The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek...

43
The ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen [email protected]

Transcript of The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek...

Page 1: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

The ambiguity of wh-expressions

First year report

Radek Šimík CLCG

Department of Linguistics University of Groningen

[email protected]

Page 2: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

1

Contents Contents............................................................................................................................................1 1 Introduction ..............................................................................................................................2 2 The point of departure: movement, binding, and lexicalization ..............................................2

2.1 Wh-movement and locality...............................................................................................2 2.2 Between movement and unselective binding ...................................................................4 2.3 A systematic approach to lexical ambiguity ....................................................................6 2.4 Conclusion........................................................................................................................8

3 Wh-phrases: What and where are they? ...................................................................................9 3.1 Wh-phrases with no operator properties.........................................................................10

3.1.1 Bare wh-variables ...................................................................................................10 3.1.2 Wh-phrases in quantifier-derivation .......................................................................19 3.1.3 Summary ................................................................................................................24

3.2 Wh-phrases as quantifiers? .............................................................................................25 3.3 Wh-phrases as specific DPs............................................................................................27 3.4 Wh-phrases and maximalizing semantics.......................................................................29 3.5 Conclusion......................................................................................................................29

4 Further related issues..............................................................................................................29 4.1 Multi-step wh-movement ...............................................................................................30

4.1.1 Cases of extraction out of finite CPs......................................................................30 4.1.2 Cases without extraction ........................................................................................30

4.2 Complementizers as invariant pronominals ...................................................................33 5 Conclusion..............................................................................................................................34 Appendix A Activities first year .................................................................................................36

A.1 Attended schools and conferences .................................................................................36 A.2 Talks ...............................................................................................................................36 A.3 Papers .............................................................................................................................37 A.4 Other academic activities ...............................................................................................37 A.5 Reading and orientation in the relevant literature ..........................................................37

Appendix B Plans for the coming years .....................................................................................38 B.1 Rough chronological and topical schedule.....................................................................38 B.2 Schools ...........................................................................................................................38 B.3 Talks and papers.............................................................................................................39 B.4 A semester abroad ..........................................................................................................39

References ......................................................................................................................................40

Page 3: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

2

1 Introduction Many languages possess a class of wh-expressions. Such a class is typically defined in morphological terms, i.e. there is a morpheme (possibly realized by a small number of allomorphs, e.g. wh- and h- in English), which has the potential to combine with a category-specific stem (mostly nominal, adjectival, or adverbial). The result is a wh-expression, often called a "question word". Indeed, being a question word, or more technically a "question operator", can be regarded as the core function of wh-expressions. It is not the only one, however. Wh-expressions serve to express a whole range of functions, even within one language. Thus, the English wh-words are also used as relative operators and in some dialects also comparative operators. Other languages extend the usage to existential quantifiers (German and Dutch), or even complementizers (Romance and some Slavic languages). It is this ambiguity potential that lies at the heart of my project. Below, I present a number of properties of wh-expressions in various syntactic contexts and pose many empirical and theoretical questions. Some well-known as well as novel issues are raised in the discussion. The project proposal is organized as follows. Section 2 sets the point of departure: it characterizes the syntax of A-bar dependencies and its most important aspects, namely locality (constraints), wh-movement, and unselective binding. The last subsection presents a framework for dealing with lexical ambiguity, in rather non-technical terms. Section 3 is the main body of this proposal. It contains a lot of well-known as well as novel data and brings in a number of interesting problems. I take an empirically determined perspective and examine the ambiguous behavior of wh-expressions in a range of constructions. Naturally, no definitive conclusion is reached but many interesting questions are raised and some tentative proposals are made. Section 4 discusses two further issues: the multi-step nature of wh-movement and the pronominal nature of complementizer-like heads. Some more theoretically appealing questions are raised. Section 5 concludes the project proposal, summarizes the main hypotheses and suggests some ways in which they could be tested or further elaborated. 2 The point of departure: movement, binding, and lexicalization In this section, I set the general theoretical and methodological background for my proposal. First, I introduce the notions of A-bar movement, unselective binding, and A-bar dependencies. After that, I outline the basic methodology for dealing with lexical ambiguity. 2.1 Wh-movement and locality One of the defining properties of all natural languages is the so-called displacement property—parts of sentence structures may be pronounced somewhere else than they are (semantically) interpreted. (1) a. I like to read historical novels

b. [What kind of novels]1 do you like to read t1? This tension between the place of interpretation and the place of pronunciation is recognized as soon as we make the theoretical assumption that the semantic properties of sentences which we traditionally refer to as theta-roles are computed in a strictly local manner, namely as local (sister) relations of predicates (verbs) and their arguments (nominal phrases). This assumption is

Page 4: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

3

sometimes referred to as uniform theta assignment hypothesis (UTAH) (Baker 1988). I stick to the convention here and adopt this hypothesis as an axiom.1 In the example above, what kind of novels is unambiguously interpreted as the thing that I read. Translated to a theory which uses UTAH, there is an occurrence of the wh-phrase in the sister position of the verb read. In (1b), this local relation is captured by the gray t, traditionally referring to a trace. The trace shares an index with the understood object, which is a means of capturing the non-local relation. It has been assumed that non-local relations of the kind above are unbounded (in syntax), as for the length of the intervening structure (in short path, Abels 2002), see (2a). However, not every kind of path can intervene between the position of interpretation and pronunciation, as shown in the examples below. (2) a. What kind of novels1 does Peter think that Mary said that you like to read t1?

b. * What kind of novels1 did you enjoy the time when you could read t1? c. * What kind of novels1 did you get angry because Mary kept reading t1? The conditions on paths are more or less well-known, at least on the empirical level. Forty years ago, Ross (1967) translated the core empirical findings into the first comprehensive theory of syntactic islands; the example in (2b) is an instance of complex NP (relative clause) island, (2c) is an adverbial clause island. Numerous efforts have been made since then in order to find a unifying theory of locality (conditions on paths). One of the most promising lines of thought exploits the notion of relativized minimality (Rizzi 1989)—a condition which prohibits a movement of an element of class � whenever the path contains another element of the same class (a precise formulation makes use of the c-command relation). A most obvious example involves a wh-island, where � is [wh] feature: (3) *Who[wh]

1 did you ask where/whether[wh] Mary met t1 Currently, the dominant view is that a principle of relativized minimality must be supplemented by a theory of phases2, i.e. syntactic heads which render their complements as inaccessible for operations triggered by any higher (c-commanding) head. Abels (2002) proposes that a phase is a head which (potentially) contains all syntactic features and therefore functions as a universal intervener. In line with Chomsky (1999), the only elements which can escape the complement of a phase are the ones that move to its specifier. The most influential line of theoretical thinking about the long-distance dependencies of the kind illustrated above started with Chomsky (1977), who introduced the notion wh-movement (later also A-bar movement, i.e. movement to A-bar positions, i.e. not case- or theta-positions). Chomsky proposed that wh-movement is involved in the derivation of root and embedded questions, comparatives, topicalization, clefts, pseudoclefts, and relative clauses: (4) a. Who1 did John see t1? [root wh-question]

b. I asked who1 John saw t1 [embedded wh-question]

1 It is good to keep in mind that the theory of syntax would look very different if the axiom was abandoned. This has recently been proposed by Neeleman and van de Koot (2006). 2 However, see the theory of Starke (2001), who aims to subsume all locality under a relativized minimality principle, in his approach translated into a general restriction on the basic structure-building operation Merge.

Page 5: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

4

c. John is taller than (what1) Mary is t1 [comparative] d. This book1, I really like t1 [topicalization] e. It is this book Op1 that I really like t1 [cleft] f. This book is what1 I asked Bill to read t1 [pseudocleft] g. I found a book (which1) you can read t1 [relative clause] All these constructions, despite some superficial complications (e.g. a wh-word does not appear everywhere), share the property of leaving a gap in the position of interpretation and are subject to the same locality restrictions. It is important to note that wh-movement not only affects the place of pronunciation. It also radically changes the interpretation of the moved element and virtually the whole sentence. More specifically, the position in the left periphery of (English) sentences seems to be responsible for assigning discourse (or information structure) properties; for example, it forces a phrase of a sentence to be interpreted in contrast to other elements of the same (contextually determined) class (4d), shows which part of the sentence is unknown to the speaker (4a), or unambiguously determines anaphoric relations across clauses (4g). This view leads us to what we could call universal discourse-role assignment hypothesis (UDAH), possibly a natural counterpart of UTAH. A radical version of this hypothesis (although not called this name) recently became a part of the cartographic approach to syntax. Starting basically with Rizzi (1997), the cartographic program makes the assumption that not all wh-/operator-phrases in (4) move into the same position of the CP-domain (often referred to as left periphery). A rich functional domain may look as follows:3 (5) ForceP > TopP* > FocP > (TopP*) > FinP > IP (Rizzi 1997: p297) A single C-head is replaced by a sequence of five functional projections (and potentially more, since * shows that the domain can be extended). The highest one (ForceP) is argued to be involved in the coding of the external relation of the clause (in cases of embedding) and in so-called clause-typing (declarative, interrogative, imperative, etc.); the lowest one (FinP) is involved in the finiteness distinction (complementizers in finite and non-finite clauses may differ). The domain in between is reserved for information-structure tampering. Thus, topics (e.g. (4d)) are placed in the specifier of Top and foci (analogues of (4e/f)) are placed in the specifier of Foc. It has further been proposed that wh-phrases in questions sit in SpecFocP, whereas relative operators are placed in SpecTopP. In other words, interrogative operators are foci and relative operators are topics. It is useful to note that this syntax-based approach is not a priori incompatible with intonational accounts of information structure. To give an example, it is plausible to analyze (contrastively) focalized phrases in situ as being in a distant relation with the Foc head. This long-distance relation (which can be technically captured e.g. by the operation Agree, cf. Chomsky 2000) can be interpreted as high pitch at PF. 2.2 Between movement and unselective binding

3 I put the second Top* in brackets because the subsequent literature, Rizzi’s as well as other authors’ puts its existence into doubt. At different points in the discussion below, I will also assume the hierarchy Top > Foc.

Page 6: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

5

Chomsky characterized A-bar movement in terms of (i) leaving a gap in the argument (trace) position and (ii) locality conditions. However, some languages create analogous semantics without conforming to the observation (i). A classical example of such a language is Chinese (since Huang 1982). On the surface of things, Chomsky’s claim concerning A-bar movement remains correct because Chinese fails to conform to the locality conditions (ii), too:4 (6) strong island (relative clause)

Akiu kan-bu-qi [DP [CP Op1/i [IP t1 zuo shenme]] de reni]? Akiu look-not-up do what PNM person ‘What is the thing/job x such that Akiu despises [people [who do x]]?’

(Tsai 1994: p13) (7) weak island (whether-clause)

ni xiang-zhidao [CP Akiu mai-bu-mai shenme] (ne)? you want-know Akiu buy-not-buy what Q a. ‘What is the thing x such that you wonder whether Akiu will buy x?’ b. # ‘Do you wonder what Akiu will buy?’

(Tsai 1994: p52) It seems inappropriate to use the term movement for languages that exhibit neither of the properties suggested by Chomsky. Therefore, I will adopt a more general and traditional term, subsuming both movement and non-movement long-distance dependencies: A-bar dependency. The Chinese kind of A-bar dependency has often been analyzed in terms of unselective binding. In questions like those above, a question operator is base-generated in the left periphery and binds a variable within its scope (this variable normally takes the form of an indefinite NP). Binding is a process which is not sensitive to any locality restrictions. Note that the Chinese state of affairs casts serious doubts on the uniformity of discourse-properties assignment (UDAH), simply because the phrase in question shenme ‘what’ does not occupy a position in the left (or right) periphery of the sentence. Nevertheless, a weaker version of UDAH is still in play if it can be shown that what causes shenme ‘what’ to be interpreted as a question word, actually dwells in the left periphery. According to Cheng (1991) and Tsai (1994), this is indeed the case: there is a question (Q-)quantifier in the C/SpecCP, which can even be phonetically realized, as in (7) above. This brings us to a very rough cross-linguistic typology of discourse-property assignment: (8) a. [CP QP1 (C) [IP I [VP V <QP1>]]] [English-type, movement] b. [CP Qi (C) [IP I [VP V NPi]]] [Chinese-type, binding] It is no surprise that such a clear cut will not take us far. To begin with, not all Chinese wh-phrases function in the same way as shenme ‘what’. Tsai (1994) argues that only those A-bar dependencies which involve an individual variable (or referential variable) have the license to obviate locality restrictions and shows that adverbial phrases like ‘why’ (with causal reading) or ‘how’ (with manner reading) cannot appear in islands:

4 In the following examples, I adopt Tsai’s (1994) analysis, mainly for the purposes of clarity. Only sometimes, I omit irrelevant details. The distinction between number-indexing for movement and letter-indexing for binding/coreference is mine and will be used throughout the report.

