[Rocio Perez Tattam] Second Language Competence T(BookZZ.org)

Second Language Competence

Second Language Competence: The Acquisition of Complex Syntax in Spanish

By

Roco Prez Tattam

Second Language Competence: The Acquisition of Complex Syntax in Spanish,

by Roco Prez Tattam

This book first published 2011

Cambridge Scholars Publishing

12 Back Chapman Street, Newcastle upon Tyne, NE6 2XX, UK

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library

Copyright 2011 by Roco Prez Tattam

All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or

otherwise, without the prior permission of the copyright owner.

ISBN (10): 1-4438-2701-0, ISBN (13): 978-1-4438-2701-0

To family and friends

TABLE OF CONTENTS

Acknowledgements .................................................................................... ix

Introduction ................................................................................................. 1

Chapter One................................................................................................. 7 A Cross-linguistic Account of Control, Raising and Exceptional Case Marking

1.1. Control and Raising Structures in Generative Syntax 1.1.1. Control and Raising Structures in the Theory of Principles and Parameters 1.1.2. Control and Raising Structures in the Minimalist Program 1.1.3. Exceptional Case Marking in the Minimalist Program

1.2. Raising and Control Structures in Spanish and English 1.2.1. Optional Control Structures 1.2.2. Raising to Object Structures 1.2.3. Other differences between Spanish and English

1.3. Summary

Chapter Two .............................................................................................. 39 Learnability in Second Language Acquisition

2.1. Language Acquisition and Learnability 2.2. The Acquisition of Infinitival Clauses in Complex Structures

2.2.1. The Acquisition of the Complementizer 2.2.2. The Acquisition of Control

2.3. The Learnability Hypothesis of Control, Raising and Exceptional Case Marking 2.4. Summary

Chapter Three ............................................................................................ 65 The Acquisition of Control, Raising and Exceptional Case Marking

3.1. Predictions for the Acquisition of Control, Raising and Exceptional Case Marking 3.2. Experimental Design

3.2.1. Method 3.2.2. Participants

Table of Contents viii

3.3. Results 3.3.1. Language Proficiency Test 3.3.2. Grammaticality/acceptability judgment task 3.3.3. Forced choice task

3.4. Conclusions of the Experimental Study 3.5. Summary

General Conclusions.................................................................................. 93

Appendix A ............................................................................................... 99 Main Verbs

Appendix B.............................................................................................. 103 Language Assessment Questionnaire

Appendix C.............................................................................................. 105 Grammaticality Judgment Task

Appendix D ............................................................................................. 113 Forced Choice Task

Bibliography............................................................................................ 121

Index........................................................................................................ 137

ACKNOWLEDGEMENTS

This volume adapts and expands the research carried out for my PhD dissertation, which I defended in 2007 at the University of Ottawa, Canada. I am greatly indebted to my advisor, Prof. Juana M. Liceras, for many of the ideas, methods and analyses presented in this volume.

I would like to thank the members of the PhD committee, Prof. Jos Camacho, Prof. Helmut Zobl, Dr. Joyce Bruhn de Garavito and Dr. Eta Schneiderman, for their comments and suggestions for further research. These have been incorporated to a certain extent into this volume. A very special thanks goes to Dr. Alex Cuza from Purdue University, Dr. Raquel Fernndez Fuertes from the University of Valladolid, as well as several colleagues, friends and family members for their invaluable help in recruiting and testing participants.

The research for my PhD dissertation was partially funded by several sources, which I would like to acknowledge here: an Admission Scholarship from the Faculty of Graduate and Postdoctoral Studies of the University of Ottawa, a Summer Graduate Scholarship in Humanities and Sciences from the Social Sciences and Humanities Research Council (SSHRC), and a research grant from the Academic Development Fund of the Association of Part-Time Professors of the University of Ottawa (APTPUO).

INTRODUCTION

This volume addresses the question of the nature of language transfer in second language acquisition1. By analysing the process of acquisition of complex syntax by non-native learners of Spanish, it seeks to establish what kind of knowledge is transferred from the native language and what changes occur as learners become more proficient. It focuses on a particular class of grammatical constructions which typically contain infinitival clauses that complement the main verb: Control, Raising and Exceptional Case Marking structures. It presents acquisition data obtained by means of different experimental methods from adult learners of Spanish with English as their first language. By looking at language transfer phenomena, it deals with the development of linguistic competence in second language acquisition and degree of convergence between the non-native (or interlanguage) and native (or target) grammars.

From a syntactic perspective, a complex sentence is made up of two or more simple sentences, also known as clauses. This combination may take place by means of two recursive mechanisms: coordination and subordination (or sentence embedding). Subordinate or embedded clauses are characterized by being dependent on the so-called main clause. In isolation, embedded sentences are ungrammatical, as shown in (1b) as opposed to (1a) (ungrammatical constructions are marked with an asterisk).

(1a) [John says [that he wants beer]]

(1b) *[That he wants beer]

This study focuses on so-called verbal complements, that is, embedded clauses that are subcategorized by the main verb. Verbal complements are further subdivided according to the inflection of the embedded verb into finite and non-finite complements, as shown in (2) and (3).

(2) John told me [CP that he was sick]

(3) John hopes [TP to get better soon]

Introduction 2

The embedded clause in (2) is analysed as a Complementizer Phrase (CP), headed by a functional category called a complementizer (that) which serves as a nexus between both clauses. The embedded clause in (3) is analysed as a Tense Phrase (TP) headed by the non-finite (or infinitival) verb. When main verbs combine with infinitival clauses, they give rise to different syntactic configurations. This study focuses on four such configurations, shown in (4) to (7).

(4a) Mary decided to call John.

(4b) Mary forced John to decide.

(5) Mary seems to be happy.

(6) Mary considers John to be intelligent. (7) Veo al jardinero regar las plantas.

I see the gardener watering the plants. Examples (4a) and (4b) appear in so-called Control configurations,

example (5) appears in a Raising to Subject configuration, example (6) appears in a Raising to Object configuration and example (7) appears in an Exceptional Case Marking configuration.

These constructions constitute an interesting research topic for different reasons. From a syntactic perspective, there is interaction between the semantic and thematic properties of the main verb, and the type of syntactic configuration in which the infinitival clause appears. This has consequences not only for the distribution of infinitival clauses (i.e., the syntactic positions in which they appear) but also for their interpretation (i.e., the choice of antecedent for the infinitival clause). More specifically, infinitives do not allow overt subjects, as opposed to finite verbs. Therefore, they may enter into a relation of referential dependence with a sentence-external or sentence-internal referent. Infinitival clauses in the configurations above are referentially dependent on an argument in the main clause. For example, [to call John] in (4a) is coindexed with the main subject Mary (Mary decides and Mary calls John). In this case, the interpretation of the infinitival clause is straightforward, since there is no other possible referent in the main clause. In contrast, there are two possible referents in (4b), the main subject Mary and the main object John. Syntacticians try to account for the interpretation of infinitival clauses in Control, Raising and Exceptional Case Marking structures in terms of

Second Language Competence 3

syntactic operations, whereas semanticians maintain a bigger role for semantic factors across the board.

Syntactic studies have dealt extensively with Control, Raising or Exceptional Case Marking structures from a theoretical perspective. Acquisition studies have looked at the development of infinitival clauses as verbal complements (order of acquisition)2, but have not focused specifically on these constructions. That said, certain Control phenomena have been researched and described as late-learned rules, that is, grammatical phenomena for which adult native speaker ability is not attained by children until they are six or older (Goodluck and Birch 1987). Late-learned rules are important for language acquisition with regard to computational and processing complexity, and researchers in the field have tried to establish the characteristics of late-learned rules that slow down the acquisition of Control structures compared to other types of complex sentences. The majority of the data is naturalistic, although there are a number of experimental studies which deal with late-learned Control phenomena. The data usually come from child native (L1) learners of English, although there are a limited number of studies dealing with non-native (L2) learners of English with different native languages.

The specific focus of the study presented here is the acquisition of infinitival clauses in Control, Raising and Exceptional Case Marking structures by L2 learners of Spanish, in order to address the following research questions. On the one hand, what kind of L1 knowledge gets transferred to the L2? L1 transfer will be looked at as a measure of the development of the interlanguage grammar, since it is assumed that more proficient learners will transfer less as they adopt the syntactic representations that are appropriate for the target grammar. On the other hand, how much convergence is there between interlanguage and native grammars? And a related question: can there be convergence between these grammars? Adult learners have already adopted the representations of their L1 by the time they begin learning the L2. Presumably, these representations can be accessed through the L1. The framework for the study is the Generative theory of language acquisition, which crucially assumes that the language faculty underlies both L1 and L2 acquisition, even though native and interlanguage grammars differ with regard to the development of the properties of the L1 and the L2. This study also provides a contrastive analysis of Spanish and English infinitival complementation from the perspective of the latest trends of Generative theory, as well as evidence from the acquisition data that showcases the different properties of Control, Raising and Exceptional Case Marking in these languages. Unlike many previous acquisition studies, it does not focus on constructions

Introduction 4

belonging to the same parameter3. Control, Raising and Exceptional Case Marking structures superficially appear to be identical, but the underlying syntactic configurations are very different. As a result, properties related to the distribution and interpretation of infinitival clauses also differ across these constructions, as mentioned earlier. This study applies developmental mechanisms such as the Subset/Superset principle4 to the learnability hypotheses concerning these properties, as well as other proposals such as the Competing Grammars Hypothesis for L2 acquisition, which focuses on the similarities between acquisition and diachronic change.

