the confinement time. (B79 542)

(587) Unix distributor Qualix is due to announce SayIt this week, a new productivity software product that will let Sun users operate their Sparcs by voice commands. (CSA 310)

Table 6.6: Real and projected totals for constructions containing matrix verbs allow, let and permit

Totals per 1000 Projected totals Matrix

verb BNC

to bare to bare allow 33222 467 1 15514 33 permit 3958 239 0 946 0 let 28678 0 775 0 22225 Totals 16460 22258

These six occurrences (582) – (587) were deliberately chosen to illustrate the fact that there exists a certain degree of overlap between the contexts in which all three constructions may be felicitously employed. Hudson et al. (1996) mention let and allow as a pair of words whose meaning is so similar it cannot be taken to motivate their differing patterns of complementation. In this respect they disagree with the arguments put forward in Wierzbicka (1988): indeed, they write that “her arguments rest on subtle differences of meaning which we do not share” (Hudson et al. 1996: 442). In the next two sections I will attempt to cast light on this area of disagreement between Wierzbicka and Hudson et al. by looking, firstly, at the sorts of matrix verb subjects that typically occur with the two types of construction and then at the sorts of permission they typically encode.

6.2.3.1 Types of subject and TAM relations

Tables 6.7 and 6.8 contain data on the person and animacy of matrix verb subjects and on TAM relations respectively. These data show that there are considerable differences in the contexts in which the three verbs are used. There certainly seems no reason to assume that these differences are unmotivated by semantic factors. For example, the data in Table 6.7 show that examples (582) – (584), which contain animate matrix verb subjects, are much more typical of let constructions and (585) – (587), containing inanimate matrix verb subjects, of allow and permit.


216

Table 6.7: Person and animacy of subject of matrix verbs allow, let and permit in to and bare infinitive constructions, with horizontal percentages

3rd person Total 1st person 2nd person animate inanimate

Opaque*

allow 23 5% 18 4% 125 27% 293 63% 9 2% 468 permit 5 2% 10 4% 56 23% 165 69% 3 1% 239 let 208 27% 232 30% 256 33% 19 2% 60 8% 775

*Tokens labelled ‘Opaque’ include some with non-finite matrix verbs as well as tokens where let functions as “a pragmatic particle of imperative or optative mood” in the words of Quirk et al. (1985: 148) Table 6.8: TAM of matrix verbs allow, let and permit in to and bare infinitive constructions, with horizontal percentages based on 468 tokens of allow, 239 tokens of permit and 775 tokens of let



Mod-al -ing inf

allow

11 2%

139 30%

3 1%

59 13%

1 .2%

13 3%

5 1%

73 16%

82 18%

82 18%

permit

4 2%

64 27%

0 45 19%

0 1 .4%

7 3%

40 17%

30 13%

48 21%

let

347 45%

29 4%

2 .3%

96 12%

1 .1%

8 1%

10 1%

107 14%

59 8%

116 15%

Not only does let rarely occur with inanimate matrix verb subjects (S1s), its animate S1 is very frequently the vocative target of a request on the part of the speaker, as may be seen in Table 6.8. Example (588) illustrates this usage. (589) and (590) are examples of the occurrences, labelled ‘Opaque’ in Table 6.7, in which an identity for S1 is difficult to stipulate. (588) If you have any recipes that work particularly well, please let me know for

future editions of this book. (C8P 386) (589) Let me take as a telling example the facts regarding human sexual

dimorphism. (HTP 68) (590) Oh please don't let me blush now! she prayed. (BMW 1841) In (588) the speaker clearly directs his requests at an identifiable addressee, the reader of the book. (589) is an extension of this usage: the request, though still perhaps nominally directed at a particular addressee, is, in fact, purely rhetorical. (590) is even further removed from the central usage. Not only is it not so much a request as an expression of hope (or, in this case, fear), but there is no discernible

Let, allow and permit: types of subject and TAM relations 217

addressee. The matrix verb subject must be interpreted as “the fates” or some such nebulous quantity. The data in Tables 6.7 and 6.8 reveal a considerable degree of similarity between constructions containing allow and permit. There is no doubt that there is a greater overlap between the contexts in which these two can be employed than there is between either of them and let. This is not to say that we may not detect material differences between the two. A closer look at their collocations reveals that permit is much more likely to be used than allow when the matrix verb subject embodies a legal injunction. To put it succinctly, ‘circumstances’ tend to allow, ‘Acts of parliament’ to permit.

6.2.3.2 Force dynamics

The discussion in this section will be restricted to the two most common constructions, ‘let S2 bare infinitive’ and ‘allow S2 to infinitive’. The key question is why one of these constructions contains a to infinitive complement and the other a bare infinitive complement. One theory, advanced by Duffley (1992: 85), among others, is that let always implies the realisation of the complement situation while allow opens for its non-realisation. I think that this may well be true. However, as there were no instances among the downloaded tokens of allow (or indeed permit) of non-realised complement situations other than one actually taken from Duffley (1992), and cited as (594), I hesitate to place too much weight on this one feature. Instead I will investigate the two constructions with respect to force dynamics. Before doing so, I will first exclude from consideration those tokens that occur in contexts where there is no overlap between let and allow, and then proceed to consider more closely those tokens where their usage appears more similar, at least to the naked eye. Most of the tokens to be weeded out do not encode permission as such. In the case of allow, they include (591) in which allow means admit or consider rather than permit. Also excluded are tokens which do not contain an explicit S2, such as (592), the single token of ‘allow S2 bare infinitive’, cited as (593), and one token, (594), which is not an example of actual language usage, but a made-up example from a monograph on the infinitive (Duffley 1992). (591) We can apply the test to the technical and technological subjects, and not

only those, but the professional subjects also; and the boundary line will run now on this side, now on that; but the things that it divides are different in kind, and only on one side of that line lies what we ought to allow to be education. (A69 383)

(592) The transcription start sites for the P A2b and P A3 promoters analyzed by S1 mapping (594) indicated three contiguous possible start sites for each promoter that did not allow to determine which was the first nucleotide incorporated by the RNA polymerase, a problem frequently found with S1 mapping analysis. (FTE 1006)


218

(593) A Fresco ‘canvas’ on to which gliffs can be ‘painted’ will allow developers create things like two-dimensional geometric figures: like X it has some difficulty with three dimensions at the moment. (CSP 167)

(594) He proposes the following contexts to illustrate the contrast between the way let represents permission and the manner in which allow evokes it: (216a) I allowed him to do it, but he didn't do it. (HXG 1001)

Having removed the tokens in question we are left with 462 tokens of ‘allow S2 to infinitive’ that clearly encode permission or its negative counterpart, prohibition. The let construction is more versatile than the allow construction. It is therefore necessary to weed out rather more tokens containing let. These include tokens of the two multi-word verbs ‘let x know’ (= ‘inform x’) and ‘let x have’ (= ‘give x’), illustrated in (588) and (595). Also excluded are a handful of causative tokens containing complement predicates other than know and have, such as (596) in which let it spill means ‘spill it’, first-person plural suggestions, of the sort illustrated in (597), hortatives as in (598), and what the Cambridge Grammar refers to as “open let imperatives”, as in (599) (Huddleston and Pullum 2002: 925). Finally, tokens of the construction in (600), which is restricted to scientific and for the most part mathematical discourse, are also excluded. Common to all these let constructions is the absence of parallel constructions containing allow. (595) I will let you have a list of his customers and I want them contacted, in the

first instance by telephone. (HWP 1159) (596) Checks are made at the pubs to see that the beer is actually drunk, as

serious athletes have been known to let it spill down their vests to cool them off, for which they incur a two minute penalty. (BPK 1162)

(597) Let’s assume one of your employees drinks too much both at work and at home. (A05 29)

(598) But most of all, let brick be your inspiration. (CFN 312) (599) Let Clan Diarmaid rue the day they ever raised hand against Clan

Gillian!’ (APW 144) (600) For: Let U denote the subset of N comprising all those positive integers n

for which the statement S(n) is true. (EV9 566) The total number of let tokens to be excluded amounts to 285, leaving us with 490 tokens which clearly encode either permission or prohibition. In an influential paper on causation, Kemmer and Verhagen characterise permissives as encoding the removal of a barrier preventing the permittee from realising some goal.

A fourth type [of causation], enablement/permission, involves not the exertion of force on an entity to bring about an event that otherwise would not have happened, but the removal by the causer of a conceived barrier that was preventing the causee from carrying out or undergoing the

Let and allow: force dynamics: introduction 219

effected event. Enablement refers to the case where the barrier is physical […] and permission to the case where the barrier is social or sociopolitical in nature […]; we can thus consider enablement and permission as subvarieties of one type. (Kemmer & Verhagen 1994:120)

Figure 6.3 illustrates this type of permission (or enablement), wherein S1, the permitter, removes a barrier which was blocking the path of S2, the permittee, permitting the latter to continue unimpeded on his or her way.

Figure 6.3 illustrates one of two main forms of permission described by Talmy (1986), who distinguishes between what he calls onset letting and extended letting as follows: ”onset letting correlates with the cessation of impingement and extended […] with its nonoccurrence” (Talmy 1986: 76: see also Talmy 2000: 418). While accepting Talmy’s distinction between these two types of permission, I prefer to use the term barrier-removal, based on Kemmer and Verhagen, rather than onset-letting. For the concept which Talmy calls extended letting I will use the term non-imposition (of any barrier). I will also eschew Talmy’s terminology (agonist and antagonist) for the participants in the act of permission, preferring the more specific terms permitter and permittee. The form of permission which I term non-imposition is illustrated in Figure 6.4.

S1

S2

S1 S2

Figure 6.3: Barrier-removal by the permitter (SI) enabling the permittee (S2) to pass


220

6.2.3.2.1 Tokens encoding permission

We begin our comparison with constructions encoding permission. It is only matrix verbs with positive polarity that encode barrier-removal or non-imposition. Negative polarity matrix verbs encode barrier-retention or imposition. These will be discussed in the next section. Table 6.9 contains details of the number of positive and negative polarity matrix verbs in the downloaded samples.

Table 6.9: Constructions containing positive and negative polarity active voice matrix verbs allow and let with horizontal percentages

Totals Percentage totals Matrix verb BNC

Positive Negative Positive Negative

allow 33222 414 48 89.6% 10.4%

let 28678 372 118 76.0% 24.0%

All positive polarity tokens were examined with a view to determining whether they encoded barrier-removal or non-imposition. This sometimes involved a considerable amount of trawling in the co-text in an effort to ascertain the possible prior existence of barriers. In other cases the immediate context contained sufficient information to conclude that such a barrier existed. One possible source of such information is the presence of temporal adverbials like later, this time and for once in examples (601) – (603). (601) The US pilots later allowed an Iraqi search-and-rescue helicopter to fly to

the crash site and then return to its base. (CBE 784) (602) ‘Maybe it was practising,’ Jenna said shortly, and this time when she

walked off he let her go. (HGD 2931)

S1

S2

S1 S2

Figure 6.4: Non-imposition of barrier by the permitter (SI) enables the permittee (S2) to pass

Let and allow: force dynamics: tokens encoding permission 221

(603) Elinor was silent as she allowed herself to see, for once, the soft and gentle cobweb of romance she had carefully woven around her life with Billy. (FPB 406)

The adverbial later in (601) indicates that the permitter in question had previously forbidden the permittee to carry out the flight in question. Similarly in (602) the use of the adverbial this time presupposes the presence of a previous barrier erected by the permitter to the permittee’s leaving. The barrier in the reflexive construction in (603) was self-imposed by the permitter, who decides herself to lift it on the occasion in question. Information similar to that encoded by the adverbials in (601) – (603) may be encoded by adjectives, as in (604) – (605).

(604) In an attempt to remedy this the SLORC introduced new banking laws in

July 1990 which allowed foreign banks to open branches in Myanma. (HLD 4402)

(605) New divorce legislation reported on Dec. 9 was said to grant unprecedented rights to women, allowing divorcees to seek compensation through the courts for housework which husbands had ordered them to do during the marriage. (HLT 1639)

The use of the adjective new in (604) implies the earlier existence of laws prohibiting the operation of foreign banks. In (605) both new and unprecedented contribute to establishing the implication that there used to be a barrier to the seeking of compensation on the part of divorcees. It is not only in the case of utterances containing adverbials and adjectives in the matrix clause, as in (601) – (605) that it is possible for us to infer the prior existence of barriers. In many cases we can make a similar inference based on other sorts of information in the immediate context. Consider in this respect (606) – (608). (606) She allowed herself to feel all the pain she'd denied herself for so long. (HGM 851) (607) The kind of strength you need to develop is that which allows you to move

your body or limbs through a short, fast contraction. (A0M 394) (608) ‘When do you think they’ll let us go home?’ asked Williams at the end of

the questioning. (CML 2145) In (606) it is the presence of the adverbial ‘for so long’ in the relative clause that allows us to infer the previous self-imposed barrier to the feeling of pain. In (607) it is the relative clause ‘you need to develop’ which shows the existence of a barrier in the form of a lack of the requisite type of physical strength. In (608) it is the uncertainty implicit in the enquiry ‘when do you think?’ that implies the present existence of a barrier to the participants returning home.


222

In yet other cases we can rely on our general world knowledge to enlighten us with respect to the possible existence of a prior barrier. (609) – (611) are cases in point.

(609) Claudia relaxed her fingers, letting the pencil drop to the desk. (H8J 2708) (610) WHEN that grand old man of the turf Jim Joel died last spring at the age

of 97, a condition in his will allowed the Queen Mother to choose any horse she wished from his estate. (K97 13871)

(611) THE father of Hillsborough coma victim Tony Bland pleaded yesterday: Let my son die with dignity. (CEN 6140)

In (609) our knowledge of the function of taut fingers as a container of objects allows us to see the causal relationship between the action of relaxation and the falling of the pencil. We know that prior to their being relaxed the fingers constituted a barrier to the pencil’s falling. In interpreting (609) we make use of our knowledge of the physical world. In (610) we employ our cultural knowledge. We know that people are not entitled to choose items from another person’s estate unless permission to do so has been expressly given. Example (611) also involves our broader cultural knowledge, in this case our familiarity with the function of life-support machines. The form of the request allows us to make use of this knowledge to infer the existence of a barrier to the peaceful death of the Hillsborough victim. A final locus for inference involves cases where the permitter is encoded by a sine qua non condition, as in (612) – (614).

(612) If you re recall back in nineteen eight five Tony the Government brought

in the transport bill which let operators compete. (KM8 236) (613) She says the settlement, which works out at £252,000 a year, will allow

her to keep her maid, cook and secretary and help with the costs of a plush Milan apartment. ( CEM 770)

(614) This process allows the harsh tannin to mellow and the wine to develop complexity. (C9F 2161)

(612) is similar to examples (601) – (603) in that it contains a temporal adverbial, “back in nineteen eight five”. A natural inference is that it was the introduction of the bill which allowed the operators to compete. (In the ‘real world’, of course, it was the passing of the bill into law, not its introduction, that conferred the permission on the operators in question!) However, the presence of the adverbial is not necessary for us to make the requisite inference. The very fact that it is the bill that is the permitter implies the prior impossibility of competition, in other words the existence of an earlier impediment. Similarly in (613) without the settlement in question the woman would have had to let her servants go. Thus the prior lack of these funds amounted to a barrier to her retaining her staff. And in (614) without the process in question the tannins would not mellow.


Examples (601) – (614) all encode situations of barrier-removal. To categorise them as such it is sufficient to identify the prior existence of a barrier, which may either be implicit or explicit. The prior non-existence of a barrier is less easy to stipulate, for obvious reasons. We may sometimes draw on our world knowledge, as in the case of (615). More often we must trawl the co-text before we can conclude that no such barrier existed.

(615) With the tension reaching boiling point, it was finally announced that the

French officials had allowed the result to stand and they had to be applauded for a sporting decision. (A40 42)

(616) In mid-stream Meg let Ben take the oars from her, changing seats with him nimbly as the boat drifted slowly about. (FRF 1677)

(617) So we let the blacks come down to us, we didn’t go looking for them. (FAY 933) We can infer from (615), without searching the co-text, that the officials in question had the power to alter the result but chose not to exercise this power. In other words (615) is an instance of non-imposition. An extensive search of the text in (616) did not reveal any prior wish on the part of Ben to assume the task of rowing. In this case a paraphrase without a verb of permission, such as ‘Meg handed the oars to Ben’ would be more felicitous than one implying a previous desire on his part to take over. Similarly (617) does not imply a prior prohibition on the descent of ‘the blacks’. It merely states that the permitters did not themselves make any effort to seek them out. Tokens such as (616) – (617) may appear at first sight to be ambiguous. However, this sort ambiguity usually evaporates, as it does in the case of both these tokens, when one conducts a thorough examination of the co-text. Whenever such an investigation revealed no clue as to the previous existence of a barrier to the realisation of the situation encoded in the complement clause, the token in question was labelled as encoding non-imposition. Thus all tokens containing positive polarity matrix verbs were assigned to one of two semantic classes. One might object that the overall picture yielded by this analysis is too black-and-white. However, the question of the presence or absence of a barrier is a black-and-white question. Either such a barrier existed, or it did not. If it existed one may expect it to have been either explicitly mentioned or at least implied by the speaker. Having distinguished between barrier-removal and non-imposition, and described the criteria used to classify tokens as instantiating one or other of these two types of permission, we will now proceed to consider the extent to which the two constructions encode both types of permission. The answer is shown in Table 6.10 and Figure 6.5. The table indicates the number of tokens in the downloaded samples, the figure projected totals for the BNC as a whole.


224

Table 6.10: Constructions containing positive active voice matrix verbs allow and let encoding barrier-removal or non-imposition with horizontal percentages

Totals per sample Percentage totals Construction

barrier-removal

non-imposition

barrier- removal

non- imposition

allow to inf. 365 49 88.2% 11.8%

let bare inf. 81 291 21.8% 78.2%

0

2000

4000

6000

8000

10000

12000

14000

barrier-removal non-imposition

allow

let

Figure 6.5: Projected totals for tokens encoding barrier-removal versus non-

imposition with positive matrix verbs let and allow The totals in Table 6.10 and Figure 6.5 are striking to the naked eye. The difference between the two constructions with respect to encoding barrier-removal or non-imposition is, as one might expect, significant statistically (at the level of p= 0.0001). We can therefore safely conclude that the fact that allow and let in our data predominantly encode barrier-removal and non-imposition respectively is not a matter of mere chance. Figure 6.5 is an eloquent illustration of the fact that there is a considerable difference in how the two constructions are used. Nevertheless, there is, as we have already noted, a certain degree of overlap between them. Exactly how much overlap is there? We have seen, for instance, that let encodes barrier-removal in examples (602), (609), (611) and (612). Of 81 tokens of let encoding barrier-


removal, as many as 25 contain the predicate go, as in (602) and (618) – (619).15 Of 382 tokens of allow, on the other hand, just one, cited as (620) contains the predicate go (there are two tokens of non-imposition ‘allow S2 to go’). (618) At last he let her go, and, puffing happily, she straightened her kerchief

and skirt. (CDN 274) (619) We went on a few yards, then I held back to let her go ahead of me where

the path was narrowest. (GV2 3746) (620) ‘It's the number of Africans the UN allows to go into space,’ he said. (HH3 4323) Although (620) is clearly an instance of barrier-removal rather than non-imposition, it does not encode the release sense, which is the most common meaning of ‘let S2 go’, as in (602) and (618). There is thus very little overlap between the two constructions with the complement predicate go. Of the remaining 56 examples of barrier-removal encoded by let, another 27 contain motion verbs. There are also 27 occurrences of motion verbs other than go among the 382 tokens with allow. Of these there are seven predicates that collocate with both allow and let in my material. One should, however, note that the BNC as a whole contains tokens with allow containing all the motion complement predicates that occur with let. The fact that they did not all surface in the randomly downloaded databases merely serves to indicate that they are not particularly common. The seven motion predicates other than go common to both constructions in my data are come, enter, fall, flow, fly, leave and move. (621) – (624) exemplify the latter two predicates in both constructions. (621) Many of the people on my courses on dying, for example, had never really

come to terms with the inevitability of death in their own lives, and many a time we had to stop to allow distressed and upset people to leave the room. (CCE 1552)

(622) The East German government’s decision to let the refugees leave across their own country came as a surprise to many, including some West German diplomats. (A1G 399)

(623) You can unhook by taking the weight of the rig back on to your arms, thereby allowing the rope to fall out of the hook. (AT6 905)

(624) She let her cloak fall back from her head in surprise. (FRE 328) The predications in (621) and (622) seem very similar in meaning. In this case there would appear to be a genuine overlap between the use of the two constructions. The same point does not apply to (623) and (624). If one were to

15 There were also 29 examples of the same-subject ‘let go of’ construction, as in “He let go of Leila who rubbed her wrist, holding it close to her chest.” (BNC AD9 4023) among the 1,000 downloaded tokens of let.


226

substitute allow for let in (624) one would be ascribing a greater degree of agentivity to the permitter than is implied by the utterance with let. This pair of sentences is, however, unusual in this respect. The majority of tokens encoding the permission of motion with predicates other than go resemble (621) and (622) with respect to the meaning equivalence of the two constructions. One can also detect the same sort of meaning equivalence (or rather it is hard to detect any substantive semantic or pragmatic difference) between the barrier-removal constructions containing let with non-motion predicates and similar tokens containing allow. There are only 29 instances of such tokens with let among the 81 tokens encoding barrier-removal. The only complement predicate to occur more than once in my data is listen, of which there are three examples. Other predicates encoding physical or mental action include dress, drink, seize, think, wear and write. Two others, borrow and buy, involve transfer, either physical or legal. (625) and (626) exemplify the let and allow constructions with buy.

(625) They should let them buy more land and do more council building. (KCG 807) (626) ‘The rents-to-mortgage scheme would allow tenants to buy at least half

their property from day one by paying the existing rent.’ (K55 6748) The predications in (625) and (626) seem very close in meaning. The same point applies to the let tokens containing the other non-motion complement predicates and utterances containing allow which may be found in the BNC as a whole, if not in the randomly downloaded samples. To sum up the question of overlap between the two constructions in cases where let encodes barrier-removal, we can conclude that, although a certain amount of overlap certainly exists, there is less than might at first appear to be the case given the figures in Table 6.10. This is because of a tendency to avoid using go with allow, at least in the sense of release, presumably a by-product of the degree of entrenchment of the ‘let S2 go’ construction. We turn now to the other case of overlap, instances where allow overlaps with let in encoding non-imposition. According to Table 6.10, there are 49 instances of non-imposition encoded by allow. In contrast to the case of barrier-removal encoded by let, there does not appear to be a single category of complement predicate that stands out among the various utterances. Predicates used include verbs of motion such as go, move and walk, non-motion activity verbs like work and demonstrate and accomplishment verbs like cool and wilt. (515), repeated here as (627), and (628) are typical examples of allow used to encode non-imposition.

(627) This will allow it to continue being part of a household, rather than being

confined on its own in kennels. (CJE 229)


(628) Thus, by allowing the divisions within his camp to exist and, at the same time, constantly insinuating that, without him, pre-war ‘chaos’ would return, Franco succeeded in creating the belief that he was not just indispensable to political stability, he was consubstantial with it.

(HPV 552) (627) is about getting a neighbour to look after a dog. It thus encodes non-imposition in the sense that the dog is not hindered from continuing as part of the household. In (628) it is implied that Franco could have quashed the divisions within his camp should he have so wished, but that he refrained from doing so: again, a clear case of non-imposition. In the case of both these tokens it would be possible to construct minimal pairs containing let which would be extremely difficult to distinguish either semantically or pragmatically from the originals. We are thus again faced with real overlap in usage here. As an indication of the extent to which the non-imposition tokens containing allow resemble those containing let¸ we can consider the type of permitters and permittees we find in the two constructions. Figure 6.6 contains details of the animacy of permitters (SI) and permittees (S2) in both let and allow constructions encoding barrier-removal and non-imposition.

0,0

10,0

20,0

30,0

40,0

50,0

60,0

70,0

80,0

90,0

100,0

let non-imposition let barrier-removal allow non-imposition

allow barrier-removal

animate S1

animate S2

Figure 6.6: Percentage of animate permitters (SI) and permittees (S2) in

constructions with let and allow encoding non-imposition and barrier-removal.

Figure 6.6 clearly shows that when the allow construction encodes situations of non-imposition, it is more likely to occur with animate permitters and permittees than when it encodes barrier-removal. The difference is less marked in the case of let. This is because of the restricted type of situation prototypically encoded by barrier-removing let, i.e. situations involving release, where both the releaser and the released are typically human.


228

To sum up this section on the degree of overlap between the two constructions, there is no doubt that such overlap exists, but it is not as extensive as may at first appear from Table 6.10, and certainly by no means as great as implied by Hudson et al. We would seem justified, on the basis of the evidence presented in this and the previous section, and the projected totals for the two constructions contained in Table 6.6, in concluding that:

• situations encoding barrier-removal are likely to be encoded by allow, unless

• they encode situations of release, when they are likely to be encoded by let.

• In the case of situations involving motion other than release, it is impossible to predict which of the two constructions will be employed.

• Situations encoding non-imposition are likely to be encoded by let. • In a minority of cases such situations may be encoded by allow. It is not

possible to predict which situations are likely to be encoded by allow.

6.2.3.2.2 Tokens encoding prohibition

In the previous section we were concerned with positive polarity matrix verbs in constructions encoding barrier-removal and non-imposition. Constructions containing negative polarity matrix verbs let and allow, on the other hand, encode either barrier-retention or barrier-imposition. These two forms of (refusal of) permission are illustrated in Figures 6.7 and 6.8.

S1

S2

S1

S2

Figure 6.7: Retention of barrier by S1 prohibits S2 from passing

Let and allow: force dynamics: tokens encoding prohibition 229

The criteria for distinguishing between barrier-retention and barrier-imposition are similar to those used to distinguish between barrier-removal and non- imposition presented in the previous section. These will not be enumerated again here. We begin by looking at barrier-retention as encoded by not allow. (629) It is our interests, rather than those of a degenerate and selfish minority,

that the police should protect; and if the law at present does not allow them to do so then the law must be changed. (C88 1105)

(630) After all we don’t allow employers to take money out of our bank accounts and they have no right to take money out of our pension funds!

(HDP 175) (631) In the Wedgwood Benn Case 1961 the House of Commons refused to

allow Tony Benn to take his seat although he had been duly elected. (EVK 872) In (629) the adverbial at present refers to a barrier that has been in existence for some time – in other words it encodes the extended rather than the instantaneous sense of the simple present. If one were to omit the adverbial, the sentence would still be understood to encode barrier-retention. This is because of the generic nature of the prohibition referred to. (630) also encodes a generic prohibition, the barrier in question being of a permanent nature. (631), on the other hand, refers to a specific act of prohibition. In its case, we must make use of our world knowledge to interpret it as encoding barrier-retention rather than imposition. The reason why Benn was not allowed take his seat was because of a general prohibition in existence at the time against members of the nobility sitting in the Commons.

Situations involving barrier-retention may also be encoded by not let, as in (632) – (634). (632) They don’t let women drive cars, let alone fly an aircraft. (BNV 987)

S1

S2

S1

S2

Figure 6.8: Imposition of barrier by S1 prohibits S2 from passing


230

(633) The Foreign Office promises the EC ‘is not going to let this drop’. (CAG 1347) (634) Photography wouldn’t let him go, however, and after serving out his

apprenticeship with various studios he went freelance, working to commission and supplying picture libraries with travel shots. (CRP 474)

(632) and (633) resemble (629) and (630) in encoding general validity predications. In (632) the authorities have imposed a general prohibition on driving by women, and in (633) the EC a general prohibition on the dropping of the idea in question. (634) contains the release sense of ‘let S2 go’, metaphorically extended to the domain of the mind. The prohibitor, in this case ‘photography’, has already a strong grip upon the mind of the participant whom we may term the ‘prohibitee’. This grip takes the form of a barrier blocking the way for the prohibitee to entertain alternative careers. While not let may encode barrier-retention, it is more common for it to encode barrier-imposition as in (635) – (638). (635) ‘I am not letting anyone have my car,’ I said. (ED9 1298) (636) ‘Don’t let her get away, Tim!’ he shouted. (B0B 478) (637) On the way home in the taxi (Nigel was indulging them for once - he

couldn’t let Gina use her bike under the circumstances) he took delight in telling her she had a spot on her backside. (AC3 2043)

(638) I’m right behind you, but you mustn’t let the parents hear. (H8Y 906) The use of the present progressive in (635) indicates that the prohibition is being imposed at the time of speaking. Similarly, the imperative in (636) encodes the imposition of a prohibition. The imperative is very common indeed with both positive and negated let as may be seen in Table 6.8. When a positive matrix verb is in the imperative it may either encode barrier-removal (‘let her go’) or non-imposition (‘let her be’), the latter being more common than the former.16 (637) and (638) both contain modalised matrix verbs. In (637) the adverbial ‘under the circumstances’ implies that the prohibition in question was not a permanent one. This is thus an instance of barrier-imposition rather than retention. (638) is similar in meaning to the imperative (636), differing from it in that the illocutionary force of the prohibition is buttressed by the implications of the existence of an obligation on the part of the prohibitee. The fourth and final combination of form and meaning, barrier-imposition encoded by not allow, is exemplified by (639) – (641).

16 Apart from the imperative the other TAM relation that stands out as being

markedly different with let and allow is the simple present. The ‘allow to’ tokens in question most commonly encode generic situations of barrier-removal with a sine qua non type permitter, as in (613).

Let and allow: force dynamics: tokens encoding prohibition 231

(639) After the feud he refused to allow Jamila to visit her parents. (A6V 790) (640) Unlike his predecessor, he has told Mrs Rees she cannot allow her toys to

be sold. (ACR 2289) (641) He could not allow himself to fall in love with a girl so obviously a part of

the world of wealth and consequence which he had abandoned. (HE 2369) In (639) the time adverbial “after the feud” implies that there was no prior barrier to the girl’s visiting her parents. The contrastive adverbial “unlike his predecessors” in (640) serves a similar function. Finally, the self-imposed prohibition in (641) refers to a particular girl rather than girls in general, and there is no implication that the prohibitor was generally reluctant to engage in affairs of the heart. We have now seen that, just as both barrier-removal and non-imposition may be encoded by constructions containing positive matrix verbs let and allow, both barrier-retention and imposition may be encoded by constructions containing negated matrix verbs let and allow. Table 6.11 and Figure 6.9, which may be compared to Table 6.10 and Figure 6.5, contain details of how often the two constructions are used to encode the two types of prohibition. The totals in Table 6.11 and Figure 6.9 indicate that there is a greater degree of overlap between the two constructions with negated matrix verbs than was the case with positive ones, as shown in Table 6.10 and Figure 6.5. Nevertheless, the difference between the two constructions with respect to encoding barrier-retention or imposition is still significant statistically at the level of p= 0.01, indicating that the two are by no means interchangeable.

Table 6.11: Constructions containing negated active voice matrix verbs allow and let encoding barrier-retention or imposition

Totals per sample Percentage totals Matrix verb

barrier-retention

imposition barrier-retention

imposition

allow 21 27 43.8% 56.3%

let 25 93 21.2% 78.8%


232

0

500

1000

1500

2000

2500

3000

3500

barrier-retention imposition

allow

let

Figure 6.9: Projected totals for tokens encoding barrier-retention versus

imposition with negated matrix verbs let and allow Taken together, Tables 6.10 and 6.11 provide eloquent testimony to there being a clear difference of meaning between the permissive constructions containing let and allow. The former prototypically encodes non-imposition or imposition: the latter prototypically encodes barrier-removal with positive matrix verbs but is equally likely to encode barrier-retention or imposition with negated matrix verbs. The evidence of the present section, in conjunction with the evidence of section 6.2.3.1, thus allows us to refute the claim of Hudson et al. (1996) that the meaning of let and allow is so similar it cannot be taken to motivate their differing patterns of complementation. The next section will attempt to relate the syntactic differences in complementation form to these differences in meaning.

6.2.3.2.3 Why ‘allow S2 to infinitive’ but ‘let S2 bare infinitive’?

In this section I will attempt to relate the fact that allow and let differ with respect to the form of infinitive they govern (or co-occur with) to the general sense of these two complement forms. In the present study both ‘allow S2 to infinitive’ and ‘let S2 bare infinitive’ are classified as Forward-looking constructions. This view is not shared by all scholars. Duffley (1992), for example, classifies ‘let S2 bare infinitive’ as a Same-time construction. He writes: “Let represents permission as non-intervention, i.e. as not obstructing the accomplishment of the event expressed by the infinitive, and so letting cannot be conceived as coming before the event permitted (indeed one cannot say that one has let someone do something until they have actually done it)” (Duffley 1992: 88). We saw,

Why ‘allow S2 to infinitive’ but ‘let S2 bare infinitive’? 233

however, in the previous section that let, although it prototypically encodes non-intervention (non-imposition), may also encode barrier-removal. Certainly when let encodes barrier-removal the data includes many examples in which the letting clearly precedes the carrying out of the action permitted. This is the case in both (642) and (643). (642) They should let them buy more land and do more council building. (BNC 807) (643) They let me wear a maternity dress then, because I hadn't worn one up till

then. (FU1 1095) That (642) is an example of the barrier-removal sense of let is indicated by the modal should. Clearly any possible purchase of land would necessarily be subsequent to the lifting of the barrier. Nor is there any implication in (643) that the girl donned the garment in question at the moment permission was given. Indeed quite the opposite was the case. As is made clear by the co-text in (644), she had no such garment in her possession. (644) They let me wear a maternity dress then, because I hadn't worn one up till

then. The thing that upsets me when I see all these pregnant people walking around is that I didn't do it. I never really was officially pregnant. I didn't wear maternity clothes. (FU1 1095)

(644) may be taken as evidence against Duffley’s contention cited above that that “one cannot say that one has let someone do something until they have actually done it”. In (644) not only does the permission precede the actualisation of the situation permitted, but this situation may in fact never have been actuated.17 If this is the case it would serve as evidence against another contention of Duffley’s, which is that while allow may open for the non-realisation of the situation permitted, let never does so. He writes: “As Cotte [1982] points out, allow and let evoke permission in different ways, the latter seems incapable of evoking the giving of permission without evoking at the same time the realization of the action permitted […], whereas allow is not subject to this restriction” (Duffley 1992: 85). Mittwoch makes the same point, going so far as to maintain that “the contrast between let and allow or permit is particularly striking in this respect” (Mittwoch 1990: 117). There was not a single instance among the tokens of either let or allow in which the complement situation was encoded as definitely not realised. There were, on the other hand, many tokens in which it was encoded as definitely realised. These include all tokens encoding non-imposition, irrespective

17 This depends on our interpretation of the statements “I didn’t do it” and “I

didn’t wear maternity clothes”, whether we understand them as preterite or pluperfect.


234

of whether the matrix verb is let or allow, and most of the tokens encoding barrier-removal, as in (645) and (646). (645) Once safe in the haven of the kitchen, she allowed George to slip from her

grasp and ran, sobbing, to her mother. (C98 823) (646) The handshake was a stupid idea as she was still holding Darren, but she

let him slide down to the floor and we shook. (A0F 1148) There is no doubt that both (645) and (646) encode barrier-removal. Equally there is no doubt that in both cases this removal leads to the definite realisation of the complement situation. However, unlike the case of non-imposition, it is possible for tokens encoding barrier-removal to leave open the question of the eventual non-realisation of the complement situation. Consider in this respect (647) – (652). (647) Apart from an interval for lunch, the meeting continued until 3.30 p.m.

and then adjourned until 7.00 p.m. to allow representatives to report to their parties and organisations. (CCC 985)

(648) Sharpe had stopped at the ford to let the horses drink. (CMP 1414) (649) While waiting for this money to come through, he asked the owners if they

would allow him to order goods for the coming summer season. (B17 762)

(650) Well if they would let us borrow it, that's a different matter to buying it from then because we then wouldn't make any money on it. (D97 649)

(651) Its data-holding potential allows it to be used to help ensure that all pupils are involved in the game as well as the traditional competitive classroom activity, testing. (EUW 198)

(652) Two of them, cc:Mail and Notes, work together on office networks, letting employees collaborate as they shuffle documents around electronically - a market growing by 60% a year. (CR9 2246)

In (647) even though the barrier to the representative’s reporting to their respective parties has been removed by the meeting’s being adjourned, there is no guarantee that they will actually do so. Similarly in (648) even though Sharpe removes the barrier to the horses’ drinking by stopping there is no guarantee that they will do so. As is well known, one can lead a horse to the water…..! (649) – (652) may be analysed in a similar fashion. We may conclude that insofar as there is any difference between the let and allow constructions with respect to the eventual realization of the complement situation, this difference is related to a greater tendency for allow to be used to encode barrier-removal. However, in the case of both verbs the situation permitted is generally realized by the promissee. In all Forward-looking constructions containing the bare infinitive form of complement the situation in the complement clause is profiled as (almost) certain of realisation in the projected future. This is true of the construction in section 6.2.1 containing bid. It also true of the let construction. Figure 6.4 illustrates non-

Why ‘allow S2 to infinitive’ but ‘let S2 bare infinitive’? 235

imposition, which is the prototypical sense of let. In it the permittee (S2) progresses past the barrier without any interruption. This progression is automatic. In other words the employment of the bare infinitive in the case of let, at least when encoding non-imposition, is motivated by the schematic sense of this complement form. This is also the case with the ‘allow S2 to infinitive’ construction. We saw in section 6.2.3.2.1. that this construction prototypically encodes barrier-removal, as illustrated in Figure 6.3. In cases of barrier-removal there is a very real contrast between the situation pertaining before and after the lifting of the barrier. The alternative of the barrier’s not being lifted is implied in a way that is not the case with non-imposition. It is the implication of this alternative, I would suggest, that motivates the employment of the to infinitive to encode situations of barrier-removal. This interpretation of the rationale behind the use of the two different forms of complement with let and allow begs three questions. The first is why some tokens of barrier-removal are encoded by let. The second is why some tokens of non-imposition are encoded by allow. The third is why barrier-imposition, which resembles barrier-removal in the foregrounding of an alternative situation, is normally encoded by let rather than allow. To begin with the third question, there is no doubt that barrier-imposition carries an implication of an alternative situation in a way that barrier-retention does not. One might therefore expect it to be encoded by a to rather than a bare infinitive clause. In fact barrier-imposition patterns syntactically with non-imposition and barrier-retention with barrier-removal, at least in the case of let. It would appear that the negation in the not let construction is external to the whole predication in the complement clause, in other words that barrier-imposition is interpreted as not (non-imposition). Similarly, barrier-retention is interpreted as not (barrier-removal). This interpretation is, no doubt, motivated by the fact that the positive polarity constructions are much more common than their negated counterparts, as shown in Table 6.9. Note however that not allow is more likely to be used to encode imposition than its positive counterpart to encode non-imposition. This tendency is presumably motivated by the foregrounding of the alterative situation of non-imposition, as illustrated in Figure 6.9. The next question that needs to be addressed is how the fact that ‘let S2 bare infinitive’ sometimes encodes barrier-removal harmonises with the thesis advanced here that the to infinitive form is more suited to express this sense, implying as it does the existence of a latent alternative to the situation in the complement clause. We saw in section 6.2.3.2.1 that the ‘let S2 bare infinitive’ construction may be employed in complement situations involving motion. Most of the tokens of ‘let S2 go’ encode a situation involving the sudden release by the permitter of the permittee. If we think in terms of Figure 6.3, the permittee (S2) has his or her nose right up against the barrier, so to speak, prior to the permitter’s lifting it. As soon as the barrier is lifted, the permittee is on his or her way. The bare infinitive form of complement was said in section 4.6 (p. 142) to encode a situation that is profiled as certain to occur in some domain. In force dynamic terms the lifting of the barrier in the ‘let S2 go’ construction guarantees the


236

occurrence of the situation permitted. Thus the employment of the bare infinitive complement with let in this type of situation is motivated by the meaning of the complement form. This is not to deny that a to infinitive construction would be equally well suited to encode the situation in question. However, when it comes to situations involving release, it would appear that the possible use of go with allow is pre-empted by the degree of entrenchment of the ‘let S2 go’ construction. There is, on the other hand, a certain degree of overlap in the encoding of situations involving release with other complement predicates, such as leave, in (622) and fall in (624). It is exactly in cases where there exist several motivated possibilities for encoding a situation that one would expect to encounter overlap in usage. And it is in just these areas that overlap actually exists, as was shown in section 6.2.3.2.1. The third and final question to be addressed is why the ‘allow S2 to infinitive’ construction, which prototypically encodes barrier-removal, is occasionally used to encode non-imposition. As was pointed out in section 6.2.3.2.1 there is no single type of complement predicate that stands out in these cases. The schematic sense of the to infinitive construction was alleged in Chapter 4 to carry an implication of the existence of a latent alternative to the situation encoded in the complement clause. This alternative may be more or less foregrounded. In constructions like ‘choose to infinitive’ or ‘prefer to infinitive’ the alternative may well be explicitly mentioned in the co-text. In other cases it may be more in the background. In the case of the allow construction the alternative may be focused upon, as in cases where the existence of a previous barrier is implied by a temporal adverbial, such as in (601), or an adjective, as in (604), or it may be defocused, as it is when allow encodes non-imposition. In the prototypical instances encoding barrier-removal the sense of an alternative tends to be more present, closer to the surface, as it were, than in the case of more peripheral instances, encoding non-imposition. To sum up this section, the fact that ‘let S2 bare infinitive’ prototypically encodes non-imposition and ‘allow S2 to infinitive’ prototypically encodes barrier-removal may be related to the schematic sense of these two complement forms. We have also seen that more peripheral uses of the two forms can be explained with reference to the existence of alternative motivations for the encoding of certain situations by one form or the other.

6.2.3.3 Let and allow: a summary

In this section we have seen that the contention of Hudson et al. (1996) that let and allow are to all intents and purposes interchangeable is not supported by the evidence of the corpus data. There are, indeed, clear differences in meaning between the constructions in question, both with respect to the types of subject with which they typically occur and with respect to force dynamics. Such differences do not necessarily preclude their employment in similar contexts. While the matrix verb let prototypically encodes non-imposition, allow prototypically encodes barrier-removal. The difference in the meaning of the

Let and allow: a summary 237

matrix verbs motivates the employment of the former with bare infinitive complement form which prototypically encodes a situation as certain to occur in some domain, and the latter with the to infinitive form which prototypically encodes a situation as the more/most likely of two or more alternatives in some domain. In the case of both let and allow the domain in question is argued to be that of the projected future. Thus in the case of one of the pairs of constructions listed by Hudson et al. their objections to the thesis of Wierzbicka (1988) that grammar is not semantically arbitrary have been shown in this section to be unfounded.

6.2.4 Constructions containing dare and need

There are two Applied Attitude matrix verbs, dare and need, that take both bare and to infinitive complements. (Need also takes an -ing complement but this construction involves trajector-landmark realignment and will be discussed at the end of this section with other constructions that instantiate this semantic switch.) We will look at these two matrix verbs in turn, starting with dare, exemplified initially by (653) – (655). (653) During my first stay in China in 1984, it was very rare indeed to see young

people holding hands in public and I clearly remember the expressions of shock when a student dared to ask, in a packed lecture hall, if students in Britain were allowed to ‘talk love’. (CG0 1149)

(654) But she didn’t dare ask him about Spiderglass and what he was doing there. (FP0 256)

(655) because er so few, so few of you have admitted to doing it, well five people here have, dare I ask if any one would like to say whether or not it was successful, yes. (FLB 218)

Dare is sometimes described as a marginal modal auxiliary. Such auxiliaries fall outside the scope of the present study. One might therefore raise the question as to whether we ought not to exclude dare completely from our remit. Although in some doubt, I have chosen not to do so because of its close semantic resemblance to ‘venture to infinitive’, which is certainly not an auxiliary construction.18 When dare clearly functions syntactically as a full auxiliary, when, that is, it lacks otherwise-obligatory do-support, as in (655), I have deemed it to fall outside the scope of the study proper. I mentioned above that ‘dare to infinitive’ resembles ‘venture to infinitive’. The two constructions share the following three semantic features: firstly, the

18 Biber et al. (1999: 484) do not include dare in their discussion of semi- modals. Duffley (1992: 102ff) contains a discussion of dare in its auxiliary role.


238

subject actually realises the complement clause situation, secondly, the speaker considers that the subject is taking a chance in so doing, and thirdly the speaker usually views this chance-taking on the part of the subject in a positive light. The two constructions with dare as a full matrix verb may occur with the same complement clause predicates. One such pair was cited as (653) – (654). Three more are cited as (656) – (661). (656) My father was the only one of her relations who dared to tell her that this

man should not be trusted. (FPU 1080) (657) ‘And don't dare tell me it's going to be too tough for a woman,’ Mariana

shouted, as she glared down at him from the saloon. (AMU 1879) (658) For one thing he was an impatient sort of eagle, inclined to get angry and

feel insulted at the smallest thing; for another – and this took Creggan a while to realize - he was preoccupied with thinking about what had happened to Minch, not daring to hope that what Creggan had so boldly said about her coming back was true. (FP3 303)

(659) She was everything he had dared hope for, and more: she had perfect features and limbs, an endearing, screwed-up little face and a trace of dark hair on her pate, just to show she took after her parents. (EWH 458)

(660) She kept her eyes on his shirt-front, not daring to look at his face; to see his anger. (H85 1265)

(661) Under cover of the porridge business, when all his attention was fixed on Finn, Melanie dared look at her uncle at last. (FRC 1124)

To begin with ‘dare to infinitive’, an analysis of the downloaded tokens indicates that it is most often used in the sense ‘possess the courage to do (and do)’. This meaning may be detected in (653), (656) and (660), although not in (658). A very common meaning of ‘dare bare infinitive’ is ‘have the effrontery to’.19 This is encoded in (657), which is representative of the many examples containing ‘don’t you dare’. However, there are also some few examples of ‘dare

19 Dixon suggests another semantic difference between the two constructions. He writes: “The lexical-verb sense tends to refer to an inner state of the subject, as in (1a), and the modal use to some external circumstance, as in (1b).

