The Left Periphery
Linguistik Aktuell/Linguistics Today (LA) Linguistik Aktuell/Linguistics Today (LA) provides a pla...
33 downloads
626 Views
2MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
The Left Periphery
Linguistik Aktuell/Linguistics Today (LA) Linguistik Aktuell/Linguistics Today (LA) provides a platform for original monograph studies into synchronic and diachronic linguistics. Studies in LA confront empirical and theoretical problems as these are currently discussed in syntax, semantics, morphology, phonology, and systematic pragmatics with the aim to establish robust empirical generalizations within a universalistic perspective.
General Editors Werner Abraham
University of Vienna / Rijksuniversiteit Groningen
Elly van Gelderen
Arizona State University
Advisory Editorial Board Cedric Boeckx
Christer Platzack
Guglielmo Cinque
Ian Roberts
Günther Grewendorf
Lisa deMena Travis
Liliane Haegeman
Sten Vikner
Hubert Haider
C. Jan-Wouter Zwart
Harvard University University of Venice
J.W. Goethe-University, Frankfurt University of Lille, France University of Salzburg
University of Lund
Cambridge University McGill University
University of Aarhus University of Groningen
Volume 129 The Left Periphery. The interaction of syntax, pragmatics and prosody in Czech by Anne Sturgeon
The Left Periphery The interaction of syntax, pragmatics and prosody in Czech
Anne Sturgeon
John Benjamins Publishing Company Amsterdam / Philadelphia
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data Sturgeon, Anne. The left periphery : the interaction of syntax, pragmatics and prosody in Czech / by Anne Sturgeon. p. cm. (Linguistik Aktuell/Linguistics Today, issn 0166-0829 ; v. 129) Includes bibliographical references and index. 1. Czech language--Syntax. 2. Czech language--Discourse analysis. 3. Pragmatics. I. Title. PG4361.S78 2008 491.8'65--dc22 isbn 978 90 272 5512 9 (Hb; alk. paper)
2008019876
© 2008 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
Dedicated to
Zelma Molnár Kline 1905–2005
Table of contents Acknowledgements
chapter 1 Introduction 1. Left dislocation in Czech 1 2. Previous research on Czech 3 3. Organization 4 chapter 2 The syntax of the left periphery 1. Introduction 7 2. Clause structural assumptions 8 3. Identifying the clause-initial position: Clitics in I0 10 4. Locating the lexical verb in v0 11 4.1 VP adverbs 11 4.2 VP ellipsis 14 5. The left periphery: Discoursal-syntactic mapping 16 5.1 Discourse function and [Spec, IP] 16 5.2 V0-to-I0 raising: Satisfying the EPP 20 6. Discoursally motivated A-bar movement 23 6.1 Wh-movement 23 6.2 Focus 27 6.3 Contrastive topicalization 27 6.3.1 Weak Crossover 30 6.3.2 Operator-variable contrastive topicalization 32 6.3.3 Syntactic analysis of CT constructions 33 7. Another look at multiple Wh-questions 34 7.1 Multiple pre-clitic XP’s 35 7.2 The position of non-initial wh-phrases 37 8. Conclusion 37 chapter 3 The syntax of left dislocation 1. Overview of left dislocation in Czech 39 2. Grohmann 2003 and left dislocation in German 41 3. Syntactic approach to left dislocation in Czech 43
xi
1
7
39
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
4.
5.
6. 7. 8.
9.
10.
Evidence for a movement account of CLD 44 4.1 Conditions A and C 45 4.2 Quantifier binding 46 Evidence for non-movement analyses of HTLD I, II 47 5.1 Reconstruction and HTLD I 47 5.2 Reconstruction and HTLD II 49 5.3 Conclusion 50 Gender and resumption 50 Phrasal categories 53 Evidence for A-bar movement of resumptive pronouns 56 8.1 Long distance dependencies 56 8.2 Island sensitivity 58 Syntactic positions of CLD and HTLD I, II 60 9.1 The left periphery 60 9.2 Embedded contexts 62 9.3 The syntax of HTLD 64 Conclusion 66
chapter 4 67 The discourse functions of left dislocation 1. Introduction 67 2. Contrastive topic 68 2.1 Prince 1981, 1997 70 2.2 Hajičová and Sgall 2004 71 2.3 Büring 2003 73 2.3.1 Discourse trees 74 2.3.2 CT congruence 75 2.3.3 Focus on clausal polarity 78 2.3.4 Conventional meanings vs conversational implicature 79 3. CLD as contrastive topic marking 81 3.1 Overview of the discourse function of CLD 81 3.2 Conversational implicature vs. conventional meaning 86 3.3 Conclusion 88 4. The discourse function of HTLD 88 4.1 Overview of the requirements of “topic promotion’’ 90 4.2 The preceding context 91 4.3 Perseveration in the discourse 93 4.4 Considering HTLD II 95 4.5 Conclusion 96 5. Conclusion 96
Table of contents
chapter 5 The syntax of resumption 1. Introduction 99 2. CLD: Locating the CT interpretation in [Spec, IP] 99 2.1 Prosodic realization of contrastive topic 100 2.2 Comparing CLD and HTLD II 104 2.3 Interpretation of [Spec, TopP] 105 3. The syntax of CLD 107 4. Resumption at the syntax-phonology interface 109 4.1 V(P) fronting in Hebrew 110 4.2 PF requirements on spell out in Czech CLD 113 4.2.1 Pronouncing the top copy 113 4.2.2 Pronouncing the resumptive 114 5. Spelling out the resumptive pronoun 116 5.1 Resumption as partial spell out of movement copies 116 5.1.1 Grammatical gender matching 118 5.2 Why a demonstrative pronoun? 118 6. Resuming other categories 120 6.1 Properties and propositions 121 6.2 Manner, place and time adverbials 123 6.3 Prepositional phrases 124 6.4 Conclusion 125 7. Previous accounts of resumptive spell out 125 8. Conclusion 127 chapter 6 Conclusion 1. Overview 129 2. The syntax and pragmatics of left dislocation in Czech 129 2.1 Movement vs. non-movement derivations 130 2.2 The pragmatics of left dislocation 130 3. The syntax of resumption 131 4. Conclusion 131 References Index
99
129
133 141
Acknowledgements The most important contributor to my development as a linguist and the final product of this work is my advisor, Judith Aissen. She has always encouraged me to engage in field research and to be creative in my solutions to linguistic puzzles. Through her own work and her excellence as a teacher and advisor, I learned to think and write precisely. Her generosity with her time and her ability to carefully consider drafts is legendary and this book benefited significantly from her input. My other committee members also contributed greatly to my dissertation. James McCloskey provided important insight into the formal aspects of the analysis of resumption. I thank Donka Farkas for her input on the pragmatics of contrastive topic. The final committee member, Eva Hajičová, hosted me at the Institute of Formal and Applied Linguistics at Charles University in Prague in Fall, 2003. Much of the data that provided inspiration for this project was gathered and analyzed during that time. I thank Eva Hajičová and Petr Sgall for the time they gave me for consultation, as well as their long-distance support during the research and writing of this work. This work would, quite literally, not have been possible without the help of my Czech consultants: Jana Hejduková, Věra Kafková, Tomáš Koukolík, Jitka Malíková, Hana Pariser, Jakub Šíma, Karolina Wencelová, and employees at Café Prague in San Francisco. I especially thank Martina Šímová and Kristina Valendinová for their regular help over the past few years. Their patience and careful consideration of the data made this project not only possible, but enjoyable.
chapter 1
Introduction 1. Left dislocation in Czech Czech has two main types of left dislocation which differ in their syntax, discourse function and prosody: Contrastive Left Dislocation (CLD) and Hanging Topic Left Dislocation (HTLD). I provide a unified syntactic account which is motivated by the pragmatics and prosody of these constructions. In both types of left dislocation, there is a left dislocated constituent, separated from the following clause by an intonational break. As in left dislocation generally, this constituent binds a resumptive pronoun. One distinctive feature of left dislocation in Czech is that, when the resumptive is a full pronominal element, it obligatorily appears at the left edge of the clause and is associated with a gap somewhere within the clause. Czech left dislocation, then, involves three linked elements: the left dislocate, the resumptive and the gap. The examples in (1) and (2) illustrate the two types of left dislocation, with the left dislocate, the resumptive and the gap underlined. In HTLD, there is no case matching between the left dislocate and the resumptive element: the dislocate is always in the nominative case, (1).1 (1) Anička1? Té1 se nic nestalo t1. HTLD Anička.nom that.dat refl-cl nothing neg-happened “Anička? Nothing happened to her.” (Czech National Corpus2)
In CLD, the left dislocate matches the resumptive element in case, see (2).
. Abbreviations in the book are as follows: c (complementizer), fut (future), pst (past), inf (infinitive), cond (conditional), aux (auxiliary), excl (exclamative), cl (clitic), refl (reflexive), rel (relative pronoun), prt (particle), fem (feminine), masc (masculine), neut (neuter), sg (singular), pl (plural), acc (accusative), dat (dative), nom (nominative), instr (instrumental), loc (locative), dir (directional). . The Czech National Corpus is an online corpus of over 100 million words collected from online texts in a variety of styles (novels, newspaper articles, magazine articles, academic journals). It was developed and is managed by researchers at Charles University (Prague, Czech Republic). It can be found online at: http://ucnk.ff.cuni.cz/.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(2)
Ale tu dívku1 … tu1 znám t1 určitě. CLD but that girl.acc that.acc know.1sg for-sure “But that girl … I know her for sure…” (http://www.cswu.cz/jediland/ffcz/sunny/pochopitlez11.htm)
Research on left dislocation, beginning with Ross 1967, 1973; Rodman 1974/1997; and van Riemsdijk and Zwarts 1974/1997,3 has focused on two questions: whether the left dislocate is base-generated at the left periphery or moves there and how the resumptive is introduced into the structure. Ross 1973 was among the first to note that left dislocation is not a unitary construction. In some cases, German left dislocation, for instance, there is evidence from case matching that a tight syntactic connection exists between the clause-internal gap and the left dislocate, but in others, such as English, there is no such evidence. To account for case matching in German, Ross suggested that the left dislocate moves from a clause-internal position to a position at the left edge of the clause. Van Riemsdijk and Zwarts 1974/1997 discuss connectivity as another test for determining the appropriate analysis of left dislocation in Dutch (see also Cinque 1977 for Italian). In left dislocation constructions which exhibit connectivity, a movement analysis is warranted, in others, a base generation account is needed. I provide evidence that Czech has both types of left dislocation: one involving movement of the dislocate to its surface position and one involving base-generation. In CLD, there is a close syntactic connection between the left dislocate and the clause-internal gap: this is evident in (2) which exhibits case matching. Such a connection is absent in HTLD, which exhibits no case-matching between the left dislocate and the resumptive or gap, see (1). I suggest that CLD is generated through movement, HTLD through base generation. Resumption in left dislocation constructions involving movement is theoretically problematic. Resumptives were inserted transformationally in the early literature (Ross 1967, 1973; Emonds 1970; among others). In the 1990’s, machinery available in Government and Binding, operator-variable chains in particular, motivated researchers to posit chain formation accounts of left dislocation. Wiltschko 1997 and Anagnostopoulou 1997 consider CLD in languages such as German and Dutch in which the resumptive element has moved to a position at the left edge of the clause and is related to a left peripheral constituent by an operator-variable chain (for other chain formation accounts see also Demirdache 1991, among others).
. Seminal papers on left dislocation from the 1970’s and 1980’s, as well as more recent work, are collected in Materials on Left Dislocation (eds. Elena Anagnostopoulou, Henk van Riemsdijk & Frans Zwarts 1997).
Chapter 1. Introduction
Anagnostopoulou 1997, for example, suggests that the demonstrative pronoun is an operator which moves to the left edge of the clause and enters into a predication relationship with the left peripheral constituent in CLD constructions. The chain formed between the left dislocate and the resumptive element licenses the connectivity effects found in CLD constructions. With the advent of the Copy and Delete Theory of movement (Chomsky 1995), it became possible to account for these connectivity effects with movement, avoiding the machinery associated with chain formation. The most current work on left dislocation, Grohmann 2003, builds on previous work in Ross 1967, Cinque 1977 and Vat 1981/1997 and revives a movement analysis of left dislocation to account for the connectivity effects in German CLD. The left dislocate first moves to a position at the left edge of the clause, then this element undergoes a subsequent movement to a higher position in the left periphery. The possibility of reconstruction accounts for connectivity effects. An analysis of this type, however, poses questions for resumption. Under a chain formation account, the resumptive is the head of a movement chain, and is pronounced for that reason. Under a movement account, though, the resumptive element is not the head of a movement chain. Rather, it is a movement copy which would be expected to remain unpronounced at PF. I consider this question and provide an answer at the syntax-prosody interface.
2. Previous research on Czech A central goal of researchers in the Praguian tradition is to investigate and describe the ways languages encode information structure. For Czech, this involves both word order and intonation. Work on Czech has principally focused on Topic Focus Articulation which has been implemented within the Functional Generative Description framework developed in the 1960’s by Petr Sgall (for developments of this theory see Sgall et al. 1973, 1980, 1986; Hajičová et al. 1998; among many others). The Topic Focus Articulation of the clause is indicated in tectogrammatical tree structures; dependency relations, morpho-logical information and thematic roles are also annotated in the trees. The point of intersection of my work on left dislocation and work within the Praguian tradition is the interpretation and syntactic representation of contrastive topic constructions in Czech. Hajičová et al. 2003; Hajičová and Sgall 2004; and Veselá et al. 2003 discuss the construction, illustrated in (3), in which an element with a contrastive topic interpretation appears at or near the left edge of the clause. Note that (3) contains no resumptive element. This is not an instance of left dislocation, but a simpler construction which I call contrastive topicalization.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(3) Domácím se dařilo ze začátku. Hostům se home-team refl-cl succeeded from beginning visitors refl-cl povědl až druhý poločas. succeeded only second half “The local team was successful at the beginning. The visitors succeeded only in the second half.” (Hajičová & Sgall 2004: (31))
Hajičová et al. 2003 suggest that the discourse referent of a contrastive topic must be a member of a contextually salient set of alternatives in the discourse; they relate this concept of alternatives to the “theme alternative” set of Steedman 2002. Here, I investigate a related construction which also exhibits a contrastive topic discourse function: CLD. Though most contemporary work on Czech is within Functional Generative Description, several researchers have also considered Czech syntax from a derivational syntactic perspective, primarily within the Government and Binding framework (see Veselovská 1995; Lenertová 2001; Kučerová 2007, among others). Veselovská’s 1995 dissertation considers issues in the clause structure of Czech, including the position of the lexical verb, the structure of the Czech DP, among other topics. I draw heavily from Veselovská’s work in chapter 2 when developing an analysis of Czech clausal structure. Lenertová 2001 considers the left periphery of the Czech clause and adopts a Rizzi 1997-style analysis with an “exploded” CP domain. In chapter 2, I present evidence that a more economical view of the left periphery in Czech is possible and propose a structural analysis in which the central left peripheral position is not a position in the C-domain, but, rather, a specifier within the inflectional domain, [Spec, IP]. 3. Organization The first goal of this work is to develop an understanding of Czech clause structure and the syntax of the left periphery. In chapter 2, I present evidence to support Veselovská’s 1995 claim that the lexical verb in Czech remains low within the clause, raising from V0 to v0, but no higher. The evidence comes from the position of vp adverbs and from vp Ellipsis. Under my account of the left periphery, it is unnecessary to posit an exploded CP (pace Lenertová 2001). In most cases, left peripheral positions are located exclusively in the inflectional domain, with [Spec, IP] functioning as the pivotal left-edge A-bar position. [Spec, IP] can be filled by XP’s instantiating one of three discourse functions: topic, contrastive topic, or focus (foci or wh-phrases). One result of this chapter is a novel approach to positioning non-initial wh-phrases. In Czech, there is evidence that they appear, not in the IP domain, but in the vp domain.
Chapter 1. Introduction
In chapter 3, I turn my attention to the syntax of CLD and HTLD. I start by introducing a third left dislocation construction, HTLD II, which has a second position clitic resumptive, rather than a left-edge full pronominal element. Connectivity evidence supports my claim that hanging topics are base generated in their left-peripheral position, while CLD’ed elements move to the left periphery from a clause-internal position. Additional evidence for the movement nature of CLD is found when the behavior of the resumptive element is considered. The resumptive in CLD is required to match the left dislocate in grammatical gender when there is a mismatch between the natural and grammatical gender of the dislocate. This is not the case with the non-movement HTLD constructions. Resumption in HTLD behaves like cross-sentential pronominalization; the resumptive can match the dislocate in either grammatical or natural gender. In chapter 4, I establish that CLD instantiates a contrastive topic discourse function while HTLD I, II serve to promote a discourse referent to topic status. These conclusions are based primarily on a corpus of approximately 100 attested textual examples of left dislocation collected from the Czech National Corpus and through Google searches, and secondarily on work with native speaker consultants. My analysis of CLD is framed within the Büring 2003 formal pragmatic analysis of contrastive topic. In the case of HTLD, I follow Gregory and Michaelis’s 2001 methodology in classifying the discourse referents of hanging topics as promoted topics. The necessary components are now in place to develop a syntactic analysis of resumption in CLD that links leftward movement of the dislocate in CLD to the pragmatics of the construction. In chapter 5, I develop an analysis in which the dislocated constituent moves first to [Spec, IP] and then onto a higher position in the left periphery, [Spec, TopP]. Both movements are motivated by the pragmatics of the construction. In order to predict spell out of the intermediate copy of the left dislocate, the resumptive, I adopt an approach similar to Landau 2005b in which spell out of the intermediate copies is motivated in the PF component of the grammar. The copy of the dislocate in [Spec, IP] spells out as the resumptive element due to the prosodic requirements of the contrastive topic discourse function of the construction. Concluding remarks are found in chapter 6.
chapter 2
The syntax of the left periphery 1. Introduction Though SVO in unmarked utterances, Czech is primarily a discourse configurational language in which structural positions at the left and right edges of the clause are identified with specific discourse functions. The left edge is associated with continuing topic, contrastive topic and narrow focus, the right edge with both wide and narrow focus. In the left periphery, two structural positions serve to delineate important domains within the clause: the position of the lexical verb in the vP domain (v0) and the inflectional head (I0). The first part of this chapter provides evidence that supports a low position of the lexical verb; lexical verbs are shown to head the lowest vP projection. Once the position of the lexical verb has been established, we turn to the inflectional head which hosts second position clitics. This position delimits the left peripheral A-bar position ([Spec, IP]) which hosts XP’s instantiating continuing topic, contrastive topic and focus discourse functions, as well as most wh-expressions. My approach to the syntax of the left periphery is similar to that of Rizzi 1997 and Puskás 2000 in that I assume that XP’s associated with a topic, a contrastive topic or a focus discourse function, as well as wh-expressions, are found in structural positions at the left periphery (for a similar approach to Czech see Lenertová 2001). I depart from these approaches, however, in that, with the exception of dislocated XP’s, all these elements are found primarily within the IP projection. Not only is an exploded CP not necessary for Czech, the CP projection hosts only wh-expressions (and even then only under certain circumstances); [Spec, IP] is the target left peripheral position for leftward A-bar movements. It may appear that I have simply moved the various elements down one projection. However, I argue that this is not the case. Given the low position of the verb in Czech (it heads the lowest vP projection) and the fact that only one left peripheral XP is possible, it is not necessary to posit an exploded CP or to even make use of the CP projection in most cases. Independent evidence suggests that languages require both an inflectional and a verbal domain. If no additional functional projections are regularly needed by the language, it is preferable to avoid them for reasons of economy of representation (see Chomsky 1991).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
2. Clause structural assumptions I adopt standard assumptions about the verbal domain: internal arguments of the verb are base generated within the VP and the external argument is located in the specifier position of a dominating functional projection, vP. The head of this projection, v0, licenses structural accusative case on verbal objects within the VP. In the following sections I provide evidence for several other assumptions. First, the verb remains low in the syntax; as in English, it raises from V0 to v0 but no higher. Second, nominative case is valued through static Agree between the external argument in [Spec, vP] and I0. Third, [Spec, IP] is not a position associated with case licensing, but, rather, it is an A-bar position associated with an EPP feature. This position is filled in one of two ways: by raising the highest XP within the vP domain or by attracting an XP with a particular discourse function (contrastive topic or focus). Consider the unmarked transitive sentence in (1). (1) Petr políbil Hanu. Petr kissed Hana.acc “Petr kissed Hana.” IP
(2) DP1 Petr
I′ I []
vP DP t1
v′ v2 políbil (kissed)
VP V t2
DP Hanu
The derivation in (2) proceeds as follows. The DP, Hanu, Merges as a complement to the verbal head, políbil (“kiss.pst”). Since the verb is transitive, a vP shell dominates the VP, and the external argument, Petr, Merges into its specifier position. By assumption, the verb raises from V0 to v0. Φ-features on Petr are checked with the inflectional head, I0, via static Agree, and nominative case is licensed on that argument. The highest XP, the DP, Petr, raises to [Spec, IP] to check the EPP feature on I0. Once in [Spec, IP], this element is associated with a topic interpretation.
Chapter 2. The syntax of the left periphery
Another position of the internal argument of the verb, Hanu, is possible, with little change in interpretation. In (3), Hanu precedes the verb. (3)
Petr Hanu políbil. Petr Hana.acc kissed “Petr kissed Hana.” Lit: “Petr Hana kissed.”
Czech, like German, allows scrambling of arguments and adjuncts to a preverbal position in the middlefield. I assume that these elements are adjuncts to vP because, unlike elements in specifier positions, ordering between multiple middlefield XP’s is unrestricted. The tree for (3) is provided in (4). (4)
IP DP1 Petr
I′ I []
vP DP1 t1
vP DP3 Hanu
vP DP t1
v′ v2 políbil (kissed)
VP V t2
DP t3
Hanu and Petr both scramble to adjoined positions in the middlefield. Since Petr is in the highest adjoined position, it raises to [Spec, IP] to satisfy the EPP feature on I0. Additional evidence for this syntactic analysis of the Czech middlefield is provided in section 5.1. I assume a Copy and Delete theory of movement following Chomsky 1995. Movement is an operation composed of Agree and Remerge and takes place when there are uninterpretable features on the goal and/or the probe and the probe has an EPP feature. Moved elements leave behind a full copy in each position they occupy and, when movement is overt, lower copies remain unpronounced at PF. For illustrative purposes, I represent lower copies as traces in this chapter. However, the fact that full copies are present in lower positions plays a crucial role in the syntactic analysis of resumption in Contrastive Left Dislocation presented in chapter 5.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
3. Identifying the clause-initial position: Clitics in I0 Following the literature, I assume Czech clitics are positioned syntactically in I0 (for arguments supporting a syntactic, rather than a phonological, analysis see Fried 1994; Veselovská 1995 & Lenertová 2001). The clause-initial position ([Spec, IP]) is filled by an XP associated with one of three discourse functions: contrastive topic, topic or focus. Czech has verbal and pronominal clitics which are restricted in both their syntactic position and their relative ordering. We first turn to pronominal clitics. Clitic pronominal forms exist for two of the seven nominal cases: dative and accusative; for nominatives, null pronouns, rather than clitics, alternate with strong forms. Only strong pronouns are used to express the remaining cases: genitive, locative, instrumental and vocative. Reflexive clitics are also used and can be both anaphoric to argument positions and used with lexically reflexive verbs. Among pronominal clitics there is an obligatory ordering principle: reflexive > dative > accusative. In (5), a reflexive clitic is followed by dative and accusative clitics. (5) Nelíbí se mi to. neg-please refl-cl me.dat.cl it.acc.cl “I don’t like it.”
In addition to clitic pronominals, Czech has two types of clitic auxiliaries, both of which occur only with the past participle: clitic auxiliaries marking person in the past tense and those marking conditional mood. To express the past tense, the past participle occurs with a clitic auxiliary that marks person (jsem (“aux.1sg.cl”), jsi (“aux.2sg.cl”), jsme (“aux.1pl.cl”), jste (“aux.1sg. cl”)). This clitic is required for first and second person subjects, (6), but is absent with third person subjects. (6) Vy jste mluvili s Janou v pondělí. you.pl aux.2pl.cl talked.pst with Jana in Monday “You talked to Jana on Monday.”
Conditional mood is expressed by the combination of the past participle and a conditional auxiliary clitic which is inflected for 1st, 2nd and 3rd person (bych (“cond.1sg.cl”), bys (“cond.2sg.cl”), by (“cond.2sg.cl”), etc.), (7). (7) Chtěla bych si přečíst tu novou knížku. want.pst cond.1sg.cl refl-cl read.inf that new book “I would like to read that new book.”
Auxiliary clitics obligatorily precede pronominal clitics, see (8). (8) Řekla jsem mu to. told aux.1sg.cl him.dat.cl it.acc.cl “I told it to him.”
Chapter 2. The syntax of the left periphery
I assume that verbal clitics are base generated in I0. It is beyond the scope of the book to provide an analysis of pronominal clitic placement, though I assume they appear in I0 in the surface syntax.
4. Locating the lexical verb in v0 Now that basic assumptions about clause structure and the system of clitic pronouns and auxiliaries have been established, I consider the structure of the verbal domain. Evidence for a low position of the lexical verb and for positioning auxiliaries within the vP domain comes from three sources: from the position of VP adverbs, from the position of middlefield XP’s and from VP ellipsis (VPE). In VPE, the fact that only verbs which correspond to the English modals, but not lexical verbs, can be stranded points to a low position for lexical verbs. The position of the lexical verb is important because it is one of the positions that delimits the middlefield (the span between I0 and v0). 4.1 VP adverbs To demonstrate that the verb in Czech remains low, Veselovská 1995 appeals to work by Emonds 1978 and Pollock 1989. They diagnose the position of the verb in English and French using the structural position of VP adverbs such as often. VP adverbs adjoin to the highest projection of the VP domain, for me, the vP. Thus, if they precede the verb, the verb must remain within the verbal domain. If, on the other hand, a VP adverb can follow the verb, the verb must raise into the inflectional domain. Familiar evidence from English and French is given below. The VP adverb, often, appears before the verb in English, (9a), and after it in French, (9b); this suggests a low position for the English verb (no V0-to-I0 raising) and a higher position for the verb in French. (9) a. b.
James often kisses (*often) Marie. Jean (*souvent) embrasse souvent Marie Jean often kisses often Marie “Jean often kisses Marie.”
(Pollock 1989: (4))
The position of VP adverbs in Czech patterns with that of English. The adverb often must precede the verb if it has sentential scope, example (10). (10) Honza často líbá (*často) Marii. Honza often kisses often Marie “Honza often kisses Marie.”
(Veselovská 1995: 83 (7))
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
This evidence suggests that the verb remains within the vP in Czech, raising from V0 to v0, by assumption, but no higher.4 Czech has several verbs which correspond in meaning to English modals: moct (“can.inf”), muset (“must.inf”), and the future form of být (“be.inf”).5 Unlike their English counterparts, these verbs inflect fully. Rather than analyzing these verbs as occupying I0, I propose that they are verbs which simply subcategorize for infinitival vP complements. Evidence presented below suggests that, like lexical verbs, Czech modals do not raise to I0, but remain within the vP domain. Since moct (“can.inf”), muset (“must.inf”), and the future form of být (“be.inf”) take vP complements, we would expect their relative position with respect to VP adverbs to differ from that of lexical verbs. Often adjoins to vP; if there is more than one vP projection, this element could adjoin to any of them. Further, if the modals appear in v0 (and not in I0), VP adverbs should be able to precede them, follow them and intervene between them. This is what is found. Possible positions of the VP adverb with respect to these verbs is shown in the textual examples in (11). In (11a), often follows the stacked verbs and adjoins to the lowest vP shell, the shell headed by the lexical verb. (11) a.
Mág bude muset často udělat víc než wizard will must.inf often finish-doing.inf more than
si naplánoval… refl-cl planned
“The wizard will often have to finish doing more than he planned…” (things.magick.cz/magick/index.php?id=2&co=celej)
Other orders of this adverb with respect to these verbs are attested, but are less common. In (11b), the VP adverb precedes the clitic cluster. b. Naopak, mozek často bude muset dělat on-the-other-hand brain often will must.inf do.inf
složité kotrmelce… complicated somersaults
“On the other hand, the brain will often need to do complicated somersaults…” (www.rydval.cz/phprs/view/php?cisloclanku=2005123154)
. For Veselovská 1995, the verb raises to the head of an AgrO projection which immediately dominates the VP. . Veselovská 1995 analyzes the Czech correspondent to will (the future form of být (“be. inf”)), but not the correspondents to must and can, as an auxiliary verb which takes a complement headed by an infinitival verb. One reason for this is that this verb has only an inflected form and must precede all verbs in a sequence of infinitival verbs.
Chapter 2. The syntax of the left periphery
In (11c), the VP adverb intervenes between the stacked verbs. c.
Tento člověk se bude často muset vypořádávat s that person refl-cl will often must.inf struggle-out.inf with
náročnými situacemi… exacting situations
“That person will often have to struggle with exacting situations…” (www.volny.cz/rosalynda/asryby.htm)
XP’s in the Czech middlefield also adjoin to vP. The expectation would, again, be that these elements can occur in any order with respect to verbs which subcategorize for vP’s. In (12) we see that a PP in the middlefield can follow a sequence of modals, (12a), or intervene between them, (12b). (12) a.
…že bych se mohl muset k dialupu ČTc vrátit… c cond.1sg.cl refl-cl may.pst must.inf towards dial-up ČTc return.inf “… that I would have to go back to ČTc [telecom] dial-up …” (mobil.idnes.cz/tiskni.asp?c=A030808_5221467_mob_ prakticky&r=mob_prakticky)
b. Marie bude za to muset platit. Marie will after it must.inf pay.inf “Marie will have to pay for it.”
The tree in (13) exemplifies the analysis of the verbal domain using (11c) as illustration.6 (13)
IP DP1 tento člověk (that person)
I se (refl-cl) []
I′ vP v bude (will)
vP AdvP často (often)
vP
v muset (must.)
vP t1
v′
v2 vypořádávat (struggle-out.)
VP ...
. Bude (“will.3sg”) and muset (“must.inf”) are Merged into the structure as V heads that raise to the head of a dominating vP projection. This is not illustrated in the tree for space reasons.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
The subject DP, tento člověk (“that person”), raises to [Spec, IP] to satisfy the EPP feature associated with I0, and the reflexive clitic is positioned in I0. In the verbal domain, the VP adverb, často (“often”), intervenes between the first two verbs and adjoins to an intermediate vP projection. The lexical verb raises from the VP domain to head the lowest vP. The low position of verbal elements in Czech leaves I0 available to house second position clitics, and permits the first position to be identified as [Spec, IP], rather than [Spec, CP]. Any additional CP projection is unnecessary in most cases in Czech. 4.2 VP ellipsis In this section, I propose an analysis of VPE in which the target of ellipsis is a vP complement to a modal, (i.e., moct (“can.inf”), muset (“must.inf”), and the future form of být (“be.inf”)) (for other work on V-stranding and VPE see McCloskey 1991; Goldberg 2002, 2005; among many others).7 Consider (14a–b). (14) a. “Vždytˇ to bude bolet,” zaúpěl Timmie…“Nebojte, nebude [vPbolet].” for-sure it will hurt.inf groaned Timmie neg-fear neg-will hurt.inf ““It will hurt for sure,” groaned Timmie… “Don’t worry, it won’t [vPhurt].”” (Czech National Corpus) b.
…každej chce jít domů, ale já nemůžu [vP jít domů]. every wants go.inf home but I neg-can go.inf home “Everyone wants to go home, but I can’t [vP go home].” (www.mujweb.cz/www/vyzyvatel/VYZVA/SOUBORY/V49.htm)
Like VPE in other languages, Czech VPE allows the identity relationship between the antecedent and the elided material to be inexact. There are two possible interpretations of (15): either Pavel and Honza differ in their attitudes toward Pavel’s dog or each differs in his attitude towards his own dog. Note that the form of the elided possessive pronoun would differ under the two possible interpretations.
. Czech also has a type of ellipsis (possibly another type of VPE) in which a polarity head, rather than an auxiliary verb, is stranded, (i).
(i) Já v tomhle stádiu nemám co ztratit. Ty ale jo [vP máš co ztratit]. I in that stage neg-have what lose.inf you excl yes have what lose.inf “I don’t have anything to lose at this stage. But, boy, you do [vPhave something to lose].” (Czech National Corpus) Further research is necessary to determine the exact nature of this construction.
Chapter 2. The syntax of the left periphery
(15) Pavel1 nebude mít rád svého1 nového psa ale Honza2 bude Pavel neg-will have.inf joy self ’s new dog but Honza will [mít rád svého2/jeho1 nového psa]. have.inf joy self ’s/his new dog “Pavel1 won’t love his1 new dog, but Honza2 will [love his1/2 new dog].”
As in English, Czech VPE requires that the ellipsis site be sister to a prosodically strong head. This explains why the complement to I0 cannot be elided. Clitics, which appear in I0, are not prosodically strong, and thus, cannot license VPE. Combined with earlier assumptions, analyzing VPE as elision of the vP complement to a modal makes two predictions. First, we saw in the previous section that Czech allows sequences of modals, each taking a vP complement. In such cases, there should be a choice as to which complement vP is elided. This is correct. In (16) there are two ellipsis possibilities: stranding the first modal or stranding the first two modals. (16) Já budu muset udělat zkoušku, ale ty nebudeš (muset) [udělat zkoušku]. I will must.inf pass.inf exam but you neg-will must.inf pass.inf exam “I will have to pass the exam, but you won’t [have to pass the exam].”
Ellipsis can target the vP complement to either head.8 The second prediction concerns the stranding of lexical verbs in VPE. If lexical verbs raised out of the verbal domain to I0 we would expect to be able to elide the remnant vP, stranding just the verbal head.9 Under the present analysis, however, lexical verbs raise only from V0 to v0. Since VPE does not target VP complements, only vP complements, it should be impossible to strand a lexical verb in VPE. This is correct as shown by the ungrammatical (17). (17) *Honza si koupil nové auto, ale Petr nekoupil1 [VP t1 nové auto]. Honza refl-cl bought new car, but Petr neg-bought new car Intended: “Honza bought a new car, but Petr didn’t [buy a new car].”
. There is one remaining issue. If clitics are in I0 and ellipsis targets the vP complements, we would expect clitics to survive ellipsis. This is not the case.
(i)
Já to budu studovat ale ty *to nebudeš [studovat]. I it.cl will study.inf but you it.cl neg-will [study.inf] “I will study it, but you won’t [study it].”