In summary, the present study investigates the acquisition of infinitival clauses in Control, Raising and Exceptional Case Marking structures by adult L2 learners of Spanish (L1 English). It looks at the effects of L1 transfer and level of language proficiency in the distribution and interpretation of these clauses by means of two types of behavioural experiments: a self-paced (offline) grammaticality judgment task and a time-paced (online) forced choice task.

Chapter One presents a contrastive analysis of infinitival clauses in Control, Raising and Exceptional Case Marking structures in English and Spanish. It focuses on a number of properties that have been studied in the Generative literature, from the earliest work (Chomsky 1965; Chomsky 1973; Chomsky and Lasnik 1977) to the latest tendencies (Chomsky 1981, 1982, 1995, 2001). These properties concern the distribution and interpretation of infinitival clauses, that is, (a) the syntactic positions in which they appear in Control, Raising and Exceptional Case Marking structures, and (b) the choice of antecedent for the infinitival clause in these structures.

Chapter Two presents previous literature on the acquisition of Control, Raising and Exceptional Case Marking structures within the Generative framework, with the aim of providing the backdrop for the learnability hypotheses and the predictions for the acquisition of Control, Raising and Exceptional Case Marking in Spanish as an L2. This literature is presented in the context of past and present L2 acquisition theories, which address the issues of non-native competence and convergence between native and interlanguage grammars, and aim to predict which areas of language are more likely to show linguistic transfer.

Chapter Three presents the experimental study derived from the learnability hypotheses for the acquisition of Control, Raising and Exceptional Case Marking, which are based on the Competing Grammars Hypothesis for L2 acquisition (Zobl and Liceras 2004) and markedness in terms of subset/superset relationships between grammatical features across

Second Language Competence 5

English and Spanish. The two experimental tasks by which the L2 data was collected (the grammaticality judgment task and the forced choice task) deal with the distribution and interpretation of infinitival clauses in these syntactic configurations. The data analysis will indicate whether there are competing representations in the interlanguage grammar where Spanish and English differ (effect of the L1), and to what extent this competition persists over time (effect of the level of language proficiency). General conclusions derived from the data analysis are presented after this chapter, as well as several issues concerning the acquisition of complex structures.

Notes 1 Bilingual acquisition includes simultaneous and sequential bilingualism. Meisel (2001), and Butler and Hakuta (2004) define simultaneous bilingualism as the acquisition of two native languages (L1+L1) from birth and in a natural context. It is also known as L1 bilingualism (De Houwer 1990, 1995, 1998; Genesee 2001, Lanza 1993, Meisel 1989, 2001; Swain 1972). Sequential bilingualism is defined as the acquisition of a native language (L1), followed by the acquisition of a non-native language (L2) , mostly but not necessarily in an institutional context. Sequential bilingualism has also been termed consecutive or successive bilingualism (Albrecht 2004; McLaughlin 1978; Lindholm and Padilla 1978). Sequential bilingual learners can be children or adults. The present study deals with adult sequential bilinguals, which will be termed non-native or L2 learners, as opposed to native or L1 speakers. 2 The development of the ability to produce and understand complex sentences (including finite and non-finite complements) by child learners has been considered to be one of the most interesting and important aspects of language acquisition, as a distinctive trait of human languages (Limber 1973). 3 As observed in Ayoun (2005), parameters (...) represent the range of variation that can be found in natural languages (...) (71), they subsume (...) a cluster of phonological, syntactic, lexical, or morphological properties or structures governed by an abstract principle (...) and they are (...) traditionally assumed to be binary in that they have two (mutually exclusive) values or settings. (...) (72). For instance, the pro-drop or null subject parameter (Belletti 1982; Chomsky 1981; Jaeggli 1982; Jaeggli and Safir 1989; Perlmutter 1971; Rizzi 1982) distinguishes languages according to whether they allow null subjects in finite clauses or not (e.g., we go to the movies every weekend in English as opposed to vamos al cine todos los fines de semana in Spanish), and according to other related syntactic and morphological properties, such as subject-verb inversion (Ha llamado tu hermanavs. *Called your sister), that trace sequences (Quin crees que ha llamado? vs. *Who do you think that called?), and pleonastic pronouns ( Est lloviendo vs. Itis raining).

Introduction 6

4 The Subset/Superset principle (Berwick 1985) was originally developed to define the relationship between the options of a parameter. According to this principle, where the language defined by one parametric option forms a subset of a language defined by another parametric option, children should initially entertain the narrowest hypothesis to account for the input data, and should only extend their grammars on the basis of evidence that cannot be accounted for.

CHAPTER ONE

A CROSS-LINGUISTIC ACCOUNT OF CONTROL,RAISING AND EXCEPTIONAL CASE MARKING

This chapter presents a contrastive analysis of infinitival clauses in Control, Raising and Exceptional Case Marking structures in English and Spanish. It focuses on a number of properties that have been studied in the Generative literature, from Standard Theory (Chomsky 1965; Chomsky 1973; Chomsky and Lasnik 1977) to the Theory of Principles and Parameters (Chomsky 1981; Chomsky 1982) and the Minimalist Program (Chomsky 1995, 2001). These properties concern the distribution and interpretation of infinitival clauses, that is, (a) the syntactic positions in which they appear in Control, Raising and Exceptional Case Marking structures, and (b) the choice of antecedent for the infinitival clause in these structures.

1.1. Control and Raising Structures in Generative Syntax

To quote Davies and Dubinsky, "(...). Raising and Control are among a handful of syntactic phenomena (including anaphora and question formation) which have been central concerns of generative syntax since the 1960s (...)." (vii)1. As explained previously, these syntactic phenomena are manifested in (but are not limited to) complex structures where the main verb is complemented by an embedded infinitival clause, as shown in (8) and (9).

(8a) John tried [to understand the article] (8b) John persuaded the doctor [to examine Mary]

(9a) John seemed [to understand the article] (9b) John believed the doctor [to have examined Mary]

Chapter One 8

On the surface, the constructions in (8a) and (9a) are practically identical: both contain an intransitive main clause followed by an infinitival clause. Likewise, the constructions in (8b) and (9b) both contain a transitive main clause followed by an infinitival clause. The difference is that the main subject (John) in (9a) is semantically linked only to the embedded verb (understand), whereas in (8a) it is semantically linked to both the main verb (try) and the embedded verb. In (9b), the main object (the doctor) is linked only to the embedded verb (examine), whereas in (8b) it is linked to both the main verb (persuade) and the embedded verb.

In the Generative literature, the constructions in (8) are referred to as Control structures (Subject Control in the case of [8a], Object Control in the case of [8b])2. The constructions of (9) are referred to as Raising structures (Raising to Subject in [9a], Raising to Object in [9b]). Whether a given construction is Control or Raising depends on the lexical properties of the main verb. These constructions are attested in many languages other than English, among them Spanish as shown in (10) and (11).

(10a) Mara decidi [llamar a Juan] "Mary decided [to call John]"

(10b) Mara oblig a Juan a [tomar una decisin] "Mary forced John [to make a decision]"

(11) Mara parece [estar contenta] "Mary seems [to be happy]"

Both the infinitive and the gerund are non-finite forms that can combine with finite verbs, but only the infinitive can appear in complex or bi-clausal structures. Gerunds appear in so-called periphrastic constructions, that is, structures where two verbs constitute a single head of the verbal predicate (Gmez Torrego 1999)3. This means that periphrastic constructions are by definition mono-clausal, similarly to simple structures with a single verb. Since the topic of this research is the acquisition of complex structures, constructions with non-finite forms such as gerunds and participles, or infinitives that appear in mono-clausal structures (e.g., modal structures such as you must go to school) will not be discussed further in this volume.