(1a) He doesn’t dare to touch Mary (he hasn’t the courage, since she is so beautiful and he is too shy)

(1b) He doesn’t dare touch Mary (for fear of catching AIDS).”

(Dixon 1991: 184)

I must admit I found it hard to detect evidence of this putative distinction among the downloaded tokens.

Constructions containing dare and need 239

to infinitive’ used in this sense. Finally, the type of daring evoked by both (658) and (659) involves neither courage nor effrontery but presumption. What are we to make of the fact that to is present in one construction and not the other? Is the to infinitive complement more goal-like than the bare infinitive one, and, if so, how? Recall that it is a recurrent theme of this study that the use of to evokes an alternative to the actual complement clause situation, either a possible alternative in the projected future, or an alternative in some (other) mental space. The sort of alternative evoked depends on the semantics of the matrix verb and of the construction type, on whether it is Judgement, General or Forward-looking. In a General construction such as ‘like to x’ the complement situation is implicitly contrasted with ‘to y’. In a Judgement construction such as ‘pretend to do x’ the complement situation is implicitly contrasted with ‘actually do x’. The Forward-looking ‘dare to infinitive’ resembles the latter: ‘she dared to x’ is implicitly contrasted with ‘she did not x’. In (653), for instance, the implicit alternative is ‘not asking’, in (656), ‘not telling’. In examples with negated dare, such as (658) and (660), the alternatives are ‘actually hoping’ and ‘looking’ respectively. What then may be said of the ‘dare bare infinitive’ examples? Do not they too evoke implicit alternatives? The answer to this question is that many of them in fact do so. The reason for this is to be found in the semantics of the matrix verb itself, irrespective of which construction it occurs in. The very idea of daring something, of possessing sufficient courage or effrontery to do it, carries the implication that the possession of such a sufficiency is not something to be taken for granted. From this point of view, we may compare dare with prefer, which was discussed in section 5.5.3. Whether it be followed by a to infinitive clause or an -ing clause, prefer always evokes the idea of an alternative. Indeed we might go so far as to assert that the to after dare is to some extent redundant, and that minimal pairs with and without to are semantically equivalent. This would certainly tally with my intuitive impression that one form can normally be substituted for the other. One exception is after the -ing form of dare, which seems to strongly favour the to complement (all 61 tokens of daring in my data are followed by to). Just as was the case with dare, I also considered excluding need from the study on the grounds that it is often described as a modal auxiliary and undoubtedly inhabits to some extent the same semantic field as ‘must’. On the other hand ‘need to infinitive’ bears a close semantic resemblance to ‘require to infinitive’, which I have never seen described as a semi-modal. Moreover, ‘need -ing’ resembles ‘deserve -ing’, and no other modals take -ing complements. Biber et al. (1999:164) state that ‘dare and need are probably obsolescent as modal auxiliaries, at least in AmE’. My data indicate that of the two need is the more obsolescent in present-day British English.20 For these reasons it was decided to retain ‘need (to)’ in the study. It is exemplified by (662) – (664).

20 A comparison of the incidence of the two constructions in the LOB/FLOB and Brown/Frown corpora shows that the 'need to infinitive' construction


240

(662) You need to check on the tax position. (B08 217) (663) However, some of the sections, such as the consumer trade magazines,

will need to be broken down still further into more useful groups. (EX6 291) (664) Don't think you need actually stand in the mud, though - if we get back to

the hard it will be all right.’ (H0D 69)

In the ‘need to infinitive’ construction the speaker exclaims upon the necessity of the subject's realising the targeted alternative in question. In the matter of TAM relations, ‘need to infinitive’ is normally either in the present (68%) or modalised (20%). Need resembles dare in that it also takes a bare infinitive complement, as in (664). However, although need is over twenty times more common than dare in the BNC, the projected total of bare infinitive complements with need is less than a third of those with dare (see Table 12 in Appendix 2 for details). While Biber et al. (1999: 164) are of the opinion that need is on the wane as a modal auxiliary, they do still detect a preference in negative clauses for the use of need as an auxiliary (without do insertion) in all of the three forms of written British English in their corpus. In my material, constructions with do insertion were much more common than the alternative in written as well as spoken contexts. The fact that my figures do not discriminate between genres may be partly responsible for this discrepancy. To find out whether this was the case I investigated the ‘written imaginative’ portion of the BNC and compared the results with those for fiction in the Longman Grammar of Spoken and Written English. Biber et al. found that the construction without ‘do’ was seven times more common in British fiction than the one containing ‘do’. The figures given (Biber et al. 1999: 164) are 35 and 5 occurrences per million words respectively. The comparative figures for the written imaginative part of the BNC are 16 and 9. In other words my figures confirm the tendency on the part of fiction writers to favour the ‘need not’ form although this tendency is not as marked as in the fictional component of the Longman corpus. The discrepancy between the figures for the written part of the BNC as a whole and for the written imaginative part shows that the ‘don’t need’ alternative is very much favoured in at least some written genres. ‘Need -ing’ is one of eight Applied Attitude constructions that display trajector-landmark inversion in the complement clause. As this type of construction is rather uncommon in present-day British English, I cite one occurrence of each matrix verb as (665) – (672).

(665) In the case of investment trusts, this warning needs repeating with force.

(H8E 839)

was spreading at the expense of its bare infinitive counterpart in both British and American English (see Egan 2006: 101).


(666) Vertus produces fruitier Champagnes than those of its northern neighbours on the Côte and, although the Chardonnay does not achieve the remarkable perfume and definition of les Mesnil or Cramant, the wines are of very high quality and well deserve upgrading. (C8M 996)

(667) Because God does not require reminding that we were just as good Jews without ostentation. (BN3 184)

(668) Their clothes felt damp, corsets defied fastening by chilling fingers, socks and stockings were recalcitrant, the bathing attendant was shouting impatiently. (H8A 3667)

(669) A text well within a reader's competence may not invite reading because it is dull, predictable, and appears to have no relevance. (BML 1074)

(670) Below that the car risked bogging down in the mud or grounding on an obstacle, above it the tyres might lose adhesion on the continual twists and turns or cliff-like descents, or one of the vicious pot-holes or rock outcrops rupture the suspension or pierce the sump. (HTT 931)

(671) Hair that dries out resists styling and is more difficult to manage. (CDR 2418) (672) He did not look like a wolf, but he did look like a young man with an eye

for a girl, and techniques that would bear watching. (H8L 368)

These eight constructions all encode in their complement clauses situations in which the subject is the participant that would normally be encoded as the landmark rather than trajector. As Conrad puts it: “After the verbs need, require, want, bear the gerund has passive meaning, i.e. the subject of the main verb is semantically the object of the gerund” (Conrad 1982: 182). Conrad’s “semantic object” corresponds to Cognitive Grammar's “landmark”. Langacker writes: “The crucial feature of a passive [is] that the subject (trajector) is identified with a processual participant that would otherwise be the direct object (primary landmark)” (Langacker 1991: 201). In present-day English the reversal of the canonical trajector-landmark alignment is a feature of the passive participle. In these eight constructions, on the other hand, it is encoded by the -ing form. Thus, in (669) it is the text that extends the invitation, but not the text that will be the agent of any reading that materialises in the projected future. Similarly, in (671), it is the hair that does the ‘resisting’ but not the styling. These constructions differ considerably in frequency. ‘Need -ing’ is the second most common Applied Attitude -ing construction after ‘avoid -ing’. Indeed, considering the comparative paucity of occurrences in the other constructions, it might appear that ‘need -ing’ bears most of the burden for ensuring the survival of this interpretation of the -ing form in present-day English.21 Only half of the ‘deserve -ing’ examples in my

21 This statement rests on an assumption, that token frequency plays a more important role than type frequency in the retention of minority forms, that may well be unwarranted. There is some evidence that type frequency is


242

material, one of which is cited as (666), contain this sense of the -ing form. The other two are like (673) in that the trajector of the matrix verb corresponds to the trajector of being and only thereafter with the landmark of kept. There are no such passive ‘being’ constructions after need.

(673) It is now finished and fits perfectly although, as yet, I have not worn it,

feeling it deserves being kept in a glass case. (CGX 1469)

Of these eight matrix verbs, only deserve and need also occur in passive to infinitive constructions as in (674) and (675), which may be compared to (665), (666) and (673).

(674) If post cold-war Europe is not to be violent and conflictual then the

sources of rivalry and mistrust need to be addressed head-on. (A2W 182) (675) His is a book that deserves to be updated, and deserves to be read. (K5C 2087)

Semantically the -ing constructions share with their to infinitive counterparts in (674) and (675) the encoding of the expression of an opinion by the speaker that the subject is worthy/needful of the realisation of the complement situation. In so far as the verb need itself encodes a lack of some sort or other, it carries the implication of the existence of an alternative to the prevailing situation. It is this alternative that is encoded as the targeted alternative in the to infinitive construction. However, it is also encoded as a targeted alternative in the -ing construction. The main difference between the two constructions is that in the one case there is a greater degree of overlapping between the profile of the matrix verb and the complement situation than in the case of the other. Are there then any other differences at all between ‘need V-ing’ and ‘need to be V-ed’? There is at least one and that is that the -ing predicate resists modification by adverbials of manner or degree, although not of time or place. That the former are not totally excluded is proven by the inclusion of with force in (665), but this is the only such example among the downloaded tokens. For this reason it is not always possible to paraphrase an occurrence of ‘need to be -ed’ with a felicitous ‘need -ing’ construction, whereas the opposite is always possible. There are some 90 examples of the ‘need to be -ed’ construction in my material, as opposed to 32 of ‘need -ing’. If, however, we restrict our attention to complement clauses containing a verb (phrase) and nothing else, such as ‘to be prepared’ and ‘preparing’, we find that there are only 9 of the former type and 23 of the latter. We may therefore conclude that, of the two, ‘need to be -ed’ is the default choice in all cases except for those in which the speaker wishes to profile

the more important of the two. Bybee (2001) discusses the importance of both factors in relation to phonological and morphological change.


a situation that may be encoded by a simple verb phrase. In the latter case the scales of probability tip towards ‘need -ing’.

6.2.5 To infinitives versus bare infinitives: a summary

In this section we have looked at a variety of Forward-looking constructions containing to infinitive and bare infinitive complements. We have focussed on constructions containing the matrix verbs bid, help, allow, let, dare and need. Common to the bare infinitive constructions is the intimation that the complement situation will be realised in the (near) future. Thus, we saw in section 6.2.1 that the ‘bid S2 bare infinitive’ construction is normally used to encode trivial requests with which one would expect the complement subject to comply at once. Similarly, if one lets someone do something, we can be almost certain that they will in fact do it. On the other hand, if one bids or allows some one to do something, there is at least a possibility that they may not do so, that what we have termed the targeted alternative may not be realised. This distinction between the two construction types is blurred in the case of help, discussed in 6.2.2. This is because the complement situation is encoded by the matrix verb itself as at least partially realised. The exact opposite is true in the case of need and dare, discussed in 6.2.4, which both themselves imply the existence of more or less latent alternative situations to the one encoded in the complement clause.

6.3 To infinitive versus -ing complements

According to the data in Table 2.7 (p. 43) there are 220 Forward-looking to infinitive constructions and 67 Forward-looking -ing constructions in the present study. Sixteen matrix verbs occur in both forms of construction. Just as was the case with some of the pairs of constructions in the previous section, some of these verbs are very common indeed. For instance, ‘begin to infinitive’ is the fourth most common to infinitive construction according to the figures in Table 13 in Appendix 2. And both ‘begin -ing’ and ‘start -ing’ are among the five most common -ing complement constructions, according to Table 14 in the same appendix. We will start our discussion of these constructions in 6.3.1 by looking at Mental Process constructions containing anticipate and intend. Then in 6.3.2 we look at Attitude constructions containing dread, fear and want. Section 6.3.3 is devoted to Aspect constructions containing begin and start.

6.3.1 Mental Process constructions

There are two Mental Process matrix verbs, anticipate and intend, that occur with both to infinitive and -ing complements in Forward-looking constructions. These are by no means the only two Mental Process matrix verbs that occur with these two complement forms, but in the other cases the two complement forms belong


244

to different construction types. Thus, in the case of verbs like forget and remember, to be discussed in section 7.2, the constructions containing to infinitive complements are Forward-looking, the constructions containing -ing complements Backward-looking. And, in the case of verbs like conceive and imagine, the constructions containing to infinitive complements belong to the Judgement category, the constructions containing -ing complements to the Contemplation category. Anticipate and intend are thus the only two mental Process matrix verbs which occur in two non-finite Forward-looking constructions. Anticipate occurs with both sorts of complement in both same- and different-subject constructions, while intend only occurs with both types in same-subject constructions (as a different-subject construction it only occurs with a to infinitive complement). As may be seen in Tables 2 and 8 in Appendix 2, the -ing form of complement is far more common than the to infinitive one in the case of anticipate, while the to infinitive form is more common in the case of intend. We will discuss them in turn below, starting with anticipate, exemplified by (676) – (681). (676) To offset this, there were usually a few well-known local characters

hanging about, anticipating their presence being required. (BN3 1592) (677) The agreements anticipate deliveries starting in volume later this year, and IBM will also make available low-end disks for Bull mid-range systems and Zenith microcomputers. (CTG 180) (678) Erm it’s certainly grown faster than we anticipated it growing and we are

not able to er we’re not able to carry out what we like to do with Neighbourhood Watch. (GY4 217)

(679) Because of the ideal lively location and facilities, we anticipate the Magamar apartments to be very popular this year with Club Clients. (AM0 716) (680) Project Video is provisionally scheduled for publication in September

1989 and we anticipate sales over the first three years to be in the region of 2,000 cassettes. (AP1 402)

Both ‘anticipate S2 to infinitive’ and ‘anticipate S2 -ing’ are Forward-looking constructions in the sense that they imply that the complement situation is profiled by the matrix verb subject as a likely candidate for realisation in the projected future. There are only two Mental Process matrix verbs, anticipate and foresee that encode complement situations in this domain by means of -ing clauses. As a rule -ing Mental Process complements are either located in the past, as is the case with remember and recall, or are mere objects of contemplation, as with envisage or fancy.22 The fact that the subject of anticipate views the complement situation as likely to come about is, however, no guarantee that such

22 When fancy is employed as an Attitude matrix verb, the -ing complement encodes a situation in the projected future, as in “Do you fancy coming jogging?” (KBU 1057)

To infinitive versus -ing complements in Mental Process constructions 245

will be the case. Indeed, the construction may be employed in contexts where the expectations of the subject are explicitly seen to have been confounded, as in (678). More common, however, are examples like (676), in which it is implied that the subject’s expectations were met, and (677), which is open in this respect. Nor are the expectations of the subject completely confounded in (678), as the anticipated growth did in fact materialise - it just came about sooner than expected. In the ‘anticipate S2 to infinitive’ construction, instantiated in (679) and (680), it is also an open question whether the complement situations will be realised. It is certainly debatable whether the future situation in these examples is encoded as less certain of realisation than is the case with the -ing construction. There are just not enough examples of the construction in the BNC on which to base a generalisation. We can, on the other hand, be more certain that in the case of ‘anticipate -ing’ (and ‘foresee -ing’) the matrix verb subject takes it for granted that events in the future can be predicted with a considerable degree of certainty. (In this connection, one may note one of the definitions of anticipate in the OED: “9. To look forward to, look for (an uncertain event) as certain.”) In other words we are dealing with something akin to what Bolinger (1968) in the following passage calls “metaphorical reification”:

Verbs such as want, wish, hope, expect, command, etc. apply to unrealized possibilities; the complement that goes normally with them is the infinitive. Verbs such as enjoy, visualize, detest, understand, deny, approve, etc. usually apply to actualities or to possibilities conceived as actualities – literally or metaphorically they reify the action; their normal complements are -ings. (Bolinger 1968: 127)

Compared to the ‘anticipate S2 -ing’ construction, ‘anticipate S2 to infinitive’ is very much a minority option in present-day English. This is reflected in the fact that there were only two such tokens among the 1,000 downloaded utterances containing anticipate. Nor is it mentioned in any of the reference works I consulted. My own initial reaction on encountering these examples in the corpus was to view both same-subject and different-subject to infinitive constructions with anticipate as anomalous because of what I perceived to be a contradiction between the element of forward projection which I took to be connoted by the matrix verb and the intimation of a possible alternative connoted by the to infinitive. That is, use of the verb anticipate implies that the complement situation is profiled from a point in the future into which the subject projects him- or herself. From such a vantage point one perceives a situation as it is, without evoking the existence of possible alternatives. Note that this element of forward projection is not implied by verbs such as expect that only occur in to infinitive constructions. Tokens such as (679) and (680) are the product of speakers who are using anticipate in the sense of expect. Mair (2003b: 243) points out that speakers like this are by no means thin on the ground. One must therefore conclude that constructions containing anticipate and a to infinitive complement were gaining a foothold in the language at the end of the twentieth century. This foothold is more tenuous in the case of the same-subject to infinitive construction,


246

illustrated in (681), which may be contrasted with the far more common -ing construction in (682) and (683). (681) (He was also anticipating to gain at least six ‘O’ levels before leaving

school.) (CL1 951) (682) Pat anticipates receiving quite a few requests in the near future. (HX8 304) (683) It was always disappointing to sight the Land Rover and anticipate being

in camp in a few minutes and then have an hour to walk. (AT3 1678) There is no mention of the structure in (681) in any of the dictionaries or grammars I have consulted. Again, there appears to be an element of semantic incompatibility in (681) between the distancing to infinitive, which locates the complement clause situation at some remove in the projected future, and the matrix verb itself, which projects the subject’s point of view into the same future. If one anticipates something, one is ahead of one’s time, so to speak. Now, the whole point about projecting oneself into the future, about anticipating or foreseeing something, is to consider it as if it were unfolding before it actually does so. However, if one sees it as unfolding, one does not see it as the targeted of several possible alternatives. Speakers who use anticipate to mean ‘expect’ have lost (or not acquired) the element of forward projection prototypically connoted by the verb. Note, by the way, that while the subject in the -ing construction expresses him- or herself with come confidence about the realisation of future situations, he or she may actually be mistaken as to the exact location of the situation in question in the projected future. At the time of anticipation in (683) the subject imagined this to be closer at hand than was in fact the case. Intend is the only other Forward-looking Mental Process verb that occurs in both an -ing and a to infinitive construction. Unlike anticipate it makes reference to a decision which has already been made. Both ‘intend to infinitive’ and ‘intend -ing’ refer to a prior decision by the subject, a decision which has not yet been realised. In over half the examples of both constructions (see Table 6.12), the matrix verb is in the present, as in examples (163) and (164), repeated here as (684) and (685), both of which contain the complement predicate continue. (684) This time we intend to continue our protest and will escalate action if

necessary. (KA5 414) (685) Manchester design company Lord & Bowes has developed its new look,

including an editorial column and news roundup, and we intend continuing with our lengthier informative features. (K9H 14)


Table 6.12: TAM of matrix verb intend in to infinitive and -ing constructions, with horizontal percentages based on 410 tokens of ‘intend to infinitive’ and 27 tokens of ‘intend -ing’



Modal -ing inf

intend to

220 54%

9 2%

114 28%

4 1%

2 .5%

37 9%

6 2%

17 4%

intend -ing

18 67%

7 26%

2 7%

(684) and (685) are very similar in meaning. In particular the subject in both utterances may be presumed to be master of his or her own fate, so to speak. In fact, the ability to influence future events to this extent is typical of the -ing construction, but not of its to infinitive counterpart. Of course, given the fact that the matrix verb is in the present in the majority of cases in both constructions, it is often difficult to predict how successful the subject may be in realising a situation in the projected future. However, almost a quarter of all tokens of the ‘intend to infinitive’ construction involve situations in which the subject is not the sole master of his or her fate. (686) and (687), which have already been cited as (207) and (150), are typical examples in the present. (686) The Serbian forces’ westward drive into Bosnia fits into a map disclosed

for the first time yesterday, which shows Serbian leaders intend to capture more than 70 per cent of Bosnia, leaving Croatian and Muslim populations isolated in small pockets of land. (AKR 527)

(687) The commanders cannot actually see what Lord Raglan intends to refer to. (CBR 1524)

In (686) it is obviously not up to the Serbian leaders to decide unilaterally how much of Bosnia they are going to control. In (687) the realisation of the complement situation is explicitly excluded, the subject patently failing to carry out his intention. If we turn our attention to matrix verbs in the past or past perfect we find many more examples of situations that were never realised. (688) and (689) are cases in point. (688) All but a few of them were Jews, refugees from a Nazi state which

intended to dispossess, transport and exterminate them. (G1X 683) (689) Michael had originally intended to run Royal Ballerina in the Prix

Vermeille at Longchamp the day after the St Leger and now Kauntze is hoping for dry conditions at Doncaster for what he regards as an easier race. (K2D 3501)

As for the ‘intend -ing’ construction, there is only one example in my material of a complement clause situation that is explicitly excluded. This is


248

(692), in which the subject herself decides to override her previous decision to remain silent. (690) Bolger intends running St Jovite over a more suitable 10 furlongs in the

Derrinstown Stud Derby Trial at Leopardstown before tackling the Epsom Classic. (AKE 1619)

(691) In the meantime, the solicitor intended administering the estate on behalf of the family. (HHC 2205)

(692) She’d intended not answering, but habit overruled her good intentions. (HGM 2672)

It is typical of the ‘intend -ing’ construction that it encodes situations which are completely up to the subject to bring about. Perhaps it would be more accurate to say that it prototypically profiles situations as if they were completely up to the subject to bring about. It is instructive in this context to contrast (689) with (690). Both refer to a racehorse trainer’s intentions with regard to running a particular horse. No trainer can be absolutely certain as to whether the fates will allow him to produce a horse in sound condition on the day of a race. Nevertheless, whereas (689) encodes this uncertainty, (690) ignores it. It profiles the running of the horse as something almost certain to come about. The same point can be made with regard to matrix verbs in the past. If the subject, for instance in (691) and (692), intended doing something, this something usually came to pass. When it did not, as in (692), this was due to a conscious decision of the subject herself. It may not just be the absence of the implication of latent alternatives in the -ing construction that serves to create this impression of comparative certainty. The unfolding nature of the situation profiled by the -ing form as opposed to the more static image evoked by the infinitive may also contribute to this effect. To sum up, ‘intend to infinitive’ typically profiles an intention on the part of the subject to proceed towards a goal that may or may not be reachable: ‘intend -ing’ typically profiles an intention to realise a situation without even raising the possibility that it may not be realisable. Note that this conclusion is exactly the opposite of that reached by Wood. He writes: “She didn’t intend paying her bill this month implies ‘such was her design, but she might be induced to alter her mind’. If we wish to convey the notion that she had firmly determined not to pay, we must use the infinitive” (Wood 1956: 13). This is yet another illustration of how unsafe it is to rely on intuition in these matters, even such a normally sound judge as Wood proving fallible on this occasion. As has been noted with regard to other matrix verbs that govern several types of clause, there may well be occasions on which both constructions can be employed to convey much the same (but not the exact same) content. In particular, ‘intend -ing’ always entails ‘intend to infinitive’, so one can normally substitute the latter for the former with the only subsequent change of meaning being a diminution in the subject’s commitment to the probability of the coming about of the complement situation. It is presumably this one-way entailment that has led scholars such as Conrad (1982: 17 & 178) to assert that the two constructions are ‘free variants’. But the


entailment is only one-way. The opposite does not hold. It would, for example, be impossible to substitute ‘intends referring’ for ‘intends to refer’ in (687). The same point applies to other contexts which explicitly exclude the realisation of the complement clause situation.

6.3.2 Attitude constructions

There are more instances of to infinitive and -ing forms of complement with the same matrix verb in the case of Attitude matrix verbs than any other type. The most common form of contrast is between Same-time -ing and General to infinitive constructions, discussed in section 5.5. There are, however, three matrix verbs that occur with both complement forms in Forward-looking constructions, These are dread and fear which occur with both forms in same-subject (but not different-subject) constructions, and want, which occurs with both in different-subject (but not same-subject) constructions.23 We begin by looking at fear and dread. Before exemplifying the two constructions containing fear and dread, one may note the fact that these matrix verbs are very much in the minority among Forward-looking Attitude verbs in so far as they encode a negative attitude on the part of the subject to the complement situation. The vast majority of Forward-looking Attitude constructions encode a positive attitude on the part of the subject towards a situation in the projected future. Thus, seventeen of twenty same-subject Forward-looking to infinitive constructions encode an attitude of some sort of ‘wanting’ on the part of the subject. The remaining three (regret is the third) encode the opposite attitude of ‘not wanting’. Of the four -ing constructions, only one, ‘fancy -ing’, is positive, the other three (face is the third) encoding a negative attitude. The negative fear and dread take both to infinitive and -ing complements. Both verbs are clearly Forward-looking in both constructions. They both contain an element of anticipation in their semantic make-up and in both cases whatever is anticipated is seen as extremely undesirable. Where then does the distinction, if any such there be, between the to infinitive and -ing constructions lie? We begin by citing, in (693) – (696), an example of each. (693) They may fear to extend love and acceptance to the new partner as it seems a betrayal of the absent parent. (BLW 1396) (694) I swim here with Byron because I dread to swim alone, and tolerate all his impudent remarks. (HGS 1477)

23 Want does occur with a same-subject -ing complement in a construction showing trajector/ landmark realignment, in a similar fashion to those discussed in section 6.2.4, p. 240, but no tokens of this construction surfaced among the 1,000 tokens of want downloaded from the BNC.


250

(695) The old person may fear being cut off, and the family may feel they are abdicating responsibility, experiencing some guilt. (CGD 1810) (696) Our players are beginning to dread playing at home. (CEP 3629) The most obvious difference between (693) and (694) on the one hand and (695) and (696) (previously cited as (129), p. 67) on the other is that in the case of the former pair it is up to the subject to realise the complement situation should he or she wish to do so. In other words, in the to infinitive constructions the complement situation is profiled as a targeted alternative, one of several theoretically possible alternative situations in the projected future. There is no indication that any such alternatives exist in the case of the -ing complements in (695) and (696). Employing the two above-mentioned notions of anticipation and undesirability, and ignoring for the time being the difference between the two matrix verbs, we may sketch out the difference between the constructions in the following manner. (a) S fears/dreads to do X: S does not wish to anticipate the doing of X (one of several possible future alternatives) because it is very undesirable. (b) S fears/dreads doing X: S anticipates the doing of X which he or she finds very undesirable. We now consider the two types of complement in turn, starting with ‘fear to infinitive’ and ‘dread to infinitive’, exemplified by (697) – (701). (697) What we will do next Season I dread to think. (J19 2697) (698) Mr Kennedy says: ‘If that’s how they treat a witness, I dread to

imaginehow they treat a suspect.’ (CH1 8541) (699) Angels might well have feared to tread where Charlotte habitually planted

her size four sandals with zest and aplomb. (H8L 22) (700) Perhaps the re-adoption of Augustus had triggered more perturbation and

he feared to sleep. (FP1 1380) (701) Ancient law, it seems, was on their side; thousands were struck off, and

more feared to be. (G3C 379) What sort of activities does one ‘fear to’ and ‘dread to’? To begin with the latter, we appear to dread mental activities. Of thirty-three examples in the downloaded random sample, twenty-eight contain the phrase ‘dread to think’ and two the semantically related ‘dread to imagine’: see (697) and (698). One may choose whether or not one wishes to think about something, and in all of these utterances the subject chooses to distance him- or herself (rhetorically) from entertaining the thought in question. This element of choice is clearly apparent in (694), one of the few examples of ‘dread to infinitive’ containing a verb other than ‘think/imagine’. There is a greater variety of complement clause verbs in the case of ‘fear to infinitive’ than ‘dread to infinitive’. There are six instances of ‘fear to tread’,

To infinitive versus -ing complements in Attitude constructions 251

one of them cited as (699), among the downloaded tokens, but this is presumably due to the popularity of Pope’s phrase rather than any inherent element in the semantic make-up of the matrix verb itself. Although there is one instance of imagine, there are none of think. This absence may well be due to the fact that ‘dread to think’ is so firmly entrenched in the mental grammar that it serves to block the production of ‘fear to think’ in most contexts where the latter might suitably be produced. In any case, mental process verbs do not appear to be particularly favoured in the case of ‘fear to infinitive’. What most of the complement clauses do have in common, both with each other and with the complements of ‘dread to infinitive’, is the implication that it is within the power of the subject to bring about, or rather to avoid bringing about, the process in the complement clause. Thus, in (693) one can choose to whom one wishes to extend one’s love and acceptance. There are, however, varying degrees of control one can exercise over one’s own fate, and ‘fear to infinitive’, unlike ‘dread to infinitive’, is not confined to contexts where the subject exercises complete control. Take (700), for example: on the basis of our general knowledge we know that one may put off sleep, but not indefinitely. And (701) encodes a complement situation the control of which is outside the remit of the subject. In it the subject may not interrupt the progress towards the targeted alternative signalled by to: what it does express is the hope of the subject that this progress will be interrupted by some outside agency. Unlike some of the other ‘fear to infinitive’ examples, (701) entails ‘fear -ing’, to which we now turn. (702) An employer might respond to this by offering his workers higher wages

to foster effort so that they will fear losing their jobs because the alternative jobs are less well paid (see Greenwald and Stiglitz 1987).

(HWH 950) (703) Many women from immigrant communities are hidden by their men, who

may be entrepreneurs, and they fear offending both their men and the community by complaints about their work. (HH3 9420)

(704) He dreads telling Martha he is going away, but he promised Babur. (HGU 1190) (705) A lot of American guys dread coming to Europe, but it’s not an excuse.’ (K3H 762) Common to all the downloaded tokens of ‘fear -ing’ is that they describe situations in the complement clause which are beyond the control of the subject. In all cases the object of the fear is envisaged as a real possibility and the subject shudders at the prospect of its being realised. The powerlessness of the subject is underlined in many instances, 50% to be exact, by the passive nature of the verb phrase in the complement clause, as in (695). Moreover, even when the complement clause has the syntactic form of a transitive clause, as in (702) and (703), many of the elements of prototypical transitivity (as described, for example, in Taylor 1995: 206 ff.) are lacking. ‘Losing one’s job’ and ‘offending


252

somebody’ do not contain PATIENT objects, for one thing, nor do they describe activities initiated by the subject of his or her own free will. Some instances of ‘dread -ing’ resemble ‘fear -ing’ in that, while they describe situations in the complement clause that may or may not happen, their happening or not is not a matter for the subject to decide. More common in the case of ‘dread -ing’ are situations which are portrayed as well nigh unavoidable. In other words, it may well be that the reason why they are realised so vividly is that they are considered almost real, even though situated in the projected future. The intensity of the emotion encoded in the ‘dread -ing construction’ is pointed out by Wierzbicka.

Dread refers to the future, and yet ING is possible (and in fact virtually obligatory). However, as many dictionaries of English point out […] dread is an ‘intense’ emotion, which involves a vivid ‘picture’ of the dreaded event. This suggests that dreading involves an act of imagination; and the terrifying event happening in our imagination is of course simultaneous with the feeling itself. In other words, dread behaves syntactically like imagine […] because ‘dreading’ implies imagining. (Wierzbicka 1988: 70)

While agreeing with Wierzbicka with respect to the vividness with which the dreaded situation is pictured, I would go further and suggest that this vividness is at least partly due to the situation being profiled as virtually unavoidable. Take (696) for example: given that every second match is normally played at home, the situation dreaded by the players is bound to arise. Similarly, in (704) the subject is committed to telling Martha, and in (705) these American guys are scheduled to come to Europe, whether they like it or not, and they anticipate this relocation with acute displeasure.24 To sum up, both ‘fear to infinitive’ and ‘dread to infinitive’ profile situations that may be avoided. As a rule their avoidance is within the competence of the subject to ensure. As for ‘fear -ing’ and ‘dread -ing’, they both profile situations that are almost certain to occur. The subject exercises no influence over their possible occurrence. In all cases he or she is an unwilling victim. Both to infinitive constructions encode targeted alternatives, with the focus in these cases being upon targets that the subject wishes to avoid. Both -ing constructions encode situations seen as having a certain extension in the projected future. The same point may be made with respect to different-subject -ing constructions as exemplified by (706) – (709).

24 Note that the evidence of the ‘dread -ing’ construction runs counter to the following principle enunciated by Givón : “If an event is epistemically less expected, chances are that one’s emotional valuative reaction to it – surprise, preference, aversion – is stronger.” (Givón 1995: 152)

To infinitive versus -ing complements in Attitude constructions 253

(706) ‘I dread mine reaching their teens. (BLW 1071) (707) She says he must have been released early, and she’s dreading him coming to see her and demanding his parents’ address. (CKF 2137) (708) But she fears the embers’ flicker and the night closing in over chill ash. (A0L 3702) (709) The most probable reason for this unjust behaviour by the Roman

Governor was that he feared some sort of riot breaking out in the capital city during the festival. (CEJ 2983)

Just as in the case of same-subject -ing complements, these different-subject constructions carry a strong implication that the complement situation is likely to come about. There is no avoiding children growing up, as in (706), or night falling, as in (708). In (707), the matrix verb subject would seem to be completely certain that the complement subject is about to arrive on her doorstep. It would appear that it is this subjective certainty, or at least very strong belief, on the part of the matrix verb subject (S1), that sanctions the use of the construction, and not some sort of objective likelihood. This point may be nicely illustrated by (709), in which the subjective expectations of S1 lead to his taking steps to actually avert the realisation of the complement situation. A look at Table 3 in Appendix 2 shows that want is by far the most common matrix verb used in different-subject Attitude constructions. Indeed, ‘want S2 to infinitive’ is over four times as common as all the other Attitude constructions put together. More surprising, perhaps, is the fact that ‘want S2 -ing’ accounts for almost 40% of the different-subject Attitude -ing constructions. (710) – (714) exemplify both constructions. (710) ‘Bob,’ said Tessa, when he had put the phone down again, ‘do you want

me to come to the funeral or not?’ (G12 2341) (711) Bobby’ll be here in a minute, and I want you all to give him a performance that’ll blast him out of his seat!’ (H92 1006) (712) ‘I want you to be somebody else!’ (JY8 3100) (713) On Saturday morning I want them thinking, I don’t want them going into

their shells. (K5A 3415) (714) ‘The car was so badly crushed that I thought about asking the boss if he

wanted it posting back.’ (CBF 1839) Both want constructions point to situations in the projected future. In the case of the to infinitive complements, the situation is normally one that the matrix verb subject has a reasonable expectation will come about. It is this element of possible realisation that gives rise to the paradox in (712). In (710), the matrix verb subject (S1) is presented with two alternative future courses of action for the complement subject (S2), and it is up to him to decide which to bring into effect. Similarly, in (711), it is within the power of S2 to fulfil the wishes of S1. As for the ‘want S2 -ing’ construction, its complement situations are also located after the act of wanting. In their case, however, it would seem that the situation is


254

played out before the mind’s eye. In (713), for instance, S1 projects himself forward, mentally, to the coming weekend and envisages what he wishes to see taking place, and not taking place, at that time. It is the second of these two perspectives, that of envisaging something one would wish to avoid seeing realised, that is the more common of the two, accounting for six of the nine occurrences in my data. And of the three exceptions, one, cited here as (714), is an instance of the extremely marginal (in present-day English) different-subject -ing construction in which S2 encodes the landmark rather than the trajector of the complement clause predicate. The impression we have of the situation in the complement clause being played out is, no doubt, to some extent due to the ongoing durative nature of -ing complements, described in section 4.4.

6.3.3 Aspect constructions

There are four Aspect matrix verbs that occur in Forward-looking constructions with both to infinitive and -ing complements. Two of them are the very common begin and start. The other two are get and commence. Get is a very versatile verb, only a small number of tokens of which are aspectual in character. In addition, the Aspect ‘get to infinitive’ construction would appear to be limited to the complement predicator work, exemplified by (715), which may be compared to ‘get -ing’ in (716). (715) If the ombudsman decides the case should be examined, investigators will get to work. (CBC 6496) (716) Now, if you think you’ve got the answers, get dialling. (C9J 1774) ‘Commence to infinitive’ occurs with a greater variety of complement predicates than does ‘get to infinitive’. It is exemplified by (717) – (718), ‘commence -ing’ by (719) – (720). (717) Judith rolled up her sleeve and commenced to clean out the hen’s entrails

preparatory to stuffing the carcass. (FRJ 907) (718) I commenced to talk in English, whereupon he said, ‘Speak in Burmese.’ (CDC 1452) (719) Almost as soon as he commenced taking evidence, Beveridge was

inundated with memoranda on the needs of the elderly. (CKP 434) (720) In April 1982 the IOM commenced trading in the S&P 500 stock index

futures contract, with an associated option contract in 1983. (HY2 364) The meaning of these two commence constructions is in many cases very similar. One does, however, get the impression that the to infinitive construction often encodes the taking of the very first steps towards the realisation of the situation, whereas the -ing construction implies both that the subject gets the ball rolling, as it were, and keeps it rolling. Evidence for this analysis may be found in the possibility of interruption in the case of the to infinitive complement, as in (718).

To infinitive versus -ing complements in Aspect constructions 255

We will return to this question of possible interruption below in our discussion of begin and start. Dixon (1991: 176) states that: “both commence and complete tend to be used for some definite and significant piece of work, not just any everyday job.” There is little evidence to back up this statement in the corpus in the case of the to infinitive construction, but the -ing construction does seem to be very common in the fields of legal and commercial discourse, with ‘commence trading’, as in (720), accounting for 20% of all tokens of the construction. The remainder of this section will be devoted to the four constructions containing begin and start. Their TAM relations are listed in Table 6.13 and they are exemplified initially by (721) – (724).

Table 6.13: TAM of matrix verbs begin and start in to infinitive and -ing constructions, with horizontal percentages, based on 452 tokens of ‘begin to infinitive’, 74 tokens of ‘begin -ing’, 154 tokens of ‘start to infinitive’ and 190 tokens of ‘start -ing’



Modal -ing inf

begin to

1 .2%

38 8%

27 6%

254 56%

45 10%

14 3%

31 7%

22 5%

11 2%

9 2%

begin -ing

0 2 3%

0 57 77%

0 3 4%

3 4%

2 3%

0 7 10%

start to

1 2%

35 23%

12 8%

64 42%

9 6%

7 5%

4 3%

14 9%

4 3%

4 3%

start -ing

0 37 20%

0 84 44%

0 2 1%

7 4%

25 13%

1 1%

34 18%

(721) Note that when power is first applied to the circuit, there will be a slight

delay before the amplifier begins to work. (C92 1360) (722) ‘You don’t have to begin working, doing my jobs and setting my house to

rights. (AD1 3158) (723) Evidently, there is more to our belief that the flipping caused the wipers to

start to work than has so far been specified. (EVX 218) (724) I started working as a caddie in 1944 when I was about eight, and I earned

a penny for every year. (ASA 971) A first glance at examples (721) – (724) and the data in Table 6.13 may incline us to conclude that the four constructions in question are very similar both in meaning and in distribution. We would not be the first to conclude this. Dixon, for example, writes:

In many sentences start and begin may be substituted one for the other with little or no change in meaning […]. But there do appear to be some


256

semantic preferences for each verb, which motivates their use to a considerable extent. (Dixon 1991: 176 & 2005: 181)

Unfortunately, Dixon does not go on to spell out these semantic preferences. Quirk et al. (1985: 1192), commenting on the choice of to infinitive or -ing complements, note that in the case of many examples “there is no observable difference of meaning between the constructions”. They do, however, go on to point out that “in other cases, a contrast between ‘potentiality’ and ‘performance’ may influence the choice”. They also note a preference for the -ing form to describe what they term “multiple” activities, which correspond to what Freed (1979) refers to as “series”. Freed (1979), herself, also underlines the similarity between the various constructions in certain contexts. She writes:

The differences between to V and V-ing are extremely subtle and can be convincingly portrayed only in elaborately described contexts […] there are cases or contexts in which the difference between them does not matter. We may say that in such cases there is an intersection of the truth conditions associated with the two forms. (Freed 1979: 75)

There is no doubt that such an “intersection of truth conditions” pertains in (721) – (724) and in many other similar examples. That is, there are many cases where a begin construction may be seen to entail a start construction, a to infinitive construction an -ing construction, and vice versa (i.e. there are many utterances of the type ‘she began to do X’ which, if true, guarantee the truth of ‘she started to do/doing X’). However, it does not follow from this that the meanings of the various constructions are identical. Table 6.13 provides us with some evidence of differences in their distribution, with start matrix verbs being more likely to occur in the present and -ing complements after infinitive matrix verbs. The latter would also appear to be very rare after progressive and non-finite -ing forms of the matrix verb. To these we may add the data concerning the person and animacy of subjects contained in Table 6.14.

Table 6.14: Person and animacy of subjects of begin and start in to infinitive and -ing constructions with horizontal percentages

3rd person 1st person 2nd person animate inanimate

Total

begin to 55 12% 6 1% 242 54% 149 33% 452 begin -ing 3 4% 2 3% 60 81% 9 12% 74 start to 14 9% 13 8% 82 53% 45 29% 154 start -ing 43 23% 21 11% 110 58% 16 8% 190

Table 6.14 shows a marked tendency for inanimate subjects of both verbs to be followed by a to infinitive complement. The same point applies to commence: 29% of tokens of ‘commence to infinitive’ contain inanimate subjects, as opposed


to only 4% in the case of ‘commence -ing’. Another tendency is for ‘start -ing’ to occur with first- and second-person subjects. If we restrict our focus to animate subjects and enquire to what extent these play the role of Agent, we will again note a considerable divergence between the figures for to infinitive and -ing complements, with ‘begin to infinitive’, in particular, favouring non-agentive subjects to a much greater extent than any of the other three constructions (see Table 6.15). Discussing ‘aspectualizers’ in general, and begin and start in particular Freed writes: “all of these verbs allow (and even in some cases require) animate agentive subjects. It may be assumed that this is related to the causative nature of these verbs” (Freed 1979: 81). This is true as far as it goes. However, as Table 6.15 shows, begin and start differ with regard to the extent of their “causative nature”.

Table 6.15: Agentive and non-agentive animate subjects of matrix verbs begin and start with horizontal percentages

Agentive Non-agentive Total begin to 205 68% 98 32% 303 begin -ing 65 100% 0 65 start to 94 86% 15 14% 109 start -ing 166 95% 8 5% 174

One thing is to note these differences in distribution, both those between begin and start and those between the constructions containing to infinitive and -ing complements; another thing is to explain them. To begin with the difference between the two matrix verbs, the greater incidence of start following first- and second-person subjects may be related to the higher incidence of start in spoken conversations. A glance at the projected totals in Table 11 in Appendix 2 will show that there are slightly more occurrences of all forms of the verb begin than of start in the BNC as a whole. A search restricted to the spoken dialogue part of the corpus reveals that start is ten times as popular as its rival. Even ‘start to infinitive’, which is outnumbered three-to-one by ‘begin to infinitive’ in the corpus as a whole, outnumbers ‘begin to infinitive’ in the spoken dialogue sub-corpus by almost two-to-one. These findings would appear to be at odds with those of Biber et al. (1999: 711) who include begin but not start in their list of the most common verbs controlling to infinitive clauses in conversations. The second difference between start and begin that calls for attention is the greater incidence of the former in the simple present. There would seem to be little difference between ‘begin to infinitive’, ‘start to infinitive’ and ‘start -ing’ as regards the semantic context of these examples. Nor does an analysis of the subjects of the various occurrences in the simple present reveal any marked differences between the begin and start constructions. It appears that we must look elsewhere for the reason for the imbalance between the number of begin and start constructions in the present (or indeed the past: if we include figures for


258

progressive and perfect forms, we find that over 75% of ‘begin to infinitive’ and 85% of ‘begin -ing’ encode past situations, as opposed to only 55% of ‘start to infinitive’ and less than 50% of ‘start -ing’). Perhaps a clue to the reasons behind this distribution may be found in Freed (1979), who points to the following difference between the two matrix verbs.