This pattern is also found in Spanish and is analyzed as Null Complement Anaphora (NCA) in Depiante (2001). More work is needed on this construction in Czech to clearly distinguish VPE from NCA. The availability of strict and sloppy readings suggests VPE, while the fact that clitics do not survive ellipsis suggests NCA. . See section 5.2 for an instance when the lexical verb appears in I0 and can license ellipsis.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
The impossibility of lexical verb stranding in VPE provides further evidence for an analysis of the verbal domain in which lexical verbs remain low in the syntax, raising from V0 to v0, but no higher. Other support for this analysis was found in the position of VP adverbs and middlefield XP’s, both of which adjoin to vP. Since both types of elements obligatorily precede lexical verbs, the lexical verb must raise no higher than the head of the lowest vP projection.
5. The left periphery: Discoursal-syntactic mapping Now that the syntax of the anchoring head positions (I0 and v0) has been established, we turn our focus to the location of XP’s with respect to these anchors. With an exception to be discussed in section 7, only a single phrasal constituent can occupy the position before I0, which I identify as [Spec, IP]. This privileged structural position is associated with at least two specific discourse functions (narrow focus and contrastive topic). When not occupied by an element bearing one of those functions, it is associated with a constituent whose discourse referent is given or topical. 5.1 Discourse function and [Spec, IP] Narrowly focused elements corresponding to the answers to wh-questions can appear in a left peripheral position preceding the clitic cluster.10 Consider (18b). (18) a.
Co jsi dala Honzovi k narozeninám? what aux.2sg.cl gave Honza.dat towards birthday “What did you give Honza for his birthday?”
b. KnížkuF jsem mu dala. book aux.1sg.cl him.cl gave “I gave him [a book]f.”
Narrowly focused elements can also appear at the right edge of the clause. The example in (18c) is also a felicitous answer to (18a). c.
Dala jsem mu knížkuf. gave aux.1sg.cl him.cl book “I gave him [a book]f.”
. Other types of focused elements can appear in this clause initial position as well. See Lenertová and Junghanns 2007 for a discussion of wide focus and intonationally prominent clauseinitial elements. Here, however, I limit myself to narrowly focused XP’s.
Chapter 2. The syntax of the left periphery
Both positions of focused elements, left and right edge, are equally felicitous for speakers.11 Another discourse function associated with [Spec, IP] is that of contrastive topic (CT). Contrastive topicalization exhibits the following characteristics. First, the discourse referent of a CT must be a member of a previously evoked set. Second, a clause containing a CT must also contain a focused element. Third, CT’s are associated with an intonational rise (see Veselá et al. 2003). The discourse function associated with CT’s is the subject of chapter 4. In (19b), the dative arguments, Honzovi (“Honza.dat”) and Petrovi (“Petr. dat”), are CT’s. The discourse referents that they denote are members of a previously mentioned set, “sons”, and each clause contains a focused element (autíčko (“car”) and knížku (“book”)). (19) a.
Co jsi koupila synům k svátku? what aux.1sg.cl bought sons.dat towards name-day “What did you buy your sons for their name days?”
b. Honzovict jsem koupila autíčkof a Petrovict knížkuf. Honza.dat aux.1sg.cl bought car.dim and Petr.dat book “Honzact I bought [a toy car]f and Petrct [a book]f.”
Note that when a CT occurs at the left edge, the focus appears at the right, (19b). These data suggest that [Spec, IP] can be associated with two distinct discourse functions: focus and CT. But, can more than one XP be found at the left periphery of the IP domain? In general, no. Assuming that the second position clitic marks the boundary between [Spec, IP] and the rest of the clause, only one XP can occur in the preclitic position, see (20). These judgments are robust for all speakers consulted. (20) a.
*Honzovif knížkuct jsem dala. Honza.dat book aux.1sg.cl gave Intended: “I gave Honzaf [a book] ct.”
b. *Knížkuct Honzovif jsem dala. book Honza.dat aux.1sg.cl gave Intended: “I gave Honzaf [a book]ct.”
That CT’s and focused elements cannot co-occur in the preclitic position suggests that elements with these two discourse functions compete for the same structural position, [Spec, IP].
. The clause-initial position of focus may be associated with a slightly lower register.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
We now turn to the position between I0 and v0, the Czech middlefield. Elements that occur between the second position clitic cluster and the verb have been previously mentioned or evoked in the discourse context and are not subject to any ordering restrictions. This is illustrated in (21a–b). Internal arguments of the verb, nepověsil (“neg-hung”), prádlo (“laundry”) and z okna (“from the window”), have raised out of the VP. Both orderings of these elements are felicitous and have the same interpretation. (21) a.
Já bych prádlo z okna nikdy nepověsil. I cond.cl laundry from window never neg-hung “I would never hang my laundry from the window.” (www.okoun.cz/boards/nikdy_bych…)
b. Já bych z okna prádlo nikdy nepověsil. I cond.cl from window laundry never neg-hung “I would never hang my laundry from the window.”
Since elements in this domain can appear in any order, I assume that they adjoin to vP. In section 4, I provided data which showed that modals, which appear in v0, can appear in any order with respect to these adjuncts. This is expected if middlefield XP’s adjoin to vP. When there is no CT or focused element in the clause, the highest XP in the middlefield raises to satisfy the EPP feature on I0. Consistent with traditional approaches to Czech, this clause-initial element is associated with a topic interpretation. The discourse function of topic has sometimes been conflated with that of CT under the terms topic or topicalization; these two discourse functions are differentiated here. Unlike CT’s, topical elements need not co-occur with a focused element and are not associated with a rising intonational contour. Typically, discourse referents with a topic interpretation have been previously mentioned in the discourse context and continue to be discussed in the following clauses. These are not characteristics associated with CT’s. CT’s tend to be compared with other discourse referents in the context and are generally not discussed in subsequent clauses. A textual example in (22) provides illustration. The DP referring to Vítek appears in [Spec, IP] in (22b), and, as a result, Vítek is interpreted as topical. I underline all references to this discourse referent. (22) a.
Maminka Vítka zvedla a postavila ho do dveří vagónu. mom Vitek.acc brought and stood him to doors wagon.gen
b. Vítek se těšil, jak vyleze po schůdkách sám. Vitek refl-cl looked-forward-to how climbs.3sg through stairs alone
c.
Chapter 2. The syntax of the left periphery
Místo něho šla po schůdkách maminka. in-place-of him went through stairs mom
“Mom took Vítek and put him in the door of the train compartment. Vítek was looking forward to climbing the stairs alone. Instead of him, mom went up the stairs.” (Řiha 1992)
Vitek is mentioned in the clause preceding its appearance in [Spec, IP], (22a), and continues to be discussed in the following clauses, (22b–c). Any argument can appear in [Spec, IP]; there is no requirement that the nominative raise to satisfy the EPP. Since elements first scramble to an adjoined position in the vP, it is the highest adjoined element which raises, whatever that may be. Consider (23b) in which an accusative-marked DP fills [Spec, IP]. (23) a.
V téhle stanici přistoupil do kupé pán s rezavým psíkem. in that station boarded to compartment man with rust-colored dog
b. Toho pána Vítek skoro neviděl. that man.acc Vítek almost neg-saw c.
Měl hlavu moc nahoře. had head very high
“At that station a man with a rust-colored dog entered the train compartment. Vítek could barely see him. He was very tall.” (Řiha 1992)
In this case it is the accusative argument that is topical. The discourse referent corresponding to toho pána (“that man.acc”) is introduced in (23a), and, since it is the only given element in the second clause, the DP which refers to it appears in [Spec, IP] in (23b). Another sign that this discourse referent is topical in (23b) is that fact that it is referred to by a null pronominal in the following clause, (23c). Once again, only a single XP can occur in the left peripheral position in the clause. Like CT’s and foci in (20), topical elements cannot co-occur with either CT’s or focused XP’s, (24). (24) a.
*Honzovict/f knížkut jsem dala. Honza.dat book aux.1sg.cl gave Intended: “I gave Honzact/f the book.”
b. *Knížkut Honzovict/f jsem dala. book Honza.dat aux.1sg.cl gave Intended: “I gave Honzact/f the book.”
The data in (20) suggested that there is one structural position at the left edge for which CT’s and foci compete. The data in (24a–b) suggest that topics compete for the same position.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
The schematic in (25) illustrates the proposed structure of the left periphery. (25)
IP XP
Focus, Contrastive Topic, Topic I′
I clitics
vP Middlefield vP
XP XP
vP v′ v
VP
It follows from the preceding discussion that the syntactic position of an XP has interpretive consequences. For instance, CT’s and foci must be in specific clausal positions in order to be interpreted felicitously. In section 6, a syntactic analysis is developed which motivates movement of these XP’s to the left periphery. 5.2 V0-to-I0 raising: Satisfying the EPP I assume that an EPP feature is obligatorily associated with the highest inflectional head and is normally satisfied by movement of an XP into [Spec, IP]. Verb-initial sentences, however, are very common in Czech. One context in which they occur is when the subject is a null pronominal. Consider (26). (26) Pojedu do Francie přiští týden. go.fut to France next week “I am going to go to France next week.”
Sentences of this type bring into question the assumption that the EPP is obligatorily associated with the highest inflectional head, I0. Since there is no apparent overt element that satisfies the EPP, one possible solution would be to assume that it is a null subject pronominal that satisfies the EPP. This solution, though, is problematic because another common context for verb-initial clauses is when the subject argument appears in a right-edge focus position, see (27b). (27) a.
Kdo si koupí pivo? who refl-cl bought beer “Who’s buying the beer?”
Chapter 2. The syntax of the left periphery
b. Koupí ho Marekf. bought it.cl Marek “Marek’s buying it.”
In (27b), as opposed to (26), there is no null argument which could potentially fill [Spec, IP].12 Examples (26) and (27b), then, appear to contradict the generalization that the highest inflectional head is associated with the EPP.13 Alexiadou and Anagnostopoulou 1998, however, propose that there are two ways that languages can satisfy the EPP: Move/Merge XP or Move X0. Languages that require Move/Merge XP are of the familiar type; movement of an XP to [Spec, IP] satisfies an EPP feature on I0, or, alternatively, an expletive is Merged into that position. Another class of languages are those with null subject pronouns and rich verbal morphology. Due to this rich inflection, the finite verb in these languages has the requisite pronominal features to satisfy the EPP through V0-to-I0 raising. Czech provides an interesting test case for the Alexiadou and Anagnostopoulou proposal. Unlike many languages that they consider, the lexical verb in Czech remains within the vP when an overt XP appears at the left edge. If the finite verb raised only when there was no other XP available to satisfy the EPP, there should be evidence of its higher position in just those cases. Such evidence exists. It was shown in section 4.1 that VP adverbs obligatorily precede the lexical verb when an XP appears in [Spec, IP], see (28). (28) Honza to často vidí (*často) kolem sebe. Honza it.cl often sees often around self “Honza often sees it near him.”
But, just in cases when the verb appears in the initial position, VP adverbs follow the lexical verb. Consider the textual example in (29a). . A possible solution would be to posit a null expletive. However, the fact that Czech has optional expletives makes this solution unsatisfying. Consider (i). There is no interpretational difference between these sentences.
(i)
Prší. / Ono prší. rains / expletive rains “It’s raining.”
To pursue an analysis of this type, one would have to account for the fact that expletives are sometimes covert and sometimes overt. . Veselovská 1995 argues that verb-initial sentences are the result of topicalization of the verb to a left-edge head position. I suggest that, though this may be true in some cases, it cannot account for all verb-initial sentences. I follow the formal analysis of Alexiadou and Anagnostopoulou 1998.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(29) a.
Vidím to často kolem sebe. see it.cl often around self “I often see it near me.” (www.pismak.cz/dilo.php?num=20538)
The same pattern is found with sentence-initial modals, (29b). b. Budu to často vidět kolem sebe. will.1sg it.cl often see.inf near self “I will often see it near me.”
In both (29a–b), the highest finite verb precedes not only the adverbs, but also the second position clitic; this suggests that the first verbal element has raised out of the vP to I0. This raising of the verb, Alexiadou and Anagnostopoulou argue, can satisfy the EPP feature on I0. This type of analysis also predicts that just when the lexical verb appears in I0, it should be able to license VPE. This is what is found, see (30). (30) Jan chtěl jít do krámu, ale nakonec nešel. Jan wanted go.inf to store but after-all neg-went “Jan wanted to go to the store, but didn’t after all.”
When the verb raises from v0 to I0 to satisfy the EPP, it is in a structural position that is high enough (I0) to license elision of its complement vP. In the examples considered above, (26), (27b) and (29a–b), the verb fronted to satisfy the EPP feature on I0 even when there were other XP’s in the sentence which could, in principle, have raised to satisfy the EPP. Though a comprehensive analysis of the restrictions on this sentence-initial position is beyond the scope of this book, there appears to be a principle at the syntactic-pragmatic interface which requires that a sentence-initial XP be pragmatically linked to the previous clause (see also Sgall et al. 1973, 1980, 1986). This restriction may provide an explanation for the fact that, in some verb-initial contexts, the verb has fronted when there are other XP’s in the clause which could move. If these XP’s cannot be associated with a topic interpretation, fronting is not felicitous.14 Given this analysis, it is possible to assume that the EPP is obligatorily associated with I0. Czech provides two ways to satisfy the EPP: movement of an XP to [Spec, IP] or head raising of the highest finite verb. Verb raising is a last resort; it
. Additional evidence for this restriction comes from Cummins 1998. Czech lacks obligatory definite and indefinite determiners and Cummins argues that syntactic position serves to mark definiteness. Clause-initial DP’s are obligatorily associated with a definite interpretation. One reason for this interpretative restriction may be that the sentence-initial position is associated with a topic interpretation, an interpretation generally unavailable for indefinites.
Chapter 2. The syntax of the left periphery
occurs only when there is no overt XP in the clause which can be associated with a topic discourse function. 6. Discoursally motivated A-bar movement Under the Minimalist Program, all movements are motivated by uninterpretable features on either the goal or the probe. In the case of discoursally motivated displacement, these features correspond to discourse features. In the following sections, I present evidence that suggests that, in Czech, all discoursally motivated movements are A-bar movements. To provide a context for the discussion, I first establish the A-bar character of wh-movement in Czech. 6.1 Wh-movement Czech wh-movement exhibits two familiar characteristics of A-bar constructions: long distance dependencies and island sensitivity. Extraction of wh-expressions out of all types of embedded clauses is possible. Consider the examples in (31a–b). Extraction out of the finite complement to the bridge verb myslet (“think.inf ”) is illustrated in (31a). (31) a.
Koho1 myslíš, že Marie pozvala t1 na tu párty? who.acc thinks c Marie invited on that party “Who1 do you think Marie invited t1 to the party?”
In (31b), a wh-phrase is extracted out of an embedded subjunctive clause in an object control structure. b. Co1 jsi požádala Marii, aby mi koupila t1? what aux.2sg.cl requested Marie.acc c.cond.3sg.cl me.dat.cl bought “What1 did you ask Marie to buy me t1?”
As expected, wh-movement also obeys well-known island constraints. (32a) illustrates a wh-island and (32b) a Complex NP island. (32) a. *Komu2 by tebe zajímalo koho1 Marie představila t1 t2? who.dat cond.3sg.cl you.acc interested who.acc Marie introduced Lit: “To who2 do you wonder who1 Marie introduced t1 t2?” b. *Komu1 viděla Hana chlapa, který dal t1 kytky? who.dat saw Hana guy rel gave flowers Lit: “To who1 did Hana see a guy who gave flowers t1?”
Following standard analyses, I assume that Czech wh-movement is an perator-variable A-bar construction (for current approaches to operator moveo ment in Minimalism see Reinhart 1998; Fox 2002; Adger & Ramchand 2005). An interpretable wh-operator feature, q, appears on a functional head at the left edge, as well as an uninterpretable wh feature. Wh-expressions have an uninterpretable q
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
feature and an interpretable wh feature on the head of their phrase. Thus, both the probe and the goal are active as both are associated with uninterpretable features. Czech has overt wh-movement, so the probe is associated with an EPP feature which motivates overt movement of the wh-expression into its specifier. In Czech, it appears that wh-movement targets [Spec, IP] rather than [Spec, CP], as is often assumed.15 Evidence for this structural position is that fronted wh-phrases cannot occur with topical elements in the pre-clitic domain, see (33); wh-phrases appear to compete for the same position, [Spec, IP] (see also Lenertová 2001).16 (33) *Co Honzat by na to řekl? what Honza cond.3sg.cl on it say Intended: “What would Honzat say about that?”
A illustrative tree for (34) is given in (35). (34) Co by na to řekl? what cond.3sg.cl on it say “What would he say about that?” IP (35) DP I′ [wh/q, u]1 (what) I [wh, u/q][EPP] vP by (.) PP na to3 (on it) DP Pro
vP v′ v řekl2 (said)
VP PP t3
V′ V t2
DP t1
. Manetta 2006 also assumes a low target for wh-movement in Hindi-Urdu. She assumes that the wh-phrase target [Spec, vP]. . In section 7, I return to this issue and show that wh-phrases can occur with CT’s in the preclitic domain. An expanded left periphery is posited at that point.
Chapter 2. The syntax of the left periphery
The uninterpretable features on the wh-phrase and I0 are valued through the movement operation (Agree and Remerge) of the wh-phrase to [Spec, IP]. The PP, na to (“on it”), scrambles to an adjoined position in the vP domain. There is an additional complication, however. Czech is among the languages that allow multiple wh-fronting. Rudin 1988 divides multiple whfronting languages into two categories: +Multiply Filled Specifier, [+MFS], and –Multiply Filled Specifier, [–MFS]. Under her proposal, which is adopted in Richards 2001, languages either allow multiple movement to [Spec, CP] ([+MFS]) or restrict movement to [Spec, CP] to one wh-expression and adjoin other wh-phrases to a lower projection. Bulgarian and Romanian are languages of the first type, while Czech, along with Polish and Serbo-Croatian, are of the second type. For me, one wh-phrase in Czech fronts to [Spec, IP] and enters into a checking relationship with I0, other wh-phrases adjoin to a lower projection. Czech patterns with the [-MFS] languages according to two of the key diagnostics. The first diagnostic Rudin 1988 uses to distinguish between [+MFS] and [–MFS] languages is wh-island effects. The prediction is that [+MFS] languages should not show wh-island effects since there are multiple specifier positions of C0 in the embedded domain through which multiple wh-phrases can move, thereby escaping embedded clauses. Consider (36) for Bulgarian. It is fairly acceptable for a heavy wh-phrase to escape a wh-island. (36) ?[Koja ot tezi knigi]2 se čudiš koj znae koj1 prodava which of these books refl-cl wonder who knows who sells t1t2? “[Which of these books]2 do you wonder who knows who1 sells t1t2?” (Rudin 1988: (20b))
[–MFS] languages, on the other hand, exhibit wh-island effects since there is only one specifier position of C0. If this position is filled, as it is in embedded whquestions, additional wh-phrases are restricted from moving out of the embedded clause. This restriction holds in Czech, see (32a). The second diagnostic is Superiority. Czech patterns with other [–MFS] languages in that it lacks Superiority effects in wh-questions. [+MFS] languages, on the other hand, respect Superiority; this is expected as movement to specifier positions is motivated by features of the attracting head (C0 in Rudin’s account). The highest wh-phrase within the clause is attracted first and moves into the highest specifier position; lower wh-phrases move to lower specifier positions of C0. Wh-movement in Bulgarian, [+MFS] language, respects Superiority, (37a–b).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(37) a. b.
Koj kogo vižda? who.nom who.acc sees “Who sees who?” *Kogo koj vižda? who.acc who.nom sees Intended: “Who sees who?”
Czech, in contrast, entirely lacks Superiority effects in multiple wh-questions, (38). (38)
Komu by kdo co dal? who.dat cond.cl who.nom what.acc gave “Who would give what to whom?” Lit: “To whom would who what give?”
Richards 2001 argues that the lack of Superiority effects is a direct result of the possibility of adjunction to a lower functional head in [-MFS] languages. If whphrases first adjoin to a lower functional projection and then the highest whexpression raises to the specifier position of the attracting head, Superiority effects are not expected. The initial adjunction operation allows free ordering of wh-XP’s, obviating superiority effects. Under this assumption, leftward movement of whphrases parallels that of non-wh-expressions which also first adjoin to a lower functional projection, see §5.1. What structural position do the lower wh-phrases occupy? Rudin 1988 and Richards 2001 assume that they adjoin to IP. New evidence from Czech, however, suggest that lower wh-phrases target a lower functional projection, vP. The first wh-expression is separated from subsequent ones by the clitic cluster. Since clitics are, by assumption, in I0, this suggests that the lower wh-phrases adjoin to vP, rather than IP.17 I assume that all wh-phrases first adjoin to vP, and from there, the highest wh-phrase raises to satisfy the EPP on I0. The q-feature on other whphrases is also checked by I0 via static Agree, but there is no additional leftward movement. If wh-phrases adjoin to vP, rather than IP, the prediction is that they can intervene between other non-wh elements in the middlefield. This is indeed possible. In (39), the interleaving of wh- and non-wh-phrases in the middlefield suggests that lower wh-expressions do adjoin to vP. (39)
Kde jsi (včera večer) koho (včera večer) komu představila? where aux.2sg last night who.acc last night who.dat introduced “Where did you introduce who to who last night?“ Lit: “Where (last night) who (last night) to who introduced?”
. Additional evidence for the lower adjunction site is provided in section 7.
Chapter 2. The syntax of the left periphery
The structure of (39) is illustrated in (40). IP
(40) DP kde[wh/q, u] (where)
I′ I [wh, u/q][EPP] jsi (.2.)
vP
DP vP koho (who.) AdvP vP včera večer (last night) DP komu (who,) DP pro
vP v′
v1 představila (introduced)
VP ...
In the following sections, I consider focus movement and contrastive topicalization. I show that properties of contrastive topicalization in Czech parallel those of whmovement, suggesting that contrastive topicalization is also an A-bar movement. 6.2 Focus Because the subject of this book is contrastive topic constructions, not focus, I will not provide a detailed syntactic analysis of focus movement. I assume that leftward movement of elements for interpretation as foci is an operator-variable A-bar movement motivated by syntactic features in I0. Focused elements, however, can also be interpreted in their base positions at the right edge of the clause, suggesting that syntactic movement is not the whole story for focus in Czech. Some researchers, e.g., Büring 2001 and Arregi 2001, have argued for a prosodic analysis of focus placement in which focused elements are positioned in clausal positions where they can receive the required intonational rise. It is likely that intonational constraints play a role in motivating the structural position of focused elements in Czech, but I will not develop such an account here. 6.3 Contrastive topicalization Czech contrastive topicalization exhibits typical characteristics of A-bar movement, such as reconstruction, long distance dependencies and island effects. To illustrate
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
that contrastive topics are evaluated for Condition A in their base position, and not in [Spec, IP], I first show that Czech possessive and non-possessive reflexives obey Condition A. A version of this constraint is provided in (41–43). (41) Binding Domain:18 A binding domain for α is the most deeply embedded Complete Functional Complex (CFC) containing α in which the basic binding
requirements for α can be met. (42) Complete Functional Complex: An XP in which all the θ-roles compatible with a lexical head are assigned in
A-positions. (43) Condition A: For α, an anaphor, its basic binding requirements can be met in principle in a category C iff there is an assignment of indices to DP’s within C (perhaps different from the actual assignment of indices within C) according to which α is A-bound.
A reflexive possessive must be grammatically bound within its binding domain. This condition holds in (44).19 (44) Honza1 dal svému1 přítele dárek. Honza gave self ’s friend.dat present.acc “Honza1 gave his1 friend a present.”
Czech has no logophoric uses of reflexive pronouns (see Sturgeon 2003). The possessive reflexive in (45) is ungrammatical because it appears across a sentence boundary from its antecedent and thus, c-command between the antecedent and the reflexive cannot hold. (45) Honza1 rád čte. *Knížky o svém1 psovi ho velmi zajímají. Honza joy reading books about self ’s dog him.acc.cl very-much interest “Honza1 likes to read. Books about himself1 interest him very much.”
Possessive reflexives occur freely in [Spec, IP] as subparts of contrastive topics, (46). (46) [Svůj1 sešit]ct
jsem
ztratil1, ale [svou1 učebnici]ct ještě mám1.
self ’s notebook.acc aux.1sg.cl lost but self ’s textbook.acc still have “I lost [my notebook] ct, but still have [my textbook] ct.”
. A slight reformulation of Chomsky 1986 due to James McCloskey. . All Czech reflexive pronouns are subject oriented and, thus, must be anteceded by a subject argument (see Sturgeon 2003).
Chapter 2. The syntax of the left periphery
After movement of the contrastive topic to the left edge, the reflexive possessive is no longer c-commanded by its antecedent. Given the grammaticality of (46), Condition A must be evaluated after reconstruction of the contrastive topic to its base position within the VP. Reconstruction for quantifier binding also holds for contrastively topicalized elements. Like reflexive binding, quantifier binding requires that a quantificational DP c-command a bound pronominal. The bound pronominal in (47) is ungrammatical because the quantifier does not c-command it.20 (47) *Jeho1 pes kousnul každého1 majitele. his dog bit every owner Intended: “His1 dog bit every1 owner.”
In (48), the bound pronominal reading is available because the quantified NP ccommands the bound pronominal. (48) Každý1 majitel krmil svého1 psa. every owner fed self ’s dog “Every1 owner fed his1 dog.”
The structural requirements for quantifier binding must also be evaluated after reconstruction, since contrastive topics can contain bound elements, see (49). (49) [Své1 dceři]ct každá matka1 dala do školy nové šaty, ale self ’s daughter.dat every mother gave to school new clothes but [svému1 synovi]ct ne [dala každá matka1 do školy nové šaty]. self ’s son.dat no gave every mother to school new clothes “Every mother gave her daughter new clothes for school, but not her son.”
Another diagnostic for A-bar movement is the possibility of long distance dependencies. Though it is generally less felicitous than long distance wh-movement, contrastive topics at the left edge of a matrix clause can be associated with a position within an embedded clause, (50). (50) a.
Ve třídě jsou čtyři žáci a tři ještě nemají učebnici. in class are four students and three still neg-have textbooks
b. [Honzovi1]ct mi řikali, že učitel ji už dal t1, ale Honza.dat me.dat told.pl c teacher it.cl already gave but
ostatnímct ještě ne. others still no
. There are no nominative reflexives in Czech. A personal pronoun is used in these types of contexts in which a reflexive is ruled out by the Czech lexicon.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
“There are four students in the class and three still don’t have textbooks. HonzaCT they told me the teacher gave a textbook to, but the othersCT she hasn’t yet.”
Finally, contrastive topicalization also obeys well-known island constraints. (51a) illustrates a wh-island and (51b) a Complex NP island. (51) a. *Honzovi2 by tebe zajímalo koho1 Marie představila t1 t2. Honza.dat cond.3sg.cl you.acc interested who.acc Marie introduced Lit: “Honza2 you wonder who1 Marie introduced t1 to t2?” b. *Janě1 viděla Hana chlapa, který dal t1 kytky? Jana.dat saw Hana guy rel gave flowers Lit: “Jana1 Hana saw a guy who gave flowers to t1.”
That constrastive topics exhibit long distance dependencies, obey island constraints and reconstruct for binding suggests that [Spec, IP] is an A-bar position and that the contrastive topic moves to this position for reasons unrelated to case assignment. 6.3.1 Weak Crossover Since the early 1970’s, researchers have distinguished various types of A-bar dependencies. Rizzi 1997 divides A-bar movements into two classes: operator movement constructions that bind a variable (wh-movement) and those that don’t (topicalization). To distinguish these two classes, he uses the diagnostic introduced in Lasnik and Stowell 1991, Weak Crossover (WCO). WCO is used to identify “true quantifier” constructions, A-bar constructions in which a syntactic operator binds a variable in the position of the trace. For Lasnik and Stowell, English wh-movement is a true quantifier construction because it exhibits WCO effects. Co-variation between lawyers and that lawyer’s clients is not possible in (52). (52) *Which lawyer1 do his1 clients hate?
English topicalization, however, does not exhibit WCO effects, (53). (53) That first grader1 I saw his1 teacher scold, but that second grader1 I didn’t [see his1 teacher scold].
Given evidence of this type, Lasnik and Stowell 1991 conclude that the syntactic nature of the movement operations in wh-movement and topicalization differ. In wh-constructions, an operator binds a variable in the position of its trace, but in English topicalization constructions there is no such binding. WCO does not distinguish these two types of A-bar movements in Czech, however. Richards 2001 presents evidence that [-MFS] languages (for him, IPabsorption languages), such as Serbo-Croatian, do not exhibit WCO effects, even in
Chapter 2. The syntax of the left periphery
wh-constructions. This also true for Czech. There is an available reading of (54) in which lawyers and their clients do co-vary. (54) Kterého právnika1 nenávidí jeho1 klienti? which lawyer hates his clients Lit: “Which lawyer1 do his1 clients hate?”
Richards 2001 proposes that WCO effects are ameliorated by local A-scrambling of wh-phrases to adjoined positions in the IP (for me, the vP). From this adjoined position, binding relations are established and WCO effects obviated.21 Independent of wh- and CT-movement, otherwise expected WCO effects are obviated by overt scrambling to vP. We know that neither DP in (55) is in [Spec, IP] due to the presence of the adverbial, včera (“yesterday”) at the left edge. (55)
Včera každého kluka1 jeho1 matka napomenula. yesterday every boy his mother scolded “Yesterday every boy was scolded by his mother.” Lit: “Yesterday every boy1 his1 mother scolded.”
The accusative marked DP, každého kluka (“every boy.acc”), can bind the nominative DP, jeho matka (“his mother”), even though the nominative argument c-commands its base position. Some binding relationships appear to be established in scrambled positions in the vP domain. WCO effects are absent in contrastive topicalization in Czech, (56). Since WCO effects are not found in wh-constructions, the absence of such effects in this construction is not surprising, and does not imply that wh-movement and CTmovement do not involve operator-variable A-bar movement.
. Richards follows Bošković 1998 in noting that A-scrambling does not affect binding conditions such as Condition A. Condition A is still evaluated after reconstruction, see (i). Binding of his son by Honza is possible even though c-command between the binder and the bindee does not hold at the surface; these relations are established in the base positions of the two arguments.
(i)
Včera ji svému1 synovi Honza1dal. yesterday it.cl self ’s son.dat Honza gave “Yesterday Honza1 gave it to his1 son.”
For German, Hinterhoelzl and Pili 2003 find variation among speakers with respect to the availability of reconstruction for reflexive binding from within the middlefield. For some speakers, this is available. New reflexive binding relationships, however, cannot be established in the middlefield. They link this to the subject orientation of (some) reflexive pronouns in German. Structural positions in the vP domain do not appear to be adequate to establish reflexive binding within the vP domain. This is an area for future research in Czech.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(56) Prváka1 jsem neslyšela jeho1 učitelku napomenout, ale first-grader aux.1sg.cl neg-heard his teacher scold.inf but toho druháka1 jo [jsem slyšela jeho1 učitelku napomenout]. that second-grader yes aux.1sg.cl heard his teacher scold.inf “The first grader1 I didn’t hear his1 teacher scold, but that second grader1 I did [hear his1 teacher scold].”
To account for the symmetry between wh- and CT-movement, I assume parallel syntactic derivations. XP’s scramble to positions adjoined to VP in both constructions. This ameliorates WCO. Then the highest CT or wh-expression in the vP domain raises to [Spec, IP] and binds a variable in the position of the trace. If WCO does not determine the type of A-bar movement instantiated by CTmovement, alternative means must be found. In the following section, I adopt Authier’s 1993 strategy for identifying operator-variable constructions. 6.3.2 Operator-variable contrastive topicalization Lasnik and Stowell 1991 suggest that one characteristic of “true quantifiers” in operator-variable constructions is that they range over a possibly non-singleton set. CT constructions in Czech exhibit just this characteristic. Part of the conventional meaning of contrastive topicalization is that the discourse referent of a CT is a member of a non-singleton set (see chapter 4 for a more thorough discussion of the pragmatics of this construction). The CT construction in (57) conventionally implicates that there are other discourse referents under discussion with respect to the same open proposition, “I know/don’t know x”; the same is true for the English equivalent. (57) [Toho pána]ct neznám, ale tu dívku znám. that man.acc neg-know but that girl.acc know “That man I don’t know him, but that girl I do.”
This characteristic of CT constructions follows from an operator-variable analysis. Authier 1993 identifies another diagnostic for operator-variable constructions: participation in pairings of members of two sets. (58a) illustrates a felicitous multiple wh-question. The structure of the question indicates that a felicitous answer is a potluck attendee-potluck dish pair, (58b). (58) a. Who brought what to the potluck? b. Mark the bean dip, Sally the casserole.
This participation in multiple pairings follows from the operator-variable analysis of wh-movement in English. However, unlike regular wh-questions, echo questions do not participate in multiple wh-questions in which there are pairings between members of two
Chapter 2. The syntax of the left periphery
(or more) sets. (59) is, thus, infelicitous. Authier indicates an echo wh-phrase with an “ ”. (59) *Who did what? (Authier 1993: (10b))
Authier concludes from this that echo wh-constructions are non-quantificational and, thus, are not operator-variable constructions. The prediction, then, is that they will not exhibit WCO effects, as opposed to quantificational non-echo whquestions. This is borne out in (60a). Compare the wh-question in (60b) to the echo question in (60a). (60) a.
Who1 did [the claim that he1 was drunk] bother?
b. *Who1 did [the claim that he1 was drunk] bother? (Authier 1993: (8a))
CT constructions behave like non-echo wh-questions in that pairings between members of two sets are not only possible, but extremely common. Multiple whquestions are often answered by CT constructions, see (61a–b).22 (61) a.
Kdo co přinesl? who what brought “Who brought what?”
b. Honzact přinesl [vepřové maso s knedlíky]f a Martact bábovkuF. Honza brought pork meat with dumplings and Marta babovka “Honzact brought [the pork and dumplings]f and Martact [the pound cake]f”
In (61b) guests are paired with dishes as an answer to (61a) (the opposite pairing of discourse function and participant is also possible). CT constructions instantiate two characteristics of operator-variable constructions. I, therefore, conclude that CT movement in Czech is an operator-variable A-bar movement. 6.3.3 Syntactic analysis of CT constructions I analyze operator movement in contrastive topicalization on a par with operator movement in wh-constructions. The head of the contrastively topicalized XP has
. Topicalization in English also exhibits these two characteristics of operator-variable constructions: their discourse referents are members of a non-singleton set and they participate in pairings between members of two or more sets. This suggests that the role of WCO as a diagnostic of operator-variable constructions may be overstated (see Lasnik & Stowell 1991).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
an interpretable CT feature (comparable to an interpretable wh feature) as well as an uninterpretable CT operator feature, which I will call ct (comparable to the q feature). The landing site for CT’s is [Spec, IP] and the head of that projection bears an interpretable ct feature and an uninterpretable CT feature, as well as the EPP. The tree in (63) illustrates the analysis of (62). (62) [Tu červenou tašku]ct si koupila JanaF. that red bag.acc refl-cl bought Jana “[The red bag]ct Janaf bought.” (63)
IP [DP tu červenou tašku][CT/ct, u] (that red bag) I [CT, u/ct] []
I′ vP
[DP tu červenou tašku][CT/ct, u]
vP
First, the CT’ed DP, the red bag, scrambles to an adjoined position in the vP domain. It then raises to [Spec, IP] to satisfy the EPP feature on I0 and check the uninterpretable features associated with the moved element and the head of the landing site, I0. Assuming a Copy and Delete theory of movement, the bottom copy deletes and the top copy is pronounced.