Page 7: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

6

(9) * Akiu xihuan [DP [CP Op1/i [IP Luxun weishenme xie t1]] de shui]?

Akiu like Luxun why write PNM book ‘What is the reason x such that Akiu likes [books [that Luxun wrote for x]]?’

(Tsai 1994: p13) At least from the view of locality conditions, a subset of Chinese wh-phrases behave as though they were moving. Japanese exhibits mixed strategies even to a greater extent. A-bar dependencies in Japanese may obviate only strong islands, while being sensitive to weak (wh-) ones: (10) strong island (relative clause)

John-wa [[[t1 dare-o aisiteiru IP] Op1/i CP] onna-oi DP] nagutta no? John-Top who-Acc loves woman-Acc hit Q

‘Who is the person x such that John hit the woman who loves x?’ (11) weak island (whether-clause)

?? John-wa [ Mary-ga nani-o katta ka-dooka CP] Tom-ni tazuneta no John-Top Mary-Nom what-Acc bought whether Tom-Dat asked Q

‘What is the thing x such that John asked Tom whether Mary bought x?’ (Tsai 1994: p50) Thus, we have to admit that at least Chinese and Japanese mix the two strategies in (8) in an interesting and non-trivial way. It has been argued that the parameter that is at stake here is the structural placement of question operators (Cheng 1991, Tsai 1994, Hagstrom 1998). The semantic interface (or UDAH) requires the positioning of this operator at the structural periphery of the matrix clause. A language like Chinese base-generates the operator directly at the periphery but for some reason, this operator cannot bind adjunct wh-phrases. English generates it together with the question word and movement is required. Japanese generates it in such a way that it has to move but not so far as the English one (arguably in SpecDP, i.e. at the edge of strong nominal islands), leading to the mixed pattern. Let me conclude this subsection with a remark on the development of unselective binding in the linguistic theory. It was introduced for the treatment of indefinite NPs by Lewis (1975) and was later popularized by Kamp (1981) and Heim (1982). As far as I know, Pesetsky (1987) was the first one to have made use of unselective binding for the purposes of (multiple) wh-questions (in English). Recently, Kratzer (2005) has suggested a more general application of this mechanism, arguing for the existence of existential concord (�negative concord), drawing evidence from German indeterminate phrases. Finding the crux of the distinction between unselective binding and movement is a hot issue in the present research and I hope that my thesis will contribute something in this respect, too (see mainly section 3.1.1). 2.3 A systematic approach to lexical ambiguity As I mention in the introduction, languages which possess a morphological class of wh-words typically do not use these for a single function. The examples in (4) show a range of functions of the English wh-words: they function e.g. as question, relative, and comparative operators, and

Page 8: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

7

more detailed distinctions can be further drawn (e.g. root vs. embedded questions, restrictive vs. appositive relative operators). I already suggested that the different semantic functions of wh-operators can be formally expressed by different positioning within the syntactic tree (FocP, TopP, ForceP, etc.). It remains to be asked why wh-words express the functions that they express. This is a question of cross-linguistic relevance, since not every language with wh-words uses them for the same range of functions. German is a case in point for the comparison with English. In colloquial German, wh-words are excluded from the relative operator functions. On the other hand, they exhibit the behavior of existential quantifiers, something not found in English. Importantly, these issues have to be studied carefully as it proves to be true that fine distributional considerations matter, especially in the borderline cases just mentioned (relative or existential operators). Thus, the English relative operators can be expressed by a wh-word only on the condition that the target of relativization has a specific reading (see section 3.3); the German existential quantifiers, on the other hand, can only be expressed by a wh-word only if they are not focalized or topicalized (see section 3.1.1.1).5 In my project I would like to follow the hypothesis that lexical ambiguity of the above type is systematic and predictable to a certain extent. In particular, there is a hierarchy of functions (i.e. functional heads) such that more than one function can be spelled-out by the same exponent. The spell-out function conforms to the (syntactic) hierarchy in the following way: if F3 gets spelled-out as x and F1 gets spelled out as x then F2 also spells-out as x. The possible and impossible spell-outs are listed below: (12) F1 > F2 > F3 (or [F1 [F2 [F3]]])

a. x � F1; x � F2; x � F3 b. x � F1; y � F2; z � F3

c. x � F1; x � F2; y � F3 d. y � F1; x � F2; x � F3 e. * x � F1; y � F2; x � F3 All structurally related functions may share the same spell-out (the same lexical item) as in (12a), or they may all be different as in (12b). The syncretism may also be partial but then it must respect the hierarchy in the way suggested above, i.e. (12c/d) is all right, as opposed to (12e).6 Consider the following tentative application of the pattern in (12) to the empirical issue discussed here. The hierarchy proposed in (13) is supported by the findings made within the cartographical program: relative operators are associated with Top (relatively highest in the hierarchy), question operators with Foc (which is lower than Top), whereas existential operators seem to be independent of the left periphery altogether (thus being structurally low). Hence, (13) plausibly correlates with TopP > FocP > IP.7

5 Dutch expresses relative operators by wh-words only if they are headed by a preposition. 6 A spell-out restriction of this kind has been referred to as the ABA restriction and has been used for explaining verbal morphology paradigms in English and nominal morphology paradigms in Czech (see Caha 2007 for examples and references). 7 Furthermore, the position of the wh-phrase within the hierarchy may correlate with the structural make-up of the wh-morpheme itself (e.g. in German, (13a), wh � [Q [∃]] ∨ [∃]). To dispense with the apparent redundancy in notation (the label of the position in which the wh-phrase occurs necessarily matches the label of the wh-phrase), it can be argued that the wh-phrase itself is responsible for marking the position; cf. the general proposal of Starke (2001) to treat specifiers as complex heads, i.e. projecting/labeling elements.

Page 9: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

8

(13) relative operator > question operator > existential operator

a. x � relative; wh � question; wh � existential / C [German] b. wh � relative / C; wh � question; x � existential [English] c. * wh � relative; x � question; wh � existential [an impossible language] C in (13) refers to a set of (arguably syntactic) conditions which are associated with the spell-out rules: e.g. English relative operators are only wh- if also specific; German existential operators are wh- if not topic/focus. These conditions should eventually be dispensed with in favor of a systematic explanation, formulated in terms of a more fine-grained hierarchical structure.8 The system/hierarchy predicts that there are no languages which make use of wh-morphology for expressing existential and relative operators to the exclusion of question operators. It remains to be answered, however, why the question operator function of wh-words appears to be the basic (or even defining) one, in other words, why there is no language (known to me) that has a robust morphological class, which is reserved for quantification/determination and at the same time displays ambiguity to the extent witnessed with wh-words. 2.4 Conclusion In this section, I have set the general theoretical background for my project. In order for the project to be successful, the following questions have to be answered in a satisfactory way: (i) What does it mean for a construction to involve an A-bar dependency? (ii) What is the difference between A-bar movement and unselective (A-bar) binding? (iii) What does it mean to be a wh-expression? The answer to question (i) may be found in examining the syntax-semantics interface: A-bar dependencies are defined by a natural class of operator-variable relations where the operators involved correspond to discourse properties of sentences and constituents. Operator-variable relations are standardly treated as one-to-one relations (and one-to-many relations in cases of unselective binding). Towards the end of this project proposal (see section 4.1), we will see that the standard approach may prove insufficient in the light of some data. In particular, I will tentatively propose that operators may be syntactically (and hence semantically) complex expressions. Under this view, A-bar dependencies would involve a many-to-one (or even many-to-many) relations. The answer to question (ii) was claimed to lie the position where operators are base-generated. Languages may base-generate operators either together with their variables or separated from them. In the former case, the operator phrase has to undergo a semantically-driven movement to the appropriate position of interpretation. In the latter case, operators are generated in the position where they get interpreted and unselectively (and at a distance) bind their variables. There are three empirical distinctions that correspond to the theoretical/analytical distinction proposed.

8 A state of affairs which is comparable to German is reported for some wh-in situ languages, e.g. Chinese, in that indefinite NPs can be bound by all sorts of quantifiers (universal, existential, question) but are ruled out from being bound by relative operators (Gillian Ramchand, p.c.).

Page 10: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

9

First, both processes of creating operator-variable dependencies are subject to locality conditions of their own: the movement of low base-generated operators observe well-known locality conditions; unselective binding, on the other hand, involves a restriction on variables, in that they have to stay in the scope of corresponding operators. Second, movement targets specific positions in the tree whereas unselective binding, in principle, renders any position that is in the scope of the operator (and which is not ruled out for independent reasons) as a possible occurrence of the bound variable. This predicts strict positioning in cases of movement but rather liberal positioning in case of unselective binding. Third, while movement seems to be able to license the (interrogative) force of any kind of wh-phrase, there appear to be some restrictions on the type of wh-phrase that is to be unselectively bound (Chinese adjuncts). Note that this last distinction does not follow from anything that has been said so far and even though there have been attempts to provide explanations, an adequate solution is still to be found. Below, I will propose a preliminary analysis of non-operator wh-phrases in German and Czech as unselectively bound variables (see section 3.1). I will illustrate that their distribution is more liberal than the distribution of interrogative (or relative) wh-words, but still restricted. It will also be shown that there are some relevant restrictions on the type of wh-phrase, some that have hardly be recognized so far. In the first step to answering the question (iii), I suggested that wh-morphemes potentially spell-out a of number syntactic structures (lexical items) which are related in a systematic and predictable way (subsets). Some spell-outs, even though in principle possible, will take place only under certain conditions. A few cases of such conditions will be discussed below (see e.g. 3.1.1.2). The discussion related to questions (ii) and (iii) will often bring us out of the domain of A-bar dependencies (as they are usually defined), since wh-expressions which are semantic variables are often bound by logical (existential, universal) operators, which have no intrinsic discourse properties. Nevertheless, the relevance of such a discussion in the context of examining the ambiguity of wh-expressions is self-evident. 3 Wh-phrases: What and where are they? My project consists in a detailed investigation of operator-like phrases (mainly with wh-morphology) in Slavic and Germanic languages. I would like to approach them from two different perspectives, which may eventually prove to be two sides of the same coin: (i) their internal morpho-syntactic makeup and (ii) their distribution within the functional domain of sentences. This section, which forms the main body of my project proposal, is organized as follows. Section 3.1 discusses wh-phrases which seem to function as variables, i.e. have no quantificational power. I deal with the syntactic and semantic behavior of bare wh-phrases in English, German, and Czech. I also make some interesting observations concerning Czech, English, German, and Dutch wh-phrases as morphological building blocks in quantifier-derivation. In section 3.2 it is shown that wh-phrases in root questions function as quantifiers. Section 3.3 is devoted to a brief discussion of wh-phrases in English relative clauses, which apparently behave as specific DPs. Section 3.4 is concerned with the relation of wh-phrases/constructions with maximalization semantics. Section 3.5 concludes the discussion.

Page 11: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

10

3.1 Wh-phrases with no operator properties In this section I will show that some wh-phrases may function as semantic variables, i.e. elements with no operator or quantificational properties. I will show that there are some appealing restrictions on non-operator wh-phrases which call for explanation. 3.1.1 Bare wh-variables I assume that bare, non-operator wh-expression, as exemplified below, are essentially non-quantificational indefinites, which are (unselectively) bound by c-commanding quantifiers, as proposed by Kamp (1981) and Heim (1982).9 3.1.1.1 German wh-existentials10 I start by discussing some interesting restrictions on the distribution of German wh-words with the behavior of indefinite pronominals.11 Consider the case of German was ‘what’, which can mean ‘something’ in certain configurations: (14) a. … dass Peter in diesem Geschäft was kaufen wollte

that Peter in this store what buy wanted ‘… that Peter wanted to buy something in this store’

b. … dass Peter was in diesem Geschäft kaufen wollte that Peter what in this store buy wanted ‘… that Peter wanted to buy something in this store’ c. * … dass Peter WAS in diesem Geschäft kaufen wollte ( aber ich weiß nicht mehr was) that Peter what in this store buy wanted but I know no more what ‘… that Peter wanted to buy SOMETHING in this store (but I don’t know what)’ d. * Was wollte Peter in diesem Geschäft kaufen [, und etwas nicht] what wanted Peter in this store buy and something not ‘Something, Peter wanted to buy in this store [, and something not]’ The first two examples above show that was ‘what’ can be interpreted as ‘something’ in its argument position (14a) as well as in a scrambled position (14b). However, as soon as was is supposed to be interpreted specifically (or perhaps with some kind of contrast), it is unacceptable, see (14c). The last example shows that was cannot be topicalized (stress on wollte) or contrastively topicalized (stress on was). The behavior of bare was ‘what’ stands in stark contrast with the DP-internally quantified etwas ‘something’, whose range of interpretations is considerably bigger. The following examples show a non-specific etwas in its base-position (15a) as well as scrambled position (15b), as in the case of was. Etwas also survives the assigning of a strong specificity property (15c) or (contrastive) topicalization (15d).