[…] only from a sentence with begin does it necessarily follow that the nucleus (or characteristic activity) of the event has been initiated; a sentence with start followed by a to V complement can have as a consequence that only the onset of the event named in the complement has been initiated. We may conclude, therefore, that start refers to the onset of an event while begin refers to the initial temporal segment of the nucleus of an event. (Freed 1979: 71)

In support of the above contention Freed (1979: 72) maintains that a search of over 200 tokens in her corpus did not reveal a single one in which an activity whose initiation was encoded by ‘begin to infinitive’ was cancelled before it got under way. The evidence of the over 450 tokens of ‘begin to infinitive’ downloaded from the BNC corroborates this assertion. My material does, however, contain a handful of examples of cancelled ‘start to infinitive’, two of which are cited as (725) and (726). (725) As he disappears behind a safety curtain of blue smoke he starts to say something, but he coughs on the smoke. (H8M 4114) (726) She started to bristle, but then she looked at him and saw that his smile

was open and good-humoured. (JY7 2107) In (725) the complement clause situation of ‘saying something’ never actually gets underway. In (726) the subject never actually gets round to ‘bristling’. Freed is undoubtedly correct in asserting that ‘begin to infinitive’ could not be substituted for ‘start to infinitive’ in such a context and that this contrast points up a clear difference between the two constructions. Although in the vast majority of cases of ‘start to infinitive’ the context will confirm that the complement situation went on to be realised, at least in part, the construction would seem to leave open the possibility of an interruption before the situation proper gets underway. It follows from this difference between the two matrix verbs that the language user who wishes to focus attention on the starting point of a situation, disregarding its possible continuance, may well prefer to employ a ‘start to infinitive’ construction. Let us consider some present tense examples of ‘begin to infinitive’ and ‘start to infinitive’: to facilitate comparison they are all ‘time when’ adverbials.


(727) The birth weight of newborn infants, either male or female, first increases steeply with maternal age up to about age 25 years (on the average by 300 grammes), then slower to about age 35, when it reaches the zenith and, from this age, begins to decrease, continuing to the end of the childbearing

ages. (H0P 323) (728) This could explain why some babies only have colic in the evening, when

the morning feed reaches the intestines and the bacteria that live there begin to feed on the undigested lactose. (BM1 1388)

(729) And that is the moment when you begin to decide what you are going to do about going grey. (G2V 3406)

(730) Many good exercise programmes stop when life and the usual routine starts to get chaotic or change for some reason. (AD0 1097)

(731) When the weather starts to improve in the New Year, note that we have, by popular request, another talk by Graham Vincent, and talks by John Massey and our own member Len Davies. (HHN 306)

(732) ‘But all these women had better hurry along though before his looks start to fade.’ (CD5 939)

In (727) the average weight of newborn babies not only begins to decrease, it continues to do so. Indeed, this is stated explicitly in the final clause. Similarly, it is not the actual first steps of the bacteria’s breaking down the lactose in (728) that is the cause of the colic, but the activation and continuation of this process. And, while (729) focuses on the moment at which a decision-making process is initiated, it strongly implies that this will be continued until a decision has actually been reached. Contrast these examples with (730), in which it is sufficient that life has started to get chaotic for the main clause situation of stopping to be triggered. Whether or not the chaos continues is immaterial. Similarly, in (731) and (732) the focus is on the starting point of improvement in the weather and disimprovement in physical appearance, respectively. Although we know from our world knowledge that weather on the whole tends to continue to improve throughout the spring, and that looks tend to continue disimproving with age, the continuance of these processes is not focussed upon in these sentences. These examples, then, would seem to support the assumption that the ‘start to infinitive’ construction is often preferred when one wishes to focus on the very starting point of a process to the exclusion of any consideration of its possible continuation. ‘Begin to infinitive’, on the other hand, is often used to describe the commencement of a process that is likely to continue. Strictly speaking, of course, it is only of situations in the past that we can be sure to what extent they actually continue. One might venture the tentative suggestion that this very fact influences the speakers’ preference for ‘begin to infinitive’ in the past and ‘start to infinitive’ in the present. Duffley (1999) proffers another explanation for the difference between begin and start. He writes:


260

Begin is the more neutral and general of the two verbs and simply designates the initial segment of an event, which means that it is compatible with the evocation of only the very first instant of something (cf. at the very beginning of the chapter), but of course will tend to suggest a more extended portion of an event […]. Start on the other hand, contra Freed (1979:71), who defines it as evoking an onset, does not refer inherently to a segment of an event at all, but rather evokes the notion of breaking out of a state of rest or inactivity, or, in its transitive use, initiating an event by breaking out of a state of rest or inactivity. (Duffley 1999: 309)

Of the two explanations, I think Freed’s is better able to account for utterances in which the complement situation is not realised, such as (725) and (726). Duffley, too, recognises the possibility that complements of ‘start to infinitive’ may not be realised. He writes: “Due to the fact that start, unlike begin, does not inherently designate a segment of an event, the notion of breaking out of a state of rest or inactivity which it denotes can also be construed merely as movement towards the first moment of the infinitive’s event, in which case the latter will be understood to be non-initiated” (Duffley 1999: 316). If we take this “first moment” of Duffley’s to correspond to Freed’s “onset”, there may not be much of a difference between their two approaches after all. The next question to require our attention is why ‘start to infinitive’ takes a higher percentage of Agent subjects (61%) than ‘begin to infinitive’ (45%). (Table 6.15 shows that both ‘begin -ing’ and ‘start -ing’ strongly favour Agent subjects.) We may gain a clue to the answer to this question in the semantics of the matrix verbs themselves. When not functioning as a same-subject matrix verb start is much more causative than begin: whereas “she started a course in linguistics” might be spoken of a professor, “she began a course in linguistics” is more likely to refer to a student. Spoken of a student, “she started” might well be taken to imply “but did not finish”. Moreover, of the two, only start takes a different-subject complement clause, as in “Much before that the whole biblical use that starts the whole sales project going.” (J3U 68). Now, one of the defining features of the prototypical Agent in a transitive construction is that he or she causes a change in a Patient. It is symptomatic of how little causativity is discernible in many examples of the ‘begin to’ construction that at one stage generative grammarians (Perlmutter 1970, for instance) postulated the existence of two separate ‘begin to’ constructions, one transitive and one intransitive. For present purposes it suffices to point out that causativity and agentivity tend to go hand in hand. According to Dowty (1991: 572), for instance, causativity and volition (which presupposes animacy) are both entailments of the role of Proto-Agent. As start is the more causative of the two verbs, in the sense that it can be used causatively in other constructions, it is not surprising that ‘start to infinitive’ occurs with a high percentage of Agent subjects. Having now looked at three ways in which start differs from begin, to wit the incidence of examples with first- and second-person subjects, of examples in


the present versus past and of examples with agentive subjects, we now turn to the other parameter evincing differences in distribution, that of to infinitive versus -ing complements. Here there are two aspects in particular of their distribution that call for an explanation. The first is the high correlation between agentive subjects and -ing complements, the second the almost complete absence of such complements after -ing forms of both matrix verbs. With regard to the first question, it was noted above that ‘start to infinitive’ occurs with Agent subjects in 61% and ‘begin to infinitive’ in 45% of cases. The corresponding figures for ‘start -ing’ and ‘begin -ing’ are 87% and 88%, respectively. Lest we be led astray by these percentages, it ought to be pointed out that, when faced with a situation in which an Agent initiates an activity or process, the raw figures in Table 6.15 indicate that the speaker is more likely to encode it using ‘begin to infinitive’ than ‘begin/start -ing’. (If, however, we distinguish between ‘writer’ and ‘speaker’, only the former would be expected to choose ‘begin to infinitive’: the latter would be more likely to favour ‘start -ing’.) The point here, however, is that animate subjects would appear to be almost always agentive with ‘begin -ing’. The question is why this should be so. Let us consider in (733) – (738) some examples of both -ing constructions.25 To facilitate comparison, they are all past tense. (733) A little boy came out of the woods opposite and began skiing down the

slope towards the road. (B0U 993) (734) He came out from refreshments, and we were all bored to death, waiting to go round again, flings his cape round his shoulder, took his helmet off, and started dancing down the road, throwing flower petals about. (B24 830) (735) Then they began following my daughter and she was forced to leave the

country as well. (EV5 955) (736) When she was 15 Mandy started taking the pill. (CBF 10123) (737) Things began fitting damningly into place. (K8R 1149) (738) Erm I started losing my hair at twenty nine (HVE 157) The first point to be made in connection with ‘start -ing’ and ‘begin -ing’ is that the distinction noted above between the two matrix verbs with respect to whether they encode the necessary enactment of the complement situation is no longer operative. Both -ing constructions necessarily entail the initiation of the complement clause situation. The -ing form in (733) – (738) encodes an ongoing situation. In each case the complement situation either has a certain duration, if the meaning is continuative, or must recur a few times, if it is iterative.26 The

25 Some of these examples have been cited earlier in section 3.5, (733) as (102), (734) as (103), (739) as (104) and (740) as (105).

26 Note, in this connection, the following two examples of Dirven‘s: “ (54)a. The clock began to strike twelve. (= the first stroke). b. The clock began striking twelve. (= various strokes).” (Dirven 1989: 130)


262

situations of skiing and dancing in (733) and (734) are single unfolding activities. following and taking in (735) and (736), on the other hand, encode recurring situations. It is this latter type with regard to which Quirk et al. (1985: 1192) note a preference for the-ing complement. My data indicate that this preference is particularly marked in the case of ‘start -ing’, ‘begin -ing’ being more commonly employed to encode continuative situations. (737) and (738) illustrate the fact that both -ing constructions may be employed with non-agentive subjects. (739) – (744), the first four of which contain to infinitive constructions with agentive subjects, the other two to infinitive constructions with non-agentive subjects, are cited here for purposes of comparison with (733) – (738). (739) They began to ascend the staircase, holding Jimmy between them. (G0E 3461) (740) He started to haul the punt up the rollers. (BMR 1617) (741) Nobody bought them so I began to repair fridges. (A95 138) (742) Alexander started to direct his actions by inventing orders which

he gave to himself when standing, sitting and reciting. (BM0 131) (743) Locals began to fear for her safety. (ABC 79) (744) She curled up into a small ball, too tired to stay awake, but too scared to

go to sleep in case she started to dream again. (H8F 1886) Let us start by asking ourselves which of the complement situations in (733) – (744) could have been felicitously encoded using the alternative form of complementation. The -ing complements in (733) and (734) entail their to infinitive counterparts and could be replaced by these with little change in meaning. They may be compared in this respect to (739) and (740), which also describe once-off activities and which could also be encoded by means of -ing complements. The context might however have precluded this in a case like (740) if it were to transpire that no actual hauling took place. The difference between the two pairs of constructions is that one focuses on the starting point of the situation, either the onset or the first part of the nucleus in Freed’s (1979) terms, whereas the other profiles the situation as continuing to unfold after it has been kick-started, so to speak, by the matrix verb. The picture is similar in the case of repeating situations, such as those in (735) and (736). These -ing complements leave us in no doubt that the situations they encode are repeated on several occasions. The to infinitive complements in (741) and (742) must also be taken to refer to repeated actions. There are, however, only a handful of these in my material, confirming the assertion of Quirk et al. referred to above about the preference for -ing forms to describe “multiple” activities. Several scholars are of the opinion that the choice between the two complement types is related to differences between the situations to be encoded in terms of concreteness or specificity. Conrad, for instance, maintains his usual distinction between -ing complements as referring to concrete events and to infinitive complements as denoting a disposition. He writes: “When the gerund is used after these verbs it refers to one or more concrete, individual instances of the


action denoted by the gerund. The infinitive has the negative characteristic that it does not refer to concrete, individual instances of the action. It is non-referring and it usually mentions dispositions” (Conrad 1982: 146). Although this analysis may suffice to account for the use of the -ing form in (733) – (734), it cannot account for the use of the to form in (739) and (740), to mention just two utterances in which this form appears to refer to concrete events. Freed, in a passage quoted in section 5.5.5 (p. 185), maintains that the to infinitive after verbs such as begin, start and continue evokes a generic reading while the -ing form is preferred for encoding specific once-off or repeated situations. While this distinction holds up in the case of continue, the evidence of the data tends to invalidate the hypothesis in the case of start and begin. We have already encountered, in (739) – (740), tokens with the to infinitive which clearly encode specific predications. (745) – (748) resemble these in this respect. (745) The two men started to dig down among the scattered bones to find other

signs of interest. (CE9 978) (746) With a meekness that made her feel mean, he knelt down on the floor and

started to gather his cuttings together. (AC4 565) (747) Recovering, Emmie seized the thin old hearthrug and began to beat at the

flames. (HH9 3666) (748) The leopard flattened his belly to the ground and began to creep forward. (CK2 393) There can be no doubt that the situations encoded by ‘start to’ in (745) and (746) refer to concrete situations which are to be realised immediately on the occasion in question, as are the situations encoded by ‘begin to’ in (747) – (748). Equally, general predications in the projected future may be encoded by ‘start -ing’ and ‘begin -ing’, as in (749) – (752). (749) ‘When cocaine was made cheaper in the form of crack, that’s when things

started dying, that’s when more crimes started happening, that’s when more people started dropping out of school to make that dollar.’ (ED7 940)

(750) Even before autumn sets in, gardeners start thinking about spring and putting in bulbs for next year (A0G 408)

(751) Like many other coastal shipping companies it met hard times after the First World War, when motor lorries began taking the general cargoes that used to go by sea to all the little ports. (H0D 546)

(752) Thus, for instance, Italy began building in a Romanesque style earlier than most nations - for its emergence from Roman work was a natural corollary - continued it later, produced little Gothic work and then burst forth into the Renaissance a century and more before the rest of Europe. (HR1 21)

The predication of things dying, crimes happening and people dropping out in (749) are all higher order predications. They do not encode concrete instances of


264

the realisation of these situations on specific occasions, but rather the likelihood of their being realised on suitable occasions. Similarly (750) does not encode concrete instances of thinking by specific gardeners, but rather the likelihood of any gardener realising such a situation just before the advent of (every) autumn. (751) does not encode specific acts of transport by specific transporters. Nor does (752) encode specific acts of building in Italy. Tokens like these are by no means rare in the corpus. Figure 6.10 shows the percentage number of tokens of each of the four constructions that encode both specific and general predications.

0,0

10,0

20,0

30,0

40,0

50,0

60,0

70,0

begin to begin -ing start to start -ing

General

Specific

Figure 6.10: Percentage of tokens of ‘begin to’, ‘begin -ing’, ‘start to’ and ‘start -ing’ encoding general and specific predications

While the percentages in Figure 6.10 show that both to infinitive constructions are more likely to encode general predications and both -ing constructions specific predications, only in the case of begin is the difference significant at the level of p=0.05. In any case, there is so much overlap between the two complement types that we can conclude that Freed (1979: 152) overstates her case with respect to this distinction. (745) – (748), (750) and (752), like (733) – (736) and (739) – (742), are all examples of constructions with agentive subjects. We have yet to address the question as to why this form of subject should be so predominant in the case of -ing complements. Before doing so, however, let us just note that -ing clauses may follow subjects that encode other semantic roles such as Patient, as in (737), and Experiencer, as in (738). Both of these could be paraphrased felicitously by to infinitive complements. The reason for employing the -ing complement form in (737) and (738) is presumably to emphasise the continuative nature of the process initiated. These examples with non-agentive subjects are, however, exceptions to


the general tendency. A tentative attempt to explain this general tendency might begin by pointing out that all -ing clauses following begin and start, whether they be continuative or iterative, encode a process which has been set in motion and which is continuing to evolve. Whereas an ingressive to infinitive construction may merely mark the starting point of a process, the -ing constructions imply that the process continues unfolding. It appears that when a language user is aware both of the identity of the causer of an initiated situation, and the fact of its continuation, he or she tends to encode this combination by means of a Forward-looking -ing construction. The co-occurrence of these two elements may at the very least be assumed to be a strong motivating factor in the choice of this form of construction. Before rounding off this consideration of begin and start, we should look briefly at the almost complete absence of -ing complements after -ing forms of the matrix verbs, the so-called “double -ing constraint”, to employ the term coined by Ross (1972). We may also note the scarcity of to clauses after to infinitive matrix verbs. With regard to the latter, the figures in Table 6.13 may be misleading. The table contains nine instances of infinitive forms of ‘begin to infinitive’, but seven of these matrix verbs are bare infinitives. Before looking at this reluctance to duplicate either to infinitive or -ing, I will cite some examples of exceptions to this constraint. (753) looks nice, no burnt or anything, and I says ee it’s lovely is it, I said

alright, would been down for this morning like and I says ooh come on then don’t bother when you get home starting making anything I says just say, let’s ask for steak pies, so I says er can we have er two steak and kidney pies please?, she said I’m sorry we don’t, there isn’t any kidney an all, I’m glad there isn’t I didn’t want, well I’d of eaten ’em like but I’d rather have them without (KSS 4028)

(754) However, it is also important to begin to distinguish what is real from what is fictional. (HXF 378)

(755) ‘Leith never said anything about agreeing to marry you when I saw her in the -’ Travis left his own unhappy thoughts to start to exclaim.

(JY1 1809) In the case of prototypical Aspect verbs, such as begin and start, the matrix verb and the complement clause predicate both refer to one and the same situation. The complement clause predicate indicates the nature of the situation in question and the matrix verb indicates whether it is beginning, continuing, ending or of regular occurrence. So close is the relationship between the two that Brinton (1988) maintains that these matrix verbs function semantically as auxiliaries. If we keep this fact in mind, we will be able to see why the implication of either a possible alternative to, or the continuative nature of, the complement clause situation need not be reduplicated. Contrast this with cases where the matrix verb and complement verb encode completely separate situations: then there is no such restriction; we may readily say such things as “I am considering taking civil


266

action.” (from K5D 12726), for example. The same point applies to the to infinitive constructions: repetition of the marker of a possible alternative would in most cases be semantically redundant. However, redundancy in these contexts is not absolutely excluded by the grammar, as examples (723) and (753) – (755) prove. In (753) there is no obvious reason why a to complement would not have served the speaker’s purpose just as well. I have certainly been able to detect no stylistic or syntactic reasons motivating the repetition in this or any of the other examples. Presumably the repetition of the continuative marker may be put down to the pressures of speech production in the conversational context. In (754) an -ing complement would perhaps have been equally felicitous, although the choice of the to infinitive does have the merit of pointing to the very starting point of the distinguishing activity and not its continuation, the speaker emphasising thereby that it is taking the first step that is of crucial importance. Similarly, in (755) the duplication of to may be motivated by the speaker’s wish to show that Travis breaks off before completing his exclamation. The first to, indicating the purposive goal, might be interpreted as ‘right up to’, the second, indicating the goal of starting, as ‘towards’. In this, admittedly subtle, manner, the ‘double to’ construction may convey a shade of meaning that would be absent from a ‘to start -ing’ construction. I conclude this discussion of the various complements of start and begin with a short summary of the questions addressed. I have tried to shed some light on the following five questions: (1) why start occurs more in the present, begin in the past tense, (2) why start occurs more often with first- and second-person subjects, (3) why start occurs more with Agent subjects, (4) why both -ing constructions occur more with Agent subjects, and (5) why there are so few examples of ‘double -ing’ and ‘double to’. The difference in person and tense between the two verbs has been linked to the tendency to prefer start to begin in face-to-face conversations. The difference between them with respect to the incidence of agentive subjects has been related to the semantic factor that start in general is more causative then begin. As for the differences between the complement types in terms of agency, we have related this to the ongoing/ repetitive nature of a situation encoded by -ing. Finally, the avoidance of ‘double -ing’ and ‘double to’ was related to the semi-auxiliary-like status of both verbs. Finally, one should mention the fact, pointed out by Mair (2002: 117) and Skandera (2003: 349), that both ‘start -ing’ and ‘begin -ing’ are spreading at the expense of their to infinitive counterparts in American English. There are certain indications that the same development may be expected to materialise in British English, at least in the case of begin, if it is not yet underway. Although there has been no change in the ratio of -ing to to infinitive complements in the case of begin in British English from the early sixties to the early nineties, there has been a statistically significant change in the Full Nominal : Pronominal Subject Ratio, to the effect that there are more full nominal subjects in FLOB than LOB. This sort of change often precedes, or accompanies, the spread of a construction in a language (see Egan 2006). Needless to say, this sort of development would serve to further blur the distinctions between the constructions.

To infinitive versus -ing clauses: a summary 267

6.3.4 To infinitive versus -ing clauses: a summary

In this section we have looked at a variety of Forward-looking constructions containing to infinitive and -ing complements. We have focussed on constructions containing the matrix verbs anticipate, intend, fear, dread, want, start and begin. The -ing complements in all these constructions encode situations profiled as very likely indeed to come about in the projected future. In the case of the Mental Process and Attitude verbs it is the (matrix verb) subject who is almost certain that the complement situation will come about. This certainty with respect to realisation implies that the eventual realisation of the complement clause situation is within the power of the subject to bring about in Mental Process constructions like ‘intend -ing’, but outside the control of the subject in the case of Attitude constructions like ‘dread -ing’. In the latter case it is actually the to infinitive form of complement that implies greater independence on the part of the matrix verb subject. In the case of the Aspect matrix verbs, it is the speaker who is sure of the realisation of the complement situation. In all Forward-looking -ing constructions the complement situation is profiled as imperfective, durative and ongoing if it encodes an activity, repetitive if it encodes an achievement or accomplishment. As for the to infinitive complements of these verbs, these profile the future situation as less certain of realisation. This is particularly true of the Mental Process and Attitude constructions. The complements focus on one possible future scenario as more likely of realisation than any alternatives. This is the targeted alternative, which is either something to be striven for, in the case of intend and want, or striven to be avoided, in the case of fear and dread. The to infinitive complements of begin and start are more like their -ing counterparts in so far as their complement situations are almost always realised (only ‘start to infinitive’ admits of non-realisation). We have, however, seen that there are material differences between the two forms of construction with respect to the types of subject with which they typically occur, with the -ing constructions favouring Agent subjects more than the to infinitive ones.

6.4 Causation constructions

The present study contains thirty-six Causation constructions. They are all listed in Table 6 in Appendix 2. Causation constructions were defined in section 2.2.1 as constructions in which S1 (the matrix verb subject) directs the realisation of the complement situation by S2 (the complement subject), who has no independent say in the matter. This definition is rather too rough-hewn in two important respects. In the first place, S1 may direct the non-realisation, rather than the realisation, of the complement situation, as is the case with matrix verbs like prevent and stop. In the second place, the assertion that S2 has no independent say in the matter is too sweeping. It is, for instance, true of ‘compel S2 to infinitive’, but less so of ‘impel S2 to infinitive’. There is, however, one feature that is shared by all Causation constructions. There is never any doubt as to whether the


268

complement situation is realised. Moreover, apart from six matrix verbs like prevent that always encode the non-realisation of the situation, realisation is always the result. This section will focus on the four matrix verbs cause, force, get and make. The reasons for this are that they are all quite common and that they differ as to their preferred forms of complementation. Have would also have been included, had there been sufficient occurrences of causative have among the 1,000 downloaded tokens. As may be seen in Table 6.16, get is represented by only twenty-seven occurrences in all, but it was decided to include it, in order that at least one verb taking an -ing complement would be present in the discussion. One occurrence of ‘cause S2 bare infinitive’ is excluded from the discussion below, as it may be considered somewhat idiosyncratic in present-day English. (The token is cited in Appendix 1.) This leaves us with five constructions to consider, ‘cause S2 to infinitive’, ‘force S2 to infinitive’, ‘get S2 to infinitive’, ‘get S2 -ing’ and ‘make S2 bare infinitive’. Figures for the animacy of both subjects for each construction are given in Table 6.16.

Table 6.16: Animacy of S1s and S2s after matrix verbs cause, force, get and make, with vertical percentages

cause to force to get to get -ing make bare Animate S1s 14 10% 130 52% 20 100% 5 71% 28 33% Animate S2s 67 46% 243 98% 18 90% 3 43% 52 61% Total 147 249 20 7 85

The data in Table 6.16 display more variation with regard to the animacy of S1 than that of S2. However, if one considers the whole semantic field of Causation constructions, one finds that there is a closer correlation between the animacy of S2 and the form of the complement than is the case for S1. In particular, constructions in which S2 is prototypically animate, containing such matrix verbs as drive and compel, always contain a to infinitive form of complement. Our five constructions are situated on a cline with regard to the animacy of S2 from ‘force S2 to infinitive’ to ‘get S2 -ing’, as illustrated in Figure 6.11.

S2 most force get (to) make cause get (-ing) S2 least animate animate

Figure 6.11: Animacy of S2s after matrix verbs cause, force, get and make

Causation constructions 269

As may be seen from the data in Table 6.16 and Figure 6.11, the -ing complement form of get is situated at one extreme of the S2 animacy cline with two of the to infinitive complements at the other extreme. ‘Cause S2 to infinitive’ appears to be the ‘odd man out’, as it were, being situated to the right of ‘make S2 bare infinitive’. One of the hypotheses about the meaning of the to infinitive complement form advanced in Chapter 4 is that it evokes the idea of an alternative course of action to that actually pursued in the complement clause. One might expect such an alternative course to be particularly salient in the case of S2s suborned by S1 into doing something against their own will. Langacker (1990: 257) points out that the case ending of S2, which he terms the “pivot”, in causative constructions in Japanese varies according to whether or not S2 is a willing participant in the realisation of the complement situation. An investigation of the data with a view to ascertaining the degree of reluctance to comply with the will of S1 exhibited by the S2s in the various constructions showed that all five constructions contain such occurrences. (756) – (760) illustrate this sort of situation. (756) ‘If he causes my cousin to fail her exams, I will indeed fire him.’ (JXT 1146) (757) In the end, the Know Nothings, led by Senator Jesse Helms, managed to

force Washington’s Corcoran Gallery to cancel a 150-picture Mapplethorpe retrospective and started a full-scale attack on the kinds of artistic projects which the National Endowment of the Arts (no radical force) reluctantly and sparingly backed. (A5K 264)

(758) ‘It’s very difficult to get farmers’ wives to take exams,’ says Sue. (ACR 2223) (759) I think you ought to go I’ll get you going real quick. (KDH 372) (760) ‘Tories made us axe 3,000 town hall jobs’ (CBF 12422) Occasionally, reluctance on the part of S2 may be signalled overtly by the speaker. (758) is a case in point, where it is indicated by the adjective difficult. On the whole, however, listeners must make use of their world knowledge to interpret the degree of reluctance correctly. We know, for example, that students do not normally fail exams willingly, that galleries dislike having to cancel exhibitions and that local politicians are averse to cutting the municipal workforce. It is knowledge of this sort that enables us to conclude that S2 in (756), (757) and (760) is reluctant to submit to the will of S1. (This may not be strictly true of S2 in (756), in so far as the possibility exists that the cousin may not actually be aware of the causation at work.) In none of the constructions, however, is reluctance invariably present. No such reluctance may be detected, for instance, in (761) – (765). (761) Watching her, Lucy was struck by a sudden thought, her intuition causing

her to say slyly, ‘Affair being the operative word. (HHB 2331)


270

(762) Coming out of Beckley a wandering Triumph Herald forced us to check the anti-lock brakes (they work, and well). (ED9 1741)

(763) It is obviously important to get people to join the G M B and it’s just as important to keep them as members once they have joined. (HUC 573)

(764) ‘They have lost confidence, but Martin is just the player who can get them going and I think you will see their true character soon. (CH3 5784) (765) I thought of Biblical characters dressed in white flowing robes walking

about the arid plains of Palestine or striding along the shores of the Sea of Galilee, making lepers leap into the air and hurl away their bells.

(B19 777) There is no reason to assume that any of the S2s in (761) – (765) would be reluctant to realise the complement situation. In (762) S2 would no doubt have preferred not to have to put the brakes to the test, but having been put in that position, their personal safety obviously required the realisation of the complement situation. As for the remaining four examples, the attitude of S2 to the complement situation presumably varies from indifference in the case of (763) to downright enthusiasm in (765). Examples (756) – (765) show that all five constructions may contain both reluctant and what we may label more ‘neutral’ S2s. Table 6.17 contains figures for both types of S2 for each construction, the ‘neutrals’ being given the benefit of the doubt where any such existed.

Table 6.17: Reluctance exhibited by S2s after matrix verbs cause, force, get and make, with vertical percentages

cause to force to get to get -ing make bare Reluctant S2 27 18% 238 96% 11 55% 1 14% 25 29% Neutral S2 120 82% 11 4% 9 45% 6 86% 60 71% Total S2s 147 249 20 7 85

Table 6.17 reveals a much more marked difference between the constructions than does Table 6.16. With the exception of ‘cause S2 to infinitive’, the constructions containing the to infinitive exhibit, as expected, more instances of coercion by S1. At this point it may prove of interest to enquire as to the nature of the complement situation, and in particular whether it is one that, under other circumstances, S2 could have realised as a result of the conscious exercise of will, in other words whether S2 is to some extent agentive. Of course, given that one of the characteristics of prototypical agency is the ability to control one’s actions, and that Causation constructions encode the usurpation of this ability by S1, it is obvious that no S2 functions as a fully-fledged Agent in these constructions. As Givón puts it: “The more successful a manipulation is, the less control does the manipulee have, and the less agent-like the manipulee is” (Givón 1990: 51).


Nevertheless, the agentivity of S2 has been shown to be relevant to the choice of complement structure in other languages. Achard (1998: 98), for instance, considers the degree of agentivity of S2, which he relates to the salience of S2, to be a major factor in influencing the choice between the two French causative constructions which he labels “VV” and “VOV”. Table 6.18 contains details of the agentivity of S2 in all five constructions.

Table 6.18: Agentivity exhibited by S2s after matrix verbs cause, force, get and make, with vertical percentages

cause to force to get to get -ing make bare Agentive S2 26 8% 233 94% 18 90% 3 43% 26 31% Non-agentive S2 121 82% 16 7% 2 10% 4 57% 59 69% Total S2s 147 249 20 7 85

Comparing Table 6.18 to Table 6.17, we find that ‘get S2 to infinitive’ has taken up a position closer to that of ‘force S2 to infinitive’ with ‘cause S2 to infinitive’ now at the other extreme. And the reason why cause differs so much from the other to infinitive constructions in Table 6.17 is now evident. Coercion by S1 presupposes possible agentivity on the part of S2. In the case of ‘cause S2 to infinitive’ most S2s are simply not agentive. I will now temporarily narrow the focus of inquiry to concentrate on the three constructions exhibiting the greatest degree of arm-twisting by S1, so to speak, as indicated by Table 6.17, ‘force S2 to infinitive’, ‘get S2 to infinitive’ and ‘make S2 bare infinitive’. Of these three, ‘force S2 to infinitive’ obviously encodes the greatest degree of antagonism between S1 and S2, ‘make S2 bare infinitive’ the least. Not only is the prototypical S2 noticeably more reluctant to carry out the wishes of S1 in the force construction, but he or she is also subjected to bending their will to that of S1, in a matter over which they would normally themselves have full dominion, to judge from the evidence of Table 6.18. (766) – (771) are some typical examples of the three constructions in this respect. (766) If you refuse to co-operate I can force you to pay me the money you owe

me. (H97 821) (767) Fergie’s worries are compounded by suspension, injuries and the

foreigners rule which will force him to bring in one 17-year-old youth player - and have three others on the bench. (CH3 2574)

(768) Now that’s different you’ll not get em to cancel it but to get them to spin it out over a longer period is a possibility and that’s what we’re gonna be

working towards. (K6W 295) (769) Well I think we’ll have to get some other quotes, because Peter isn’t the the roofer, he gets someone else to do it, so I think we ought to have some other quotes. (KLS 145) (770) Then he’ll make her wash her face and we can have tea. (EFJ 1926)


272

(771) He looked surprised when everyone laughed, then joined in himself, with very hoarse, loud laughter which made everyone start laughing all over again. (G12 2052)

These six examples differ from one another in various respects, both with regard to the animacy of both subjects, the agentivity and reluctance of S2, the immediacy of the resulting complement situation and the extent to which an alternative is evoked. Table 6.19 sums up these five features for the six examples.

Table 6.19: (766) – (771) contrasted with regard to animacy, agentivity, reluctance, immediacy and latent alternatives

‘force S2 to’ ‘get S2 to’ ‘make S2 bare’ (766) (767) (768) (769) (770) (771) Animate S1 + - + + + - Agentive S2 + + + + + - Reluctant S2 + + + - + - Immediacy of result - - - - + + Implied alternative + + + + - - (766) – (771) may be taken to be typical of the three constructions. Over 90% of tokens of ‘force S2 to’ resemble either (766) or (767) in terms of the five features in Table 6.19. 90% of the tokens of ‘get S2 to’ resemble either (768) or (769) and 86% of the tokens of ‘make S2 bare’ resemble either (770) or (771). One piece of information that may be gleaned from Table 6.19 is that ‘make S2 bare infinitive’ differs from both the other constructions in two respects. In the first place it always encodes a complement situation that follows immediately upon the action of S1. If one makes somebody laugh, for instance, the effect is always immediate. Duffley (1992: 60) interprets this to mean that the complement situation is “concurrent” with the matrix verb make, Dirven (1989: 124) that it is “contemporaneous”. Langacker notes that the strict temporal coincidence, which we find in the case of the bare infinitive complements of Perception verbs, is seldom to be found with Causation verbs. He goes on to make the point that “even here zero signals a kind of immediacy that is lacking […] with the to-marked complement of cause” (Langacker 1991: 444). The same point is made by Verspoor: “After causative verbs such as make, the event expressed in the plain infinitive construction is not co-temporal, but one that immediately (without time interval) follows the causative event” (Verspoor 2000: 215). Not only does the complement situation follow hard upon the act of causation, but there is normally no evocation of an alternative scenario, such as making somebody cry (or not making them laugh). If, on the other hand, one forces somebody to hand over a sum of money, or gets someone to cancel something, the resulting situation may well follow its cause at some temporal remove. Moreover, in both cases the inadmissible alternatives of ‘not paying’ and ‘not cancelling’ are hovering in the


background.27 Given these two differences between the constructions one may be tempted to ask whether the contribution of to to the ‘force S2 to infinitive’ and ‘get S2 to infinitive’ constructions is primarily to indicate lack of immediacy, or the idea of a latent alternative. The answer is, I think, that it indicates both. Recall the assumption, originally stated in section 4.3.1 (p. 95), that to instantiates a path-goal schema and that the two salient characteristics of such a schema are the twin elements of distance and direction. It is the element of distance that has been metaphorically extended to enable it to encode a lack of immediacy, the element of direction to encode the existence of a latent alternative. These two elements may be more or less prominent in the various constructions incorporating to infinitive complements. In the present instance, it just happens that both have a role to play. A third respect in which ‘make S2 bare’ clearly differs from the other two is in the incidence of non-agentive S2s (see Table 6.18 and example (771)). Although both to infinitive constructions may take such S2s, they rarely do so. If we ask what is the main difference between ‘force S2 to infinitive’ and the other two constructions, we may conclude, on the basis of the evidence in Tables 6.16 and 6.17, that it is the preponderance in the case of the former of reluctant S2s.28

27 Mittwoch maintains that one of the reasons why matrix verbs that take bare infinitive complements do so is that “on semantic grounds their complements either cannot be negated, as with perception verbs, or such negation is problematic, as with causatives” (Mittwoch 1990: 123). I would hesitate to accord much weight to this factor since causatives taking to infinitive complements are also very unlikely to be followed by a negated complement clause situation. There are no such examples in my material, and I have come across only one such in the BNC as a whole: “They forced her not to accept the house.” (A6V 741).

28 According to Dixon (1991: 194 & 2005: 197), force can imply coercion: in my material it almost invariably does so. Achard (1998: 99ff.), contrasting the causative French matrix verbs ‘faire’ and ‘forcer’, also emphasises the reluctant attitude of S2 in the case of the latter. He writes: “An important part of the meaning of forcer is that its subject makes it necessary for the causee to act in a certain way. In other words, forcer obligatorily involves a strong interaction between the main subject and the causee which, as a result of the interaction, must reluctantly initiate the infinitival process” (Achard 1998: 103). Achard goes on to explain the presence of the preposition à in the ‘forcer’ construction in terms very similar to those posited for to in the present study: “The figure is construed to follow a path to the ground via the process designated by the main verb. With forcer, the causee’s path towards the infinitival process needs to be made specific, because it is not naturally initiated by the causee, but induced by the main subject” (Achard 1998: 104).


274

One can get someone to do something or make someone do something they actually do not mind doing, as is the case in (769) and (771), but one can hardly force someone to do so. As for the ‘get S2 to infinitive’ construction, it differs from the other two in that it invariably takes animate S1s, as may be seen in Table 6.16. Both ‘force S2 to infinitive’ and ‘make S2 bare infinitive’ occur frequently with inanimate S1s, as in examples (767) and (771). Another area in which the three constructions differ is in relation to the incidence of reflexive S2s. All three occur with a large percentage of pronominal S2s as may be seen in Table 6.20.

Table 6.20: Pronominalisation of S2s after matrix verbs force, get and make, with vertical percentages

force to get to make bare no. of personal pronouns 88 35% 8 40% 4 5% no. of reflexive pronouns 36 15% 0 1 1% Total no. of S2s 249 20 85

Table 6.20 shows that the three constructions differ greatly with respect to the number of reflexive S2s. Indeed, although it is hazardous to base projected totals on a figure of 1%, we may, nevertheless, be tempted to assay the tentative conclusion that ‘force –self to infinitive’ exceeds ‘make –self bare infinitive’ by some two-to-one in real terms. Why should the construction exhibiting the greatest degree of polarity between S1 and S2 be preferred to encode situations where they both refer to the same person? One possible answer might be that the choice of the ‘force S2 to infinitive’ construction enables the speaker to highlight the very fact that this ‘same person’ plays two quite different roles in the whole on-stage situation. (772) – (775) illustrate both constructions with reflexive S2s. (772) Forcing herself to climb the stairs, she eventually regained the turret room. (JXU 1026) (773) It wasn’t easy, but she forced herself to go on meeting that blue gaze

without flinching. (JY7 2991) (774) ‘I forced myself to write a chapter or two about the good things he did:

getting rid of dead wood in the bureau; eliminating corruption among his agents; setting up a fingerprint system, an FBI laboratory that could serve as a technical resource for police forces all over the country. (H46 1049)

(775) ‘How has it been, Wayne?’ asked Rachel, making herself concentrate as she removed the old dressings. (JXY 68)

The three examples of the ‘force -self to infinitive’ construction in (772) – (774) would appear, at first sight, to resemble closely in meaning the ‘make -self bare infinitive’ construction in (775). We may, nevertheless, detect a slight difference stemming from the greater degree of reluctance exhibited by the prototypical S2


in the ‘force S2 to infinitive’ construction. Put simply, in (775) S1 may be engaged in causing herself to do something which she finds difficult to do. The same thing applies to (772) – (774). But in these examples there is the added implication that she feels a considerable degree of antipathy towards the action in question. This is not to say that a certain amount of antipathy may not be implied by the ‘make S2 bare infinitive’ construction. My tentative assertion is only that it is likely to be more pronounced in the case of ‘force -self to infinitive’. It is time to return to the ‘cause S2 to infinitive’ construction. The figures in Tables 6.16 – 6.17 reveal that in its case S1 is overwhelmingly inanimate, that its S2s are almost equally divided between animates and inanimates, and that, of the animate S2s, only about 40% (27 of 67) may be described as reluctant realisers of the complement or as playing an agentive role (26 of 67) in its realisation. It also takes fewer personal pronoun S2s than any of the other four constructions. One final difference between ‘cause S2 to infinitive’ and the other constructions is the incidence of matrix verbs in the simple present. Table 6.21 contains the TAM details for all five constructions. (776) – (778) exemplify some typical uses of ‘cause S2 to infinitive’ in the simple present, (779) and (780) some related uses with a modalised matrix verb.

Table 6.21: TAM of matrix verbs cause, force, get and make, with horizontal percentages, based on 147 tokens of ‘cause S2 to infinitive’, 249 tokens of ‘force S2 to infinitive’, 20 tokens of ‘get S2 to infinitive’, 7 tokens of ‘get S2 -ing’ and 85 tokens of ‘make S2 bare infinitive’



Mo-dal -ing inf

cause to 0 38 26%

0 35 24%

0 3 2%

6 4%

27 18%

26 18%

12 8%

force to

1 .4%

12 5%

3 1%

88 35%

2 1%

10 4%

5 2%

28 11%

57 23%

43 17%

get to

0 2 10%

0 2 10%

0 1 5%

0 5 20%

1 5%

9 45%

get -ing

0 0 0 0 0 2 29%

0 2 29%

0 3 43%

make bare

2 2%

10 12%

0 32 38%

0 3 4%

2 2%

10 12%

9 11%

17 20%

(776) This causes the nodes to contract regularly, thereby propelling the lymph around the body. (B7H 946) (777) A statement that causes the rest of the line to be ignored thereby allowing comments to be included in a program. (JXG 1625) (778) Formed from this association, her subsequent discourse invents

anthropomorphically and causes her production to be invested with fantasy and wit. (HAD 394)


276

(779) For large grammars this is especially the case since the interaction of many rules can cause the system to explode in a combinatorial manner. (HGR 2332) (780) For example, a rise in national income will cause L 1 to increase, and thus

will shift to the right. (J15 1187) (776) – (780) all encode statements of fact or opinion (disguised as fact!). This statement is often scientific in nature and may refer to ‘facts’ pertaining to various domains, such as biology in (776), technology in (777), art criticism in (778), grammar in (779) and economic theory in (780). In none of these examples is S2 animate, and therefore no reluctance or agentivity on its part may be detected. An enquiry as to whether the result encoded in the complement clause follows immediately upon the act of causation will elicit answers that vary from a definite ‘yes’, in the case of (776) and (777), to a possible ‘no’ in the case of (780). However, just as is the case with both ‘force S2 to infinitive’ and ‘get S2 to infinitive’, there are many occurrences in which there is no doubt that a time lapse exists between the act of causation and its subsequent consequence. One such instance has already been cited as (756). (781) and (782) are others. (781) The Preces and sermons, together with his controversial works, were to

cause Andrewes to become regarded as the leader in time and eminence of the Anglo-Catholic movement. (CFF 563) (782) Feathers, 20, hurled four-month-old Harry Gurney against a wall in a fit of temper, causing the tot to die in agony. (CH2 8240) It may be of interest to note that all three of the cited examples that encode an unmistakable delay in the resulting consequences contain animate S2s. This was not the case with the examples from more ‘scientific’ discourse. Perhaps in the latter case the consequences are felt to be unavoidable, rendering the question of the exact moment of their occurrence less salient. The last of the five constructions in this case study, ‘get S2 -ing’, may be illustrated by (783) – (784). (783) Help with feeding in the hospital had been poor, mainly because, says

Kim, the staff were hard-pressed to give even basic care, let alone the sort of unhurried support often needed to get breastfeeding going. (H07 243)

(784) Questions should be open ended so as to get the candidate talking. (B0M 2054) (785) We’re up and running with, we’ve got the test rig going, new spindle

bearings up, we’re pushing it. (JTA 1116) The figures in Tables 6.16 and 6.18 show that ‘get S2 -ing’ may take both animate and inanimate S1s and S2s, and that S2 may be either agentive or non-agentive. Table 6.17 reveals the biggest difference between it and the other four constructions. It is rarely used to encode the imposition of a situation on S2


against the will of the latter. (759) is the only example of this in my material and the context makes clear that it is said tongue-in-cheek. Another difference between the ‘get S2 -ing’ construction and the others is the nature and number of semantic fields from which its complement predicates are drawn. In my material 6 of 7 occurrences contain verbs of motion, either go, come or move, the only exception being (784). The most common of these verbs, go, tends to be used in its extended sense of start, ‘get x going’ meaning ‘start x’, as in (783) and (785), but it may also be used in its prototypical sense, as in (759). Common to all tokens of ‘get S2 -ing’ is the sense that the complement situation is ongoing and begins unfolding immediately after the act of causation. Let us now attempt to draw together the various threads of this discussion of the most common causation verbs, have excepted. I begin by summing up, in tabular form, the information to be gleaned from Tables 6.16– 6.21. Table 6.22 shows that ‘cause S2 to infinitive’ is the construction that differs most from the other four. It has the lowest percentage score with respect to animacy of S1, agentivity of S2 and pronominalisation of S2. In addition it is the only construction to favour the simple present to any extent. Of the other constructions ‘get S2 to infinitive’ and ‘force S2 to infinitive’ resemble one another quite closely, differing mainly with respect to the incidence of animate S1s and reluctant S2s. ‘Get S2 -ing’ and ‘make S2 bare infinitive’ resemble one another, and differ from ‘get S2 to infinitive’ and ‘force S2 to infinitive’, in the frequency of their occurrence with non-reluctant and non-agentive S2s.

Table 6.22: cause, force, get and make compared with respect to the various parameters discussed in this case study

Anim-

ate S1

Anim-ate S2

Reluct-ant S2

Agent- ive S2

Pronom- inal S2

TAM of matrix verb

Total

cause to

10% 46% 18% 18% 18% Present 26% Past 24% Non-fin. 26%

147

force to

52% 98% 96% 94% 50% Past 36% Non-fin. 40%

249

get to 100% 90% 55% 90% 40% Non-fin. 50% 20 get -ing

71% 43% 14% 43% 29% Non-fin. 43% 7

make bare

33% 61% 29% 31% 53% Past 38% Non-fin. 31%

85

The fact that there are considerable differences between the five constructions does not, of course, preclude their sometimes being used to encode very similar situations. A speaker is often able to construe a situation involving causation in various ways. His or her choice of construction will depend on which aspect of


278

the situation they wish to emphasise. A speaker who wishes to emphasise the antagonistic nature of a causation situation may choose to employ ‘force S2 to infinitive’, a speaker who wishes to tone down this aspect, ‘get S2 to infinitive’ or ‘make S2 bare infinitive’. Table 6.22 tells us something about the arsenal which the English language places at the disposal of individual speakers. How they choose to dispose of this arsenal is their own affair (and that of their interlocutors).