7. Another look at multiple Wh-questions Now that the syntactic behavior of contrastive topicalization constructions has been shown to be that of operator-variable A-bar movement, it is possible to turn to a puzzle concerning multiple wh-movement in Czech.23 Though, normally, two XP’s cannot appear before the clitic cluster, the order shown in (64) is fully grammatical. (64) WH > CT > clitic cluster
. Lenertová 2001 asserts that wh-expressions can be interpreted in situ. I have not found examples of wh-in situ to be grammatical for my consultants and, therefore, do not consider examples of that type here.
Chapter 2. The syntax of the left periphery
In this section, we consider what syntactic characteristic of wh-movement allows wh-phrases to co-occur with CT’s in the preclitic domain, but restricts the co-occurrence of non-wh-XP’s, (20) and (24). 7.1 Multiple pre-clitic XP’s The question in (65) is an attested example. Both the wh-phrase, co (“what”) and the contrastive topic, Ema, precede the second position clitic. We know that Ema is a CT because that DP is associated with intonational rise. An example of this type would occur in a context in which the speaker is considering several individuals with respect to their feelings about a particular topic. The sentence preceding (65) is: “And Josef, he likes the idea”. (65) A co Emact by na to řekla? and what Ema cond.3sg.cl on it say “And what would Ema say about that?”
(Lenertova 2001: (8))
The grammaticality of (65) contrasts robustly with the question in (66) in which a topic and a wh-phrase precede a clitic. (66) *Co Honzat mu dal? what Honza him.cl gave “What did Honza give him?”
There is an important difference between the co-occurrence of a CT and a wh-phrase and that of a wh-phrase and a topic.24 Both CT’s and wh-phrases participate in operator-variable movement associated with particular feature matrices ([wh, q] for wh-movement and [CT, ct] for contrastive topicalization). XP’s associated with these features must move to the left periphery in order to be interpreted felicitously. If there are no such XP’s in the structure, the highest XP in the vP domain raises to [Spec, IP] to satisfy the EPP on I0. To account for the grammatical co-occurrence of wh-phrases and CT’s, I assume that it is a fact of the Czech lexicon that the featural matrices associated with wh-constructions ([wh,u/q]) can appear in either C0 or I0. The features associated with CT-movement, however, are obligatorily associated with I0. A C0 associated with wh-features may select for an I0 associated with CT features. Thus, in a derivation involving wh-movement and CT-movement, wh-features appear in C0 and those associated with CT-movement appear in I0. This raises the question whether wh-features always have the option of appearing on C0. I assume that
. I assume that wh-phrases and focused elements cannot co-occur in the left periphery since they instantiate the same discourse function.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
structural economy conditions restrict wh-features from appearing on C0 when they could appear on I0. The tree in (67) illustrates the proposal. (67)
CP [DP co]1[WH/q, u] (what)
C′
C [WH, u/q] [EPP]
IP
[DP Ema]2[CT/ct, u]
I′
I [CT, u/ct] [EPP]
vP
t1
vP t2
vP
It is also possible for the CT to appear in a post-clitic position, (68). We know that Ema is a CT and not a scrambled element in the vP domain because it is associated with the same intonational rise here as in (65) and would be used in the same contexts. (68) A co by Emact na to řekla? and what cond.3sg.cl Ema on it say “And what would Ema say about that?”
To account for the two positions of the clitic cluster, I assume that clitics can appear in I0 or optionally raise to the next higher head, C0, when it is projected. One last question remains. Why are CT’s and foci restricted in the pre-clitic domain? Consider the ungrammatical (69), repeated from (20a). (69) *Honzovif knížkuct jsem dala. Honza.dat book aux.1sg.cl gave Intended: “I gave Honzaf [the book]ct.”
Under the assumption that features associated with focus movement, as well as those associated with CT-movement, are restricted to I0, this pattern is expected. There is only one preclitic position which can host discoursal features. If I0 is associated with either focus or CT-features, it is not possible for features associated with the other type of discoursal movement to appear in the left periphery. Thus, only a CT or a focused element can appear in the preclitic position.
Chapter 2. The syntax of the left periphery
7.2 The position of non-initial wh-phrases The possibility of two elements in the pre-clitic domain provides more direct evidence that non-initial wh-phrases adjoin to vP, rather than IP. Let us consider the facts. CT’s obligatorily appear in [Spec, IP]. When they co-occur with whexpressions in wh-questions, they immediately follow the first wh-phrase, (70a). Other wh-expressions follow both the CT and the clitic cluster, suggesting that they appear below I0, in the vP domain. (70) a.
Kde tyct jsi koho komu představila? where you aux.2sg.cl who.acc who.dat introduced “Where did youct introduce who to who?” Lit: “Where youct who to who introduced?”
Alternative orders of the adjoined wh-phrases and the CT are considerably degraded, (70b). b. ??Kde koho komu tyct jsi představila? where who.acc who.dat you aux.2sg.cl introduced Intended: “Where did youct introduce who to who?”
If we assume that non-initial wh-phrases adjoin to vP, the facts presented in (70a–b) follow. Assuming that they adjoin to IP would incorrectly predict (70b) to be grammatical. The assumption that wh-features in the lexicon are associated with either I0 or 0 C , while discoursal features are restricted to I0, allows us to account for the cooccurrence restrictions found in the left periphery. Wh-expressions can co-occur with CT’s in the preclitic position, but no other combination of XP’s is allowed. Now that the relative positions of wh-phrases and CT’s has been established, additional evidence for a low position of these wh-expressions is available. Since non-initial wh-phrases obligatorily follow the CT (in [Spec, IP]), they must hang in an adjoined position within the vP, below I0.
8. Conclusion In this chapter, I have established the basic assumptions about Czech clause structure that underlie the analysis to come. With the exception discussed in the last section, all elements in the Czech clause appear within the IP. Unlike other approaches to the left periphery, for Czech, positing an exploded CP domain (or even making use of CP in all clauses) is unnecessary. Facts which support this analysis are the low position of the lexical verb (in v0) and the fact that the left peripheral position is restricted to one XP. All left-peripheral XP’s (wh-expressions, CT’s, focused
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
elements and topics) can be accommodated within the IP. Avoiding a proliferation of functional projections is thus possible in the left peripheral domain. Two head positions anchor Czech clause structure: I0 and v0. The lexical verb appears in the lowest v0 position and clitics appear in I0. Various diagnostics establish the position of the lexical verb in the verbal domain: the positions of VP adverbs and VPE. VP adverbs must precede the lexical verb and lexical verbs cannot be stranded in VPE. Additionally, positioning verbal and pronominal clitics in I0 allows us to understand the distribution of XP’s in the Czech clause. One specifier position, which can be filled by contrastive topics, topics, foci or wh-phrases, precedes the clitic cluster. The middlefield of Czech is the domain between v0 and I0, and XP’s which are given in the discourse context can scramble to adjoined positions in this span of the clause. Given the analysis presented in this chapter, the pivotal left-peripheral A-bar position in Czech is [Spec, IP], not [Spec, CP].
chapter 3
The syntax of left dislocation 1. Overview of left dislocation in Czech Left dislocation constructions, to my knowledge, have not been previously discussed in the literature on West Slavic (see Arnaudova 2004 for a discussion of left dislocation in Bulgarian). Given the diversity and complexity of these constructions in Czech, left dislocation in Slavic merits further research. There are three types of left dislocation constructions in Czech: Contrastive Left Dislocation (CLD) and two varieties of Hanging Topic Left Dislocation (HTLD I, II) (for a similar inventory in German, see also Grohmann 2003). In all three constructions, a constituent on the left periphery is separated from the main clause by an intonational break, and a coreferent resumptive pronoun is found within the main clause. CLD and HTLD I share certain characteristics. The coreferent resumptive element (usually a demonstrative) appears obligatorily at the left edge of the clause and binds a clause internal gap. Example (1) is ambiguous between the two constructions. The left dislocated DP, Petr, appears on the left periphery and is resumed by a demonstrative pronoun at the left edge of the main clause. Both the left dislocated element and the resumptive pronoun are underlined.1 (1) Petr, ten si koupil chleba v krámě. Petr.nom that.nom refl.cl bought bread in store “Petr, he bought bread at the store.”
Note that the left dislocated element does not participate in the calculation for second-position clitics; the clitic in (1), si (refl-cl), follows the demonstrative and not the left dislocated element. In some sense, left dislocated elements are extra-clausal. HTLD II differs from the other left dislocation constructions in that the coreferent resumptive element is not a strong demonstrative pronoun, but a clitic
. Both personal and demonstrative pronouns are accepted as resumptives in HTLD I, but in CLD, demonstratives are strongly preferred by all speakers consulted.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
pronoun in the second position, (2). This construction is less common than the other two left dislocations, CLD and HTLD I. Generally, people accept constructions like (2), but judgments tend to be less clear. (2) Tu kočku, dala jsem ji sousedovi. that cat.acc gave aux.1sg.cl her.acc.cl neighbor.dat “That cat, I gave it to my neighbor.”
HTLD II
In other respects, HTLD II behaves like CLD and HTLD I. There is an extra-clausal XP separated by an intonational break from the main clause, and the left dislocate antecedes a clause-internal pronominal element. CLD and HTLD I, II constructions also differ in several important ways. We first consider CLD constructions. A distinguishing characteristic of CLD is that it exhibits obligatory case matching between the left dislocate and the left-edge resumptive demonstrative. In (3), both the resumptive pronoun and the left dislocated element are in the accusative, the expected case of the clause-internal gap. (3) Ostatně toho draka, toho by si besides that dragon.acc that.acc cond.3sg.cl refl.cl
CLD
mohl tak jedině namalovat… can so only draw.inf “Besides, that dragon, he would only be able to draw him…” (Czech National Corpus)
In HTLD I constructions, however, case matching is absent. Left dislocated elements appear in the default case/citation form, nominative, (4). (4) Anička? Té se nic nestalo.2 Anička.nom that.dat refl.cl nothing neg-happened “Anička? Nothing happened to her.” (Czech National Corpus)
HTLD I
The left dislocated constituent in (4), Anička, does not appear in the case of the resumptive demonstrative, the case associated with the clause-internal gap, the dative. In HTLD II case matching between the left dislocate and the resumptive clitic is optional. In (5a) the left dislocate is in the nominative case, the resumptive in the accusative. (5) a.
Ta dívka, znám ji ze školy. that girl.nom know her.acc.cl from school “That girl, I know her from school.”
HTLD II
. In this example, the left dislocated element is followed by a question mark in the punctuation and is associated with a question intonation pattern when spoken. This intonation pattern is possible with both CLD’ed and HTLD’ed elements.
Chapter 3. The syntax of left dislocation
In (5b) both the hanging topic and the resumptive clitic are in the accusative case.3 b. Tu dívku, znám ji ze školy. that girl.acc know her.acc.cl from school “That girl, I know her from school.”
HTLD II
Additional differences between these three constructions are found in the intonational break which separates the left dislocate from the main clause; speakers report that the intonational break between the left dislocated element and the main clause is longer in HTLD I, II than in CLD. Since case matching alone does not distinguish these three constructions, additional syntactic diagnostics must be explored. In sections 4 and 5, evidence from reconstruction is presented. CLD constructions exhibit reconstruction effects (such as reconstruction for Conditions A and C), but HTLD I, II do not. Drawing on these facts, I suggest that CLD constructions are generated through movement, while HTLD I, II are not.
2. Grohmann 2003 and left dislocation in German Grohmann 2003 follows a long tradition of research on left dislocation in Germanic languages (for Dutch, see van Riemsdijk & Zwarts 1974/1997; Vat 1981/1997; for German, see Altmann 1981; Scherpenisse 1986, among many others). Grohmann builds on this previous work and succeeds in describing the full range of left dislocation constructions found in German and characterizing them according to criteria which include: connectivity effects, size of the intonational break and type/position of the resumptive element. I follow his typology of left dislocation constructions in my description of Czech and conclude that the system of Czech left dislocation is very similar to that of German Grohmann reports that German, like Czech, exhibits three types of left dislocation constructions: CLD and HTLD I, II. In all three types there is a left-edge constituent which is coreferential with a clause-internal resumptive element.
. In Grohmann’s 2003 description of left dislocation in German, he suggests that HTLD I and II both exhibit optional case matching. Since HTLD I with case matching and CLD are indistinguishable in the surface syntax, he uses intonation to tease apart these two constructions. German CLD does not have an obligatory intonational break between the left dislocate and the main clause, but HTLD I, II do. Since both CLD and HTLD I exhibit intonational breaks in Czech, I feel that this is an unreliable means of distinguishing these two constructions. I assume, for the purposes of this book, that HTLD I lacks case matching.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
German CLD and HTLD I are similar in that the resumptive element is found at the left edge of the clause; they differ in that CLD, but not HTLD I, requires case matching between the resumptive and the left dislocate. Also, in CLD the resumptive is obligatorily a demonstrative pronoun, while the resumptive in HTLD I can be either a demonstrative or a personal pronoun. (6) a.
Diesen Mann, den habe ich noch nie gesehen. this.acc man that.acc have I yet never seen “This man, I’ve never seen him before.”
b. Diese-r/-n Mann, den/ihn habe ich noch this.nom/acc man that.acc/him.acc have I yet
nie gesehen. never seen
“This man, I’ve never seen him before.”
CLD
HTLD I
HTLD II is similar to HTLD I in that case matching is not required. It differs, though, in that the resumptive element appears not at the left edge of the clause, but in the middlefield. c.
Diese-r/-n Mann, ich habe den/ihn noch. this.nom/acc mann I have that.acc/him.acc yet
nie gesehen never seen
“This man, I’ve never seen him before.”
HTLD II
(Grohmann 2003: (25a–c))
German HTLD II differs from the parallel construction in Czech in that the resumptive element is a full pronoun, rather than a clitic. Building on previous work in Cinque 1977 and Vat 1981/1997, Grohmann 2003 revitalizes a movement analysis of left dislocation to account for connectivity effects which hold between the left dislocate and the clause-internal gap in CLD. Grohmann 2003 suggests that CLD constructions are generated by movement of the left dislocate from a clause-internal position to a position at the left edge of the clause. Under Grohmann’s analysis, the demonstrative resumptive is a spelled out trace of movement (see chapter 5 for a thorough discussion of resumption). German left dislocation constructions have also been considered recently in Frey 2004a, b; Shaer and Frey 2004; and Frey 2005. Frey 2005 adds to Grohmann’s discussion by suggesting that there is a common topic interpretation associated with the two positions in which the resumptive can appear. The clause-initial position (in which the left-edge resumptive is found) is a structurally prominent position in German which can be filled by topical elements. According to Frey 2004a, the middlefield position (where the resumptive is found in HTLD II) is also a structural position associated with topical elements.
Chapter 3. The syntax of left dislocation
Chapter 4 considers the interpretation of left dislocation constructions and finds similarities between Frey,s analyses of German and dislocation in Czech. From a syntactic perspective, like Grohmann 2003, I argue that Czech CLD constructions, but not HTLD I, II, are formed through movement of the left dislocate from its base position to a position at the left edge. In support of a movement analysis of CLD, Grohmann presents evidence suggesting that the left dislocate and the clauseinternal gap are syntactically connected. However, if this construction is, indeed, derived through movement of the left dislocate from a clause-internal position, through a functional specifier, to a position on the left periphery, there should also be evidence of a syntactic connection between the resumptive and the left dislocate. Grammatical gender matching effects in Czech provide new evidence of such a connection, §6. 3. Syntactic approach to left dislocation in Czech Evidence from reconstruction, to be presented in sections 4 and 5, suggests that CLD, but not HTLD I, II, is derived by movement. To account for connectivity effects in CLD, I propose that CLD involves movement of the left dislocated element from its base position within the clause through [Spec, IP] to the specifier position of a functional projection in the C-domain, TopP. The resumptive pronoun is a Spelled Out copy of the left dislocated element, (7).
(7) CLD TopP XP
IP dem<−XP
I′ I
vP ...XP...
HTLD I, II constructions have a different derivation. Since reconstruction effects are absent in both constructions, I suggest that hanging topics are not generated by movement but are base generated in a functional projection (FP) adjoined to the Top projection and related to the resumptive pronoun through coreference. This resumptive is either a demonstrative which has topicalized to [Spec, IP], (8a), or a second position clitic, (8b).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(8) a.
HTLD I FP TopP
DP
IP dem1
I′ I
vP ... DP1 ...
b. HTLD II FP DP
TopP IP YP
I′ I clitic1
vP ... DP1 ...
In (8a–b) the hanging topics appear in a higher position than do CLD’ed elements. The fact that HTLD’ed elements can precede CLD’ed suggests that this is the correct analysis; co-occurrence data are presented in §9. A syntactic analysis of CLD constructions is presented in chapter 5, and that of HTLD is discussed in §9 of this chapter. In the next section, I provide evidence from reconstruction that CLD exhibits connectivity effects, while HTLD I, II do not.
4. Evidence for a movement account of CLD There is a tight syntactic connection between the CLD’ed element and the clauseinternal gap. In addition to case matching between the CLD’ed element and the
Chapter 3. The syntax of left dislocation
resumptive pronoun, left dislocated elements in these constructions reconstruct to clause-internal positions. In this section, evidence for reconstruction is provided from Conditions A and C, and from quantifier binding. It is important to note that Czech lacks logophoric pronouns entirely. Both pronominal and possessive reflexives must be c-commanded by their antecedent at some point in the derivation, see (45) in chapter 2 (see also Sturgeon 2003 for additional evidence). 4.1 Conditions A and C Reflexive pronouns can be dislocated in CLD constructions. This suggests that these elements undergo reconstruction to a clause-internal position. In chapter 2, I showed that Czech possessive and non-possessive reflexives obey Condition A; a version of the binding conditions is provided in (9–10), repeated from (41–43) in chapter 2. (9) a.
Binding Domain: A binding domain for α is the most deeply embedded Complete Functional Complex (CFC) containing α in which the basic binding requirements for α can be met.
b. Complete Functional Complex: An XP in which all the θ-roles compatible with a lexical head are assigned in A-positions. (10)
Condition A: For α, an anaphor, its basic binding requirements can be met in principle in a category C iff there is an assignment of indices to DP’s within C (perhaps different from the actual assignment of indices within C) according to which α is A-bound.
Possessive reflexives can be left dislocated in CLD constructions, suggesting that reconstruction to a clause-internal position for interpretation of the possessive is possible, (11). Accusative case matching between the left dislocated element and the reflexive pronoun provides evidence that this is an example of CLD. (11) Svého1 nejlepšího přítele, toho má Honza1 rád. selfś best friend.acc that.acc has Honza joy “His1 best friend, Honza1 likes him.”
CLD
If reconstruction did not occur in (11), the reflexive possessive would not be ccommanded by its antecedent at any level of the derivation and (11) would be expected to be ungrammatical. The same pattern is found with non-possessive reflexive pronouns. Again, a c-commanding antecedent is required. The reflexive pronoun embedded in the picture NP in (12) is ungrammatical because it lacks a c-commanding antecedent (see Sturgeon 2003 for a discussion of binding and picture NP’s in Czech).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(12) Honza1 rád čte. *Knížky o sobě1 ho velmi zajímají. Honza joy reading books about self him.acc.cl very-much interest “Honza1 likes to read. Books about himself1 interest him very much.”
Non-possessive reflexive pronouns also grammatically occur in CLD constructions, (13). In order to be c-commanded by its antecedent, Honza, the nominal containing the reflexive pronoun must originate in a clause-internal position. Note, again, that there is case matching between the left dislocated element knížku (“book”) and the resumptive pronoun. (13) Jenom knížku o sobě1, tu má Honza1 rád. only book.acc about self, that.acc has Honza joy “Only the book about himself1, that Honza1 likes.”
CLD
Given that Czech lacks logophoric uses of reflexive pronouns, reflexive pronouns in Czech must be c-commanded by their antecedent at some level of the derivation. Reconstruction to a clause-internal position accounts for the grammaticality of the left dislocated reflexives in (11) and (13). R-expressions in Czech respect Condition C; the example in (14) is ungrammatical if the two R-expressions corefer. (14) Honza1 podal Honzovi2/*1 pivo. Honza passed Honza.dat beer. “Honza1 passed Honza2/*1 a beer.”
CLD’ed elements are also subject to Condition C. Reconstruction of the left dislocated element in (15) triggers a violation of Condition C if the null subject and Martin co-refer because the null subject pronoun c-commands the base position of the left dislocated element. (15) Faktu, že Martin1 je chudý, tomu nepřikládá2/*1 fact.dat that Martin is poor that.dat neg-attach
CLD
velký význam. big meaning “The fact that Martin1 is poor, he2/*1 doesn’t attach much importance to it.”
The sentence in (15) is grammatical, however, if the R-expression, Martin, and the subject of the matrix verb do not corefer. Again, Condition C provides evidence for reconstruction of the CLD’ed element to a clause internal position. 4.2 Quantifier binding A pronoun interpreted as a bound variable must be c-commanded by the quantifier that binds it, as was shown in chapter 2.4 Bound pronominals can occur . Bound pronominals are homophonous with reflexive pronouns when they appear in the same clause as their antecedent.
Chapter 3. The syntax of left dislocation
in CLD constructions, (16). Note accusative case matching between the CLD’ed element and the resumptive pronoun. (16) Svého1 nejlepšího přítele, toho má každý1 rád. self ’s best friend.acc, that.acc has every joy “One’s1 own best friend, everyone1 loves them.”
CLD
The grammaticality of (16) suggests that the bound pronominal was c-commanded by its antecedent at some level of the derivation. Reconstruction of the left dislocated element would account for the grammaticality of this example. Reconstruction for quantifier binding is also possible in contexts of negation, (17). No one grammatically binds the pronominal within the left dislocated DP. (17) Svého1 psa, toho nikdo1 nemá rád. self ’s dog.acc that.acc no-one neg-has joy “One’s1 own dog, no one1 likes it.”
CLD
As in the previous examples, the grammaticality of (17) is expected if the CLD’ed element reconstructs to a clause-internal position. That CLD’ed constituents are evaluated in their base position for purposes of reflexive and quantifier binding follows from a movement analysis of CLD constructions. 5. Evidence for non-movement analyses of HTLD I, II Neither HTLD construction, on the other hand, exhibits reconstruction effects: it is not possible to left dislocate reflexive or bound pronouns in either. This suggests that hanging topics are base generated in a clause-peripheral position and related to their resumptive element through a process akin to cross-sentential pronominalization. The appearance of optional case matching between the resumptive element and the hanging topic in HTLD II, then, is not related to a movement derivation; there is no obligatory correlation between case matching and movement in left dislocation. This raises the question of the mechanism by which case is transmitted between a resumptive element and its antecedent. I return to this issue in §9. 5.1 Reconstruction and HTLD I Reflexive pronouns cannot be left dislocated in HTLD I; the possessive reflexive in (18) is ungrammatical. This ungrammaticality is expected if hanging topics are
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
not generated by movement, but are base generated at the left edge. In that case, the reflexive is not c-commanded by its antecedent at any stage in the derivation. Note, also, the lack of case matching; the hanging topic is in the nominative case and the resumptive pronoun is in the accusative. (18) *Svoje1 sestřenice Anička, tu má Honza1 rád. self ’s cousin Anička.nom that.acc have Honza joy Intended: “His1 cousin Anička, Honza1 likes her.”
HTLD I
The same results hold for non-possessive reflexives, (19). Again, there is no case matching between the hanging topic and the resumptive pronoun. (19) *Knížka o sobě1, tu má Honza1 rád. book.nom about self, that.acc has Honza happy Intended: “The book about himself1, Honza1 likes it.”
HTLD I
Reconstruction to satisfy Condition A is not possible in HTLD I constructions. HTLD’ed elements do not reconstruct to their base position for evaluation of Condition C either, (20). It is possible for an R-expression (Honzův (“Honza’s”)) contained within the left dislocate to be related to a coreferential pronoun which would have c-commanded it had it undergone reconstruction. (20) Honzův1 kůň, toho má1 velmi rád. Honza’s horse.nom that.acc has much joy “Honza’s1 horse, he1 loves it very much.”
HTLD I
These facts suggest that HTLD I is not generated through movement from a clauseinternal position to a position high in the left periphery. The same results are found with quantifier binding: a bound pronoun cannot be contained within a hanging topic. The example in (21) is HTLD I as evidenced by the lack of case matching between the left dislocated element (nominative) and the resumptive pronoun (accusative) and the appearance of the strong pronominal at the left edge of the main clause. (21) *Svůj1 nejlepší přítel, toho má každý1 rád. self ’s best friend.nom, that.acc has every(one) joy Intended: “Self ’s1 own best friend, everyone1 loves him.”
HTLD I
Reconstruction judgments in HTLD I and CLD were extremely robust across consultants. All speakers consulted agreed that reflexive elements were possible in the left dislocate in CLD, but not HTLD I, constructions. The fact that these judgments were so robust, given the usual variability involved in judgments on reflexive pronouns, provides strong support for the analysis.
Chapter 3. The syntax of left dislocation
5.2 Reconstruction and HTLD II Judgments for HTLD II are more difficult to obtain because these constructions tend to be a bit degraded when compared with CLD and HTLD I. The fact remains, though, that HTLD II patterns with HTLD I with respect to reconstruction effects. Bound pronominals contained within left dislocates are strictly ungrammatical in HTLD II. Consider the ungrammatical (22a–b). The left dislocated element is in the nominative case in (22a); case matching between that element and the resumptive clitic is absent. (22) a. *Svoje1 dcera, každá1 matka jí dala do self ’s daughter.nom every mother her.dat.cl gave to
školy nové šaty. school new clothes
“Her 1 daughter, every1 mother gave her new clothes for school.”
HTLD II
In (22b) the left dislocate and the pronominal match in dative case; the result is ungrammatical as well. b. *Svojí1 dceři, každá1 matka jí dala do self ’s daughter.dat every mother her.dat.cl gave to
školy nové šaty. school new clothes
“Her 1 daughter, every1 mother gave her new clothes for school.”
HTLD II
When reflexive binding is considered, however, the facts are not quite as clear. Reconstruction for the evaluation of Condition A is not possible when case matching between the left dislocate and the resumptive clitic is absent, (23a). (23) a. *Ta svoje1 holka, maminka1 jí dala sušenky. HTLD II that self ’s daughter.nom mother her.dat.cl gave cookies “Her1 daughter, mom1 gave her cookies.”
In contexts of case matching, however, the example is degraded, rather than strictly ungrammatical, (23b).5 b. ?Té svojí1 holce, maminka1 jí dala sušenky. HTLD II that self ’s daughter.dat mother her.dat.cl gave cookies “Her1 daughter, mom1 gave her cookies.”
. Since all other tests point to a lack of reconstruction in ‘‘case matching’’ HTLD II, I leave this inconsistency as a outstanding issue for future research and assume that neither HTLD construction undergoes reconstruction.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
HTLD II lacks reconstruction for the evaluation of Condition C in contexts with case matching as well as those without, (24a–b). Coreference between the possessive within the left dislocate and the pronominal subject is grammatical in both examples. (24) a.
Janina1 fenka, má1 ji moc ráda. Jana’s dog.nom have her.acc.cl much joy “Jana’s 1 dog, she1 loves it very much.”
b. Janinu1 fenku, má1 ji moc ráda. Jana’s dog.acc have her.acc.cl much joy “Jana’s1 dog, she1 loves it very much.”
HTLD II
HTLD II
With the possible exception of (23b), HTLD II patterns with HTLD I in that there is no evidence that the left dislocate has a syntactic connection with the clauseinternal gap. Hanging topics appear to be base generated at the left edge of the clause and associated with the clause-internal resumptive through coreference. 5.3 Conclusion The data presented in this section and the previous one suggest that a tight syntactic connection exists between the left dislocated element and the clause internal gap in CLD constructions. The same is not true of HTLD I and II. In those constructions there is no evidence of reconstruction of the hanging topic to a clauseinternal position, (23b) excepted, and, thus, no connection between hanging topics and the clause-internal gap. I suggest that these differences stem from their distinct syntactic derivations. CLD constructions involve movement of the left dislocated element from a clause-internal position to the left edge; hanging topics are base generated on the left periphery of their clause. 6. Gender and resumption In the previous sections I showed that there is a syntactic connection between the left dislocated element and the clause-internal gap in CLD; HTLD I, II both lack such a connection. If the left dislocated element in a CLD construction moves from its base position through [Spec, IP] to [Spec, TopP], we would expect evidence of a tight syntactic connection not only between the CLD’ed element and the clause-internal gap, but also between the CLD’ed element and the resumptive pronoun in [Spec, IP]. These effects should be absent in HTLD constructions; the relationship between the hanging topic and the left edge resumptive is one of coreference, similar to that found in cross-sentential pronominalization. Evidence of a syntactic connection between a left dislocate and its corres ponding resumptive can be found through consideration of the pronominalization
Chapter 3. The syntax of left dislocation
behavior of a certain class of nouns: nouns which exhibit a grammatical/natural gender mismatch. Nouns of this type occur cross-linguistically and are often unique in their behavior as antecedents; in some languages, including Czech, they can antecede a pronoun which matches either the grammatical gender of the antecedent DP or the natural gender of its discourse referent. Děvče (‘‘girl.neut’’) is a noun of this type in Czech.6 Its grammatical gender is neuter, but it refers to discourse referents whose natural gender is female.7 We know that this noun is grammatically neuter because the gender of its modifiers is neuter: to chytré děvče (“that.neut clever.neut girl.neut”). However, cross-sententially, pronominals referring to this discourse referent can be either feminine or neuter, (25). (25) Představila jsem ho tomu děvčeti a introduced aux.1sg.cl him.acc.cl that girl.dat.neut and ta/to mu stiskla/o ruku. that.fem/that.neut him pressed.fem/neut hand “I introduced him to that girl and she shook his hand.”
We first consider the predictions for HTLD constructions with respect to this class of nouns. If the left dislocate in HTLD I, II were base generated at the left edge and related to the resumptive through a process of pronominalization, we expect that both the feminine and neuter pronouns would be possible under coreference with the DP to děvče (“that girl.neut”). This is, indeed, what is found. Consider (26a). The demonstrative resumptive can be either feminine or neuter. (26) a.
To děvče, toho/tu jsem that girl.neut.nom that.neut.acc/that.fem.acc aux.1sg.cl
představila Janě. introduced Jana
“That girl, I introduced her to Jana.”
HTLD I
The HTLD II example in (26b–c) illustrates the same pattern. Resumption of the left dislocate can match the grammatical gender of the DP or the natural gender of
. Other such nouns in Czech include slang terms for teenaged girls: pulec (‘‘tadpole.masc’’), žabec (‘‘young frog.masc’’). Both of these terms are grammatically masculine, but have discourse referents which are feminine. I use děvče (‘‘girl.neut’’) instead of these other terms since this word is a part of standard Czech and not a slang term. . Laura Janda (p.c.) suggests that there is no mismatch in this case because děvče (‘‘girl’’) refers to sexually immature members of the female gender whose natural gender is, in fact, neuter. Whatever the analysis of the gender mismatch, the fact remains that pronominals refering to the discourse referent of děvče can be either feminine or neuter.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
its discourse referent. (26b) illustrates an example without case matching between the left dislocate and the resumptive and (26c) an example with case matching. b. To děvče, představila jsem that girl.neut.nom introduced aux.1sg.cl
ho/ji Janě. it.neut.acc/her.fem.acc Jana.dat
“That girl, I introduced her to Jana.”
c.
Tomu děvčeti, představila jsem that girl.neut.dat introduced aux.1sg.cl
mu/jí Janu. it.neut.dat/that.fem.dat Jana.acc
“That girl, I introduced Jana to her.”
HTLD II
HTLD II
The results found in (26) are compatible with the analysis for HTLD I, II presented thus far. Resumption in these constructions parallels cross-sentential pronominalization because HTLD’ed elements are base generated in their surface position and are in a co-referential relationship with the clause-internal pronominal element. Given the analysis outlined above, CLD constructions are predicted to differ. The left dislocated XP moves from its base position within the clause, through [Spec, IP], to a left peripheral position; I assume that the resumptive element is a spelled out copy of the left dislocate. If the resumptive element in [Spec, IP] is a syntactic copy of the left dislocate, the gender of the resumptive element should obligatorily match the grammatical gender of the left dislocate, not the natural gender of its discourse referent. This is what is found, (27). The only possible gender of the resumptive is a neuter; the feminine form is ungrammatical for all speakers consulted. Case matching between the left dislocated XP and the leftedge resumptive element indicates that this is, indeed, a CLD construction.8 (27) Tomu děvčeti, tomu/*té that girl.neut.dat that.neut.dat/that.fem.dat
CLD
jsem dala jedničku. aux.1sg.cl gave one “That girl, I gave her an A.”
. Researchers have suggested that a factor conditioning the choice between the genders on the pronoun in cases like those described above is recency of mention. If the antecedent has been mentioned recently, grammatical gender is more likely to override the natural gender of the discourse referent in pronoun choice. However, if the antecedent is farther away, the natural gender has a bigger impact on pronoun choice. I have found this to be true in Czech in field work on crosssentential pronominalization. However, the fact that the resumptive and its antecedent are equidistant in HTLD I and CLD suggests that what we are seeing in (27) is not simply a recency effect.
Chapter 3. The syntax of left dislocation
The data presented in this section provides a type of support for the proposed syntactic connection between the left dislocate, the resumptive pronoun and the clause-internal gap in CLD constructions which has not yet been discussed in the literature. Reconstruction effects suggest that the CLD’ed element is interpreted in its base position for binding, and obligatory grammatical gender matching suggests a syntactic connection between the resumptive and the left dislocate. These effects are absent in HTLD I, II. There is no evidence from reconstruction of a syntactic connection between the hanging topic and the clause-internal gap nor evidence from gender matching of a syntactic connection between the hanging topic and its coreferent resumptive.