9 Below, I only discuss cases involving an abstract existential quantifiers (existential closure). However, the analysis should be easily translatable to cases with other kinds of quantifiers (e.g. adverbials). 10 I would like to thank Linda Gerlach, who kindly provided me with the present judgments. She is a speaker of Hochdeutsch, possibly influenced by the Niederrhein dialect. 11 A parallel state of affairs is found in Dutch.

Page 12: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

11

(15) a. … dass Peter in diesem Geschäft etwas kaufen wollte that Peter in this store something buy wanted ‘… that Peter wanted to buy something in this store’

b. … dass Peter etwas in diesem Geschäft kaufen wollte that Peter what in this store buy wanted ‘… that Peter wanted to buy something in this store’ c. … dass Peter ETWAS in diesem Geschäft kaufen wollte that Peter what in this store buy wanted ‘… that Peter wanted to buy SOMETHING in this store’ d. Etwas wollte Peter in diesem Geschäft kaufen [, und etwas nicht] what wanted Peter in this store buy and something not ‘Something, Peter wanted to buy in this store [, and something not]’ How can we account for the distributional restrictions on was? Assume that there is a default unselective existential quantifier located somewhere in the middle-field of the German sentence, as schematized in (16). (16) [CP *whP C [XP ∃ (*Foc)whP T [VP whP V]]]] (17) [CP ∃∃∃∃P C [XP (Foc)∃∃∃∃P T [VP ∃∃∃∃P V]]]] The German was must be bound by the abstract unselective quantifier, i.e. it cannot appear too high—outside of its scope (e.g. in SpecCP, (14d)). On the other hand, etwas can spell-out more features. Particularly, it can have a referential (specific indefinite) reading, as in (15d).12 Further, I assume that the ungrammaticality of (14c) stems from the presence of a silent (focus) operator (plausibly expressed by intonation), which prevents the wh-variable from binding by the abstract existential quantifier. Again, etwas can exploit its referential reading in such contexts, cf. the grammaticality of (15c). Importantly, not all kinds of wh-words allow for the kind of interpretation/distribution as was ‘what’ above. The examples below show a human-denoting wh-word in nominative and accusative (wer/wen ‘who’), locative wh-word (wo ‘where’), and finally temporal and manner wh-words (wann ‘when’ and wie ‘how’). While the locative wh-word patterns more or less with the referential ones, we observe a sharp decline in acceptability in the case of non-referential wh-words.13 (18) a. … dass mich da wieder wer getroffen hat

that me there again who.nom met has ‘… that someone has met me here again’ b. … dass ich da wieder wen getroffen habe that I there again who.acc met have ‘… that I met someone there again’

12 This does not mean that etwas cannot function like Heim’s (1982) indefinites at all. E.g. in donkey-sentences, it gets quantified by an abstract universal quantifier (Wenn du etwasi findest, gib dasi mir ‘If you find somethingi, give iti to me’ / ∀x. you find x � give x to me). It is the potential to be used referentially that makes etwas ‘something’ different from was ‘what’. 13 I use the terms referential vs. non-referential only for expository reasons because we have just seen that wh-words are not referential but rather set-denoting. It would be more precise to say “(not) involving an individual variable”, or “with/without referential potential”.

Page 13: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

12

c. ? … dass sie das wieder wo gemacht haben that they it again where done have ‘… that they did it somewhere again’ d. * … dass sie das wieder wann machen wollen that they it again when do want ‘… that they wanted to do it some time again’ e. * … dass sie das wieder wie machen wollen that they it again how do want ‘… that they want to do it some way again’ The unacceptability of non-referential wh-words reminds us of the unacceptability of some wh-in situ adjuncts in Chinese questions. I would like to argue that the impossibility to use non-referential/adjunct wh-words is indicative of unselective binding, in other words, I claim that non-referential wh-words resist unselective binding. At this point, I have no reasonable explanation of this state of affairs.14 The last point I would like to make in this section concerns the complexity of a non-operator wh-phrase. So far, I gave examples of simple wh-phrases like was ‘what’ or wo ‘where’. Note that complex wh-phrases are ruled out in similar contexts: (19) Gestern habe ich wieder was // * welche Bücher gekauft

yesterday have I again what which books bought ‘Yesterday I bought something // some books again’

The idea that first crosses one’s mind is that the wh-expressions used in the complex phrases above are inherently specific and are therefore ruled out from position where non-specificity is crucial. There are two arguments to deny this idea. Firstly, the German was-für construction is used for non-specific reference but is ruled out of the given context anyway: (20) * Gestern habe ich wieder was für Bücher gekauft

yesterday have I again what for books bought ‘Yesterday I bought some books again’

Secondly, German actually allows welche in the context above, as soon as the noun is elided: (21) [ Bücher,] gestern habe ich wieder welche gekauft

books yesterday have I again which bought ‘Yesterday I bought some (books) again’

In the next section, we will see that similar restrictions are observed in a Czech construction involving non-operator wh-expressions.

14 A remark on Dutch is due here. Even though Dutch is a language closely related to German, the only wh-word that can be used existentially, is wat ‘what’. At this point, I have no idea why this should be the case. It seems natural, though, that Dutch is in between German and English, which cannot use any wh-word like this. It would be very useful to look at some microvariation in Dutch and German dialects.

Page 14: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

13

3.1.1.2 Czech irrealis free relatives Another candidate for variable/indefinite wh-expressions which are not necessarily analyzed as displaced to the left periphery are irrealis free relatives (IFR). In IFR, a wh-expression is typically bound by an existential or negative quantifier, expressed by an auxiliary verb (‘(not) be’ or ‘(not) have’), which selects the “wh-clause”.15,16 (22) a. Je/Není s kým mluvit

is/is-not with whom talk.inf b. Mám/Nemám s kým mluvit have/have-not.1sg with whom talk.inf

‘There is no one / I have no one to talk to’ The distribution of wh-expressions in IFR resembles the distribution of the bare wh-variable in German. They can only occur in relative adjacency to the existential/negative auxiliaries ‘(not) be’/’(not) have’.17 They are ruled out from in-situ (focus) positions, as well as left-peripheral (contrastive) positions: (23) a. Mám s kým mluvit

have.1sg with whom talk.inf b. * Mám mluvit s kým have.1sg talk.inf with whom

c. * S kým mám mluvit [, ne o �em] with whom have.1sg talk.inf not about what ‘I have someone to talk to’ Again, as in the German case, potentially referential non-wh-phrases are much more flexible in their distribution: (24) a. Mám možnost s n�kým mluvit

have.1sg possibility with someone talk.inf b. Mám možnost mluvit ( jen) s n�kým have.1sg possibility talk.inf only with someone

c. S n�kým mám možnost mluvit [, a s n�kým ne] with someone have.1sg possibility talk.inf and with someone not ‘I have someone to talk to’ Thus, the above distinction can be captured by the following schema (note the analogy to the schemata in (16) and (17)):

15 The attribute irrealis captures the fact that this kind of relatives expresses an epistemic modality. As modality can play a role in structural considerations, it should be taken into account eventually. 16 The quantificational power can also stem from adverbial quantifiers of the always type. 17 In infinitival IFRs second position clitics can intervene between the existential/negative predicate and the wh-word, with no decrease in acceptability: (i) Nem�l jsem [IFR {kde} mu ho {kde} dát] not.had aux1sg him.dat it.acc where give.inf ‘There was no place (for me) to give it to him’

Page 15: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

14

(25) [CP *whP C [TP ∃-have [XP whP [VP Foc V *whP]]]] (26) [CP ∃∃∃∃P C [TP T [XP ∃∃∃∃P [VP Foc V ∃∃∃∃P]]]] Note that I leave open the question what XP exactly is, i.e. to which position the bare wh-expression (can/must) move(s). However, below I will show that the XP is not a CP and wh-words are therefore not standard operators in SpecCP but rather scrambled indefinites. Let me take an existing analysis as a starting point. Grosu (1994, 2002) argues that IFR are CPs, where the wh-expression appears in SpecCP, thus being an A-bar operator. This would make IFRs very similar to embedded questions (see section 3.1.1.3), except that they can be selected by a very limited set of verbs (namely, verbal existential/negative quantifiers). Grosu puts IFRs into a direct opposition with standard free relatives (SFR), which he takes to be DPs. Let us mention two born-out predictions of such an analysis: (i) IFRs do not show matching effects; (ii) IFRs are transparent for wh-extraction. (27) a. Je/mám [CP s kým mluvit]

is/have.1sg with whom talk.inf ‘There is / I have someone to talk to’ b. O �em1 není/nemáš [CP s kým mluvit t1]? about what is-not/have-not.2sg with whom talk.inf ‘About what is there someone / have you someone to talk to?’

The (auxiliary) verbs ‘is’ and ‘have’ normally assign nominative and accusative, respectively. However, in IFR, there is no case-requirement on the wh-expression at all (in the case above it is in instrumental, assigned by the preposition s ‘with’), and the case of the matrix existential verb apparently remains unassigned, i.e. matrix and embedded requirements on case do not need to match. SFRs are expected to behave in the opposite way exactly because there is an additional D-layer in their representation. (28) a. Petr pozval [DP koho (koli) potkal // * kdo (koli) mu zavolal]

Petr invited who.acc ever met who.nom ever him called ‘Petr invited who(ever) he met / called him’ b. * Kde1 Petr pozval [DP koho( koli) potkal t1]? where Petr invited who ever met ‘Where did Petr invite who(ever) he met t?’

However, Czech irrealis free relatives display a much more intricate pattern, which defies a simple CP-approach. The following discussion mainly aims to illustrate this fact. In the examples above we saw that the verb in the IFR is an infinitive. But if the IFR is selected by the ‘(not) have’ predicate, then it may also be subjunctive (the Czech subjunctive (also called conditional mood) consists of a subject-agreeing conditional auxiliary by- and a past participle verb form). (29) a. Mám/Nemám s kým mluvit // bych mluvil

have/have-not.1sg with whom talk.inf would.1sg talk.prt ‘I have someone / no one to talk to’

Page 16: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

15

b. Je/Není s kým mluvit // * bych mluvil is/is-not with whom talk.inf would.1sg talk.prt ‘There is someone / no one for me to talk to’ The difference in verb form implies a different syntactic structure. Indeed, what we find is that the infinitive form is transparent for pronominal clitic-extraction, which very strongly suggests that no CP is involved (Jakub Dotla�il, p.c.); the relevant elements below are boldfaced: (30) a. V�era jsem mu1 nem�l [IFR co dát t1]

yesterday aux.past.1sg him had-not what give.inf ‘Yesterday I had nothing to give to him’ b. V�era mu1 už nebylo [IFR co dát t1] yesterday him already was-not what give.inf ‘Yesterday there was nothing to give to him’

The following example shows that pronominal clitics typically do not escape CPs transparent for wh-extraction: (31) a. Koho1 si-s myslel, že Petr ošet�oval t1?

who refl-2sg thought that Petr treated ‘Who did you think that Peter was treated by?’ b. * Myslel si-s ho1, že Petr ošet�oval t1? thought refl-2sg him that Petr treated ‘Did you think that Petr treated him?’

On the other hand, the status of conditional-mood IFR is unclear as well because it not only disallows clitic-extraction but also wh-extraction: (32) a. * V�era jsem mu1 nem�l [IFR co bych dal t1]

yesterday aux.past.1sg him had-not what would.1sg give.prt ‘Yesterday I had nothing to give to him’ b. * O �em1 nemáš [IFR s kým bys mluvil t1]? about what have-not.2sg with whom would.2sg talk.prt ‘About what do you not have someone to talk to t?’

Again, note that generally there is nothing wrong with extraction out of embedded clauses in Czech: (33) Koho1 jsi �íkal, že bys cht�l pozvat t1?

who aux.past.2sg said that would.2sg want.prt invite ‘Who did you say that you would like to invite?’

The distinction between negative and existential quantification proves to be syntactically relevant, too. Firstly, negatively quantified structures are more liberal as for wh-extraction than existentially quantified ones:18

18 The reader could wonder whether the lower wh-phrases in (34) are ambiguous between an existential and interrogative reading, given the possibility to assign interrogative force at a distance (by unselective binding) in

Page 17: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

16

(34) a. O �em si nemáš/není s kým promluvit?

about what refl have-not.2sg/is-not with whom talk.inf ‘About what do you have / is there no one to talk to?’ b. ? O �em si máš/je s kým promluvit? about what refl have.2sg/is with whom talk.inf ‘About what do you have / is there someone to talk to?’

Secondly, the negative quantifier binds non-referential (adjunct) wh-expressions more readily than the existential one:19 (35) a. Nemám/Není tady kdy/jak/pro� uklidit/uklízet.

have-not.1sg/is-not here when/how/why clean up.prf/imprf ‘I have / There is (here) no time/way/reason to clean up.’ b. ?* Mám/Je tady kdy/jak/pro� uklidit/uklízet. have.1sg/is here when/how/why clean up.prf/imprf ‘I have / There is a time/way/reason to clean up.’