6.5 Forward-looking constructions: a summary

This chapter has been concerned with Forward-looking constructions. The projected future, defined in section 2.3 as the domain of situations after the time of the matrix verb as the subject or speaker expects these to evolve, is the only domain in which situations may be encoded using all four non-finite construction types. Indeed, one can in many cases also encode situations in this domain by means of finite complement clauses. The number of profiling options available to the speaker wishing to refer to future events is obviously related to the variety of ways in which we can view these same events. We can profile them as static wholes or as evolving processes, as more or less likely of realisation. Some matrix verbs only allow of one complement type, while other verbs allow of several. The focus in this chapter has been on the latter sort. In section 6.2 we compared bare and to infinitive complements with Communication, Effort, Enablement and Applied Attitude verbs. We saw that bare infinitive constructions tend to be used to intimate that the complement situation will definitely be realised in the (near) future. For instance, we saw that the ‘bid S2 bare infinitive’ construction is normally used to encode trivial requests with which one would expect the complement subject to comply at once. Similarly, if one lets someone do something, we can be (almost completely) certain that they will in fact do it. The to infinitive complement constructions, on the other hand, tend to be used to encode at least the possibility that what we have termed the targeted alternative might not be realised. We also saw that the distinction between the two construction types was less clear in the case of constructions containing help, dare and need. The reason for this is that the matrix verb in these constructions carries clear implications with respect to the realisation of the complement situation. In section 6.3 we looked at three sorts of Forward-looking constructions containing to infinitive and -ing complements, Mental Process constructions, Attitude constructions and Aspect constructions. We saw that the -ing complements in these constructions resembled bare infinitive complements in that they encode situations profiled as very likely indeed to come about in the projected future. They differ from bare infinitive complements in that the complement situation is not profiled as a whole but profiled as imperfective and durative. Again we saw that the to infinitive complements profile the future situation as less certain of realisation. The complements focus on the targeted alternative and encode it as more likely of realisation than any possible

Forward-looking constructions: a summary 279

alternatives. The two verbs begin and start resemble help, dare and need in that the differences in meaning between constructions containing various complement forms are harder to detect in certain contexts. Finally, in section 6.4, we contrasted three types of complement that occur in causative constructions. We saw that the complements of make resembled the complements in other bare infinitive constructions in encoding situations as following hard upon the time of the matrix verb. Nor is there any implication that the complement subject has an independent say in the realisation of the complement situation. In the to infinitive constructions, especially those with agentive complement subjects, we saw that the complement situation is profiled as just one of several alternatives, albeit one that the complement subject is under extreme pressure to realise. The -ing complement form in the ‘get S2 -ing’ construction encoded the complement situation not as a whole but as evolving in the future domain. Figure 6.12 summarises the most salient characteristics of the three forms which we have looked at in this section, as well as the to -ing construction, which may also be used to encode future situations. complement situation

profiled as a unitary whole

complement situation profiled as extended

almost certain to occur in projected future

bare infinitive -ing

likely to occur in projected future

to infinitive to -ing

Figure 6.12: Schematic characteristics of bare and to infinitive and -ing Forward-looking complement constructions

Figure 6.12 is based on Figure 4.11 (p. 143), the only difference being that the pertinent domain in Figure 4.11 is now given as the projected future. The characterisations of the four construction types in the table are not just valid for the constructions reviewed in this chapter. They are also true of the other Forward-looking constructions, containing matrix verbs which only take one sort of non-finite complement. For instance, in section 6.2 we looked at Communication constructions containing beg, beseech and bid, which take both bare and to infinitive complements. There are, however, twenty-seven different-subject Communication matrix verbs in the present study that only occur in to infinitive constructions, including such common verbs as ask, order and tell. These to infinitive constructions resemble ‘bid S2 to infinitive’ in profiling the complement situation as a whole and as likely, but by no means certain, to come about in the future. Similarly, there are twenty-four same-subject Forward-looking Attitude matrix verbs in the study. In section 4.5, we discussed aspire which occurs in both a to infinitive and a to -ing construction. In section 6.4 we


280

discussed fear and dread which occur in both to infinitive and -ing constructions. In section 7.3 we will discuss regret which occurs in a Forward-looking construction with a to infinitive complement and a Backward-looking construction with an -ing complement. Eighteen of the remaining twenty verbs (the exceptions are face and fancy) only occur in to infinitive constructions. And they always encode a targeted alternative in the projected future. The degree of likelihood of its realisation varies from construction to construction. The characteristics of the various constructions listed in Figure 6.12 may motivate the choice of complement form when the speaker is able to choose between various construals of a situation in the future domain. When faced with such a choice he or she may be guided by considerations which we have discussed in this chapter. Of course there is always the possibility that the mode of construal chosen is not of much consequence to the speaker in some contexts. In such cases, which are likely to occur most often with more auxiliary-like matrix verbs, such as help or start, the speaker may well reach for the construction ‘closest to hand’, so to speak. The question of which construction is closest to hand raises issues of degree of entrenchment and processing strategies which are far beyond the scope of this book to address.

Chapter 7 Constructions in contrast: Backward-looking constructions

7.1 Introduction

There are only five matrix verbs in the present study that occur both in Backward-looking constructions and in some other construction type. These are the Mental Process verbs remember and forget and the Attitude verb regret, which all occur in Forward-looking to infinitive as well as Backward-looking -ing constructions. The fourth verb is report which occurs in a Judgement to infinitive as well as a Backward-looking -ing construction. The fifth and final verb is cease, which occurs in a General to infinitive and a Backward-looking -ing construction. Because of the small number of these verbs the discussion below will also include mention of some semantically related verbs that only occur with one form of non-finite complement, such as recall and recollect in the case of remember, describe in the case of report and stop in the case of cease.

7.2 Mental Process constructions

There are eight Backward-looking Mental Process constructions in the present study, four same-subject constructions, and four different-subject ones. Both groups of four constructions contain the same four matrix verbs, recall, recollect, remember and forget. Examples (786) – (789) instantiate the four same-subject constructions. (786) I recall feeling an empty numbness on that day and for many months after. (B11 472) (787) Eva recollects making a commitment to Christ when she became a junior

soldier. (H7E 241) (788) I remember standing at the back of The Lyceum and looking down at the

crowd. (A6E 18) (789) Never would he forget sitting beside her in the night air watching an open-

air performance of The Parvenue and listening to the band at Fort House on Thursday evening. (H8A 34)

Recall, recollect and remember are quite similar in meaning. They each encode a summoning up to the conscious surface of the mind of a situation experienced by the subject in the past. As Conrad puts it: “There is a group of psychological

Constructions in contrast: Backward-looking constructions

282

verbs (remember, imagine, &c.) which, with the gerund, mean ‘to see in the mind’s eye’, ‘to contemplate mentally’. The gerund after these verbs indicates the mental ‘picture’ contemplated, or, in connection with remember and some others, the real event which was the original of a mental re-creation” (Conrad 1982: 176). The predicate in the complement situation may be stative, as in (786), or punctual, as in (787), but in all cases the use of the -ing form prompts a durative interpretation of the complement situation, entailing a repetitive or cyclical construal of accomplishments and achievements (see section 4.4.1.6 for a discussion of this form of construal). As long as the action of remembering lasts, the situation is played out before the mind’s eye. There are only two relevant tokens of same-subject ‘forget -ing’ among the downloaded tokens of forget (a third token, a Contemplation construction, is cited as example (93) in section 3.4, p. 52). Both of these are negated. The lack of examples of positive ‘forget -ing’ is presumably due to a semantic conflict between the profiling of a situation as unfolding in the past domain and the assertion that this same situation has left no trace in the subject’s memory. The other three verbs may also be negated, as in (790) – (792). (790) ‘I don’t recollect writing poems for anyone else for twenty year.’ (A0L 3006) (791) ‘I don’t recall mentioning that fact.’ (HHB 1717) (792) But, I don’t ever remember seeing him sit down to have any. (KBB 1355) When the matrix verbs are negated, as in (790) – (792), the subjects obviously have no memory of the situations on which to build their present imaging of them. Nevertheless, they still construct an image of them and, while not actually denying that they have taken place, strongly imply that this is the case. All three examples are in the first person, the first two being from written and the third from spoken dialogue. Note how in (792) the adverbial in the matrix clause actually refers to the situation in the complement clause; i.e. the sentence is a metonymic extension of “I don’t remember ever seeing…”. This is actually quite a common form of extension in spoken English: there are some forty examples of ‘ever remember’ in the BNC. Although recall, recollect and remember are similar in meaning there are some differences between their use. For instance there are twice as many first-person tokens with remember. And recollect tends to be modalised/negated more than twice as often as the other two. As for the question of meaning there would appear to be a slight difference between remember and recall/recollect with respect to the depths the subject has to plumb in order to retrieve the image of the unfolding situation in the complement clause. One gets the impression that in the case of remember the memory of the situation is often closer to the surface: in other words, remember is less likely to imply a previous act of forgetting than is recall. Adverbials used in the corpus with recall, such as ‘belatedly’, ‘now’, ‘suddenly’ and ‘later’, presuppose that the situation has been lost to the conscious mind for some time. There is no evidence that such adverbials cannot occur with

Mental Process constructions 283

remember, but the fact that none such do occur in the sixty-seven downloaded tokens would certainly indicate that they are not as common as is the case with recall. The four matrix verbs all occur in different-subject Backwards-looking -ing constructions, exemplified by (793) – (796).29 (793) I recall her pacing the sitting-room while I am doing my homework,


(794) I recollect one girl putting on dark glasses as a disguise in the hope that she would be luckier than Oliver Twist. (AMC 217)


(796) I will never forget Kevin McMonagle (Old Blue Eyes) singing Tom Leonard’s brilliant version of ‘My Way’ (‘though some may mock/the macho talk/upon the Walk/of No Surrender/I’ve drank the rent/I’ve clocked the wife/I’ve spewed my ring upon the fender’) or John Cobb as Alasdair’s sad flowerseller, or Siobhan Redmond’s Bo’ness hippy.

(B38 1411) The complement clauses in (792) – (796) all encode a situation involving S2 (the complement subject) that has occurred before the time of the matrix verb. In each case this situation has been experienced by S1 (the matrix verb subject). There seems to be no doubt that in each case we are presented with a situation which S1 recalls to his or her mind’s eye. The occurrence of adverbials such as while I am doing my homework in (793) and non-stop in (795) indicate that the complement situation is reviewed by S1 as unfolding over a certain time interval. These situations are clearly given a durative/cyclical/repetitive form of construal. In the case of situations with a more clearly defined right-hand boundary, such as those in (794) and (796), the notion of repetition may seem somewhat less apt. Although the actions of donning spectacles, in (794), and singing a song, in (796), may take some time to carry out, they do both come to a definite full stop (they are both accomplishments). However, from the perspective of an S1 operating in memory mode, as it were, the actions in question can be replayed again and again. Wierzbicka maintains that “If […] I remember DOING something, then I […] have a mental picture of myself doing it” (Wierzbicka 1988: 71). In a similar vein Verspoor writes: “[...] an -ing structure symbolizes that at the moment that the act of remembering or imagining is taking place, a mental representation of at least part of the event itself causes the recollection” (Verspoor 1997: 445). I would argue that irrespective of how great a portion of an event may have triggered the recollection, it is up to the ‘recollector’ to shape the content of his or

29 These tokens have all been cited previously, (793) as (40), (127), (236) and (294), (794) as (116), (795) as (237) and (796) as (249).


284

her own mental picture. It is in this sense that the -ing complement form may reasonably be interpreted as facilitating a cyclical construal. Only two of these four Mental Process verbs occur in to infinitive constructions. These are remember and forget. While remember outnumbers forget with regard to the incidence of -ing clauses, the opposite applies in relation to to infinitive clauses. There is an obvious reason for this. The person who has done something and then forgotten having done it is normally incapable of communicating this fact, for the very reason that he or she has forgotten it! The person who has forgotten to do something may well have now realised, or been made aware of, his or her sin of omission. At all events a decision to act in a certain manner has not been fulfilled and a speaker aware of the original decision may well feel moved to communicate about this. Table 7.1 contains the TAM details for the two constructions ‘forget to infinitive’ and ‘remember to infinitive’.

Table 7.1: TAM of matrix verbs forget and remember in to infinitive constructions with horizontal percentages, based on 99 tokens of ‘forget to infinitive’ and 19 tokens of ‘remember to infinitive’



Mo-dal -ing infin.

forget to

26 26%

6 6%

1 1%

33 33%

0 6 6%

13 13%

4 4%

7 7%

3 3%

remem-ber to

10 53%

2 11%

1 5%

4 22%

1 5%

1 5%

Note in Table 7.1 the number of examples of ‘forget to infinitive’ in the simple past and perfect and contrast this with the number of examples of ‘remember to infinitive’. If one has recalled to mind a prior decision to ‘put the cat out’, for example, and acted upon this recollection, one may subsequently communicate this fact by stating ‘I remembered to put the cat out’. However, unless one suspects that one’s interlocutor harbours suspicions about one’s memory, one is much more likely just to say ‘I put the cat out’. In the case of forget, a bald ‘I didn’t put the cat out’ would beg the obvious question of motive underlying the omission. Both ‘remember to infinitive’ and ‘forget to infinitive’, exemplified by (797) – (800), usually make reference to a previous decision which has been reactivated, or not, as the case may be. (797) Beryl sends her apologies - she forgot to give me your message. (ANY 1101) (798) I was half-way home before I realised that, with all the worry and talk about the accident, I had forgotten to ask Mrs McDougall about Ewen

Mental Process constructions 285

Mackay. (CKF 1103) (799) But at least, thought Charles, I remembered to pull up my socks. (ASE 2111) (800) so er we must remember to buy him a card and something (KD8 6309) In the case of ‘forget to infinitive’ the targeted alternative is never reached. In (797) the message was not given, and in (798) the question not asked. In (798) the action of asking is the goal or target of a previous decision towards which the subject should have been heading. The act of forgetting interrupts the subject’s progress towards this target. As for ‘remember to infinitive’ the complement always encodes the progress of the subject right up to his or her target. The decision previously taken is recalled and acted upon. Thus, in (799), the socks have been ‘pulled up’. Being in the simple past, (799) is unusual, in so far as it really only makes sense to use ‘remember to infinitive’ in the past if one wishes to contrast this particular situation with other real or imaginary acts of forgetting. The most common mood for ‘remember to infinitive’ is the imperative. As may be seen in Table 7.1 this accounts for over half of the downloaded tokens of the construction, as well as over a quarter of the ‘forget to infinitive’ examples (all negated). (801) – (805) are typical examples of both constructions with imperative matrix verbs. (801) ‘And don’t forget to bow to your partner, Mr Craven,’ our long-suffering

instructor would call out. (ACK 151) (802) No matter what you drink, remember to do your bit for the environment

byusing the bottle and can banks which are now available in most Sainsbury’s car parks. (BN7 1389)

(803) Now eat up and don’t forget to take your litter home with you.’ (ACK 2950) (804) Remember to keep rhubarb that has not been stripped of its

poisonousleaves well away from similar-looking cabbage or spinach; remove green parts and sprouting eyes from your potatoes. (BMD 1618)

(805) Don’t forget to take some Yorkshire parkin home with you - this delicious dark gingerbread is available from any good cake shop! (A70 2653)

These occurrences with imperative matrix verbs diverge from the occurrences considered so far in that they do not always refer to the reactivation of decision already made by the subject. In (801) we may presume that Mr Craven was previously aware of his obligation to bow, but is there any reason to think that the subject in (802) has made a conscious decision to do something for the environment and that the speaker is reminding him of this prior decision? What about (803)? Is there any evidence of a prior decision having been made by the subject to bring his litter home? The answer must be ‘no’. What the speaker is referring to is a decision made collectively by society, or at least by the environmentally friendly portion of society, and implying that the subject belongs to this group, thereby putting more pressure on the subject to comply with


286

society’s wishes. The point is that ‘remember to infinitive’ and ‘n’t forget to infinitive’, when used in the imperative, often do not refer to a genuine prior decision on the part of the subject, but to a, possibly unfounded, presumption by the speaker that such a decision has been reached. Why should the speaker make such an assumption? He or she may wish to avoid talking down to the subject, and may avoid this by implying that the subject has greater knowledge or foresight than is, in fact, the case. An example of this is (804). This sentence occurs towards the end of a magazine article on the dangers of poisonous plants. Nowhere else in the article is the need to keep rhubarb separate from other plants alluded to. The writer just presumes that his or her readers are already aware of this need, and is merely reminding them of it. This rhetorical device is also employed by advertisers who, instead of trying to convince potential customers to decide to try their wares, pretend to be merely reminding them of a decision already taken to do so. (805) is a typical example of this. To sum up, both ‘remember to infinitive’ and ‘forget to infinitive’ always encode in their complement clauses situations in the projected future vis-à-vis the acts of remembering and forgetting. In this respect, as has often been pointed out, they differ completely from their Backward-looking -ing counterparts. The former two encode the calling to mind of a decision or a commitment to act in a certain fashion, the latter two the calling to mind of an experience in the past. It was mentioned above that in the case of remember the memory of the situation often seems closer to the surface than with recall and recollect. If this is so, it would go some way towards explaining why these two matrix verbs do not occur in Forward-looking constructions. In so far as they both imply a previous act of forgetting, their subjects no longer have access to a prior decision or commitment to act in a certain manner.

7.3 Attitude constructions

The Attitude verb regret resembles the Mental Process verbs forget and remember in that it may be used to encode a reference to a past situation or to a situation that is about to be realised. It is the only Attitude matrix verb that may be used in these two ways. (806) – (809) exemplify the Backward-looking -ing construction. (806) But she did regret making her mother cry. (H7P 779) (807) Louise now greatly regretted having made Fleury the green coat, which

she feared made him too conspicuous ... and it was a fact that the sepoy sharpshooters could seldom resist trying to hit this brilliant green target. (EFW 285)

(808) Not that I regretted taking on the responsibilities, but it meant shelving any dreams I’d had.’ (JY8 2833)

(809) The mother herself appears to have thought so, for she later stated that she regretted having taken that step at that time. (FDL 132)

Attitude constructions 287

The complement situations in (806) – (809) are all clearly situated prior to the time of the matrix verb. This element of anteriority is doubly underlined in (807) and (809) by the use of the perfect auxiliary having. Almost a quarter (24 of 108) of the downloaded tokens are of this type. Langacker maintains that the use of the perfect form implies that a situation has current relevance: “Use of the perfect conveys that the event is not simply over and done with but continues to have some ‘current relevance’” (Langacker 1991: 211-212). This property may be detected in some, but not all, of the tokens in the corpus. In (807) for instance the present consequence of making Fleury into a brilliant green target is the reason for the subject’s regret. Other tokens where there is a clear intimation of current relevance (i.e. relevance at the time of the matrix verb) include (810) – (812). (810) Second-placed Ilford, unbeaten in the four matches played, must regret

thaving conceded their match with Clacton. (E9P 394) (811) He can laugh about it now, and has referred to it in many speeches since to

very good effect; but at the time it hurt. Not only that, it was brought up time and again by the media and used to undermine the serious work that he was trying so hard to do. Charles bitterly regretted having allowed the cameras in. (A7H 1315)

(812) Hodge and those around him believed Syngman Rhee should be encouraged and believed he could be controlled; they were later to regret having aided Rhee’s ambitions to the extent that they had done when they found themselves exposed to Rhee’s mordant censure. (EDP 344)

In (810) it is their current position in the league table that causes the football team to regret the concession of a match. In (811) it is the current hounding by the media that causes the subject to regret exposing himself to the camera. And in (812) it is the current fact of their being subject to censure by Rhee that causes the subjects to regret their pervious support. One cannot make too much of the question of current relevance in distinguishing between the two constructions, however, as there are also many tokens without having where the state of regret is clearly triggered by the current situation. (813) – (814) are cases in point. (813) Karen had a cheeseburger and lived to regret opting for it to be well done,

as it turned up dry as a bone inside a bap as wrinkled as Sid James’ face. (K34 37) (814) Vernon regretted giving Harcourt the cold shoulder; he would have been

someone to laugh with. (FNU 2430) Leaving aside the difference between the simple and perfect -ing form, there is no doubt that both of them are used to encode situations prior the time of the matrix verb and, as such, are clearly different to the ‘regret to infinitive’ construction, exemplified by (815) – (817).


288

(815) I turned the switch and I regret to inform you that out of the loudspeaker came a lot of old Irishmen singing maudlin songs. (CAS 1006)

(816) I regret to say that I forgot to give the Vice Chairman’s apologies for non- attendance. (J43 347) (817) Brown Windsor, I regret to add, is extremely well handicapped with

Ghofar. (AHC 240) In (815) – (817) the subject expresses an attitude towards a situation he or she is on the brink of realising. All thirty-two of the downloaded tokens of this construction resemble these three examples in having a first-person subject and a complement predicate encoding an act of communication. In both respects they resemble a minority use of the ‘hate to infinitive’ construction, discussed in section 5.5.2 (p. 175) and exemplified in (460), repeated here as (818), and (819). (818) I hate to say this guys but lets compare the Irish with the English squads

for tomorrow. (JIG 1509) (819) I hate to tell you this but the er the forecast for the weather is pretty awful

this year. (KPK 138) To sum up, both regret constructions resemble the Mental Process constructions discussed in the previous section with respect to the domains of realisation of their complement situations. The main difference between them is the more restricted distribution of ‘regret to infinitive’ with respect to type of subject, which is always first-person, and type of complement situation, which is always an act of communication.

7.4 Communication constructions

Communication constructions may be divided into two main sorts, those that encode a communication addressed to an on-stage participant, who is identical with the complement subject (S2), and those that encode some sort of comment on the onstage situation to another member of the audience, as it were. These two sorts of communication were introduced in section 2.2.1 (p. 22) and details of the incidence of both types are contained in Table 4 in Appendix 2. There are fourteen matrix verbs that are used to encode the communication by the matrix verb subject (S1) to a more or less explicit third party of some information about the onstage situation involving S2. This information may take the form of the expression of a judgement, in which case it is encoded by a to infinitive clause, or it may merely report developments onstage, in which case it is encoded in an -ing clause. One verb, report occurs in both types of construction. Rather than discuss it by itself, we will look at it in the context of both construction types, starting with the Judgement constructions. There are eleven matrix verbs in the study that are used to encode the expression of a judgement by S1 on some situation involving S2. These are acknowledge, announce, claim, declare, own, proclaim, profess, pronounce,

Communication constructions 289

report, suggest and tip. The communication is not addressed to S2 but to a third party in the audience, which may often be presumed to encompass the world at large. With the single exception of ‘tip S2 to’, which differs semantically from all the other constructions in that it encodes the expression of a judgement by S1 with regard to a future situation (in other words, a prediction), rather than a judgement about a past, present or timeless state of affairs, the complement clause predicate is invariably be. (820) – (823) are typical examples of Communication Judgement constructions. (820) After only three weeks Richter announced his experiment to be a success.

(CER 513) (821) The attitude of many older musicians and critics to science and technology is nothing more, of course, than the stale residue of the romantic, fin de

siècle aesthetic that, in the phrase of Villiers de L’Isle-Adam, claims science to be ‘the religion of the suburbs’. (ADP 318)

(822) He didn’t know ‘what statement Mr. Gibbon had made to the other gentlemen or what reasoning they could have employed so as to sign a paper declaring the fact of urinous vomiting to be utterly impossible - hereI must be at issue with them believing as I do ... from my little physiological knowledge, that vomiting of urine for 26 weeks is by no means impossible...’. (CMG 1388)

(823) What we in effect register when pronouncing something to be ‘the same’ X are merely certain similarities. (FTV 593)

In (820) – (823) S1 gives voice to a judgement he or she has arrived at concerning S2. In (820) S2 is the experiment in question, in (821) science in general, in (822) a putative fact, and in (823) the very general something. In each case S1 is human and the situation predicated of S2 is, in theory, one of several possible such situations. S1 maintains that it is the correct one. It may be of interest to note that the speaker often chooses to employ a passive construction in cases like this, thereby lending the judgement expressed the authority of a frequently unnamed, but presumably authoritative, judge! Four verbs, describe, mention, mime and report occur in constructions in which S1 merely describes an onstage situation to a fellow member of the audience without passing any judgement upon it. The first three of these verbs only take an -ing form of complement. All four are illustrated in (824) – (827), the first of which was previously cited as (238). (824) Interestingly, John Clare also describes old women singing and telling stories during the weeding and haymaking Evidently Complaining Daphne, though a pastoral, has an actual connection with the life of agricultural labour in Northamptonshire. (AN4 1088) (825) You just mentioned these salt ladies coming round. (K63 219) (826) Romany mimed them going about the boat, looking here and looking there. (FEM 1961)


290

(827) A potential failure of contraception had been recognised by 171 patients: 93 reported a split or leaking condom; 13 reported a condom coming off during intercourse; 32 admitted inconsistent use of condoms; 32 reported forgetting to take contraceptive pills or taking antibiotics concurrently with the pill; and one reported a late injection of medroxyprogesterone acetate. (FT1 729)

In (824) – (827) S1 calls the attention of other members of the audience to the enactment of some situation on stage, as it were. This situation is normally located in the past and normally encoded in an -ing clause. As Dixon puts it: “The meaning of describe implies reference to the unfolding of an activity, and this verb is restricted to an ING complement” (Dixon 1991: 218). However, as mentioned above, one of these verbs, report, also takes a to infinitive complement. There were three such occurrences among the downloaded data. All three are cited here: two of them occur in the same long extract, (828). (828) Thus (858), with an adjective which can occur predicatively (but not in the

prenominal attributive position), is quite acceptable; but (55), containing galore which is excluded from ordinary predicative position, is not: (858)the observers reported Her Majesty (to be) asleep (55) the observers reported the whisky (to be) galore We may further note that there are certain inherently restrictive adjectives (including superlatives and, under certain conditions, ordinals) which imply selection out of a known set, and which appear to occur (unlike most adjectives) in ordinary predicative position accompanied by the definite article: (860) (a) your slice is the largest (b) your slice is the small (See Chapter 7; and Ferris, in preparation.) (HPY 563)

(829) Pat’s Jester was defeated by Twin Oaks at Haydock on Wednesday, with jockey Neale Doughty reporting the horse to have finished sore.

(CBG 8299) The first point to be made about these examples is that two of them are not examples of actual language usage at all. They have been constructed for purposes of illustration, not of use. This does not mean that they may not reflect actual usage, although as the author of the textbook from which these examples are taken points out, one of them is not found in English (galore never occurring as a predicative).30 The second point is that in both (828) and (829) the complement is predicative, the to pointing to one of several possible states which might conceivably pertain to S2. In this respect they resemble the other Judgement constructions discussed above. Note that in (829) the judgement

31 The book in question is The meaning of syntax: a study in the adjectives of English by Conor Ferris (Longman 1993).

Communication constructions 291

pertains to a situation in the past, being encoded by a perfect form of the infinitive. To sum up, the ‘report S2 -ing’ construction resembles the other Backward-looking Communication constructions in encoding a situation as evolving in the period prior to the time of the matrix verb. The ‘report S2 to infinitive’ construction, on the other hand, encodes the expression of a judgement by S1 on some situation involving S2. When this situation is located in the past, as it is in (829), one of the features that distinguish the two constructions is blurred. In the next section we will consider an Aspect to infinitive construction that poses a similar problem.

7.5 Aspect constructions

There are six Backward-looking Aspect constructions in the present study. These consist of the matrix verbs cease, complete, discontinue, finish, quit and stop followed by the -ing form of complement. There are three general Aspect constructions. One of these, ‘continue to infinitive’ was discussed in section 5.5.5, where it was compared to the Same-time ‘continue -ing’ construction. The other two General Aspect constructions contain the matrix verbs use(d) and cease. Both constructions are used to encode general validity predications located prior to the time of the matrix verb. They thus obviously have a certain amount in common with the Backward-looking constructions. In order to tease out the differences in meaning between the two types of constructions, I decided to focus on cease which is the only verb to occur with both forms of complement and compare it to stop. Tables 7.2 and 7.3 contain details of the TAM relations and person and animacy of the subjects of the matrix verbs for the three constructions ‘cease to infinitive’, ‘cease -ing’ and ‘stop -ing’.

Table 7.2: TAM of matrix verbs cease and stop, with horizontal percentages, based on 563 tokens of ‘cease to’, 67 tokens of ‘cease -ing’ and 212 tokens of ‘stop -ing’


Imper-ative smp prog smp prog pres past

Mo-dal -ing inf

cease to

2 .4%

129 23%

5 1%

153 27%

5 1%

56 10%

77 14%

95 17%

29 5%

12 2%

cease -ing

2 3%

6 9%

0 17 25%

0 4 6%

11 16%

10 15%

1 1%

16 24%

stop -ing

25 12%

27 13%

1 .5%

45 21%

0 7 3%

26 12%

44 21%

0 37 18%


292

Table 7.3: Person, animacy and agentivity of subjects of matrix verbs cease and stop with horizontal percentages, based on 563 tokens of ‘cease to’, 67 tokens of ‘cease -ing’ and 212 tokens of ‘stop -ing’

1st person 2nd person 3rd person

animate

ag-ent

non-agent

ag-ent

non-agent

ag-ent

non-agent

3rd person inan-imate

Total anim-ate

Total agentive

cease to

11 2%

9 2%

4 1%

9 2%

105 19%

130 23%

295 52%

268 48%

120 21%

cease -ing

1 1%

0 0 0 54 81%

4 6%

8 12%

59 88%

55 82%

stop -ing

30 14%

3 1%

40 19%

2 1%

107 51%

10 5%

20 9%

192 91%

177 84%

A glance at the data in Table 7.2 reveals that ‘cease to’ tends to occur quite frequently in the Present and ‘stop -ing’ in the Imperative. In addition both -ing constructions occur quite frequently with infinitive matrix verbs. More revealing, perhaps, of the differences between the behaviour of the three constructions is the data on the person and animacy of matrix verb subjects contained in Table 7.3. The data reveal a considerable difference between ‘cease to’ and the other two constructions in terms of the frequency of animate and, more particularly, agentive subjects. The two -ing constructions in their prototypical uses would appear to share the following semantic component: ‘somebody was doing something at point x: at point y (y>x) they were no longer doing it’. The ‘cease to’ construction, on the other hand, prototypically contains the semantic component: ‘a certain situation had been occurring at intervals or had pertained for some time at point x: at point y (y>x) this was no longer the case’. It is this element of something having pertained ‘in general’, as it were, which has led to the classification of ‘cease to’ as a General construction. (830) – (834) are typical of ‘cease to’ in this respect. (830) The Labour Party ceased to contest by-elections and ruled out extra-

Parliamentary agitation of any kind as incompatible with its membership of the Government. (CE7 1487)

(831) I'm going to love you and make love to you till the sunset ceases to happen.’ (JY4 4181)

(832) Companies may cease to exist by effluxion of time, by appropriate resolution or by court order for dissolution of the company. (EEH 710)

(833) The order will terminate when the child ceases to be of compulsory school age or if a care order is made. (AN5 357)


(834) So the man on the street ceased to be the prime target of charter-train salesmen. (A11 669)

(830) and (831) encode habitual situations which have been realised by the subject on appropriate occasions, in the case of (830) whenever a by-election is called, in the case of (831) every evening. (832) – (834) have generic subjects, and the predications in these may also be expected to have been realised by appropriate subjects on suitable occasions. These general validity predications differ from those which we have hitherto encountered in one important respect. Located as they are prior to the time of the matrix verb they entail that the situation in the complement clause has actually been realised on some (non-specific) occasion(s). Thus they share one important semantic property, that of actuation, with Backward-looking constructions. Indeed in a very small number of cases, ‘cease to’ itself encodes a specific Backward-looking predication. (835) and (836) are two of these. (835) Dr Hadley, who had asked the questions, ceased to play with the two-

toned ball pen he had been using as a remote muscle of his tongue. (ADA 2367)

(836) Because the fingers within her own warm hands ceased to flutter agitatedly, a brief smile flickering over the other girl's pale face as she seemed to drift off into a more comfortable, easy sleep. (JXX 1083)

Both (835) and (836) encode the ending of specific (serial) activities realised on a single occasion in the past. There are in all 7 of these tokens in my material, representing 1.2% of a total of 563 tokens of ‘cease to’. They are clearly exceptional, this sort of predication being normally encoded in an -ing clause, as we shall see in our discussion of ‘cease-ing’ and ‘stop -ing’ below. The ‘cease to’ construction often encodes situations that have lasted for some time. An indication of this is the fact that the two stative verbs be, with 244 occurrences, and exist with 73 occurrences, account for over half of the complement clause predicates of ‘cease to’. Duffley (1999) cites Webster’s Dictionary (1986) on the distinction between stop and cease: “Webster’s (1986: 780) brings out an interesting distinction between stop and cease: the former “applies primarily to action or progress or to what is thought of as moving or progressing”; cease, on the other hand, “applies primarily to states or conditions or to what is thought of as being or having existence” (Duffley 1999: 310). The incidence in my material of complement predicates be and exist would seem to indicate that cease in British English resembles cease in American English in this respect. Of the two matrix verbs, stop is the more common. The ‘stop -ing’ construction is over three times more common than ‘cease to’ according to the projected totals contained in Table 11 in Appendix 2. With regard to cease, there would appear to be a difference between American and British English. Freed notes that her “analysis of cease was actually hampered by limited intuitions


294

about the word and its infrequent use” (1979: 108). This problem did not arise in the present study. Contrasting cease with stop (and quit), Freed goes on to state: “cease appears to differ from stop in the same way that quit does, namely, the cessation of the complement-event is definitive. This is again difficult to substantiate, but most speakers attest to the fact that cease “means” complete termination” (1979: 121). Other scholars to have contrasted cease with stop include Wierzbicka, who states: “stop implies a sudden and unpredictable change; cease, by contrast, suggests an ongoing process which at first can be expected to continue but which at a certain point can be expected to come to an end” (1988: 81), and Dixon, who writes:

Cease and stop have “subject orientation”; the activity terminates (perhaps only temporarily) because of something to do with the person doing it […..] Cease and stop most often refer to the volition of the subject but need not always do so […..] The difference in meaning between these two verbs is that stop tends to refer to something happening suddenly (often, unexpectedly) while cease may describe a general winding down to nothing. (Dixon 1991: 175-6 & 2005: 180)

Girard (1994: 61) maintains that while stop has subject orientation, the ‘cease to’ construction differs from both ‘stop -ing’ and ‘cease -ing’ in having speaker orientation. She maintains that the situation realised by the complement clause subject may not actually stop at all in the real world, as it were, but that what comes to an end is the speaker’s perception of it.31 As she herself points out, her analysis is based on a limited number of examples and it must be admitted that the assertion that the complement situation may continue in the real world is not substantiated in the case of most of the tokens in my material. Nevertheless the data in Table 7.3 on the agentivity of the subjects in the three constructions do show that the complement situation in the case of the ‘cease to’ construction is much less likely to be terminated as the result of a willed action on the part of the subject. Thus Dixon‘s assertion regarding the tendency of cease to refer to the volition of the subject is not true with respect to ‘cease to’ in present-day British English. Dixon’s assertion that the cessation encoded by ‘cease to’ may be of a temporary nature is at odds with that of Freed noted above and will be considered next, after which we will look at the assertion that the process of cessation is often gradual (both Dixon and Wierzbicka) and to be expected (Wierzbicka). Let us start by looking at some more examples of ‘cease to’, (837) – (840), three of which contain the most common complement clause predicate, be. 31 Her exact formulation is as follows: “Dans cette optique, nous remarquons

que dans le cas d’emploi de “cease” ([cease + TO + V]) le referent de l’élément qui va devenir le sujet grammatical dui verb ne “cesse” aucune activité dans l’extralinguisiique, alors que le referent de l’élément qui va devenir le sujet de “stop” est bel et bien à l’origine de la cessation d’activité note pår l’énonciateur.” (Girard 1994: 61)


(837) If the situation is not resolved within months, the rescue plan might cease to be viable. (A14 1163)

(838) By the late 1960s Pakistan had ceased to participate in military exercises within the framework of SEATO and CENTO and it eventually withdrew from SEATO. (GVK 847)

(839) As land is passed on over the generations, it gradually ceases to be a viable unit. (AN3 582)

(840) The National Party had in March 1990 committed itself to maintaining the ban which then ceased to be a domestic political issue. (HLD 5409)

I suggested above that ‘cease to’ prototypically means something like “a certain situation had been occurring at intervals or had pertained for some time at point x; at point y (y>x) this was no longer the case”. This definition may be further refined by stating that this ‘point y’ need not necessarily refer to the point at which the actual ending took place. In (837) the speaker maintains that a time will come when the plan in question is no longer viable. The exact moment at which this change in viability occurs is not relevant. Similarly, in (838) the speaker focuses on the fact that the subject no longer takes part in military exercises. The exact point at which the transition was made is not mentioned, though a date before which it took place is. It is this very property, that of focussing on change but not the exact point of change, that allows ‘cease to’ to encode situations where the change is not sudden but gradual, a property noted by both Dixon and Wierzbicka. (839) is an example of the construction being used in this manner. This property does not, however, mean that the construction may not be used in contexts where the point of transition is explicitly stated. In (840), for instance, this point is communicated by anaphoric then. Nevertheless, such references to the point of transition are less frequent in the case of ‘cease to’ than they are with stop. More common in the case of ‘cease to’ are references of a vaguer nature to a period during which the change came about, as in utterances where the predications is modified by long (since/ago) as in (841) – (843) which may be compared to (844) and (845). (841) The fact that King Arthur had been Romano-British and engaged in a

struggle against Saxon invaders made the whole thing even more nonsensical, but Schellenberg had long since ceased to be amused by the excesses of the Third Reich. (HTW 587)

(842) It's all good-humoured teasing and winding up, and for my part I’ve long since ceased to care whether or not anyone regards me as the worst climber in the world or some sort of antediluvian relic with no rights to any opinion on ‘rock climbing as it is done these days.’ (ECG 1393)

(843) John George had long ceased to play the father to John: not only did he put his son first at all times, but his alcoholism made him extremely difficult to deal with. (B34 74)

(844) They had long since ceased sharing anything of personal value, beyond the children. (FR8 1976: NB. omitted from World edition)


296

(845) Not that she could explain it in so many words, but Katherine had long since stopped trying to win her mother's love. (FNT 121)

There are in all 37 tokens of ‘long (since/ago) cease to’ in the BNC as a whole compared to 13 tokens of ‘long (since/ago) stop -ing’ and not a single token of ‘long (since/ago) cease -ing’, the text from which (844) is taken having been omitted from the World edition. Another adverbial that occurs frequently with ‘cease to’ is never, of which there are 46 tokens in my material, making up 8% of the total number of utterances, and as many as 87% of the tokens with negated matrix verbs. (846) – (849) exemplify ‘never cease to’, (851) – (850) ‘never cease-ing’ and ‘never stop -ing’. (846) Joyce's continuing appeal never ceases to delight me. (H9Y 2148) (847) Symmetry never ceases to fascinate the human being. (ADX 1189) (848) So as one goes down the stratigraphical column, if one leaves behind the

spectacles of the specialist and looks about one with the wondering eyes of a child, one never ceases to be amazed at the diversity and yet the uniformity of it all. (H7K 151)

(849) Anselm never ceased to urge him to decide, but he issued no ultimatum, and made no threats: he insisted only on fulfilling his personal responsibilities. (CKR 28)

(850) Millie thought that they would never cease praying. (CK9 1916) (851) Although he had been a sobering and restricting influence in a Germanic

way upon the Queen, she took to wearing black immediately after his death and, in a sense, appeared never to cease mourning for him until the day she died. (CBJ 931)

(852) At one of the camps Miss Picon asked a woman at her concert why she had brought her baby, who never stopped crying. (AJ8 524)

(853) The wheel never stopped turning, the machines rattled backwards and forwards, backwards and forwards, backwards and forwards, never slowing, never slacking, never tiring; backwards and forwards, backwards and forwards, backwards and forwards, ever anxious, ever urgent, ever eager; full of anger and ambition with a rhythm so consistent, irresistibly insistent, ever beating, never cheating, beating, beating through the brain. (AEB 80)

(846) – (849) all encode general predications. It is not implied in (846) that the speaker is preoccupied with Joyce’s appeal 24 hours a day. In (847) both the subject and object of cease to fascinate are generic. (848) also contains a generic subject. It also contains, in amaze, the third most common complement clause predicate (after be and exist) collocate of ‘cease to’. There are just 3 tokens of ‘never cease -ing’ in the BNC as a whole, compared to 114 tokens of ‘never cease to’. (850) encodes a specific predication, the praying in question being profiled as occurring on a specific occasion. It would thus be unlikely to be encoded by ‘cease to’. The predication in (851), on the other hand, is not restricted to a single


occasion; quite the contrary in fact, as Victoria never appeared in public without her widow’s weeds. The fact that the matrix verb is itself infinitival in form may have influenced the choice of complement form in this case. (852) is similar to (850) in that the predication is clearly restricted to a single occasion, while the predication in (853) is serial. There are 146 tokens of ‘never stop -ing’ in the BNC as a whole, making it slightly more common than ‘never cease to’. However, ‘stop -ing’ is five times more likely to be negated by n(o)t than never, differing significantly from ‘cease to’ in this respect. We have already noted that Dixon and Wierzbicka would appear to be correct in their observations about the ability of ‘cease to’ to occur with gradually changing situations. We now proceed to consider their other assertions about the construction, regarding the temporality or permanence of the cessation and the extent to which it might have been expected. There are only two examples among the 563 downloaded tokens which explicitly state that the situation whose end is signalled by ‘cease to’ may be resumed at a later date. One of these is cited as (854). (854) The two men ceased for a time even to acknowledge one another in the

street; and though they later resumed formal courtesies, close friendship was dead. (B0R 458)

(855) When a plant or animal dies it ceases to participate in carbon exchange with the biosphere and no longer takes in $sup14; C. If $sup14; C were stable, its concentration relative to $sup12; C would remain constant after death, but, since it is not. (AC9 718)

The fact that my material contains only one other token similar to (854) would indicate that while although Dixon may be correct when he asserts that cease may be used to encode a mere interruption in a situation, it clearly differs from stop in that such ‘interruption’ readings are very rare indeed. As a rule endings signalled by ‘cease to’ are terminal. The plant in (855) is unlikely to be reborn and resume its interaction with the biosphere. The reason why the cessation is normally interpreted as terminal can be related to the semantics of general validity predications. General construction encodes situations as likely to occur on appropriate occasions in the past, present and future. The ‘cease to’ construction, being a General construction, normally excludes present and future recurrences of the situation in question. In this way they differ from Backward-looking constructions, such as ‘cease -ing’ and ‘stop -ing’, which only make reference to the past and the present (in relation to which the past is defined). They never refer to the domain of the projected future. So constructions that encode the ending of a situation by means of a Backward-looking form carry no implications about that situation’s possible future reactivation. What are we to make of Wierzbicka’s assertion that the transition signalled by ‘cease to’ is likely to be less unexpected than the change signalled by ‘stop -ing’? It is not easy to put one’s finger on what exactly Wierzbicka is getting at here, but in so far as she wishes to convey that there is something


298

‘organic’ in the nature of the change in the situation, it is hard to detect this in many of the examples in my material. Consider (855) again. It is obvious from our world knowledge that plants do not continue to exist forever: thus we may conclude that their demise is inevitable. This does not, however, mean that this inevitability is part of the semantics of the construction. Indeed, the opposite may apply in the case of (856). (856) If there is only one adult resident, a discount of 25 per cent can be claimed

(s79); it is important therefore that if husband and wife cease to live together each of them should notify the local authority that separation has taken place so as to end joint liability and for the remaining resident spouse to claim the discount (if applicable). (JXH 51)

(857) His old friend Barfield ruefully suggested that ‘at a certain stage in his life he deliberately ceased to take any interest in himself except as a kind of spiritual animus taking stock of his moral faults ... (A7C 308)

In modern Western culture, which is the culture of speakers whose output is represented in the BNC, it is presumed that marriages will last until the death of one of the parties, despite copious evidence to the contrary. And how could one possibly have expected the subject in (857), one of the minority of subjects of ‘cease to’ to exhibit volition, to desire to take such a limited interest in himself? At the very least, the evidence of these examples forces us to conclude that Wierzbicka’s assertion about the expectedness of the cessation is unproven. Let us now take a closer look at the ‘cease -ing’ construction, already exemplified by (844), (850) and (851) and here by (858) – (859). The data in Tables 7.2 and 7.3 show that ‘cease -ing’ resembles ‘stop -ing’ more than it does ‘cease to’ both with regard to TAM relations and with regard to the animacy and agentivity of its subjects. It differs from ‘stop -ing’ in that it seldom occurs in the imperative or with any (other) second- or first-person subjects. This is no doubt related to the rarity of its use in spoken contexts. (858) True, it had taken her several minutes to decide whether the keys should

be bunched or splayed but she had ceased fidgeting with them and settled into an abnormal quietude. (GWG 1201)

(859) Some two to three minutes later, when halfway between the RAF Wittering airfield boundary and the main runway, with the aircraft in a right turn at only 100 feet above the ground, the engine ceased providing power. (J1B 1041)

(860) On ceasing smoking many people gain weight, which can be 10lb (4.5kg) or even more. (FEX 1558)

‘cease -ing’ shares three features with ‘cease to’. In the first place, it does not normally focus on the actual point of cessation, although it may well do so, as in (859). In the second place, the cessation is not necessarily sudden. Again, it may well be so, as in (859), or it may not. (858), for example, does not imply a sudden


stop. Thirdly, the break is usually seen as definitive. This connotation of finality is a property of the matrix verb itself rather than the Backward-looking construction which, as was mentioned above, carried no reference to the domain of the projected future. Even the smokers in (860) may be presumed to have stopped for good. We have noted some differences in distribution between ‘cease to’ and ‘cease -ing’. What, if any, are the differences in meaning between the two? Firstly, ‘cease -ing’, like the other -ing Aspect constructions, usually implies that the subject is agentive. An exception is example (859). Secondly there is the assumption, pointed out by Freed, that the situation to be discontinued “was occurring at the time, or up until the time of the cessation of the event”: this is in contrast with ‘cease to’, where “the event in question is understood as having occurred sporadically or intermittently prior to its cessation but not necessarily at the precise moment that cease operates” (Freed 1979: 122). It must be admitted that it is not always easy to detect this principle in operation for the simple reason that neither cease construction carries an inherent reference to the actual point of cessation. Nevertheless, one may agree to the extent that the subjects in (858) – (860) were all engaged in the situation up to the point that they discontinued it. We see the distinction most clearly in the case of punctual verbs such as acknowledge in (854) and fidgeting in (858). Here is manifest the same distinction between repetition on several as opposed to a single occasion that was noted in the case of the General ‘continue to’ and the Same-time ‘continue -ing’ in section 5.5.5. Duffley’s (1999) explanation of the difference between the two constructions is couched in rather different terms, but is nonetheless compatible with the explanations advanced here. He writes:

According to the analysis of to developed here, the to-infinitive after cease has the function of a goal/result circumstantial. This implies that cease itself is conceived as involving the notion of movement, with the infinitive being represented as the term of that movement […] with the infinitive ceasing is conceived as a transition to a new state of affairs in which the infinitive’s event no longer exists […]. The -ing complement construction, on the other hand, involves the transitive sense of cease which can be paraphrased as ‘to existentially terminate (an event)'. Compared to the infinitival construction, this conceptualization shifts the focus slightly from the state of affairs ensuing upon cessation to the event whose existence is terminated by being ceased. (Duffley 1999: 325-6)

Duffley’s contrast between ‘cease to’ and ‘cease -ing’ in terms of transitivity corresponds to our distinction in terms of agentivity. His point about the focus in the -ing construction being more on the situation that has been terminated corresponds with my assertion that Backward-looking constructions carry no inherent reference to the future, whereas General constructions do so. The third construction to be considered in this section, ‘stop -ing’, is exemplified by (845), (852) and (853) and again here by (861) – (864). It was


300

noted above that both Dixon and Wierzbicka maintain that ‘stop -ing’ implies a sudden clean break in a situation rather than a gradual one. To what extent is this assertion well-founded? (861) She stopped speaking abruptly and stared at Clare with a look of fixed

terror in her eyes. (H8N 1632) (862) Certainly neither the KGB nor the GRU are going to stop spying, for that

would leave them as exposed to criticism as if the CIA suddenly stopped spying on Russia. (AN0 419)

(863) ‘Stop talking you two and look sharpish,’ said the corporal. (K8T 587) (864) I’ve stopped smoking now for four month. (H53 13) Adverbials like abruptly and suddenly in (861) and (862) occur freely with ‘stop -ing’. They do not sit easily with either of the cease constructions. The use of the imperative, which is common with ‘stop -ing’, according to Table 7.2, also singles out a point for the desired ending of the situation, the default point being the point of utterance. (863) is a case in point. If the ‘stop -ing’ construction normally refers to a point at which, rather than a period during which, the cessation took place, what are we to make of the use of the durative adverbial for four months in (864)? In fact this does not refer to the length of a period during which the process of stopping took place, but to the exact point in the past at which the cessation came into effect: the sentence may be paraphrased: ‘It is now four months since I stopped smoking’. We may conclude on the basis of these examples that Dixon and Wierzbicka are basically correct in their assertion that ‘stop -ing’ usually involves a clean break in the complement situation. In line with the discussion of ‘cease to’ above, where it was pointed out that this construction did not carry an inherent reference to the actual point of cessation, we may now state that the opposite applies in the case of ‘stop -ing’. The latter normally does make reference to such a point. Two other points were noted above regarding ‘stop -ing’, the assertion that it sometimes encodes the mere interruption rather than the complete cessation of the complement situation, and its tendency to occur overwhelmingly with agentive subjects. With respect to the latter, Wierzbicka writes:

Stop is not […] restricted to intentional acts, but due to its implication of suddenness it invites volitional interpretations much more readily than cease does:

a. He stopped breathing (probably intentionally).

b. He ceased to breathe (probably unintentionally).