7. Phrasal categories Another difference between CLD and HTLD I constructions is the inventory of phrasal categories that appear in dislocation constructions.9 Czech patterns with German LD in that any category can figure in CLD, but only DP’s can appear in HTLD constructions (see Grohmann 2003). Grohmann uses the larger intonational break associated with HTLD to show that only DP’s can dislocate in HTLD constructions. I employ connectivity tests which are more reliable for Czech. When considering phrasal categories other than DP’s, HTLD and CLD constructions are, on the surface, indistinguishable because case matching is not a relevant diagnostic. Therefore, we must turn to connectivity effects. Consider the left dislocated prepositional phrase in (28). It is grammatical to embed a reflexive within a dislocated PP; this suggests a CLD derivation. In order for the reflexive to be grammatical, Condition A must be evaluated at the base position of the PP; at that point in the derivation the reflexive is c-commanded by its antecedent. (28) [Se svým1 bratrem]3, s tím pojede Honza1 t3 na výlet. with self ’s brother.instr with that.instr will-go Honza on trip “His1 brother, Honza1 is going to go on a trip with him.”
The example in (28) suggests only that a CLD derivation is available; HTLD is not ruled out. When an R-expression is left dislocated, Condition C is violated if a pronoun c-commands the base position of the left dislocate. If an HTLD derivation, as well as a CLD derivation, were available, the expectation would be that (29) would
. We know that HTLD II is limited to nominals because clitic pronouns are restricted to DP’s; there are no clitic pronouns which refer to, for example, locative or temporal XP’s.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
be judged grammatical under both coindexations since a derivation would exist under which reconstruction did not occur. (29) [S Honzovým1 psem]3, s tím pojede2/*1 t3 na výlet. with Honza’s dog.instr with that.instr will-go on trip “Honza’s1 dog, he2/*1 will go on a trip with it.”
The ungrammaticality of a coindexation under which the R-expression embedded within the left dislocate (Honzovým (‘‘Honza’s.instr’’)) corefers with the subject of the matrix verb suggests that an HTLD derivation of (29) is unavailable. Given the above examples, I assume that PP’s can left dislocate in CLD, but not HTLD I, constructions. Other categories can dislocate in CLD as well. As expected, the resumptive element matches the left dislocate in category. Consider the left dislocated CP in (30). (30) [Abych Honzovi1 uvařila vepřové maso]3, to po c.cond.1sg Honza.dat cooked pork meat that.neut from mě chce2/*1 t3. me wants “For me to cook Honza1 pork, he2/*1 wants that from me.”
Reconstruction for the evaluation of Condition A is not a relevant diagnostic when considering fronted finite clauses because reflexive pronouns are clause-bounded (see Sturgeon 2003). However, Condition C is evaluated after reconstruction as evidenced by the ungrammaticality of coreference between the matrix subject and the embedded object, Honza. Regardless of the position of the null-pronominal within the matrix clause ([Spec, vP] or [Spec, IP]), it c-commands the embedded object and induces a violation of Condition C. Since (30) is ungrammatical under co-reference, the matrix subject must c-command the embedded clause at some point in the derivation. Strong evidence supporting a movement approach to left dislocated CP’s (i.e., that they occur in CLD, but not HTLD) is presented in (31–32). In (31), a dislocated CP appears at the left edge of the clause. (31) Jestli se máš dobře, to bych se zajímala. if refl-cl have.2sg well it.neut cond.1sg.cl refl-cl wonder “Whether or not you are doing well, I wonder about that.”
There are two possible derivations for (31). One in which the CP was base-generated at the left edge (HTLD) and one in which it moved there from a clause-internal position. If it were base generated at the left, the sentence in (32), in which the verb, zajímat se (“wonder.inf”), takes a DP complement, should be grammatical.
Chapter 3. The syntax of left dislocation
However, this is not what is found. This verb, like the English equivalent, takes only a CP or PP complement; a DP complement is ungrammatical, (32). (32) *Zajímám se to. wonder.1sg refl-cl it.cl Lit: “I wonder it.”
Given that the matrix verb in (31) requires a CP (or PP) complement, (31) must have a CLD derivation. If the CP dislocate moves from a base-position within the matrix vP, the subcategorization requirements of zajímat se (“wonder.inf”) are satisfied before movement. However, if the CP is base generated at the left edge, the pronominal resumptive, to (‘‘it.neut’’) would not be able to satisfy the subcategorization requirements of the matrix verb. Given that Condition C appears to be evaluated after reconstruction in (30) and that the subcategorization requirements of zajímat se (‘‘wonder.inf’’) can only be satisfied under a movement analysis, (31), I conclude that CP’s can be left dislocated only in CLD constructions. Manner adverbials can also appear at the left edge in CLD constructions, see (33). The corresponding demonstrative pronoun for manner adverbials is tak (‘‘thus’’). (33) Potichu, tak zavřel dveře. quietly thus closed doors “Quietly, I closed the door like that.”
Reconstruction, however, is not a possible test for manner adverbials because their attachment point is too high in the clause-internal syntax to contain a reflexive pronoun. A reflexive possessive is not possible in (34), though the potential antecedent linearly precedes the reflexive pronoun. (34) Honza1 vešel do pokoji hlučně jako jeho1/*svoje1 kočka. Honza entered to room loudly like his/self ’s cat “Honza entered the room loudly like his cat.”
Though it is not possible to provide evidence from reconstruction that manner adverbials appear in CLD and not HTLD constructions, given the fact that there is evidence for other categories (such as CP’s and PP’s), I will assume that manner adverbials are only possible in CLD constructions. The evidence provided in this section that any category can dislocate in CLD constructions parallels the behavior of other A-bar movement operations, such as wh-movement. Additional evidence is provided in the following section that CLD, as well as the movement operation associated with the resumptive in HTLD, involves A-bar movement.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
8. Evidence for A-bar movement of resumptive pronouns The previous sections have detailed a range of differences between CLD constructions on the one hand and HTLD I, II on the other. Movement vs. non-movement derivations were posited to explain these differences. If the difference between these two types of constructions is indeed one of movement vs. non-movement, we would expect that the behavior of these constructions would also differ with respect to islands. The expectation would be that the movement construction, CLD, would exhibit island sensitivity, while the non-movement HTLD I, II would not. This prediction, however, is only partially borne out in the data. It is true that CLD’ed constituents cannot be related to a gap within an island, but this restriction also holds for HTLD I. The reason for this apparent inconsistency is that, in both CLD and HTLD I, the resumptive pronoun moves from its base position to a position at the left edge of the main clause. In other words, regardless of whether the left dislocate moves to its surface position or is base generated there, there is movement of the resumptive. Therefore, island effects emerge in both constructions. HTLD II constructions differ, however. Since the resumptive pronoun remains in the clause within which it was generated as a clitic and does not move to the left periphery, island effects should be absent. This is what is found. Before demonstrating these effects, however, I establish that long distance dependencies are, in fact, possible in CLD and HTLD I, II constructions. 8.1 Long distance dependencies That contrastive topicalization constructions (i.e., movement to [Spec, IP] without resumption) involve A-bar movement was established in chapter 2. This construction exhibits long distance dependencies and island sensitivity, and constituents of any category can front. In this and the following section, I show that fronting of the demonstrative resumptive in HTLD I and CLD involves a parallel movement. We first consider long distance dependencies. The left dislocate can be associated with a gap in an embedded finite or non-finite clause in CLD.10 Consider (35a–c). Case matching between the left dislocate and the left edge resumptive
. Unlike wh-movement, which is fully grammatical out of all embedded indicative and subjunctive clauses, extraction out of an embedded indicative clause in CLD and HTLD tends to be slightly degraded. This is also found with long distance dependencies in contrastive topicalization (see chapter 2). I do not have an explanation of this. Petr Sgall (p.c.) has suggested that HTLD may be more felicitous from a finite indicative clause than CLD. Sgall commented that the prototypical case would be HTLD I as in (i), but (ii) (CLD) is also possible.
Chapter 3. The syntax of left dislocation
indicate that these are instances of CLD. The examples provided below illustrate finite, subjunctive and non-finite embedded clauses respectively. (35) a.
Chemii, tu1 Honza potvrdil, že jeho dcera bude chemistry.acc that.acc Honza confirmed c his daughter will
studovat t1. study.inf
“Chemistry, Honza confirmed that his daughter is going to study it.”
b. Tu zkoušku, tu1 chci aby Petr udělal t1. that exam.acc that.acc want c.cond.3sg Petr passed “That exam, I hope Petr passes it.” c.
CLD
CLD
Písničku od Beatles, tu jsem Petra slyšela zpívat t1. song.acc from Beatles that.acc aux.1sg Petr.acc heard sing.inf “The Beatles song, I heard Petr sing one.” CLD
Long distance dependencies are also grammatical in HTLD I, (36a–c). The accusative resumptive pronoun has moved out of an indicative embedded clause in (36a). Note the lack of case matching between the left dislocated element and the resumptive pronoun. (36) a.
A ten druhej? No, toho1 myslím, že nemůžu and that second.nom well that.acc think c neg-can
zapomenout t1. forget.inf
“And the second? Well, I think I can’t forget him.”
HTLD I
(36b–c) illustrate long distance dependencies in subjunctive and non-finite clauses. b. A ten druhej? No, toho1 chci abych and that second.nom well that.acc want c.cond.1sg
HTLD I
mohla zapomenout t1. could forget.inf “And the second? Well, I wish I could forget him.”
(i)
Zdeněk, toho1 myslím, že Marie pozvala t1 na tu párty. Zdeňek.nom that.acc think c Marie invited on that party ‘Zdeňek, I think Marie invited him to the party.‘
HTLD I
Zdeňka, toho1 myslím, že Marie pozvala t1 na tu párty. Zdeňek.acc that.acc think c Marie invited on that party
CLD
(ii)
This observation also provides additional support for the movement/non-movement distinction between CLD and HTLD. Coreference is more likely to be available across a distance than is movement.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
c.
Ta peněženka, tu1 jsem viděla Honzu that wallet.nom that.acc aux.1sg.cl saw Honza.acc
ukrást t1. steal.inf
“That wallet, I saw Honza steal it.”
HTLD I
Like other A-bar movements in Czech, the resumptive in left dislocation can be associated with a gap in an embedded clause. Though there is no movement out of the embedded clause, antecedence over a distance is also grammatical in HTLD II constructions. The examples in (37a–c) illustrate both case matching and non-case matching HTLD II in the three types of embedded clauses considered above: finite, subjunctive and non-finite. (37) a.
Honza/Honzovi, doufám, že Martina mu Honza.nom/dat hope c Martina him.dat.cl
pozve na ten večírek. invites on that party
“Honza, I hope Martina invites him to the party.”
b. Honza/Honzovi, chci abys mu Honza.nom/dat want c.cond.2sg him.dat.cl
představila Janu. introduced Jana.acc
“Honza, I want you to introduce Jana to him.”
c.
Ta holka/Tu holku, viděla jsem maminku dovést that girl.nom/acc saw aux.1sg.cl mom bring
ji do školy. her.acc.cl to school
“That girl, I saw her mother bring her to school.”
HTLD II
HTLD II
HTLD II
Now that it has been established that all three left dislocation constructions allow long distance dependencies, we turn to island sensitivity. 8.2 Island sensitivity Both the movement left dislocation construction, CLD, and the non-movement HTLD I exhibit island effects due to movement of the resumptive element to the left edge. Though long distance dependencies are allowed in CLD constructions, the gap cannot be separated from the dislocate by an island boundary. The example in (38a) illustrates this with respect to a wh-island. (38) a. *Petrovi, tomu2 by mě zajímalo Petr.dat that.dat c.cond.3sg me interested
CLD
Chapter 3. The syntax of left dislocation
koho1 Marie představila t1 t2. who.acc Marie introduced
Intended: “Petr, I wonder who Marie introduced to him.”
(38b) involves a Complex NP Island; the gap in a CLD construction cannot be found within a relative clause. b. *Zdeňka, toho1 Hana viděla chlapa, který Zdeněk.acc that.acc Hana saw guy rel
zná t1. knows
Intended: “Zdeněk, Hana saw a guy who knows him.”
CLD
HTLD I parallels CLD with respect to island sensitivity. Consider (39a–b). The gap associated with the hanging topic cannot be within a wh-island, (39a), or a relative clause, (39b). (39) a. *Petr, tomu2 by mě zajímalo koho1 Petr.nom that.dat c.cond.3sg me interested who.acc
Marie představila t1 t2. Marie introduced
Intended: “Petr, I wonder who Marie introduced to him.”
b. *Zdeněk, toho1 Hana viděla chlapa, který zná t1. Zdeněk.nom that.acc Hana saw guy rel knows Intended: “Zdeněk, Hana saw a guy who knows him.”
HTLD I
HTLD I
HTLD II is a good testing ground for the non-movement nature of HTLD. Since there is no movement of the resumptive out of its clause, island effects should not appear. As expected, HTLD II differs from both CLD and HTLD I in that a resumptive element can be separated from its antecedent by an island boundary. Consider (40a–b); the hanging topic is associated with a resumptive element within a Complex NP island. In both (40) and (41), an example of HTLD II without case matching between the left dislocate and the resumptive precedes HTLD II with case matching. (40) a.
Honza, viděla jsem chlapa, který ho zná. Honza.nom saw aux.1sg guy rel him.acc.cl knows “Honza 1, I saw a guy who knows him1.”
HTLD II
b. Honzu, viděla jsem chlapa, který ho zná. HTLD II Honza.acc saw aux.1sg guy rel him.acc.cl knows “Honza 1, I saw a guy who knows him1.”
Similar effects are found with wh-islands, (41a–b).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(41) a.
Honza, Anička se ptala kamarádky, co Honza.nom Anička refl.cl asked friend what
mu dala k vánocům. him.dat.cl gave towards Christmas
“Honza , Anička asked her friend what she gave him for Christmas.”
b. Honzovi, Anička se ptala kamarádku, co Honza.dat Anička refl.cl asked friend.acc what
HTLD II
HTLD II
mu dala k vánocům. him.dat.cl gave towards Christmas
“Honza , Anička asked her friend what she gave him for Christmas.”
Lack of island effects in HTLD II constructions provides further evidence that, in this construction, there is no movement of the left dislocate from a clause-internal position to a position at the left edge. Island effects are not a diagnostic of movement of the left dislocate in CLD and HTLD I, however, due to movement of the resumptive demonstrative to the left edge of the main clause in both constructions. Evidence from island sensitivity and long distance dependencies suggest that movement of the resumptive element in CLD and HTLD I is an instance of A-bar movement. A base generated DP is coreferent with a resumptive element in HTLD I. In CLD, however, there is an additional movement; from its position at the left edge, the resumptive raises to [Spec, TopP]. 9. Syntactic positions of CLD and HTLD I, II Up to this point I have assumed that CLD’ed elements are in the specifier position of a functional projection above CP, TopP, while hanging topics adjoin to that projection. To probe the relative syntactic positions of CLD’ed and HTLD’ed elements, I consider cases in which the two co-occur. 9.1 The left periphery We can determine immediately that CLD and HTLD I cannot co-occur. This is because both involve a resumptive in [Spec, IP], and that position can only hold a single element, (42a). The same holds for multiple HTLD I’s, (42b).11
. If both left dislocates exhibit case matching, their appearance at the left edge is marginally possible, (i). To account for the syntactic positions of multiple resumptive elements, we must assume that the second resumptive hangs not in the IP domain, but is adjoined to vP. That the second demonstrative follows the second position clitic supports this contention.
Chapter 3. The syntax of left dislocation
(42) a. *Honza knihu, tu tomu chci koupit. Honza.nom book.acc that.acc that.dat want buy.inf Intended: “Honza, a book, I want to buy him one.”
HTLD I, CLD
b. *Honza kniha, tu tomu chci koupit. HTLD I, HTLD I Honza.nom book.nom that.acc that.dat want buy.inf Intended: “Honza, a book, I want to buy him one.”
It is more telling to consider co-occurrence patterns with HTLD II, since the complicating effect of multiple resumptives in [Spec, IP] is absent. If hanging topics adjoin to the functional projection that hosts CLD’ed XP’s, HTLD II should obligatorily precede CLD. This is what is found. I restrict my attention to HTLD II with case matching because this variety of HTLD II is preferred by my consultants. Consider (43a). (43) a.
Honzu2 Janě1, té1 ho2 představila. Honza.acc Jana.dat that.dat him.acc.cl introduced “Honza Jana, she introduced him to her.”
HTLD II, CLD
The hanging topic obligatorily precedes the CLD’ed XP. The alternate order of the left dislocates, CLD preceding HTLD II, is ungrammatical, see (43b). b. *Janě1 Honzu2, té1 ho2 představila. Jana.dat Honza.acc that.dat him.acc.cl introduced Intended: “Jana Honza, she introduced him to her.”
CLD, HTLD II
Additionally, multiple hanging topics are predicted to be grammatical since, in principle, any number of adjoined elements are allowed. I limit myself to two adjoined DP’s since there are pragmatic restrictions on the number of hanging topics that speakers will accept. In (44) multiple HTLD II’s appear at the left edge. (44) Honzovi knížku, chci mu ji dát zítra večer. Honza.dat book.acc want him.dat.cl it.acc.cl give.inf tomorrow night “Honza the book, I want to give it to him tomorrow night.”
HTLD I and HTLD II are also predicted to co-occur. Since case matching in left dislocation is preferred by speakers and because there are pragmatic restrictions on multiple dislocates, I have not been able to generate a construction with this pattern that my consultants will accept. I attribute this gap to the combined effects of these two degrading factors.
(i)
?Honzovi knihu, tu jsem tomu chtěla koupit. Honza.dat book.acc that.acc aux.1sg.cl that.dat wanted buy.inf Lit: “Honza, the book, I wanted to buy him one.”
I leave the analysis of this example as an issue for future research.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
We have established that hanging topics are positioned higher in the syntax than are CLD’ed XP’s and are found in an adjoined position. The next question to consider is the location of these positions relative to other elements that occur in the C-domain. In the previous chapter, we established that wh-elements which co-occur with contrastive topics appear in [Spec, CP]. Can left dislocated elements precede wh-phrases in [Spec, CP]? Given the analysis presented above, LD’ed elements should be able to precede these wh-expressions in wh-questions, and, indeed, CLD’s XP’s can. A common way to form wh-questions in Slavic languages, and Czech in particular, is for an LD’ed element to precede the clause. Consider (45a). A CLD’ed element precedes a wh-expression which is positioned in [Spec, CP]. (45) a.
Tu knížku, komu tu jsi dala? that book.acc who.dat that.acc aux.2sg.cl gave “That book, who did you give it to?”
CLD
In (45b), a left dislocate precedes a wh-question in an HTLD II construction. In this case the wh-expression is in [Spec, IP]. b. Toho psa/ Ten pes, komu jsi ho dala? HTLD II that dog.acc/ that dog.nom who.dat aux.2sg.cl it.acc.cl gave “That dog, who did you give it to?”
HTLD I cannot occur with wh-questions, however. This is because the resumptive pronoun in [Spec, IP] in these constructions is interpreted obligatorily as a contrastive topic; this interpretation is indicated by an intonational rise (see chapter 2 for a discussion of this). The resumptive element in HTLD I is incompatible with this discourse function, and, as a result, HTLD I cannot occur with wh-questions. This discoursal incompatibility is discussed in detail in chapter 4. HTLD II is grammatical, though, because the resumptive element, a clitic, does not appear in [Spec, IP]. The proposed syntactic structures for HTLD I, II and CLD in (7–8) make the correct predictions. HTLD I, II obligatorily precede CLD and multiple hanging topics are allowed at the left periphery. The fact that hanging topics are base generated in a position adjoined to TopP while CLD’ed elements move into the specifier of this projection accounts for these facts. Further evidence for these structural positions is found when embedded clauses are considered. 9.2 Embedded contexts CLD’ed elements, but not hanging topics, can appear at the embedded clause boundary of complements to bridge verbs. Consider (46). The CLD’ed Pan Kopyto immediately follows the complementizer in the embedded clause.
Chapter 3. The syntax of left dislocation
(46) Pana Nováka oni neznají, ale babička řekla, Mr. Novak.acc they neg-know but grandma said
CLD
že Pana Kopyto, toho znají. c Mr. Kopyto.acc that.acc know “Mr. Novak they don’t know, but grandma said, Pan Kopyto, they know him.”
The same is not true for HTLD I, II. These elements cannot be found at this embedded clause boundary, (47a–b). (47) a. *Honza mě řekl, že škodovka, tu si Honza me.dat told c škoda.nom that.acc refl-cl
Marie koupila. Marie bought
Intended: “Honza told me that the Škoda Marie bought it.”
b. *Honza mě řekl, že škodovka, Marie si ji Honza me.dat told c škoda.nom Marie refl-cl it.acc.cl
koupila. bought
Intended: “Honza told me that the Škoda Marie bought it.”
HTLD I
HTLD II
This is what is expected if these elements are adjoined to, rather than in the specifier of, TopP. Chomsky 1986 argues that it is not possible to adjoin to lexically selected projections (see also McCloskey 2005). If we assume the that TopP is part of an expanded C-domain which includes the Top projection as well as CP, these projections together form the C-domain complex which is selected by the governing bridge verb in (46–47). Adjunction to an embedded TopP, then, is predicted to be ungrammatical. CLD’ed elements, however, can appear in embedded positions because they appear in a specifier, rather than an adjoined, position. Though hanging topics cannot appear at this clause boundary, they can be associated with a gap within an embedded clause. It is grammatical for a hanging topic at the left edge of the clause to be coreferent with an resumptive element in an embedded position. (48a) is an example of HTLD and (48b) HTLD II. (48) a.
Škodovka, Honza mě řekl, že tu si Škoda.nom Honza me.dat told c that.acc refl-cl
Marie koupila. Marie bought
“The Škoda, Honza told me that Marie bought it.”
b. Škodovka, Honza mě řekl, že si ji Škoda.nom Honza me.dat told c refl-cl it.acc.cl
Marie koupila. Marie bought
“The Škoda, Honza told me that Marie bought it.”
HTLD I
HTLD II
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
The patterns presented above are consistent with a structure in which CLD’ed elements occupy [Spec, TopP] and HTLD I, II are adjoined to that projection. A schematic tree is provided in (49). (49)
FP HTLD II1
TopP
CLD2
IP dem<-XP2 I clitic1
I′ vP v′ v
VP V
XP2
In chapter 5 a more detailed syntactic account of movement in CLD constructions is presented. 9.3 The syntax of HTLD The analysis of HTLD constructions is straightforward. The hanging topic is base generated in an adjoined position dominating the main clause. There is an additional requirement that the clause to which the hanging topic is adjoined contain a coreferential element.12 In the examples provided in this chapter we have seen that
. Note that Czech also has the type of left dislocation van Riemsdijk 1997 terms ‘Loose Aboutness’ Left dislocation. In these LD constructions, there is no coreferential resumptive element within the clause, but, rather an ‘aboutness’ relationship holds between the left dislocate and the main clause. An example is given in (i). (i)
Pokud jde o ryby, mám ráda smaženého kapra. as-far-as goes about fish have joy fried carp “As far as fish go, I like fried carp.”
Chapter 3. The syntax of left dislocation
this coreferential element can be a left edge demonstrative (or, for some speakers, a pronominal) or a clitic resumptive element. A coreferential epithet, positioned at the left edge, is also allowed, see (50). (50) Honza, toho pitomce nikdy nechci vidět. Honza.nom that.acc idiot.acc never neg-want see.inf “Honza, I never want to see that idiot.”
Given that there is a requirement that a coreferential element appear within the clause to which the hanging topic is adjoined, I assume that hanging topics must enter into a binding relation with this clause-internal element. This binding relation provides a mechanism for characterizing the optional case-matching that is found in HTLD II; apparently, morphological case can be optionally transmitted through this relation between the hanging topic and the pronoun it binds. Crucially, while this binding relation provides a mechanism for case agreement, it does not provide a mechanism for reconstruction. This will follow from the analysis to be developed more fully in chapter 5, where reconstruction is a consequence of the copy theory of movement. Since HTLD involves no movement, it can involve no reconstruction. There is one remaining issue. In HTLD I constructions, the resumptive obligatorily appears at the left edge of the main clause. I suggest that this does not reflect a syntactic requirement, but a pragmatic one related to the discourse function that HTLD I instantiates: topic promotion (see chapter 4). In the preceding chapter, I posited that elements in [Spec, IP], if not associated with a CT or a focus interpretation, were construed as topical. If HTLD I promotes the discourse referent of the HTLD’ed DP to topic status, then the next occurrence of a DP referring to that discourse referent is expected to be in a position associated with the topic discourse function. As noted in §2, German places the same requirement on resumptive pronouns: they must appear either in a left edge topic position or in a topic position within the middlefield. Frey 2004b, 2005 also suggests that the discourse function of HTLD in German is topic promotion. Presumably, the position of the resumptive in German is also linked to this discourse function of the hanging topic.
These constructions often begin with the phrase pokud jde o (lit: ‘‘as far as goes about”) and only require that the left dislocate is semantically related to the following clause; a coreferential resumptive element is not required. Usually this semantic relationship is one of proper inclusion between the sets denoted by the two related elements (i.e., fish~carp). I assume that (i) instantiates a different type of left dislocation construction in which the left dislocate is base generated in a functional projection above TopP and do not consider it here.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
10. Conclusion Three left dislocation constructions were considered in this chapter: CLD, HTLD I, II. CLD, but not HTLD I or II, exhibits connectivity effects between the clauseinternal gap and the left dislocated element. This suggests that CLD’ed XP’s move from a clause-internal position to a position at the left edge. That this movement involves an intermediate position, [Spec, IP], in which a spelled out copy of the moved XP appears as a resumptive demonstrative, is supported by grammatical gender matching facts; in CLD, the resumptive element must match in grammatical gender with the left dislocate even when the natural and grammatical genders conflict. This requirement does not hold of HTLD I, II. In these constructions, the behavior of the resumptive mirrors cross-sentential pronominalization in that the resumptive can match the left dislocate in either natural or grammatical gender. These contrasts follow from a movement account of CLD and a base-generation account of HTLD I, II. In HTLD I, II, the left dislocate is base-generated in an adjoined position on the left periphery. It then enters into a binding relationship with a clause-internal resumptive element. Through this relation, case can be optionally transmitted from the resumptive to the left dislocate in HTLD II. In the following chapters I focus primarily on CLD and link the pragmatics of this construction to its syntax. Chapter 4 develops a pragmatic analysis of CLD as contrastive topic marking, and supports it with corpus data. In chapter 5, I focus on the syntactic analysis of CLD and link the appearance of the left-edge resumptive with the discourse function of CLD.
chapter 4
The discourse functions of left dislocation 1. Introduction In addition to being different syntactically, the three left dislocation constructions in Czech, CLD, HTLD I and II, differ pragmatically. In this chapter, I provide a formal framework within which to understand their discourse functions. The chapter focuses on CLD and HTLD I, but the distribution of HTLD II constructions suggests that their discourse function mirrors that of HTLD I. Many discoursal analyses of English left dislocation constructions (LD) have been proposed and most suggest a topic/comment structure (see Gundel 1974, 1985; Halliday 1967; Reinhart 1981; among many others). Chafe 1975 and Geluykens 1992, on the other hand, suggest that left dislocation constructions mark contrast. Prince 1997, in an influential in-depth corpus study, argues against researchers who suggest that left dislocation constructions in English are “topic-marking”. She argues, instead, that English left dislocation has three functions: simplifying the discourse by left dislocating a discourse new entity which would otherwise occur in the subject position, resumptive pronoun LD’s which rescue potential island violations and poset LD’s. Prince’s characterization of poset LD’s has been the most influential and has laid the framework for a new understanding of the pragmatics of left dislocation structures. According to Prince, the discourse referent of the left dislocate in poset LD is related to the previous context through membership in a cognitively salient set. Gregory and Michaelis 2001, through a corpus study of spontaneous spoken examples from the Switchboard corpus, consider Prince’s characterization of left dislocation in English and argue that, though there are some differences between the three sub-cases, there is one over-arching discourse function: topic promotion. Unlike English, Czech has three distinct LD constructions, CLD and HTLD I, II, which contrast syntactically in that the preposed element in CLD constructions moves to its surface position while hanging topics are base generated at the left edge. These syntactic differences correlate with discoursal differences. Through textual analysis of the distribution of two of these constructions, CLD and HTLD I, I show that HTLD constructions exhibit the discourse function normally associated
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
with LD constructions in English, that of topic promotion (see also Gregory & Michaelis 2001). CLD constructions, on the other hand, pattern with contrastive topic constructions in English such as topicalization (Prince 1981) and the B-accent (Jackendoff 1972). Researchers such as Frey 2005 and Grohmann 2003 also suggest that the discourse function of CLD in German is contrastive, though a thorough analysis of the discourse function of this construction has not yet been undertaken. I start with CLD and present evidence that it expresses the contrastive topic discourse function. To this end, I build on earlier work on contrastive topic, Prince 1981 and Büring 2003 in English, and Hajičová and Sgall 2004 on Czech. Büring’s 2003 formal pragmatic analysis serves as the basis for my discussion of CLD constructions in Czech. The discourse function of HTLD constructions contrasts sharply with that of CLD. The referents of HTLD’ed XP’s do not contrast with other entities in the discourse. Instead, HTLD constructions promote entities to topic status. Despite the distinct discourse functions of these two constructions, there is a common thread. Though in neither construction is the discourse referent the topic of the preceding discourse, in both constructions the discourse referent of the left dislocate has been either previously mentioned or is a member of a salient set in the discourse which has been previously mentioned. To illustrate the discourse function of these two constructions, spontaneous textual examples have been identified from the Czech National Corpus and through Google searches. These examples have been supplemented with field research.
2. Contrastive topic Two influential strands of research on contrastive topic marking have developed separately. Formal semanticists have focused primarily on the contrastive topic interpretation associated with certain intonational patterns (the fall-rise intonational contour, the hat or bridge accent, the B-accent) (Jackendoff 1972; Roberts 1996, 1998; Krifka 1999; Steedman 2000; Büring 1997, 1999, 2003; among others). The functional literature has focused primarily on certain preposing constructions, including English topicalization, which has been argued to have a contrastive topic marking function (Chafe 1975; Prince 1981, 1997; Ward 1988; Lambrecht 1994; Birner & Ward 1998; among others). One goal of this work is to bring together these distinct strands of research which make compatible claims about contrastive topics. The B-accent in English is a fall-rise intonational contour which Jackendoff 1972 argues contrasts with the A-accent, a focus-marker. One requirement on the use of the contrastive topic-marking B-accent is that the sentence also contain
Chapter 4. The discourse functions of left dislocation
a focus-marked element. In (1) the subject argument is marked as a contrastive topic while the direct object is focused.
(1) a.
What about Fred? What did he eat?
b. What about Susan? What did she eat? (adapted from Jackendoff 1972: 261)
FredB ate the beansA. SusanB ate the eggplantA.
The speaker is pairing eaters with what was eaten. First, Fred is considered with respect to a food item; then the speaker moves on to consider another person, Susan. The situation is described “eater-by-eater”. This example is typical of constructions containing contrastive topic marking. Two (or more) elements intonationally marked with the B-accent are considered with respect to the same open proposition, in this case: a ate x, a = contrastive topic. Certain preposing constructions in English have the same discourse function. Prince 1981 considers topicalization, as in (2). Again, the speaker is considering an event involving food, this time from the perspective of food items. In this case the speaker considers food items and pairs them with people. Contrastively topicalized elements are subscripted as CT and focused ones as F.
(2) a.
What about the beans? Who brought them? The beansCT MaryF brought.
b. What about the eggplant? Who brought it? The eggplantCT SusanF brought.
Again, contrastive topics co-occur with focused elements. Contrastive topics are preposed and focused elements are marked with an intonational rise. Other languages also have contrastive topicalization constructions in which the contrastive topic appears at the left edge of the clause. Szabolcsi 1981, Kiss 1987 and Molnár 1998 propose that elements in the initial position in Hungarian can be interpreted as contrastive topics. The same is shown for Finnish and Catalán in Vallduví and Vilkuna 1998. Hajičová and Sgall 2004 argue that the initial position in Czech, though part of the topic portion of the sentence, can also be interpreted contrastively and is accompanied by an optional intonational contour similar to the B-accent. In the remainder of this section I consider analyses of contrastive topic in greater detail. First, the Prince 1981, 1997 approach to topicalization in English is outlined. My analysis of Czech CLD as a contrastive topic marking constructions builds on the analysis of another contrastive topic marking construction in Czech, contrastive topicalization. The discourse function of this construction is discussed in depth in Hajičová et al. 2003 and Hajičová and Sgall 2004. I then turn to the Büring 2003 formal analysis of the B-accent (what he calls “contrastive topic marking”) which builds on the analysis of A- and B-accents in Roberts 1996 and is reminiscent of the approach to contrastive topic taken in Krifka 1999. Once these analyses are in place I use the analysis developed in Büring 2003 to show that the CLD construction in Czech marks the left dislocated element as a contrastive topic.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
2.1 Prince 1981, 1997 A strength of the Prince 1981, 1997 study of English topicalization is her use of spontaneous textual examples. Such examples force the researcher to consider questions that might not otherwise arise if only constructed examples were used. Common to both the formal semantics and the functional proposals is the requirement that contrastive topic marking co-occur with focus marking. An important theoretical contribution of functional approaches is that the discourse referent of a topicalized element must be a member of a cognitively salient set which has been previously evoked in the discourse.1 This relationship between the contrastive topic and the preceding discourse is an important one and has not, to my knowledge, been incorporated into formal pragmatic accounts of this construction. Consider the example in (3) as illustration. There are three groups of mice in the discourse context; each group is considered in turn with respect to the same open proposition concerning what it will be fed in a feeding experiment.
(3) a.
She has an idea for a project. She’s going to use three groups of mice.
b. Onect she’ll feed mouse chowf, just the regular stuff they make for mice. c. Anotherct she’ll feed veggiesf. d. And the thirdct she’ll feed junk foodf.
(Prince 1997: (14))
The relevant open proposition is shown in (4) for (3b).
(4) a.
Open proposition: She’ll feed one (∈ {the three groups of mice}) X.
b. Instantiation: X = mouse chow
(Prince 1997: (13))
The topicalized elements one, another and the third denote groups (i.e., subsets) of mice. These groups of mice are members of the previously introduced set, three groups of mice. Each sentence contains a focused element which corresponds to the different foods each group will be fed: mouse chow, veggies and junk food. Each alternative group of mice is paired with a distinct focus value (food item). Another example is considered in (5); focus falls on a degree phrase in this example.
(5) a. “My father loves crispy rice,” says Samboon, “so we must have it on the
menu. b. And Mee GrobCT he loves just as muchF.” Mee Grob ($4.95) is a rice noodle… (Prince 1997: (18a))
. Prince 1981, 1997 characterize this set relationship as a “partially-ordered set” (poset). Other researchers have characterized this set relationship as that of a “cognitively salient set” (Hajičová & Sgall 2004, Frey 2005). I assume this characterization of the relevant set relationship and do not consider the question of whether “poset” is the correct formalization.