The fact that non-referential wh-words less readily obtain the necessary force in these constructions, together with the relative free positioning of the wh-word within the IFR (see footnote 17), are indicative of unselective binding. The aim of the discussion above was to show that IFR themselves do not form a single class, at least syntactically, and defy a simple CP-approach.20 Thus, wh-words in some IFRs probably do not even have the possibility to appear in SpecCP and are better analyzed as non-operator indefinites, which are unselectively bound by an existential or negative quantifier (in the form of an auxiliary verb). I summarize the data above in the following table. Note that the two parameters (be vs. have and negation vs. existential quantification) are sensitive to (i) the kind of wh-clause (infinitive vs. conditional), (ii) clitic- and wh-extraction possibilities, and (iii) the kind of wh-expression bound by the verbal quantifier (referential vs. non-referential21).

multiple questions (May 1985). The answer is negative, i.e. the lower wh-phrase, being c-commanded by the existential/negative predicate obligatorily receives existential/negative force. At the same time, the predicate (or possibly the lower wh-phrase) fails to intervene in creating an interrogative wh-chain. This clearly shows that the force of the two wh-phrases is substantially different, despite the identity of form. Thanks to John Nerbonne for pointing out this interesting issue. 19 I add the imperfective verb since it sounds more felicitous with the reason wh-phrase pro� ‘why’. 20 The fact that at least a subset of the wh-expressions in IFR occupy a lower position than SpecCP generates further questions. Czech left periphery behaves in a quasi-templatic way in the sense that there is a strict ordering within the second, so-called Wackernagel position: verbal auxiliaries (agreeing with subject) are followed by reflexive pronominals, then dative and accusative clitics come, and the right-most position is reserved for non-focused (temporal/place/manner) adverbs, like tady ‘here’ in the example below. The wh-expressions in IFR seem to be located somewhere in the right-most domain and certain amount of optionality seems to be involved: (i) Nebylo {kdy} tady {kdy} pouklízet. was-not when here when clean up 21 It has been argued that this distinction is more relevant than the classical argument vs. adjunct distinction (see e.g. Tsai 1994). However, a structural position may be important, too.

Page 18: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

17

(36) Czech irrealis free relatives: Summary of the data inf cond plain cl-extr wh-extr plain cl-extr wh-extr ref-wh non-ref-wh be � � ? * x22 x � ?* not be � � � * x x � � have � � ? � * * � ?* not have � � � � * * � � To conclude, let me show that Czech IFR disallow complex wh-expressions: (37) Nemám si s kým // * s kterým studentem promluvit

have-not.1sg refl with whom with which student talk.inf ‘I have no one / no student to talk to’

The situation strongly resembles the one observed in the preceding section for German. Namely, a non-specific wh-word does not contribute to the improvement of grammaticality but eliding the NP does: (38) * Nemám si s jakým studentem promluvit

have-not.1sg refl with what student talk.inf ‘I have no student to talk to’

(39) ? Nemám si s kterým promluvit

have-not.1sg refl with which talk.inf ‘I have no one to talk to’

Before I turn to the summary of this section, let me mention a case that has been argued to be another instance of wh-variables. 3.1.1.3 Embedded questions: Berman (1991) Berman (1991) argued that wh-expressions in embedded wh-clauses behave on a par with indefinites in the sense of Kamp (1981) and Heim (1982), i.e. elements of the same kind that I described in the preceding two sections. Let us now look at some examples (Berman 1991: p62/65): (40) a. The principal usually finds out which students cheat on the final exam b. MOST [student’(x) & cheat-on-the-final-exam’(x)] [find-out’(tp, [student’(x) & cheat on-the-final-exam’(x)])] c. For most students who cheat on the final exam, the principal finds out of them that they cheat on the final exam (41) a. With few exceptions, Mary knows which students submitted which abstracts to which conferences.

22 “x” means “irrelevant”. This is because even plain conditionals are ungrammatical with ‘(not) be’, therefore no further tests can be applied here.

Page 19: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

18

b. MOST [student’(x) & abstract’(y) & conference’(z) & submit-to’(x, y, z)] [know’(m, [student’(x) & abstract’(y) & conference’(z) & submit-to’(x, y, z)])] c. For most triples of a student, an abstract, and a conference such that the student submitted the abstract to the conference, Mary knows that the student submitted the abstract to the conference Berman argues that the whole wh-clause forms a restriction of the matrix-clause adverbial quantifier (usually, with few exceptions, in Berman’s notation translated as MOST). Because all wh-phrases appear in the wh-clause, they are unselectively bound by the quantifier. Berman does not discuss cases with no overt adverbial quantifier. In such cases wh-words are understood universally (John Nerbonne, p.c.).23 As far as I can see, there are a number of problematic aspects with Berman’s analysis. First of all, embedded questions, along with root questions, belong to the distributional core of wh-words in all languages with wh-morphology that are known to me. There is a tradition of analyzing wh-words in root questions as quantifiers (see section 3.2 for discussion and references). If wh-expressions in embedded questions behave as indefinites with no quantificational power, then their distribution is expected to be fundamentally different from those in root questions, which is clearly false. I will briefly return to the comparison of root and embedded questions in section 3.2. Furthermore, if my analysis in the preceding two sections is on the right track, we would expect to find some similarities between wh-expressions in embedded clauses and the cases discussed before. However, almost no such similarities can be found: embedded wh-expressions can be complex and they are licit independently of their referential properties. The only similarity can be attributed to the fact that not all verbs that embed clauses, can also embed wh-clauses. 3.1.1.4 Summary In this section I described three cases where wh-words seem to behave as variables/indefinites: German sentences containing wh-existentials, Czech irrealis free relatives, and (English) embedded wh-clauses. I observed that the first two cases exhibit remarkable similarities, even though they are superficially quite different constructions. Namely, they both display restrictions on non-referential wh-expressions (adjuncts) and on complex wh-expressions. The tentative hypothesis is that these restrictions are due to the non-operator status of the wh-expressions (or the fact that they are unselectively bound). On the other hand, the third case, even though similar in some semantic respects, has properties virtually identical to those in root questions, which is a canonical case of operator-variable relation.24 This section can be taken to be the first step towards the understanding of non-operator wh-phrases and the restrictions associated with them. It seems to me promising that the restrictions observed here are not completely isolated and are observed elsewhere. For example, there are

23 The quantification variability analysis is challenged by Ginzburg (1995) and Ginzburg & Sag (2000: section 4.4.2). However, the criticism is based on judgments of paraphrases involving quantifiers like to some/a limited extent, which are arguably attached in a different position than quantifiers like sometimes, which are used by Berman. There is not enough space to fully evaluate this discussion. 24 The operator-variable relation does not necessarily involve quantification. It has been argued that A-bar chains in relative clauses do not involve quantification (Lasnik and Stowell 1991); some also argue that even wh-words in root questions are no quantifiers (e.g. Ginzburg and Sag 2000). See section 3.2 for further remarks on the quantificational status of wh-phrases in questions.

Page 20: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

19

well-known restrictions on the extraction of non-referential wh-phrases (their stronger sensitivity to weak islands) as opposed to referential ones; as for the complex vs. simple matter, complex wh-phrases are known to behave differently from simple ones in certain respects, too; to name just one example, they are ruled out from some special cases of sluicing (van Craenenbroeck 2004). 3.1.2 Wh-phrases in quantifier-derivation In many Indo-European languages wh-words act in the derivation of universal, existential, negative, and other kinds of quantifiers. Consider the following two examples of forms derived from wh-words or their parts. (42) shows a regular English based on the form where. (43) displays an exactly parallel pattern for the Czech form jak ‘how’. (42) a. where wh ere

b. somewhere some wh ere c. nowhere no wh ere d. everywhere every wh ere e. wherever wh ere ever f. there th ere

(43) a. jak j ak

b. n�jak n� j ak c. nijak ni j ak d. všelijak všeli j ak e. jakkoli j ak koli f. tak t ak

The green part is the actual wh-form, one phoneme in both English and Czech, [w] and [j] respectively; it is missing in the last case where it seems to be replaced by a definiteness morpheme, phonemes [θ] in English and [t] in Czech.25 The violet part -ere- and -ak-, on the other hand, is present in all six forms. I will assume that it expresses a general quantifier-restriction, something like ‘place’ in the English case and ‘manner’ in the Czech case. The blue (leftmost or rightmost) part is a quantifier-like morpheme, determining the interpretation of the restricted variable. Let us further assume that the morphemes can be translated to a primitive hierarchical structure: (44) a. [some [wh + ere]] b. [n� [j + ak]] I assume that the quantifier c-commands the rest of the structure (so the (e) free-choice forms must be derived, the wh-part presumably moves to its Spec). It is plausible that the wh-part further c-commands the variable restriction but I remain careful in making any definite claims. Arguably, the restriction-part can be a spell-out of a bigger structure, as suggested by Kayne

25 However, see below the Czech derivates from ‘when’, which may suggest that the t(h)-part actually selects the wh-part.

Page 21: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

20

(2005), who proposes that the [r]-part in English is comparable to the German and Dutch R-pronoun, which is indefinite and can spell-out a number of restriction types, namely ‘place’, ‘thing’, or even ‘reason’. The restriction-part can definitely have lexical complements, consider examples like (some)where in the city. Let us now have a look at the paradigms that we find in English, Czech, German, and Dutch. Note that below I only exemplify apparently complex forms, i.e. such that have a (wh-) base selected by an overt quantifier. Therefore, I omit the bare wh-existentials presented in the preceding sections (even though from the semantic point of view they belong to the existential row in Czech, Dutch, and German).26 (45) English who what where when how why which wh wh-o-m wh-at wh-ere wh-en how why which free choice

whom-ever

what-ever

where-ever

when-ever

how-ever *why-ever

which-ever

D th-e-m th-at th-ere th-en so there-fore

th-is

∃ some-body

some-thing

some-where

some-time

some-how some

¬ no-body no-thing no-where

no-time ?no-how, no-way

no

∀ every-body

every-thing

every-where

every-time

?every-how, ?every-way

every, each

Observations about English: • A near-regular pattern is found in free-choice items (the only exception is *why-ever). • When we ignore differing vowels in the stems of whom vs. them and what vs. that, there

seems to be a fairly regular pattern for the wh-/th-variation. • Quantifiers some, no, and every generally defy selecting wh-forms. This may be because they

are actually NP-quantifiers (some/no/every chicken). There are two apparent exceptions: where can be selected across the board and how can be selected by some. I have nothing to say about the latter. As for where, however, note that it can be replaced by place (someplace), so in fact, as already hinted, the restriction-part in where probably spells out nominals (Kayne 2005). It seems tenable that there are no “real” wh-form-selecting quantifiers in English (which may eventually prove to be in correlation with the fact that English, as opposed to German and Dutch, does not have bare wh-variables, as described in section 3.1.1.1).

• There is a plausible decomposition for which, which could be wh+each (especially in the light of Dutch).

26 Dark green highlights the basic wh-word row. Light green highlights the cases where a quantifier/D-word is derived from the corresponding wh-word and furthermore, has more or less the expected semantics. Gray fields contain either ungrammatical forms of the expected quantifiers or grammatical forms, which are not derived from the corresponding wh-word. The gray fields are left empty when there is no straightforward/simple variant.

Page 22: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

21

(46) Czech who

(dative) what (dative)

where when how why how much

which what (kind of)

wh k-o-mu �-e-mu k-de k-dy j-ak pro-� k-olik k-terý j-ak-ý free choice

komu-koli

�emu-koli

kde-koli kdy-koli jak-koli *kolik-koli

který-koli

jaký-koli

D t-o-mu t-o-mu t-a-dy t-e-h-dy t-ak pro-to t-olik ten t-ak(ov)ý ∃ n�-ko-

mu n�-�e-mu

n�-kde n�-kdy n�-jak n�-kolik n�-který n�-jaký

¬ ni-ko-mu ni-�e-mu ni-kde ni-kdy ni-jak *ni-kolik žádný ni-jaký ∀ vš-e-mu vš-e-mu vš-u-de v-ž-dy vš-e-li-

jak *vše-

kolik každý vš-e-li-

jaký else n�-komu

jinému n�-�emu jinému

jin-de jin-dy jin-ak *jin-olik jiný jin-aký, jin-a�í

Observations about Czech: • The Czech pattern looks vastly regular. However, one has to be careful because some slots in

the paradigm may fit only apparently, i.e. ni-jaký ‘no-wh-kind’ has an idiomatic reading (it means something like ‘bland’, ‘flabby’, etc.) and semantically is not derived from n�-jaký ‘some-wh-kind’.

• The who-/what-series are in dative because nominatives often have suppletive forms and do not reveal the similarities.

• The ni-prefix from the negation series expresses negative concord, rather than real negation. The issue is not trivial and I leave it aside.