(Wierzbicka 1988: 81)

One may be tempted to agree with Wierzbicka’s ascription of intentionality to the prototypical subject of the ‘stop -ing’ construction, yet still find her examples in


this quotation somewhat bizarre. There are six occurrences of ‘stop breathing’ in my material as against one of ‘cease to breathe’. In the BNC as a whole there are forty-four occurrences of the former as against three of the latter. That people should be ten times more likely to stop breathing intentionally, or that speakers should be so much more likely to profile them as so doing, seems absurd. Certainly, none of the tokens in the BNC indicate that such was the speaker’s intention. (867) is a typical example. However, the fact that we do not approve of Wierzbicka’s examples does not mean that we have to abandon her general thesis. Indeed, Table 7.3 shows that the subjects of ‘stop -ing’ constructions are overwhelmingly agentive, and the subjects of (861) – (864) all display agentive properties. Let us consider, in (867) – (868), some examples of the construction with non-agentive subjects. (865) When the ship’s gun stopped firing, I went back to the stockade. (FSJ 713) (866) Tessa stopped crying and sighed a deep, uncontrollable sigh like a yawn. (G12 1551) (867) The remaining tumour is benign, but the damage to her brain is still causing her to stop breathing between 10 and 15 times a day. (K3K 1477) (868) ‘What do you know, it had stopped raining outside. (GVL 3413) In the tables on the subjects of the various constructions I have only ascribed agentivity to animate subjects, animacy being one of the characteristics of prototypical Agents. There is no doubt, however, that some of the inanimate subjects in the ‘stop -ing’ construction evoke the idea of agentivity through metonymic relationships, as in (865). As for animate subjects, these have been classed as non-agentive unless agentivity is explicitly signalled in the context. There are some border-line cases, in particular sentences containing verbs of physical and emotional reaction such as laugh and cry. Laughing and crying are not usually conceived of as actions initiated by a willing actor: in other words they more often encode reactions than actions. On the other hand, their cessation may be willed. The context of (866) leaves open the question of whether the subject stopped crying as a result of the conscious exertion of will, and I have therefore decided to classify it as non-agentive. Whereas one may wish to stop crying, there is little doubt, contrary to the assertion of Wierzbicka mentioned above, that most people do not normally wish to stop breathing. In none of the tokens containing breathing in the BNC, one of which is cited as (867), is there any doubt that the termination of the activity is unwilled. Finally, there are a minority of contexts in which ‘stop -ing’ is used with a subject which is clearly non-agentive. One such is (868). One final question about ‘stop -ing’ remains to be addressed. Does it normally signal a complete cessation of, or a mere interruption to, the complement clause situation? The short answer is that it signals a complete cessation for the duration of the period focussed upon. In (861), for example, the subject stopped speaking on the occasion in question. There is no implication that


302

she would never speak again. The same point may be made with regard to all the other examples, (862) – (868), with the exception of (864). In (864) the very fact that the speaker employs a durative adverbial may be an indication of the possibly contingent nature of the cessation, but there is no doubt that the break is intended to be definite. At the other extreme is a sentence like (867) which explicitly refers to repeated interruptions in the breathing process. The time has come to sum up this discussion of constructions containing matrix verbs cease and stop. In the first place we have seen that neither cease construction makes inherent reference to the exact moment at which the situation in the complement clause comes to an end. However, they differ with respect to whether the situation in the complement clause is presumed to have been in progress at the (unspecified) moment of its cessation. Whereas ‘cease -ing’ normally encodes once-off or serial situations which may generally be presumed to have been in progress at the time of their ceasing, this is not true of the higher-order general validity predication encoded by ‘cease to’. The ‘stop -ing’ construction resembles the ‘cease -ing’ construction in normally encoding a once-off or serial situation, but differs from it in also making inherent reference to the exact point of cessation. These similarities and differences between the prototypical senses of the three constructions are summarised in Figure 7.1.

complement situation in progress at time of cessation

complement situation not necessarily in progress at time of cessation

matrix verb refers to point of cessation

stop -ing

matrix verb does not refer to point of cessation

cease -ing cease to

Figure 7.1: Illustrating how cease and stop constructions differ with respect to their encoding of the time of cessation of complement situations and to whether theses situations were actually in progress at the time in question

Figure 7.1 illustrates the prototypical uses of all three constructions. We have seen that all three may occasionally be extended to encode situations that may not be subsumed under these prototypical senses. Note that the empty cell in the figure is due to the logical impossibility of referring to the exact location in time of a complement situation without being aware of whether that same situation was in progress or not at this point. Another difference we noted between the ‘cease to’ construction and the two -ing constructions was the extent to which the former, unlike the latter, occurs with non-agentive subjects. Thus ‘cease to’ does not normally encode the willed ending of the complement situation. The fact that ‘stop -ing’ refers to the


point at which an agent-driven situation in progress comes to an end explains why it is normally taken to encode a clean break in this situation. This in turn is related to the possibly non-terminal nature of the break. An agent who discontinues an activity at a particular point may well decide to reactivate it at some future point: if one says that something that was being done was stopped, one does not imply that it may not be resumed. If, on the other hand, one cancels a higher-order predication that something will tend to be realised on suitable occasions in the past, present and future, one implies that it may not be revived in the future domain.

7.6 Backward-looking constructions: a summary

Backward-looking constructions encode situations as having occurred prior to the time of the matrix verb. Unlike the Same-time situations discussed in Chapter 5, which may be encoded by both bare infinitive and -ing constructions with some matrix verbs, and the future situations discussed in Chapter 6, which afford the speaker in many cases a variety of encoding options, situations in the past only allow of one form of non-finite complement, the -ing form. There are two exceptions to this, the ‘use(d) to’ and ‘cease to’ constructions, both of which prototypically encode general validity predications. There are twenty-nine Backwards-looking constructions in the present study, most of them either Mental Process or Aspect. Twenty-four of these matrix verbs do not occur with any form of non-finite complement other than -ing. The five exceptions are remember, forget, regret, report and cease. In section 7.2 we looked at remember and forget which both have two main senses. They may be used to encode a reference to some situation in the past or to a commitment to carry out a certain course of action in the future. Both verbs may be used in both senses with nominal and finite clausal objects as well as non-finite ones. When used in the first, Backward-looking sense, they take an -ing complement, when used in the second, Forward-looking sense, they take a to infinitive complement. The Attitude verb regret, discussed in section 7.3, resembles remember and forget in that it can be used with reference to something that has happened or something that is about to happen. In section 7.4 we looked at the verb report, which is normally employed to encode a communication about a past event. In this sense the speaker uses an -ing (or a finite) complement. It may also be used to communicate a judgement about something. In this, much less common, sense, the speaker may use a to infinitive complement. Finally in section 7.5 we looked at the Aspect matrix verb cease which may be used to encode the end of an ongoing situation in the past, in which case the situation in question may be encoded in an -ing clause, or the rescindment of a generic predication, in which case a speaker can employ a to infinitive clause.

Chapter 8

Summary and Conclusions

In an article entitled “Is cognitive grammar a usage-based model? Towards a realistic account of English sentential complements”, Teresa Fanego concludes that “the wide-ranging nature of the claims in cognitive linguistics creates a particular need for converging evidence from empirical work that can help substantiate those claims. A theory of language that prides itself on its attention to actual linguistic use is expected to rely on facts, rather than on one or two people’s intuitions about a few sentences” (Fanego: 2004). As was mentioned in section 1.1 recent years have seen the publication of many studies combining the methods of corpus linguistics and cognitive linguistic theory. This book also attempts to link cognitive linguistic theory to empirical evidence in the form of corpus data. As such it is intended as a contribution to the pool of knowledge that will enable cognitive linguists to answer in the affirmative the question posed by Fanego in the title of her article. Of course, the book can only serve as a contribution to the satisfaction of Fanego‘s demands to the extent that it can be shown to be both empirical and cognitive. The work upon which the book is based is undoubtedly empirical in that the starting point was a large body of corpus data, which were categorised according to similarities and differences in the distribution of the various constructions. An attempt to reproduce the various steps of this process in book form would have rendered the final product unreadable. Nevertheless, the reader has been presented in the course of the book with large amounts of real language data. He or she may well disagree with some of the author’s interpretations of these data. One of my aims, indeed, has been to provide the reader with sufficient data to do just that. Examples of language usage that might be taken to contribute to the invalidation of the proposed characterisations have been cited wherever possible - there has been no sweeping under the carpet of tokens that diverge from the norms of general usage. For my own part, I have not taken the presence of one deviant token, such as the use of ‘try -ing’ in the sense of ‘try to infinitive’ in example (377), or the use of ‘cannot bear to infinitive’ in the sense of ‘cannot bear -ing’ in example (470), both in Chapter 5, to constitute sufficient evidence to overturn the proposed characterisations of the construction types. If all grammars leak, as Sapir asserted, it would be surprising if some complement constructions didn’t occasionally leak into the neighbouring pond, as it were! This book not only presents the results of an empirical investigation into non-finite complement constructions, it is also written within a cognitive linguistics framework. It is, in the first place, usage-based, not just in the practical sense that the characterisations proposed are based on actual usage data, but also in the theoretical sense. Langacker writes: “Cognitive Grammar [...] is a usage-

306 Summary and conclusions

based theory. The grammar lists the full set of particular statements representing a speaker’s grasp of linguistic convention, including those subsumed by general statements. Rather than thinking them an embarrassment, cognitive grammarians regard particular statements as the matrix from which general statements (rules) are extracted” (Langacker 1987: 46). The general statements which I have made about Forward-looking to infinitive constructions, for example, have been extracted from statements about Attitude Forward-looking to infinitive constructions, Mental Process Forward-looking to infinitive constructions, Enablement Forward-looking to infinitive constructions, etc.. The general statements made about Mental Process Forward-looking to infinitive constructions were extracted from statements about the ‘intend to infinitive’ construction, the ‘expect to infinitive’ construction, the ‘remember to infinitive’ construction, etc. To what extent these extractions mirror the extractions likely to be made by a particular speaker of English is impossible to decide on the basis of corpus evidence. What the corpus investigator can ensure, by careful attention to the linguistic evidence, is the avoidance of generalisations on the basis of unwarranted attestations of similarity between constructions. Another respect in which this book conforms to the basic theses of cognitive linguistics is in its adherence to the content requirement. Langacker describes this requirement in the following terms:

The only structures permitted in the grammar of a language (or among the substantive specifications of universal grammar) are (1) phonological, semantic or symbolic structures that actually occur in linguistic expressions; (2) schemas for such structures; and (3) categorizing relationships involving the elements in (1) and (2). Hence no descriptive constructs are permitted that lack both phonological and semantic content. Furthermore, overt structures cannot be derived from hypothetical ”underlying” structures having a substantially different character. Although the full import of this restriction may not be immediately apparent, I believe it inherently places far more stringent limitations on the descriptive options available to the analyst than are found in any established theory. (Langacker 1987: 54)

In this book no reference has been made to “underlying” structures: all proposed structures are extracted from actual utterances and generalisations based on these, as described in the previous paragraph. Two other respects in which the approach taken in the book is typically cognitive are its adherence to the prototype model of characterisation and the dynamic evolutionary model for tense and aspect. With respect to the former, Langacker states that: “The prototype model [....] avoids the problems inherent in the criterial-attribute model. First, it does not require that every member of a category possess a given feature (or even that there be a salient property shared by all members)” (Langacker 1987: 17). In the discussion of the -ing form in section 4.4, we saw that there were in fact two salient properties shared by all members of this particular category, imperfectivity and duration. However, the

Summary and conclusions 307

category prototype, occurring in Same-time constructions with complements containing activity or accomplishment predications, shared three more features with the -ing of the progressive, and some of these other features with some of the other members of the network of -ing complement constructions. As for the dynamic evolutionary model, the domain referred to as the ‘projected future’ in this book is based on Langacker’s notions of potential and projected reality (see Langacker 1991: 278). This is the domain that affords the producer of an utterance the greatest choice with respect to construal of the situation to be encoded. The mention of construal brings me to the final important aspect in which this book may be said to be firmly rooted in the cognitive paradigm. Langacker writes: “Every linguistic expression, at its semantic pole, structures a conceived situation (or scene) by means of a particular image. In conceptualizing a scene for expressive purposes, the speaker (and secondarily the hearer, in reconstructing the speaker’s intent) is obliged to make choices with respect to [...] various parameters” (Langacker 1987: 128). A speaker’s decision with respect to the construal of a scene will influence his or her choice of construction in cases where various options are available. This is particularly true in the case of situations in the projected future, which is also the domain which affords us the most options in terms of non-finite constructions. Note that Langacker states that the speaker “is obliged to make choices”. This obligation also pertains when this choice is of no great import to the speaker in the actual context of communication. In such cases, which were encountered at intervals throughout Chapters 5 and 6, the speaker just has to opt for one of several available alternatives and rely on his or her addressee to weed out any undesired implications. I have now mentioned five respects in which the approach in this book may be said to be cognitive: it is usage-based, it satisfies the content requirement, its classifications are prototype-based, it employs the dynamic evolutionary model and it allows for various construal options on the part of the speaker. I have not carried out systematic tests with a view to determining whether taking these features into account can help us provide a better account of the system of non-finite complementation in English than alternative approaches. On the other hand, I have made use of all of them in struggling with the corpus data and in formulating explanations to account for the distribution of the various types. The strengths and weaknesses of various other approaches to non-finite complementation were illustrated in Chapter 3. To the extent that the classifications and explanations in this book account satisfactorily for the data, this may be taken as evidence of the efficacy of the cognitive approach. I turn now to the particular goals of the study, articulated in section 1.4., where it was stated that one of three goals was to establish, with a greater degree of certainty than has hitherto been the case, the relative distribution of the various constructions containing infinitive and -ing complement clauses in present-day British English. The data in appendices 1 and 2 constitute an attempt to satisfy this goal. Throughout the course of writing the book, and the dissertation on


which it is based, I have been struck by how neat this distribution actually is, how form and function seem to go hand in hand, in an area sometimes said to be characterised by no little degree of chaos and arbitrariness. The discussion in Chapters 5 - 7 centred on the more ‘difficult’ cases, those that might be expected to feature a greater degree of blurring of the boundaries between the various construction types than is otherwise the case. Most of the time form and function do, indeed, seem to go hand in hand and Table 2.6 (p. 42) is, I think, eloquent testimony to the considerable isomorphism of the system of non-finite complement clauses in English today. This should not be taken to imply that the system has reached some sort of stable, optimal form. Scholars such as Mair (2002) have provided ample evidence that the system is still evolving. No doubt it will continue to evolve in the future. The second goal of the study was to elucidate the meanings of the various constructions, paying particular attention to those constructions containing the same matrix verb. Chapters 5 to 7 contained detailed examinations of such constructions. Two types of matrix verb stood out in this respect, in the first place verbs that participate in both Same-time and General constructions, such as like and continue, in the second place verbs that participate in different sorts of Forward-looking construction, such as intend, fear and begin. The first contrast was discussed in section 5.4, the second in Chapter 6. The range of profiling options available to the speaker in the case of complement situations located in the projected future was one factor that emerged in the course of the study. Another point that emerged was the degree to which the -ing form occurs in constructions other than the Same-time ones which are often presumed to be dominant for the form. The discussion in section 4.4 shows that, while Same-time activities are prototypical, there is actually a broad network of senses for -ing complements, with the exact sense of the form varying according to the meaning of both the matrix verb and the complement predicate. One respect in which the present study differs from previous ones is in the number of basic construction types proposed to account for the various uses of infinitive and -ing complements. Four categories of -ing complement were described in section 2.3, and we saw in section 3.10 that these four classes are largely co-extensive with four classes proposed by Smith and Escobedo (2001). The main difference between their classification and mine is in the labelling of the class of constructions containing matrix verbs such as postpone, delay and avoid which I have classified as Forward-looking on the grounds that their complement situations would be expected to come about in the projected future if something had not happened to hinder their realisation, either temporarily or permanently. As for to infinitive complements, the main difference between the present study and previous ones is their division into three rather than two classes. The distinction between Forward-looking to and Judgement to is well-established in the literature. Both Wierzbicka (1988) and Dixon (1991), for instance, make use of these two categories, although they employ different labels for them. The type of to infinitive complement which I have chosen to label General is recognised by Freed (1979) in her discussion of aspectuals. In this

Summary and conclusions 309

book we have seen how the category also includes to infinitive constructions containing such Attitude matrix verbs as like, love and hate. This brings me to my third and final goal, which was to establish the meaning of the complementisers to and -ing. In section 4.3 the meanings of the to infinitive in the three construction types in which it occurs were compared both to one another and to the meanings of the form in subject position and as an adverbial. The common meaning extracted from these was that the to infinitive encoded a situation, viewed as a whole, and profiled as the targeted of several theoretically possible situations in some specified domain. At the maximally schematic level, then, to points to a targeted alternative in some domain. The targeted alternative is identical to the goal of the path-goal schema, which many scholars have seen as underlying the basic sense of the to infinitive. The term targeted alternative has been adopted in preference to goal in an attempt to render more precisely exactly what is implied when we describe a complement situation as the goal of a matrix verb. As for the -ing form, this was discussed in section 4.4, in which the four construction types in which it occurs were compared to one another, as they are used with four types of complement predicate. Finally we compared the common schematic meaning extracted from this matrix of sixteen senses to the meaning of the form when it occurs in subject position and as an adverbial. The common meaning extracted from these was that the -ing form encoded a situation profiled as extended in some specified domain. This definition is almost identical to that of Dixon (1991:218, 2005: 240), the main difference being the substitution of the more general ‘in some specified domain’ for his ‘in time’. This substitution is necessitated by the fact that the -ing form occasionally occurs in constructions, such as Contemplation constructions, which do not make reference to time as such. Neither the characterisation of the to infinitive nor of the -ing form proposed in the present study are wholly new. They are both firmly rooted in characterisations in the earlier studies reviewed in Chapter 3. They are both, however, more precise in some respects than those posited in previous studies. This is not to say that they may not be improved upon. This leads me to the question of avenues previously explored or unexplored and whether some of these might warrant (re)exploration in the light of the results of the present study. One obvious such area is the use of both infinitive and -ing complements in other clausal and phrasal contexts. I made a brief pass at the former of these in sections 4.3.2 and 4.4.2. It would be interesting to see whether the characterisation of the forms proposed in this book would be able to cast any light on their employment in other contexts. When I first embarked on this study I intended to include (I might even go so far as to say that I ‘intended including’, which choice of complement form would, in retrospect, have indicated an unwarranted element of hubris on my part!) a diachronic dimension. Considerations of both time and space rendered this impracticable. I still think that it would be very interesting to produce a version of Table 2.6 based on the distribution of various forms of non-finite complement clause in earlier stages of English and to plot developments over the


course of the centuries in terms of the categories in this table. Rohdenburg (1995: 385), Fanego (1996: 66) and Egan (in press) have all pointed out that there was in many contexts a greater variety of syntactic options available to the speaker in the form of non-finite clauses some centuries ago. There is no doubt that the system represented in Table 2.6 is neater than it was in the comparatively recent past. Tracing the steps by which the present situation evolved might well deepen our understanding both of the present state and of processes of language evolution. It would also be of interest to investigate to what extent other languages carve up possible functional domains in a similar fashion. Finally, the discussion in Chapters 5 - 7 of matrix verbs taking two or more forms of non-finite complement would undoubtedly benefit from being extended to include finite complements.

References Achard, M. (1998), Representation of cognitive structures: syntax and semantics

of French sentential complements. Berlin: Mouton de Gruyter. Andersson, E. (1985), On verb complementation in written English. Lund:

Gleerup. Bailey, D. (1992), ‘The problem of the alternation of to V/V -ing after ‘aspectual’

verbs’, in: J. Chuquet and D. Roulland (eds.), Subordination, subordinations. Rennes: Presses Universitaires de Rennes ll. 185-197.

Beukema, F. and M. Verspoor, (1991), ‘Semantic triggering of sentential complement selection: The case of to/-ing non-finite clauses’, in: J. Frankis and I. Tieken-Boon van Ostade (eds.), Language : usage and description : studies presented to N.E. Osselton on the occasion of his retirement. Amsterdam: Rodopi. 149-158.

Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan (1999), Longman grammar of spoken and written English. Harlow: Longman.

Bladon, R. A. W. (1968), ‘Selecting the “to” or “-ing” nominal after “like, love, hate, dislike” and “prefer”’, English Studies, 49: 203-214.

Bolinger, D. (1968), ‘Entailment and the meaning of structures’, Glossa, 2 (2): 119-127.

Bolinger, D. (1974), ‘Concept and percept: two infinitive constructions and their vicissitudes’, in: World Papers in Phonetics: Festschrift for Dr. Onishi’s Kizyu. Tokyo: Phonetic Society of Japan. 65-91.

Brinton, L. J. (1988), The development of English aspectual systems aspectualizers and post-verbal particles. Cambridge: Cambridge University Press.

British National Corpus (2001), Oxford: Oxford University Computing Services. Bybee, J. L. (2001), Phonology and language use. Cambridge: Cambridge

University Press. Carter, R. and M. McCarthy, (2006), Cambridge grammar of English: a

comprehensive guide: spoken and written English grammar and use. Cambridge: Cambridge University Press.

Chomsky, N. (1957), Syntactic structures. The Hague: Mouton. Chuquet, J. (1986), To et l’infinitif anglais: dètermination et opérations

énonciatives. Paris: Ophrys. Conrad, B. (1982), Referring and non-referring phrases : a study in the use of the

gerund and the infinitive. København: Akademisk Forlag. Cotte, P. (1982), ‘Autour de ‘to’’, CIEREC Travaux, xxxv: 57-80. Croft, W. (2000), Explaining language change: an evolutionary approach.

Harlow: Longman. De Smet, H. and H. Cuyckens, (2005), ‘Pragmatic strengthening and the meaning

of complement constructions’, Journal of English Linguistics, 33 (1): 3-34.

312 References

Dirven, R. (1989), ‘A cognitive perspective on complementation’, in: D. Jaspers, W. Klooster, Y. Putseys, and P. Seuren (eds.), Sentential complementation and the lexicon: studies in honour of Wim de Geest. Dordrecht: Foris. 113-139.

Dixon, R. M. W. (1991), A new approach to English grammar, on semantic principles. Oxford: Clarendon Press.

Dixon, R. M. W. (2005), A semantic approach to English grammar. 2nd ed. Oxford: Oxford University Press.

Dowty, D. R. (1991), ‘Thematic proto-roles and argument selection’, Language, 67 (3): 547-619.

Duffley, P. J. (1992), The English infinitive. London: Longman. Duffley, P. J. (1995), ‘Defining the potential meaning of the -ing form in a

psychomechanical approach’, Langues et linguistique, 21: 1-11. Duffley, P. J. (1999), ‘The use of the infinitive and the -ing after verbs denoting

the beginning, middle and end of an event’, Folia Linguistica, 33: 295-331.

Duffley, P. J. (2000), ‘Gerund versus Infinitive as Complement of Transitive Verbs in English’, Journal of English Linguistics, 28 (3): 221-248..

Duffley, P. J. (2003), ‘The Gerund and the to-Infinitive as Subject’, Journal of English Linguistics, 31 (4): 324-352.

Duffley, P. J. (2004), ‘Verbs Of Liking With The Infinitive And The Gerund’, English Studies, 85: 358-380.

Duffley, P. J. (2006), The English gerund-participle: a comparison with the infinitive. New York: Peter Lang.

Duffley, P. J. and J.-F. Joubert, (1999), ‘The gerund and the infinitive with the verbs intend, mean, propose and their close synonyms’, The Canadian Journal of Linguistics, 44 (3): 251-266.

Duffley, P. J. and R. Tremblay, (1994), ‘The infinitive and the -ing as complements of verbs of effort’, English Studies 75: 566-575.

Eagleston, R. D. (1972), ‘Selecting the to- and -ing nominal after prefer’, English Studies, 75: 141-144.

Edelman, G. M. (1989), The remembered present: a biological theory of consciousness. New York: Basic.

Egan, T. (2003), Distance and direction : a usage-based study of infinitive and -ing complement clauses in English. Dr.Art. dissertation: University of Oslo.

Egan, T. (2006), ‘Pronominal and full nominal subjects in competing constructions’, in: C. Dalton-Puffer, D. Kastovsky, N. Ritt and H. Schendl (eds.), Syntax, Style and Grammatical Norms: English from 1500-2000, Bern, Berlin etc.: Peter Lang. 91-102.

Egan, T. (in press), ‘Emotion verbs with to-infinitive complements: from specific to generic predication’, in: M. Gotti, M. Dossena and R. Dury (eds), ENGLISH HISTORICAL LINGUISTICS 2006: Selected papers from the 14th International Conference on English Historical Linguistics (14

References 313

ICEHL), Bergamo, August 2006. Tome I: Syntax and Morphology. Amsterdam: John Benjamins.

Fanego, T. (1992), Infinitive complements in Shakespeare’s English: synchronic and diachronic aspects. Santiago de Compostela: Universidade de Santiago de Compostela.

Fanego, T. (1996), ‘On the Historical Development of English Retrospective Verbs’, Neuphilologische Mitteilungen, 97: 71-79.

Fanego, T. (1997), ‘On patterns of complementation with verbs of effort’, English Studies, 78 (1): 60-67.

Fanego, T. (2004), ‘Is Cognitive Grammar a usage-based model? Towards a realistic account of English sentential complements’, Miscelánea, A Journal of English and American Studies, 29: 23-58.

Farkas, D. (1988), ‘On obligatory control’, Linguistics and Philosophy, 11: 27-58.

Ferris, C. (1993), The meaning of syntax: a study in the adjectives of English. London: Longman.

Fischer, O. (2003), ‘Principles of grammaticalization and linguistic reality’, in: G. Rohdenburg and B. Mondorf (eds.), Determinants of Grammatical Variation in English. Berlin: Mouton de Gruyter. 445-478.

Freed, A. F. (1979), The Semantics of English aspectual complementation. Dordrecht: Reidel.

Girard, G. (1994), ‘Cease + to + v /stop + v + ing et la notion de ‘‘sujet identique”’, Sigma, 16: 59-70.

Givón, T. (1984), Syntax: a functional-typological introduction. 1; Amsterdam: John Benjamins.

Givón, T. (1990), Syntax: a functional-typological introduction. 2; Amsterdam: John Benjamins.

Givón, T. (1995), Functionalism and grammar. Amsterdam: John Benjamins. Goldberg, A. E. (1995), Constructions : a construction grammar approach to

argument structure. Chicago: University of Chicago Press. Greenbaum, S., G. Nelson, and M. Weitzman, (1996), ‘Complement Clauses in

English’, in: M. Short and J. Thomas (eds.), Using corpora for language research : studies in the honour of Geoffrey Leech. London: Longman. 76-91.

Gries, S. T. (2003), Multifactorial analysis in corpus linguistics: a study of particle placement. New York: Continuum.

Hamawand, Z. (2002), Atemporal complement clauses in English: a cognitive grammar analysis. München: LINCOM Europa.

Hammett, D. (1970), The Maltese Falcon. London: Pan Books Ltd. Horie, K. (2000), ‘Complementation in Japanese and Korean: A contrastive and

cognitive linguistic approach’, in: K. Horie (ed.), Complementation: cognitive and functional perspectives. Amsterdam: John Benjamins. 11-31.

Huddleston, R. and Pullum, G. K. (2002), The Cambridge grammar of the English language. Cambridge: Cambridge University Press.

314 References

Hudson, R., A. Rosta, J. Holmes and N. Gisborne. (1996), ‘Synonyms and syntax’, Journal of Linguistics, 32 (2): 439-446.

Jackendoff, R. (1990), Semantic structures. Cambridge, Mass.: MIT Press. Jespersen, O. (1940), A modern English grammar : on historical principles.

Copenhagen: Munksgaard. Jørgensen, E. (1990), ‘Remember and forget with gerund and infinitive as

objects’, English Studies, 71: 147-151. Karttunen, L. (1971), ‘Implicative verbs’, Language, 47 (2): 340-358. Kemmer, S. and Verhagen, A. (1994), ‘The grammar of causatives and the

conceptual structure of events’, Cognitive Linguistics, 5 (4): 115-156. Kempson, R. M. and Quirk, R. (1971), ‘Controlled activation of latent contrast’,

Language, 47: 548-572. Kiparsky, P. and C. Kiparsky (1971), ‘Fact’, in: L. A. Jakobovits and D. D.

Steinberg (eds.), Semantics: an interdisciplinary reader in philosophy, linguistics and psychology. Cambridge: Cambridge University Press. 345-369.

Kjellmer, G. (1985), ‘Help to/help revisited’, English Studies, 66: 156-161. Krug, M. G. (2000), Emerging English modals: a corpus-based study of

grammaticalization. Berlin: Mouton de Gruyter. Kruisinga, E. (1931), A handbook of present-day English. 5th ed. Groningen:

Noordhoff. Lakoff, G. (1987), Women, fire, and dangerous things: what categories reveal

about the mind. Chicago: University of Chicago Press. Langacker, R. W. (1987), Foundations of Cognitive Grammar: Theoretical

prerequisites. Berlin: Mouton de Gruyter. Langacker, R. W. (1990), Concept, image, and symbol: the cognitive basis of

grammar. Berlin: Mouton de Gruyter. Langacker, R. W. (1991), Foundations of Cognitive Grammar: Descriptive

applications. Berlin: Mouton de Gruyter. Langacker, R. W. (1997), ‘Generics and Habituals’, in: R. Dirven and A.

Athanasiadou (eds.), On conditionals again. Amsterdam: John Benjamins. 191-222.

Langacker, R. W. (1999), Grammar and conceptualization. Berlin: Mouton de Gruyter.

Leech, G. N. (1971), Meaning and the English verb. London: Longman. Lind, Å. (1983), ‘The variant forms help to/help’, English Studies, 64: 263-273. Mair, C. (1990), Infinitival complement clauses in English: a study of syntax in

discourse. Cambridge: Cambridge University Press. Mair, C. (1995), ‘Changing patterns of complementation, and concomitant

grammaticalisation, of the verb help in present-day English’, in: B. Aarts and C. F. Meyer (eds.), The Verb in contemporary English : theory and description. Cambridge: Cambridge University Press. 258-272.

Mair, C. (2002), ‘The changing forms of complementation in late Modern English: a real-time study based on matching text corpora’, English Language and Linguistics, 6 (1): 105-131.

References 315

Mair, C. (2003a), ‘Gerundial complements after begin and start: Grammatical and sociolinguistic factors, and how they work against each other’, in: B. Mondorf and G. Rohdenburg (eds.), Determinants of grammatical variation in English. Berlin: Mouton de Gruyter. 329-345.

Mair, C. (2003b), ‘Thomas Egan: Distance and Direction: a usage-based study of infinitive and -ing clauses in English: doctoral defence’, Norsk Lingvistisk Tidsskrift, 21 (2): 237-246.

Matthews, P. H. (1981), Syntax. Cambridge: Cambridge University Press. Matthews, J. and Stewart, R. B. (1988), Warriors of Christendom. Poole:

Firebird Books Ltd. McGregor, W. B. (1997), Semiotic grammar. Oxford: Clarendon Press. Mittwoch, A. (1990), ‘On the distribution of bare infinitive complements in

English’, Journal of Linguistics, 26 (7): 103-131. Mukherjee, J. (2005), English ditransitive verbs: aspects of theory, description

and a usage-based model. Amsterdam: Rodopi. Noël, D. (2003), ‘Is there semantics in all syntax? The case of accusative and

infinitive constructions vs. that-clauses’, in: G. Rohdenburg and B. Mondorf (eds.), Determinants of Grammatical Variation in English. Berlin: Mouton de Gruyter. 347-377.

OED (1994), The Oxford English dictionary: on compact disc 2nd edn. Oxford: Oxford University Press.

Palmer, F. R. (1965), A linguistic study of the English verb. London: Longman. Palmer, F. R. (1988), The English verb. 2nd ed. London: Longman. Pearsall, J. and P. Hanks (1998), The New Oxford dictionary of English. Oxford:

Clarendon Press. Perlmutter, D. M. (1970), ‘The two verbs begin’, in: R. A. Jacobs and P. S.

Rosenbaum (eds.), Readings in English transformational grammar. Waltham, Mass.: Ginn. 107-119.

Popper, K. R. (1972), Objective knowledge : an evolutionary approach. Oxford: Clarendon Press.

Poutsma, H. (1904), A grammar of late modern English. Groningen: Noordhoff. Procter, P. (1995), Cambridge international dictionary of English. Cambridge:

Cambridge University Press. Quirk, R., S. Greenbaum, G. Leech & J. Svartvik (1985), A Comprehensive

grammar of the English language. London: Longman. Quirk, R. (1995), Grammatical and lexical variance in English. London:

Longman. Rohdenburg, G. (1995), ‘On the replacement of finite complement clauses by

infinitives in English’, English Studies, 76 (4): 367-388. Rohdenburg, G. (1996), ‘Cognitive complexity and increased grammatical

explicitness in English’, Cognitive Linguistics, 7: 149-182. Rohdenburg, G. (1999), ‘Clausal Complementation and Cognitive Complexity in

English’, in: F.-W. Neumann and S. Schülting (eds.), Anglistentag Erfurt 1998. Trier: Wissenschaftlicher Verlag. 101-112.

316 References

Rohdenburg, G. (2003), ‘Cognitive Complexity and horror aequi as Factors Determining the Use of Interrogative Clause Linkers in English’, in: B. Mondorf and G. Rohdenburg (eds.), Determinants of grammatical variation in English. Berlin: Mouton de Gruyter. 205-249.

Ross, J. R. (1972), ‘Doubl-ing’, Linguistic Inquiry, 3 (1): 61-86. Rudanko, J. (1984), ‘On some contrasts between infinitival and that complement

clauses in English’, English Studies, 65 (2): 141-161. Rudanko, J. (1989), Complementation and case grammar: a syntactic and

semantic study of selected patterns of complementation in present-day English. Albany: State University of New York Press.

Rudanko, J. (1996), Prepositions and complement clauses: a syntactic and semantic study of verbs governing prepositions and complement clauses in present-day English. Albany: State University of New York Press.

Rudanko, J. (1998), ‘To infinitive and to -ing complements: a look at some matrix verbs in late modern English and later’, English Studies, 79 (4): 336-348.

Rudanko, J. (2000), Corpora and complementation: tracing sentential complementation patterns of nouns, adjectives and verbs over the last three centuries. Lanham, Md.: University Press of America.

Rudanko, J. (2003), ‘Comparing alternate complements of object control verbs: evidence form the Bank of English corpus’, in: P. Leistyna and C. F. Meyer (eds.), Corpus analysis: language structure and language use. Amsterdam: Rodopi. 273-283.

Sapir, E. (1921), Language : an introduction to the study of speech. New York: Harcourt, Brace & Co.

Simon-Vandenbergen, A.-M. (2003), ‘Thomas Egan: Distance and Direction: a usage-based study of infinitive and -ing clauses in English: doctoral defence’, Norsk Lingvistisk Tidsskrift, 21 (2): 246-257.

Sinclair, J. (ed.) (1987), Collins COBUILD English language dictionary. London: Collins.

Skandera, P. (2003), ‘Start doing or start to do: Is the gerund spreading in American English?’ in: C. Tschichold (ed.), English Core Linguistics: Essays in honour of D. J. Allerton. Bern, Berlin, etc.: Peter Lang. 343-352.

Smith, M. B. and Escobedo, J. (2001), ‘The semantics of to-infinitival vs. -ing complement constructions in English’, in: M. Andronis, et al. (eds.), Proceedings from the Thirty-seventh Meeting of the Chicago Linguistic Society: 37: The Main Session. Chicago: The Society. 549-563.

Stubbs, M. (1996), Text and corpus analysis: computer-assisted studies of language and culture. Oxford: Blackwell.

Summers, D. (2003), Longman dictionary of contemporary English. New ed. Harlow: Pearson Education.

Sweet, H. (1891), A new English grammar: logical and historical. Oxford: Clarendon Press.

Talmy, L. (1986), ‘Force dynamics as a generalization over ‘causative’’, in: D. Tannen and J. E. Alatis (eds.), Languages and linguistics: the

References 317

interdependence of theory, data, and application. Washington: Georgetown University Press. 67-85.

Talmy, L. (2000), Toward a cognitive semantics. Cambridge, Mass.: MIT Press. Taylor, J. R. (1995), Linguistic categorization: prototypes in linguistic theory.

2nd ed. Oxford: Clarendon Press. Vendler, Z. (1967), Linguistics in philosophy. Ithaca, N.Y.: Cornell University

Press. Verspoor, M. (1990), Semantic Criteria in English complement selection, Ph.D.

dissertation. University of Leiden. Verspoor, M. (1997), ‘The story of -ing: a subjective perspective’, in: M. Pütz

and R. Dirven (eds.), The Construal of space in language and thought. Berlin: Mouton de Gruyter. 417-454.

Verspoor, M. (1999), ‘To infinitives’, in: L. De Stadler and C. Eyrich (eds.), Issues in cognitive linguistics: 1993 proceedings of the International Cognitive Linguistics Conference. Berlin: Mouton de Gruyter. 505-526.

Verspoor, M. (2000), ‘Iconicity in English complement constructions: conceptual distance and cognitive processing levels’, in: K. Horie (ed.), Complementation : cognitive and functional perspectives. Amsterdam: John Benjamins. 199-225.

Visser, F. T. (1973), An historical syntax of the English language. Leiden: E.J. Brill.

Wierzbicka, A. (1988), The semantics of grammar. Amsterdam: John Benjamins. Wierzbicka, A (2006), English: meaning and culture. Oxford: Oxford University

Press. Wood F.T. (1956), ‘Gerund versus infinitive’, English Language Teaching, 11:

11-16.

Appendix 1

The Matrix verbs

In this appendix all the matrix verbs in the study, including those no tokens of which surfaced in the downloaded data, are listed alphabetically with 1. the number of tokens of the verbs in the BNC 2. the various syntactic types of non-finite clause they occur with 3. an indication of the frequency of each 4. their semantic classification in the present study 5. page references, where applicable 6. an example of each type. The frequency of the various constructions is based on projected totals in the BNC as a whole and is indicated by asterisks, each of which indicates a multiple of 5. Thus * = 1 to 5 tokens ** = 6 to 25 tokens *** = 26 to 125 tokens **** = 126 to 625 tokens ***** = 626 to 3125 tokens and so on. The reason for employing asterisks rather than real or projected totals is that the latter cannot be taken at their face value as indicating with exactitude the distribution of the various constructions in present-day British English. Rather than lend a false patina of authenticity to the figures, I have chosen to convert them into a form which is designed to give a fairly reliable indication of the relative frequency of the various constructions. The totals are based on the results returned for searches for tokens tagged as verbal by the customised search engine SARA for the original version of the BNC. However, unless stated otherwise, all the quotations are from the World version. The totals returned for most verbs in the World version differ somewhat from the totals in this appendix. This is due to improvements in the tagging mechanism in the newer version.