The constructions in (8) to (11) differ in several ways, which have been used to make the case for distinguishing between Control and Raising. From a semantic point of view, Control and Raising structures are

Cross-linguistic Account of Control, Raising and Exceptional Case Marking 9

distinguished by their thematic structures and selectional restrictions. In Control structures, an argument in the main clause is semantically linked to both the main verb and the embedded verb, as mentioned above. In the case of intransitive main clauses, the argument is the main subject. In the case of transitive main clauses, the main object4. In terms of thematic structure, the argument in the main clause receives a thematic role from both the main verb and the embedded verb (John is the agent of try and the experiencer of understand in [8a]; the doctor is the agent of examine and the theme of persuade in [9a]). In contrast, the main argument is semantically linked only to the embedded verb in Raising structures. Verbs such as seem do not assign thematic roles (John is the experiencer of understand, but receives no other thematic role in [8b]). Verbs such as believe assign two thematic roles, agent to its subject and theme to the clausal complement (the doctor is the agent of examine in [9b]), but no thematic role to the main object. Main verbs that appear in Control structures typically select animate subjects, since they assign the role of agent (# the article tried to be interesting is therefore semantically odd). This selectional restriction does not apply in the case of Raising structures, as main verbs that appear in these constructions assign no thematic role (the article seems interesting)5.

From a syntactic point of view, Control and Raising structures show certain differences that are language specific, such as the use of the embedded passive or pleonastic subjects in English6, the use of partitive clitics (en) in French (Ruwet 1991), or clitic climbing in Spanish (Lujn 1980)7. Other differences, which are related to the distribution and interpretation of the infinitival clause are common to English and Spanish.

1.1.1. Control and Raising Structures in the Theory of Principles and Parameters

Together with gerunds and participles, infinitives are characteristically non-inflected (or non-finite) verbal forms (Chomsky 1970). In Spanish, an infinitive consists of a verbal root plus a thematic vowel (a or e) and an infinitival affix r. Within the Government and Binding framework (Chomsky 1981), the non-finite verbal affix r is analysed as the functional head of the inflection node in the derivation. The infinitival particle to in English is similarly analysed as a non-finite verbal form marker belonging to inflection8. The main difference is that Modern English has no bound infinitival morphology, as opposed to Spanish9.

"Case Theory is the module of the grammar concerned with the distribution of NPs. (...)" within the Principles and Parameters framework

Chapter One 10

(Haegeman 1994, 193). It establishes the Case filter, which "(...) imposes a requirement on the licensing of NPs (...)" (Ibid., 193). The Case filter states that all phonetically realized (or overt) nominal elements must be assigned abstract case10. Noun Phrases (NPs) receive case under government or by specifier-head agreement. According to Chomsky (1981), functional categories such as Tense and Agreement, which are present in finite inflection (e.g., inflected verbal affixes in Spanish, auxiliaries in English) license overt NP subjects by assigning them nominative case. Non-finite inflection lacks Tense and Agreement, meaning that infinitives cannot assign nominative case and therefore do not license overt subjects (*John be tired of commuting every day).

At the same time, the Extended Projection Principle, according to which all clauses must have subjects applies to both inflected (or finite) clauses and non-finite clauses. Therefore, lexically and phonetically null categories have been posited in the Generative literature as the subject of the infinitival clause, particularly after the introduction of Standard Theory (Chomsky 1973), as shown in (12) and (13).

(12) Johni tried [S'[SPROi to understand the article]]

(13) John seemed [Sh to understand the article]

The category of the null subject is different in Control and Raising structures. In Raising structures such as (13), the NP John is generated in the embedded clause. Since the embedded verb is an infinitive, it cannot assign abstract case and therefore does not license overt subjects. In order to comply with the Case filter, the NP is forced to raise to the main clause, where it is assigned nominative case by the main verb. In addition to the Case filter, derivations have to comply with the Thematic Criterion11, which links thematic roles with syntactic positions. The NP receives a thematic role (agent) from the embedded verb. Since verbs such as seem do not assign thematic roles, this derivation does not violate the Thematic Criterion. The NP leaves behind a trace (h) in the subject position of the embedded clause, thus complying with the Extended Projection Principle.

In Control structures such as (12), the NP has to be generated in the main clause. If it were to raise from the embedded clause, it would receive a thematic role from both the main verb and the embedded verb (agent), violating the Thematic Criterion. At the same time, this argument is semantically linked to both verbs. In order to reflect this semantic relationship and comply with the Thematic Criterion, a null pronominal (PRO) is posited as the subject of the infinitival clause. PRO fulfills the


Case filter (it does not require abstract case to be licensed in that position), the Extended Projection Principle, and the Thematic Criterion by receiving the thematic role from the embedded verb. The NP in the main clause is the controller or antecedent of PRO in the embedded clause, and thus the link is established in syntactic terms.

In constructions such as (14), the object NP in the main clause was initially proposed to raise from the embedded clause12, similarly to the subject NP in (13). However, if the NP were to raise from the embedded clause, it would receive a thematic role from both the main verb (theme) and the embedded verb (agent), violating the Thematic Criterion. At the same time, the Case filter does not allow the presence of an overt NP in the subject position of the embedded clause.

(14) *John believed the doctor [Sh to have examined Mary]

Chomsky (1981) suggests that the NP remains in the embedded clause, but at the same time is assigned case by the main verb, as shown in (15). Since main verbs do not usually assign case to the subject of the embedded clause, these constructions are known as Exceptional Case Marking structures.

(15) John believed [the doctor to have examined Mary]

In English, some verbs (such as want, prefer, like, among others) may appear in structures with intransitive main clauses such as (16), or structures with transitive main clauses such as (17). With certain main verbs, the main object is preceded by the preposition for (e.g., the company would prefer for him to retire early). These constructions are called Optional Control structures. This refers to the ability of these main verbs to alternate in two different syntactic configurations (San Martn 2004), a Control configuration in the case of (16), and a non-Control configuration in the case of (17).

(16) John wants to retire early.

(17) The company wants him to retire early.

Chomsky (1981) proposes that the embedded clause contains a phonologically empty complementizer that can optionally license the overt NP (him) by assigning it accusative case. Since the embedded clause is non-finite, it is the prepositional complementizer for. This prepositional

Chapter One 12

complementizer is eliminated in the Phonological Form, giving rise to the configuration in (18)13. The rule allowing the elimination of the prepositional complementizer applies only to postverbal positions and certain complements (for the children to run would be dangerous / *the children to run would be dangerous).

(18) The company wants [S' [Shim to retire early]]

In summary, Control and Raising structures contain a null category in the subject position of the embedded infinitival clause. In contrast, Exceptional Case Marking and Optional Control structures contain overt NPs, licensed either by the main verb or a complementizer.

Chomsky (1981) further distinguishes Control from Raising according to clause structure. As shown in (19) as opposed to (20), the infinitival subject is introduced by a null complementizer in Control structures, but not in Raising structures.

(19) Mary decided [CP[C[TPPRO[Tto] call John]]]

(20) [TPMary [T[VP[Vseems][TP[Tto][VPh[Vbe] happy]]]]

Stowell (1982) notes that infinitival clauses in these configurations show interpretative differences relating to tense and mood. More specifically, certain infinitival clauses may be understood as being unrealized with respect to the main clause. In Control structures such as (19), the event denoted by the infinitival clause (call) is not interpreted as necessarily taking place at the same time as the event denoted by the main clause (decide). In these cases, the tense/mood interpretation of the embedded clause would be similar to modal elements such as should or would in English, or the subjunctive in Spanish. In Raising structures such as (20), both events are interpreted as necessarily taking place at the same time (seem and be happy). Infinitival clauses in Control structures are thus said to denote irrealis mood, and infinitival clauses in Raising structuresare said to denote realis mood. Stowell (1982) affirms that these interpretative differences are not only related to the semantic content of the main verb, but also to the syntactic representation of infinitival clauses in these configurations. According to this author, infinitival clauses with C(omplementizer)-positions (i.e., Control structures) contain a tense operator, which results in the irrealis interpretation of the infinitival clause. Infinitival clauses with no C-positions (i.e., Raising structures) lack


this tense operator, which results in the realis interpretation of the infinitival clause.

Regarding the interpretation of infinitival clauses, separate modules are needed in the Principles and Parameters framework to account for null infinitival subjects in Control and Raising structures. Traces in Raising structures are c-commanded by the NP antecedent in the main clause, and are therefore anaphoric in nature. Their interpretation is regulated by Binding Theory. PRO is anaphoric in the sense it is obligatorily bound by its antecedent, but it is also pronominal in the sense that it is bound by an antecedent outside the embedded clause. Control Theory regulates the interpretation of PRO. PRO has special properties: for instance, it may have an arbitrary reference if the embedded clause is not governed by the main verb (to err is human). As explained earlier, its reference can be determined either by the subject or the object of the main clause. If the main clause contains more than one possible antecedent, it is usually the closest NP to the infinitival clause, in accordance with Rosenbaum's (1968) Minimal Distance Principle14.