• D-words are generally derived by selecting the restriction-part, to the exclusion of the wh-part, as in English. The same holds for the ∀-words. A potential exception are the when-cells,

which seem to involve a allophones of the phoneme [g], namely ž [�] and h [h].27 • There are interesting cross-paradigmatic relations, note for example that jaký ‘wh-kind’ is

derived from jak ‘how’ by conversion (only an adjectival case/gender/number ending is added).

• Despite the superficial regularity, there are interesting gaps, e.g. kolik ‘how much’ can only figure in definite-determiner and existential-Q derivation.

• Czech has an else-prefix. This may be an interesting thing to investigate. Note that Germanic expresses this prefix by a free morpheme attached related to an existential quantifier: somewhere else, ergens anders, irgendwo anders, a strategy used in Czech for what and who series.

• As in Germanic languages, the word který ‘which’ seems to be morphologically (and hence syntactically) complex. Unfortunately, it cannot be decomposed in a straightforward way.

(47) Dutch who what where when how why which wh w-ie w-at w-aar wanneer hoe waarom w-elk- free choice

wie ook (maar)

wat ook waar ook wanneer ook

hoe ook waarom ook

welk- ook

27 This can be seen more explicitly in other Slavic languages (e.g. Russian). In future, I plan to extend my database.

Page 23: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

22

D d-ie d-at d-aar toen, dan zo daarom ∃ iemand iets ergens ooit ene N ¬ n-

iemand n-iets n-ergens n-ooit geen

∀ allen alles over-al al-tijd, immer

alle, elke

Observations about Dutch: • Dutch looks very much like English in the sense that it uses wh-words only for the derivation

of free-choice items. However, the free choice items are free morphemes in Dutch. Plausibly, they can be treated as floating quantifiers.

• D-words exploit the restriction-part quite regularly, as in English. • The existential-quantifier series have a different derivational basis. Importantly, the negative

quantifiers are derived from the existential series (by adding n-) and not from the wh-series. • Welke ‘which’ seems to be derived from elke ‘each/every’ by adding a wh-element. • There is a plausible decomposition of ergens ‘somewhere’ into er-gens. This would allow us

to say that this existential quantifier actually contains the R-pronoun, similarly as waar ‘where’. A support for this analysis comes from the fact that ergens figures in prepositional contexts (ergens mee ‘with something’ ergens over ‘about something’, etc.), so it must contain a nominal element anyway. This analytical step actually connects the existential and negative series of ‘where’ with the wh-series, coming to the same result as in English.

(48) German who what where when how why which wh w-er w-as w-o w-ann w-ie warum welch- free choice

wer auch immer

was auch immer

wo auch immer

wann auch immer

wie auch immer

warum auch immer

irgend-welch-

D d-er d-as d-a d-a, d-ann

so darum, deswegen

∃ irgend-wer, jemand

irgend-was

irgend-wo

irgend-wann

irgend-wie

irgend-welch-

¬ niemand nichts nirgend-wo

nirgend-wann

nirgend-wie

keine

∀ alle alles über-all immer, all-zeit

alle, jede

Observations about German: • Like in Dutch, also German free choice quantifiers are free morphemes. • There is an additional regular morphological existential-series, based on the morpheme

irgend (formally corresponding to the Dutch ergens). However, one has to be careful because there are some (probably important) semantic and syntactic differences: irgendwas vs. etwas and irgendwer vs. jemand. Furthermore, irgend- is not limited to selecting wh-forms (the Dutch ergens, by the way, is even excluded from that), since we have irgend-jemand ‘someone’ and even irgend-ein Mann ‘some man’.

• German irgendwo can freely be used with prepositions, similarly as the Dutch ergens.

Page 24: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

23

Global observations: • D-words are derived from the same stem as wh-words but NOT directly from the wh-words. • The 4-language sample gives rise to the following implication: If existential quantifiers are

not derived from wh-words, then negative quantifiers are not, either (English, Dutch). This seems to suggest that negative quantifiers are extended existentials.

• Germanic languages do not like to form their definite ‘manner’ demonstratives via derivation. They all have a suppletive form: so/so/zo.

• Wh-reason adverbs like why generally do not take part in quantifier-derivation. Languages where ‘why’ involves an argument-like wh-element (Czech, Dutch, and German) allow it to be used for D-word derivation. Only Dutch and German allow their ‘why’ to directly act in free-choice quantifier derivation. No other forms are attested anywhere.

• The fact that Dutch and German use a free morpheme to express a free choice quantifier could correlate with the fact that these two languages also allow unselective binding of wh-expressions by an abstract existential quantifier (see section 3.1.1.1).

Apart from looking for the fine structure of quantified and non-quantified wh-phrases, a paradigm investigation like the one above gives rise to other interesting questions. Let me mention one of them below. What kind of existential quantification are we dealing with in the cases above? We know that indefinites may be non-specific or specific. The former can be identified by taking a narrow scope with respect to a universal quantifier, while the latter take a wide scope. The following examples show a scopal ambiguity of the Czech quantifier n�koho ‘somebody’, n�kam ‘somewhere’, n�kdy ‘sometime’, and n�jak ‘someway’ with respect to a universal quantifier každá hol�i�ka ‘every girl’. All the sentences may potentially be uttered when talking about a number of girls who have played hide-and-seek. In the first set of examples, the surface order makes the narrow reading of the indefinite more readily available, i.e. the sentences are fully grammatical under the surface-scope reading and very marginal or ungrammatical under the inverse scope reading. In the second set of examples, the existential quantifiers are topicalized so that a specific reading of the existential is more readily available (the inverse scope reading is still acceptable). We can see that a wide-scope reading of the existential is significantly demoted with the existentials ‘sometime’ and ‘someway’.28 (49) a. Každá hol�i�ka n�koho našla [∀ > ∃, ??∃ > ∀]

every girl found someone ‘Every girl found someone’ b. Každá hol�i�ka se n�kam schovala [∀ > ∃, ??∃ > ∀] every girl refl somewhere hid ‘Every girl hid somewhere’

c. Každá hol�i�ka n�kdy pykala [∀ > ∃, *∃ > ∀] every girl some-time sought ‘Every girl sought at some time / sometimes’ d. Každá hol�i�ka se n�jak schovala [∀ > ∃, *∃ > ∀] every girl refl someway hid ‘Every girl hid in a way’

28 In my intuition, the wide scope reading is much less problematic when a different bound existential morpheme is used, namely -si. This opens up the path for further investigations.

Page 25: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

24

(50) a. N�koho našla každá hol�i�ka [∃ > ∀, ∀ > ∃]

someone found every girl ‘There was someone who was found by every girl’ b. N�kam se schovala každá hol�i�ka [∃ > ∀, ∀ > ∃] somewhere refl hid every girl ‘There was a place where every girl hid’

c. N�kdy pykala každá hol�i�ka [??∃ > ∀, ∀ > ∃] some-time sought every girl ‘There was a time when every girl sought’ d. N�jak se schovala každá hol�i�ka. [??∃ > ∀, ∀ > ∃] someway refl hid every girl ‘There was a way in which every girl hid’ That the problem is in the existential quantification and not with the wh-word in general can be shown by corresponding root questions, where the ambiguity arises quite clearly: (51) c’. Kdy pykala každá hol�i�ka? [wh > ∀, ∀ > wh]

when sought every girl ‘When did every girl seek?’ d’. Jak se schovala každá hol�i�ka? [wh > ∀, ∀ > wh] how refl hid every girl ‘How did every girl hide?’

Note that this is not the first selective behavior of existential quantifiers that we have encountered here. In sections 3.1.1.1 and 3.1.1.2 we saw that existential quantifiers behave selectively also in cases of unselective binding (i.e. they fail to bind non-referential wh-phrases).29 3.1.3 Summary In this section, I showed that there are some quite interesting and open issues related to wh-expressions with no inherent quantificational power or A-bar properties. I started with discussing bare wh-expressions which arguably get their quantificational power by the means of unselective binding, as opposed to movement. In the second subsection I had a look at the paradigms and morphological make-up of quantifiers that are derived from wh-words. We saw that wh-words themselves may be treated as syntactically complex structures, containing a stem-like morpheme, which functions as a general restriction on the variable (place, time, etc.). This morpheme is further extended by an (indefinite) wh-morpheme or a (definite) th-morpheme. Indefinite, wh-bases can further be selected by quantifier morphemes. Despite the idiosyncratic noise, which is undoubtedly present in the observed paradigms, I believe that a detailed morphological analysis,

29 A way of unifying these phenomena would be to say that unselective binding (and therefore distributional restrictions) is also present in examples (49)–(50). In such an approach, recently put forth by Kratzer (2005), the element that I have been calling a quantifier-morpheme here, e.g. n�- in n�kdo ‘someone’, is actually an uninterpretable existential-concord element. A generalized concord-approach (involving negative, existential, but also modal concord) is also argued for by Zeijlstra (2007).

Page 26: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

25

in connection with syntactic correlations, may be an important step towards understanding the syntax and semantics of wh-expressions.30 On the basis of a number of observations made in this section, we can formulate the following tentative generalizations: • There is a group of existential quantifiers that are incompatible with non-referential wh-

phrases, namely the abstract existential (unselective) operator in German (3.1.1.1), the verbal existential operators in Czech (3.1.1.2), and the specific n�- ‘some’ operator in Czech.

• The abstract existential operator in German and the verbal ones in Czech are incompatible with complex wh-phrases.

• The ability of the abstract existential operator to bind (referential) wh-words seems to correlate with the existence of free-morpheme free choice operators (i.e. the German auch immer and the Dutch ook (maar)). The strength of this correlation may easily be tested by looking at other languages.

• Free choice and definite operators are generally more liberal in binding wh-words than existential or negative operators (i.e. they also bind adjunct/non-referential wh-words). This suggests that question operators belong to the same quantifier-family.

3.2 Wh-phrases as quantifiers? In the preceding section, I discussed wh-expressions which do not appear at the root of the syntactic tree. They were all embedded in structures which contained quantifiers to bind them. In this section, I will discuss some aspects of wh-expressions in questions, i.e. at the root of the syntactic tree. There has been a long-standing assumption, going back to Chomsky (1973, 1977) and surviving till modern theories of syntax, that wh-expressions in questions are quantifiers. The following examples illustrate the background intuition of this idea: (52) Who did John see?

LF: for which x, x a person, John saw x (Chomsky 1977: p83) (53) John saw everyone.

LF: for every x, x a person, John saw x Note further that quantifiers like every are associated with a property traditionally referred to as weak crossover (WCO), which is a restriction on quantifier-binding. WCO prohibits a binding configuration where a pronominal variable (or a constituent containing it) c-commands the trace of the raised quantifier.31 Thus, his cannot be bound by every student in the following sentence: (54) *Hisi mother loves every studenti

LF: for every x, x a student, x’s mother loves x

30 John Nerbonne (p.c.) points out to me that one could obtain more interesting results by examining wh-words and their derivatives in a number of typologically different languages. Given the degree of idiosyncrasy, even strong tendencies (e.g. regarding the distinction referential vs. non-referential) could be used as relevant data and may lead to interesting generalizations. 31 For a more detailed discussion and a more precise formulation of WCO see Lasnik and Stowell (1991: p691).

Page 27: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

26

Compare this with a topicalized non-quantified R-expression Peter, where the binding is unproblematic. (55) Peter1/i, hisi mother loves t1 Crucially, wh-expressions in root questions behave like quantifiers, which supports the analysis in (52). (56) *Who1/i does hisi mother love t1?