Same-subject Different-subject

BNC to inf. bare inf. -ing to inf. bare inf. -ing

abhor 117 * *

abhor 319

abhor occurs in Same-time Attitude constructions. There was one same-subject token in the original version of the BNC, cited as 1. below. The text from which it is taken has been deleted in the World Edition. 1. Since David lived in north London and abhorred rising early, he grunted and asked for another destination. (GVC 437) 2. I abhor sentences being increased sneakily. (K5C 515)



abide 352 * * abide occurs in General and Same-time Attitude constructions. 1. ‘You know I cannot abide to light my Woodbine from the campfire.’ (HWN 1538) 2. While he wasn’t able to resist indulging his rather immature sense of humour by selling me a few dummies, my father couldn’t abide a son of his not being a credit to him in some way; my body was a forlorn hope for any improvement, so only my mind was left. (HWC 163) abominate is listed in Rudanko (1989: 45) as taking an -ing complement. There

are 7 occurrences of abominate in the BNC, none of them with a non- finite complement.

accomplish is listed in Rudanko (1989: 45) as taking an -ing complement. There

are 909 occurrences of accomplish in the BNC, none of them with a non-finite complement.



ache 603 * ache occurs in a Forward-looking Attitude construction. 1. You know how much I’ve wanted you, ached to make you mine.’ (HA6 3849)

320 Appendix 1: The matrix verbs



acknowledge 4168 ** (*)

acknowledge occurs in Backward-looking and Judgement Communication constructions. The ‘acknowledge S2 to infinitive’ construction possibly verges on the archaic (hence the brackets in the table). The only occurrence among the 1,000 downloaded tokens, cited below, actually dates from 1702. See p. 288. 1. Pentagon officials acknowledge using two Stealth fighter bombers to drop 1,000 pound bombs. (ARW 1300) 2. Thus in 1702 the Lower House of Convocation issued a declaration in which they denied that they were ‘in Opposition to Episcopacy’, asserting ‘that we acknowledge the Order of Bishops ... to be of Divine Apostolical Institution’. (HY9 482)



adjudge 69 * adjudge occurs in a Judgement Mental Process construction. 1. In many cases, the common law will control Acts of Parliament, and sometimes adjudge them to be utterly void: for when an Act of Parliament is against common right and reason, or repugnant, or impossible to be performed, the common law will control it, and adjudge such an Act to be void. (FP8 1286)


BNC to inf. -ing to -ing to inf. -ing to -ing

admit 11070 ** **** ** admit occurs in Judgement, Backward-looking and Same-time Communication constructions. See pp. 108, 110, 140-141. 1. He admitted to have been a virgin at twenty-two. (ACS 828) 2. ROCKER Bill Wyman has admitted knowing his relationship with Mandy Smith while she was still a teenager was ‘morally wrong’. (CBF 11789) 3. Lind later admitted openly to being an official of the Shipping Federation which, he declared, had an entirely undeserved reputation for ‘union smashing’. (FES 705)

adore 321



adore 528 * ** * adore occurs in General and Same-time Attitude constructions. See pp. 37, 51, 54, 107, 110, 161-162, 169-171. 1. I mean it’s lovely because he er he adores to get into does Edmund and that means that he and George go about this lovely big er grounds that they’ve (KC0 4993) 2. I adored getting drunk and I adored reading in the papers what I had done the night before. (CDG 702) 3. ‘I adored her being there in a place I equate with hell.’ (K32 1033)



advise 5023 ***** (*) advise occurs in Forward-looking Communications constructions. The construction in 2. only occurs in this one example and is possibly idiolectal. 1. ‘I advise you to answer all my questions truthfully. (G15 1818) 2. Can I advise your planning to take this particular debate. (HUD 161)



affect 12340 *** affect occurs in a Judgement Applied Attitude construction 1. They affected to despise her, but they did not find their contempt a strain, whereas the other really bad disciplinarian in the school, a Geography mistress, one Miss Riley, inflicted on them an intolerable suffering, for they felt themselves compelled to torment her, and she would sit before them, thin and pretty and anguished, making vain attempts to restore order, miserably transparent in her misery, and unable to conceal the depths of her humiliation - depths which frightened them, but which they could not leave unplumbed. (EFP 146)




afford 5321 ***** ** afford occurs in Forward-looking Applied Attitude and Enablement constructions. 1. Just a minute darling it’s alright I can afford to buy you a packet of Polos. (KBH 1213) 2. Yes, Chair, I, I mean I support the criteria that’s been laid out in five two there, I think it’s very important that we do actually, important that we do actually highlight the issue of low pay, and where it’s occurring, and in, in some way that then prevents companies, the, the unscrupulous type of companies that would come in to exploit that, from actually doing it, since most people would then be aware it, and I think that the last speaker’s just suggested that it’s an extremely cheap way of obtaining masses of information about Shropshire’s erm, earnings levels, and I think that’s very important that we are, and do come to grips with that, and clearly couldn’t afford officer time to be spent on, on merely collating that all the time. (HYX 235)



agree 23316 ****** agree occurs in a Forward-looking Communication construction. See pp. 8, 27, 50. 1. These three countries agreed to cooperate soon after the Second World War. (G24 293)



aim 7180 ****** aim occurs in a Forward-looking Effort construction. 1. I never aimed to be an independent, career-woman type, but on being jerked into it by Simon’s death I’ve had to become one. (CK0 1428)

allow 323



allow 33222 ** ******* *** allow occurs in Forward-looking Enablement constructions. See pp. 23, 40-41, 107, 200, 214-237, 243. 1. The transcription start sites for the P A2b and P A3 promoters analyzed by S1 mapping (594) indicated three contiguous possible start sites for each promoter that did not allow to determine which was the first nucleotide incorporated by the RNA polymerase, a problem frequently found with S1 mapping analysis. (FTE 1006) 2. The standard file is accessed by many applications to provide the dialogue box which allows files to be opened and saved. (HAC 9605) 3. A Fresco ‘canvas’ on to which gliffs can be ‘painted’ will allow developers create things like two-dimensional geometric figures: like X it has some difficulty with three dimensions at the moment. (CSP 167)



announce 12453 ** announce occurs in a Judgement Communication construction. See pp. 288-289. 1. After only three weeks Richter announced his experiment to be a success. (CER 513)



anticipate 2265 * *** * *** anticipate occurs in four Forward-looking Mental Process constructions. See pp. 13, 83, 110-111, 243-246, 267. 1. (He was also anticipating to gain at least six ‘O’ levels before leaving school.) (CL1 951) 2. I didn’t anticipate spending days mucking out some of the dirtiest piggeries I had ever seen. (G35 1906) 3. Project Video is provisionally scheduled for publication in September 1989 and we anticipate sales over the first three years to be in the region of 2,000 cassettes. (AP1 402)


4. Erm it’s certainly grown faster than we anticipated it growing and we are not able to er we’re not able to carry out what we like to do with Neighbourhood Watch. (GY4 217)



ape 64 * ape occurs in a Same-time Applied Attitude construction. 1. Rather than shout, Trent aped driving and raised his eyebrows in enquiry. (AMU 2117)



apply 18496 **** apply occurs in a Forward-looking Communication construction. 1. Oh aye he applied to go to the Gulf. (G5J 1852)



appoint 5854 **** appoint occurs in a Forward-looking Enablement construction. 1. Perhaps the answer might be to appoint referees from neutral countries to officiate in future competitions,’ said Kew. (CBG 1157)



arrange 6857 **** *** ** arrange occurs in a Forward-looking Applied Attitude or Enablement construction. 1. Most people who arrange to work flexible hours do so because they want to spend time with their families, as well as pursue a career. (CBW 2156)

arrange 325

2. Arrange a net to entangle game when it springs the trap. (ADY 2134) 3. Hoomey wondered who he would go along with when the day came his parents arranged his getting married, for Sikhs arranged it for their sons, Hoomey knew. (AT4 699)



ask 60618 **** ****** ask occurs in two Forward-looking Communication constructions. See pp. 8-10, 32, 279. 1. The former Dutch captain had asked to be left out of the squad in September. (CBG 8421) 2. Toby had evidently asked the doctor to be quiet. (FEE 1577)



aspire 534 **** ** aspire occurs in two Forward-looking Attitude constructions. See pp. 136, 138-140, 279. 1. Oswald Mosley aspired to be the Great Dictator, Sir Charles Chaplin played the role better. (ACH 1128) 2. They are the ones who aspire to being hooligans but lack the ‘bottle’ to succeed in such a role. (ECN 1534)



assay 143 * assay occurs in a Forward-looking Effort construction. 1. Jerome crept to the foot of the steps, and there halted, baulked, rather, like a startled horse, drew hard breath and assayed to mount, and then suddenly threw up his arms to cover his face, fell on his knees with a lamentable, choking cry, and bowed himself against the stone of the steps. (G0M 2038)




assign 1744 ** assign occurs in a Forward-looking Enablement construction. 1. Arghatun assigned a burly leader of ten to look after them. (FSE 1378)



assist 3930 **** assist occurs in a Forward-looking Enablement construction. 1. It is important to assist the patient to change position regularly while in bed and to encourage ambulation as quickly as possible to assist the circulation. (EV5 638)



assume 10932 ** **** assume occurs in Judgement Applied Attitude and Mental Process constructions. See pp. 39, 49, 53, 58, 97, 103-104. 1. (2) Provided that where by the partnership contract an option is given to surviving or continuing partners to purchase the interest of a deceased or outgoing partner, and that option is duly exercised, the estate of the deceased partner, or the outgoing partner or his estate, as the case may be, is not entitled to any further or other share of profits; but if any partner assuming to act in exercise of the option does not in all material respects comply with the terms thereof, he is liable to account under the foregoing provisions of this section. (J6P 1055) 2. ‘I have always assumed it to be apocryphal.’ (AB9 1931)



attempt 8364 ****** **

attempt 327

attempt occurs in two Forward-looking Effort constructions. See pp. 16, 79, 155-160. 1. Do not attempt to contact me. (CDA 1183) 2. And she couldn’t do anything about Castelfonte, she couldn’t even get them to attempt making it profitable - she never understood about the land, that wasn’t in her blood. (F9R 302) attribute is listed in Rudanko (1989:44) as taking an -ing complement. There are 2178 occurrences of attribute in the BNC. Of the 1000 downloaded, not one occurred with a non-finite complement.



authorise 1457 **** authorise occurs in a Forward-looking Enablement construction. 1. Although officials remain resolutely silent, there is a widespread belief that the British Government will authorise the colony’s administration to go ahead with the first batch of mandatory repatriations this week. (A9N 585)



avoid 11461 ***** *** avoid occurs in Forward-looking Applied Attitude and Causation constructions. See pp. 24, 52, 82-83, 85, 308. 1. I decided to throw myself into buying a flat, and avoid thinking about Lebanon as much as I could. (FS0 215) 2. Meanwhile, Premier John Major was walking a diplomatic tightrope in Washington yesterday to avoid Britain being drawn into a ground war with Serbia. (CBF 13395)



battle 845 * battle occurs in a Forward-looking Effort construction.


1. More than 150 officers battled to end the violence at a remand centre in Reading, Berks. (CBF 3219)



bear 6909 **** *** *** ** bear occurs in General (1), Forward-looking (2 & 5) and Same-time (3 & 6) Attitude constructions, as well as a Forward-looking Applied Attitude construction (4). See pp. 29-30, 37, 40, 161, 175-176, 241, 305. 1. ‘I can’t bear to see or hear women cry. (HHA 3383) 2. I just can’t bear to tell him. (ADG 100) 3. ‘Some people can’t bear being closed in,’ said Clarissa. (G1D 3054) 4. The tune can, of course, always bear strengthening, and this can be done either by giving it to the two trumpets in unison, and the alto and tenor parts to the tenor trombones, or by giving the alto and tenor parts to the 2nd trumpet and 1st trombone respectively and doubling the tune in the lower octave on the 2nd trombone. (GVS 1041) 5. I can’t bear Mrs Howard to go. (CA6 1662) 6. I’ve got a very stubborn streak and I discovered that I couldn’t bear people telling me what I could and couldn’t eat. HRT 767)



beg 1945 **** **** * beg occurs in three Forward-looking Communication constructions. See pp. 8, 200-202, 279. 1. He was on his knees, begging to stay. (HR8 2902) 2. I beg you to keep this in mind. (BMN 1161) 3. Master Hussey, he begs you also be in attendance.’ (HGG 58)



begin 43172 ******* ****** begin occurs in two Forward-looking Aspect constructions. See pp. 27, 38, 56, 76, 93, 108-110, 114, 185, 189, 243, 254-267, 279, 308.

begin 329

1. Locals began to fear for her safety. (ABC 79) 2. Algy began abusing her mother. (FSP 1186)



begrudge 69 * * begrudge occurs in Same-time Applied Attitude and Attitude constructions. 1. None seemed to begrudge handing over the thousands of dollars in tolls payable in cash by each ship to the Canal Commission, which brought in more than £150 million in 1988. (CJD 971) 2. Believe me, whatever happened between you, he wouldn’t berudge your being here. (JY8 4364)



believe 34516 ***** believe occurs in a Judgement Mental Process construction. See pp. 3, 22, 58. 1. Most scientists believe the infill to be lava, and though there are no features yet seen on the Moon that are definitely volcanic, floods of lava from fissures that are buried beneath their own lava are known on the Earth and could account for the mare infill on the Moon. (GW6 803)



beseech 64 * ** * beseech occurs in three Forward-looking Communication constructions. See pp. 200-202, 279. 1. In tears they beseeched to be evacuated; their one terror to be labelled untransportable’. (K91 1041) 2. Mr Cottle looked up, his eyes beseeching Mary Ann to find another object for her well-meaning pity. (ACV 1765) 3. I beseech you say not one word but ‘ yes’ or ‘no’ till I have said all I have to say. (FU4 1774)




bid 1342 *** ** *** bid occurs in Forward-looking Effort and Communication constructions. See pp. 91, 200-204, 243, 278-279. 1. SCOTLAND will bid to lift their third Triple Crown in nine years when they travel to Twickenham on Saturday week with the same line-up which did them proud in the 20-0 victory over Wales. (K5A 6403) 2. His father died on 16 July 1580, bidding Alford in his will to study till he was twenty, and then to ‘seek’ Burghley. (GTE 57) 3. He poured her a long cool gin sling, and bade her sit. (FPX 2028)



bother 3905 ***** **** bother occurs in a Forward-looking Applied Attitude construction. 1. Sandison did not even bother to check his pulse or try to revive him. (ASN 2393) 2. I didn’t bother hopping from slippery stone to slippery stone, just waded, almost ankle deep, and emerged completely dry. (CCP 1141)



bribe 254 *** bribe occurs in a Forward-looking Enablement construction. 1. Then in 1986, another prisoner bribed a guard to smuggle out a note. (CFH 390)



brook 73 * brook occurs in a Same-time Attitude construction.

brook 331

1. On Sunday afternoons, Mr. Russ took the Church Sunday School and he would not brook any boy not attending either his or some other Sunday School of another denomination. (B22 28)



burn 4341 ** burn occurs in a Forward-looking Attitude construction. 1. They were burning to know how he had got on in New York, and Isobel asked him. (CDN 1080)



care 7269 **** care occurs in a General Attitude construction. 1. ‘When I remember it first it was a playroom for the Glynn children and any others they cared to let in; a sort of clubhouse. (GWB 2479)



catch 14342 **** catch occurs in a Same-time Perception construction. See p. 24. 1. She had caught him flirting with another woman. (ASV 1710)



cause 18984 ***** ** cause occurs in two Forward-looking Causation constructions. See pp. 268-269, 275-277.


1. These activities reinforced the general fear that apathy might cause the poll to be unduly low which it was thought would be a greater handicap to SDLP than to other parties. (H7C 131) 2. The oil spilt from the Braer was unusually light and toxic, and his, combined with fierce storms which mixed it into the seawater and caused it combine to form clumps with fine partcles churned up by the waves, meant that rather than floating to the surface, as is normal with spilt crude oil, it was carried by ocean currents far from the spill site and later redeposited in deep’ sumps’ on the seabed. (J3G 633)



cease 2915 **** **** cease occurs in General and Backward-looking Aspect constructions. See pp. 38, 40, 97, 105-106, 185, 281, 291-303. 1. On Wednesday, when Ladbrokes ceased to call odds on Rovers lifting the long-awaited championship, he was pulling pints at the Blackie’s. (K4T 237) 2. Delaney’s opponent ceased struggling, looked wildly, unbelievingly, at him as they drew apart. (BPA 2957)



chafe 85 * chafe occurs in a Forward-looking Attitude construction. 1. Though chafing to fly single-engined fighters, by the outbreak of war he was still flying twin-engined Blenheim F1 night-fighters in 29 Squadron from Debden, Essex. (GT2 1302)



challenge 3628 **** challenge occurs in a Forward-looking Communication construction. 1. Mr Kinnock challenged the Prime Minister to disown Mr Baker. (AJD 345)

choose 333



choose 16478 ****** **** choose occurs in Forward-looking Applied Attitude and Enablement constructions. See p. 106, 180. 1. Birth parents have to wait for adoptees to choose to contact them and may therefore experience feelings of desperation and anticipation around the time when the adoptee would be 18 years old. (EE8 214) 2. (But in 1923, George V chose Stanley Baldwin to become prime minister over the more experienced Lord Curzon, and in 1957 the Queen passed over R.A. Butler to succeed Harold Macmillan.) (ADB 183)



claim 17031 ***** *** claim occurs in two Judgement Communication constructions. See pp. 80-81, 288-289. 1. Men also claimed to have had four times more sexual partners in their lives than women - 12.1 versus 3.2. - Reuter (K5D 9947) 2. The attitude of many older musicians and critics to science and technology is nothing more, of course, than the stale residue of the romantic, fin de siècle aesthetic that, in the phrase of Villiers de L’Isle-Adam, claims science to be ‘the religion of the suburbs’. (ADP 318)



command 1554 *** command occurs in a Forward-looking Communication construction. See p. 22. 1. If your dog tries to pick up pebbles, you must command it to drop them before any harm results. (CJE 1163)



commence 1413 ** ***


commence occurs in two Forward-looking Aspect constructions. See pp. 54, 254-257. 1. She commenced to promenade up and down the stage, punctuating her remarks with hammy gestures of her lorgnette and preposterously long cigarette holder. (FR3 610) 2. She commenced wrapping a blood pressure cuff around his arm. (BPA 1572)



compel 1070 **** compel occurs in a Forward-looking Causation construction. See p. 267. 1. Registering your name does not compel you to buy shares. (HJ4 7887)



complete 9605 *** complete occurs in a Backward-looking Aspect construction. See pp. 82, 255, 291. 1. She twirled around the room but Mary wasn’t interested and returned upstairs with her friends to complete dressing. (CDM 896)



conceive 1610 *** * conceive occurs in Judgement and Contemplation Mental Process constructions. See pp. 111, 244. 1. ‘I adopted what I conceived to be a classless accent,’ he said, ‘as that is what I thought Byron ought to be and that is what I am. (HRF 576) 2. If and again I probably overestimate the average journey time does not exceed thirty minutes, it is difficult even with Merseyrail to conceive trains being delayed in excess of an hour. (K3T 111)

condemn 335



condemn 2264 *** condemn occurs in a Forward-looking Causation construction. 1. So you’re virtually condemning some people to operate below subsistence (KM6 581)



condescend 64 **

condescend occurs in a Judgement Applied Attitude construction. See pp. 13-14, 27-28. 1. Are you going to condescend to tell anyone when you’re coming back?’ (EDN 1254)


BNC to inf. -ing to-ing to inf. -ing to-ing

confess 1609 ** ****

confess occurs in two Judgement Communication constructions. See pp. 136-137, 141. 1. ‘It is dramatic stuff, I believe, though I confess not to have read any. (ADS 995) 2. He also he confessed to participating in the bombing of army barracks in Belfast and England. (HJ3 3915)



consent 613 *** consent occurs as a Forward-looking Communication construction. 1. He consented to give the address at the funeral of Angela Carter, the novelist. (AK4 6)




consider 28282 ***** *****

consider occurs in Contemplation and Judgement Mental Process constructions. See pp. 12-13, 34, 38-39, 42, 49, 82, 94, 111, 113, 115, 117, 120, 266. 1. THE Princess of Wales seriously considered walking out on the Royal Family, according to royal biographer Andrew Morton. (K97 15472) 2. I had considered my body to be lumpy, untidy, anomalous and entirely unsuited to the person within. (CEE 908)



conspire 346 ****

conspire occurs in a Forward-looking Effort construction. 1. History and geography have conspired to bring Greece to a moment of decision. (CRB 3349)



constrain 857 *** constrain occurs in a Forward-looking Causation construction. 1. In some cases, a Zuwayi who had unwittingly offended someone would constrain that person to forgive him by offering compensation and sacrificing animals to him. (ADW 759)



contemplate 1535 **** *

contemplate occurs in two Contemplation Mental Process constructions. See pp. 27, 38, 40, 68, 82, 111, 113, 115, 120, 127. 1. For a wild moment she contemplated locking the door against him, then,

contemplate 337

pulling herself together, she prepared to face him. (JYE 3415) 2. It is pleasant to contemplate the Fabii learning Greek, while the Greeks were admiring Roman fides. (H0K 227)



continue 28711 ****** *****

continue occurs in General and Same-time Aspect constructions. See pp. 27, 35-36, 38, 40, 50, 64, 74, 93, 105-106, 108, 126, 161, 184-197, 291, 308. 1. UNHCR officials said meanwhile that hundreds of Slavic Muslims continued to arrive each day in Srebrenica, where the mood was increasingly one of panic. (K5M 12314) 2. The two men continued walking, trying to fathom the mystery of the fog- bound train. (CE9 279)



contract 1286 *** ** contract occurs in Forward-looking Communication and Enablement constructions. 1. The captains who contracted with the king to provide troops were not, however, directly responsible for recruiting all the troops they agreed to provide. (E9V 780) 2. He blamed the suppliers of the chemicals and British Steel, who contracted him to do the work, for not warning him about the safety procedure. (K54 2715)



contrive 396 **** contrive occurs in a Forward-looking Effort construction. 1. First important question: how many of the young people contrive to alert us to their cultural soundness by working ‘motherf-er’ into their interpretations? (CK5 2400)




convince 3604 *** convince occurs in a Forward-looking Communication construction. 1. After the Milan performance, Koons had a translator convince Cicciolina to do a series of soft-core photo shoots with him, sessions that evolved during the later photo shoots that Illona succumbed to Koons’ charms. (ACP 1910)



counsel 248 ** counsel occurs in a Forward-looking Communication construction. 1. Kemalpasazade sought the advice of Mueyyedzade who counselled him to feign acceptance of Hacihasanzade’s proposal; and when, on the next day, the latter petitioned the sultan, Bayezid II, to appoint Kemalpasazade to a kadilik, Mueyyedzade intervened, speaking highly of Kemalpasazade’s abilities and asking that he not be wasted in a kadilik but rather that he be given the vacant Taslik medrese so that he might busy himself with the pursuit of learning. (H75 233)



crave 301 ** * crave occurs in Forward-looking Attitude (1 & 3) and Communication (2) constructions. 1. One step, and she could be in his arms, where she craved to be. (HH8 3545) 2. The author craves to be forgiven for a very brief self-quotation in illustration of this from a work of his own: (GVS 672) 3. Girls with unsatisfactory parenting may crave a baby of their own to provide the affection lacking in their own upbringing; but then themselves repeat the problems which they are trying to get away from (Kiernan 1980b). (EDK 1202)

dare 339



dare 2703 ***** **** *** dare occurs in Forward-looking Applied Attitude and Communication constructions. See pp. 91, 200, 237-240, 243, 279. 1. But what really bugs our Sharon is that they dare to show any football at all and thus interrupt the daily stream of soap sludge. (K52 7525) 2. She was everything he had dared hope for, and more: she had perfect features and limbs, an endearing, screwed-up little face and a trace of dark hair on her pate, just to show she took after her parents. (EWH 458) 3. The youngsters dared one another to touch the hot ship. (CJA 2803)



decide 24221 ****** **

decide occurs in Forward-looking Mental Process and Causation constructions. 1. One afternoon we became so depressed that we decided to drown our sorrows in drink. (G3B 659) 2. The look he gave her decided her not to press him. (HGY 2515)



declare 5996 **** declare occurs in a Judgement Communication construction. See pp. 288-289. 1. They declared the video to be obscene, confiscated it and imposed an on-the- spot fine on my friend. (C95 1663)



decline 3049 ***** decline occurs in a Forward-looking Communication construction.


1. He declined to comment when asked whether someone was trying to smear Di by faking the letter. (CH2 29)



defer 556 ** defer occurs in a Forward-looking Applied Attitude construction. See p. 84. 1. Tired by her journey from the north of England and the heat, she had decided to defer unpacking until the next day, standing her cases temporarily to one side of the curtain that divided her cabin from the small galley. (HHA 31)



defy 790 ** *** defy occurs in Forward-looking Applied Attitude and Communication constructions. See p. 241. 1. Insect ‘cuckoos’ defy listing; they are so numerous and their habit has been reinvented so often. (ARR 1412) 2. Her breasts swing in a way I defy any artist to capture in paint. (H8M 2456)



deign 73 *** deign occurs in a Judgement Applied Attitude construction. See p. 106. 1. Then, perhaps, the finicky felines will deign to dip their tongues into the clean water you offer them. (BMG 454)



delay 2235 **** **

delay 341

delay occurs in Forward-looking Applied Attitude and Causation constructions. See pp. 49, 83-84, 109, 308. 1. Do not delay requesting veterinary attention for any young foal which is not sucking and has diarrhoea. (BPB 1027) 2. Corporate lawyers managed to delay the case coming to court for two years and then they prolonged the proceedings for a further two years. (CHL 656)



delight 836 ** delight occurs in a General Attitude construction. See p. 39. 1. It is not merely a question of its vastness (it is the second largest country in the world, measuring 9.2 million square kilometres), but its breathtaking grandeur: God’s own country, as the Canadians delight to call it. (A0P 217)



demand 5633 **** ** demand occurs in two Forward-looking Communication constructions. See p. 100. 1. The sheriff demanded to know how the solicitor had got hold of a confidential social inquiry report. (K5D 7321) 2. Ken asked to see Robert Bolt and demanded the speech to be cut. (J0W 1373)



deny 6933 **** deny occurs in Backward-looking and Same-time Communication constructions. See pp. 108, 110, 115. 1. He also denies raping a 26-year-old woman in her home earlier the same day. (K5D 13420) 2. In spite of her love of reading Eva denies being an egghead. (H7E 417)




describe 23413 *** *** describe occurs in two Backward-looking Communication constructions. See pp. 54, 112, 281, 289-290. 1. During the trial a witness described seeing bodies ‘stacked like dominoes’ amid the rubble but the cold-hearted killer from the Bogside defiantly displayed no remorse when arrested for mass murder in 1983. (HJ3 4076) 2. Under its image of the jeweller who has lost his pearl in an orchard, this describes a father lamenting his dead infant daughter in the graveyard where she is buried. (CDV 659)



deserve 3077 **** ** ** deserve occurs in Forward-looking Applied Attitude and Attitude constructions. See pp. 241-242. 1. Those journalists who recklessly write false stories deserve to be made to grovel in apology. (J78 485) 2. The second hopeful sign is that the state and its criminal justice system (broadly conceived to include criminal, administrative, and civil law) has shown itself receptive to arguments that corporate crime victims deserve protection and that corporate criminals deserve sanctioning, particularly when those arguments have been well orchestrated, empirically supported, and contain implicit electoral threats. (CHL 958) 3. The dip-shit little bastards deserve the fucking Micks turning them into mince; saw the inside of an armoured car once; been blown to buggery; thrown a hundred feet into the air and then rolled all the way down a hill; we took turns looking inside just to prove we were real men; looked like the inside of a fucking slaughter-house ... ‘ (FP6 1631)



design 11380 *** design occurs in a Forward-looking Enablement construction.

design 343

1. Knight Williams, a retirement income specialist, has designed four unit trusts to meet the needs of those in or approaching retirement. (CBU 2283)



desire 908 *** desire occurs in Forward-looking Attitude and Communication constructions. 1. In this speech, Cassius reveals part of his reasoning for desiring to assassinate Caesar, and we begin to see his true character. (KA1 1394) 2. When James dispatches Paul and others on missionary expeditions, he desires them to convert people to Jesus’s form of Judaism. (EDY 1286)



detail 791 ** detail occurs in a Forward-looking Causation construction. 1. He detailed two of his sergeants to check the other trolleys and walked over to the surviving terrorist. (HTJ 3100)



determine 10242 **** *** determine occurs in Forward-looking Mental Process and Causation constructions. 1. He determined not to give Gina any housekeeping this week to make up for his losses. (AC3 348) 2. But for the activists in the Party and its affiliates, the integratory and mobilizing functions of the ‘Hitler myth’ were not confined to support for current attainments, but rested on the incorporation in Hitler of the ‘idea’ of Nazism itself, determining future utopias to be won as well as past glories achieved. (ADD 1122)




detest 249 ** * detest occurs in two Same-time Attitude constructions. 1. I had enjoyed Hope and the work we did at night, but I detested working at night, as I had never managed to sleep well during the day. (CK0 60) 2. He is generally fed microfeed with sugar beet, and he loves carrots and sugar, but he detests his feed being mixed. (BPB 533)



die 21688 *** die occurs in a Forward-looking Attitude construction. 1. ‘I am sorry to interrupt your cosy chat, Martin, but Daddy’s dying to go up to the bar for a drink and we’ve promised to stop by Carolyn Roach’s house later on.’ (FRS 807)



direct 5248 *** direct occurs in a Forward-looking Communication construction. 1. Pennethorne’s version was that Hall had approved the scheme and directed him to draw it out in more detail, which he did, and submitted a set of drawings along with perspectives and an estimate to Hall in February 1856. (B1D 122) disclaim is listed in Rudanko (1989: 46) as taking an -ing complement. There are

89 occurrences of disclaim in the BNC, none of them with a non-finite complement.



discontinue 313 **

discontinue 345

discontinue occurs in a Backward-looking Aspect construction. See p. 291. 1. their general practitioner agreed both to discontinue prescribing Charles’ barbiturates and not to give him any further medication. (B30 518)



discourage 1064 * * discourage occurs in two Forward-looking Communication (or Enablement) constructions. 1. The fact that in three cases concomitant gastric ulcers were found should not discourage the endoscopist to look for further possible bleeding sites despite a seemingly obvious source of bleeding. (HU2 6935) 2. The Liverpool Six - Professors Sir Alan Walters, Patrick Minford, Tim Congdon, Bill Martin, Gordon Pepper and Peter Warburton - aim to discourage Britain returning to the exchange rate mechanism. (CEK 2355)



discover 10438 ** *** discover occurs in Judgement and Same-time Perception constructions. See pp. 16, 62, 154. 1. Too late Yartek discovers one key to be the fake one. (F9Y 530) 2. She was ashamed to think that Rupert might have discovered her looking it up. (HA4 1276)



discuss 14697 *** discuss occurs in a Forward-looking Communication construction. See pp. 51, 83, 108. 1. We’d discussed taking a 1-iron, but he insisted on a 3-wood and, although he got a bad bounce, put it in the sand and took a bogey. (ASA 826)




disdain 55 ** disdain occurs in a Judgement Applied Attitude construction. 1. Bach himself did not disdain to transcribe Vivaldi concertos for organ or harpsichord and to borrow fugue-subjects from Legrenzi and Corelli. (GUH 68)



dislike 1119 *** ** dislike occurs in two Same-time Attitude constructions. See pp. 73, 82, 114, 118, 125-126. 1. They disliked having so young a couple in the vicarage and they made that plain. (CMJ 630) 2. ‘I’m not getting stuffy,’ retorted Edward, ‘but I do dislike outsiders criticising something they don’t understand. (ASE 1231)



dread 462 *** ** ** dread occurs in three Forward-looking Attitude constructions. See pp. 22, 67, 80, 82-84, 107-108, 119, 243, 249-253, 267, 280. 1. I dread to think what Ben’s gonna be like on chr christmas. (KC8 335) 2. He dreads telling Martha he is going away, but he promised Babur. (HGU 1190) 3. She dreaded the inspector coming because he always asked a lot of questions, and told her how lucky she was that she would soon be able to go to a big place in London where there were lots of other girls and boys. (BP1 1842)



drive 12961 ***

drive 347

drive occurs in a Forward-looking Causation construction. 1. Yet all this was a direct response to the same basic insecurity that drove the German people to support Völkisch opinion and later to support Hitler. (BN2 561)



elect 3912 **** *** elect occurs in Forward-looking Applied Attitude and Enablement constructions. 1. Taking a deep breath we elected to go the whole hog and print 16 pages. (HAP 387) 2. But first priority for the pool fraternity is to elect a new men’s team manager to replace Trevor Harte who has stepped up to Association chairman. (HJ3 7656)



employ 7752 **** employ occurs in a Forward-looking Enablement construction. 1. Wealthy people buy the houses from the Trust, employ builders to do the restoration and move in when all is ready. (G2E 104)



empower 491 **** empower occurs in a Forward-looking Enablement construction. 1. The will empowered her to appoint her life interest (after her death) to her husband. (CBW 3458)



enable 9633 ******


enable occurs as a Forward-looking Enablement construction. See p. 24. 1. Help with child care, therefore, is important to enable women to continue in, or return to, paid work. (CRF 347)



encourage 10438 ***** encourage occurs in a Forward-looking Communication (or Enablement) construction. 1. Heads should encourage governors to visit with specific aims in view. (B23 1309)



endeavour 594 ****

endeavour occurs in a Forward-looking Effort construction. 1. The Americans endeavoured to persuade the Russians to reconvene the Joint Commission but without success. (EDP 480)



endure 1111 * ** * endure occurs in Forward-looking Attitude (1), Same-time Applied Attitude (2) and Attitude (3) constructions. See pp. 76, 161. 1. He might endure to be forsworn and dishonoured, but he would not endure to be defeated. (K8S 493) 2. ‘I had to endure watching him dismantle all the safeguards Ayling had created. (GWG 1010) 3. Once Tam darted a terrified glance over his shoulder and saw Kim sprawled across the writing-table; white faced and trembling from head to toe, he had his eyes closed and his fists were clenched tight as he summoned up every last ounce of courage in his eleven-year-old body to endure his father’s beating without weeping or crying out. (FU8 815)

engage 349



engage 3918 ** *** engage occurs in Forward-looking Communication and Enablement constructions. 1. A magnate would engage to serve with a specified number of men for a particular time in return for wages which were agreed in advance and paid by the Exchequer. (E9V 768) 2. Shortly after moving into The Kilns, Minto had engaged a gardener to help her with the eight acres of ground. (A7C 828)



enjoin 77 ** enjoin occurs in a Forward-looking Communication construction. 1. Rock stars punch the air, embrace one another, pray; and enjoin the world to use a condom. (CAL 1493)



enjoy 14437 ***** enjoy occurs in a Same-time Attitude construction. See pp. 14, 51, 61, 65, 71, 82. 1. How are you enjoying being director of this Association? (KRF 153)



entitle 5090 **** entitle occurs in a Forward-looking Enablement construction. 1. This would entitle the tenant to have a new tenancy at the then prevailing market rent. (J6R 592)




entreat 37 * * entreat occurs in two Forward-looking Communication constructions. 1. Amelia fluttered against his lips, entreating to be let back in. (ALJ 2777) 2. Strapping himself into the vacated chair, he entreated Rogal Dorn to guide him - then emptied his mind of all but titanic tech thoughts. (CJJ 2198)



envisage 1746 *** *** envisage occurs in two Forward-looking Mental Process constructions. See pp. 38, 49, 108, 244. 1. He envisaged combining farming and family life with military service in idyllic rural settlements. (EA6 500) 2. Chris Nixon said that he could not envisage the school buying college time unless the college could buy school time. (HHX 18410)



equip 1690 *** equip occurs as a Forward-looking Enablement construction. 1. His address implied that schools and teachers were no longer equipping young people to meet the requirements of industry. (CLY 438)



escape 5030 *** escape occurs in a Forward-looking Applied Attitude construction. See pp. 52, 83. 1. PUNK violinist Nigel Kennedy has escaped having a good behaviour pledge slapped on him after smashing up a hotel room. (CBE 469)

eschew 351



eschew 218 * eschew occurs in a Forward-looking Applied Attitude construction. 1. With Clinton carrying the campaign into the hitherto safe Republican areas, Perot eschewed campaigning in person in favour of buying huge amounts of television time for lengthy ‘info-commercials’ designed to inform the electorate about his analysis of the country’s economic difficulties. (HLR 507)



essay 24 * essay occurs in a Forward-looking Effort construction. 1. This she protested against very strongly, on the ground that she had known many missed and some murdered who essayed to travel round that way in the night; not that she thought such dark deeds as robbery would be perpetrated by any of the islanders, ‘But you see, sir, the island is very near the sea, and all sorts of men are sailing round about, an’ when they’ve spent all their own earnings on drink, it’s hard to know what they’ll no’ do to try and get more.’ (BIN 196)



evade 511 ** evade occurs in a Forward-looking Applied Attitude construction. 1. Fourth, until 1981, trade unions could evade having to make any payments to their members on strike since they were in receipt of supplementary benefit. (FR4 1289)



exhort 184 ***


exhort occurs in a Forward-looking Communication construction. 1. Lord Arran exhorted businessmen to be more risk-taking and more entrepreneurial. (KJ4 1809)



expect 25191 ****** ******

expect occurs in two Forward-looking Mental Process constructions. See pp. 13, 32, 38, 93, 103-104, 246, 306. 1. Italians fully expect to watch skiing from Australia at 5 o’clock in themorning and would be surprised not to follow the Japanese Formula One Grand Prix live. (A33 370) 2. Students should expect the Government to meet living costs, but not to cough up for the extra beer and nightclubbing which Richard regards as an important part of student life. (A8K 485)



face 15157 *** face occurs in Forward-looking Attitude (1) and Applied Attitude (2) constructions. See p. 249, 280. 1. What she really wanted was to go on the pill, but as her doctor was a close family friend, she couldn’t face asking him. (BP4 597) 2. The 78 women at Avdel Systems in Welwyn Garden City, Herts, said they faced losing 20 per cent of their pensions if they retired at 60. (CEN 444)


BNC

to inf. bare inf. -ing to inf. bare inf. -ing

fail 15864 ******

fail occurs in a Forward-looking Effort construction. See pp. 80, 96. 1. Neurophysiological accounts of consciousness fail to address the problem of the unities of human consciousness. (A0T 1454)

fall 353



fall 23610 *** ** fall occurs in Forward-looking Applied Attitude and Aspect constructions. See p. 136. 1. Whether the same principle should be held to apply in the case of public authorities other than the Crown charged with the enforcement of the law falls to be decided in the present case. (FC6 253) 2. ‘Oh yeah,’ said Lydia gloomily and fell to plucking at the turf between her feet. (G0X 205)



fancy 1645 **** ** *** fancy occurs in Forward-looking Attitude (1) and Judgement (2) and Contemplation (3) Mental Process constructions. See pp. 25, 71, 76, 113, 127, 244, 249, 280. 1. There is no truth in the rumour that some of their number fancied joining in for a bit of a busman’s/policeman’s holiday. (CB3 1596) 2. He thinks it’s a riding-crop handle, while my detecting partner and I fancy it to be a parasol-holder. (G2Y 1427) 3. Fancy that lot ganging up on me. (ALH 2950)



fear 4438 *** *** *** fear occurs in Forward-looking Attitude constructions. Note that the BNC contains tokens of ‘fear S2 to infinitive’ (an example is cited as 3.), although none such surfaced among the 1,000 downloaded tokens. See pp. 76, 80, 85, 107, 114, 117, 119, 243, 249-253, 267, 280, 308. 1. Elrond says ‘I fear to take the Ring to hide it. (KAY 435) 2. The French goal ace had feared being out of the game for two months after hobbling out of the 3-0 first leg defeat with a damaged hamstring. (CH3 4783)


3. But I'm pleased to say that Steg isn't the hair-tearer I initially feared it to be. (EB6 440) 4. Partnerships could henceforth be established between consenting adults so that ‘two men could live permanently together without fearing prattling informers bringing down the criminal law upon them’. (ASK 298)



feel 60570 **** ***** ***** feel occurs in a Judgement Mental Process (1) and Same-time Perception (2 & 3) constructions. See pp. 23, 25, 114, 146, 150-153. 1. Women often feel that they are not at risk because they do not feel themselves to be associated with any of the so called ‘high risk groups’ and, although we are encouraged to have safer sex, AIDS does not really seem to be an issue for us. (CJ9 650) 2. ‘You aren’t usually on this bus,’ he said and Sally felt her cheeks grow hotter. (BMW 1840) 3. His heart was banging against his chest wall so hard that she could feel it thudding into her as well. (FSC 1726)



feign 108 * * feign occurs in Judgement and Same-time Applied Attitude constructions. See pp. 154-155. 1. Corbett rubbed his eyes wearily, feigning to be more exhausted than he really was. (BMN 1775) 2. The Spurs striker is alleged to have feigned being head-butted by Coventry’s Andy Pearce. (CH3 7434)



find 76982 *** ***** find occurs in Judgement Mental Process and Same-time Perception constructions. See pp. 25, 59, 62, 153-154.

find 355

1. I found them (almost universally) to be gross without openness, and cunning without refinement. (B0R 168) 2. He found him sitting in his canvas chair beside a brazier, drinking Guinness with the chief armourer. (HRA 1326)



finish 11427 (*) ***** finish occurs in a Backward-looking Aspect construction. The projected total for the to infinitive construction is enclosed in brackets, since the example quoted is the only occurrence of this form in the whole of the BNC. It would appear to be idiolectal, or possibly dialectal. See pp. 19, 115, 127, 291. 1. And would he have to keep the churchyard? Well m mother used to clean for the church clean Yes. and er filled up the lamps and all that you know. Oh I see. And father used to dig the graves. And my three brothers went after father Mm. er finished to dig the graves. I see. Yeah. Yes. (K6U 417) 2. She finished scribbling her details on the back of the old envelope that she had extricated from her bag. (H97 219)



fit 8537 ** fit occurs in a Forward-looking Enablement construction. 1. Try to round off your answer by demonstrating how the old job has fitted you to take on the job you are applying for. (BNA 805)



forbear 50 *** * forbear occurs in two Forward-looking Applied Attitude constructions. 1. Clara forbore to point out that cremation did result, precisely, in ashes, because she took, expertly, her mother’s meaning, which was that cremation was an unnatural practice and that bodies ought to rot quietly at their own leisure. (EFP 57)


2. On the part of the plaintiff, it has been urged that the cases cited for the defendant were not cases where actions had already been brought, but only cases of promises to forbear commencing proceedings. (H81 535)



forbid 1269 **** * forbid occurs in Forward-looking Communication and Enablement constructions. See pp. 19, 25, 108. 1. He expressly forbade men to call other leaders ‘Father’. (C8L 958) 2. However the custody order expressly forbade Miss T. being brought up as a Jehovah’s Witness, the intention being that she should make her own decision when she was old enough to do so. (FDD 148)



force 11169 ***** force occurs in a Forward-looking Causation construction. See pp. 23, 93, 107, 200, 268-278. 1. The bad weather has also forced Rebecca Stephens to delay her attempt to become the first British woman to climb Everest. (K1M 1856)



foresee 780 * *** foresee occurs in two Forward-looking Mental Process constructions. See pp. 116, 119, 244. 1. The director, Jean-Christophe Amman, foresees having to close a floor at a time if the city is unwilling to subsidise its costs. (EBV 617) 2. Louis Shores foresees the student shaping his own course and his own curriculum, with the library providing the main means of achievement of his objectives as he sees them; the teacher is to be available ‘when needed’. (EW7 861)

forget 357



forget 12699 ***** *** ** forget occurs in Backward-looking and Forward-looking Mental Process constructions. See pp. 29, 52, 76, 80, 82, 96, 114, 244, 281-286, 303. 1. And don’t forget to borrow extra tables and chairs if necessary. (C9F 2500) 2. ‘I’ll never forget cheering her on. (CBC 1578) 3. I will never forget Kevin McMonagle (Old Blue Eyes) singing Tom Leonard’s brilliant version of ‘My Way’ (‘ though some may mock/the macho talk/upon the Walk/of No Surrender/I’ve drank the rent/I’ve clocked the wife/I’ve spewed my ring upon the fender’) or JohnCobb as Alasdair’s sad flowerseller, or Siobhan Redmond’s Bo’ness hippy. (B38 1411) forswear is listed in Rudanko (1989: 46) as taking an -ing complement. There are

24 occurrences of forswear in the BNC, none of them with a non-finite complement.



fret 267 * fret occurs as a Forward-looking Attitude construction. 1. Although he enjoyed ARP work my husband was really fretting to be back with the Navy. (B2E 1118) funk is listed in Rudanko (1989:45) as taking an -ing complement: there are 21

occurrences of funk in the BNC, none of them with a non-finite complement.