Chapter 4. The discourse functions of left dislocation
(6) a. Open proposition: He loves Mee Grob (∈ {Thai dishes}) to-X-degree.
b. Instantiation: X = just as much
(Prince 1997: (19))
The relevant set in this example is inferred from the context, the set of Thai dishes, which includes crispy rice and Mee Grob. The open proposition contains a variable over degrees whose value in the topicalized example is denoted by just as much. In this case the focus value does not change between alternative members of the set; the speaker indicates that his father loves both dishes equally. The two requirements of topicalization structures that Prince 1981, 1997 identifies are exhibited in both (3) and (5). The discourse referent of the topicalized element must be a member of a previously evoked or inferred cognitively salient set and the sentence must contain a focused element. The focused element often differs across topicalized alternatives, (3), but that is not a requirement of the construction, see (5). 2.2 Hajičová and Sgall 2004 An important precursor to my analysis of CLD as a contrastive topic marking construction is Hajičová and Sgall’s 2004 approach to Czech contrastive topicalization. Contrastive topicalization in Czech is characterized by a preposed XP in [Spec, IP] that is associated with an optional intonational rise (see chapter 2 on the syntax of this construction).2 This construction differs from CLD in that it has no resumptive element. The analysis of contrastive topicalization presented in Hajičová and Sgall 2004 is compatible with Prince 1981, 1997 for English topicalization. Their analysis is framed within the Topic Focus Articulation (TFA) framework.3 Contrastive topicalization is a special instance of a contrastive element appearing in the topic portion of the sentence (often corresponding to the preverbal domain under TFA); this portion of the sentence is generally reserved for given (i.e., contextually bound) information. The contrastive element has the requirement that it must be in a cognitively salient set relationship with a set which has been previously established in the discourse (see also the “theme alternative set” of Steedman 2002). Hajičová and Sgall consider in detail the types of relationships between the contrastive topic and the previous context that license contrastive topicalization. Among these are: explicitly enumerated sets, implicit sets and part-whole relations. In (7) the set from which the discourse referents of the contrastively topicalized elements are chosen is overtly enumerated in the previous context: the sports teams from Brno and Ostrava.
. For Hajičová and Sgall, contrastive topics are base generated at the left edge of the clause. . Details about this framework are given in Chapter 1.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(7) a.
Včera se hrál zápas mezi Brnem a Ostravou. yesterday refl-cl played match between Brno and Ostrava
b. Domácímct se dařilo ze začátkuf. Hostůmct home.dat refl-cl succeeded at beginning visitors.dat
se povedl až druhý poločasf. refl-cl succeeded in second half
“Yesterday the match between Brno and Ostrava took place. The home teamct was successful at the beginningf. The visitorsct succeeded only in (Hajičová & Sgall 2004: (30–31)) the second halff.”
The contrastively topicalized DP’s (home team and visitors) are compared with respect to the period in the game in which they were winning: the home team in the beginning, the visitors in the second half. The periods in the game are the focused elements in the construction and, as such, appear in the position generally associated with narrow focus in Czech: the sentence final position. The discourse referents of contrastively topicalized elements can also be members of an implicit set in the discourse, see (8). The speaker is comparing the opinions of people regarding the development of a case. Most people were happy about it, but the addressee was not. (8) a.
People were happy about the development of that case, but you have a different opinion.
b.
Přiznal ses, že tebect osobně to mrzíf. admitted.1sg refl-cl/aux.2sg.cl that you.dat personally it bothers “You admitted that youct, personally, were botheredf by it.” (Czech National Corpus, Hajičová et al. 2003: (26))
Again, the contrastive topic appears at the left edge of the (embedded) clause and the focused element at the right. Focus in this case falls on the right-edge embedded finite verb. Another relationship that holds between contrastively topicalized elements is that of part-whole. In (9), the speaker describes renovating their house. (9) a.
Loni jsme renovovali celý dům. last-year aux.1pl.cl renovated whole house
b. Omítkuct jsme natřeli na žlutof, stěchuct jsme plaster aux.1pl.cl painted on yellow roof aux.1pl.cl
vyspravilif… fixed
“Last year we renovated the whole house. The plasterCT we painted yellowF, the roofCT we fixedF…” (Czech National Corpus Hajičová et al. 2003: (22))
Chapter 4. The discourse functions of left dislocation
The contrastive topics, the plaster and the roof, are in a part-whole relationship with the previously evoked, house. Focus falls on the VP in both clauses; painted yellow and fixed are in clause-final positions. The felicity of this example suggests that the relationship that holds between the contrastive topic and the preceding context must be revised. It is not just that the contrastive topic must be a member of a previously evoked set; it can also be in a part-whole relationship with an element in the preceding context. In the following section I present Büring’s 2003 analysis of contrastive topic constructions in English; this analysis is compatible with the intuitions behind Prince’s characterization of English topicalization and that of contrastive topicalization in Czech (Hajičová & Sgall 2004). Büring 2003 provides the foundation for the analysis of Czech CLD developed below. 2.3 Büring 2003 Büring 2003 develops a formal pragmatic analysis of the contrastive topic intonation (the B-accent) in English. His approach to contrastive topic marking is similar in some respects to the functional analysis of preposing constructions presented above. Both strands of research require that contrastive topic marking co-occur with focus marking. What the functional approaches provide is the additional requirement of a link between the contrastive topic and the previous context. Researchers have suggested that the contrastive topic accent indicates that the statement containing the contrastive topic is a partial answer to a (possibly covert) question in the discourse. The question to which (10a–b), repeated from (1), are partial answers is: Who ate what? (10) a.
What about Fred? What did he eat?
b. What about Susan? What did she eat?
FredCT ate the beansF. SusanCT ate the eggplantF.
The utterance in (10a) not only functions as a partial answer to Who ate what?, it also suggests to the hearer that additional questions, such as What did Susan eat?, What did X eat?, are active in the discourse. One such question, What did Susan eat?, is answered overtly in (10b). It is not necessary for additional alternatives to be considered overtly in the discourse. Büring 2003 provides an illustrative example, (11). (11) a.
Where were you at the time of the murder?
b. ICT was at homeF.
(Büring 2003: (22))
The contrastive topic marking on the subject, I, suggests to the hearer that there are others in the discourse context who may have been at different locations at the time of the murder. A speaker might use CT-marking in (11) to suggest that other people might not have such a solid alibi.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
Büring 2003 builds on the partial answer aspect of the use of contrastive topic marking and develops a formal analysis using a notion of Contrastive Topic (CT) congruence. His analysis presupposes a model of discourse in which questions, subquestions and assertions form a hierarchical structure. 2.3.1 Discourse trees Büring represents the hierarchical structure of discourse through discourse trees (d-trees). The discourse in (12) involves a series of questions and answers whose structure is represented by the d-tree in (13). (12)
How was the concert? Was the sound good? How was the audience? How was the band? How was the drummer? And what about the singer? Did they play old songs?
No, it was awful. They were enthusiastic. Just fantastic. Better than ever. Not a single one.
(13)
(Büring 2003: (4))
Discourse
How was the concert?
Was the sound good? How was the audience?
No, it was awful.
question ...... How was the band?
They were enthusiastic. How was the drummer?
subquestion ...... subsubquestion ......
Just fantastic
Each node in the tree is called a Move (cf. Carlson 1983) and each Move consists of sentences expressed as syntactic objects. D-trees are subject to certain general well-formedness conditions, two of which are given in (14). (14) a.
Informativity: Don’t say known things, don’t ask for known things.
b. Relevance: Stick to a question until it is sufficiently resolved. (Büring 2003: (8))
Chapter 4. The discourse functions of left dislocation
Büring implements Informativity using Stalnaker’s 1978 concept of common ground. He follows Roberts 1996 in defining Relevance with respect to a question under discussion (QUD); “for any move M, the QUD is the move M’ immediately dominating it” (Büring 2003: 5). The definition of Relevance used is given in (15). (15) Relevance a. An assertion A is relevant in a d-tree DT iff A is an answer to the QUD for A in DT. b. A question Q is relevant in a d-tree DT iff at least one answer to Q is an answer to the QUD for Q in DT. (Büring 2003: (9))
In order for a Move to be relevant in a given d-tree, it must answer the QUD or, if the Move is a question, an answer dominated by that Move must address the QUD for that question. For example, the subtree in the d-tree in (13), How was the band?, How was the drummer?, Just fantastic satisfies Relevance. The answer, Just fantastic is relevant because it answers its QUD, How was the drummer? That subquestion is relevant because an answer to it ((The drummer was) Just fantastic) addresses the QUD, How was the band? And, finally, the subquestion How was the band? is relevant because an answer to the subsubquestion it dominates addresses its QUD, How was the concert? With this model of discourse in place, I turn to Büring’s approach to contrastive topic marking (CT-marking). 2.3.2 CT congruence Contrastive topic marking does two things. First, it indicates that the utterance is a partial answer to a larger question, the question under discussion. Second, it suggests to the hearer that alternatives to the contrastive topic are under consideration in the discourse. The first conjunct in (16) is a partial answer to the question, Who bought what?, and implicates that alternatives to the contrastive topic, Katie, are under consideration. The second conjunct provides more information about an alternative to Katie, Bill. (16) KatieCT bought the SUVF, but BillCT bought a hybridF.
It is possible to map the utterance in (16) onto a d-tree which provides information about the context in which the utterance is felicitous. The d-tree in (17) is rooted in the question to which (16) is a partial answer: Who bought what? An utterance containing a contrastive topic is dominated by both the question under discussion (QUD) and an intermediate question (a subquestion) concerning the CT-marked element. Sisters to the subquestion are questions concerning alternatives to the contrastively marked element.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(17)
Who bought what?
What did Katie buy?
What did Bill buy?
KatieCT bought the SUVF
BillCT bought a hybridF
CT- and F-marking are shown in the tree in (17), though they are not technically properties of Moves. Under Büring 2003, utterances (which do contain CT- and F-marking) map onto Moves. Büring proposes a CT-congruence condition to address the question of what d-trees license utterances with CT-marking. The first requirement of CTcongruence is that the Move which dominates an utterance with CT-marking be a member of the CT-value of the utterance. Büring draws on Rooth’s 1985 analysis of focus to calculate the CT-value of an utterance such as (16), Katiect bought [the SUV]f. First, the focus value is calculated, (18). (18) Focus value of (16) [[KatieCT bought [the SUV]F]]f = {Katie bought x | x ∈ De} {‘‘Katie bought an SUV’’, “Katie bought a hybrid’’, “Katie bought a station wagon’’}
The focus value of a sentence containing a constituent xF is a set of propositions in which x is replaced by alternatives to it. Note that this focus value corresponds to the ordinary value of the question, What did Katie buy? To form the CT-value, a set of sets of propositions is formed. The focus value is a set of propositions and when the contrastive topic is also replaced by a variable, the result is a set of sets of propositions, (19). (19)
CT-value of (16) [[[Katie]CT bought [the SUV]F]]ct = {{x bought y | y ∈ De | x ∈ De} {{‘‘Katie bought the SUV’’, “Katie bought the hybrid’’, “Katie bought the station wagon’’}, {‘‘Bill bought the SUV’’, “Bill bought the hybrid’, Bill bought the station wagon’’}, {‘‘Fran bought the SUV’’, ‘‘Fran bought the hybrid’’…}…}
The CT-value of (16), this set of sets of propositions, corresponds to the ordinary value of the set of questions: {What did Katie buy?, What did Bill buy?, What did Fran buy? …}. Does the tree in (17) satisfy the first requirement of CT-congruence: the Move dominating an utterance, U, with CT-marking must be a member of the
Chapter 4. The discourse functions of left dislocation
CT-value of U? The left branch of the tree satisfies this requirement; the question, Q, immediately dominating U is a member of the CT-value of U. What did Katie buy? is a member of the CT-value of Katiect bought [the SUV]f. The second requirement is that Q must have at least one sister, Q’, which is also a member of the CT-value of U. This also holds in the tree in (17). The sister to What did Katie buy?, What did Bill buy?, is also a member of the CT-value of U. Büring’s formulation of CT-congruence is given in (20). (20) CT-congruence An utterance U containing a contrastive topic can map onto a move MU within a d-tree D only if U indicates a strategy around MU.
Büring defines “indicate a strategy” in (21). (21) U indicates a strategy around MU in D iff there is a non-singleton set Q’ of questions such that for each Q ∈ Q’ a. Q is identical to or a sister of the question that immediately dominates MU. b. [[Q]]o ∈ [[U]]CT
The tree in (17) satisfies the requirements of CT-congruence: the utterances are immediately dominated by questions which are members of the CT-value of those utterances and the set of questions is a non-singleton set. The subquestions associated with the example in (16) contrast with an utterance in which the intonational contours have been reversed. In (16), the speaker is pairing people with cars, person by person. The opposite would be true if the intonation were reversed. In (22) the same situation is under discussion, but the speaker is pairing cars with people, car by car. The subquestions related to these two utterances illustrate the observation made by Jackendoff 1972 that CT-marking is important in framing the discourse. (22) KatieF bought the SUVCT.
The super question (Who bought what?) is the same for both (16) and (22), but CT marking on the car (rather than on the buyer) structures the discussion differently. In this case, the speaker is presenting the information “car by car’’, rather than “person by person”. The subquestions for (22) are: Who bought the SUV?, Who bought X? Kuno 1982 calls this means of structuring the discourse the “sortal key”. A question to consider is what influences the choice of sortal key. This question, Büring does not address. Returning to the functional research on this topic provides insight into factors influencing the choice of sortal key. Prince 1981, 1997 and Hajičová and Sgall 2004 both note that the discourse referents of contrastive topics are members of a cognitively salient set in the discourse (see also Birner & Ward 1998; Gregory &
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
Michaelis 2001; Frey 2005). I follow these approaches and suggest that the element which is the contrastive topic must be related to the previous context in a particular way. This relationship is characterized in (23) under the term, familiar. (23)
Familiar: The referent or property denoted by the XP stands in a cognitively salient (set) relation or a part-whole relation to an already introduced referent or property.4 (modified from Frey 2005: (40))
I purposefully distinguish this relationship with the previous context from that of topicality. The notion of topicality is used in section 4 to refer to discourse referents which are not only familiar in the sense of (23), but also continue to be discussed in the following context. Though contrastive topics share some features with foci (the construction of an alternative set), they are familiar in that they are related to elements already introduced in the discourse context. The constructed alternative set for CT’s has either already been introduced or evoked in the discourse context or one or more of its members has been evoked. Focused elements have no such relationship to the previous discourse. Since CT’s are linked in a particular way to the larger discourse context in a way that focused elements are not, the speaker can influence the following discourse through his or her choice of which element to mark as the CT; this element is the sortal key. In (22), the speaker chose to mark the car (the SUV), rather than the person buying the car as the contrastive topic. That the DP referring to the car has been marked as a CT suggests to the hearer that the relevant alternative set consists of alternatives to the SUV. The speaker is likely, then, in the following discourse to contrast that type of car with other types of cars, rather than contrasting the car buyer with other car buyers. The alternate intonational pattern, in which the DP referring to the car buyer is CT marked, (16), would involve an alternative choice of sortal key. This would be appropriate if the speaker were discussing a group of car buyers. 2.3.3 Focus on clausal polarity The focus in a contrastive topic construction can fall on any element, including the polarity of the clause. This is the case in the majority of the Czech corpus examples
. In this definition, I distinguish between a cognitively salient set and a part-whole relationship to a previously mentioned referent or property. Prince 1997 and Hajičová and Sgall 2004 consider the part-whole relationship to be another set relationship among elements in the discourse. However, this is not generally the approach taken in the formal semantics literature. The part-whole relation has been discussed in the literature on bridging inferences (Clark & Haviland 1977, among others).
Chapter 4. The discourse functions of left dislocation
to be discussed, so I will show how these fit into the framework developed in the previous section. Consider (24). This utterance would be felicitous in a context in which the speaker is considering certain people with respect to whether or not Mary knows them. (24) Mary knowsF PaulCT, but she doesn’tF know BillCT.
The CT-marking in (24) suggests to the hearer that there are alternatives to Paul in the context who Mary does not know. The focus value of Mary knowsf Paulct is the set of propositions, {Mary knows Paul, Mary doesn’t know Paul}. Note, again, that the focus value of Mary knowsf Paulct corresponds to the ordinary value of the question, Does Mary know Paul? (Hamblin 1973). The CT-value is the set of sets of propositions in which the CT-marked element has been replaced by alternatives to it: {{Mary knows Paul, Mary doesn’t know Paul}, {Mary knows Bill, Mary doesn’t know Bill} …}. The two parts of the utterance in (24) map onto Moves in (25). (25)
Who does Mary know?
Does Mary know Paul?
Does Mary know Bill?
Mary knowsF PaulCT
but she doesn’tF know BillCT
The tree in (25) is licit because the subquestion dominating Mary knowsf Paulct is a member of the CT-value of that utterance. The sister to that subquestion is also a member of this CT-value. 2.3.4 Conventional meanings vs conversational implicature Another aspect of this analysis is the division of labor between the conventional meaning of CT marking and the conversational implicatures associated with it. An observation about CT-marking is that hearers will typically assume, given an utterance such as Katiect bought the SUVf, that the questions raised by this utterance concern people other than Katie and cars other than the SUV. Büring 2003 asks how we get from the conventional meaning of contrastive topics, What did X buy?, to this implication. He concludes that this is a conversational implicature. If the speaker knew that another person, besides Katie, bought an SUV, it would be simpler to say, “Katie and Matt bought SUV’s”. A naturally occurring example from the television drama Six Feet Under illustrates the defeasibility of this i mplicature, (26).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(26) OneCT’s in a nursing homeF, and the otherCT, I think, is also in a nursing homeF. (Six Feet Under (HBO): season 3 episode 2)
The use of the CT-marking in (26) suggests to the hearer that alternatives to the contrastively accented element are associated with different focus values.5 However, this is a conversational implicature associated with CT-marking, not part of the conventional meaning, and can, therefore, be defeated. Another implicature associated with CT-marking is that alternatives to the CT-marked element are available in the discourse context. Under the Büring 2003 proposal this is also a conversational implicature. While I agree that it is an implicature that the focal values associated with alternatives to the contrastive topic differ, I maintain that the inference that alternatives to the contrastively marked element are available is not a conversational implicature, but should be part of the conventional meaning of CT-marking. This is not the case in Büring 2003. Büring’s definition of CT-congruence, (20), requires that the question dominating the assertion containing CT-marking be a member of a non-singleton set. However, there is no requirement that members of this set differ in their meaning. Büring defines these questions not as semantic objects, but as syntactic objects. Two utterances of the same question could satisfy this requirement as there could be two distinct syntactic objects in the set representing the same semantic object. If CT-marking merely conversationally implicated the existence of alternatives to the CT-marked element in the discourse context, the implicature would be defeasible. But, in fact, it is very difficult, I suggest impossible, to defeat this implicature. (27) #GodCT created the world in sevenF days.
(Büring p.c.)
The utterance in (27) is odd in a context in which there are no alternatives to God, but it improves dramatically if the context is enriched, see (28). (28) Maybe your godCT created the world in sevenF days, my godCT could (Büring p.c.) have done it in fiveF.
As I understand Büring 2003, there is no explanation for the infelicity of the intonational pattern associated with (27). However, if it were part of the conventional meaning of CT marking that there are alternatives to the CT marked element in the discourse, the infelicity of (27) would be explained. As shown in (26), an utterance with CT-marking does not commit the speaker to a change in focus value changes between alternatives. There is a commitment,
. Krifka 1999 suggests a connection between the use of the additive particle, also, and contrastive topic marking; this appears to hold in this example.
Chapter 4. The discourse functions of left dislocation
however, to the existence of alternatives to the CT-marked element in the discourse context. It would be possible to capture this commitment if a strategy consisted of a non-singleton set of questions which were semantic, rather than syntactic, objects. If this were the case, it would not be possible for there to be two questions with the same semantic interpretation in a strategy. Keeping in mind other restrictions, such as Informativity and Relevance, this change requires that an alternative to the CT be considered in a strategy in order for it to be well-formed and thus, captures the intuition that the existence of alternatives to the CT is part of the conventional meaning of CT-marking. 3. CLD as contrastive topic marking In this section, I provide evidence that the discourse function of CLD constructions in Czech is that of contrastive topic marking. CLD constructions exhibit all the features associated with contrastive topics: a focused element occurs obligatorily within the utterance and the discourse referent of the CLD’ed element is contrasted with alternatives in the discourse context. Twenty-three spontaneous corpus examples provide solid evidence that this is the correct interpretation of the construction; these examples are supplemented by constructed examples from field research. 3.1 Overview of the discourse function of CLD That Czech CLD constructions mark contrastive topics is supported by native speaker intuitions. Offered a CLD construction in elicitation contexts, speakers insist on continuations in which the CLD’ed element is compared to alternatives with respect to an open proposition. In addition to insisting on contrastive continuations, when presented with a contrastive context (of the type discussed in Büring 2003), speakers overwhelmingly prefer to use a CLD and not an HTLD construction; this supports the hypothesis that CLD’ed, but not HTLD’ed, elements are contrastive topics. Consider the elicited example in (29). Accusative case matching between the preposed element and the resumptive pronoun indicates that this is CLD. (29) Tu taškuct tu si koupila Hanaf ale tu peněženkuct, that bag.acc that.acc refl-cl bought Hana but that wallet.acc tu si koupila Janaf. that.acc refl-cl bought Jana “That bagCT, HanaF bought it and that walletCT, JanaF bought it.”
The speaker is discussing buying events and is considering them purchase by purchase. First that bag is considered and paired with a buyer, then that wallet.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
The contrastive topics, that bag and that wallet, occur in the CLD positions of their respective clauses. Each of the focused subjects, Hana and Jana, occurs at the right edge of its clause, a position linked with narrow focus, and is associated with an intonational rise. The utterance in (29) maps onto Moves in the d-tree in (30). The CLD’ed construction in the first conjunct indicates that the question, Who bought that bag? is active in the discourse and that related questions, such as: Who bought that wallet?, are as well. Both questions are overtly addressed in (29). (30)
Who bought what?
Who bought that bag?
Who bought that wallet?
HanaF bought that bagCT
JanaF bought that walletCT
According to Büring’s CT-congruence condition, these utterances licitly map onto these Moves. Who bought that bag? is a member of the CT-value of Hanaf bought that bagct. (31) CT-value of (30) [[[Hana]F bought [that bag]CT]]ct = {{x bought y | y ∈ De | x ∈ De } {{‘‘Hana bought that bag’’, “Jana bought that bag, “Jitka bought that bag.”}, {‘‘Hana bought that wallet’’, “Jana bought that wallet.’’, Jitka bought that wallet.’’}…}
Additionally, the set containing these two questions (Who bought that bag? and Who bought that wallet?) make up a strategy in the sense of (20). They are members of a non-singleton set of questions whose members dominate the Moves, Hana bought that bag and Jana bought that wallet. The context in which this Czech CLD example occurs is very similar to the contexts in which the English B-accent is found. The speaker is considering answers to a multiple wh-question and one element, in this case the purchase, serves as the CT-marked sortal key. CT-marking in this case takes the form of a CLD construction. In the remainder of this section, I consider additional examples of CLD which support my contention that CLD is a CT-marking construction in Czech. The majority of these data are spontaneous textual examples identified from the Czech National Corpus and through Google searches. In (32) a set of two sisters (representing two historically significant railway lines) has been previously introduced into the discourse. The speaker considers
Chapter 4. The discourse functions of left dislocation
these two discourse entities in turn with respect to distinct predications. Again, instrumental case matching between the left dislocated element and the resumptive pronoun indicate that it is a CLD construction.6 (32) a.
Sudičky totiž oběma přisoudily stejně: stát se fairies namely both awarded same become.inf refl-cl
ovliňovatelkami stovek lidských osudů a výrazně působit influencers hundreds people fate and strongly affect.inf
na tvárnost svého okolí. Ne, není tu řeč o on appearance self ’s surroundings no isn’t here talk about
konkrátních osobách. concrete people
b. Tou prvníct, z dvojice známou a populární, That first.instr of two well-known and popular
tou je železniční trat’ ž Pardubic do LiberceF. … that.instr is railway route from Pardubice to Liberec
c.
Skromným sourozencemct, prožívajícím svůj bohatý život modest sibling living self ’s rich life
bez okázalé publicity, je pak trat’ ž Jaroměře do SvatoňovicF. without showy publicity is then route from Jaroměř to. Svatoňovice
a.
“The fairies awarded the same amount of good fortune to both sisters: to become influencers of hundreds of people’s fates and strongly affect the appearance of their surroundings. No, we are not talking about specific people’.’
b. The first oneCT of the two, well known and popular, that is a railway route from Pardubice to LiberecF.… c.
The modest siblingCT, living its rich life without showy publicity, is a route from Jaroměř to SvatoňoviceF.” (Czech National Corpus)
In this example, the discourse referent of the CLD’ed element, that first (sister), is a member of the previously evoked set of sisters, overtly referred to in the discourse by the phrase, both. In (32b), the first sister appears in a CLD construction and is paired with the focused predication, a railway route from Pardubice to Liberec. In (32c), the second sister is paired with a distinct predication, a route from Jaroměř . In copular constructions, the subject DP can appear in the instrumental or the nominative case. For a discussion of the distinction between these two possibilities see Veselovská 2003.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
to Svatoňovice. The context in which this CLD construction appears is reminiscent of those found in the previous examples of contrastive topics. Distinct alternatives (the two sisters) are considered with respect to wh-questions with the same focused constituent, in this case: What is the first sister?, What is the second sister? The CLD construction in (32b) is a partial answer to the question under discussion: What are the two sisters? The statement in (32c) concerning the second sister also provides a partial answer; the conjunction of these two statements provides a complete answer to the question under discussion. This example provides evidence for the hypothesis that the choice of sortal key is related to discourse familiarity. The set of two sisters has been introduced in the discourse whereas the distinct railway lines they are associated with have not. The element which is linked to the prior discourse, in this case, the first (sister) and the modest sister, serve as sortal keys. They are, in turn, related to two railway lines, new information in the discourse context. The first sister is referred to by a CLD’ed DP (i.e., a construction involving left dislocation), the second by a contrastively topicalized DP (no left dislocation). Both of these constructions have the same discourse function, marking contrastive topics, and often appear together. In most corpus examples in which more than one member of the contextually salient set is overtly considered and a CLD construction introduces a member of the set, the CLD construction appears first, followed by a contrastive topicalization construction. Though a thorough analysis of the discoursal differences between CLD and contrastive topicalization is outside the scope of this book, I hypothesize that they have different distributions which are related to the activation status of the contrastive topic. I suggest that CLD constructions are used to refer to elements with a lower activation status. Lambrecht 1994, among others, has reached similar conclusions for left dislocation constructions; they are often used to raise the activation state of a discourse referent. In other words, the use of CLD appears to establish the connection between the contextually salient set and a member of that set. Once that connection is established, contrastive topicalization (a simpler construction) is used to refer to additional members of that set. In (33) focus falls on the polarity of the clause. Case matching shows that this is a CLD construction; both the CLD’ ed element and the resumptive pronoun are in the accusative. The example is from an online chat room and the context is a posted photograph. Speaker A writes (33a) and Speaker B responds with (33b). (33) a.
Jinak kdo zná toho kluka, že se aspoň zasměje… alternatively who knows that guy c refl-cl at-least will- smile
b. Toho klukact??? Toho neznámf, ale Prochoract v něm poznávámF. that guy.acc that.acc neg-know but Prochor in it recognize
Chapter 4. The discourse functions of left dislocation
a. “Anyway, whoever knows that guy will at least laugh…
b. That guyCT, I don’tF know him, but ProchorCT I recognizeF from the picture.” (http://www.dfklub.cz/gallery/opinion.php?id = 10917)
Again, the first contrastive topic is CLD’ed and the second is contrastively topicalized. Focused elements, the matrix verbs, know/recognize, occur at the right edge in both clauses. As in previous examples, the sortal key, members of the set of elements in the photograph, is linked to the discourse context. As shown in chapter 3, XP’s of any category can be CLD’ed. The expectation, then, is that these elements would also be associated with a contrastive topic interpretation. Consider an example with a preposed AP, (34). (34) “…milostiváCT, to jsemF. Ale pardon! ochlasta a blázenCT nejsemF.” gracious that.neut am. but pardon drunk and lunatic neg-am “GraciousCT, that I amF. But, excuse me! A drunk and a lunaticCT I am notF.” (Czech National Corpus)
The speaker contrasts properties that she has with properties that she does not have. The property gracious appears in a CLD construction and is paired with the positive polarity of the copula; the properties evoked by the nominals drunk and lunatic, on the other hand, are paired with the negative polarity of the copula. Focal intonation in this case is associated with the copula, which appears at the right edge in both clauses. The speaker is providing partial answers to the question under discussion: What am I? The two clauses in (34) address the subquestions, Am I gracious? and Am I a drunk and a lunatic? The same results are found when a VP is fronted. Consider (35), an elicited example. (35) Hrát kuželkyCT, to umíF moc dobře, ale hrát fotbalCT, to neF. play.inf bowling that.neut know very well but play.inf soccer that.neut no “BowlingCT, she knowsF that well, but soccerCT, she doesn’tF.”
The fronted VP in this case functions as a contrastive topic. The VP to bowl is compared with the VP to play soccer. These contrastive topics are paired with different polarities of the matrix VP: know well, not know well. The speaker is addressing the question under discussion, How well do you play sports? As with other categories, a CLD’ed PP indicates that alternative questions are active in the discourse; consider the elicited example in (36).7 . CLD’ed PP’s are also attested in online corpora, see (i). (i) K tragédii. K té došlo třeba až za sto let. towards tragedy towards that got-past maybe even after 100 years “That tragedy, we won’t get past it even after 100 years.” (Czech National Corpus)
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(36) V hezkých lesíchct, v těch jsme se procházeliF, in pretty forests.loc in that.loc aux.1pl.cl refl-cl walked ale v ošklivýchct, v těch neF. but in ugly.loc in that.loc no “In pretty forestsCT, we went walkingF in some, but in ugly onesCT, we didn’tF.”
The speaker asserts in the first conjunct that they had been walking in pretty forests. The use of CLD suggests to the hearer that alternative locations are under consideration. This implication is overtly satisfied in the discourse. The second conjunct addresses another location, in ugly forests. Both alternative PP’s appear in CLD constructions and are associated with opposing focus values. In pretty forests is paired with positive polarity, indicated by focal intonation on the matrix verb; in ugly forests is paired with negative polarity, indicated by a negative particle. The same results are found when AdvP’s are preposed. In the elicited (37), the comparison is between manner adverbials. The speaker compares the manner in which the door and the window are closed. (37)
Potichuct, tak zavřel dveřef, ale hlučněct, tak zavřel oknof. quietly thus closed doors but loudly thus closed window “QuietlyCT, he closed the doorF like that, but loudlyCT, he closed the windowF like that.”
The manner adverbials function as contrastive topics and the focused elements within the clauses, door and window, occur at the right edge. In sum, a wide range of phrasal categories instantiate the left dislocated element in Czech CLD: DP, AP, PP, VP, AdvP. Regardless of category, CLD is always associated with a contrastive topic interpretation. 3.2 Conversational implicature vs. conventional meaning In the previous examples, the focus value changed across alternatives. Due to the strong conversational implicature associated with contrastive topic constructions, we would expect this to be true in most cases. In the majority of the corpus examples (and all elicited examples) the alternatives are associated with focused elements that have different values (including values for clausal polarity). However, there are several examples in which the focus values paired with alternatives to the contrastive topic are the same. One such example is given in (38). Note that in this example, the contrastive topic in the first clause is identified through contrastive topicalization and the second through a CLD construction. We know that the first conjunct involves contrastive topicalization because a strong pronoun, nám (“us.dat”) appears in [Spec, IP]. The polarity of the focused VP’s is positive in both clauses.
Chapter 4. The discourse functions of left dislocation
(38) a.
“Tak námct nalejte dva stokyf a [támhle tomu pánovi]CT, so us pour two drinks and over-there that man.dat
tomu nalejtef taky.” that.dat pour also
b. “Samořejmě, samořejmě…” slabikoval hostinský… of-course of-course said bartender a.
“OK, pourF usCT a drink and [that man over there]CT, pourF him drink, too.”
b. “Of course, of course, said the bartender…” (Czech national Corpus)
The maxim of manner suggests that if you knew that both us and that man were paired with the same focus value, it would be simpler to say: “Pour a drink for us and for that man over there”. However, the maxim of manner is a conversational implicature, hence defeasible. Recall that it was also violated in the naturally occurring English example given in (26). That examples of this type exist and that they are rare is predicted by Büring 2003. In section 2.3.4, I posited that it is part of the conventional meaning of CT-marking, not a conversational implicature, that there are alternatives to the contrastive topic in the discourse context in English and suggested a modification to the Büring 2003 analysis. This appears to hold for Czech as well. In the examples considered above, alternative members of the set evoked by the contrastive topic are overt in the discourse. There are, however, examples in which neither the set of which the contrastive topic is a member nor overt alternatives are mentioned. Those examples share a common characteristic, however. The contrastive topic is modified by an element which suggests the existence of (unmentioned) alternatives: na přiklad (“for example”), třeba (“for instance”). Consider example (39). (39) a.
Též já mám své představy jak pomoci… also I have self ’s ideas how help.inf
b. Na přiklad JindruCT, toho všech strastí ušetřímf. on example Jindra.acc that.acc all woes will-save.1sg a.
“I also have my ideas about how to help…”
b. “For example, JindraCT, I will saveF him from his woes.” (Czech National Corpus)
CT-marking in (39) indicates that other questions are active in the discourse which concern alternatives to the discourse referent of the CLD’ed DP, Jindra. Answers to such question are covert, however, as in the example in (11).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
I have found no corpus examples in which there are neither overt alternatives to the contrastive topic nor a modifier of the type in (39). Interestingly, in cases in which an element such as for example precedes the left dislocated XP, speakers strongly prefer to use a CLD construction when given a choice between CLD and HTLD, supporting the idea that for example suggests the existence of alternatives to the contrastive topic. Examples of this type follow from the contention that the existence of alternatives to the contrastive topic is part of the conventional meaning of the contrastive topic construction and not a conversational implicature. 3.3 Conclusion In this section, I conclude that Czech CLD is, in fact, a CT-marking construction and is compatible with the analysis of contrastive topic found in Büring 2003. Two contributions have been made to the understanding of contrastive topic presented in Büring 2003. First, contrastive topics serve as sortal keys and structure the discourse that follows. Second, the implicature that there are alternatives in the discourse to the CT is shown to be part of the conventional meaning of an utterance containing CT-marking, not a conversational implicature. Data from English and Czech support this. Native speaker judgments support these findings as well: when presented with a CLD construction native speakers require a context in which questions are raised about alternatives to the contrastive topic. There is also a correlation between a contrastive topic context and a preference for a CLD construction. When speakers are given the choice between using a CLD or an HTLD construction in a particular context, they overwhelmingly prefer a CLD construction when a contrastive topic interpretation is available in the discourse. With this established, I turn to HTLD constructions.