The interpretation of infinitival clauses with null subjects is at the heart of the debate between semantic and syntactic accounts of Control and Raising. Generally speaking, theories formulated within the Generative framework rely on purely syntactic operations to account for the distribution and interpretation of infinitival clauses in these configurations. These theories usually follow or adapt the Minimal Distance Principle (Hornstein 1999, 2001, 2003; Landau 2000, 2003; Manzini 1983; Manzini and Roussou 2000, among others), and only accept semantic accounts for highly exceptional cases. In contrast, other theories propose a bigger role for semantics, particularly the meaning of the main verb (Culicover and Jackendoff 2001; Jackendoff 1972; Sag and Pollard 1991). According to authors such as Culicover and Jackendoff (2001), the Minimal Distance Principle does not fully account for the interpretation of infinitival clauses, particularly with verbs such as promise as shown in (21).

(21) Johni promised the doctor [PROi to examine Mary]

The main clause contains two possible NP antecedents for PRO. The closest antecedent in terms of the Minimal Distance Principle is the main object (the doctor). However, the main subject (John) is interpreted as the antecedent here. In contrast, in constructions with verbs such as persuade, which also contain two possible NP antecedents in the main clause, the main object is interpreted as the antecedent of PRO (i.e., John persuaded the doctori [PROi to examine Mary]).

Chapter One 14

In order to account for the interpretative properties of infinitival clauses with promise-type verbs from a syntactic perspective, Larson (1991) analyses promise as a double object verb and persuade as a single object verb in Deep Structure, as shown in (22) and (23).

(22) [VP[NPJohn[V[Vpromised[VP[NPMary[Vto come[V[Ve[NPe]]]]]]]]]]

(23) [VP[NPJohn[V[Vpersuaded[VP[NPMary[Vto come[Ve]]]]]]]]

In promise-type constructions, the main object (Mary) is an indirect object that raises from a VP internal position in the embedded clause. In Deep Structure, the closest NP antecedent in Minimal Distance terms is the main subject (John) (i.e., John promised to come to Mary would be the Deep Structure equivalent of John promised Mary to come). In persuade-type constructions, the main object is generated directly in the main clause. It is the closest NP antecedent in both Deep and Surface Structure. Larson (1991) concludes that Control structures with promise-type verbs are structurally more complex than Control structures with other transitive verbs. He supports his analysis by pointing out that promise-type verbs can also appear in small clauses with double objects, as opposed to persuade-type verbs (e.g., Mary promised John the book vs. *Mary persuaded John the book).

In summary, the null category in the subject position of the embedded clause is different in Control and Raising structures (PRO vs. an NP trace) and its interpretation is also regulated by a different module of the grammar: Control Theory in the case of PRO, Binding Theory in the case of the NP trace. Control and Raising are not only distinguished semantically by their thematic structures and selectional restrictions, but also syntactically by their clause structure and the category of the subject of the infinitival clause.

The adoption of the Minimalist Program will challenge many of the differences proposed within the Principles and Parameters framework. Some analyses go as far as attempting to provide a single account for Control and Raising (Hornstein 1999, 2001). Other analyses continue providing different accounts (Landau 2000, 2003).

1.1.2. Control and Raising Structures in the Minimalist Program

According to the syntactic analyses formulated within the framework of Generative syntax, and particularly the Theory of Principles and


Parameters, infinitival clauses in Control, Raising and Exceptional Case Marking structures complement the main verb. This verb may assign one thematic role (as in decide), two thematic roles (as in force), or no thematic roles (as in seem). Control and Raising structures are proposed to contain null infinitival subjects; Exceptional Case Marking structures contain overt infinitival subjects. In Control structures, the infinitival subject is the null pronominal PRO; in Raising structures, the infinitival subject is an NP trace.

The Minimalist Program (Chomsky 1995, 2001) brings about some changes to the analysis of Control, Raising and Exceptional Case Marking. Even though it adopts the basic ideas of the Theory of Principles and Parameters with regard to linguistic knowledge and the model of language acquisition (Eguren and Fernndez Soriano 2004)15, it conceptualizes the properties of linguistic systems rather differently. Crucially, the Minimalist Program eliminates the distinction between representational levels (Deep and Surface Structure), and the distinction between lexical and syntactic information (assuming that structurally relevant features have lexical content, that is, that they are interpretable). It also aims to develop some ideas on the economy of derivations16 and the economy of representations17, which have the effect of reducing the number and length of syntactic operations to three essential ones: Merge, Agree and Move.

The proposals on the economy of derivations and representations are particularly relevant for the syntactic analysis of Control, Raising and Exceptional Case Marking structures. Null complementizers, which were posited within the Theory of Principles and Parameters to prevent PRO from being governed by the main verb, have no place in the context of the Minimalist Program. Chomsky and Lasnik (1993) propose the Null Case Hypothesis, according to which non-finite (infinitival) inflection checks (null) structural case on PRO via specifier-head agreement. They assume that infinitival inflection can only check null case, and that PRO is the only element that can receive null case. As shown in (24), PRO moves from a VP-internal position to the specifier of T in order to have its null case features checked by infinitival inflection on T, that is, by a phrasal head with matching case features.

(24) John tried[TPPROi(Spec)[T[Tto understandi[VPti[V[Vthe article]]]]]

The Null Case Hypothesis does not appear to distinguish between Control and Raising structures, as it gives no reason why infinitival inflection should not check null case in Raising structures as well. In order to solve this problem, Martin (1992, 1996, 2001) proposes that infinitival

Chapter One 16

inflection can only check null case in syntactic configurations where the infinitival clause denotes irrealis (i.e., Control structures) on the basis of Stowell's (1982) observations on tense/modal differences in infinitival clauses. Therefore, null case can only be checked in Control structures, and PRO is the only element that is licensed as infinitival subject in these configurations.

Bokovi (1997) also adopts the Null Case Hypothesis, pointing out that the type of complement (CP or TP) depends on the selectional properties of the main verb. Unless there are lexical or structural factors that require a C-position in the derivation (such as an overt complementizer), clauses without a C-position (TP) as shown in (25) and (26) are the default option in accordance with Minimal Structure Principle18.

(25a) John tried [TPPRO[Tto][VP[Vunderstand] the article]]

(25b) John persuaded the doctor [TPPRO[Tto][VP[Vexamine] Mary]]

(26) [TPJohni[T[VP[Vseems][TP[Tto][VPti[Vunderstand] the article]]]]

In Control structures such as (25a) and (25b), PRO checks null case in the specifier of TP, as opposed to Raising structures such as (26). Here, there is no null case for PRO to check and therefore PRO cannot be the infinitival subject. The NP moves from a VP-internal position to the specifier of T, where it can receive structural case from the main verb, leaving a trace behind.

This analysis has been adopted with some modifications in more recent versions of the Minimalist Program. More specifically, it is accepted that infinitival affixes or markers have abstract tense properties, that is, unspecified Tense values that must be determined from the context provided by the main clause. As explained in Radford (2004), if the Tense value of the infinitival clause is independent of that of the main clause, null case can be valued on a goal with matching phi-features (PRO), as shown in (27)19. If this is not the case, infinitival inflection cannot value null case on PRO, and therefore PRO cannot be the infinitival subject.

(27) John tried [TPPRO[Tto understand the article]] [3-pers] [3-pers]

[Sg-num] [Sg-num] [Null Case] [Ts-nf] [EPP]


Some authors such as Wurmbrand (2001) do not accept that infinitival inflection has Tense features. This author claims that infinitival clauses may receive an irrealis interpretation if they involve a future modal (similar to woll in German), or realis if they lack the future modal. As shown here, the infinitival clause in (28a) is not future-oriented, and therefore the construction in (28b) is ungrammatical, as opposed to (29).

(28a) John is trying to call Paul. [-Tense, -Modal] (28b) *John is trying to call Paul later.

(29a) John decided to call Paul. [-Tense, +Modal] (29b) John decided to call Paul later.

Wurmbrand (2001) suggests that complements in constructions such as (28a) have been restructured20 in the sense that they involve truncated infinitival clauses (i.e., they lack TP or CP layers). As opposed to infinitival clauses in constructions such as (29a), they denote bare events or actions, and therefore lack illocutionary force or thematic properties. They do not license PRO as infinitival subject and thus do not give rise to Control, as shown in (30).

(30) [TPJohn[vP[VP[Vis trying[VPto call Paul]]]]]

Wurmbrand's (2001) proposal implies that some verbs such as try, which have been classified as Control verbs in previous studies, would instead be more akin to periphrastic verbs21, which typically give rise to single clause sentences.