LF: for which x, x a person, x’s mother loves x Lasnik and Stowell (1991) were the first to note that there is a similar distinction between wh-expressions (and A-bar operators in general) in different contexts. Thus, while wh-movement in questions is quantifier-like and triggers WCO, wh-movement in appositive relative clauses lacks this property (data from Rizzi 1997: p291):32 (57) a. ?* Who1/i does hisi mother really like t1?

b. John, who1/i hisi mother really likes t1 Note further that (at least contrastive) focus also behaves in a quantifier-like manner and thus differs from the topic in (55): (58) *PETER1/i hisi mother loves t1 [, not JOHN] On the basis of data above, Lasnik and Stowell (1991) propose that A-bar dependencies split into two classes: quantificational (involving focus-movement structures, root wh-questions, and QR) and non-quantificational (involving topicalization, wh-movement in (appositive) relative clauses, but also null operator movement in parasitic gap and tough-movement constructions). Rizzi (1997) formalized their proposal by the means of a fine left-peripheral syntax with a split C-head, (at least) into Foc and Top heads (see the end of section 2.2). The specifier of Foc hosts quantificational operators and the specifier of Top hosts non-quantificational ones. The present discussion gives us a diagnostics for determining the status of a wh-expression in embedded clauses (cf. section 3.1.1.3). The data at stake are the following:33 (59) a. ?*Who1/i does hisi mother really like t1?

b. x I don’t know who1/i hisi mother really likes t1

32 Chomsky (1982) assigns the same properties to restrictive relatives. However, his judgments are disputed elsewhere in the literature (Higginbotham 1980, Safir 1986). So after all, there seems to be a distinction between appositive and restrictive relatives in this respect. 33 I have not obtained any judgments yet, which is why “x” is used instead of the judgment marker. Still, Lasnik & Stowell (1991: p703) give one example of an embedded wh-clause and their judgments concerning WCO do not differ from other examples involving root questions: (i) * Mary asked me [whoi [PRO to persuade hisi motherj [PROj to vouch for ti]]]

Page 28: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

27

If (59b) is unacceptable in the very same way as (59a), then either Berman (1991) cannot be right claiming that embedded wh-expressions lack quantificational power, or all other authors cannot be right in claiming that weak crossover violations are due to a quantifier “crossing” a pronominal. This is not the right place to decide this controversy but it is important to keep it in mind. In conclusion to the discussion of weak crossover effects, let me mention that they are not observed universally (see Richards 2001 for Serbo-Croatian, Sturgeon 2005: ch2 for Czech). This fact would directly translate to a non-quantificational analysis of question operators in some languages. Within the cartographical inquiry, Grohmann (2006) proposes that German wh-phrases in root questions are topics, i.e. non-quantifiers in terms of Lasnik and Stowell (1991) and Rizzi (1997). Currently, I am not aware of any other syntactic proposal denying the quantificational status. Within more semantics-based theories, however, there is a whole tradition of analyzing wh-words in questions as non-quantifiers. These proposals go back to analyzing questions as “open propositions” (Hamblin 1958). In such a case, wh-words are analyzed as non-quantificational indefinites, i.e. expressions denoting sets. The set is not “closed up” by a quantifier, rather, it “percolates” all the way up to the node denoting a proposition, which consequently denotes a set of propositions, defined by the range of denotations of the indefinite wh-phrase. 34 This idea has been recently followed by Ginzburg & Sag (2000), within the framework of HPSG, but also Kratzer (2005). The problem of the quantificational status of wh-phrases in root (or embedded) questions is complex and I am far from being able to evaluate all the arguments that have been put forth. At this point, I would only like to note that one point of criticism considered at the end of section 3.1.1.3 holds here, too. In particular, some distributional restrictions, concerning the referentiality and complexity of the wh-phrase, are observed with unselectively bound indefinite wh-phrases (discussed in 3.1.1) but not with wh-phrases in questions, which undergo movement to the left periphery.35 3.3 Wh-phrases as specific DPs In this section, I will show that in the context of relative clauses wh-expressions, apart from being operators, behave as specific DPs, so that their behavior is directly comparable to the one of personal pronouns. This can be seen most obviously on the example of appositive relatives clauses, which can often be paraphrased using a personal pronoun in place of the wh-operator: (60) a. I also invited Mark, who really loves classical music

b. I also invited Mark. He really loves classical music

34 There is broad consensus of analyzing questions as sets of (sets of) propositions, whatever the actual technical implementation of this idea may be. An overview of approaches with a lot of references can be found in Hagstrom (2003). 35 The criticism is somewhat weaker here. Note that the restrictions on complexity and referentiality can be attributed either (i) to the mechanism of unselective binding or (ii) to wh-phrases denoting indefinites. Non-quantificational analyses of wh-phrases in (root) questions would only have problems if the latter was true, while Berman’s (1991) analysis of embedded wh-phrases as unselectively bound indefinites runs into troubles in either case.

Page 29: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

28

English appositive relative clauses exhibit a rather special behavior; from the three possibilities of expressing the relative-clause left periphery, i.e. overt wh-operator, that-complementizer, and zero, they only allow wh-operators: (61) I also invited Mark, who/*that/*0 really loves classical music This can be directly translated to the fact that only wh-expressions can be interpreted specifically in appositive relatives, as opposed to a zero-operators. This assumption is further supported by the behavior of wh-operators in restrictive relative clauses: English restrictive wh-relatives, as opposed to that- and 0-relatives, force a specific reading of the head with respect to a RC-internal universal quantifier in the subject position (analogous data are reported by Aoun and Li 2003): (62) a. the book which every student should read, namely…

… Syntactic Structures [individual reading, ∃ > ∀] … the one he refers about [functional reading, ∃ > ∀] * … X in case of A, Y in case of B, Z in case of C, etc. [pair-list reading, ∀ > ∃] b. the book (that) every student should read… … Syntactic Structures [individual reading, ∃ > ∀] … the one he refers about [functional reading, ∃ > ∀] … X in case of A, Y in case of B, Z in case of C, etc. [pair-list reading, ∀ > ∃]

The potential continuations exemplify the possible readings of the relative clause operator (which and 0, respectively) with respect to the universal quantifier. In a wh-relative (62a), the operator only has wide reading (either individual or functional). In a that-relative (62b), the universal quantifier can scope over the zero relative operator, which generates the pair-list reading.36 Note further that the behavior of wh-operators in root questions does not correspond to wh-relatives but rather to that- and 0-relatives. (63) Which book should every student read? (∃ > ∀; ∀ > ∃) In a paper of mine (Šimík 2007b), I argue that the state of affairs encountered in the wh-restrictive relatives, if real, is explained when we assume that wh-operators in English relatives (both restrictive and appositive), as opposed to questions, contain a D-layer on top of the wh-operator. This D-layer automatically translates to specific/pronominal semantics. Indeed, as for their distribution and semantics, English wh-relatives are analogous to relatives with resumptive pronouns in other languages (Czech, Hebrew, Greek, etc.). Other instances of D-like wh-words can be found in free relatives, which have been argued to involve an empty D, too (Caponigro 2003 and references therein). This D can in some languages be realized as a so-called light head (Citko 2004). These assumptions are quite well-known and not particularly controversial, so I am not going to discuss them here in detail. 36 There is a controversy in the literature as for how the pair-list reading, esp. in questions, should be generated. Some argue that the pair-list reading is not a result of reversed scope relations but rather it is a variant of the functional reading; cf. Engdahl (1986), Chierchia (1993), or Ginzburg & Sag (2000).

Page 30: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

29

3.4 Wh-phrases and maximalizing semantics A number of authors have argued that wh-phrases (and generally operators) are associated with so-called maximalizing semantics. Generally and very informally said, maximalization is a process of quantification such that its outcome is reference to something maximal. Thus, maximalization can apply both to scalar sets (where it yields the highest element on a certain scale, e.g. the highest degree for gradable adjectives) and sets of individuals (where it yields the maximal number of elements in a set). According to von Stechow (1984), maximalizing semantics is involved in comparative constructions. The informal semantic representation of the sentence below is as follows: there is a degree d such that a linguist can be maximally d-rich and John is richer than d. (64) John is richer than a linguist can be (Rullmann 1995: p54) Rullmann (1995) adopts von Stechow’s analysis of comparatives and applies it to root and embedded questions and free relative clauses. Jacobson (1988, 1995) also argues that maximality is involved in the semantics of free relatives. Grosu and Landmann (1998) introduced a new and by now widely accepted category of relative clauses, maximalizing relatives, where a syntacto-semantic operation of maximalization is the defining property. Something like maximalization is arguably involved also in exclamatives (Zanuttini and Portner 2003 use the notion of "widening", see also Rett 2007). If maximalization semantics is real, it is necessary to ask how it arises. The general logic of the linguistic system which is assumed here implies that it should be related to some structural position. If maximalization is involved in the constructions suggested above (i.e. free relatives, maximalizing relatives, comparatives, questions), then these should have something in common structurally, more particularly there should be a “maximalizing” functional head. Given the range of constructions for which maximalization has been reported, it is probable that it is a rather core property of A-bar operators and therefore sits quite low in their functional hierarchy. Of course, these considerations become meaningful only if maximalization proves to be a syntactic property and not a purely semantic one. 3.5 Conclusion This section showed that wh-expressions in Czech, German, and English can bear quite various functions—ranging from variables, through quantifiers, all the way to specific DPs. I assumed that the functions are generally determined by a c-commanding quantifier/determiner, which may be related to the wh-expression either very locally (phrase-internally) or at a distance (unselective binding). In other words, the semantic character of wh-expressions seems to be derivative from the kind structure in which they appear. Before I come to the conclusion and the summary of the main hypotheses made in this proposal, I will deal with two issues which are closely related to the discussion so far. 4 Further related issues In this section I will discuss two issues which are closely related to the preceding discussion and which bring in some novel analytical and theoretical views. In section 4.1 I focus on cases of

Page 31: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

30

multi-step operator movement. Section 4.2 deals with the pronominal nature of complementizer (left-peripheral) heads. 4.1 Multi-step wh-movement 4.1.1 Cases of extraction out of finite CPs It has been argued that A-bar movement proceeds in successive steps and not in one fell swoop. This assumption has been widely accepted for cases of so-called long A-bar movement, i.e. extraction out of finite CPs. Rizzi (2006) summarizes the evidence for the existence of intermediate landing sites in A-bar movement: intermediate complementizers may show agreement with the operator phrase in the matrix SpecCP; there may be subject-auxiliary inversion in embedded clauses (Belfast English); intermediate positions must be available for anaphor-binding purposes. Another piece of evidence comes from wh-doubling (sometimes referred to as wh-scope marking). Some dialects of German and Dutch realize a wh-phrase in both matrix and embedded clauses (Barbiers, Koeneman, and Lekakou 2007): (65) Wie/wat denk je wie/die ik gezien heb?

who/what think you who/this.rel.pron I seen have ‘Who do you think I saw?’

Given the evidence for intermediate landing sites, we should ask why wh-phrases stop at intermediate positions on their way to the matrix SpecCP at all. The two major explanations on the market are as follows. Some authors take intermediate positions as evidence that there are phases: if something is to move out of a phase, it has to move through the phase’s edge, since this is the only accessible position for further operations. Rizzi (2006), following McCloskey (2002), stipulates that there are pseudo-criterial features (pseudo-focus, pseudo-topic, etc.) in the embedded clauses, which trigger intermediate movements, without having interpretive consequences. The former explanation dwells on the postulation of phases—a device whose naturalness has been questioned. The second solution seems even worse—an intermediate movement is totally vacuous and the postulation of pseudo-criterial features (formal features par excellence, i.e. EPP) is nothing but a stipulation. In the following subsection, I look at “short”, one-fell-swoop A-bar movement. The question I want to ask is whether it is crystal clear to assume that a wh-phrase moves to the left-periphery only once. If we find evidence that there is a multi-step movement even without extraction out of finite CPs and if such “intermediate movements” are meaningful, then a serious and novel hypothesis would arise: intermediate movement is meaningful. 4.1.2 Cases without extraction Let us now look at the evidence that A-bar movement to SpecCP is not trivial and possibly proceeds in more steps. As far as I know, Boeckx (2003) is the first one to have argued for this option; his subsequent work on this topic includes Boeckx and Grohmann (2004) and Boeckx and Stjepanovi� (2005). He argues that in some languages and in some constructions, there is a

Page 32: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

31

scrambling-like movement, which feeds the final A-bar movement to SpecCP. He and Grohmann call this step submove. Arguably, this step is given support from both interfaces: seen from the PF side, the “intermediate” landing site can be realized as a (resumptive) pronoun (which is analyzed as a stranded D-layer of the operator phrase and yields interpretive effects); on the LF side, this movement has consequences for the interpretation, namely, the operator is interpreted as specific.37 Czech provides evidence for a partial movement that takes place very high in the left periphery of interrogative clauses. Observe the following pair: (66) a. Kterého u�itele jsi v�era potkal?

which teacher aux.past.2sg yesterday met b. Kterého u�itele že jsi v�era potkal? which teacher that aux.past.2sg yesterday met

‘Which teacher (did anyone say) did you meet yesterday?’ The example (66a) shows a standard Czech question. The question in (66b), on the other hand, contains an additional morpheme že, placed behind the wh-phrase kterého u�itele ‘which teacher’. This element is homomorphous with the standard Czech declarative complementizer ‘that’ and when used in a root question contexts, it triggers a ‘speech-act’ presupposition, which can be roughly expressed as follows: ‘it has been said that you met some teacher yesterday’. The position as well as the morphological form of this element suggest that it realizes a high functional head in the CP-domain, say Rizzi’s (1997) Force (for the sake of simplicity). The wh-phrase then sits in SpecForceP. Now, note that this že can appear in between the wh-word and the rest of the wh-phrase. The following examples also show that the positioning is not arbitrary and obeys at least standard constituency restrictions: (67) a. Kterého že ( starého) kamaráda jsi navštívil?

which that old friend aux.past.2sg visited b. * Kterého starého že kamaráda jsi navštívil? which old that friend aux.past.2sg visited

‘Which old friend (did anyone say) did you visit?’ As far as I can see, the only way to cope with such data in the analysis is to say that there is more than one movement to SpecCP. Either the wh-part moves independently of the rest (and to a different position), or the whole wh-phrase moves to a certain position, from which the wh-part is further raised to SpecForceP. Že is not the only element that forces a multi-step analysis. Another element like that is to ‘it/this’, an invariant demonstrative-like morpheme, which arguably realizes a Foc head in Rizzi’s left periphery (Šimík 2007a), because it is only licensed in wh-questions and cases of contrastive focus dislocation:

37 This “submove hypothesis” generates interesting empirical questions concerning the relation between resumption and (long) scrambling. There are some apparent problems to be solved, e.g. in Hebrew, resumptive pronouns can be spelled out in both argument position and a scrambled position (and apart from that also in the operator position).