4. The discourse function of HTLD In this section, I argue that the HTLD construction is not associated with a contrastive topic interpretation, but is, rather, a topic promotion device. To show this, I follow Gregory and Michaelis’s 2001 analysis of HTLD constructions in English and utilize their algorithm for identifying sentence topics (see also Reinhart 1981; Davison 1984; Gundel 1985; Geluykens 1992). The contexts of eighteen spontaneous corpus examples support this claim. The main difference between CLD and HTLD constructions is that in HTLD constructions, there is no evidence that any alternative questions are being raised. In some cases this is due to the fact that there are no other discourse referents under
Chapter 4. The discourse functions of left dislocation
discussion. In (40), the discourse referent of the hanging topic has been previously mentioned and is the only referent under discussion by the participants. We know that this is HTLD and not CLD because case matching is absent: the hanging topic is nominative, the resumptive accusative. (40) a.
“Jonatáne, a ted’ vezmeš do huby hadr a vyleštíš Jonatán and now bring to mouth rag and shine
nám us
hezky auto, zadarmo tě živit nebudem,” a well car for-free you support.inf neg-will.1pl and
Šebestová šeptala za plotem Machovi, človeče, to je Šebestová whispered behind fence Mach.da man it is
děsný, oni si z něho udělali úplnýho otroka, horrible they refl-cl from him made totally slave
b. chudinka malej, toho čeká pěknej život. poor-thing small.nom that.acc waits nice life c.
a Mach vytáhl sluchátko a řekl, Šebestová, víš co, and Mach took receiver and said Šebestová know what
radši z něho uděláme třeba vrabce nebo sýkorku, better from him make maybe sparrow or chickadee
aspoň bude volnej jako pták… at-least would-be free like bird
a.
“Jonatán, and now bring a rag in your mouth and shine our car, we aren’t going to support you for free.” And Šebestová whispered behind the fence to Mach, “hey, it is horrible how they are making him a slave,”
b. poor little thing, a nice life was awaiting him.” c.
And Mach pulled out the telephone receiver and said, “It would be better for us to make him a sparrow or a chickadee, at least then he would be as free as a bird…” (Czech National Corpus)
Two individuals, Mach and Šebestová, are discussing a third, Jonatán, with respect to the proposition: a nice life was awaiting him. There is no alternative to Jonatán that has a different life situation (for instance: ‘‘lucky dog, a nice life is still awaiting him”). In section 2, a definition of familiarity was presented, (23). The discourse referents of CLD’ed elements are familiar under that definition; this is also true of the discourse referents of HTLD’ed elements. These elements are members of a cognitively salient set which has been previously introduced into the discourse. In (40) this is the singleton set containing Jonatán. Discourse referents of HTLD’ed
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
e lements, however, can be members of non-singleton sets without raising alternative questions in the discourse. The hanging topic in (41) is a member of the set of party guests. Note the lack of case matching between the hanging topic (nominative) and the resumptive element (accusative). (41) a.
Mechanicky stiskla ruku Krulišovi a manželce, docentu mechanically shook hand Kruliš and wife, docent
Melužinovi a manželce…Aaronu Cohenovi, tomu zastupci Melužin and wife Aaron Cohen.dat that deputy
Metro- Goldwyn- Mayer, který vážně seděl se snoubenkou u baru… Metro- Goldwyn- Mayer who seriously sat with fiance at bar
b. Jarouši, ten lokaj v předsíni, toho já Jarouši.voc that lackey.nom in lobby that.acc I
vodněkud znám… from-somewhere know
c.
Ba ne, tendle ksift je mi hrozně povědomej. Mám even no that face is me.dat horribly familiar have.1sg
dojem, že můj bracha mu jednou dal přes dršku. impression that my brother him.dat once gave across mug
a.
“Mechanically, she shook hands with Kruliš and his wife, with Docent Melužin and his wife, … , with Aaron Cohen, that deputy from Metro-Goldwyn-Mayer who sat seriously with his fiancé at the bar,…”
b. “Jarouš, that lackey in the lobby, I know him from somewhere…. c.
“No, that face is really familiar to me. I have the feeling my brother punched him in the face once.” (Czech National Corpus)
In this example the speaker is considering party guests in turn. That lackey in the lobby is considered with respect to the open proposition, “I know x from somewhere’’; there are no alternatives in the discourse context which are considered with respect to the same proposition. Instead, (different) information is provided about different party guests: who is sitting at the bar, who is a deputy from Metro-Goldwyn-Mayer, etc. 4.1 Overview of the requirements of “topic promotion’’ The previous two examples suggest that since alternative questions are not generally raised by HTLD constructions, these constructions have a discourse function other
Chapter 4. The discourse functions of left dislocation
than marking a contrastive topic. I provide evidence that this construction can be characterized as that of topic promotion by following the criteria that Gregory and Michaelis 2001 appeal to in identifying topic promotion in English hanging topic constructions. First, it is necessary to show that the discourse referents of hanging topics are sentence topics. Gregory and Michaelis consider both the preceding and the following contexts of left dislocation constructions. They note that topical elements have generally been evoked in the previous discourse either by prior mention or by membership in a previously mentioned set. This follows the definition of familiar in (23). Topical elements also tend to perseverate in the following context. Once these two properties have been established it is also necessary to show that the discourse referents of hanging topics are not already established sentence topics – otherwise HTLD would not be “topic promoting”. I use the notion of backward looking center from Centering Theory to probe whether the discourse referents of hanging topics are non-topical in the context preceding the HTLD construction (for Centering Theory see Grosz, Joshi & Weinstein 1983; Grosz, Joshi & Sidner 1986; Grosz, Joshi & Sidner 1995; among others). 4.2 The preceding context Researchers have pointed out that the role of topic is distinct from the discourse status of the referent (given/familiar). Evoked elements are not necessarily topical and topics need not be discourse-old. Lambrecht and Michaelis 1998 and Prince 1992 provide evidence that discourse-new entities can function as topics if they are inferable from the preceding context (see also Birner & Ward 1998). One way of being inferable from the preceding context is to be a member of a previously evoked set; this corresponds to the notion of familiarity in (23). Gregory and Michaelis 2001 characterize the extent to which an entity is given by rating the “anaphoricity” of tokens (the degree to which a referent has a discourse antecedent) on a scale of 0–2. A score of “0” means that there is no discourse antecedent, a score of “1” means that the discourse referent is a member of a previously evoked set, and “2” means that the discourse referent has been previously mentioned. In all of the Czech HTLD examples in my corpus, the discourse referent of the hanging topic has been either previously mentioned or is a member of a previously evoked set (scores of “1” or “2”). Consider again (41). In this case the discourse referent of the hanging topic, ten lokaj v předsini (“the lackey in the lobby”), has not been mentioned in the previous context, but it is a member of an evoked set: the set of party guests. This discourse referent receives a rating of “1”. In (40), the discourse referent of the hanging topic, Jonatán, has been previously mentioned in the discourse, and receives a rating of “2’’.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
Example (42) below is similar. Again, there is only one entity under discussion: Mr. Kopyto, the discourse referent of the hanging topic, (42b). The hanging topic appears in the nominative case, while the resumptive is in the accusative. (42) a.
A tu paní Štěrnová zvolala: “Pan Kopyto!” A and suddenly Mrs. Sternova called Mr. Kopyto and
ostatní se přidali. Pan Kopyto, pan Kopyto, kdo others refl-cl joined-in Mr. Kopyto Mr. Kopyto who
jiný!”… Otto seděl a mlčel. Pomyslil si, co je to else Otto was-sitting and quiet thought refl-cl what is it
za divné jméno a kdo to vůbec je. from strange name and who it at-all is
b. “Pan Kopyto, toho oni neznají. Mr. Kopyto.nom that.acc you neg-know c.
To je jeden znamý od Backrů,” řekla babička. “Je to it is one acquaintance from Backra said grandma is it
nějaký baaadatel [sic] překládá z jazyka německého some scholar translated from language German
…Germanist a sběratel… Nosí velký brejle a má hluboký Germanist and antiquarian wears large glasses and has deep
hlas. Prý měl krční operace…” voice they-say had neck operation
a.
“And all of a sudden Mrs. Sternova called, “Mr. Kopyto!” And the others joined in. “Mr. Kopyto, Mr. Kopyto. Who else!” … Otto was sitting and stayed quiet. He was thinking, what a strange name and who is that.
b. “Mr. Kopyto, you don’t know him. c. “He is an old acquaintance from Backra,” said grandma. “He is some sort of scholar; he translates German…Germanist and antiquarian… He wears large glasses and has a deep voice. They say he had a neck operation.” ” (Czech National Corpus)
The discourse referent of Mr. Kopyto, (42b), has been mentioned in the previous context, and receives a score of “2’’. For a topic promotion account such as Gregory and Michaelis 2001, cases in which the discourse referent of the hanging topic has been previously mentioned could be problematic. If that discourse referent is already topical, then HTLD would not be topic-promoting. Gregory and Michaelis distinguish between
Chapter 4. The discourse functions of left dislocation
iscourse referents with no antecedent (those with a “0” rating) and those with d an antecedent (a rating of “1” or “2”). There is, however, a way to tease apart previously mentioned entities (those with a rating of “2”) from established topics under Centering Theory. The Centering Theory notion of backward looking center has been equated with that of continuing topic (see Beaver 2004, among others). Backward looking centers tend to be the discourse referents of DP’s which have been in the syntactic subject position in one or more of the past two clauses. In (40), the discourse referent of the hanging topic, Jonatán, was mentioned once in the previous two clauses; the DP associated with it was the object of a preposition in the immediately preceding clause. The same is true of the discourse referent of the hanging topic in (42), Mr. Kopyto. This referent was also mentioned once in the preceding two clauses; in the immediately preceding clause the DP associated with it was an object clitic. Under Centering Theory, neither Jonatán nor Mr. Kopyto are the backward looking centers of the clauses preceding the left dislocation c onstructions (and, thus, are not topics). Summarizing, the discourse referents of hanging topics in Czech have either been overtly mentioned in the preceding context, (40) and (42), or are members of a previously evoked set, (41). This generalization, however, is also true of the discourse referents of CLD’ed elements. There are two important differences, though. A key difference already discussed is that alternative questions are raised by the use of a CLD construction, but not by HTLD constructions. Another difference is outlined in the following section: perseveration. 4.3 Perseveration in the discourse Once it has been established that an entity has been mentioned in the previous context (is familiar), but is not yet topical, it is necessary to consider whether the discourse referent of the hanging topic is topical in the following context. If the discourse function of HTLD is to promote a discourse referent to topicality, this referent should be topical in the context following the HTLD construction. One measure of topicality that will be useful in determining if this is the case is perseveration in the discourse. To determine whether a discourse referent perseverates in the discourse, Gregory and Michaelis follow the quantitative measures introduced in Givón 1983. Givón defines topic persistence as: “the number of times the referent persists as an argument in the subsequent ten clauses following the current clause” (Givón 1983: 908). Gregory and Michaelis considered only the five subsequent clauses due to the nature of the spoken examples they considered, but found that considering ten clauses provided no more insight into the discourse status of the entity under consideration. I follow them in considering the five following clauses.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
CLD and HTLD constructions differ dramatically in the perseverance of the discourse referent of the left dislocated element. The discourse referents of hanging topics are much more likely to perseverate in the following discourse than those of CLD’ed elements. In twelve of the eighteen HTLD examples under consideration, the discourse referent persists in two or more clauses (67% of cases). Perseveration in the discourse is much less common with respect to CLD constructions: in only three of the twenty-three examples does the discourse referent of the CLD’ed element perseverate in two or more of the following clauses (13% of cases). Consider the passages discussed above. In (40b), there are two references to Jonatán in the three clauses following the HTLD construction; those mentions are underlined. In (41c), the discourse referent of the hanging topic, the lackey in the lobby, is mentioned overtly in two of the following clauses. Example (42c) is even more striking; Mr. Kopyto, is mentioned pronominally in all of the following five clauses. The hanging topic construction promotes Mr. Kopyto to topic status and sets the stage for a discussion of details about his life: he is an old acquaintance, he is a Germanist, he wears large glasses, etc. The fact that the discourse referents of the hanging topics in (40–42) persist in the discourse suggest that these elements are, indeed, topics. What can be said about the contexts following CLD examples? Though the discourse referents of both CLD’ed and HTLD’ed elements must be familiar, there is a striking contrast between the contexts which follow these two constructions. Unlike HTLD, the discourse referents of CLD’ed elements are not generally mentioned in the subsequent discourse. In example (43), the speaker considers two individuals in turn with respect to the question, “Who do I know?’’, but additional information is not provided about either one. This example contrasts robustly with the HTLD examples considered above. (43) a.
Neznám ho, a přece bych řekl, že už jsem neg-know him and exactly would say c already aux.1sg.cl
ho někde him somewhere
viděl… saw
b. Ale tu dívku … tu znám určitě… but that girl.acc that.acc know for-sure c.
ta tvář je mi velice povědomá … když jsem přišel that face is me.dat very familiar when aux.1sg.cl came
blíž, málem jsem se rozesmál. closer a-little aux.1sg.cl refl-cl started-to-laugh
a.
“I don’t know him, but I would say that I have seen him somewhere…”
b. But that girl … I know her for sure… c. that face is very familiar to me … when I came a little bit closer I almost laughed.” (http://www.cswu.cz/jediland/ffcz/sunny/pochopitlez11.htm)
Chapter 4. The discourse functions of left dislocation
The discourse referent of the CLD’ed element, that girl, does not perseverate in the discourse; her face is mentioned in the following clause, but no further information is provided about her. Most examples of CLD exhibit a similar pattern: various alternatives are considered in turn and the discourse does not focus on any one in particular. In (32), the discourse concerns two sisters (railway lines) and addresses the question: what is she? The discourse, however, focuses on neither sister. The same is found in (38). The discourse referent of the CLD’ed element, that man over there, is not mentioned in any of the following clauses. After the CLD construction, the discourse returns to a discussion of the larger bar scene. Though the discourse referents of both CLD’ed and HTLD’ed elements are members of a previously evoked set, only HTLD constructions promote a discourse referent to topic status. I conclude that the HTLD construction is a topic promotion device. Hanging topics have been evoked (either overtly or as members of a previously evoked set) in the preceding discourse, but are, as yet, non-topical. Appearing in the left edge hanging topic position promotes them to sentence topic status. Their status as sentence topic is confirmed by the fact that they perseverate in the following discourse. CLD’ed elements, on the other hand, exhibit a contrastive topic discourse function. The discourse referents of CLD’ed elements do not perseverate in the discourse, but, are, instead, contrasted with other members of a set of alternatives with respect to an open proposition. 4.4 Considering HTLD II Data collected from elicitation suggests that HTLD II, like HTLD I, is a topic promoting construction. This construction is infelicitous in contrastive contexts. In (44), the speaker is contrasting two books with respect to their lengths, and HTLD II is infelicitous. (44) #Tu krátkou knížku, dočetla jsem ji včera, ale tu that short book.acc read aux.1sg.cl it.cl yesterday but that dlouhou, dočtu ji přiští týden. long.acc read her.cl next week “The short book, I read it yesterday, but the long one, I will read it next week.”
However, HTLD II is felicitous in a context in which the speaker provides additional information about the discourse referent of the HTLD II’ed element in the following sentence, see (45). (45) Tu krátkou knížku, dočetla jsem ji včera. Byla moc zajimavá. that short book.acc read aux.1sg.cl it.cl yesterday was very interesting “The short book, I read it yesterday. It was very interesting.”
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
What appears to be the case in (45) is that the discourse referent of the HTLD’ed DP has been promoted to topic status by the use of this construction. Evidence of this comes from the fact that this element is referred to by a null pronoun in the following clause. Though a thorough analysis of corpus data has not been conducted, these data suggest that HTLD II patterns with HTLD I, and contrasts with CLD, with respect to discourse function. 4.5 Conclusion There are some discoursal similarities between CLD and HTLD constructions. In both the discourse referent of the left dislocated XP is a member of a cognitively salient set which has been previously mentioned (i.e., both are familiar according to the definition in (23)). However, in key ways, these two constructions differ. First, CLD, but not HTLD, indicates that alternative questions are active in the discourse: CLD marks contrastive topics. Second, the discourse referents of hanging topics persist in the discourse, while those of CLD’ed elements do not. This, in conjunction with the fact that the discourse referents of hanging topics have nontopical antecedents, suggests that HTLD functions as a topic promotion device. The comparison between these two constructions points to a difference between various notions of topicality. As Gregory and Michaelis 2001 discuss, topicality is two-fold: a connection to the previous and following contexts. Both constructions exhibit the first notion of topicality (familiarity), that of previous mention. Only HTLD exhibits the second, that of perseverance in the discourse. The differing behaviors of these two constructions with respect to topicality suggests that a more nuanced understanding is needed. It is possible for a construction to be sensitive to one component of topicality, but not the other.
5. Conclusion Two theoretical contributions were made in this chapter. First, a modification of Büring 2003 was suggested which captures the intuition that the existence of alternatives to the contrastive topic is a part of the conventional meaning of a contrastive topic construction, rather than a conversational implicature. Second, I imported from the functional literature the requirement that contrastive topics must be familiar in the discourse, in the sense provided in (23). This notion of familiarity is relevant to both CLD and HTLD. The discourse referent of the LD’ed element in both these constructions must be familiar. However, only HTLD promotes this familiar discourse referent to topic status.
Chapter 4. The discourse functions of left dislocation
CLD and HTLD are members of a larger typology of preposing constructions. The evidence presented in this chapter suggests a connection between preposing (movement) constructions and a contrastive topic interpretation. The construction involving movement (CLD) exhibits a contrastive topic interpretation, but the base generated construction, HTLD, does not. This pattern is not particular to Czech. Gregory and Michaelis 2001 show that English preposing constructions differ significantly in their discourse functions. English HTLD (non-movement) is similar to Czech HTLD in that it is a topic promotion construction. Topicalization (movement), on the other hand, is not topic-promoting. Prince 1981, 1997 argues that it is, instead, a contrastive topic construction. The same is true for Bulgarian and German. Arnaudova 2004 analyzes two Bulgarian preposing constructions, Clitic Left Dislocation Constructions (CLLD) and topicalization. She finds that the movement construction, topicalization, has a contrastive topic interpretation along the lines of Büring 2003, while CLLD (non-movement) has a topic/comment structure. The same appears to be true of German. Although Frey 2004a, 2005 is cast in a different framework, his analysis of German CLD8 and HTLD suggests that CLD constructions in German have a “contrastive flavor” along the lines of contrastive topic, while HTLD does not. Under his analysis, HTLD constructions serve to introduce new discourse topics; this is compatible with what I have established for HTLD in Czech. A question to consider is why this pairing between movement and contrastive topic and non-movement and topic/comment might hold across languages.
. Frey uses the term, German Left Dislocation (GLD), rather than Contrastive Left Dislocation, but I follow Grohmann 2003 and use CLD.
chapter 5
The syntax of resumption 1. Introduction Resumption in left dislocation has been a puzzle in the generative literature since it was first considered in Ross 1967 and van Riemsdijk and Zwarts 1974/1997. The fact that two instances of an overt element referring to the same discourse referent appear in the construction (the left dislocate in [Spec, TopP] and the resumptive in [Spec, IP] for Czech CLD) has resisted a straight-forward analysis. I take a novel approach to spelling out the resumptive element at the prosodic-syntactic interface. Vat 1981/1997 is an important precursor to the movement approach advocated here. Their account is based on Vergnaud’s 1974 analysis of relative clauses. Both overt exponents of the left dislocate (the resumptive and the left dislocate) are base generated as sisters within the VP. The full XP raises to a specifier position outside the IP domain and the resumptive element raises to the highest specifier position within the IP. While my analysis does not involve the base generation of two copies of the left dislocate, it does involve the creation of multiple copies through movement. As discussed in chapter 3, in Czech CLD, the left dislocate moves from its VPinternal position, to a position in the left periphery, leaving behind a copy in [Spec, IP]. I assume that the resumptive element is an overt instantiation of this copy. Thus, the question under consideration here is the mechanism which ensures spell out of both the initial and intermediate copies of the left dislocate. To guarantee that both copies have an overt exponent at PF, I consider not only the syntactic characteristics of CLD, but also its prosodic realization. Following other research at the prosodic-syntactic interface (Zubizaretta 1998; Franks 1998; Bošković 2001; Stjepanović 2003; Landau 2005b, among others), I pursue an analysis in which it is the prosodic rise associated with the resumptive which motivates spell out at PF. 2. CLD: Locating the CT interpretation in [Spec, IP] In the previous chapter it was established that CLD, like contrastive topicalization, has a contrastive topic (CT) interpretation. The difference between these two
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
constructions, then, is the appearance of the left dislocated constituent in CLD constructions. The question under consideration in this section is the locus of the CT interpretation in CLD. In contrastive topicalization constructions, the CT interpretation of the fronted XP clearly arises due to its appearance in [Spec, IP], as there is only one copy of this element in the left periphery. However, in CLD constructions it is less apparent which structural position, [Spec, TopP] or [Spec, IP], contributes the CT interpretation. Considering the intonational contours associated with the XP’s in these two positions sheds light on this question because XP’s with a CT interpretation in Czech are known to bear an intonational rise (Veselá et al. 2003). Through a prosodic study, I found that it is the resumptive element and not the left dislocate which carries the prosodic rise in this construction. Assuming that this contour falls on the element with the CT interpretation, it follows that it is [Spec, IP], and not [Spec, TopP], which is associated with that interpretation. 2.1 Prosodic realization of contrastive topic Veselá et al. 2003 analyze the pitch accent associated with contrastively topicalized elements in Czech as a rising pitch contour.1 To demonstrate this, Veselá et al. use an annotated corpus of spoken Czech which was compiled from recordings of three television talk show interviews. From these recordings, the researchers identified 111 tokens of utterances containing elements with a CT interpretation (contrastive topicalization constructions). They isolated the CT portion of these utterances and measured the F0 value across that constituent. Most often this constituent (or sector) consisted of a single phonological word (either a lexeme or a lexeme and a preclitic preposition); in more complex cases a focalizer, such as only, even, or also, was included in the sector. When measuring the change in the F0 value across these CT sectors, the researchers found that the value rose by an average of 19Hz.2 This pitch rise contrasted with the change in F0 values across other types of sectors. For instance,
. In English, this contour is termed the B-accent or rise-fall accent (for the earliest discussion, see Jackendoff 1972). . Vesela et al. 2003 divide CT’s into three sub-types, “kontrast 1–3”. Kontrast 1 includes examples in which several members of a set of alternatives are enumerated in analogous positions across several clauses (rise of about 18Hz). In the second group, kontrast 2, the CT is a member of a set of alternatives which has been established in the discourse (rise of about 19Hz). The third sub-group consists of examples in which the CT is in a part/whole relationship with an overtly established element in the discourse (rise of about 20Hz). They conclude that there is not a significant difference between the rise in these three cases. For me, all three types of “kontrast” instantiate the contrastive topic discourse function (see chapter 4).
Chapter 5. The syntax of resumption
previously established topics showed a very small rise in pitch (approximately +2Hz) and a falling pitch accent of about ‒14Hz was associated with new information at the end of the utterance (what is considered the “focus” portion of the sentence within the Topic Focus Articulation framework). In a phonetic experiment, I found that the resumptive demonstrative in CLD constructions is associated with an average prosodic rise of 22Hz; this rise is comparable to that of contrastively topicalized constituents as found in Veselá et al. 2003.3 The prosodic contour of the CLD’ed element, on the other hand, is generally falling. This suggests that the prosody of the resumptive pronoun, rather than the left dislocate, contributes the CT interpretation. Prosodic data were collected from two native Czech speakers in a phonetic experiment.4 The speakers were presented with 26 sentence tokens in written form, given time to review them and asked to read the tokens in sequence (without pauses). The tokens consisted of HTLD I and CLD tokens interspersed with filler examples.5 All filler and HTLD I tokens were naturally occurring examples and 7 out of the 15 CLD tokens were as well; the remaining 8 CLD tokens were constructed in conjunction with a native speaker. Each CLD and HTLD I example appeared twice in the experiment. Since expressions containing CLD and HTLD’ed elements are highly sensitive to context, the appropriate context was created by introducing these tokens with one or more sentences. Certain phonetic precautions were taken when constructing the example sentences. In order to get the clearest results, I limited the tokens to sentences with demonstrative resumptives containing several sonorant sounds: “tomu”, [tomu] (“that.masc.dat”); “tím”, [ci:m]6 (“that.masc.instr”); and “tam”, [tam] (“there. dir”). Additionally, I disregarded sharp rises in intonation that could be associated with the initial [t] and [c] in the demonstratives, as well as other voiceless obstruents. Intonational information for the HTLD I and CLD examples was extracted using Praat (Boersma & Weenink 1999–2006). As in Veselá et al. 2003, I measured
. I thank Jaye Padgett for his help in designing the experiment. All errors are my own. . The speakers are from Moravia and a northwest suburb of Prague (Kladno), and, as the time of the experiment, were recent immigrants to the Santa Cruz, CA area. One speaks fluent English; the other has limited English skills and, at the time of the experiment, lived in a Czech-speaking household. . Both HTLD I and CLD have left edge resumptive demonstratives. The third left dislocation construction in Czech, HTLD II, is compared with CLD in the following section. . The symbol, [c], represents a voiceless palatal stop.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
the F0 contour across the demonstrative taking measurements at the beginning and end of the sector and calculating the difference.7 I first consider the intonational contour associated with CLD constructions. An illustrative example of CLD is given in (1). The question in (1a) provides the context for the CLD construction in (1b).8 (1) a.
Komu pomůžeš? who.dat help.2sg “Who are you helping?”
b. Jindrovi, tomu ze všech problémů pomůžu… Jindra.dat that.dat from all problems help.1sg “Jindra, I help him with all his problems…”
(2) Pitch track for (1b)
Pitch (Hz)
500
Jindrovi 0
0
tomu ze všech problémů pomůžu [yndrov tomu ze všex proble:mu: pomu:žu] Time (s)
2.81027
. In some cases, the rise associated with the demonstrative appears to “leak” onto the following syllable. In those cases I considered that syllable to be the end of the sector. Researchers have suggested that a rising prosodic contour associated with one element often appears on the following element. . In chapter 4 I showed that CLD constructions are felicitous as answers to wh-questions, particularly when the answer contains more than one possible correspondent to the whexpression. The token sentence provided in (1b) was followed by a sentence which presented an alternative to Jindrovi.
Chapter 5. The syntax of resumption
This example illustrates a typical token. The demonstrative resumptive, tomu (“that.masc.dat”), is associated with a significant rise in pitch (41Hz), while the left dislocated constituent is linked with a small rise in pitch (9Hz). As noted in Veselá et al., the pitch falls at the end of the clause. When all 15 tokens were considered, the intonational contour associated with the demonstrative resumptive in CLD rose by an average of 22Hz. This rise is comparable to the rise of about 19Hz across contrastively topicalized elements found in Veselá et al. The left dislocate, on the other hand, was not associated with a very large prosodic rise. Due to these prosodic facts, I conclude that the locus of the CT interpretation of CLD constructions is the demonstrative pronoun in [Spec, IP] and not the left dislocate. Given that CLD, but not HTLD constructions, are associated with a CT inter pretation, the expectation is that the resumptive in HTLD I is not associated with a prosodic rise. This prediction is borne out. The prosodic characteristics of the intonational contour across the demonstrative resumptive in HTLD I constructions differ significantly. Across these resumptive demonstratives, the pitch rose an average of 2Hz when 8 tokens were analyzed. The pitch track in (4) for the example in (3) illustrates these findings. (3) a.
“Jindra, Jindra, kdo jiný?” Otto seděl a mlčel. Jindra Jindra who else Otto sat and was-quiet “Jindra, Jindra, who else?” Otto sat and was silent.
b. “Jindra, tomu by nevolal.” Jindra that.dat cond.3sg.cl neg-call “Jindra, I would never call him.” (4) Pitch track for (3b) 500
Pitch (Hz)
Jindra 0 0.110998
tomu by [yndra tomu b nevolal] Time (s)
nevolal. 1.39158
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
In this example the pitch across the demonstrative falls slightly (4Hz), though, on average, pitch rises slightly across the demonstrative. What is absent is the marked pitch rise associated with the resumptive in CLD constructions. This is the expected result because HTLD is not associated with a contrastive topic interpretation. In fact, the resumptive element in HTLD bears a pitch contour (a small rise in pitch) comparable to that found for topical elements by Veselá et al. These intonational contours provide additional evidence for the discourse functions outlined in chapter 4: HTLD I promotes a discourse referent to topic and CLD exhibits a CT discourse function. 2.2 Comparing CLD and HTLD II HTLD II constructions differ from HTLD I in that the resumptive element is a second position clitic. Given that the CT interpretation in CLD and contrastive topicalization constructions correlates with an intonational rise on the constituent in [Spec, IP], we would expect a CT interpretation to be unavailable in HTLD II constructions. This is what is found. Compare the examples in (5) and (6) (repeated from (44–45) in chapter 4). The CLD example in (5) contains a strong resumptive element in [Spec, IP] and requires a contrastive interpretation. (5) Tu krátkou knížku, tu jsem dočetla včera, ale that short book.acc that.acc aux.1sg.cl read yesterday but tu dlouhou, tu dočtu přiští týden. that long.acc that.acc read next week “The short book, I read it yesterday, but the long one, I will read it next week.”
When a clitic, rather than a full pronoun, resumes the left dislocated element, as in HTLD II, a contrastive interpretation is not available. Speakers agree that the continuation in (6) is infelicitous. (6) #Tu krátkou knížku, dočetla jsem ji včera, ale tu that short book.acc read aux.1sg.cl it.cl yesterday but that dlouhou, dočtu ji přiští týden. long.acc read it.cl next week “The short book, I read it yesterday, but the long one, I will read it next week.”
The preference, when using HTLD II, is to continue discussing the discourse referent of the left dislocate, rather than to consider another discourse referent, see (7), repeated from (45) (chapter 4). (7) Tu krátkou knížku, dočetla jsem ji včera. Byla moc zajimává. that short book.acc read aux.1sg.cl it.cl yesterday was very interesting “The short book, I read it yesterday. It was very interesting.”
Chapter 5. The syntax of resumption
Like HTLD I, HTLD II promotes a discourse referent to topic status. Parallel facts are found in questions. In Slavic languages, a very common way to form a wh-question is for a left dislocated XP to precede a clause which contains a resumptive element. The CLD example in (8) is felicitous in a context in which multiple items are paired with recipients. Consultants agree that (8) would most likely be followed by a question of the type: “And your gerbil, who did you give it to?”9 (8) Toho svého psa, komu toho jsi dala? that self ’s dog.acc who.dat that.acc aux.2sg.cl gave “That dog, who did you give him to?”
An alternative realization of the question in (8) is one in which a clitic resumes the left dislocate, (9). (9) Toho psa, komu jsi ho dala? that dog.acc who.dat aux.2sg.cl it.cl gave “That dog, who did you give it to?”
This example does not evoke a contrastive topic interpretation. Speakers prefer that the answer is followed by more information about the dog, not another question about alternatives to the dog. The pairs of declaratives and interrogatives in (5–6) and (8–9) are minimal pairs, differing in the nature of the resumptive element. The correlation between the appearance of a demonstrative resumptive in [Spec, IP] and a CT interpretation provides further evidence that it is the element in [Spec, IP] and not the left dislocate that carries the CT interpretation in left dislocation. 2.3 Interpretation of [Spec, TopP] If the element in [Spec, IP] contributes the contrastive topic interpretation, what motivates the additional movement of the left dislocated element to [Spec, TopP]? To answer this question it is instructive to compare the relative distributions of CLD and contrastive topicalization. Each construction can appear alone (i.e., not followed by another CT-marking construction), but when more than one discourse referent is explicitly associated with a CT interpretation in adjacent clauses, it is common to find either a sequence of contrastive topicalization constructions or CLD followed by contrastive topicalization. What is not common in corpus searches, however, and what is considered degraded by consultants, is to have two
. The contrastive topic interpretation of the CLD’ed element in this question suggests that contrastive topics are felicitous in questions, pace Büring 2003. He suggests that contrastive topics are not found in questions in German and English.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
consecutive CLD constructions, see chart in (10). Examples (32), (33), and (34) in chapter 4 illustrate the pattern in (10b). (10) Patterns of Co-occurrence: CLD, Contrastive Topicalization
Common Pattern
a. Contrastive Topicalization > Contrastive Topic. b. CLD > Contrastive Topicalization
Uncommon Pattern
c. CLD > CLD d. Contrastive Topicalization > CLD
I propose that this pattern of attested and degraded combinations stems from the additional discourse function associated with the left dislocate in CLD constructions. As shown in chapter 4, the discourse referent of the CLD’ed element is necessarily a member of a set which has been previously evoked in the discourse. This is captured in the definition of familiar in (23) (chapter 4). Lambrecht 1994, among others, has also suggested that left dislocation constructions are used to raise the activation status of a discourse referent. If the use of CLD served to raise the activation status of the cognitively salient set of which the discourse referent was a member as well as the activation status of the discourse referent itself, the patterns illustrated in (10) could be understood (see also Prince 1997 for cognitively salient sets and LD). Once this cognitively salient set has been activated (through the use of a CLD construction), raising the activation status of additional members of that set is unnecessary and contrastive topicalization can be used instead of CLD. Consider the textual example in (11), repeated from (33) (chapter 4). (11) Toho klukact??? Toho neznámf, ale Prochoract v něm poznávámf. that guy.acc that.acc neg-know but Prochor in it recognize “That guyct, I don’tf know him, but Prochorct I recognizef from the picture.” (http://www.dfklub.cz/gallery/opinion.php?id=10917)
The discourse context of this online chat-room example is a posted photograph, and the left dislocated DP, that guy, evokes the set of individuals pictured in the photograph. Once this set has been evoked, it is possible to refer to additional members of that set (Prochor, for example) in a simpler construction, contrastive topicalization. More corpus research on the distribution of CLD and contrastive topicalization constructions is necessary, but this preliminary study of the data suggests that CLD constructions, when followed by a construction with a CT interpretation, are followed by contrastive topicalization, rather than a CLD construction. Associating this pattern with the fact that LD constructions raise
Chapter 5. The syntax of resumption
the activation status of their discourse referent (and the set to which this discourse referent belongs) is promising.