Independently of the proposed syntactic analysis, all these authors accept that Control and Raising are different. For theoretical as well as empirical reasons, Generative analyses have traditionally distinguished Control from Raising, from Rosenbaum (1968) and Postal (1974) to the Minimalist Program. For instance, according to Standard Theory, Control and Raising structures should have different representations in Deep Structure because they do not take the same number of arguments. This is based on the assumption that there is a link between structural differences and empirical distinctions such as thematic structure or selectional restrictions. As explained earlier, within the Theory of Principles and Parameters and particularly the Government and Binding framework, Chomsky (1981) invokes the Thematic Criterion, which links thematic roles with syntactic positions. Null subjects in Control structures such as John tried [PRO to understand the article] in (25a) are posited because an

Chapter One 18

argument is needed to receive the thematic role from the embedded verb (in this case, Agent), in accordance with the Thematic Criterion. An alternate analysis involving Raising, that is, movement of the main object out of the embedded clauses as shown in (31), is excluded in Control structures by the Thematic Criterion.

(31) *Johni tried [Sti to understand the article]

In this construction, the argument John receives a thematic role from the embedded verb. If it moved to the main subject position, it would receive another thematic role from the main verb. The Thematic Criterion states that an argument can receive one and only one thematic role. Therefore, the analysis in (31) is incorrect for Control structures.

Hornstein (1999, 2001) proposes an analysis that accounts for both (Obligatory) Control22 and Raising, motivated basically by theoretical reasons. According to this author, the distinction between Control and Raising structures in the frameworks before the Minimalist Program "(...) relies on the thematic requirements d-structure places on derivations (...)" (Hornstein 2003, 11)23. He argues that the elimination of Deep Structure as a level of representation makes this distinction meaningless within the Minimalist framework. In addition, he argues that a single analysis for Control and Raising structures is conceptually attractive, since it (a) eliminates the restrictions imposed by the Thematic Criterion24, and (b) eliminates the need to posit different modules (such as Control Theory25) to account for the distribution and interpretation of the infinitival clause in these configurations. More specifically, Hornstein (1999, 2001) proposes that Control structures are derived by movement, similarly to Raising structures. He assumes that movement between thematic positions is possible, and that thematic roles function as features that trigger movement between thematic positions. At the same time, Control and Raising structures are distinguished by their thematic properties. According to Hornstein (2003) and Boeckx and Hornstein (2004), the argument in Raising structures would involve a single chain with its copies and therefore have only one thematic role, as shown in (32), whereas the argument in Control structures would involve multiple chains and at least two thematic roles, as shown in (33).

(32) [TPMaryi [TPseems to [VP be tired]]]]

(33) [TPMaryi [VP tried to [VP call John]]]]

(cf. Boeckx and Hornstein 2004, 432)


The subject of the infinitival clause (Mary) moves from a VP-internal position to the specifier of TP to check case. In Raising structures, it moves to a non-thematic position in the main clause; in Control structures, it moves to a thematic position in the main clause via another thematic position. As a result, the infinitival subject in both configurations is a DP-trace, the residue of movement at Logical Form which is coindexed with the antecedent DP in the main clause.

Since PRO has been excluded as the subject of the infinitival clause in Control structures, Hornstein (1999, 2001) accounts for the interpretation of the infinitival clause by reformulating the Minimal Distance Principle in terms of the Minimal Link Condition26. According to this modified Minimal Distance Principle, derivations such as the one shown in (34) would not be possible.

(34) *Maryi persuaded Peter [ei to call John] [TP2Mary [T[VP3Mary v+persuaded [VP2Peter persuaded [TP1Mary [VP1Mary to call John]]]]]]

(cf. Hornstein 2001, 45)

If Mary were the antecedent of the infinitival clause, it would have to merge with call to check its thematic role (Agent) and move to the specifier of VP3 to check the thematic role of persuade (Agent). It would have to cross Peter on the specifier of VP2, thus violating the Minimal Link Condition. Therefore, the main subject could not be interpreted as the antecedent of the infinitival clause in this configuration.

In contrast, Landau (2000, 2003) assumes that (a) the syntactic category of the infinitival subject in Control structures is distinct from that of the infinitival subject in Raising structures, and (b) the interpretation of infinitival clauses in these configurations is regulated by different components of the grammar. Landau's (2000, 2003) analysis shares many similarities with other proposals formulated within the Generative framework that take the so-called "standard view" of Control (e.g., Chomsky 1981, 1995, 2001; Manzini 1983). Its main contribution is the fine-tuning of the interpretation of infinitival clauses in Control structures. More specifically, not only does it distinguish between Obligatory and Non-Obligatory Control27, it also subdivides Obligatory Control into two types: Exhaustive Control and Partial Control. In the former, the subject of the infinitival clause (PRO) must be identical to the antecedent or controller, as shown in (35). In the latter, PRO must include the controller, but the reference of PRO need not be exhausted by the reference of the controller. As shown in (36), a Partial Control interpretation can be observed when the speaker has a group in mind, where it is understood

Chapter One 20

that PRO refers to the chair and the person or persons he or she wishes to meet at six.

(35) *John1 managed [PRO1+ to meet at 6].

(36) The chair1 wanted [PRO1+ to meet at 6]. (Landau 2000, 27)

According to Landau (2000, 2003), Exhaustive Control and Partial Control are distinguished by the presence of tense contrasts between the event denoted by the infinitival clause and the event denoted by the main clause. With Partial Control, the event denoted by the infinitival clause does not necessarily occur in the same time frame as the event denoted by the main clause. With Exhaustive Control, tense contrasts are not allowed between the events denoted by the infinitival clause and the main clause. Main verbs that give rise to Exhaustive Control tend to belong to certain semantic classes (in English): implicative (e.g., manage), aspectual (e.g., begin) or modal (e.g., need). Main verbs that can give rise to Partial Control tend to be factive (e.g., regret), propositional (e.g., claim), desiderative (e.g., want) or interrogative (e.g., ask). It is important to note that verbs that receive a Partial Control reading can also receive an Exhaustive Control reading, but not vice versa. Nevertheless, the Partial Control reading can only be established unambiguously when a singular controller co-occurs with an embedded collective plural as in (36). Landau (2000, 2003) concludes that a movement analysis such as Hornstein's (1999, 2001) for Obligatory Control does not account for the Exhaustive and Partial Control distinction. On his part, Hornstein (2003) assumes that infinitival clauses in Obligatory Control constructions denote irrealis and contends that a movement approach can include Partial Control if it is treated "(...) as a result of an optionally applicable meaning postulate licensed by certain matrix verbs when taking Control complements (...) (namely) non-finite [+Tense] (...)" (42). This meaning postulate would be a lexical property tied to specific lexical predicates and would not interact with the syntax.

Culicover and Jackendoff (2001) and Jackendoff and Culicover (2003) also argue against a single analysis for Control and Raising, but not for the reasons mentioned above. They accept that Control and Raising configurations may well have the same syntactic structure, and at the same time that Control verbs are distinct for Raising verbs by virtue of the thematic roles they assign to the main subject. Their main criticism to Hornstein's (1999, 2001) analysis is that the interpretation of infinitival


clauses cannot be accounted for by purely syntactic operations such as the Minimal Distance Principle. As mentioned previously, this is apparent with main verbs such as promise (as in Johni promised Mary [ei to leave]), where the antecedent of the infinitival clause is the main subject instead of the closest argument, that is, the object of the main clause. These authors account for the choice of antecedent in terms of the thematic relations of all the possible antecedents, thus favouring a semantic account for the interpretation of infinitival clauses in Control and Raising structures.

In response to Culicover and Jackendoff (2001) and Jackendoff and Culicover (2003), Boeckx and Hornstein (2004) argue for a syntactic account of the distribution of the infinitival subject, although they concede that semantics might play a role in the interpretation of infinitival clauses. With regard to the criticism of the Minimal Distance Principle, Boeckx and Hornstein (2004) contend that verbs such as promise are highly marked from a syntactic point of view. This is based on acquisition studies such as Carol Chomsky (1969), who shows that children acquire the interpretative properties of constructions with promise later than those of constructions with persuade (such as John persuaded Mary to leave, where the main object is the antecedent of the infinitival clause), or Courtenay (1998), who observes that some adults give the same interpretation to constructions with promise and constructions with persuade (i.e., they choose the main object as the antecedent of the infinitival clause in both constructions) or they consider constructions with promise to be ungrammatical.

1.1.3. Exceptional Case Marking in the Minimalist Program

The changes brought upon by the Minimalist Program have been especially profound with regard to Exceptional Case Marking. It was first proposed in Chomsky (1981) for theoretical reasons, specifically to account for apparent violations of the Thematic Criterion in sentences such as John believed the doctor to have examined Mary. Raising to Object, that is, movement of the infinitival subject (the doctor) to the main object position would mean that the argument would receive more than one thematic role. Since this would constitute a violation of the Thematic Criterion, Chomsky (1981) proposes that the object remains in situ and receives accusative case from the main verb, at the same time remaining the subject of the infinitival clause as explained previously.