BNC


get 220940 ***** **** ***** *****

get occurs in Forward-looking Applied Attitude (1), Aspect (2 & 3) and Causation (4 & 5) constructions. See pp. 15, 144, 254, 268-279. 1. BOB MARLEY’S eldest daughter, Cedella, never got to do the one thing she


always desperately wanted - join her late father on tour. (CH1 6873) 2. Whistling to himself, Henry laid the table, while, in the corner of the kitchen, Maisie finished her last chocolate bar and got to work on a packet of crisps, a tube of Rollos, half a pound of jellybabies and a jumbo bar of Turkish delight. (ASS 1371) 3. Now, if you think you’ve got the answers, get dialling. (C9J 1774) 4. He refused all food and she could hardly get him to drink. (H8N 2153) 5. Questions should be open ended so as to get the candidate talking. (B0M 2054)



grow 17320 **** grow occurs in a Forward-looking Aspect construction. 1. Nellie had grown to detest the very mention of Galloway and she turned to her friends in disgust. (EA5 2042)



grudge 75 * grudge occurs in a Same-time Applied Attitude construction. 1. There are occasions when we have grudged paying a top rate, but been too cowardly to refuse. (K4V 67)



guess 3777 ** guess occurs in a Judgement Mental Process construction. See pp. 97-98. 1. He guessed her to be worrying about money. (FRH 1511)

hanker 359



hanker 99 * hanker occurs as a Forward-looking Attitude construction. 1. He could see that, even now, she hankered to go back. (GUX 1535)



hasten 482 **** hasten occurs as a Forward-looking Applied Attitude construction. 1. When he spoke, Elves hastened to obey. (CM1 438)



hate 4620 **** **** *** *** hate occurs in General, Forward-looking and Same-time Attitude constructions. See pp. 20, 27, 37, 61, 74, 81, 125, 161, 171-175, 177, 183, 288, 309. 1. Whether the Council I, I personally would be quite keen to go down there and see A, and we’d need permission of the land owners, to do this, to see where the link could go across, you know, the best position, so that we, and I believe this is what Councillor is saying, so that we can actually come forward and maybe this ought to be a meeting with the Amenities Committee, maybe the Ramblers and bear in mind as I say again I hate to do these things and the land owner think we’re steamrollering ’em into something without their knowledge. (HYG 490) 2. I hate to say this but I do rather think from the bruising that someone did it to him.’ (HA2 1202) 3. He hated showing his bald head. (CDY 78) 4. He was so soft-hearted, he hated anyone to cry. (FSF 86) 5. ‘I hate your making a joke of it,’ says Rose, pulling her hair away. (G0F 1741)




have 1213785 ***** ****** have occurs in two Forward-looking Causation constructions. See p. 15, 268. 1. Yeah I had it go over there (KBW 14919) 2. I’ll have a look, but I’ve had him doing actual words, like first words er, you know erm, Ben and large and in and out and just the, the easy words and Luke and his own name, so he is already got the idea that when you write a letter you can’t just write any letter, you’ve got to make letters say something either as erm, as an alphabet or in the form of words, so of course she says to, would you all like to write me a sentence, well he’s already passed that stage, he thinks himself well I can’t write a sentence, what he knows as a sentence consists of words that make sense, so they’re all sitting there going a N, N, N, Q, R, S, N, T, T, and they’re saying a sentence like I went to the shops with my nanny, (KDW 6509)



hazard 114 * hazard occurs as a Judgement Mental Process construction. See p. 24. 1. Liz had joked that they were sure, in the rafters, to discover a dead baby, and indeed they did find there a mummified cat, which a pathologist friend hazarded to be at least a hundred years old. (FB0 376)



hear 35975 ***** ***** hear occurs in two Same-time Perception constructions. See pp. 66, 91, 110, 114, 128, 146, 150-151. 1. She heard him gasp at the sheer beauty of her superb feminine body. (FPX 2606) 2. Pascoe heard the mechanics working hard in rooms close by. (FP7 1838)

help 361



help 37480 ****** ****** **** ****** ****** help occurs in Forward-looking Effort (1 & 2), Applied Attitude (3) and Enablement (4 & 5) constructions. See pp. 14, 16, 30, 91, 93, 104-105, 160-161, 200, 204-214, 243, 279-280. 1. When massaged into the scalp, it helps to slow down hair loss by strengthening each shaft, thus lengthening the life of each hair. (CGN 682) 2. They helped build Kuwait and they know about Kuwait and we know about them and we trust each other. (KRL 1264) 3. She could not help observing that more coins fell into the guitar case behind her than had done before she began. (EDN 948) 4. For a fee of whisky he will help mortals caught in Fairyland to escape, or show them antidotes to spells and curses. (CAC 1045) 5. ‘My top priority is to do all that I can to help those out of work find jobs. (AL6 488)



hesitate 2200 **** hesitate occurs in a Forward-looking Applied Attitude construction. 1. He was murdered in 1969 and his assassin did not hesitate to put the blame on the ‘big men’. (A6M 163) hold occurs in a Judgement Mental Process construction. The BNC contains a

handful of tokens of ‘hold S2 to’ among a total of 45,803 occurrences of the verb. However, none of these surfaced among the 1,000 downloaded tokens.



hope 20542 ******

hope occurs in a Forward-looking Attitude construction. See pp. 50, 79. 1. Morrice hoped to write the history of Puritanism. (GTE 954)




imagine 8219 **** **** ***** imagine occurs in Contemplation and Judgement Mental Process constructions. See pp. 16, 34, 54, 58, 68, 82, 109, 111, 113, 115, 117, 126, 244. 1. I feel a thrill of fear as I imagine frying his evening meal (FAS 1376) 2. He had imagined the nuns to be very particular whom they took into their house as pupils. (CK9 2032) 3. Such a tiny little woman, I thought as I watched her, imagining her building and repairing her mental walls, her endless labour. (CDX 1751)



impel 119 *** impel occurs in a Forward-looking Causation construction. See p. 267. 1. Indeed, they have a psychological mechanism in their brains which impels them to follow the first large moving thing they see after they emerge from their shells, even if it is not their mother. (F9F 409)



implore 141 * *** implore occurs in two Forward-looking Communication constructions. See p. 9. 1. Modigliani stopped her taxi as it was crossing Montparnasse and implored to be allowed to ride with her, and even that was relayed to her readers. (ANF 212) 2. His blue eyes implored her to take off the gag. (G15 3426)



incite 170 *** incite occurs as a Forward-looking Communication construction.

incite 363

1. It is an offence to organise, to participate in or to incite someone to participate in a banned march. (EVK 1475)



induce 1888 **** induce occurs in a Forward-looking Causation construction. 1. Indeed, in some species such as rabbits, the very act of mating induces the female to ovulate. (CJ3 926)



inspire 2244 **** inspire occurs in a Forward-looking Causation construction. 1. I wondered what inspired the young Phillips to pick up the guitar? (C9N 689)



instruct 1708 **** instruct occurs as a Forward-looking Communication construction. 1. It all began when God instructed Jonah to go and preach against Nineveh, a place which, despite God’s substantial record of annihilating wicked cities, was still - obstinately, unaccountably - a wicked city. (G1X 593)



intend 10623 ****** **** **** intend occurs in three Forward-looking Mental Process constructions. See pp. 52, 75, 79, 85, 90, 104, 106, 243-249, 267, 306, 308. 1. It may make sense to be tested if you are or intend to become pregant, as


about 13% of babies born to HIV positive mothers are themselves infected. (CJ9 2170) 2 There was nothing that could be used for a cradle so obviously she intended keeping it in bed with her, but Sarah had heard tales of babies being smothered like that and she was surprisingly anxious that her newly expected brother or sister wouldn’t end up that way. (BP1 336) 3. Constantinople took fright, but the tsar intended his troops to move west into Transylvania and attack the Austrian emperor’s rebellious Hungarian subjects in the rear. (HY7 359)



invite 5951 ** ***** invite occurs in a Forward-looking Applied Attitude and Communication construction. See p. 241. 1. A text well within a reader’s competence may not invite reading because it is dull, predictable, and appears to have no relevance. (BML 1074) 2. One day he wrote to Linton inviting him to visit the Grange. (GWH 1610)



involve 28953 **** involve occurs as a Same-time Enablement construction. See pp. 14, 100. 1. Many of these projects have involved teachers visiting homes rather than parents attending school meetings. (CJG 125)



judge 4029 * *** judge occurs in two Judgement Mental Process constructions. See p. 39. 1. Fertility conditions play a part then, and as hard pruning provokes the greater growth response, we should judge to ease back a little on poor, dry soils compared with more fertile conditions. (CMM 799) 2. The head, which he judged to weigh just over two pounds, was blackened on

judge 365

one side with congealed blood from which one or two coarse grey hairs sprouted like whiskers. (G3E 2333)



justify 3505 *** *** justify occurs in Same-time Applied Attitude and Attitude constructions. 1. Nor could she then justify employing a cleaner and the odd-job man who kept all the local gardens more or less under control. (G06 805) 2. CAN anyone of your readers justify any parent leaving two young children unattended in a main shopping centre on a busy Saturday afternoon? (K4J 1167)


BNC


keep 49597 ****** *****

keep occurs in Same-time Aspect and Forward-looking Causation constructions. See pp. 82, 184, 186, 191. 1. Why do we keep referring to it as ‘that summer’ - it was only last summer, after all. (EDJ 829) 2. She said she tried to keep her attacker talking and not to be aggresive. (KIN 72)



know 184765 **** *****

know occurs in Forward-looking and Judgement Mental Process construction. See pp. 58-59, 63, 97-98. 1 They all knew to move slowly. (BMX 1241) 2. The does were retained in hessian sacks and I have known some to bite a hole in the sack and escape. (BNY 1731)




labour 474 *** labour occurs in a Forward-looking Effort construction. 1. As Lambert laboured to get back on course another blast flung him lopsided. (HRA 3013)



lead 30344 ***** lead occurs in a Forward-looking Causation construction. 1. Perhaps the lack of self-confidence led the woman to marry a man she knew to be unsupportive in case she was not asked by anyone else. (J14 284)



learn 17218 ****** learn occurs in a Forward-looking Effort construction. 1. Think of the toddler learning to walk and how often he falls down only to pick himself up and try again. (AYK 723)



leave 60533 ***** ***** leave occurs in two Forward-looking Causation constructions. 1. But leaving others to bury the dead can be a dangerous exercise. (CL7 1015) 2. The type that would be perfect for someone who was sedentary could leave you feeling sweaty on a skiing holiday. (ED4 2160)

let 367



let 28678 ******* let occurs in a Forward-looking Enablement construction. See pp. 41, 91, 107, 200, 214-237, 243, 278. 1. For example, the mother of a five-year-old who always cries when bedtime comes may give in every so often and let him stay up late because she cannot bear to see tears. (B10 759)



lie 19145 **** lie occurs as a Same-time Aspect construction. See p. 184. 1. If anyone is preparing an ambush or some stealthy assassin lies lurking in a tavern, their wait will be both long and fruitless.’ (HU0 1964)


BNC


like 41732 ****** ***** ***** ****

like occurs in two General (1 & 4), one Forward-looking (2) and two Same-time (3 & 5) Attitude constructions. See pp. 4, 24, 27, 34-37, 46-47, 50-51, 53, 61, 65-66, 77, 93, 106-107, 161-171, 177, 183-184, 308-309. 1. With clothes, most of us like to believe that our tastes are idiosyncratic, which suggests that we will often go out of our way - and pay through the nose - to buy something a bit different. (A9D 727) 2. I don't like to remind you of this, but if it wasn't for me, you'd be in a bedsit somewhere now, living off the DHSS. (FAB 2141) 3. I really like doing things together. (A74 318) 4. We do like people to give us a commitment on the spot on on all of those three counts. (JA4 151) 5. I don’t like her being there, and I’m lonely on my own. (B0W 1028)




loathe 387 * ** loathe occurs in General and Same-time Attitude constructions. See pp. 37, 97, 161, 172-173. 1. ‘You loathe to be categorized. (ECU 3509) 2. ‘I loathe being called Mademoiselle.’ (F9R 2046)



long 1308 **** long occurs in a Forward-looking Attitude construction. 1. Long lashes lay in a fan on his cheeks, and she longed to smooth the ruffled dark hair away from his brow. (HGM 3291)



look 106824 ***** look occurs in a Forward-looking Attitude construction. 1. He must be looking to get away. (J1F 437)



love 13683 ***** **** *** love occurs in two General and one Same-time Attitude construction(s). See pp. 29, 35-37, 47-48, 51, 64, 71, 74, 81, 97, 116, 125, 161-162, 169-171, 177, 309. 1. My husband suffered a stroke two years ago which has left him partially sighted, but he loves to meet up with old colleagues. (K98 532) 2. ‘I love being able to open my front door onto a garden. (HJ4 4741) 3. I cannot believe that I’ve cut up the dress he used to love me to wear and all I get for thanks is cheek. (CEY 904)

make 369



make 213476 ***** *******

make occurs in Forward-looking Applied Attitude and Causation constructions. See pp. 15, 41, 91, 107, 200, 268-279. 1. I feigned lack of interest and made to go. (FEM 1588) 2. I couldn’t sleep, you made me laugh too much, (G1V 1065)



manage 13622 ****** manage occurs in a Forward-looking Effort construction. See pp. 50, 71, 96. 1. After a splendid dinner on Friday evening most people managed to find the bar. (HAB 4)



mean 65690 ***** **** mean occurs in a Forward-looking Mental Process and a Same-time Enablement construction. The BNC contains tokens of ‘mean S2 to’ (a Forward-looking Mental Process construction, an example of which is cited as 2.), although none such surfaced among the 1,000 downloaded tokens. See p. 100. 1. If you’ve been meaning to see a few West End shows, why not try a theatre package from a company like Forte Hotels (0345-543555)? (ED4 1954) 2. ‘Yes, and I am sure it is because Sir John meant him to have the house. (HGV 3335) 3. It was not a risk he was likely to take, even if it meant him staying at his desk for the next twenty-four hours. (CML 1585)



mention 10804 ** **


mention occurs in two Backward-looking Communication constructions. See p. 289. 1. He did not mention meeting Constable when the latter came up to stay at Brathay in 1806, where he stayed with the Hardens. (B3H 765) 2. You just mentioned these salt ladies coming round. (K63 219)



mime 102 ** * mime occurs in Same-time Applied Attitude and Communication constructions. See pp. 24, 108, 154-155, 289. 1. He mimed knocking in a nail with a hammer. (ACV 1536) 2. Romany mimed them going about the boat, looking here and looking there. (FEM 1961)



mind 7392 ***** **** mind occurs in two Same-time (or Forward-looking?) Attitude constructions. See pp. 40, 116. 1. I mind writing housewife on a form. (EBR 735) 2. Not that she minded Jack playing around because she knew that one day he would be hers. (BP7 426)



miss 9652 **** ** miss occurs in Forward-looking Applied Attitude (1) and Perception (2) constructions, and a Same-time Attitude (3) construction. See pp. 25, 51. 1. The terrified girl narrowly missed being hit as two gunmen opened fire on Catholic Martin Lavery, 40. (CBF 14122) 2. BLINK at the wrong moment on Christmas Eve and you could miss a new world record being set. (CBF 1740)

miss 371

3. ‘Sometimes I think I don’t miss people throwing stones at my windows. (K52 1764)



move 34288 **** move occurs in a Forward-looking Applied Attitude construction. Note that the BNC contains some tokens of the causative construction ‘move S2 to’ (an example is cited as 2.), although none such surfaced among the 1,000 downloaded tokens. 1. The Congress-run state government in Uttar Pradesh has also moved to defuse Muslim ire by formally endorsing Urdu, spoken by most Muslims, as the official second language in the state. (A4H 141) 2. Ana just drifted to a halt, turning without demur as Maggie moved her to face the house. (HGK 2807)



name 5619 **** name occurs in a Forward-looking Enablement construction. 1. Digital Equipment Corp has named Mark Brown to head its Cambridge Research Lab in Massachusetts. (CSM 426)



need 59627 ******* **** ***** **** *** need occurs in Forward-looking Applied Attitude and Enablement constructions. See pp. 200, 237-243, 279. 1. But you need to be careful in using them, because they can introduce meanings you may not intend. (HXH 1172) 2. Ministers have warning of questions and need disclose no more information than they think fit and are obliged only to answer those matters which fall within their particular areas of responsibility. (EVK 720) 3. She would then face him and, saying her own sickle needed sharpening,


neatly slice off his head. (CAC 375) 4. Sir James needed a man to run his political affairs, to manage the undercover campaigns of entertainment and bribery. (CAW 78) 5. ‘We are in synch in terms of where we’re going, and we don’t need anybody else trying to change our views.’ (CD5 1378)



neglect 1237 *** * neglect occurs in two Forward-looking Applied Attitude constructions. 1. Mr Major got off to a stickier start with President Mitterrand, who neglected to tell him about the Gulf peace plan that France was about to launch in mid- January. (ABE 1659) 2. I think it was really because Paul felt that everything was going really well with the partnership board, and he’d been erm putting a lot of work in with people like erm from Scarborough, and John and er other people who are on the board, but he’d rather been neglecting building up a really good working relationship with N Y T E C. (H5E 195)



notice 9040 *** **** notice occurs in two Same-time Perception constructions. See pp. 22, 110, 151-153. 1. Henry brushed and spat into the basin, noticing the blood darken the snow- white saliva. (ASS 360) 2. At first, I thought it was spilt wine but then it spread and I noticed little splashes coming down from the ceiling above. (H90 2505)



oblige 1956 **** oblige occurs in a Forward-looking Causation construction. 1. To his embarrassment, the lawyer at the hearings then obliged him to go

oblige 373

through a whole code-sheet to get the names into the record, or perhaps just to entertain the audience. (ADL 664)



observe 7223 *** ** *** observe occurs in Judgement and Same-time Perception constructions. See pp. 62-63, 73, 146, 151. 1. When I was told this, I recalled hearing it said, back in the 1930s, that these visits had been observed by a young writer or two who, out of rather callous curiosity, would follow Eliot at a discreet distance, and observe him to turn round, as he approached the door, in order to see whether anybody had been spying on him. (H9X 1335) 2. Eye-witnesses observed the Cherokee descend on final approach to a height of thirty feet before the engine power was heard to increase as a go-around was initiated. (CAU 1474) 3. Camille, returning from the Ladies, observed her stepfather sipping mineral water and reading the paper and made a detour round the tables to avoid him. (G1D 141)



offer 25704 ***** offer occurs in a Forward-looking Communication construction. See p. 8. 1. My husband did not have enough marks to pay the bill, so he offered to pay the equivalent amount in Sterling. (A3T 451)



omit 1282 **** omit occurs in a Forward-looking Applied Attitude construction. 1. He’d agreed to ring her and had omitted to leave a phone number. (HDC 1103)




order 6859 ***** order occurs in a Forward-looking Communication construction. See p. 279. 1. Hartlepool magistrates ordered him to carry out 240 hours community service and pay £250 compensation. (K55 6565)



overhear 414 ** *** overhear occurs in two Same-time Perception constructions. See pp. 146, 150-151. 1. ‘I overheard her say to her friend that she had lost something, but she has not reported it to me. (GVP 2215) 2. Diana had accidentally overheard her husband talking on a portable telephone while having a bath. (ECM 1670)



overlook 2020 * overlook occurs in a Backward-looking Mental Process construction. 1. She felt bad enough as it was, so it just wasn’t the moment to remember how he’d ‘overlooked’ having agreed to see her last Friday! (JYF 835)



own 5999 ** own occurs in a Judgement Communication construction. See p. 288. 1. You have owned me to be your superior before, and you will do so now.’ (G10 265)

pant 375



pant 359 * pant occurs in a Forward-looking Attitude construction. 1. Brian (twelve, knock-kneed and beginning to sprout acne) has the unshakable conviction that you’re panting to hear the entire plot of last night’s horror movie. (BMS 697) pantomime is listed in Rudanko (1989: 45) as taking an -ing complement. There

are 4 occurrences of pantomime in the BNC, none of them with a non- finite complement.



pay 34759 **** **** pay occurs in Forward-looking Applied Attitude and Enablement constructions. 1. Even though I was earning only -20 a week (before tax) at that time, I would have paid to run in that race. (BMM 577) 2. I even had to pay someone to cut the grass the other day - I wanted to do that. (CBC 6596)



perceive 2671 *** ** perceive occurs in Judgement and Same-time Perception constructions. See pp. 25, 66. 1. They now had the means to tailor curricula to suit what they perceived to be the abilities and interests of individual children. (FAM 45) 2. Third, as we monitored our representative sample of schools over a three year period, we could perceive some of them undergoing a process of structural transition, shifting away from the simple two-tier model towards departmentalism and/or a matrix structure. (G1F 872)




permit 3958 ***** permit occurs in a Forward-looking Enablement construction. See pp. 200, 214-217. 1. For a year he had neither seen her nor permitted her name to be spoken in his hearing. (K8S 2110)



persuade 4912 ***** persuade occurs in a Forward-looking Communication construction. 1. ‘You’ve come to persuade me to accept the pension, haven’t you?’ (CN3 1729)



pine 92 * pine occurs in a Forward-looking Attitude construction. 1. At that point, the Clinton the world pines to see - leader, rather than scrambler - should, with luck, appear. (CR7 155)



plan 10015 ***** *** plan occurs in two Forward-looking Mental Process constructions. See p. 96. 1. The activists also plan to lobby Congress. (J3D 5) 2. They planned each meeting to last an hour normally, and having gathered the information the team withdrew separately to ask themselves ‘Why is this a problem now?’ (CGD 1571)

plead 377



plead 1671 ** * plead occurs in two Forward-looking Communication construction. 1. On learning of her mother’s decision, Anne pleaded to go with her. (CCD 1642) 2. Before I go back to playing the game I suggest you visit you’re local software store and plead them to show you a demo. (HAC 6396)



pledge 907 **** *** pledge occurs in two Forward-looking Communication constructions. 1. Mr Patten pledged to continue fighting for greater democracy for the colony. (CEM 33) 2. The Pact did not provide any machinery for enforcement: it merely pledged the nations to outlaw war. (EWG 312)



plot 937 *** plot occurs in a Forward-looking Mental Process construction. 1. WIDOW Sandra Wignall appeared in court yesterday, accused of plotting to murder her 55-year-old husband. (K97 16582)



postpone 1093 *** *

postpone occurs in Forward-looking Applied Attitude and Causation constructions. See pp. 76, 80, 82-84, 110, 308. 1. I should be grateful if you could postpone making a decision on this licence


application until after that date. (HPK 1614) 2. So the dream becomes a symbolic expression of this conflict and what very often happens is the there’s a kind of compromise in which you go off and look for the bathroom or the drink of water or whatever it is you want, but the dream keeps postponing you finding it, in order to lengthen the dream and the state of sleep, so you go on sleeping for a bit longer. (HUL 496)



pray 2591 *** pray occurs in a Forward-looking Communication construction. 1. At last, but not until May 1941, did the much-feared message which we had prayed never to receive, arrive. (CLU 1708)



preclude 633 *** preclude occurs as a Forward-looking Causation construction. 1. This of course, does not preclude parents contacting the School whenever necessary. (K9X 59)



prefer 6696 ***** **** **** prefer occurs in General and Same-time Attitude constructions. See pp. 41, 81, 97, 106-107, 143, 161, 177-183, 185, 197. 1. Some committees prefer to concentrate on broader issues of policy and administration. (FRB 241) 2. I much prefer being assessed as I go along, rather than having big exams at the end of the year. (HX0 739) 3. Most importantly, to his way of thinking, he preferred his staff to be men. (FSR 995)

press 379



press 7068 ** **** press occurs as a Forward-looking Communication construction. 1. Reports from the Wehrmacht ‘Armaments Inspectorates’ claimed that workers on reserved occupations in armaments factories did not want to be left out of the army and were pressing to join up. (ADD 56) 2. He would press the Home Office to reconsider its decision. (CBF 4325)



presume 1105 *** ** presume occurs in Judgement Applied Attitude and Mental Process constructions. See p. 39. 1. A remarkable proportion regard the technical investigator as an unwelcome intruder who presumes to usurp the coroner’s function. (CN2 585) 2. I presumed the countless minor mishaps to be part of any climber’s learning process. (ECH 246)



pretend 2731 ***** pretend occurs in a Judgement Applied Attitude construction. See pp. 39-40, 51, 72-73, 80-81. 1. She wrote that note while pretending to look in her handbag. (HR7 3533)



prevent 11031 ***** prevent occurs in a Forward-looking Causation construction. See p. 109, 267-268. 1. It is always worth while covering both the pitot and static vents to prevent


rain entering the systems. (A0H 232)



proceed 3639 ***** proceed occurs in a Forward-looking Aspect construction. 1. But having talked his reluctant owner and trainer into running, Davies proceeded to kid and cajole their unpredictable 11-year-old into running the race of his life. (AHC 212)



proclaim 1167 * ***

proclaim occurs in two Judgement Communication constructions. See pp. 288. 1. He’s in a sense living there is in indictment of an attitude and the love and the care that the church and a society that proclaims to be civilized has for itself, and I think we’ll have to leave it there. (KRL 3277) 2. For Christianity has always proclaimed Christ to be an inclusive concept. (EF0 812)



profess 292 *** ** * profess occurs in three Judgement Communication constructions. See pp. 80-81, 136, 288. 1. He professes to be an atheist but met his wife at the Youth Fellowship at a church in Edinburgh. (HAK 203) 2. She professes to having been a little sad when it all ended, but stressed that all involved are still friends - to the point, indeed, where Trevor Hutchinson is now a lynchpin of her own band. (ED7 811) 3. Although prepared to get into the occasional lather over the destruction of ‘heritage’, many have little time for old money, and profess their heroes to be self-made men who have succeeded by their own efforts. (ADB 464)

promise 381



promise 6297 ***** promise occurs as a Forward-looking Communication construction. 1. In concluding he promised to go to prison rather than pay his fine. (AD2 129)



prompt 2043 **** prompt occurs in a Forward-looking Causation construction. 1. According to the bulletin, the army offensive had prompted as many as 100 Achinese civilians to flee by boat to Malaysia. (HL8 1341)


BNC


pronounce 1262 **

pronounce occurs as a Judgement Communication construction. See pp. 288-289. 1. He examined Sir Thomas’s corpse, pronounced him to have been poisoned, and claimed the potion was placed in a half-drunk cup of wine beside Sir Thomas’s bed. (H98 2025)



propose 7511 ***** *** ** propose occurs in three Forward-looking Communication constructions. See p. 82. 1. ‘Do I understand that you’re proposing to advertise your products with a calendar that degrades women?’ (ANY 2649) 2. You propose sending your pregnant daughter to her ? (AT7 2734)


3. After mature deliberation your Committee venture to propose to you to entrust them with powers to digest a scheme for the medical department, to receive claims, and investigate the merits of the different candidates who may propose themselves to fill the different situations in this line, as well as of the subordinate at tendants of the house, and to report to you at your next general meeting the names of such persons as to them appear most proper, for your final approbation.’ (CMG 1181)



purpose 6 * purpose occurs in a Forward-looking Mental Process construction. 1. Just as he purposed to save you, so his perfect will and, and desire is that you receive, that you are baptized in the Holy Spirit. (J8Y 275)



push 9071 *** *** push occurs in Forward-looking Effort and Causation constructions. See p. 30. 1. In our major trade negotiations I will continue pushing to eliminate tariffs and subsidies that damage America’s farmers and workers.’ (HLF 574) 2. But in a games lesson they push you to do so much (KNY 1102) question is listed in Rudanko (1989:45) as taking an -ing complement. There are

3,768 occurrences of question in the BNC. None of the 1,000 downloaded tokens occur with a non-finite complement.



quit 1124 *** quit occurs in a Backward-looking Aspect construction. See p. 291. 1. Doctors gave him six months to live if he didn’t quit drinking. (CDG 1943)

recall 383



recall 5291 **** **** recall occurs in two Backward-looking Mental Process constructions. See pp. 32, 67, 110-112, 128, 244, 281-286. 1. Shy and introverted as a schoolboy, he recalls suffering the same teenage romantic disappointments as everyone else. (CGB 1592) 2. She recalled Burton trying to tell a few ‘feeble jokes’, but his hands were shaking. (CL2 1686)



reckon 3947 *** ** reckon occurs in Forward-looking and Judgement Mental Process constructions. 1. Toyota reckon to get 50 to 100 suggestions each year from every employee and about 50% of them are accepted. (K9L 52) 2. A second study will examine lessons which the teachers reckon to embody good practice, first to chart how good, primary school history teaching proceeds, and secondly, to clarify the relationship between the planned and the delivered curriculum, with particular reference to the way teachers combine the disciplinary demands of teaching history with the need to adapt the discipline to their conception of child development. (HJ1 22572)



recollect 227 ** ** recollect occurs in two Backward-looking Mental Process constructions. See pp. 63, 110, 120, 281-286. 1. In this he recollected lying in bed as a child and counting the enormous red flowers in the wallpaper. (F9U 1618) 2. Martyn Goff was repeatedly amazed by his knowledge of books and recollects him quoting from Shakespeare and Moby Dick at length, whilst simultaneously rolling his eyes in parodic style. (F9U 558)



BNC


recommence 69 **

recommence occurs in a Forward-looking Aspect construction. See p. 76. 1. Outside, the bad dog recommenced barking rather savagely as a car drew up in the yard. (G0X 697)


BNC


recommend 5771 ***

recommend occurs in a Forward-looking Communication construction. See p. 82. 1. I recommend the reader to read through the conversations in Appendix 2 to get a general impression of them before returning to this chapter and the next to read the discussion of detailed points. (HXY 1162)



refuse 10003 ******

refuse occurs in a Forward-looking Communication construction. 1. I refuse to sell refrigerated Brie, so when the regulations come into force that state food has to be refrigerated from manufacture to sale, it will disappear from my shop. (H06 306)



regret 1632 *** **** ** regret occurs in Forward-looking and Backward-looking Attitude constructions. See pp. 71, 82, 108, 110, 117, 120, 249, 281, 286-288, 303. 1 I turned the switch and I regret to inform you that out of the loudspeaker came a lot of old Irishmen singing maudlin songs. (CAS 1006)

regret 385

2. Karen had a cheeseburger and lived to regret opting for it to be well done, as it turned up dry as a bone inside a bap as wrinkled as Sid James’ face. (K34 37) 3. A Christian can strongly believe in a triumphant entry into eternal life for the person who has died, but this does not mean that the people who remain on earth will not still deeply regret their passing. (ADE 165)



relish 529 *** * relish occurs in two Same-time Attitude constructions. 1. Pareto and Mosca positively relished emphasizing the importance of force and manipulation (imposed consent) in political rule. (CS3 55) 2. FOREST directors wouldn’t be human if they didn’t relish Brian Clough stewing in his own juice. (CH3 4730)



remain 25435 ***** **** remain occurs in Forward-looking and Same-time Aspect constructions. See pp. 184, 191. 1. The biological activity of these compounds remains to be determined. (B0X 701) 2. She remained sitting on the end of one bench, her cloak wrapped about her. (H98 2318)



remember 26656 **** ***** ***** remember occurs in Forward-looking and Backward-looking Mental Process constructions. See pp. 4, 38, 50-51, 61, 63, 67, 69, 76, 96, 108, 112, 117, 244, 281-286, 303, 306. 1. It doesn’t give you a hangover if you remember to get down half a pint of water before you go to sleep. (FRH 3355)


2. ‘I remember thinking Hawaiian music was so bad, the corniest shit in the world. (C9M 1342) 3. I remember Yvonne saying something about being away on a job for the next few days. (FP6 863)



remind 5137 *** remind occurs in a Forward-looking Communication construction. 1. ‘Do remind me to get something for Mother before then - a potted plant or something growing. (HA4 1662)



renounce 361 *

renounce occurs in a Backward-looking Communication construction. 1. He officially renounces wanting to be a Camaldolese, a brigand or a Turk. (G1A 1599)



repeat 6483 ** repeat occurs in a Same-time Aspect construction. See p. 184. 1. You should repeat supplying valid information for all mandatory fields. (HWF 5336)



report 16922 **** *** ** report occurs in Backward-looking (1 & 3) and Judgement (2) Communication constructions. See pp. 288-291, 303.

report 387

1. Diplomats reported seeing troops surrounding Lekhanya’s home and armoured vehicles in the streets of Maseru, the capital. (HL6 346) 2. They were sighted in China, where those who attempted to eat them reported them to be tasteless or unpleasant. (HA0 3283) 3. A potential failure of contraception had been recognised by 171 patients: 93 reported a split or leaking condom; 13 reported a condom coming off during intercourse; 32 admitted inconsistent use of condoms; 32 reported forgetting to take contraceptive pills or taking antibiotics concurrently with the pill; and one reported a late injection of medroxyprogesterone acetate. (FT1 729)



request 2257 ** **** request occurs in two Forward-looking Communication construction. See pp. 8, 39. 1. Many passengers specially request to sit in the row of seats immediately behind the pilot, and you can see them checking out the instruments with the guide published in the airline’s in-flight magazine. (CAU 672) 2. He then requested Lucy and Jean to come from the kitchen, and, standing between them, he placed an arm round their shoulders as he said, ‘This is Lucy and Jean, who have attended to your refreshments on behalf of Stella, my hostess, who is indisposed. (HHB 4158)



require 28386 **** **** ***** *** require occurs in Forward-looking Applied Attitude and Causation constructions. See pp. 23-25, 100, 241. 1. He just requires to spend one night in London, he’s due to catch a flight to Vienna the next day.’ (H0R 1520) 2. And not because you’ve a foolish wee fancy in your mind that I require cosseting?’ (CK0 3310) 3. It permitted, but did not, as Wilson had intended, require local authorities to provide meals for needy schoolchildren. (EE9 1060) 4. we’re obviously there’s old people’s homes here on Drive, there are lots of small erm kindergartens that will require their appliances being serviced. (JP0 1170)




resent 914 *** *** resent occurs in Same-time Applied Attitude and Attitude constructions. See pp. 116, 118. 1. Many women resent being addressed as ‘my dear’, feeling patronised by the sense of ownership and familiarity conveyed through the description. (CEF 1602) 2. Sutherland doesn’t resent people considering him mature for his age. (HSJ 41)



resist 3326 **** resist occurs in a Forward-looking Applied Attitude construction. See pp. 83, 85, 241. 1. When distressed, black people naturally resist accepting help in frightening institutional settings from people they do not trust. (FYW 1275)



resolve 3680 **** resolve occurs in a Forward-looking Mental Process construction. 1. And now that Lucy had resolved to stay for a few days, she decided to tell her aunt to expect her when she saw her. (HHB 182)



resume 1633 *** resume occurs in a Forward-looking Aspect construction. 1. She turned back to the house and resumed searching. (GUM 4721)

risk 389



risk 1825 **** *** risk occurs in Same-time (1) and Forward-looking (2) Applied Attitude and Enablement constructions. See pp. 30, 118, 241. 1. Finding no neutral ships on the eighth day, Harry risked entering a cafe, proferring a French note from his escape kit, for cigarettes. (A67 1070) 2. He also risked being arrested and put in gaol. (BPH 468) 3. Then you risk unemployment going up again and with that a general lack of confidence. (CEK 5588)



save 11216 *** save occurs in a Forward-looking Enablement construction. 1. It saved me doing temporary work which I don’t much care for.’ (HA7 581)



scheme 100 ** scheme occurs in a Forward-looking Mental Process construction. 1. Instead of going to parties, I’d be out at night looking at badgers, and I’d always be scheming to go off birdwatching or botanising. (ANP 928)



scorn 209 ** scorn occurs in a Judgement Applied Attitude construction. See p. 94. 1. Upmarket readers generally scorn to read about somebody we’ve never heard of having it off with somebody else we’ve never heard of. (G2E 2121)




scruple 9 ** scruple occurs in a Judgement Applied Attitude construction. 1. Any compiler of a catalogue raisonné will have seen and compared all the works listed, or will scruple to state if some work has proved inaccessible. (A04 985)



see 191256 ****** ****** see occurs in two Same-time Perception constructions. There are a few tokens of the Judgement ‘see S2 to infinitive’ construction in the BNC (an example is cited as 1.), but none of these surfaced among the 1,000 downloaded tokens. See pp. 5, 13, 24, 32, 55, 94, 146-149, 151-153. 1. As they approached, Reni rose from his seat at a table near the large rectangular pool which was the centrepiece of what - 14 as Huy now saw it to be - was an unconventionally asymmetrical garden. (H84 2027) 2. I was relieved to see a dozen or so people form up in front of me. (AE0 1945) 3. We could see the ship to Hamburg coming closer. (FPU 1963)



seek 15654 ****** ***

seek occurs in Forward-looking Effort and Enablement constructions. See p. 24. 1. It sought continually to encourage teachers to reevaluate classroom practice – both in curriculum content and methodology. (CLY 497) 2. The constitutionality of two-parent notification was upheld, however, provided that a provision was included allowing minors to seek a judicial review to waive the requirement. (HKW 373)

serve 391



serve 15459 ***** serve occurs in a Forward-looking Applied Attitude construction. 1. There are occasions, though, when their admirable arrogance serves to closet the band behind a mystique bordering on élitism. (ART 1781)



set 38708 **** **** **** set occurs in Forward-looking Aspect and Causation constructions. 1. Instead, she set to stubbornly work her way through in alphabetical order, so that before long she had a small pile of pink order slips in front of her. (H8S 2362) 2. Bacon and Bungay were so exhausted that they set their servant, Miles, to watch it whilst they were asleep, under the instruction that they should be woken if the head started speaking. (CBB 214) 3. Finally, she gave her head a shake that set her ear-rings dancing and her pony-tail skimming across the back of her seat. (ACW 786)



shirk 94 * shirk occurs in a Forward-looking Applied Attitude construction. 1. But the Colonel and John-William were both Justices of the Peace, often serving on the same Bench together, and although she had so far shirked inviting them to dine - the prospect of being alone with the Colonel’s lady frankly terrifying her - she was delighted to see them here today, feeling that their presence lent great distinction. (H7P 647)



shun 234 * *


shun occurs in Judgement and Same-time Applied Attitude constructions. 1. The funny thing is a lot of the locals shunned to use it because it was such an unusual building and they’d got used to their habits of using deserted claypits and rainwater. (K1V 1787) 2. Girls wishing to avoid such a fate should shun wearing or passing near fairy flowers, such as clumps of harebells or love-in-idleness. (CAC 1130)



sit 28907 **** sit occurs in a Same-time Aspect construction. See p. 184. 1. They sat gossiping in the afternoon. (CEB 1209)



smell 2160 ** smell occurs in a Same-time Perception construction. See pp. 108, 110, 146. 1. Sharks, when the current is in their favour, can smell blood issuing from a body at a distance of nearly half a kilometre. (EFR 1664)



speed 1251 * speed occurs in a Forward-looking Applied Attitude construction. 1. Miss Mates had already sped to do his bidding, giving Sally-Anne one last sympathetic squeeze of the hand before she left her. (HGE 147)



spot 2107 * *** spot occurs in two Same-time Perception constructions. See p. 146.

spot 393

1. But it was not until three mornings later that I spotted a horse and cart arrive. (CDM 1167) 2. It seems that a passer-by had spotted a naked man frolicking on the balcony – who happened to be in the room next door - and somewhere along the line a case of mistaken identity occurred. (CK6 3035)



spy 395 ** spy occurs in a Same-time Perception construction. 1. Just before leaving the path we heard a frantic rustle and spied a baby adder slithering away. (A65 784)



stand 28665 **** *** *** stand occurs in Forward-looking (1) Applied Attitude and Same-time Attitude (2 & 3) constructions. See pp. 29-30, 37, 161, 175-176. 1. The person who stood to gain most from Pendero losing the race was Jack Butler. (BP7 1614) 2. ‘I don’t think I could stand being a nurse because I would get too involved, so I hope one day to join the fire brigade,’ she said. (CF9 154) 3. He couldn’t stand your father and he can’t stand you and the other kids being your father’s children, though he doesn’t mind you being your mother’s. (FRC 3078)


BNC


start 39838 ****** ****** ***

start occurs in two Forward-looking Aspect and one Causation construction. See pp. 29, 38, 56, 67, 76, 114, 126, 185, 243, 254-267, 279-280. 1. She started to snake out from under the bed, clutching the renegade shoe. (HTL 87)


2. I started working as a caddie in 1944 when I was about eight, and I earned a penny for every year. (ASA 971) 3. Much before that the whole biblical use that starts the whole sales project going. (J3U 68)



stay 18686 ** stay occurs in a Same-time Aspect construction. See p. 184. 1. Instead they stayed talking until the small hours, telling me of the strokes they’d pulled and the mischief they’d got into as kids - right under my unsuspecting nose! (CA9 896) stick is listed in Rudanko (1989: 46) as taking an -ing complement. There are

6583 occurrences of stick in the BNC. None of the 1,000 downloaded tokens occur with a non-finite complement.



stomach 103 * stomach occurs in a Same-time Attitude construction. 1. Then he’d have to go off and do something else for a while before he could stomach being with him again. (A6E 342)



stop 24077 ****** *****

stop occurs in a Backward-looking Aspect and a Forward-looking Causation construction. See pp. 39, 108, 112, 120, 126, 267, 281, 291-303. 1. Adam stopped walking and stared at her. (HA6 1924) 2. Now the Government is taking action to stop offenders committing crimes whilst on bail. (K22 2595)

strain 395



strain 1069 **** strain occurs as a Forward-looking Effort construction. 1. The man’s pale blue eyes peered into the fog and his ears strained to catch the sound of approaching footsteps. (EA5 128)



strive 987 ***** strive occurs in a Forward-looking Effort construction. See p. 26. 1. Tuning his psychic sense, he strove to analyse this feeling until he was virtually positive of its origin. (CM4 2662)



struggle 3251 ***** struggle occurs as a Forward-looking Effort construction. 1. Trevor Skeet struggled to make himself heard against the din. (HNK 1802)



suffer 11978 *** suffer occurs in a Forward-looking Enablement construction. 1. My conscience would not suffer me to accept any more.’ (HGV 599)



suffice 657 ****


suffice occurs in a Forward-looking Applied Attitude construction. 1. Only a concrete, political challenge of this nature would suffice to explain Herod’s anxiety. (EDY 501)



suggest 28490 *** suggest occurs in a Judgement Communication construction. See pp. 82, 289. 1. Long regarded as the pendant to ‘Chien courant’ at the Norton Simon Foundation, recent studies have suggested Tajan’s picture to be later (est. (CKY 2474)



summon 1370 *** summon occurs in a Forward-looking Communication construction. 1. He summoned Chamberlain to see him again first thing the next morning and told him his new plan. (EFN 1515)



suppose 14515 *** suppose occurs in a Judgement Mental Process Construction. See pp. 42, 97. 1. For a second he supposed them to be Prussian, then recognized the shape of the cloth-covered helmets. (CMP 417)



swear 2185 *** swear occurs as a Forward-looking Communication construction.

swear 397

1. As he finished, he smacked his lips and swore to spend Eternity in Hell, if only there was a ready supply of the drink. (CAC 617)



sweat 563 ** sweat occurs in a Forward-looking Effort construction. 1. Dorcas had given this a lot of thought as teams of nomes sweated to pry the lids off the paint-streaked tins. (HTH 1011)



take 179013 ***** ***** take occurs in Forward-looking Aspect and Judgement Mental Process constructions. 1. To avoid being recognised, he took to driving everywhere in a closed cab. (G1A 377) 2. When the male wasp spies it, he wastes no time in the attempt to mate with what he clearly takes to be a female of his own kind. (BMY 461)



taste 1350 * taste occurs in a Same-time Perception construction. See pp. 39, 146. 1. I can taste blood running down the back of my throat, then my arms are grabbed and everything just spins all colours and then black. (A74 3123)



teach 8281 ***** teach occurs in a Forward-looking Enablement construction. See p. 20.