3. The syntax of CLD The previous section established that the CT discourse function of CLD is associated with the resumptive element that appears in [Spec, IP], not the dislocate in [Spec, TopP]. This allows a unified syntactic analysis of the CT discourse function associated with both contrastive topicalization and with CLD. In both constructions, syntactic features linked with this discourse function appear exclusively on I0. Let us consider first the simpler construction, contrastive topicalization. In chapter 2, I provided evidence that contrastive topicalization involves operator movement on par with wh-movement. Movement to the left edge is motivated by a combination of operator features and CT features on the head of the CT’ed XP and on I0, as well as an EPP feature on I0. The tree in (13) illustrates the analysis of (12). (12) [Tu červenou tašku]ct si koupila Janaf. that red bag.acc refl-cl bought Jana “[That red bag]CT JanaF bought.” (13)
IP [DP tu červenou tašku][CT/ct,u] (that red bag)
I′
I [CT,u/ct] [EPP]
vP
[DP tu červenou tašku][CT/ct,u]
vP
...[DP tu červenou tašku][CT/ct,u/]...
First, the CT’ed DP, tu červenou tašku (“that red bag.acc”), scrambles to an adjoined position in the vP domain. It then moves to [Spec, IP], a move which is motivated by the matrix of interpretable and uninterpretable features on the probe and the goal. Lower copies delete. In CLD constructions, an additional functional projection, TopP, dominates the sentential domain, and the LD’ed XP undergoes leftward movement to the specifier of this projection. The appearance of an XP in [Spec, TopP]
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
evokes the contextually salient set of which the discourse referent of the CLD’ed element is a member. I associate this discourse function with a T feature on Top0. Consider (14). (14) Tu červenou tašku, tu si koupila Jana. that red bag.acc that.acc refl-cl bought Jana “That red bag, Jana bought it.”
The DP, tu červenou tašku (“that red bag”), has a familiarity feature (T), as well as the feature matrix [CT/ct, u]. Movement of the DP to [Spec, TopP] satisfies the EPP and values the uninterpretable T feature on Top0. The derivation is illustrated in the tree in (15). (15)
TopP [DP tu červenou tašku][CT/T] Top′ (that red bag) TopP [T,u][EPP]
IP
[DP tu červenou tašku][CT/ct,u/] I′ I [CT,u/ct] [EPP]
vP
...[DP tu červenou tašku][CT/ct,u/]...
The CLD’ed DP undergoes movement to [Spec, IP] to value uninterpretable features and satisfy the EPP on I0, and then, this same DP undergoes a subsequent movement to [Spec, TopP]. It is crucial that the same XP undergo movement to both [Spec, IP] and [Spec, TopP]. This result follows from economy conditions: the element in [Spec, IP] is necessarily the closest element with the appropriate featural make-up. The only DP which could intervene between [Spec, TopP] and [Spec, IP] is a wh-phrase in [Spec, CP]. We know independently, however, that wh-phrases are not topical and, therefore, are not associated with a T feature which could check the uninterpretable feature on Top0.10
. I do not consider the possibility of Merging an element from the numeration into [Spec, TopP] position for two reasons. First, it was established in chapter 3 that this construction is derived by movement and, second, economy conditions would not necessarily prefer first Merge over Move. Recently, researchers have claimed that Merge is not less costly than Move (see Adger and Ramchand 2005, among others).
Chapter 5. The syntax of resumption
The movement analysis of CLD developed here is preferable to the one proposed in Grohmann 2003. For Grohmann, both leftward movements are motivated by the same discourse feature (F) which he associates with a topic interpretation. From a theoretical perspective, positing two stacked functional projections with the same featural make-up raises the question why a language would have a CLD construction if the same discourse features could be satisfied by a simpler construction (topicalization). The analysis presented here has more explanatory power than that of Grohmann 2003 in that the two leftward movements are motivated by distinct features which are associated with distinct discourse functions. The prosody of the construction provides additional evidence for these discourse functions and related features. In (15), copies of the LD’ed DP appear in the correct positions. However, the expectation is that the top copy would spell out and the lower copies would delete. Nunes 1999 discusses the general assumption that it is the top copy that spells out in movement chains. He suggests that this preference can be derived from economy. On the top copy, uninterpretable features have deleted through the movement operations that the XP has undergone. Under the assumption that, at PF, uninterpretable features must undergo deletion before pronunciation is possible, it is more economical to pronounce a higher copy than a lower one since checking relationships with functional heads have permitted the deletion of uninterpretable features in the narrow syntax (see Franks 1998; Fox & Pesetsky 2004 for other explanations).11 I do not adopt a particular explanation for default spell out of the top copy, but, as this is the general assumption in the literature, I assume that the top copy in CLD spells out fully. The problem remains that in CLD constructions, not only does the top copy spell out, but the intermediate copy spells out as a resumptive element. I turn to the PF interface for an explanation of this additional spell out and link the appearance of the resumptive to the prosodic contour associated with it. 4. Resumption at the syntax-phonology interface Copies of the left dislocated element appear in four positions in the clause: the base position, adjoined to vP, [Spec, IP] and [Spec, TopP]. Deletion of the two lowest copies proceeds as expected, but both the top copy and the copy in [Spec, IP]
. This economy argument goes through, of course, only if we assume both that checking relationships delete uninterpretable features in the narrow syntax and that checking relationships are established through movement and not through static Agree.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
are pronounced at PF. Pronunciation of the intermediate copy in [Spec, IP] is unexpected. Recent work within the Minimalist Program has linked pronunciation of non-initial copies to PF requirements. Under Minimalism, movement leaves copies in multiple syntactic positions. This leaves open the possibility that, rather than pronouncing simply the highest copy, a lower copy or even multiple copies could be pronounced. To account for unexpected patterns of pronunciation in movement chains, researchers have disassociated movement operations from the procedure that determines which copy in a chain should be pronounced. Narrow syntax is responsible for the movement of constituents, but determining which copy to pronounce occurs at PF (see Pesetsky 1998; Franks 1998; Bošković 2001; Bobaljik 2002; Stjepanović 2003; Landau 2005b; Sato 2005; among others). In support of this approach, Landau 2005b argues that it is desirable to limit the domain of narrow syntax and, where possible, shift functions to the interfaces, LF and PF. This is particularly desirable for processes such as pronounciation, which are independently needed at PF. In Landau’s 2005b analysis of VP fronting in Hebrew, verbal elements undergo movement in the narrow syntax, but it is PF requirements, unrelated to movement operations, that motivate the spell out of two copies in the movement chain. 4.1 V(P) fronting in Hebrew The example in (16) illustrates VP topicalization in Hebrew. Two copies of the verb appear in the clause: the fronted copy is in the infinitival form and is associated with an intonational rise; the lower copy is inflected for tense and person. (16) lirkod, hu rakad dance.inf he dance.pst “As for dancing, he danced.”
(Landau 2005b: (57a))
Hebrew V(P) topicalization has two syntactic instantiations: bare-infinitive fronting (BI) and phrasal-infinitive (PI) fronting. When the verb has an internal argument, it can either front along with the verb or remain in situ. Landau analyzes the former as PI and argues that the vP fronts to [Spec, TopP]. In cases in which the internal argument of the verb remains low, Landau suggests that only the verbal head fronts (BI). Since the verb in (16) has no internal argument, this example has two possible syntactic analyses: PI and BI. Though both analyses are, in principle, possible, Landau (p.c.) suggests that locality would prefer a PI analysis in which it is the full vP that fronts. The vP properly contains the verbal head and is, thus, closer to the probe (Top0). I adopt this analysis of (16) and provide an illustration in the tree in (17).
Chapter 5. The syntax of resumption
(17)
TopP Top′
vP3 DP2 hu (he)
v′
Top
lirkod (dance.)
IP DP2 hu (he)
I′ √V+v+I rakad (danced)
vP3 DP2 hu (he)
v′ √V+v rkd
√V
The vP, consisting of a copy of the subject DP and the verb root/v0 complex, fronts to [Spec, TopP].12 Two other movements also take place. First, the verbal head raises to I0 to get tense information. Second, the subject DP, hu (“he”), raises to [Spec, IP] to satisfy the EPP feature on that head. Landau 2005a analyzes the EPP as a PF selectional feature. Due to that fact, the subject DP is required to spell out in [Spec, IP]. Landau 2005b implements spell out of the two verbal copies as follows. He argues that copies spell out overtly at PF when they are associated with phonological content. In the example in (16), phonological content comes from the structural positions in which the two copies are found. There are two chains to consider. The first chain is a head movement chain between the base position of the verb and I0. Also, there is a chain between the base position of the vP and its copy in [Spec, TopP]. There are requirements that the verbal head in the highest position of each chain be pronounced.13 V(P) fronting requires that a verbal head associated with an intonational rise appear
. Due to the morphology of Hebrew, Landau analyzes the internal structure of the verbal head as a verbal complex consisting of the root verb and v0 (see Landau 2005b for a detailed analysis). . If a long head movement analysis were implemented, the same results would hold at PF. Rather than two chains, there would be one head movement chain with copies of the verbal head in I0 and [Spec, TopP] (assuming that heads can fill specifier positions). The two highest copies in the head movement chain would be required to spell out at PF due to the phonological requirements associated Top0 and I0.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
at the left edge of the fronted constituent. The lower copy is also in a position associated with a phonological requirement. This copy of the verb adjoins to T0 and must be pronounced in order to host tense morphology; tense requires an overt exponent on a verbal head. The lowest copy of the verb (that in the base position within the VP) deletes because it is not in a structural position associated with a PF requirement. In essence, if a member of a chain is in a position associated with a phonological requirement, that chain link must be visible at PF (spell out overtly), or the derivation crashes. Landau hypothesizes that there are two ways in which an element can be associated with phonetic content. The first is a positional requirement. A head can require that the X0 that is adjoined to it, (18a), or that the XP in its specifier position, (18b), Spell out overtly. (18) a. [HP [H’ X1 [H]] […Xn]] b. [HP XP1 [H’ H]] […XPn]]
(Landau 2005b: (51))
The second possibility is for the element itself to be associated with phonetic content. If an element does not move, it has phonetic content (Landau, p.c.). Landau 2005b defines “associated with phonetic content” in (19). What is relevant for the present analysis is (19b). (19) X is associated with phonetic content iff: a. X has phonetic content, or b. X is in a position specified with some phonological requirement. (Landau 2005b: (52))
Consider again the example in (16). The two highest movement copies are in structural positions associated with phonetic content. The first is in a structural position associated with a phonological rise and the second in a position in which tense must be spelled out on an overt head. To implement the spell out of elements which are “associated with phonetic content”, Landau posits a principle of PF recoverability, see (20). (20) P-Recoverability In a chain < X1, … Xi, … Xn > where Xi is associated with phonetic (Landau 2005b: (53)) content, Xi must be pronounced.
P-recoverability, along with the definition in (19), accounts for the spell out of both copies in (16). The top copy in the vP chain spells out due to an obligatory intonational rise associated with this fronting construction, and the top copy in the head movement chain spells out to express tense and agreement, which must be realized overtly at PF. It is not the case, however, that all copies must be
Chapter 5. The syntax of resumption
ronounced. An economy condition ensures that copies not associated with phop nological requirements are deleted, (21). (21) Economy of Pronunciation Delete all chain copies at PF up to P-recoverability.
(Landau 2005b: (60))
This condition requires the deletion of the lowest copy of the verb in (16). It is not in a position associated with phonetic content and, thus, it deletes. This analysis correctly predicts multiple overt occurrences of the verb in V(P) fronting in Hebrew. In the next section we turn to CLD in Czech. 4.2 PF requirements on spell out in Czech CLD In Czech CLD, as in V(P) fronting in Hebrew, two copies of a movement chain are realized overtly at PF. Following Landau 2005b, this must be due to phonological requirements associated with the positions in which these two elements are found. Consider first the position of the resumptive. We established in §2.1 that the rising pitch accent associated with the CT discourse function of CLD obligatorily falls on a particular structural position, [Spec, IP]. In order for this pitch accent to be realized at PF, an overt copy of the contrastive topic must appear in this position. Pronunciation of the resumptive satisfies this prosodic requirement. The highest position of the movement chain, the position of the left dislocate, however, is not associated with a rising prosodic contour and there does not seem to be any other phonological requirement associated with this position. Why, then, does the top movement copy spell out? We consider this question in the following section. 4.2.1 Pronouncing the top copy Where pronunciation of the top copy is found, but is not motivated by any apparent phonological requirement on the position in which it occurs, Landau relies on a phonological interpretation of the EPP. For him, the EPP is a PF selectional requirement of a head that requires spell out of an XP in its specifier position (see Landau 2005a). Consider the example in (22). (22) Bill seems to study hard.
The DP, Bill, is assigned nominative case under static Agree with the inflectional head, I0. An EPP feature on I0 requires the movement of the DP Bill to its specifier position. Since the EPP is a PF selectional requirement, spell out of the copy of Bill in [Spec, IP] is required at PF. This interpretation of the EPP and its interaction with P-recoverability predicts, then, that the top copy in CLD constructions must be pronounced since movement of the left dislocate to [Spec, TopP] is motivated by the EPP on Top0, see (23).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(23) Petra, toho asi neznám. Petr.acc that.acc maybe neg-know “Petr, I don’t think I know him.”
However, there are problems inherent to interpreting the EPP as a PF selectional requirement. In cases in which the EPP motivates movement to an intermediate specifier position, overt pronunciation of the specifier of that head is usually not required. Consider wh-movement out of an embedded clause in English, (24). (24) Where did Mary say where Sheila bought the sweater where?
It is generally assumed that the EPP motivates movement of the wh-phrase to the embedded [Spec, CP]. But, the moved element is not pronounced in that position, as would be expected under a PF analysis of the EPP.14 Facts of this type call into question the feasibility of interpreting the EPP as a PF selectional feature. Landau 2005b makes use of this phonological EPP feature in order to avoid a default strategy such as “Pronounce highest” (Landau, p.c.). It is not clear, however, that it is possible to avoid a principle of this type. For instance, when verbal arguments are scrambled to adjoined positions, such as vP adjunction in the Czech middlefield, there do not seem to be any phonological requirements on the adjunction site which would require the top copy of the moved element to be pronounced. Since it is not clear that a default principle such as “Pronounce highest” can be avoided in the grammar and since an EPP feature interpreted as a PF selectional requirement is problematic, I do not adopt Landau’s comprehensive proposal that movement copies are pronounced only when a head in their local environment requires it. Instead, I assume that “Pronounce highest” is a default requirement on pronunciation in movement chains. Additional movement copies are pronounced when there is a phonetic requirement on the structural position the copy occupies. 4.2.2 Pronouncing the resumptive Spell out of the intermediate copy in CLD falls out from Landau’s 2005b analysis. As in Hebrew V(P) fronting, more than one position of the movement chain of the left dislocate is associated with phonetic content. The highest position must be pronounced due to the “Pronounce highest”. The intermediate position is associated with an intonational rise which requires an overt exponent. I illustrate this
. Landau 2005a suggests that the PF features associated with the wh-XP can satisfy the PF requirements of the embedded C0 before it moves on to [Spec, CP] of the matrix clause.
Chapter 5. The syntax of resumption
phonological requirement in the schematic in (25). There is a structural phonological requirement on the specifier of I0 when it is associated with a ct operator feature. (25)
IP XPCT
I′ I [ct]
Thus, the XPCT is “associated with phonetic content” via its structural position and must be pronounced due to the P-Recoverability principle. The tree in (27) illustrates the proposal for (26). (26) Ale tu dívku … tu znám určitě… but that girl.acc that.acc know.1sg for-sure “But that girl, I know her for sure…”
If no overt element appeared in [Spec, IP], the derivation would crash at PF due to the fact that the obligatory prosodic contour associated with I0ct would not be realized. Under “Economy of Pronunciation”, lower copies delete as they are not associated with any phonological requirement. (27)
TopP Top′
[DP tu dívku] (that girl) Top [T][EPP]
IP
[DP tu dívku]
I′
I [CT,u/ct][EPP]
vP
[DP tu dívku] DP
vP v′ …[DP tu dívku]…
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
There is a final issue that needs resolution. The overt intermediate copy does not surface maximally, but, instead, is phonologically reduced. I assume that this is related to the Economy of Pronunciation principle in (21). The only PF requirement on [Spec, IP] is that the prosodic contour associated with I0ct be realized. This contour can be realized with minimal spell out.
5. Spelling out the resumptive pronoun Due to P-Recoverability, (20), the intermediate movement copy in [Spec, IP] must have overt phonetic content. However, due to the economy condition in (21), spell out need only be adequate to satisfy P-Recoverability. For V(P) fronting in Hebrew, both the higher and the lower copies of the verbal element include the verb root. On the lower copy, there is additional tense morphology, while the higher copy is in the (simpler) infinitival form. In Czech CLD, it is the lower copy that has the simpler form. The motivation for spell out of this copy, like the top copy in V(P) fronting, is prosodic. Realization of the prosody associated with [Spec, IP] does not require spell out of the full movement copy. Satisfaction of both the economy condition and P-recoverability can be achieved by spelling out a resumptive element, rather than a full copy of the left dislocate. I consider how this is implemented in the next section. 5.1 Resumption as partial spell out of movement copies Various researchers (including Engdahl 1985; Koopman 1982, 1984) have suggested that resumptive pronouns alternate with movement traces and are, in some sense, overt versions of movement traces. With the revival of the Copy and Delete theory of movement, the connection between resumptive elements and copies of movement can be expressed theoretically. Since movement of an XP leaves behind copies (usually deleted at PF), spelling out these copies (formerly known as traces) under some circumstances becomes theoretically viable. McCloskey 2006 considers the procedure by which this spell out would proceed. Under Chomsky 1993, full deletion of lower copies takes place at PF. However, it is possible that deletion, under some circumstances, is partial, rather than complete. McCloskey makes the suggestion that, since resumptive elements are pronominal, and pronouns are considered to be D-heads, spell out of a resumptive element consists of deletion of the NP complement to a D0 and spell out of D0 as a pronominal element. Let us consider how this process would proceed in the example in (28). The highest copy, tu zajímavou knížku (“that interesting book.acc”), spells out fully in [Spec, TopP]. An overt exponent of the copy in [Spec, IP] is required to carry the
Chapter 5. The syntax of resumption
intonational contour associated with the construction. Partial spell out can satisfy this requirement. (28) Tu zajímavou knížku, tu zajímavou knížku that interesting book.acc that.acc interesting book dočetla Jana. finished-reading Jana “The interesting book, Jana finished reading it.”
The complement to D0 elides due to the economy condition in (21) and D0 spells out overtly as a demonstrative determiner.15 There is a problem, however, with this type of syntactic approach when we consider dislocated conjoined DP’s, see (29).16 Note that the element resuming the conjoined DP structure does not correspond to the syntactic heads of the two conjuncts but is, rather, a plural, third-person pronoun. (29) Toho svého psa a tu svou kočku, ty/ that self ’s dog.acc and that self ’s cat.acc them.acc/ *[toho a tu] mám ráda, ale ostatní zviřata asi ne. that and that.acc have.1sg joy but other animals maybe no. “My dog and my cat, I love them, but I don’t really like other animals.”
This pattern of resumption does not follow from the syntactic approach outlined above. Resumption, in this case, appears to reflect the syntactic features of the conjoined structure (third-person, plural), rather than the syntactic structure of each conjunct, suggesting that a purely syntactic approach to resumption is untenable. I suggest, instead, that the resumptive element is an overt representation of the ф-features associated with the DP in [Spec, IP]. In the case of (29), the relevant features are third-person, plural. For the simpler case in (28), the resumptive element spells out the third-person, singular features of the left dislocate.17 . Later in this section I will consider the fact that it is a demonstrative that spells out overtly as the resumptive. . I thank Sandra Chung for pointing this out to me. . One remaining issue for a movement approach to CLD is that epithets can appear in [Spec, IP]. (i)
Ale Karla, toho pitomce nikdo neviděl. but Karel.acc that idiot.acc no-one neg-saw “But Karel, no one saw that idiot.”
Researchers have suggested that the possibility of epithets in a construction rules out a movement analysis. Aoun et al. 2001, however, provide evidence from Lebanese Arabic that epithets can be used in resumption contexts which are associated with a movement derivation.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
5.1.1 Grammatical gender matching Analyzing the resumptive as the spell out of a movement copy provides an expla nation for gender-matching facts that were discussed in chapter 3. I established that certain gender mismatches that are allowed in cross-sentential pronominalization contexts are disallowed in CLD constructions. Across sentence boundaries, pronouns referring to the discourse referent of the neuter DP, tomu děvčeti (“that girl.dat”), can match either the grammatical gender of the DP (neuter) or the natural gender of its discourse referent (feminine), see (30). (30) Představila jsem ho tomu děvčeti a introduced aux.1sg.cl him.acc that girl.dat.neut and ta/to mu stiskla/o ruku. that.fem/that.neut him pressed.fem/neut hand “I introduced him to that girl and she shook his hand.”
In CLD, however, the resumptive element must match its antecedent in grammatical gender, see (31). (31) Tomu děvčeti, tomu/*té jsem dala jedničku. that girl.neut.dat that.neut.dat/that.fem.dat aux.1sg.cl gave one “That girl, I gave her an A.”
This is a consequence of the approach to resumption taken here. If the resumptive element spells out the ф-features of a movement copy, grammatical gender, rather than natural gender, will be realized on the resumptive. In this approach to resumption, no reference is made to the discourse referent of the left dislocate. 5.2 Why a demonstrative pronoun? A demonstrative resumptive is strongly preferred (possibly required) by speakers in CLD constructions. The other possible resumptive element, the strong form of the personal pronoun, is dispreferred, and even rejected, by most speakers. Given the analysis presented above, both types of strong pronouns should, in principle, be available to resume a left dislocated DP in CLD. This, however, is not what is found. I suggest that this preference can be explained by the interaction between the discourse function of CLD and the distribution of demonstrative vs. personal pronouns in the language in general. In order to develop an understanding of the interaction between discourse function and the distribution of different pronominal elements, it is necessary to
I refer the reader to their paper for the details of their analysis and leave it as an open question whether their solution can be extended to Czech CLD.
Chapter 5. The syntax of resumption
consider the activation status of the discourse referent of left dislocated elements. We established in the previous chapter that the discourse referent of the left dis locate in both CLD and HTLD is familiar (chapter 4: (23)) and that its antecedent is not the backward-looking center of its clause. Constraints on pronoun choice in Czech are also relevant. There are two strong pronominal forms that refer to third person referents: personal and demonstrative pronouns. Each of these pronominal elements respects certain constraints regarding the status of its discourse referent. Demonstrative pronouns are sensitive to the discourse status of their antecedent, while personal pronouns are sensitive to the discourse status of their referent in the following context. Personal pronouns tend to refer to elements that perseverate in the discourse context (their discourse referent is the preferred center (Centering Theory)), but are not as sensitive to the discourse status of their antecedents (see also Kresin 1994).18 Demonstrative pronouns, on the other hand, are not sensitive to the discourse status of their discourse referent in the following context, but are restricted from referring to an antecedent which is the backward-looking center of its clause (see Sturgeon 2008). This constraint is quite robust and holds of demonstrative pronouns in other languages as well (for English, Spanish, Russian, Japanese and Mandarin Chinese, see Gundel et al. 1993; for German, see Bosch et al. 2003; for Dutch, see Comrie 2000). If we combine the constraints on the use of personal and demonstrative pronouns in Czech with what we know about the discourse status of left dislocated elements, it is possible to speculate on the resumption preferences found in Czech left dislocation. A personal pronoun prefers that its discourse referent perseverate in the discourse context. This is true of the discourse referents of resumptives in HTLD, and personal pronouns can resume HTLD’ed elements. The discourse referent of the resumptive in CLD constructions, on the other hand, does not perseverate in the discourse. In CLD, the discourse referent of the left dislocate is generally not discussed in the following clauses. We saw in chapter 4 that the tendency, after a CLD construction, is to turn attention to alternatives to the discourse referent of the CLD’ed element. This characteristic of CLD constructions may contribute to the infelicity of personal pronoun resumptives. A demonstrative pronoun, on the other hand, requires that its antecedent not be the backward-looking center of its clause. Both HTLD’ed and CLD’ed elements
. Personal pronouns do, however, tend to be used in contexts in which their antecedent is not the backward-looking center of their clause; null pronouns generally refer to backwardlooking centers.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
satisfy this restriction, as established above. Thus, this type of pronominal element should be felicitous in both constructions. It is. Resuming CLD’ed elements with demonstrative pronouns is not restricted to Czech. Demonstrative resumptives are also employed in Dutch and German CLD. Work on the discourse function of CLD constructions in these languages suggests that they may also have a contrastive topic discourse function; researchers describe these constructions as “contrastive” (for Dutch see Zwart 1998; for German see Grohmann 2003; Frey 2004a, 2005; among others). Additionally, the demonstrative pronoun in these two languages is also specialized for reference to non-topical entities (for Dutch see Comrie 2000; for German see Bosch et al. 2003). Linking the choice of resumptive element to the discourse function of the construction appears to be a promising avenue for future research.
6. Resuming other categories In section 5.1, I showed that resumption of DP’s involves spell out of the phi-features associated with the movement copy. The fact that the resumptive matches the DP in syntactic, rather than semantic, features is reflected in the choice of resumptive when there is a mismatch in grammatical and natural gender (the gender of the DP and the gender of the denotation of the DP are not the same). In these cases, the resumptive element expresses the grammatical gender of the overt realization in the highest position. This is not, however, how the resumptive is determined for non-DP’s. In some cases, the resumptive appears to be determined by the denotation of the left dislocate (and its copy), not by grammatical features. In these cases, a demonstrative with the same denotation spells out the copy. In other cases, though, there appears to be a choice between choosing a demonstrative whose denotation matches that of the dislocate or, alternatively, pronouncing part of the copy itself. The resumptives used in these contexts correspond morphologically to the range of wh-expressions in the language. 19 A chart is provided in (32). (32)
Wh-Phrase a. kdo (“who”, nom) b. koho (“who”, acc) c. kde (“where”, loc) d. kam (“where”, dir)
Demonstrative ten (“that”, nom) toho (“that”, acc) tam (“there”, loc) tam (“there”, dir)
. For non-DP’s, demonstratives are the only pronominal form available.
Chapter 5. The syntax of resumption
e. kdy f. jak g. jaký
(“when”) (“how”) (“what kind of ”)
tehdy tak takový
(“then”) (“thus”) (“such”)
In the next section, dislocated predicates and propositions are shown to be resumed by a demonstrative which matches the dislocate in denotation. Other examples of resumptives which appear to be based on denotation, rather than syntactic category, are adverbials of manner, time and place. Prepositional phrases, on the other hand, are resumed either by the denotationally appropriate demonstrative or a syntactic spell out of the copy. The use of a resumptive which matches the syntactic, rather than semantic, features of the left dislocate appears to be motivated, in part, by the lexical inventory of Czech. 6.1 Properties and propositions Dislocated small clause AP’s and VP’s are resumed by a demonstrative which denotes properties. Mikkelsen 2005 argues that the pronominal forms used to denote properties in English are it and that. Evidence of this is provided in (33). (33) a. John is [considerate]. {It/That} is a good thing to be. b. John [talks quietly]. {It/That} is a good thing to do. (Mikkelsen 2005: (5.15))
The corresponding proform in Czech is the neuter demonstrative. This form is used in contexts such as those in (33), see (34). (34) a.
Jan je [silný]. To se hodí. Jan is strong that.neut refl-cl is-useful “Jan is strong. That is useful (for him).”
b.
Jan [mluví potichu]. To je dobré dělat. Jan talks quietly that.neut is good do.inf “Jan talks quietly. That is a good thing to do.” Lit: “Jan talks quietly. That is good to do.”
Mikkelsen suggests that “the form of a pronoun reflects its semantic type, which, in turn, must match the semantic type of its antecedent” (Mikkelsen 2005: (5.12)). If this is the case, the neuter demonstrative reflects the property semantic type in (34a–b). This pro-form resumes CLD’ed properties in Czech. I assume that this pro-form reflects the semantic type of the copy it spells out. The example in (35) illustrates. (35) …milostivá, to jsemf… gracious that.neut am. “Gracious, that I am.”
(Czech National Corpus)
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
The resumptive in this case is not a pro-form which matches the syntactic category of adjective (the corresponding form in Czech would be takový (“such”)), but rather a form which matches the semantic category of the dislocate, a property-denoting demonstrative. The same resumptive element, the neuter demonstrative, is found when VP’s are fronted. Consider the fronted VP in (36). (36) Zpívat Beatles, to jsem slyšela jenom Martina, ale… sing.inf Beatles that.neut aux.1sg.cl heard only Martin.acc but “Singing the Beatles, that I have only heard Martin, but…”
Again, it is not a verbal pro-form which resumes the dislocate, but a propertydenoting demonstrative. Dislocated CP’s are not resumed with a form corresponding to the syntactic category of the dislocate, but, instead are resumed by a demonstrative that matches it in semantic category. Neuter demonstratives resume dislocated CP’s, as well as dislocated property-denoting XP’s, (37a–b). (37) a.
Co přinesl Honza na večírek, to bych rád věděl. what brought Honza on party that.neut cond.1sg.cl joy knew “What Honza brought to the party, I would like to know that.”
b. Že Martin už odešel, to vím. that Martin already left that.neut know.1sg “That Martin already left, I know that.”
This is the form that is used in cross-sentential contexts to refer to propositions, (38). (38) Slyšelas, že Honza ted’ bydlí v Praze? Jo, heard.aux.2sg.cl c Honza now lives in Prague yeah slyšela jsem to. heard aux.1sg.cl it.cl “Did you hear that Honza now lives in Prague? Yeah, I heard that.”
Under the assumption that the neuter demonstrative is proposition-denoting, as well as property-denoting, again, the form used to resume propositions matches the dislocate in semantic, rather than syntactic, type. 6.2 Manner, place and time adverbials The resumptive element corresponding to dislocated manner, time and place adverbials is based on semantic, rather than syntactic category as well. This is especially clear when temporal and place adverbials, whose syntactic category is PP,
Chapter 5. The syntax of resumption
are considered. The resumptive element used in those cases is not a PP, but, rather, a demonstrative that matches the semantic category of the dislocate. First, we consider manner adverbials. To understand the nature of the resumptive element in these cases we turn to Landman et al. 2003. They suggest that there is a specialized pro-form that refers to manner adverbials in Polish, Russian and German, see the Polish (39). (39) On tańczyl tak. he danced thus “He danced like that.”
(Landman et al. 2003: (1a))
English lacks an exact correspondent to tak (“thus”) in Polish; expressions such as like that and so loosely correspond to this Polish anaphor. Landman et al. point out that this anaphor has a corresponding wh-form, jak (“how”), which is used to question manner. Czech parallels Polish in this way; its inventory of anaphors includes the manner anaphor, tak, as well as the corresponding wh-expression, jak, (40). (40) Jan tancoval tak. Jan danced thus “Jan danced like that.”
This pro-form is used as the resumptive element when an adverb is CLD’ed, (41). (41) Potichu, tak zavřel dveře. quietly thus closed doors “Quietly, I closed the door like that.”
Directional and temporal adverbials are also resumed by demonstratives that match the left dislocate in semantic, not syntactic, type. In (42), a directional PP is resumed by the corresponding demonstrative pro-form. (42) Do Budapešti, tam/*do té pojedu na výlet. to Budapest there to that go.1sg on trip “To Budapest, I am going there on a trip.”
The corresponding resumptive PP (do té (“to that”)) is ungrammatical. The same pattern is found with temporal adverbials. Though the left dislocate is a PP, the corresponding resumptive element is a temporal demonstrative: there is correspondence in semantic, not syntactic, type. The temporal PP in (43) is resumed by the corresponding temporal pro-form, tehdy (“then”). (43) V roce 1975 na jaře, tehdy/*na tom jsme se brali. in year 1975 on spring then/ on that.loc aux.1pl.cl refl-cl married “In 1975, in the spring, we got married then.”
Again, the corresponding PP (na tom (“on that”)) is ungrammatical.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
Locative adverbials behave a bit differently. These adverbials can be resumed by an element which matches the dislocate in either semantic category or in syntactic form. In (44), both the resumptive PP (v té (“in that”)), and the resumptive locative demonstrative (tam (“there”)) are possible. 20 (44)
Ve své1 kanceláři, tam/ v té pracovala Jana1 celé odpoledne. in self ’s office.loc there in that.loc worked Jana all afternoon “In her office, Jana worked there all afternoon.” Lit: “In her office, there/in that Jana worked all afternoon.”
Like directional and temporal adverbials, the resumptive can match the dislocate in semantic category (tam (“there”)). But, unlike those other PP adverbials, syntactic correspondence is also possible. This pattern of resumption is surprising, given that the Czech lexicon has a demonstrative form that corresponds the semantic category of locative. There may be some subtle semantic difference when a locative pro-form vs. a resumptive PP is used. This is a question for future research. Resumption and
the syntactic category of prepositional phrase is considered in more depth in the following section when other types of PP’s are considered. 6.3 Prepositional phrases In the previous section, we saw that there are demonstrative correspondents to manner adverbials such as directional and temporal PP’s. There is no corresponding demonstrative pro-form in the Czech lexicon for other PP’s, such as instrumentals. Due to this fact, resumption patterns with the syntactic matching found with locative PP’s. When, for example, instrumental prepositional phrases are CLD’ed, they are resumed by the corresponding P head with a complement resumptive pronoun corresponding to the syntactic category of the complement DP. In (45), a dislocated PP is resumed by the P and a resumptive element corresponding to its complement. (45) S bratrem, s tím pojede Honza na výlet. with brother.instr with that.instr go.3sg Honza on trip “His brother, Honza is going on a trip with him.”
Stranding only the head of the phrase (P0) is disallowed due to the fact that prepositions cannot strand in Czech. When there is a pro-form which corresponds to the entire prepositional phrase (as is the case with manner and temporal adverbial PP’s), that pro-form
. Note that, though the wh-expressions for the locative and the directional are distinct, the directional and locative demonstrative forms are identical, (compare (42) and (44)).
Chapter 5. The syntax of resumption
spells out in [Spec, IP], otherwise resumption proceeds as in (45). With locative adverbials, both options are available, (44). 6.4 Conclusion To sum up, resumption in Czech CLD derives from two assumptions in the grammar: null elements cannot bear a prosodic rise and economy constraints require maximal deletion of lower copies. Satisfaction of these competing constraints requires minimal spell out of the copy in [Spec, IP]. Resumption matches either the syntactic or the semantic category of the copy of the CLD’ed XP in [Spec, IP]. For DP’s, this matching is syntactic; the phifeatures of the copy of the dislocate are spelled out. However, for properties and propositions, the resumptive clearly matches not the syntactic, but the semantic type of the dislocate. The resuming element is the neuter demonstrative (the form that denotes properties and propositions), rather than a pronominal matching the category: adjective, verb or CP. Manner, time and place adverbials behave similarly. The resumptive element is not of the syntactic category PP but, instead, is a demonstrative form which denotes manner, time or place. When considering dislocated PP’s, we see that the nature of resumption is also constrained by the lexical inventory of Czech. Czech lacks corresponding pro-forms for, for example, instrumental PP’s. In those cases, the resumptive pronominal spells out the syntactic category of the movement copy. 7. Previous accounts of resumptive spell out Grohmann’s 2003 account of German CLD is the most successful recent movement approach to left dislocation and my account follows his in linking resumption to partially spelled out lower copies within a Copy and Delete theory of movement. Our analyses diverge, however, in the explanation and implementation of the spell out requirement. Grohmann appeals to a theory-internal requirement that only one copy of an XP can appear in any “prolific domain” (loosely corresponding to phase). When more than one copy of an XP appears in a single domain, the lower element must spell out as a resumptive. Ultimately, due to structural differences between the left peripheries of German and Czech, Grohmann’s approach to resumption is untenable for Czech CLD. As in my analysis, under Grohmann 2003, the CLD’ed XP undergoes two leftward movements. The CLD’ed XP moves first to the left edge of the matrix clause, [Spec, TopP], and then to a higher functional projection, [Spec, CP]. Grammatical constraints interfere with the normal deletion process associated with lower copies. The intermediate copy in [Spec, TopP] and the higher copy in [Spec, CP] both Spell out overtly, see (46).