The Minimalist Program moves away from Exceptional Case Marking and brings back Raising to Object. It proposes two non-thematic positions (Spec, AgrS and Spec, AgrO) to host the infinitival subject, and thus there

Chapter One 22

would be no violation of the Thematic Criterion. Authors such as Lasnik and Saito (1991) argue that the infinitival subject in constructions such as John believed the doctor to have examined Mary appears in a higher position in the syntactic representation than the subject of the finite embedded verb in semantically equivalent constructions such as John believed that the doctor had examined Mary. This proposal is based on empirical evidence from English presented in Rosenbaum (1968), Bach (1977) and Postal (1974), indicating that this argument might be the object of the main clause and at the same time the subject of the infinitival clause in this type of construction. Rosenbaum (1968) and Bach (1977) observe that existential there can appear in the subject position of the infinitival clause with main verbs such as believe (I believe there to be children in the park), but not with Control verbs such as force (*I forced there to be children in the park), which provides evidence in favour of subjecthood. Postal (1974) observes that a relationship between infinitival subject and main subject can be established by means of passivization (John is believed to be missing by Peter), reflexivization (John believes himself to be very clever) and reciprocal formation (They believe each other to be beautiful), which provides evidence in favour of objecthood. Lasnik and Saito (1991) conclude that the infinitival subject is in the main object position in the Logical Form28.

With regard to constructions with the verb want, Postal (1974) analyses them as Raising to Object structures. In contrast, Lasnik and Saito (1991) adopt Chomsky's (1981) analysis modified in terms of Chomsky (1989). As mentioned previously, the complementizer (or its phonologically empty equivalent) assigns case instead of the main verb. Therefore, the argument need not move to the specifier of AgrO to receive case. Bokovi (1997) agrees with this analysis and states that the infinitival subject is licensed by a phonologically empty complementizer equivalent to for in constructions such as (37).

(37) I wanted [CP[C][TPJohn/him [Tto][VP[Vsee] a specialist]]]

The complementizer assigns accusative case to the DP in the specifier of TP before moving to C (*he wanted he see a specialist is thus ungrammatical). In fact, these constructions are very similar to for-to infinitival structures (I will arrange for him to see a specialist), with the difference that the complementizer is overt in this type of constructions29.

Raising to Object analyses after Lasnik and Saito (1991) focus on specific characteristics of Non-indoeuropean languages (e.g., Japanese, Algonquin languages), such as finite complements in Raising to Object


structures. Kuno (1976) proposes that the infinitival subject raises to the main clause in these structures. Authors such as Bruening (2001) suggests that there are two possible derivations for Raising to Object structures in these languages. In one case, the NP moves to the specifier of the complementizer, leaving behind a trace. In the other case, the NP is generated in the specifier of the complementizer and it moves to an argument position in the main clause. The trace is coindexed with a null pronominal in the embedded clause. Other authors such as Hiraiwa (2001) propose that the NP can remain in situ, that is, Raising to Object is optional. Tanaka (2002) suggests that the possibility of having argument movement from a finite clause depends on whether the specifier of the complementizer is in an argument position or not.

In summary, the Minimalist Program has contributed to reformulating Raising to Object, initially proposed in Standard Theory. Similarly to Hornstein's (1999, 2001) proposal, this reformulation is attractive in the sense it eliminates the restrictions imposed by the Thematic Criterion and the need for an operation that applies to a restricted class of verbs. Nevertheless, none of the analyses described above establish the exact position of the argument in Raising to Object: the evidence appears to indicate that it could be localized either in the main clause or the embedded clause.

For the purposes of the learnability hypothesis, the syntactic category of the null infinitival subject and the syntactic representation of Control and Raising structures is not revelant. There is nothing to indicate that Spanish and English show any contrasts in these respects. In the following sections, it will become evident that some of the major cross-linguistic differences between Spanish and English concern constructions with overt infinitival subjects, such as the Raising to Object structures and structures with the verb want described above.

1.2. Raising and Control Structures in Spanish and English

Syntactic theory establishes that infinitival clauses show different distributional and interpretative properties according to the configuration in which they appear. There are a number of main verbs which can select infinitival clauses, giving rise to different syntactic configurations. The type of syntactic configuration depends on the number of thematic roles assigned by the main verb. Within the Minimalist framework, main verbs that do not assign any thematic roles generate Raising (or Raising to Subject) structures (e.g., seem). Main verbs that assign one or more

Chapter One 24

thematic roles generate Control structures (Subject Control structures in the case of verbs such as decide, Object Control structures in the case of verbs such as force), Raising to Object structures (e.g., believe) and structures with a prepositional complementizer known as Optional Control structures (e.g., want). Infinitival clauses in these configurations are distinguished by their tense/modal properties and the syntactic category of the infinitival subject. In Control and Optional Control structures, infinitival clauses denote irrealis. In Raising to Object structures, infinitival clauses denote realis. The infinitival subject is phonologically empty in both Control and Raising structures. In Standard analyses, the syntactic category of the null infinitival subject is different in Control and Raising structures. In Movement analyses, it is a trace in Raising structures and certain types of Control structures. In Optional Control and Raising to Object structures, the infinitival subject is an overt Determiner Phrase (DP).

In order to study the acquisition of these configurations, it is important to look at the contrasts between Spanish and English. The assumption is that learnability and language transfer issues are related to cross-linguistic differences.

1.2.1. Optional Control Structures

In English, infinitival clauses in Optional Control structures may appear in Control configurations such as (38a), or in configurations with a prepositional complementizer as in (38b).

(38a) I wanted [TP[Tto][VP[Vget] tickets for the show]]

(38b) I wanted [CP[C][TPJohn [Tto][VP[Vget] tickets for the show]]]

As mentioned earlier, want is not the only verb that can appear with a prepositional complementizer, but it is one of the few which can also give rise to Control structures. Other such verbs are prefer, like and expect, as shown in (39).

(39a) I would prefer [to get tickets for the early show] (39b) I would prefer for you [to get tickets for the early show] (39c) I would like [to get tickets for the show] (39d) I would like you [to get tickets for the show] (39e) I expect [to have the tickets for the show by tomorrow] (39f) I expect you [to have the tickets for the show by tomorrow]


It is important to note that these verbs do not constitute a homogenous class. For instance, the verb like can appear with a gerund (as in I like watching tv), as opposed to the other verbs (*I want watching tv). The verb expect may appear in passive constructions (John was expected to go on his own), similarly to Raising to Object verbs such as believe, but not the other verbs (*John was preferred to go on his own).

The possibility of giving rise to Control structures or structures with a prepositional complementizer is not attested in Spanish, neither for the translation equivalents of verbs such as want, as shown in (40), nor for other verbs.

(40a) Quiero ver a un especialista. "I want to see a specialist."

(40b) *Quiero Juan/l ver a un especialista. "I want John/him to see a specialist."

In Spanish complex structures, if the main subject is the antecedent of the embedded clause as in (40a), the verb appears with an infinitival clause, similarly to English. However, if the main subject is not the antecedent as in (40b), the embedded clause obligatorily contains a finite verb (in this case, in the subjunctive mood), as shown in (41).

(41) Quiero que Juan vea a un especialista.

"I want that John sees a specialist."

Unlike English, this alternation between infinitival clauses and finite clauses in contexts where the embedded clause denotes irrealis is possible in Spanish because of the existence of subjunctive mood marking on finite verbs30. In fact, this is also attested with main verbs that assign two thematic roles, as shown in (42). These verbs may give rise to an Object Control structure as in (42a), or a complex structure with a finite embedded clause as in (42b).

(42a) Mara oblig a Juan a tomar una decisin. "Mary forced John to make a decision."

(42b) Mara oblig a Juan a que tomara una decisin. "Mary forced John that he make a decision."

Optional Control structures are not possible in Spanish because it lacks prepositional complementizers that can assign structural case, such as for

Chapter One 26

in English. As observed in Fernndez Lagunilla (1987), Spanish prepositions only assign inherent case. Non-finite inflection cannot assign structural case either, therefore finite inflection is needed to license the overt DP as infinitival subject.

The majority of the prepositions in English assign inherent case only. In fact, the structural case-assigning properties of for are the result of diachronic change. Miller (2002) observes that in Old English, the infinitival marker to started appearing with the case-assigning preposition to in dative + infinitive constructions with overt infinitival subjects. As shown in (43), the ambiguity between the infinitival marker and preposition triggered the appearance of the for to infinitive31.

(43) It is good to us for to be here. (WBibl EV; Miller 2002, 204)

In Middle English, as morphological dative case was lost, the preposition for was reanalysed as a case assigner for overt infinitival subjects. The preposition for eventually replaced the preposition to, as shown in (44), giving rise to for-to infinitival constructions in Modern English. Thus preposition for came to be a complementizer assigning structural case (i.e., accusative) to overt infinitival subjects.