Page 33: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

32

(68) a. Kterého kamaráda jsi to v�era navšítivil? which friend aux.past.2sg it yesterday visitek ‘Which friend was it that you visited yesterday?’

b. PETRA[Foc] jsem to v�era navštívil Petr aux.past.1sg it yesterday visited ‘It was Peter that I visited yesterday’ c. * Petra[Top] jsem to v�era [Foc navštívil] Petr aux.past.1sg it yesterday visitek ‘Peter I visited yesterday’ To is located after all verbal auxiliaries and nominal clitics, as the following example shows: (69) Co jste se ho to snažili nau�it?

what aux.past.2pl refl him it tried tlach ‘What was it that you tried to teach him?’

Many authors have argued that wh-movement in root questions targets SpecFocP, for both LF and PF reasons (Horvath 1986 for Hungarian, Boškovic 1998 for Serbo-Croatian, Rizzi 1997 for Italian, among others). The condition on the distribution of to observed in (68) suggests that it is licensed by a focus/wh-phrase in its specifier. However, as we see in (69), to and the wh-phrase are not adjacent. Thus, we have to admit that the wh-phrase does not stay in the specifier of to (SpecFocP) but rather moves higher, to a position which precedes all the second position clitics (presumably to SpecTopP).38 Now, note that to can be used along with že and that the latter can divide the wh-phrase into two pieces, as seen above. Witness the following example: (70) Kterého že kamaráda jste to v�era navštívili?

which that friend aux.past.2pl it yesterday visited ‘Which friend was it that (anyone said) you visited yesterday?’

The discussion above leads us to the following (schematic) analysis of (70): (71) [ForceP which3 že [TopP [<which3> friend]2 jste [FocP <[which friend]1/2> to [IP yesterday

visited <[which friend]1>]]]] This analysis has to be further supported. However, if it proves to be tenable, many interesting research questions follow: • Why do wh-phrases move in a multi-step fashion? • What projections do they pass in which syntactic contexts? • Does a multi-step movement make sense semantically? • Can further movements blur the quantificational status of wh-words in root questions (cf. the

discussion in section 3.2)? • What positions are targeted in cases of extraction from finite CPs?

38 Note that if SpecTopP is a non-quantifier position, as standardly assumed, the analysis captures the fact that Czech wh-questions do not observe WCO effects (Sturgeon 2005: Ch2); cf. section 3.2.

Page 34: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

33

• Is there an equivalent state of affairs in languages that make productive use of unselective binding? I.e. is there a set of binders in the left periphery?

4.2 Complementizers as invariant pronominals Complementizers often come in the form of invariant (non--marked) pronominals. English uses that as a declarative and relative complementizer. German uses dass (phonologically identical to das ‘the/this’) and Dutch dat ‘that’ as declarative complementizers. Some dialects of German use wo ‘where’ as a relative complementizer. Russian uses �to ‘what’ as a declarative and relative complementizer. Czech and Polish use co ‘what’ as a relative complementizer. All this may be just a historical accident associated with Indo-European languages. Indeed, the generative practice so far has not provided any deep insight into why this should be the case, which is an unwanted inheritance of lexicalism. As I see it, it is much more interesting to go the paranoiac way and start out from the hypothesis that complementizers (or more generally heads in the left periphery of clauses) are pronominals minus certain nominal features (-features, Case, or both). Semantically, they may be viewed as operators, which affect their complements (clauses) and possibly are licensed by c-commanding material (selecting verbs/predicates, or their own specifiers). This is another point where it is worth doing a reasonable (wh-)pronominal decomposition. Let me put forth two very sketchy arguments in favor of my view. Firstly, I have been supposing that there are a number of functional heads in the left periphery. Obviously, not all languages express every single head phonologically. A reason for this may very well be that more functional heads get spelled out at once, in one morpheme.39 Consider the following contrast between Czech and English (the English sentence comes from the internet, found by www.google.com, and the Czech one is its translation): (72) a. It was awful to think about that he still lived with her

b. Bylo hrozné myslet na *( to) že stále žije s ní was awful think on it that still lives with her

Note that the nominal form of the English complementizer that correlates with the fact that English embedded declarative CPs can directly be selected by a preposition. The Czech complementizer že has nothing in common with nominal morphology and (perhaps therefore) defies being directly selected by a preposition—there has to be an additional (case-marked) demonstrative morpheme to ‘it/this’. A plausible analysis of this state of affairs is that English declarative complementizer realizes at least two heads within the left periphery (but see the footnote 30, this page). These two heads are expressed separately in Czech. Secondly, if the formal affinities between complementizers and pronominals have a morphosyntactic explanation, then we should expect some functional correlations. Consider the case of the Czech invariant demonstrative to (see the examples in (68) and the discussion around it). In a paper of mine (Šimík 2007a) I argue that this demonstrative actually realizes a low CP-head, namely Foc. This element, even though lacking -features and case, keeps a part of the

39 Thus, the English declarative complementizer that arguably spells-out at least the Force > (Top > Foc >) Fin sequence, since it clearly determines the declarative force of the embedded clause and at the same time can only introduce finite clauses. Note that the assumption that English complementizers spell-out a relatively big span of functional categories would also explain why English has no information structure heads and only very limited discourse related movements.

Page 35: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

34

function that it has as a full-fledged definite demonstrative pronoun. In particular, it presupposes its complement—a property of definite articles. 5 Conclusion I presented a proposal for investigating the structure and syntactic distribution of wh-expressions. The overarching hypothesis is that they are syntactically (and hence semantically) complex structures. Their complexity provides the source of ambiguity, in the sense that one and the same wh-expression may spell-out a different piece of structure, with correspondingly different syntactic distribution and semantic interpretation. Below, I summarize the main hypotheses pronounced in the proposal and suggest how they can be falsified or further elaborated. (i) The functional hierarchy in the left periphery (or its part) corresponds to what could

be called an operator-hierarchy (as proposed in section 2.3). This hierarchy is designed to capture the fine typology of operators, ranging from existential to relative operators. This is more of a background hypothesis, assumed by many other authors, some of whom I cited above. At the moment, I am unable to provide a set of criteria under which this hypothesis would be falsified. My choice to follow this hypothesis is motivated more or less by conceptual reasons.

(ii) Non-operator wh-phrases are unselectively bound by (existential) quantifiers which

have a fixed position in the operator-hierarchy. I proposed three criteria to delimit unselective binding with respect to movement: (i) distinct locality conditions; (ii) more liberal positioning of the bound variable; (iii) restrictions on the referential properties of the bound variables. My task is to find further criteria for distinguishing the two techniques of creating A-bar dependencies. Further, I hope to provide an account of the observed restrictions associated with unselective binding, both in wh-in situ languages like Chinese (non-referentiality) as well as in Czech and German (non-referentiality plus complexity). Still a more general question is why wh-words in German and Czech can only be unselectively bound by non-A-bar operators.

(iii) When wh-phrases A-bar move, they may move more than once, i.e. operators are

syntactically complex (they occupy more than one position). These movements are expected to correlate with the proposed operator-hierarchy and are semantically non-vacuous. This hypothesis predicts that there are more phenomena of the Czech type (described in section 4.1). A plausible candidate is Dutch and its complex complementizers (Zwart 2000). Whereas the evidence on the PF side seems to be convincing, it is important to come up with a plausible semantic analysis which would support these partial movements. A further task is to explore the hypothesis that intermediate movement in cases of long A-bar extraction is semantically non-vacuous. This would be proved e.g. if it were shown that intermediate occurrences behave like operators.

(iv) One and the same wh-word, within one language, may spell-out different syntactic

structures (different operators), respecting the operator-hierarchy.

Page 36: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

35

Proving this hypothesis is strongly derivative of the proposed hierarchy. Then, the task is to look for languages, where the lexicalization (spell-out) of wh-words seems to support/violate this hierarchy.

(v) Complementizers are syntactically and morphologically underspecified (pro)nominal

structures. Complementizer-like are often of pronominal nature but generally lack -features (section 4.2). It is important to assign -features their place in a (pronominal) functional head hierarchy and then see what it means for an element to lack them (are they stripped “from above” or “from below”?). The actual “category” of the complementizer is expected to reflect some of its canonical properties (operator-properties with wh-words, definiteness properties with D-words).

Page 37: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

36

Appendix A Activities first year A.1 Attended schools and conferences LOT winter school Nijmegen (January 2007) http://www.let.uu.nl/LOT/GraduateProgram/LotSchools/Nijmegen%20winterschool%202007/index07.html E. Woolford Case and agreement H. Broekhuis Derivations and evaluations: Object shift in the Germanic languages R. Jonkers Neurolinguistic aspects of aphasia LOT summer school Leuven (June 2007) http://www.lotschool.nl/GraduateProgram/LotSchools/Leuven%20summerschool%202007/index07.html A. ter Meulen Discourse analysis G. van den Wijngaerd Binding J. van Craenenbroeck Variation in ellipsis J. Lidz Learning and parsing in the acquisition of syntax and semantics E. Williams Issues in morphosyntax Alternatives to cartography Brussels (June 2007) http://www.kubrussel.ac.be/onderwijs/onderzoekscentra/crissp/bcgl/2007/index.htm EGG summer school Brno (August 2007) http://egg.auf.net/ U. Sauerland Semantics of embedded clauses E. Herburger All you wanted to know about negation and negative polarity T. Biberauer Word-order variation, optionality, and gaps M. Noonan The structure of PP O. Koeneman Syntactic doubling and the structure of chains Verb movement as a structure-creating operation H. Zeijlstra The meaning of being meaningless: the how and why of uninterpretable features A.2 Talks 2006/09/20 syntax circle Groningen The morphosyntax of the foot of relative dependency 2006/11/23 Syntax AiO Meeting III (Tilburg) Why the hell what? Remarks on the syntax and semantics of ‘why’ and ‘what’ in Czech 2006/12/01 syntax circle Leipzig (Germany) Why the hell what? A remark on the syntax and semantics of ‘why’ and ‘what’ in Czech 2007/01/13 ConSOLE XV (Brussels, Belgium)

Page 38: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

37

The sources and consequences of wh-morphology in (English) relatives and questions 2007/05/16 International Meeting of Young Linguists VIII (Olomouc, Czech Republic) Co dokáže co: Interakce vidu a p�í�inných otázkových slov v �eštin� [What can what do: The interaction of aspect and causal interrogatives in Czech] 2007/06/08 TabuDag (Groningen) Czech and Russian have an overt Foc 2007/07/28 Czech(s) in generative grammar workshop (Brno, Czech Republic) Czech and Russian have an overt Foc A.3 Papers Specificity in (Czech) relative clauses. To appear in Papers and Studies in Contemporary

Linguistics 42. Poznan. (Proceedings of Poznan Linguistic Meeting 2006) The source of wh-morphology in questions and relative clauses. To appear in Proceedings of

ConSOLE XV. Leiden. The Czech invariant demonstrative to is a Foc head. To appear in a festschrift for Lída

Veselovská. A.4 Other academic activities Taking part at the ConSOLE conference includes reviewing a paper. I reviewed a paper by Ourania Sinopoulou: Sinopoulou, Ourania. In press. Multiple questions and apparent wh-in-situ: Evidence from Greek.

To appear in Proceedings of ConSOLE XV. Leiden. Together with Marlies Kluck, I initiated a reading group for PhD students associated with the syntax circle of CLCG. The meetings took place once in two weeks. We intend to continue next semester. A.5 Reading and orientation in the relevant literature Reading as well as gathering and organizing relevant references and data has been the most time-consuming activity of the first year. A special external appendix includes a summary of what I have done (to be added).

Page 39: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

38

Appendix B Plans for the coming years B.1 Rough chronological and topical schedule By the end of the second year, I would like to have collected a fair amount of data which enable me to state an empirical generalization concerning the ambiguity of wh-expressions. The generalization should be based on a limited set of languages, say Czech, German, English, and Dutch, and should be translated to a meaningful theoretical claim, within the domain of inquiry which I outlined above. The third year should be devoted to collecting supportive as well as problematic evidence. During the fourth year, I would like to write the dissertation. As a part of forming a falsifiable theoretical claim, I have to achieve a better orientation in the theoretical positions that have been taken in the literature. Reading will thus form an important part of my research. Furthermore, I would like to work out detailed analyses of some project-related phenomena, working on Czech from a comparative point of view (mainly Germanic and other Slavic languages). Below, I list some of my reading and writing plans. Second year reading • A-bar dependencies (Cinque 1990, Cheng 1991) • Theories of quantification (Barwise and Cooper 1981, Kratzer 2005) • Information structure issues (Vallduví 1992, Büring 2003, Aboh 2004, Brunetti 2004) • The morphosyntactic structure of pronominals, R-pronouns, demonstratives, wh-forms, etc.