1. Tony was teaching Maureen to drive and Anne asked if he would teach her too. (G16 118)



tell 77112 ****** tell occurs in a Forward-looking Communication construction. See pp. 24, 279. 1. Imagine a situation in which a teacher is telling a pupil to write a longer answer. (F9W 844)



think 152596 ***** ****

think occurs in Forward-looking and Judgement Mental Process constructions. See pp. 24, 97. 1. She had thought to free herself, through nature, through the violence of nature. (FB0 1288) 2. As we have seen, by 1948 the British government seemed to have modified its attitude, and many thought the time to be more propitious for attempting to pull Britain more firmly within the European orbit. (CLR 77)



thirst 24 * thirst occurs in a Forward-looking Attitude construction. 1. Lewis denied that this would hardly satisfy the British boxing public, who are thirsting to see a showdown with Bruno. (CEP 6234)



threaten 6097 *****

threaten 399

threaten occurs in a Forward-looking Communication construction. 1. They threatened to kill the hostages if three imprisoned militants were not released, and on April 10, shortly before the expiry of their deadline, they killed Khera, and the following day the bodies of the remaining two hostages were found. (HKT 801)



tip 1415 ** tip occurs in Forward-looking Communication (1) and Enablement (2) constructions. See pp. 23, 25, 289. 1. If you achieved such a quantity of quality work in the theatre, they would be tipping you to lead the RSC. (A1D 21) 2. Either you tip the dustmen to take it away or you take it to the tip yourself. (A5N 76)



tolerate 1098 * * tolerate occurs in Same-time Applied Attitude and Enablement constructions. 1. It may be possible to distract a horse in some other way than in giving it food, so that it tolerates being shod. (ADF 745) 2. In no other circumstances would she have tolerated militant vigilantes operating with impunity so far beyond their own domains. (EDY 1242)



trouble 1050 *** ** trouble occurs in Forward-looking Applied Attitude and Causation constructions. 1. I cannot conceive that the pathologist will trouble to look there for a puncture mark and indeed, prior to that eventuality, it doesn’t seem likely that the emergency team of paramedics they’ll send out from Brighton General will be well enough acquainted with the action of this drug to hit upon the right


antidote in time to prevent her from expiring.’ (FR3 662) 2. Otherwise I must trouble you to hand them back at once. (H0R 1273)



trust 3528 * **** trust occurs in Forward-looking Attitude and Mental Process constructions. See p. 24. 1. And they wish your Grace joy, and comfort in the favour God has shown to your arms, and the Lord Henry trusts to see you in health and spirits at the assembly of your parliament at the month’s end.’ (HGG 2054) 2. ‘Why is it grown-ups never trust you to do anything properly?’ (FNW 492)



try 52868 ******* ***** try occurs in Forward-looking and Same-time Effort constructions. See pp. 50, 52, 79, 108, 156-161, 196, 305. 1. She frowned, trying to dislodge the piercing finger of doubt that stabbed inside her head. (HA6 1936) 2. I tried working myself into the ground, but I could be totally exhausted and still remember. (H8F 3155)



understand 22880 *** ***

understand occurs in Judgement and Same-time Mental Process constructions. 1. Nicholas had probably devoted those last moments, in part, to regretting that he had failed to do more for the peasantry, but if Alexander understood the late tsar to mean that he should introduce changes he was slow to take the injunction to heart. (HY7 835) 2. ‘I understand people associating me with bands I’ve worked on as an engineer, but I don’t understand why I am the only engineer in the world

understand 401

who’s expected to be discriminating about the bands he’s worked with...’ (CK4 694)



undertake 5460 **** **

undertake occurs in Forward-looking Communication and Applied Attitude constructions. 1. The artist must undertake to act reasonably at all times if offered fair advice by his or her manager. (A6A 2034) 2. Lord Donaldson then attacked Lord Mackay’s White Paper proposals for banks and building societies to undertake conveyancing. (A1J 61)



urge 3521 ***** urge occurs in a Forward-looking Communication construction. See p. 93. 1. Proud mum Maureen Brown urged everyone who knows Sara to vote for her. (K52 1826)



use 104883 ***** ***** use occurs in General Aspect and Forward-looking Enablement constructions. See pp. 35, 291. 1. Visitors used to flock to it in omnibuses and examine it with the careful scrutiny of sightseers.’ (AR0 358) 2. Staff use the poison to exterminate moles and rabbits. (K5D 6917)



value 1470 *


value occurs in a Same-time Attitude construction. 1. If therapy was needed for the child, they wanted to share in the offering of help and particularly valued taking part with the workers in life-story work. (CRW 558)



venture 925 **** venture occurs in a Forward-looking Applied Attitude construction. See p. 237. 1. I ventured to express exactly the opposite opinion and was stared at as if I were a hawker of ladies’ underwear who had accidentally strayed into a monastery.’ (EDA 1069)



volunteer 925 **** * volunteer occurs in two Forward-looking Communication constructions. 1. He volunteered to take the boys on a walk after breakfast that Sunday. (G0A 585) 2. Many of their rows had come from him volunteering Beth to do things. (HH3 12445)



vote 5035 **** * vote occurs in Forward-looking Communication and Enablement constructions. 1. 200 hospital clerical workers from all over Oxfordshire, who are working to rule, have voted to suspend their action. (KRM 2120) 2. Local councils have voted their ratepayers’ money to help executives defend their reputations, the BBC hazarded licence-money in backing Desmond Wilcox’s claim against ‘Private Eye’, and the Foreign Office footed the legal bill for diplomats accused of ‘covering up’ the truth about the death of nurse Helen Smith. (J78 720)

vow 403



vow 620 **** vow occurs in a Forward-looking Communication construction. 1. A TEENAGE mother yesterday vowed to stand by the pot-smoking body- builder who viciously battered her tiny baby to death. (CH2 8238)



want 90595 ******* ****** ***** want occurs in three Forward-looking Attitude constructions. See pp. 13-14, 60, 71, 79, 93, 107, 243, 249, 253-254, 267. 1. We wanted to build a pyramid that was going to be slightly bigger than the biggest pyramid in Egypt. (K1W 110) 2. I want you all to think very hard about whether you really want to go to the toilet.’ (HR8 47) 3. On Saturday morning I want them thinking, I don’t want them going into their shells. (K5A 3415)



watch 19254 ***** ***** watch occurs in two Same-time Perception constructions. See pp. 66, 146-149. 1. Together they sat and watched the plane circle. (HRA 3076) 2. I watched her sipping at the stuff, making faces. (FAP 264)



whine 254 * whine occurs as a Forward-looking Attitude construction. 1. Below, in the kitchen, William whined to go out, a gentle, quivering,


undemanding sound that meant, nevertheless, a fairly urgent need. (HH9 716)



wish 16043 ****** **** wish occurs in two Forward-looking Attitude constructions. See p. 53. 1. There are several simple exercises you can do if you wish to improve your powers of observation. (AYK 1474) 2. However, it appears that the singer wishes it to be known that he did something his way.’ (CEU 1514)



yearn 256 *** yearn occurs as a Forward-looking Attitude construction. 1. Getting older often means a reaching-back to beginnings, and you may yearn to return to the place of your origins, or to be in touch with old friends or relatives far away. (B3G 115)

Appendix 2

Tables

The pages that follow contain one table for each of the classes of matrix verb listed in Tables 2.2 and 2.3 in Chapter 2. There are also four tables (13-16) containing details of the most common same-subject and different-subject to infinitive and -ing constructions. Each table contains details of the types of non-finite complement constructions which occur with each of the relevant matrix verbs, which are listed alphabetically in the first column. The second column, labelled BNC, refers to the total number of tokens of the verbs in question in the BNC. The third column distinguishes between those verbs represented by over 1,000 tokens in the BNC, which are labelled ‘x’, and those represented by under 1,000, labelled ‘y’. Then come columns indicating the number of occurrences of the various constructions in question in the downloaded data, followed by columns listing the total numbers of tokens projected for the BNC as a whole on the basis of the figures in the preceding columns. As is the case in Appendix 1, the totals are based on the results returned by the customised search engine SARA for the original, rather than the World, version of the BNC.

Appendix 2: Tables

406

Table 1: Real and projected totals for Perception constructions

Totals per 1000(x)/ real totals (y)

Projected totals (x)/real totals (y)

Matrix verb

BNC x/y

to inf. -ing bare inf. to inf. -ing bare inf. scatch 14342 x 16 0 229 0 discover 10438 x 2 4 21 42 0 feel 60570 x 26 21 0 1575 1272 find 76982 x 2 38 154 2925 0 hear 35975 x 68 71 0 2446 2554 miss 9652 x 1 0 10 0 notice 9040 x 27 7 0 244 63 observe 7223 x 4 17 1 29 123 7 overhear 414 y 49 17 0 49 17 perceive 2671 x 1 0 3 0 see 191256 x 40 17 0 7650 3251 smell 2160 x 12 0 26 0 spot 2107 x 48 2 0 101 4 spy 395 y 7 0 7 0 taste 1350 x 1 0 1 0 watch 19254 x 59 131 0 1136 2522 Totals 204 16567 9690

Different-subject Mental Process constructions 407

Table 2: Real and projected totals for different-subject Mental Process constructions

Totals per 1000(x)/

real totals (y) Projected totals

(x)/real totals (y)

Matrix verb BNC x/y

to inf. -ing to inf. -ing adjudge 69 y 5 5 0 anticipate 2265 x 2 14 5 32 assume 10932 x 14 153 0 believe 34516 x 26 897 0 conceive 1610 x 20 3 32 5 consider 28082 x 25 702 0 contemplate 1535 x 2 0 3 envisage 1746 x 35 0 61 expect 25191 x 211 5315 0 fancy 1645 x 6 19 10 31 feel 60570 x 6 363 0 find 76982 x 1 77 0 foresee 780 y 41 0 41 forget 12669 x 1 0 13 guess 3777 x 6 23 0 hazard 114 y 1 1 0 imagine 8219 x 20 129 164 1060 intend 10623 x 34 361 0 judge 4029 x 27 109 0 know 184765 x 5 924 0 perceive 2671 x 39 1 104 3 plan 10015 x 3 30 0 presume 1105 x 17 19 0 recall 5291 x 24 0 127 reckon 3947 x 6 24 0 recollect 227 y 10 0 10 remember 26656 x 30 0 800 suppose 14515 x 8 116 0 take 179013 x 2 358 0 think 152596 x 1 153 0 understand 22880 x 2 3 46 69 Totals 9991 2255

Appendix 2: Tables

408

Table 3: Real and projected totals for different-subject Attitude constructions



Matrix verb

BNC x/y

to inf. -ing to inf. -ing abhor 117 y 1 0 1 abide 352 y 1 0 1 adore 528 y 1 0 1 bear 8251 x 2 1 17 8 begrudge 69 y 1 0 1 brook 73 y 1 0 1 crave 301 y 1 1 0 deserve 3077 x 1 0 3 desire 908 y 18 18 0 detest 249 y 1 0 1 dislike 1119 x 7 0 8 dread 462 y 8 0 8 endure 1111 x 2 0 2 fancy 1645 x 2 0 3 fear 4438 x 8 0 36 hate 4620 x 8 9 37 42 justify 3505 x 14 0 49 like 41732 x 29 8 1210 334 love 13683 x 3 41 0 mind 7392 x 38 0 281 miss 9652 x 1 0 10 prefer 6696 x 39 261 0 regret 1632 x 4 0 7 relish 529 y 1 0 1 resent 914 y 39 0 39 stand 28665 x 1 0 29 trust 3528 x 93 328 0 want 90595 x 129 9 11687 815 wish 16043 x 30 481 0 Totals 14081 1681

Different-subject Communication constructions 409

Table 4: Real and projected totals for different-subject Communication constructions

Totals per 1000(x)/ real

totals (y) Projected totals (x)/real

totals (y) Matrix verb

BNC x/y

to inf. -ing bare inf. to inf. -ing bare inf. acknow-ledge

4168 x 1 4 0 0

advise 5023 x 160 1 804 5 0 announce 12453 x 2 24 0 0 ask 60618 x 124 7517 0 0 beg 1945 x 178 2 346 0 4 beseech 64 y 12 2 12 0 2 bid 1342 x 12 59 16 0 79 challenge 3628 x 38 138 0 0 claim 17031 x 3 51 0 0 command 1554 x 39 61 0 0 convince 3604 x 28 101 0 0 counsel 248 y 16 16 0 0 dare 2703 x 24 65 0 0 declare 5996 x 34 204 0 0 defy 790 y 48 48 0 0 demand 5633 x 4 23 0 0 describe 23413 x 3 0 70 0 direct 5248 x 21 110 0 0 discour-age

1064 x 1 1 1 1 0

encourage 10438 x 232 2422 0 0 enjoin 77 y 23 23 0 0 entreat 37 y 13 13 0 0 exhort 184 y 94 94 0 0 forbid 1269 x 76 96 0 0 implore 141 y 66 66 0 0 incite 170 y 41 41 0 0 instruct 1708 x 329 562 0 0 invite 5951 x 251 1494 0 0 mention 10804 x 1 0 11 0 mime 102 y 1 0 1 0 order 6859 x 171 1173 0 0 own 5999 x 1 6 0 0 persuade 4912 x 539 2648 0 0 plead 1671 x 1 2 0 0 pledge 907 y 53 53 0 0

Appendix 2: Tables

410

Table 4: Real and projected totals for different-subject Communication constructions: continued




BNC x/y

to inf. -ing bare inf. to inf. -ing bare inf. press 7068 x 25 177 0 0 proclaim 1167 x 32 37 0 0 profess 292 y 2 2 0 0 pronounce 1262 x 13 16 0 0 propose 7511 x 2 15 0 0 recom-mend

5771 x 20 115 0 0

remind 5137 x 21 108 0 0 report 16922 x 3 1 51 17 0 request 2257 x 103 232 0 0 summon 1370 x 50 69 0 0 suggest 28490 x 1 28 0 0 tell 77112 x 52 4010 0 0 tip 1415 x 5 7 0 0 urge 3521 x 503 1771 0 0 volunteer 843 y 5 5 0 0 Totals 24877 105 85

Table 5: Real and projected totals for Enablement constructions



Matrix verb

BNC x/y

to inf. -ing bare inf. to inf. -ing bare inf. afford 5321 x 5 27 0 0 allow 33222 x 467 1 15515 0 33 appoint 5854 x 64 375 0 0 arrange 6857 x 5 1 34 7 0 assign 1744 x 7 12 0 0 assist 3930 x 69 1 271 4 0 authorise 1457 x 154 224 0 0 bribe 254 y 67 67 0 0 choose 16478 x 23 379 0 0 contract 1286 x 6 8 0 0 design 11380 x 9 102 0 0

Different-subject Enablement constructions 411

Table 5: Real and projected totals for Enablement constructions: continued



Matrix verb

BNC x/y

to inf. -ing bare inf. to inf. -ing bare inf. elect 3912 x 13 51 0 0 employ 7752 x 41 318 0 0 empower 491 y 142 142 0 0 enable 9633 x 873 8410 0 0 engage 3918 x 9 35 0 0 entitle 5090 x 49 249 0 0 equip 1690 x 29 49 0 0 fit 8537 x 1 9 0 0 forbid 1269 x 81 8 103 10 0 help 37480 x 84 137 3148 0 5135 involve 28953 x 9 0 261 0 let 28678 x 774 0 0 22197 mean 65690 x 2 0 131 0 name 5619 x 3 17 0 0 need 59627 x 8 2 477 119 0 pay 34759 x 9 313 0 0 permit 3958 x 239 946 0 0 require 28386 x 1 0 28 0 risk 1825 x 17 0 31 0 save 11216 x 11 0 123 0 seek 15654 x 2 31 0 0 suffer 11978 x 3 36 0 0 teach 8281 x 94 778 0 0 tip 1415 x 1 1 0 0 tolerate 1098 x 1 0 1 0 use 104883 x 23 2412 0 0 vote 5035 x 1 5 0 0 Totals 34544 687 27365

Appendix 2: Tables

412

Table 6: Real and projected totals for Causation constructions



Matrix verb

BNC x/y

to inf. -ing bare inf. to inf. -ing bare inf. avoid 11461 x 7 0 80 0 cause 18984 x 147 1 2791 0 19 compel 1070 x 302 323 0 0 condemn 2264 x 10 23 0 0 constrain 857 y 23 23 0 0 decide 24221 x 1 24 0 0 delay 2235 x 3 0 7 0 detail 791 y 7 7 0 0 determine 10242 x 4 41 0 0 drive 12961 x 6 78 0 0 force 11169 x 249 2781 0 0 get 220940 x 20 7 4419 1547 0 have 1213785 x 4 1 0 4855 1214 impel 119 y 32 32 0 0 induce 1888 x 163 308 0 0 inspire 2244 x 63 141 0 0 keep 49597 x 30 0 1488 0 lead 30344 x 59 1790 0 0 leave 60533 x 36 19 2179 1150 0 make 213476 x 85 0 0 18145 oblige 1956 x 135 264 0 0 postpone 1093 x 1 0 1 0 preclude 663 y 33 0 33 0 prevent 10031 x 198 0 2184 0 prompt 2043 x 237 484 0 0 push 9071 x 3 27 0 0 require 28386 x 93 2640 0 0 set 38708 x 7 5 271 194 0 start 39838 x 1 0 40 0 stop 24077 x 73 0 1758 0 trouble 1050 x 7 7 0 0 Totals 18653 13337 19378

Same-subject Effort constructions 413

Table 7: Real and projected totals for Effort constructions



Matrix verb BNC x/y

to inf. -ing bare inf. to inf. -ing bare inf. aim 7180 x 407 2922 0 0 assay 143 y 1 1 0 0 attempt 8364 x 798 1 6674 8 0 battle 845 y 141 141 0 0 bid 1342 x 83 111 0 0 conspire 346 y 180 180 0 0 contrive 396 y 191 191 0 0 endeavour 594 y 571 571 0 0 essay 24 y 2 2 0 0 fail 15864 x 725 11501 0 0 help 37480 x 105 122 3935 0 4573 labour 474 y 39 39 0 0 learn 17218 x 200 2 3444 34 0 manage 13622 x 499 6797 0 0 push 9071 x 3 27 0 0 seek 15654 x 391 6121 0 0 strain 1069 x 164 175 0 0 strive 987 y 627 627 0 0 struggle 3521 x 420 1479 0 0 sweat 563 y 6 6 0 0 try 52868 x 681 25 36003 1322 0 Totals 80947 1364 4573

Appendix 2: Tables

414

Table 8: Real and projected totals for same-subject Mental Process constructions



Matrix verb

BNC x/y

to inf. -ing to inf. -ing anticipate 2265 x 1 22 2 50 consider 28082 x 56 0 1573 contem-plate

1535 x 155 0 238

decide 24221 x 335 8114 0 determine 10242 x 16 164 0 envisage 1746 x 30 0 52 expect 25191 x 126 3174 0 foresee 780 y 3 0 3 forget 12669 x 99 3 1254 38 imagine 8219 x 46 0 378 intend 10623 x 410 27 4355 287 judge 4029 x 1 4 0 know 184765 x 1 185 0 mean 65690 x 13 854 0 overlook 2020 x 1 0 2 plan 10015 x 273 2734 0 plot 937 y 100 100 0 purpose 6 y 3 3 0 recall 5291 x 46 0 243 reckon 3947 x 12 47 0 recollect 227 y 12 0 12 remember 26656 x 19 68 506 1813 resolve 3680 x 98 361 0 scheme 100 y 13 13 0 think 152596 x 5 763 0 trust 3528 x 1 4 0 Totals 22817 4689

Same-subject Attitude constructions 415

Table 9: Real and projected totals for same-subject Attitude constructions



Matrix verb

BNC x/y

to inf. -ing to -ing to inf. -ing to -ing abhor 117 y 1 0 1 0 abide 352 y 1 1 0 0 ache 603 y 45 45 0 0 adore 528 y 3 20 3 20 0 aspire 534 y 145 14 145 0 14 bear 8251 x 53 6 437 50 0 burn 4341 x 5 22 0 0 care 7269 x 80 582 0 0 chafe 85 y 1 1 0 0 crave 301 y 10 10 0 0 delight 836 y 7 7 0 0 desire 908 y 151 151 0 0 detest 249 y 10 0 10 0 die 21688 x 4 87 0 0 dislike 1119 x 71 0 79 0 dread 462 y 33 25 33 25 0 enjoy 14437 x 114 0 1646 0 face 15157 x 5 0 76 0 fancy 1645 x 123 0 202 0 fear 4438 x 25 12 111 53 0 fret 267 y 3 3 0 0 hanker 99 y 4 4 0 0 hate 4620 x 94 79 434 365 0 hope 20542 x 232 4766 0 0 like 41732 x 357 41 14898 1711 0 loathe 387 y 4 16 4 16 0 long 1308 x 431 564 0 0 look 106824 x 9 961 0 0 love 13683 x 90 32 1231 438 0 mind 7392 x 99 0 732 0 pant 359 y 4 4 0 0 pine 92 y 3 3 0 0 prefer 6696 x 402 29 2692 194 0 regret 1632 x 32 108 52 176 0

Appendix 2: Tables

416

Table 9: Real and projected totals for same-subject Attitude constructions: continued




BNC x/y

to inf. -ing to -ing to inf. -ing to -ing relish 529 y 46 0 46 0 stand 28665 x 9 3 258 86 0 stomach 103 y 2 0 2 0 thirst 24 y 4 4 0 0 value 1470 x 1 0 1 0 want 90595 x 597 54085 0 0 wish 16043 x 402 6449 0 0 yearn 256 y 67 67 0 0 Totals 88115 5929 14

Table 10: Real and projected totals for same-subject Communication constructions



totals (y) Matrix verb BNC x/y

to inf. -ing to -ing to inf.

-ing to -ing

acknowledge 4168 x 3 0 13 0 admit 11070 x 1 66 2 11 731 22 agree 23316 x 231 5386 0 0 apply 18496 x 23 425 0 0 ask 60618 x 7 424 0 0 beg 1945 x 65 126 0 0 beseech 64 y 1 1 0 0 claim 17031 x 139 2367 0 0 confess 1609 x 7 79 11 0 127 consent 613 y 94 94 0 0 contract 1286 x 60 77 0 0 crave 301 y 2 2 0 0 decline 3049 x 242 738 0 0 demand 5633 x 37 208 0 0 deny 6933 x 59 0 409 0 describe 23413 x 3 0 70 0 discuss 14697 x 4 0 59 0 engage 3918 x 3 12 0 0

Same-subject Communication constructions 417

Table 10: Real and projected totals for same-subject Communication constructions: continued

Totals per 1000(x)/

real totals (y) Projected totals (x)/real totals

(y) Matrix verb

BNC x/y

to inf. -ing to -ing to inf. -ing to -ing entreat 37 y 1 1 0 0 implore 141 y 1 1 0 0 mention 10804 x 2 0 22 0 offer 25704 x 52 1337 0 0 plead 1671 x 11 18 0 0 pledge 907 y 276 276 0 0 pray 2591 x 11 29 0 0 press 7068 x 1 7 0 0 proclaim 1147 x 1 1 0 0 profess 292 y 124 6 124 0 6 promise 6297 x 302 1902 0 0 propose 7511 x 155 1164 0 0 refuse 10003 x 683 6832 0 0 renounce 361 y 1 0 1 0 report 16922 x 10 0 169 0 request 2257 x 10 23 0 0 swear 2185 x 29 63 0 0 threaten 6097 x 329 2006 0 0 under-take

5460 x 104 568 0 0

volunteer 843 y 275 275 0 0 vote 5035 x 123 619 0 0 vow 620 y 336 336 0 0 whine 254 y 3 3 0 0 Totals 25467 1474 155

Appendix 2: Tables

418

Table 11: Real and projected totals for Aspect constructions

Totals per 1000(x)/real totals(y)

Projected totals(x)/real totals(y)

Matrix verb

BNC x/y

to inf. -ing to -ing to inf. -ing to -ing begin 43172 x 452 73 19514 3152 0 cease 2915 x 563 67 1641 195 0 commence 1413 x 17 55 24 78 0 complete 9605 x 1 0 10 0 continue 28711 x 437 40 12547 1148 0 discontinue 313 y 7 0 7 0 fall 23610 x 1 0 0 24 finish 11427 x 1 81 11 926 0 get 220940 x 3 2 662 442 0 grow 17320 x 15 260 0 0 keep 49597 x 115 0 5703 0 lie 19145 x 12 0 230 0 proceed 3639 x 205 746 0 0 quit 1124 x 49 0 55 0 recomm-ence

69 y 9 0 9 0

remain 25435 x 24 9 610 229 0 repeat 6483 x 2 0 13 0 resume 1633 x 43 0 70 0 set 38708 x 4 155 0 0 sit 28907 x 23 0 665 0 start 39838 x 154 190 6135 7569 0 stay 18686 x 1 0 19 0 stop 24077 x 212 0 5104 0 take 179013 x 5 0 0 895 use 104883 x 26 2727 0 0 Totals 45032 25624 919

Same-subject Applied Attitude constructions 419

Table 12: Real and projected totals for Applied Attitude constructions



Matrix verb

BNC x/y

to inf. bare inf.

-ing to inf. bare inf. -ing

affect 12340 x 5 62 0 0 afford 5321 x 385 2049 0 0 ape 64 y 2 0 0 2 arrange 6857 x 81 555 0 0 assume 10932 x 1 11 0 0 avoid 11461 x 200 0 0 2292 bear 8251 x 8 0 0 66 begrudge 69 y 3 0 0 3 bother 3905 x 210 61 820 0 238 choose 16478 x 226 3724 0 0 con-descend

64 y 21 21 0 0

dare 2703 x 306 225 827 608 0 defer 556 y 6 0 0 6 defy 790 y 6 0 0 6 deign 73 y 66 66 0 0 delay 2235 x 58 0 0 130 deserve 3077 x 191 4 588 0 12 disdain 55 y 12 12 0 0 elect 3912 x 75 293 4 0 endure 1111 x 3 6 3 0 6 escape 5030 x 12 0 0 60 eschew 218 y 5 0 0 5 evade 511 y 11 0 0 11 face 15157 x 2 0 0 30 fall 23610 x 3 71 0 0 feign 108 y 4 3 4 0 3 forbear 50 y 29 4 29 0 4 get 220940 x 12 2651 0 0 grudge 75 y 2 0 0 2 hasten 482 y 161 161 0 0 help 37480 x 16 0 0 600 hesitate 2200 x 180 396 2 0 invite 5951 x 1 0 0 6 justify 3505 x 24 0 0 84

Appendix 2: Tables

420

Table 12: Real and projected totals for Applied Attitude constructions: continued



Matrix verb

BNC x/y

to inf. bare inf.

-ing to inf. bare inf.

-ing

make 213476 x 3 640 0 0 mime 102 y 7 0 0 7 miss 9652 x 15 0 0 145 move 34288 x 7 240 0 0 need 59627 x 386 3 32 23016 179 1908 neglect 1237 x 95 2 118 0 2 omit 1282 x 99 127 0 0 pay 34759 x 7 243 0 0 postpone 1093 x 35 0 0 38 presume 1105 x 68 75 0 0 pretend 2731 x 443 1210 0 0 require 28386 x 10 6 284 0 170 resent 914 y 77 0 0 77 resist 3326 x 75 0 0 249 risk 1825 x 271 0 0 495 scorn 209 y 12 12 0 0 scruple 9 y 9 9 0 0 serve 15459 x 101 1561 0 0 shirk 94 y 2 0 0 2 shun 234 y 1 3 1 0 3 speed 1251 x 1 1 0 0 stand 28665 x 9 3 258 0 86 suffice 657 y 87 87 0 0 tolerate 1098 x 2 0 0 2 trouble 1050 x 63 66 0 0 undertake 5460 x 4 0 0 22 venture 925 y 142 142 0 0 Totals 40433 793 6772

The 30 most common same-subject to infinitive constructions 421

Table 13: Real and projected totals for the 30 most common same-subject to infinitive constructions

Matrix verb BNC Totals per 1000 Projected totals want 90595 597 54085 try 52868 681 36003 need 59627 386 23016 begin 43172 452 19514 like 41732 357 14898 continue 28711 437 12547 fail 15864 725 11501 decide 24221 335 8114 refuse 10003 683 6832 manage 13622 499 6797 attempt 8364 798 6674 wish 16043 402 6449 start 39838 154 6135 seek 15654 391 6121 agree 23316 231 5386 hope 20542 232 4766 intend 10623 410 4355 help 37480 105 3935 choose 16478 226 3724 learn 17218 200 3444 get 220940 15 3314 expect 25191 126 3174 aim 7180 407 2922 plan 10015 273 2734 use 104883 26 2727 prefer 6696 402 2692 claim 17031 139 2367 afford 5321 385 2049 threaten 6097 329 2006 promise 6297 302 1902 Total 270183

Appendix 2: Tables

422

Table 14: Real and projected totals for the 30 most common same-subject -ing constructions

Matrix verb BNC Totals per 1000 Projected totals start 39838 190 7569 keep 49597 115 5703 stop 24077 212 5104 begin 43172 73 3152 avoid 11461 200 2292 need 59627 32 1908 remember 26656 68 1813 like 41732 41 1711 enjoy 14437 114 1646 consider 28082 56 1573 try 52868 25 1322 continue 28711 40 1148 finish 11427 81 926 mind 7392 99 732 admit 11070 66 731 sit 28907 23 665 help 37480 16 600 risk 1825 271 495 get 220940 2 442 love 13683 32 438 deny 6933 59 409 imagine 8219 46 378 hate 4620 79 365 intend 10623 27 287 resist 3326 75 249 recall 5291 46 243 bother 3905 61 238 contemplate 1535 155 238 lie 19145 12 230 remain 25435 9 229 Total 42836

The 30 most common different-subject to infinitive constructions 423

Table 15: Real and projected totals for the 30 most common different-subject to infinitive constructions

Matrix verb BNC Totals per 1000 Projected totals allow 33222 467 15515 want 90595 129 11687 enable 9633 873 8410 ask 60618 124 7517 expect 25191 211 5315 get 220940 20 4419 tell 77112 52 4010 help 37480 84 3148 cause 18984 147 2791 force 11169 249 2781 persuade 4912 539 2648 require 28386 93 2640 encourage 10438 232 2422 use 104883 23 2412 leave 60533 36 2179 lead 30344 59 1790 urge 3521 503 1771 invite 5951 251 1494 like 41732 29 1210 order 6859 171 1173 permit 3958 239 946 know 184765 5 924 believe 34516 26 897 advise 5023 160 804 teach 8281 94 778 consider 28082 25 702 instruct 1708 329 562 prompt 2043 237 484 wish 16043 30 481 need 59627 8 477 Total 92387

Appendix 2: Tables

424

Table 16: Real and projected totals for the 3 most common different-subject -ing constructions

Matrix verb BNC Totals per 1000 Projected totals see 191256 40 7650 have 1213785 4 4855 find 76982 38 2925 hear 35975 68 2446 prevent 11031 198 2184 stop 24077 73 1758 feel 60570 26 1575 get 220940 7 1547 keep 49597 30 1488 leave 60533 19 1150 watch 19254 59 1136 imagine 8219 129 1060 want 90595 9 815 remember 26656 30 800 like 41732 8 334 mind 7392 38 281 involve 28953 9 261 notice 9040 27 244 catch 14342 16 229 set 38708 5 194 mean 65690 2 131 recall 5291 24 127 save 11216 11 123 observe 7223 17 123 need 59627 2 119 spot 2107 48 101 avoid 11461 7 80 describe 23413 3 70 understand 22880 3 69 envisage 1746 35 61 Total 33936

Appendix 3

List of technical terms

Accomplishment: This term is employed in the sense of Vendler (1967:102) and refers to a situation that evolves through time and that has an inherent right-hand temporal boundary, such as ‘climbing a mountain’ or ‘eating a biscuit’. Achievement: This term of Vendler’s (1967:103) refers to punctual actions: i.e. actions which may be pinned down to a particular instant, such as ‘arriving in London’ or ‘giving birth’. Applied Attitude constructions: Same-subject constructions in which either the speaker or the subject formulates an attitude about some situation (definitely) involving the latter. Aspect constructions: Same-subject constructions in which the subject commences, continues or discontinues his or her participation in some sort of situation. Attitude constructions: Both same- and different-subject constructions in which the matrix verb subject formulates an attitude towards a situation which either involves him- or herself or some other on-stage participant. Backward-looking constructions: Constructions in which the situation in the complement clause is located before the time of the matrix verb. Causation constructions: Different-subject constructions in which the matrix verb subject directs the realisation of the complement clause situation by the subject of the complement predicate, who has no independent say in the matter. Communication constructions: Both same- and different-subject constructions in which the matrix verb subject describes, raises the question of, or expresses an attitude concerning, either their own or a third party’s possible participation in some sort of situation. Complement Clause: A clause without which an utterance would be incomplete. It is a complement of the matrix verb in the sense that the

426 Appendix 3: Technical Terms

relevant clause type is entrenched in the grammar as co-occurring with that particular sort of verb. Complementiser: A morpheme which indicates the subordinate nature of the clause in which it occurs. Complementisers are integral to most non- finite constructions as these are defined in this study. For example, the ‘like S2 to infinitive’ construction contains the complementiser to and the ‘like S2 -ing’ construction the complementiser -ing. Construal: The conceptualisation of a particular situation in a particular way by a speaker or addressee. Construction: A unit containing a combination of linguistic elements, at least one of which is schematic for more than one lexeme. The term is to be understood in the sense of Bybee (2001), who describes it as follows: “Constructions are frequently used sequences of morphemes or words that bear a particular semantic or functional relation to one another when used together – a relation that they do not necessarily have outside that construction. [...] At one end of a continuum involving constructions are fixed phrases, such as I don’t know and c’est à dire ‘that is to say’; nearer the middle are constructions with some grammatical material and a slot that is more open (e.g., the preposition dans with its noun phrase object; and, at the most general end, are constructions such as [NOUN + PLURAL + ADJECTIVE], with two slots that are very open. It seems useful to restrict the term ‘construction’ to sequences that include a relatively open slot and to classify phrases without open slots, such as c’est à dire, as fixed phrases.” (Bybee: 2001: 173) Contemplation constructions: Constructions in which the situation in the complement clause is understood as existing purely in the mind of the speaker and addressee. Dynamic evolutionary model: This term is employed in the sense of Langacker (1991: 569) who defines it as “a fundamental cognitive model which conceives the world as having structure in a particular way, and reality as having a certain evolutionary momentum that constrains its future development (certain future paths constituting potential reality, and others projected reality)”. Effort constructions: Same-subject constructions in which the subject expends energy of some sort in order to realise some sort of situation. Enablement constructions: Different-subject constructions in which the matrix verb subject either sets the stage for the realisation of the

Enablement constructions 427

complement clause situation by the subject of the complement predicate or assumes the role of a minor character in assisting the subject of the complement predicate to realise it. Forward-looking constructions: Constructions in which the situation in the complement clause is located in the projected future vis-à-vis the time of the matrix verb. General constructions: Constructions in which the situation in the complement clause is encoded as likely (or unlikely) to occur on a more or less regular basis. Judgement construction: Constructions which encode the expression of an opinion on the part of the matrix verb subject as to the veracity of the situation in the complement clause. Landmark: Langacker (1991: 549) defines ‘landmark’ as “a salient substructure other than the trajector of a relational predication”. In active sentences the landmarks of processes are usually encoded syntactically as objects. Mental Process constructions: Both same- and different-subject constructions in which the matrix verb subject thinks about either him- or herself or somebody else being/engaging in some sort of situation. Perception constructions: Different-subject constructions in which the matrix verb subject merely registers the action on stage. Profiling: The highlighting of certain features in a situation by the speaker and addressee in the course of their construal of this situation. Projected future: This term is used to refer to the ontological domains called “potential reality” and “projected reality” by Langacker in his “dynamic evolutionary model” (1991: 569). It encompasses the future as one would expect it to evolve, given one’s knowledge of the present and experience of the past. The term ‘projected future’ has been adopted in preference to ‘projected reality’ in order to indicate clearly that the ‘reality’ in question is always projected forward into the (relative) future. Prototype theory : The theory that entities are categorised with respect to the degree of family resemblance they bear to what is taken to be the most central, or core, member of the class to which they belong. In contrast to the Aristotelian categorial-attribute method of characterisation, in which all members of a class share certain necessary and sufficient

428 Appendix 3: Technical Terms

attributes and according to which all members are viewed as equally central, the prototype model allows for more central and more peripheral members. See Taylor (1995 Chs 3 & 4) for a thorough presentation of the role of prototypes in linguistic theory. Same-time constructions: Constructions in which the situation in the complement clause is understood to be concurrent with the time of the matrix verb. Schema: This is defined by Langacker (1991: 552) as “a semantic, phonological, or symbolic structure that, relative to another representation of the same entity, is characterized with lesser specificity and detail. A “coarse-grained” (as opposed to a fine-grained) reprsentation. (Equivalent to the relation between a superordinate and a subordinate structure in a taxonomic hierarchy.)” Thus, in the context of the present study, ‘General matrix verb S2 to’ is an example of a very coarse-grained schema, ‘Attitude General matrix verb S2 to’ may be viewed as a less coarse-grained schema, and ‘love S2 to’ an even finer-grained schema. The more specific schemas instantiate the less specific ones. Sequential scanning: This is defined by Langacker (1991: 553) as “a mode of processing in which a series of component states are activated successively in non-cumulative fashion (i.e. a situation is followed in its evolution through conceived time, as in watching a film)”.

Speaker: The begetter of a linguistic communication, be it spoken or written. Summary scanning: This is defined by Langacker (1991: 553) as “a mode of processing in which a set of specifications or a series of component states are activated successively yet cumulatively; thus, after a build- up phase, all facets of a complex structure are coactivated and simultaneously accessible”. Trajector: Langacker (1991: 555) defines ‘trajector’ as “the (primary) figure within a profiled relation”. In ‘processes’, encoded as clauses, the trajector is normally encoded as the subject.

Index

Page references for individual matrix verbs are included in Appendix 1. accomplishment 69, 99, 109-111,

113-116, 121-132, 136, 140, 147- 152, 185, 226, 267, 282- 283, 307

Achard 14, 20, 24-25, 92, 95, 271, 273

achievement 99, 109-111, 118-128, 131-132, 135, 147, 150-151, 267, 282

activity 54-55, 66-68, 73, 99, 109-113, 121-132, 135, 137, 147-152, 162-163, 172-173, 185-188, 226, 256, 261-262, 266, 293-294, 307- 308

Andersson 6-7 agentivity 149-150, 176, 212, 214, 226, 257-258, 260-261, 264, 270-273, 275-277, 279, 291-292, 294, 298-303 see also: semantic roles: Agent Bailey 62, 95 Biber et al. 100, 102, 237, 239-240,

257 Bladon 165-166, 180, 184 Bolinger 5, 11, 52-54, 151-153,

183, 245 Brinton 186, 265 Bybee 242 Carter and McCarthy 162 Chomsky 94-95 Chuquet 100 conceptual distance 78-86 conceptual overlap 78-86, 135 Conrad 55-57, 86, 89, 159, 164,

183, 241, 248, 262-263, 282 constructions: types of

Applied Attitude 13, 27-28, 30- 31, 40, 72, 91, 93-94, 98, 108, 160-161, 237-243

Aspect 19, 27-31, 38, 50, 54, 56, 93, 108, 136, 184-196, 254- 267, 278, 291-303

Attitude 21-31, 40, 50-51, 54, 65-66, 71, 77, 93, 106-108, 117, 131, 136, 161-184, 191, 244, 249, 253, 267, 278-279, 286-288, 303, 306, 309

Backward-looking 19, 32, 34, 39- 43, 49, 51-54, 57, 60-63, 67, 74, 78, 82-83, 87, 91-92, 94, 99, 101, 105-106, 108, 112-124, 126-127, 129-132, 141, 143, 244, 281-303

Causation 22-26, 30-31, 91, 93, 108, 132, 267-278

Communication 19, 22-31, 42, 54, 91, 93, 108, 136-137, 200- 204, 278-279, 288-291

Contemplation 38-43, 52-57, 60-61, 67-68, 74, 78, 82-83, 87, 91-92, 94, 99, 107-111, 113-129, 132, 143, 156, 158, 244, 282, 309

Effort 26-31, 91, 108, 156-161, 191, 204-214, 278

Enablement 22-26, 30, 91, 93, 100, 108, 132, 204-236, 278, 306

Forward-looking 19, 32-34, 38- 43, 46-47, 50, 52-54, 56-57, 60-62, 67, 71-87, 91- 93, 96-98, 107-127, 130-131, 135-136, 143, 156-163, 169, 171, 175-177, 190-191, 195-196, 199-280, 286, 306, 308

General 34-40, 42-43, 46-47, 50, 53-54, 56-57, 64, 71, 73, 77, 81-82, 86, 91-93, 97-98, 105-

430 Index

106, 109, 134-135, 161-196, 239, 291-299, 308

Judgement 39-40, 42-43, 49, 53, 56-59, 62-63, 71-73, 77, 80-82, 86-87, 91-94, 97-98, 109, 134-137, 141, 151-155, 196, 239, 244, 288-291, 308

Mental Process 21-25, 27-31, 38, 42, 54, 59, 63, 74-75, 93-94, 107-109, 117, 131, 153, 191, 243-249, 267, 278, 281-286, 288, 306

Perception 20-25, 30-31, 59, 62- 63, 66, 91, 94, 108, 145-155, 196, 272

Same-time 32, 38-40, 42-43, 46-52, 54, 57, 59-61, 65-67, 71, 73-74, 78, 82-83, 86-87, 91, 94, 99, 105-108, 110-127, 135, 143, 145-197, 232, 249, 303, 307-308

Cotte 233 Croft 5 De Smet and Cuyckens 162 Dirven 1, 11, 46-48, 55, 83-84, 86-

87, 148, 157, 261, 272, Dixon 1, 9, 11, 13-14, 22, 25, 53-

55, 57, 66, 86-87, 96, 109, 128, 135, 205, 238, 255-256, 273, 290, 294- 295, 297, 300, 308

Dowty 9, 260 Duffley 1-2, 5, 7, 11, 46, 48, 61-69,

76, 78, 80, 86-87, 92, 95, 109, 162, 203, 205, 211, 217, 232-233, 237, 259-260, 272, 293, 299

Duffley and Joubert 61 Duffley and Tremblay 159, 160 dynamic evolutionary model 33, 39, 54, 68, 306-307 Eagleston 180-182 Edelman 74 Egan 143, 168, 178, 240, 266, 310 factivity 48-51, 87 Fanego 1, 4, 89, 305 Farkas 14

Ferris 290 Fischer 95 force dynamics 76, 92, 114, 209, 212, 217-236 Freed 6, 38, 50, 64, 68, 105, 185-

187, 256-258, 260, 262-264, 293, 299, 308

general validity predications 34-38, 41, 47, 64, 97, 162-163, 168, 170, 173, 177-181, 184, 188-196, 230, 291, 293, 297, 302-303

see also: constructions: General Girard 294 given/new: 47-48, 140, 170, 187- 188, 191, 194 Givón 1, 9, 27, 89, 252, 270 Goldberg 70, 203 Greenbaum et al. 2, 15 Gries 1 Hamawand 3-4 holistic construal 69, 75-77, 82, 86 Horie 5 Huddleston and Pullum 65, 109-

112, 145, 162-163, 218 Hudson et al. 215, 232 hypothetical 52-53, 83 imperfective see: perfective/imperfective implicativeness 27, 50-51 interiority 64-65, 69, 109 see also: progressive: mid-interval Jackendoff 3 Jespersen 4, 5, 45 Jørgensen 4 Karttunen 50 Kemmer and Verhagen 218-219 Kempson and Quirk 51, 54, 164 Kiparsky and Kiparsky 48-50, 86 Kjellmer 204-205 Krug 95 Kruisinga 45, 183-184 Lakoff 24 landmark 20, 237, 241-242, 249,

254

Index 431

Langacker 1-3, 9, 11, 13, 20, 22, 33- 38, 54, 68, 75-78, 80, 82, 86-87, 92, 109, 115, 128-129, 131, 135, 147, 165, 241, 269, 272, 287, 305-307

Leech 65 Lind 204-206, 211 Mair 1-2, 7, 13-14, 20, 89, 204-205,

207-208, 245, 266, 308 Matthews 2, 14 McGregor 6 Mittwoch 233, 272-273 Mukherjee 1, 40-41 Noël 59, 61, 71-72, 151 Palmer 25, 68, 152 perfective/imperfective 69, 87, 118-

120, 128, 147, 154, 195 see also: progressive: imperfectivity Popper 6 potentiality 32-33, 51-54, 71, 86,

95-96, 183, 256 Poutsma 45, 183-184 progressive 60-61, 65-69, 78, 82,

87, 109-120, 124-127, 132, 145, 156-157, 230, 256, 307

characteristics of: durativity 65-66, 111-122, 124-

128, 130, 132, 148-149, 185-186, 254, 267, 278, 282-283, 300, 302, 306

dynamism 65-69, 109, 111-122, 124-125, 127, 129

imperfectivity 65-66, 109, 111-122, 125-128, 130, 135-136, 140, 267, 278, 306

limited duration 65, 109, 111-122, 124-127, 145

mid-interval 61, 65, 109, 111-122, 124-130, 132, 148, 162-163, 169, 172, 183, 185-186, 193, 196-197

ongoingness 65-69, 86, 109, 111-122, 124-127, 129, 137, 140, 146, 148, 150, 154- 155, 157,

180, 193, 195, 210, 254, 261, 266-267, 277, 294, 303

projected future 33-34, 37-38, 52, 57, 60, 77, 80, 84, 86-87, 98, 102, 133, 157-158, 161, 168, 171, 175, 178-179, 190, 197, 199, 203, 234, 237, 239, 241, 244, 246-247, 249-250, 252-253, 263, 267, 278-280, 286, 297, 299, 307-308

see also: constructions: Forward- looking Quirk 150, 164 Quirk et al. 12-13, 20, 52, 66-67,

86, 134, 138, 148, 216, 256, 261- 262

referring predications 55-57 reification 52-53, 245 Rohdenburg, 1, 159, 190, 207, 310 Ross 265 Rudanko 1- 2, 12, 50, 51, 76, 138 Sapir 41, 168, 305, 316 scope of predication 34, 75-78,

128, 146 Searle 70 semantic roles: Agent 10, 46, 136, 149, 212, 214, 241, 257, 260-261, 266-267, 270, 301, 303 see also: agentivity Experiencer 3, 66, 136, 212, 264 Patient 252, 260, 264 Simon-Vandenbergen 105-106 Skandera 266 Smith and Escobedo 78-87, 135,

308 specific predications 34-35, 45-48,

64, 86, 101, 168, 171, 183-189, 191-196, 229, 262, 264, 293, 296-297

state 22, 54, 74, 77-78, 80, 83, 92, 109-111, 113, 116-119, 122-129, 132, 136-137, 152, 162-163, 169, 172, 176, 180, 188, 293

Stubbs 89 Sweet 45

432 Index

Talmy 219 targeted alternative 96, 98, 100, 103-107, 137, 141-143, 152-153, 155, 157, 161, 164, 190, 193, 197, 207, 212, 240, 242-243, 250- 252, 267, 278, 280, 285, 309 Taylor 251 telicity 116, 120, 147-151 trajector 13, 20, 26, 30, 96, 237, 240-242, 279, 254 Vendler 69, 110, 114, 121-122 Verspoor 1, 11, 25, 61, 69-75, 86-

87, 89, 96, 147, 157, 163, 192, 272, 283

Visser 12 Wierzbicka 1, 11, 13, 57-62, 65,

68, 74, 76-78, 80, 86-87, 135, 165, 215, 237, 252, 283, 294-298, 300-301, 308

Wood 45-47, 55, 86-87, 157, 164, 211, 248

Non-finite complementation

Documents

Transcript of Non-finite complementation