(44) It is easier for a camel to pass (through a needles eye). (Miller 2002, 205)

In summary, English and Spanish complementizer systems differ in one important respect. In addition to complementizers that appear with finite clauses (that, if and whether, according to Castillo [2001]), English has prepositional complementizers which license overt infinitival subjects by assigning structural case. In Spanish, complementizers only appear with finite clauses32 and cannot license overt infinitival subjects. This is due to the different case-assigning properties of prepositions in these languages. Another important difference is the existence of finite inflection indicating the subjunctive mood in Spanish, as opposed to English. This means that there are two available options in Spanish when the embedded clause denotes irrealis (infinitival clauses or finite clauses with the verb in the subjunctive). As shown above in Control structures such as (42), finite inflection sometimes replaces infinitival inflection in translation equivalents from English. Therefore, infinitival clauses can be used in a wider variety of syntactic contexts in English compared to Spanish.


1.2.2. Raising to Object Structures

As mentioned earlier, Exceptional Case Marking is abandoned in favour of Raising to Object in the Minimalist Program. In Raising to Object structures, the argument is both the infinitival subject and the object of the main verb. In English, opinion verbs (such as believe and consider) give rise to this configuration, as shown in (45).

(45) I believe [TPJohni[Tto][VPti[Vbe] intelligent]]

In contrast with Optional Control structures, Raising to Object structures lack a complementizer. The VP-internal argument moves to the specifier of TP to receive case. In this type of constructions, the main verb can also appear with finite clauses, as shown in (46).

(46) I believe that John is intelligent.

The translation equivalents of believe and consider in Spanish are ungrammatical in configurations with infinitival clauses (*Creo a Juan [ser muy inteligente]), but not with finite clauses with the verb in the indicative mood (Creo que [Juan es muy inteligente]). However, Hernanz (1999) observes that there are certain infinitival clauses where the infinitival subject is realized as the object of the main verb33. Main verbs that give rise to these configurations tend to be causative verbs (hacer "make", dejar "let", mandar "command") or perception verbs (ver "see", or "hear", escuchar "listen", sentir "feel", observar "observe", mirar "look"), as shown in (47) and (48)34.

(47) Mara hizo regar las plantas a la asistenta. "Mary made the maid water the plants."

(48) Mara vio regar las plantas a la asistenta. "Mary saw the maid watering the plants."

According to Hernanz (1999), these constructions has been analysed either as periphrastic (restructuring) structures or as complex structures. In the latter case, there are two possible analyses: (a) the object and the infinitive constitute a unit, similarly to the subject and the verb in finite embedded clauses, or (b) the object and the main verb constitute a unit, in contrast with the infinitive. Hernanz (1999) observes that causative and perception verbs do not behave like Control verbs that assign two thematic

Chapter One 28

roles (such as obligar "force", ordenar "order", prohibir "forbid", invitar "invite"...) in the sense they do not allow two complements at the same time, as shown by the contrast between (49) and (50)-(51)35.

(49a) Mara orden a la asistenta [regar las plantas] "Mary ordered the maid [water the plants]" (49b) Mara orden a la asistenta [que regara las plantas]

"Mary ordered the maid [that she water the plants]"

(50a) Mara hizo a la asistenta [regar las plantas]"Mary made the maid [water the plants]"

(50b) Mara hizo a la asistenta [*que regara las plantas] "Mary made the maid [that she water the plants]"

(51a) Mara vio a la asistenta [regar las plantas] "Mary saw the maid [water the plants]"

(51b) Mara vio a la asistenta [*que regara las plantas] "Mary saw the maid [that she watered the plants]"

Hernanz (1999) proposes that, unlike these Control verbs, causative and perception verbs select a single argument as a syntactic object. Therefore, the object and the infinitive constitute a unit in these structures, as shown in (52).

(52) [O...[V{Vio/Hizo}] [Olas plantas regar]] "[O...[V{Saw/Made}] [Othe plants water]]"

(cf. Hernanz 1999, 2238)

To explain the fact that the infinitival subject is functioning as the main object, Hernanz (1999) proposes that the boundaries between the verb and its complement have been weakened by means of restructuring, as shown in (53). This process is linked to the lexical properties of causative and perception verbs in Spanish, and therefore it is rather exceptional.

(53a) [S...[V Vio] las plantas regar...] "[S...[VSaw] the plants water...]"

(53b) [S...[VP Hizo regar] las plantas...] "[S...[VP Made water] the plants...]

(cf. Hernanz 1999, 2257)


As shown in (53a), the clause boundary between main verb and infinitive disappears in perception structures. The infinitival subject may then receive accusative case from the main verb. In (53b), there is an additional process by which the main verb and the infinitive merge into a single unit in causative structures, similarly to the components of a periphrastic construction36.

These analyses are based on the differences between the properties of causative and perception verbs, particularly with regard to reflexivization and object pronominalization, as shown in (54) and (55).

(54a) La cantante hizo [que la vistieran]. "The singer made that they dress her."

(54b) La cantante se hizo vestir. "The singer made herself dress."

(55a) La cantante vio [que la vestan]. "The singer saw that they were dressing her."

(55b) *La cantante se vio vestir. "The singer saw herself dress."

In the causative structure of (54b), the reflexive pronoun se receives same reading as the object clitic pronoun la in (54a). Here, the causative verb appears with a finite clause instead of an infinitival clause37. In contrast, the pronoun se in the perception structure of (55b) is not equivalent to the object clitic of (55a). In fact, the construction of (55b) is behaving similarly to Control structures, which have clause boundaries between finite and infinitival clause: La cantante orden que la vistieran"The singer ordered that they dress her" vs. *La cantante se orden vestir"The singer ordered herself dress". In contrast, the construction of (54b) behaves like a simple sentence, indicating that there are no clause boundaries between finite and infinitival clauses

Main objects are pronominalized with indirect object clitics in the case of causative verbs, and with direct object clitics in the case of perception verbs, as shown in (56) and (57).

(56) Mara le hizo regar las plantas. (le = la asistenta) "Mary made her water the plants." ("her=the maid")

(57) Mara la vio regar las plantas. (la = la asistenta) "Mary saw her water the plants." ("her=the maid")

Chapter One 30

The verb hacer functions as an auxiliary in (56), forming a single unit with the infinitive regar "water". This verbal unit is ditransitive: it takes a direct object (las plantas) and an indirect object (le or la asistenta). In contrast, the verb ver takes a direct object (la or la asistenta) and the infinitive also takes a direct object (las plantas).

In summary, empirical evidence seems to indicate that causative structures show many properties of periphrastic constructions (the verb dejar "let" and mandar "command" follow the syntactic pattern of the causative verb hacer, although they are similar in meaning to Control verbs such as permitir "allow" and ordenar "order"). Therefore, causative structures will not be discussed further. Constructions with perception verbs contain an argument that is both the infinitival subject and main object, similarly to Raising to Object structures in English. They differ in that there is no movement of the argument to receive case in Spanish. Therefore, constructions with perception verbs in Spanish are not Raising to Object structures per se. Nonetheless, Hernanz's (1999) analysis of these constructions is reminiscent of Exceptional Case Marking as proposed in the framework of the Theory of Principles and Parameters. It would appear then that there is no Raising to Object in Spanish, but there is some form of Exceptional Case Marking. From now on, the term Exceptional Case Marking structures will be used to refer to constructions with perception verbs in Spanish.

The ungrammaticality of Raising to Object structures in some languages but not others has received different explanations. From a diachronic perspective, Lightfoot (1991) observes that constructions with verbs such as believe and consider (e.g., I believe John to be intelligent)represent innovations which begin to be attested in Middle English38. He analyses the emergence of these constructions as a result of the infinitival marker becoming able to transmit the structural case marking properties of the governing verb (i.e., accusative case), which he associates with the loss of dative case in English. Since Spanish makes a distinction between accusative and dative case, Raising to Object structures are not possible in this language.

Kayne (1981) associates the possibility of having these constructions with the case-assigning properties of prepositions. He proposes that verbs such as believe or consider appear in the configuration shown in (58a)39.

(58a) I believe [ [John to be intelligent]] (58b) *Je crois [ [Jean tre intelligent]] (58c) *Creo [ [a Juan ser inteligente]]


Here, the infinitival clause is headed by a null complementizer (already suggested in Chomsky and Lasnik [1977] for Control structures). Kayne (1981) proposes that the null complementizer can assign structural case, similarly to the prepositional complementizer for. In contrast, in languages such as French (or Spanish), the null complementizer cannot assign structural case as prepositions can only assign inherent case. Therefore, constructions such as (58b) and (58c) are ungrammatical in these languages.