(Dechaine and Wiltschko 2002, Bernstein 2004, Noonan 2005) Second year production • Czech irrealis free relatives • The invariable demonstrative in appositions, appositive relatives, and exclamations • The morphosyntax of wh-based quantifiers Third year reading • Multi-step wh-movement (Boeckx and Grohmann 2004, Boeckx and Stjepanovic 2005) • Left-branch extraction in Slavic languages (Corver 1990, Boškovic 2005) • The variability of placing quantification-morphemes (Hagstrom 1998, Matthewson 2001) • Semantics of maximalization (von Stechow 1984, Jacobson 1995, Rullmann 1995) Third year production • Movement in a multi-step fashion • The decomposition of complementizer heads B.2 Schools I plan to continue attending LOT summer/winter schools since they have proven to be very educative and informative. The courses I have attended so far were generally very inspiring even if they did not match my research interests fully. I will also consider attending some other summer schools, e.g. the EGG summer school or ESSLLI summer school (European Summer School in Logic, Language, and Information).

Page 40: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

39

B.3 Talks and papers At this point, I have two concrete plans: 2007/10/19 Workshop on parenthesis and sentence amalgamation Groningen Czech appositions and appositive relatives [preliminary name] 2007/11/30 Formal Description of Slavic Languages Leipzig (Germany) Slavic invariable demonstrative to/eto is Foc [abstract accepted for presentation] B.4 A semester abroad I consider spending a semester abroad, probably during the third year of my study program. I find it important to gain more experience and to some systematic feedback from linguists other than domestic. The university/institute where I would like to go remains unclear at this point. This should be decided towards the end of my second year.

Page 41: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

40

References

Abels, Klaus. 2002. Successive cyclicity, anti-locality, and adposition stranding. PhD dissertation. Aboh, Enoch. 2004. The morphosyntax of complement-head sequences. New York: Oxford

University Press. Aoun, Joseph and Yen-hui Audrey Li. 2003. Essays on the representational and derivational

nature of grammar: The diversity of wh-constructions. Cambridge, MA: The MIT Press. Baker, Mark. 1988. Incorporation: A theory of grammatical function changing. Chicago:

University of Chicago Press. Barbiers, Sjef, Olaf Koeneman, and Marika Lekakou. 2007. 'Syntactic doubling and the structure

of chains.' A paper presented at WCCFL 26, Berkeley. Barwise, Jon and Robin Cooper. 1981. 'Generalized quantifiers and natural language.' Linguistics

and Philosophy 4, 159–219. Berman, Stephen. 1991. On the semantics and logical form of wh-clauses. PhD dissertation.

University of Massachusetts, Amherst. Bernstein, J. B. 2004. 'English th-forms.' A paper presented at the Copenhagen Symposium on

Determination. Boeckx, Cedric. 2003. Islands and chains: Resumption as stranding. Philadelphia, PA: John

Benjamins Publishing Company. Boeckx, Cedric and Kleanthes K. Grohmann. 2004. 'SubMove: Towards a unified account of

scrambling and D-linking.' D. Adger, C. de Cat, and G. Tsoulas (eds). Peripheries: Syntactic edges and their effects, 241–257. Dordrecht: Kluwer.

Boeckx, Cedric and Sandra Stjepanovic. 2005. 'The wh/clitic connection.' L. Heggie and F. Ordonez (eds). Clitics and affixation, 301–314. Amsterdam: John Benjamins.

Boškovic, Željko. 1998. 'Wh-phrases and wh-movement in Slavic.' Universtiy of Connecticut, Storrs.

Boškovic, Željko. 2005. 'On the locality of left branch extraction and the structure of NP.' Studia Linguistica 59, 1–45.

Brunetti, Lisa. 2004. A unification of focus. Padova: Unipress. Büring, Daniel. 2003. 'On D-trees, beans, and B-accents.' Linguistics and Philosophy 26, 511–

545. Caha, Pavel. 2007. 'The superset principle and the spell out of non-terminals: Don't try

morphological analysis in the privacy of your own morphology.' S. Blaho, L. Vicente, and E. Schoorlemmer (eds). Proceedings of Console XV, xx–xx.

Caponigro, Ivano. 2003. Free not to ask: On the semantics of free relatives and wh-words cross-linguistically. PhD dissertation. University of California, Los Angeles.

Cheng, Lisa Lai-Shen. 1991. On the typology of wh-questions. PhD dissertation. MIT, Cambridge, MA.

Chierchia, Gennaro. 1993. 'Questions with quantifiers.' Natural Language Semantics 1, 181–234. Chomsky, Noam. 1973. 'Conditions on transformations.' S. Anderson and P. Kiparsky (eds). A

festschrift for Morris Halle. New York: Holt, Rinehart and Winston. Chomsky, Noam. 1977. 'On wh-movement.' P. W. Culicover, T. Wasow, and A. Akmajian (eds).

Formal syntax, 71–132. New York: Academic Press. Chomsky, Noam. 1982. Some concepts and consequences of the theory of government and

binding. Cambridge, MA: The MIT Press. Chomsky, Noam. 1999. 'Derivation by phase.' MIT Occasional Papers in Linguistics 18,

Cambridge, MA: The MIT Press.

Page 42: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

41

Chomsky, Noam. 2000. 'Minimalist inquiries: The framework.' R. Martin, D. Michaels, and J. Uriagereka (eds). Step by Step, 89–155. Cambridge, MA: The MIT Press.

Cinque, Guglielmo. 1990. Types of A'-dependencies. Cambridge, MA: The MIT Press. Citko, Barbara. 2004. 'On headed, headless, and light-headed relatives.' Natural Language &

Linguistic Theory 22, 95–126. Corver, Norbert. 1990. The syntax of left branch extractions. Doctoral dissertation. Tilburg

University. Dechaine, Rose-Marie and Martina Wiltschko. 2002. 'Decomposing pronouns.' Linguistic Inquiry

33, 409–442. Engdahl, Elisabet. 1986. Constituent questions. Dordrecht: Kluwer. Ginzburg, Jonathan. 1995. 'Resolving questions, II.' Linguistics and Philosophy 18, 567–609. Ginzburg, Jonathan and Ivan A. Sag. 2000. Interrogative investigations: The form, meaning, and

use of English interrogatives. Stanford: CSLI Publications. Grohmann, Kleanthes K. 2006. 'Top issues in questions: Topics–Topicalization–Topicalizability.'

L. L.-S. Cheng and N. Corver (eds). Wh-movement: Moving on, 249–288. Cambridge, MA: The MIT Press.

Grosu, Alexander. 1994. Three studies in locality and case. London: Routledge. Grosu, Alexander. 2002. 'Strange relatives at the interface of two millennia.' Glot International 6,

145–167. Grosu, Alexander and Fred Landmann. 1998. 'Strange relatives of the third kind.' Natural

Language Semantics 6, 125–170. Hagstrom, Paul. 1998. 'Decomposing questions.' MIT Working Papers in Linguistics . Hagstrom, Paul. 2003. 'What questions mean.' Glot International 7, 188–201. Hamblin, C. L. 1958. 'Questions.' The Australasian Journal of Philosophy 36, 159–168. Heim, Irene. 1982. The semantics of definite and indefinite noun phrases. PhD dissertation.

University of Massachusetts, Amherst. Higginbotham, James. 1980. 'Pronouns and bound variables.' Linguistic Inquiry 11, 679–708. Horvath, Julia. 1986. Focus in the theory of grammar and the syntax of Hungarian. Dordrecht:

Foris. Huang, C. T. J. 1982. Logical relations in Chinese and the theory of grammar. PhD dissertation.

MIT, Cambridge, MA. Jacobson, Pauline. 1988. 'The syntax and semantics of free relatives in English.' Paper presented

at the LSA winter meeting. New Orleans. Jacobson, Pauline. 1995. 'On the quantificational force of English free relatives.' E. Bach, E.

Jelinek, A. Kratzer, and B. Partee (eds). Quantification in natural languages, 451–486. Dordrecht: Kluwer.

Kamp, J. A. W. 1981. 'A theory of truth and semantic representation.' J. Groenendijk, T. Janssen, and M. Stockhof (eds). Formal methods in the study of language, 277–321. Amsterdam: Mathematical Centre.

Kayne, Richard S. 2005. 'A short note on where vs. place.' Manuscript. New York University. Kratzer, Angelika. 2005. 'Indefinites and the operators they depend on: From Japanese to Salish.'

G. N. Carlson and F. J. Pelletier (eds). Reference and quantification: The Partee effect. Stanford: CSLI Press.

Lasnik, Howard and Timothy Agnus Stowell. 1991. 'Weakest crossover.' Linguistic Inquiry 22, 687–720.

Lewis, David. 1975. 'Adverbs of quantification.' E. Keenan (ed). Formal semantics of natural language. Cambridge: Cambridge University Press.

Page 43: The ambiguity of wh- · PDF fileThe ambiguity of wh-expressions First year report Radek Šimík CLCG Department of Linguistics University of Groningen r.simik@rug.nl

42

Matthewson, Lisa. 2001. 'Quantification and the nature of crosslinguistic variation.' Natural Language Semantics 9, 145–189.

May, Robert. 1985. Logical form: Its structure and derivation. Cambridge, MA: The MIT Press. McCloskey, James. 2002. 'Resumption, successive cyclicity, and the locality of operations.' S. D.

Epstein and T. D. Seely (eds). Derivation and explanation in the minimalist program, 184–226. Oxford: Blackwell Publishers.

Neeleman, Ad and Hans van de Koot. 2006. 'Theta theory: In defense of plan B.' Lectures at the EGG summer school in Wroclav, Poland, 2005. To appear as a book.

Noonan, M. 2005. 'R-particles.' GLOW Newsletter 54. Pesetsky, David. 1987. 'Wh-in-situ: Movement and unselective binding.' E. Reuland and A. ter

Meulen (eds). The representation of (in)definiteness, 98–129. Cambridge, MA: The MIT Press.

Rett, Jessica. 2007. 'Exclamatives are degree constructions.' Manuscript. Rutgers University. Richards, Norvin. 2001. Movement in language: Interactions and architectures. Oxford: Oxford

University Press. Rizzi, Luigi. 1989. Relativized minimality. Cambridge, MA: The MIT Press. Rizzi, Luigi. 1997. 'The fine structure of the left periphery.' L. Haegeman (ed). Elements of

grammar: A handbook of generative syntax, 281–337. Dordrecht: Kluwer Academic Publishers.

Rizzi, Luigi. 2006. 'On the form of chains: Criterial positions and ECP effects.' L. L.-S. Cheng and N. Corver (eds). Wh-movement: Moving on, 97–133. Cambridge, MA: MIT Press.

Ross, John Robert. 1967. Constraints on variables in syntax. Doctoral dissertation. MIT, Cambridge, MA.

Rullmann, Hotze. 1995. Maximality in the semantics of wh-constructions. PhD dissertation. University of Massahussetts, Amherst.

Safir, Ken. 1986. 'Relative clauses in a theory of binding and levels.' Linguistic Inquiry 17, 663–689.

Šimík, Radek. 2007a. 'The Czech invariant demonstrative to is a Foc head.' Manuscript. University of Groningen.

Šimík, Radek. 2007b. 'The source of wh-morphology in questions and relative clauses.' To appear in Proceedings of ConSOLE XV, Leiden.

Starke, Michal. 2001. Move dissolves into Merge: A theory of locality. PhD dissertation. Sturgeon, Anne. 2005. The syntax and pragmatics of contrastive topic in Czech. PhD dissertation.

UC Santa Cruz, CA. Tsai, Wei-tien Dylan. 1994. On economizing the theory of A-bar dependencies. PhD dissertation.

MIT, Cambridge, MA. Vallduví, Enric. 1992. The informational component. New York: Garland. van Craenenbroeck, Jeroen. 2004. Ellipsis in Dutch dialects. Utrecht: LOT. von Stechow, Arnim. 1984. 'Comparing semantic theories of comparison.' Journal of Semantics 1,

1–77. Zanuttini, Raffaella and Paul Portner. 2003. 'Exclamative clauses: At the syntax semantics

interface.' Language 79, 39–81. Zeijlstra, Hedde. 2007. 'The meaning of being meaningless: The how and why of uninterpretable

features.' Lectures presented at the EGG summer school, Brno, August 2007. Zwart, Jan Wouter. 2000. 'A head raising analysis of restrictive relative clauses in Dutch.' A.

Alexiadou, P. Law, A. Meinunger, and C. Wilder (eds). The syntax of relative clauses, 349–385. Amsterdam: John Benjamins Publishing Company.