In the Minimalist framework, analyses such as Kayne's (1989) have been discarded in favour of Raising to Object. In fact, his analysis would imply postulating the existence of a complementizer, even though there are no lexical or structural properties that require it. This would violate principles of economy of representation such as the Minimal Structure Principle. However, this kind of analysis still applies in the case of Optional Control structures, establishing a link between the case-assigning properties of prepositions and the possibility of Optional Control as discussed earlier. In turn, the case-assigning properties of prepositions contribute to the differences between the complementizer systems in Spanish and English.

1.2.3. Other differences between Spanish and English

As mentioned previously, Control verbs that assign one thematic role are transitive (decide) and generate Control structures. In these constructions, the antecedent of the infinitival clause is the subject of the main clause (Subject Control). Control verbs that assign two thematic roles are ditransitive (force) and also generate Control structures. In these constructions, the antecedent of the infinitival clause is the object of the main clause (Object Control). In Spanish, animate and human non-pronominal objects are marked obligatorily with the preposition a. Depending on the thematic properties of the main verb, the object may receive accusative case as in (59), or dative case as in (60).

(59) Mara oblig a Susana a manejar el coche. "Mary forced Susan to drive the car."

(60) Mara prohibi a Susana manejar el coche. "Mary forbid Susan to drive the car."

In English, the object is morphologically unmarked. It can be replaced with a full pronoun (him/her/it), which appears in the same position as the object DP (to the right of the main verb). In Spanish, the object is replaced

Chapter One 32

with a clitic pronoun (lo/la/le): a direct object clitic as in (61), or an indirect object clitic as in (62). Clitics appear to the left of the main verb.

(61) Mara la oblig a manejar el coche. "Mary forced her to drive the car."

(62) Mara le prohibi manejar el coche. "Mary forbid her to drive the car."

In Exceptional Case Marking structures, the object is replaced with a direct object clitic as in (63).

(63) Mara la oy cantar en La Scala. (la=a la soprano) "Mary heard her sing in La Scala. (her=the soprano)"

This distinction between Spanish and English is relevant for the interpretation of the infinitival clause in Control, Raising and Exceptional Case Marking structures. From a semantic perspective, the sentences in (61)-(63) and their translation equivalents in English are the same: the main object is the antecedent of the infinitival clause in both cases. However, they are completely different from a syntactic perspective. In Spanish, the object pronoun is a clitic. In English, it is a full pronoun. If semantic information is the only factor in the interpretation of the infinitival clause, the syntactic category of the antecedent should not make a difference. If syntactic information is also a factor, clitic pronouns might generate difficulties for L2 Spanish learners whose native language has no clitic pronouns, such as English.

1.3. Summary

This chapter presents a contrastive analysis of infinitival clauses in Control, Raising and Exceptional Case Marking structures. This analysis reveals three crucial differences between Spanish and English: (a) case-assigning properties of prepositions (structural case/inherent case) (b) the availability of inflected forms that denote irrealis (subjunctive mood) in Spanish, and (c) the category of object pronouns (full pronouns/clitic pronouns). The first contrast concerns the grammaticality of Optional Control. The second contrast might have an effect on the usage of infinitival clauses when the embedded clause denotes irrealis. The third contrast concerns the grammaticality of Raising to Object, and might influence the choice of antecedent for the infinitival clause. These cross-


linguistic contrasts between Control, Raising and Exceptional Case Marking structures will be discussed in terms of learnability and language transfer in the following chapter.

Notes 1 For a comprehensive review of the different analyses of Control and Raising developed within Chomskian tradition, as well as other generative alternatives, see Davies and Dubinsky (2004). 2 The main subject or object "(...) is said to "control" the reference of the subject of the embedded clause (...)" (Davies and Dubinsky 2004, 3). Hence the term "Control". 3 Verbal periphrasis is defined by Gmez Torrego as "(...) la unin de dos o ms verbos que constituyen un slo 'ncleo' del predicado (...)" (Gmez Torrego 1999, 325). 4 The verb promise does not follow the pattern of verbs such as persuade. It gives rise to Control structures with a transitive main clause. However, the argument that is semantically linked to both the main verb and the embedded verb is the main subject instead of the main object: in John promised Mary to arrive early, it is John who is the agent of arrive, not Mary. As will be explained later, this asymmetry between promise and verbs such as persuade has fuelled the debate between syntactic and semantic accounts of the thematic properties of these verbs. It was also the main topic in the earliest studies on L1 acquisition of Control. 5 In English, there are several verbs (begin, threaten, allow, find, permit) that may appear in either Control or Raising configurations (Perlmutter 1970, Postal 1974), although there is a slight change of meaning: Mary began to work (on her project) vs. The printer began to work (after the toner was changed). These verbs do not impose selectional restrictions (i.e., they optionally assign the role of agent, and therefore may take either animate or inanimate subjects). 6 Rosenbaum (1968) notes that a Raising structure with a passive complement is synonymous with its active counterpart, as shown here: John seems to have understood the article = The article seems to have been understood by John. This is not so in Control structures, in those cases where the embedded passive is possible: The doctor tried to examine Mary Mary tried to be examined by the doctor. Raising structures can take pleonastic subjects such as it or existential there: there seems to be a problem. This is not possible in Control structures: *there tried to understand the article. These properties are linked to the fact that Control verbs assign thematic roles, whereas Raising verbs do not. More specifically, John is the agent in the active and passive counterparts of the Raising structure because it does not receive a thematic role from seem. As to the Control structures, the doctor is the agent in the active and Mary in the passive because tryassigns a thematic role. Pleonastic elements are semantically empty, so they are not assigned a thematic role. Therefore, they are licensed as the subject of Raising verbs, but not of Control verbs.

Chapter One 34

7 According to Kayne (1989), clitic climbing is derived via head movement. The clitic pronoun moves to the inflection node, and subsequently to the complementizer node. From this clause-initial position, it can cliticize to the main verb. He also observes that clitic climbing is not possible if the embedded clause is finite: *Mara lo quiere que Juan compre "Mary it wants that John buys". He attributes this to the presence of an overt complementizer. On the other hand, Zagona (2002) observes that Control structures allow clitic climbing: Mara decidi llamarlo "Mary decided to call him" / Mara lo decidi llamar "Mary him decided to call". In contrast, Raising structures do not allow clitic climbing in spite of the absence of an overt complementizer: Mara parece estarlo "Mary seems to be it" / *Mara lo parece estar "Mary it seems to be". According to Lujn (1980), clitic promotion (i.e., clitic climbing) is related to tense/mood specification when the embedded clause is non-finite. Only infinitival clauses denoting irrealis allow clitic climbing. As opposed to infinitival clauses in Raising structures, infinitival clauses in Control structures denote irrealis. For a more detailed account of the tense/mood properties of infinitival clauses, see Stowell (1982). 8 Infinitival to and prepositional to are distinguished by several properties (Radford 2004). Most importantly, infinitival to requires strictly verbal complements with bare forms (I managed to arrive on time / *I managed to arrival on time). In contrast, prepositional to can take other types of complements, such as nouns (Yesterday afternoon I went to the library). In fact, true prepositions in English only allow verbal complements with the gerund (Mary called Susan before leaving for work / *before to leave for work) and never bare forms (I want to leave right now / *I want to leaving right now). 9 As described in Miller (2002), Old English had the inflectional affix an (also called the plain infinitive) and the dative declined infinitive enne with proclitic to, as well as an undeclined infinitive with to. 10 The Extended Case Filter states that "*[NP ] if has no Case and contains a phonetic matrix or is a variable." (Chomsky 1981, 175). There are two types of abstract case: structural (nominative and accusative) and inherent. 11 "Each argument bears one and only one -role, and each -role is assigned to one and only one argument." (Chomsky 1981, 36). 12 Raising to Object disappears from the picture after Standard Theory (Chomsky 1973). In Chomsky and Lasnik (1977), Raising to Object structures do not show any movement. The Raising to Object operation returns in Lasnik and Saito (1991), reinterpreted within the Minimalist framework. 13 Chomsky and Lasnik (1977) analyze constructions with the verb want similarly to constructions with believe. The difference is that the embedded clause contains the prepositional complementizer for: John wanted [S'for [NPthe doctor] to examine Mary]]. 14 According to Rosenbaum (1968), infinitival clauses in Control structures select the "closest" NP as their antecedent. "Closest" is defined in terms of the number of nodes in between the embedded clause and the NP: Johni persuaded the [NPdoctorj] [Sto examinej Mary]. In Principles and Parameters, distance is defined

[Rocio Perez Tattam] Second Language Competence T(BookZZ.org)

Documents

Transcript of [Rocio Perez Tattam] Second Language Competence T(BookZZ.org)