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
(46) Diesen Mann, den kenne ich nicht. this.acc man that-one.acc know I not “This man, I don’t know him.” (Grohmann 2003: chapter 4 (2b))
Grohmann accounts for resumption in German CLD constructions within his larger theory of Prolific Domains and anti-locality effects. “Prolific Domains” correspond to three domains within the clause: the θ-domain (VP, vP); the ф-domain, (TP, AgrP’s); and the ω-domain. The ω-domain corresponds to what is generally thought of as the C-domain and, for Grohmann, includes CP and TopP (see Rizzi 1997). Based on evidence from several types of constructions (reflexivization, CLLD, among others), Grohmann proposes that when more than one copy of an XP occurs within the same Prolific Domain, the lower copy must spell out as a resumptive. Resumption in German CLD constructions follows from this proposal. Two copies of the CLD’ed XP occur in the same domain, the ω-domain: the left dislocate occurs in [Spec, CP] and another copy occurs in [Spec, TopP]. The lower copy in the ω-domain spells out as a demonstrative pronoun, see (47). (47) |ω [CP XP [TopP XP- > RP |ф [TP…XP… |θ [vP…XP…]]]]]]
The lowest two copies are the only copies in their respective domains, and, thus, do not require overt expression. This is where Czech diverges from German. In Czech CLD, the resumptive element and the left dislocate are in different prolific domains. Evidence for this comes from the co-occurrence of wh-expressions and CLD’ed XP’s. Both the left dislocate and the wh-phrase occur in what Grohmann refers to as the C-domain, but the resumptive element appears lower, in [Spec, IP], see (48). (48) Toho psa, komu toho jsi dala? that dog.acc who.dat that.acc aux.2sg.cl gave “That dog, who did you give him to?”
As shown in chapter 2, second position clitics occur in I0, a position within the ф-domain; the demonstrative resumptive occurs in the specifier position of this projection. German differs crucially from Czech in that wh-phrases cannot co-occur with CLD’ed elements in interrogatives. The reason is that the demonstrative resumptive and the wh-phrase compete for the same structural position, [Spec, TopP], which can be identified as the position preceding the second position verb, see (49). The exclamation mark in (49) indicates that it cannot be interpreted as an interrogative; only an exclamative interpretation is possible. (49) !Diesen Gast, den hat wann der Oberkellner gegrüßt? this.acc guest that.acc has when the maître d’ greeted “This guest, when did the maître d’ greet him?” (Grohmann 2003: chapter 4 (33b))
Chapter 5. The syntax of resumption
Since demonstratives appear in the same position as wh-expressions in German ([Spec, TopP]), and we know that wh-expressions are in the ω-domain, both the overt CLD’ed expression and the resumptive element appear in the same Prolific Domain. This, however, is not the case in Czech. CLD’ed elements and their resumptives appear in distinct Prolific Domains. Thus, the grammatical constraint forcing spell out of the resumptive element in German does not apply to the resumptive element in Czech. Another explanation is required. My analysis contrasts with Grohmann’s in that I link spell out to a prosodic requirement which is independently known to hold of CT elements in [Spec, IP]. The fact that null material cannot host a prosodic rise leads to the spell out of the resumptive element in CLD. It seems likely that my account for Czech CLD could extend to the parallel construction in German. German CLD has a similar discourse function (Frey 2004a, b; 2005). Also, the resumptive element in German appears in the same structural position as do German CT’s, [Spec, TopP]. Like in Czech, German CT’s are also associated with an intonational rise. Since they occupy the same position as CT’s, the resumptive in German CLD may also be associated with an intonational rise and my analysis could be extended to resumption in German CLD (for CT intonation in German see Féry 2005). 8. Conclusion The analysis of resumption presented here connects the syntax, prosody and pragmatics of Czech CLD. The interplay of these components of the construction explains the resumptive pattern. Overt resumption occurs as a direct result of the prosodic character of the construction. The CT intonational contour falls obligatorily on the element in [Spec, IP], requiring that the movement copy in that position spell out overtly. Rather than a full movement copy spelling out, however, a resumptive element spells out. I connect this to principles of economy. Where it is possible, minimize spell out. The prosody of CLD requires an overt element in [Spec, IP], however, this element need not be a full copy of the left dislocate. A resumptive element, then, satisfies both the prosodic and economic constraints within the grammar. The nature of the resumptive copy was also considered. I showed that the resumptive element must spell out the phi-features of DP’s; grammatical gender matching facts in Czech CLD provides evidence of this. However, it is the semantic, rather than syntactic, category of the left dislocate which the demonstrative resumptive spells out in the case of left dislocated properties and propositions. This semantic, rather than syntactic, matching is also evident when manner, time and place adverbials are considered; though the CLD’ed element is a PP, the demonstrative matches the denotational type, not the syntactic category, of the dislocate. Syntactic matching occurs when the lexical inventory of Czech supports it.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
The analysis presented here considers CLD at the syntactic-pragmatic-prosodic interface. As opposed to strictly syntactic approaches, like that taken in Grohmann 2003, the features motivating movement under my account are closely linked to the pragmatic function of the construction. Under Grohmann 2003, the features motivating both leftward movements of the left dislocate are identical. He is not concerned with the question of how these syntactic configurations might relate to, or interact with, pragmatic properties of the constructions. Additionally, spell out of the resumptive element is the consequence of a purely a syntactic condition within his approach. Under my account, spell out of the resumptive is a natural consequence of the prosody of the construction. Considering left dislocation at the interfaces of the grammar has made significant strides in developing an understanding of this type of construction.
chapter 6
Conclusion 1. Overview The contributions of this work lie in two areas: new empirical research on left dislocation constructions in Czech and the establishment of a close link between the formal syntax and the pragmatic interpretation of left peripheral positions. The empirical contribution furthers an understanding of Czech in particular and provides a framework within which to examine left dislocation constructions in other Slavic languages, an empirical domain that has not yet been closely examined. The formal contribution adds to an understanding of not only the syntactic motivation, but also the pragmatic motivation, for an otherwise mysterious two-step leftward movement (to [Spec, IP] and to [Spec, TopP]) in CLD. Approaching syntax from a pragmatic perspective increases the explanatory power of a syntactic account. This understanding of the left periphery can also be extended to other languages with similar left dislocation constructions. As discussed in chapters 4 and 5, Germanic languages such as Dutch and German also have CLD constructions with a contrastive interpretation (for an early analysis of Dutch, see van Riemsdijk & Zwarts 1974/1997). Linking the syntactic analyses to their pragmatic function is a promising avenue for future research.
2. The syntax and pragmatics of left dislocation in Czech I first provide an overview of Czech clausal syntax. It is not necessary to posit an exploded CP domain to account for the elements that are found at the left edge of the clause. Instead, in most cases, all left peripheral XP’s appear within the inflectional domain. The pivotal A-bar position, [Spec, IP], contains contrastive topics, topics, foci, as well as wh-expressions. In positions dominating that domain, we find left dislocated elements and, under certain circumstances, wh-phrases. In the second strand of syntactic research, I develop an analysis of three distinct left dislocation constructions: CLD and HTLD I, II. These constructions are similar in some ways (a left dislocated element binds a clause-internal pronominal
The left periphery: The interaction of syntax, pragmatics and prosody in Czech
element), but are considerably different when considered closely. I explored the syntactic, pragmatic and prosodic differences between these constructions. In CLD, the left dislocate moves from a clause-internal position to a position at the left periphery, while in HTLD I and II the left dislocate is base generated in a left peripheral position. These syntactic differences correlate with distinct pragmatic interpretations. I evaluate attested textual examples and determine that CLD has a contrastive topic interpretation, and HTLD I, II serve to promote the discourse referent of the left dislocate to topic status in the following discourse. 2.1 Movement vs. non-movement derivations There is substantial evidence in support of movement of the left dislocate from a clause-internal position to a position in the left periphery in CLD and a parallel lack of evidence of such movement in HTLD. First, CLD’ ed left dislocates are interpreted in their base positions for purposes of Conditions A and C. For instance, reflexive pronouns can appear in CLD’ ed constituents. Since there is no potential binder for the reflexive in its surface position, reconstruction of the left dislocate to a clause-internal position is required. The appearance of reflexives is strictly ungrammatical in HTLD I and quite degraded in HTLD II. This suggests an alternate syntactic structure for these constructions. Under a base-generation account, reflexive pronouns are predicted to be ungrammatical within hanging topics because they are not grammatically bound by a clause-internal antecedent. These surface differences between CLD and HTLD I, II are accounted for by the movement/non-movement nature of their syntactic derivations. 2.2 The pragmatics of left dislocation There is considerable evidence that points to a contrastive topic interpretation of CLD and a topic promotion function of HTLD I, II. From a corpus of approximately 100 attested textual examples of left dislocation constructions in Czech, using case matching as a diagnostic, I identified 23 unambiguously CLD examples and 18 examples of HTLD I. Through a phonetic experiment on intonation of CLD constructions, I determined that the resumptive element in [Spec, IP] carries the intonational rise associated with the contrastive topic interpretation. This, along with the context in which these constructions appear in textual examples, points to a contrastive topic interpretation of these constructions. HTLD I, on the other hand, lacks the prosodic rise associated with a contrastive topic interpretation. Additionally, the contexts in which these examples appear strongly suggest a topic promotion analysis of this construction. One convincing piece of evidence supporting a topic promotion function of this construction is
Chapter 6. Conclusion
that the discourse referent of the left dislocate continues to be discussed in the following clauses in 67% of attested examples. This contrasts with CLD. In those cases, the discourse referent of the left dislocate perseverates in the discourse only 13% of the time.
3. The syntax of resumption The analysis of movement in CLD is very closely linked to the pragmatics of the construction: the formal features motivating movement are pragmatically motivated, unlike in recent movement analyses of this type of construction (Grohmann 2003). The first movement of the left dislocate to [Spec, IP] is motivated by contrastive topic operator features, and the second movement by a topic feature. The appearance of the resumptive in movement left dislocation constructions has been a long-standing puzzle in the literature on left dislocation and my approach is unique in that I link the appearance of the resumptive to the prosody of the construction. Since left dislocation in CLD involves movement, the resumptive element is in the position of a movement copy of the left dislocate and should, under the Copy and Delete Theory of movement, remain unpronounced at PF. This element, however, is in a position associated with a prosodic rise. In order to realize this rise, an overt element must appear in this position or the derivation will crash at PF. What spells out, though, is a pronominal element and not the full copy. I assume that this is due to economy constraints in the PF component of the grammar. Minimal spell out, up to the satisfaction of PF constraints, is required.
4. Conclusion This book provides a careful description of previously undocumented left dislocation constructions in Czech (CLD and HTLD I, II) supported by textual examples. I also provide new insight into formal concerns that have been considered in the literature on left dislocation since Ross 1967. Does the left dislocate move to the left periphery of the clause? What mechanism requires spell out of the resumptive element in left dislocation? The answer to the first question is two-fold. Certain Czech left dislocation constructions involve movement, others do not. To address the second question, I establish a formal link between the syntax, pragmatics and prosody of CLD which provided functional motivation for spelling out an otherwise unpronounced movement chain link as a resumptive pronoun.
References Adger, David & Ramchand, Gillian. 2005. Merge and move: Wh dependencies revisited. Linguistic Inquiry 36(2): 161–94. Alexiadou, Artemis & Anagnostopoulou, Elena. 1998. Parametrizing AGR: Word order, V-movement and EPP-checking. Natural Language and Linguistic Theory 16(3): 491–539. Altmann, Hans. 1981. Formen der ‘Herausstellung’ im Deutschen: Rechtsversetzung, Linksversetzung, Freies Thema und verwandte Konstruktionen. Tübingen: Max Niemeyer. Anagnostopoulou, Elena. 1997. Clitic left dislocation and contrastive left dislocation. In Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk & Frans Zwarts (Eds), 151–92. Amsterdam: John Benjamins. Anagnostopoulou, Elena, van Riemsdijk, Henk & Zwarts, Frans (Eds). 1997. Materials on Left Dislocation. Amsterdam: John Benjamins. Arnaudova, Olga. 2004. Contrastive features, clitic doubling and the left periphery of the Bulgarian clause. In Formal Approaches to Slavic Lingusitics 13, The South Carolina Meeting 2004, Steven Franks, et al. (Eds), 13–26. Ann Arbor MI: Michigan Slavic Publications. Aoun, Joseph, Choueiri, Lina & Hornstein, Norbert. 2001. Resumption, movement and derivational economy. Linguistic Inquiry 32(3): 371–403. Arregi, Karlos. 2001. Focus and word order in Basque. Ms, MIT. Authier, J.-Marc. 1993. Nonquantificational Wh and weakest crossover. Linguistic Inquiry 24(1): 161–68. Beaver, David. 2004. The optimization of discourse anaphora. Linguistics and Philosophy 27(1): 3–56. Birner, Betty J. & Ward, Gregory. 1998. Information Status and Noncanonical Word Order in English. Amsterdam: John Benjamins. Bobaljik, Jonathan. 2002. A-chains at the PF-interface: Copies and ‘covert’ movement. Natural Language and Linguistic Theory 20(2): 197–267. Boersma, Paul & Weenink, David. 1999–2006. Praat: A system for doing phonetics by computer. Bosch, Peter, Rozario, Tom & Zhao, Yufan. 2003. Demonstrative pronouns and personal pronouns: German der vs er. In Proceedings of the EACL2003: Workshop on the computational treatment of anaphora. Bošković, Žejlko. 1998. Wh-phrases and wh-movement in Slavic. Paper presented at Comparative Slavic Morphosyntax Workshop, Bloomington IN. Bošković, Žejlko. 2001. On the Nature of the Syntax-Phonology Interface: Cliticization and related phenomena. Oxford: Elsevier. Büring, Daniel. 1997. The great scope inversion conspiracy. Linguistics and Philosophy 20(2): 175–94. Büring, Daniel. 1999. Topic. In Focus: Linguistic, Cognitive and Computational Perspectives, Peter Bosch & Rob van der Sandt (Eds), 142–65. Cambridge: CUP. Büring, Daniel. 2001. Let’s phrase it! – Focus, word order and prosodic phrasing in German double object constructions. In Competition in Syntax, Gereon Müller & Wolfgang Sternefeld (Eds), 69–105. Berlin: Mouton de Gruyter.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech Büring, Daniel. 2003. On D-trees, beans and B-accents. Linguistics and Philosophy 26(5): 511–45. Carlson, Lauri. 1983. Dialog games: An approach to discourse analysis. Ms, Cologne University. Chafe, Wallace. 1975. Givenness, contrastiveness, definiteness, subjects, topics and point of view. In Subject and Topic, Charles N. Li (Ed.), 25–55. New York, NY: Academic Press. Chomsky, Noam. 1986. Barriers. Cambridge MA: The MIT Press. Chomsky, Noam. 1991. Some notes on the economy of derivation and representation. In Principles and Parameters in Comparative Grammar, Robert Freidin (Ed.), 417–54. Cambridge MA: The MIT Press. Chomsky, Noam. 1993. A minimalist program for linguistic theory. In The View from Building 20, Ken Hale & Samuel J. Keyser (Eds), 1–52. Cambridge MA: The MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge MA: The MIT Press. Cinque, Guglielmo. 1977. The movement nature of left dislocation. Linguistic Inquiry 8: 397–411. Clark, Herbert H. & Haviland, Susan E. 1977. Comprehension and the given-new contract. In Discourse Comprehension and Comprehension, Roy O. Freedle (Ed.), 1–40. Norwood NJ: Ablex. Comrie, Bernard. 2000. Pragmatic binding: Demonstratives as anaphors in Dutch. In Proceedings of the 23rd Annual Meeting of the Berkeley Linguistics Society, 1997, 50–61. Ann Arbor MI: Sheridan Books. Cummins, George M. 1998. Definiteness in Czech. Studies in Language 22(3): 567–96. Davison, Alice. 1984. Syntactic markedness and the definition of sentence topic. Language 60(4): 797–847. Demirdache, Hamida. 1991. Resumption Chains in Restrictive Relatives, Appositives and Dislocation Structures. Ph.D. Dissertation, MIT. Depiante, Marcela. 2001. On null complement anaphora in Spanish and English. Probus 13(3): 193–221. Emonds, Joseph. 1970. Root and Structure Preserving Transformations. Ph.D. Dissertation, MIT. Emonds, Joseph. 1978. The verbal complex V’-V in French. Linguistic Inquiry 9: 151–75. Engdahl, Elisabet. 1985. Parasitic gaps, resumptive pronouns and subject extractions. Lingusitics 23(1): 3–44. Fox, Danny. 2002. Antecedent contained deletion and the copy theory of movement. Linguistic Inquiry 33(1): 63–96. Fox, Danny & Pesetsky, David. 2004. Cyclic linearization of syntactic structure. Theoretical Linguistics 31(1–2): 1–46. (Special issue on object shift in Scandinavian, Katalin É. Kiss (Ed.)). Féry, Caroline. 2005. The prosody of topicalization. University of Potsdam. Ms. Franks, Steven. 1998. Clitics in Slavic. Paper presented at Comparative Slavic Morphosyntax Workshop, Bloomington IN. Frey, Werner. 2004a. Notes on the syntax and pragmatics of German left dislocation. In The Syntax and Semantics of the Left Periphery, Horst Lohnstein & Susanne Trissller (Eds), 203–33. Berlin: Mouton de Gruyter. Frey, Werner. 2004b. A medial topic position for German. Linguistische Berichte 198: 153–90. Frey, Werner. 2005. Pragmatic properties of certain German and English left peripheral constructions. Linguistics 43(1): 89–129. Fried, Mirjam. 1994. Second-position clitics in Czech: Syntactic or phonological? Lingua 94(2–3): 155–75.
References
Geluykens, Ronald. 1992. From Discourse Process to Grammatical Construction: On left dislocation in English. Amsterdam: John Benjamins. Givón, Talmy. 1983. Topic continuity in discourse: An introduction. In Topic Continuity in Discourse, Talmy Givón (Ed.), 4–41. Amsterdam: John Benjamins. Goldberg, Lotus. 2002. An elucidation of null direct object structures in Modern Hebrew. In Proceedings of the West Coast Conference on Formal Linguistics 21, Line Mikkelsen & Chris Potts (Eds), 99–122. Somerville MA: Cascadilla. Goldberg, Lotus. 2005. Ellipsis and Identity, Ph.D. Dissertation, McGill University. Gregory, Michelle L. & Michaelis, Laura A. 2001. Topicalization and left-dislocation: A functional opposition revisited. Journal of Pragmatics 33(11): 1665–706. Grohmann, Kleanthes. 2003. Prolific Domains: On the anti-locality of movement dependencies. Amsterdam: John Benjamins. Grosz, Barbara J., Joshi, Aravind & Weinstein, Scott. 1983. Providing a unified account of definite noun phrases in discourse. In Proceedings of the 21st Annual Meeting of the Association for Computational Linguistics, 44–50. Cambridge MA. Grosz, Barbara J., Joshi, Aravind & Weinstein, Scott. 1995. Centering: A framework for modeling the local coherence of discourse. Computational Linguistics 21(2): 203–25. Grosz, Barbara J., Joshi, Aravind & Sidner, Candace. 1986. Towards a computational theory of discourse interpretation. Ms. Gundel, Jeanette K. 1974/1988. The Role of Topic and Comment in Linguistic Theory. [Outstanding Dissertations in Linguistics]. New York NY: Garland. (1974: Ph.D. Dissertation, University of Texas, Austin). Gundel, Jeanette K. 1985. ‘Shared knowledge’ and topicality. Journal of Pragmatics 9(1): 83–107. Gundel, Jeanette K., Hedberg, Nancy & Zacharski, Ron. 1993. Cognitive status and the form of the referring expression. Language 69(2): 274–307. Hajičová, Eva, Partee, Barbara & Sgall, Petr. 1998. Topic-Focus Articulation, Tripartite Structures and Semantic Content. Dordrecht: Kluwer. Hajičová, Eva, Sgall, Petr & Veselá, Kateřina. 2003. Information structure and contrastive topic. In Annual Workshop of Formal Approaches to Slavic Languages: The Amherst Meeting 2002, Wayles Brown, Ji-Yung Kim, Barbara Partee & Robert Rothstein (Eds), 219–34. Ann Arbor MI: Michigan Slavic Publications. Hajičová, Eva & Sgall, Petr. 2004. Degrees of contrast and the topic-focus articulation. Information Structure–Theoretical and empirical aspects, A. Struebe (Ed.), 1–13. Berlin: Walter de Gruyter. Halliday, Michael A.K. 1967. Notes on transitivity and theme in English, Part 2. Journal of Linguistics 3: 199–244. Hamblin, C.L. 1973. Ouestions in Montague English. Foundations of Language 10: 41–53. Hinterhoelzl, Roland & Diana Pili. 2003. Argument shift phenomena across language types. Ms. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge MA: The MIT Press. Kiss, Katalin É. 1987. Configurationality in Hungarian. Dordrecht: D. Reidel. Koopman, Hilda. 1982. Control from COMP and comparative syntax. The Linguistic Review 2: 365–91. Koopman, Hilda. 1984. The Syntax of Verbs: From verb-movement rules in the Kru language to Universal Grammar. Dordrecht: Foris. Kresin, Susan. 1994. Third Person Reference in Russian and Czech, Ph.D. Dissertation, UC, Berkeley.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech Krifka, Manfred. 1999. Additive particles under stress. In Proceedings of SALT 8, T. Mathews & D. Strolovitch (Eds), 111–128. Ithaca NY: CLC Publications. Kučerová, Ivona. 2007. The Syntax of Givenness, Ph.D. Dissertation, MIT. Kuno, Susumu. 1982. The focus of the question and the focus of the answer. In CLS: Papers from the parasession on nondeclaratives, Robinson Schneider, Kevin Tuite, Robert Chametzky (Eds), 134–157. Chicago IL: Chicago Linguistics Society. Lambrecht, Knud. 1994. Information Structure and Sentence Form: Topic, focus and the mental representations of discourse referents. Cambridge: CUP. Lambrecht, Knud & Michaelis, Laura A. 1998. Sentence accent in information questions: Default and projection. Linguistics and Philosophy 21(5): 477–544. Landau, Idan. 2005a. EPP Extensions. Ms, Ben Gurion University. Landau, Idan. 2005b. Chain resolution in V(P)-fronting. Ms, Ben Gurion University. Landman, Meredith & Morzyci, Marcin. 2003. Event-kinds and the representation of manner. In Proceedings of the Western Conference on Linguistics (WECOL) 2002, Nancy Mae Antrim, Grant Goodall, Martha Schulte-Nafeh & Vida Samiian (Eds). Fresno CA: California State University. Lasnik, Howard & Stowell, Tim. 1991. Weakest crossover. Linguistic Inquiry 22(4): 687–720. Lenertová, Denisa. 2001. On clitic placement, topicalization and CP-structure in Czech. In Current Issues in Slavic Linguistics, Uwe Junghanns Gerhild Zybatow, Grit Mehlhorn & Luca Szucisch (Eds), 294–305. Berlin: Peter Lang. Lenertová, Denisa & Junghanns, Uwe. 2007. Wide focus interpretation with fronted focus exponents in Czech. In On Information Structure, Meaning and Form: Generalizations across languages, Kerstin Schwabe & Susanne Winkler (Eds), 347–363. Amsterdam: John Benjamins. Manetta, Emily. 2006. Peripheries in Kashmiri and Hindi-Urdu, Ph.D. Dissertation, University of California, Santa Cruz. McCloskey, James. 1991. Clause structure, ellipsis and proper government in Irish. Lingua 85: 259–302. McCloskey, James. 2005. Questions and questioning in a local English. Ms, University of California, Santa Cruz. McCloskey, James. 2006. Resumption. In The Blackwell Companion to Syntax, volume IV, Martin Everaert & Henk van Riemsdijk (Eds), 94–117. Oxford: Blackwell. Mikkelsen, Line. 2005. Copular Clauses: Specification, predication and equation. Amsterdam: John Benjamins. Molnár, Valeria. 1998. Topic in focus: On the syntax, phonology, semantics and pragmatics of the so-called ‘contrastive topic’ in Hungarian and German. Acta Linguistica Hungarica 45(1–2): 89–166. Nunes, Jairo. 1999. Linearization of chains and phonetic realization of chain links. In Working Minimalism, Samuel David Epstein & Norbert Hornstein (Eds), 217–49. Cambridge MA: The MIT Press. Pesetsky, David. 1998. Some optimality principles of sentence interpretation. In Is the Best Good Enough? Optimality and competition in syntax, Pilar Barbosa, Danny Fox, Paul Hagstrom, Martha McGinnis & David Pesetsky (Eds), 337–83. Cambridge MA: The MIT Press. Pollock, Jean-Yves. 1989. Verb movement, UG and the structure of IP. Linguistic Inquiry 20(3): 365–424.
References
Prince, Ellen. 1981. Topicalization, focus-movement and Yiddish movement: A pragmatic differentiation. In Proceedings of the 7th Annual Meeting of the Berkeley Lingusitics Society, Danny K. Alford et al. (Eds), 249–64. Berkeley CA: BLS. Prince, Ellen. 1992. The ZPG letter: Subjects, definiteness and information-status. In Discourse Description: Diverse linguistic analyses of a fund-raising text, William C. Mann & Sandra A. Thompson (Eds), 295–325. Amsterdam: John Benjamins. Prince, Ellen. 1997. On the functions of left-dislocation in English discourse. In Directions in Functional Linguistics, Akio Kamio (Ed.), 117–43. Amsterdam: John Benjamins. Puskás, Genoveva. 2000. Word Order in Hungarian: The syntax of A’-positions. Amsterdam: John Benjamins. Reinhart, Tanya. 1981. Pragmatics and linguistics: An analysis of sentence topics. Philosophica 27: 53–93. Reinhart, Tanya. 1998. Wh-in-situ in the framework of the minimalist program. Natural Language Semantics 6(1): 29–56. Richards, Norvin. 2001. Movement in Language: Interactions and architectures. Oxford: OUP. Riemsdijk, Henk van. 1997. Left dislocation. In Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk & Frans Zwarts (Eds), 13–29. Amsterdam: John Benjamins. Riemsdijk, Henk van & Zwarts, Frans. 1974/1997. Left dislocation in Dutch and the status of copying rules. In Materials on Left Dislocation, Elena Anagnostopulou, Henk van Riemsdijk & Frans Zwarts (Eds), 13–29. Amsterdam: John Benjamins. (1974: MIT/ University of Amsterdam, Ms.). Řiha, Bohumil. 1992. Jak Jel Vítek do Prahy (‘How Vítek Went to Prague’). Prague: Axioma. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Elements of Grammar, Liliane Haegeman (Ed.), 281–337. Dordrecht: Kluwer. Roberts, Craige. 1996. Information structure in discourse: Towards an integrated formal theory of pragmatics. In Working Papers in Linguistics 49: Paper in Semantics, J.H. Yoon & Andreas Kathol (Eds), 91–136. Roberts, Craige. 1998. Focus, the flow of information and Universal Grammar. In Syntax and Semantics 29: The limits of syntax, Peter Cullicover & Louise McNally (Eds), 109–60. San Diego CA: Academic Press. Rodman, Robert. 1974/1997. On left dislocation. In Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk & Frans Zwarts (Eds), 31–54. Amsterdam: John Benjamins. (1974: Papers in Linguistics 7, 437–466). Rooth, Mats. 1985. Association with Focus, Ph.D. Dissertation, University of Massachusetts, Amherst. Ross, John Robert. 1967. Constraints on Variables in Syntax. Ph.D. Dissertation, MIT. Ross, John Robert. 1973. A fake NP squish. In New Ways of Analyzing Variation in English, C.J. Bailey & R.W. Shuy (Eds). Washington DC: Georgetown University Press. Rudin, Catherine. 1988. On multiple questions and multiple wh-fronting. Natural Language and Linguistic Theory 6(4): 445–501. Sato, Yosuke. 2005. Multiple spell-out, cliticization and the syntax-prosody/interface. Ms, University of Arizona. Scherpenisse, Wim. 1986. Topic, theme and the German initial field. In Topic, Focus and Configurationality: Papers from the 6th Groningen Grammar Talks, Werner Abraham & Sjaak de Meij (Eds), 277–93. Amsterdam: John Benjamins.
The left periphery: The interaction of syntax, pragmatics and prosody in Czech Sgall, Petr, Hajičová, Eva & Benešová, Eva. 1973. Topic, Focus and Generative Semantics. Kronberg Taunus: Scriptor Verlag. Sgall, Petr, Hajičová, Eva & Buráňová, Eva. 1980. Aktuální členění v češtině (‘Functional Sentence Perspective in Czech’). Prague: Academia. Sgall, Petr, Hajičová, Eva & Panevová, Jarmila. 1986. The Meaning of the Sentence in its Semantic and Pragmatic Aspects. Prague: Academia. Shaer, Benjamin and Frey, Werner. 2004. ‘Integrated’ and ‘non-integrated’ left-peripheral elements. In Proceedings of the Dislocated Elements Workshop, ZAS Berlin, November 2003, Benjamin Shaer, Werner Frey and Claudia Maienborn (Eds), 465–502. Stalnaker, Robert. 1978. Assertion. In Pragmatics, P. Cole (Ed.), 315–22. San Diego CA: Academic Press. Steedman, Mark. 2000. Information structure and the syntax-phonology interface. Linguistic Inquiry 31(4): 649–89. Steedman, Mark. 2002. The surface grammar of intonation and information structure. Paper presented at the Workshop on Contrast and Information Structure, Leipzig, Germany. Stjepanović, Sandra. 2003. A word order paradox resolved by copy deletion at PF. Linguistic Variation Yearbook 3(1): 139–77. Sturgeon, Anne. 2003. Two-tiered approach to binding domain formation: Evidence from Czech. In Annual Workshop on Formal Approaches to Slavic Linguistics, Wayles Browne, Ji-Yung Kim, Barbara Partee & Robert Rothstein (Eds), 495–514. Ann Arbor MI: Michigan Slavic Publications. Sturgeon, Anne. 2008. Topic and demonstrative pronouns. In Proceedings of Formal Description of Slavic Linguistics-5, Gerhild Zybatow, Luca Szucsich, Uwe Junghanns & R. Meyer (Eds). 514–525. Frankfurt: Lang. Szabolsci, Anna. 1981. The possessive construction in Hungarian: A configurational category in a non-configurational language. Acta Linguistica Academica Scientiarum Hungaricae 31(1–4): 261–89. Vallduví, Enric & Vilkuna, Maria. 1998. On rheme and kontrast. In Syntax and Semantics 29: The limits of syntax, Peter Cullicover & Louise McNally (Eds), 79–108. San Diego CA: Academic Press. Vat, Jan. 1981/1997. Left dislocation, connectedness and reconstruction. In Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk & Frans Zwarts (Eds), 67–92. Amsterdam: John Benjamins. (1981: Groninger Arbeiten zur germanistischen Linguistik 20: 80–103). Vergnaud, J.R. 1974. French Relative Clauses. Ph.D. Dissertation, MIT. Veselovská, Ludmila. 1995. Phrasal Movement and X0 Morphology: Word order parallels in Czech and English nominal and verbal projections. Ph.D. Dissertation, Univerzita Palackého. Veselovská, Ludmila. 2003. Analyzytické préteritum a opisné pasivum v češtině: dvojí způsob saturace silného rysu <+V> hlavy v* (‘Analysis of the preterite and the periphrastic passive in Czech: Two ways to saturate the strong feature <+V> of the v head’). Sborník Prací Filosofické Fakulty Brnenské Univerzita, Series Linguistica 51: 167–177. Veselá, Kateřina, Peterek, Nino & Hajičová, Eva. 2003. Topic-focus articulation in PDT: Prosodic characteristics of contrastive topic. The Prague Bulletin of Mathematical Linguistics 79–80: 5–22.
References
Ward, Gregory. 1988. The Semantics and Pragmatics of Preposing. New York, NY: Garland. Wiltschko, Martina. 1997. Parasitic operators in German left dislocation. In Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk & Frans Zwarts (Eds), 307–39. Amsterdam: John Benjamins. Zubizarreta, Maria Luisa. 1998. Prosody, Focus and Word Order. Cambridge, MA: The MIT Press. Zwart, Jan-Wouter. 1998. Where is syntax? Syntactic aspects of left dislocation in Dutch and English. In Syntax and Semantics 29: The Limits of Syntax, Peter Cullicover & Louise McNally (Eds), 365–93. San Diego CA: Academic Press.
Index
A Activation status, 84, 106–107, 119 Adverbs, 122–124 VP, 11–14, 21–22, 38 Agree, 8–9 Alexiadou, A., 20–22 Alternative questions see Contrastive topic Anagnostopoulou, E., 2, 3, 20–22 Anaphoricity scale, 91–93 Anti-locality, 126 Arnaudova, O., 39, 97 Authier, J., 32–33 Auxiliary verbs, 10 B B-accent, 68–69, 73, 82, 100 Beaver, D., 93 Binding Domain, 28, 45 Birner, B., 68, 77, 91 Bošković, Z., 31ff, 99, 110, 127 Bridge verbs, 23, 63 Bulgarian 25, 39, 97 Büring, D. Contrastive topic, 68, 73–81, 105ff Focus prosody, 27 C C-command, 28–29, 31, 54 Case matching, 40–41, 65 Catalán, 69 Centering Theory, 91, 93, 119 Backward looking center, 93, 119 Chain formation, 2–3 Operator variable chain, 2–3 Movement chain, 3, 110–113 Chinese, 119 Chomsky, N., 3, 7, 9, 28, 63, 116 Cinque, G., 2, 42
Clitics see Pronouns Clitic Left Dislocation, 97, 126 Common ground, 75 Complete Functional Complex, 28, 45 Comrie, B., 119, 120 Condition A, 28–29, 31ff, 45, 48, 49, 53–54 Condition C, 46, 48, 50, 53–55 Connectivity, 2, 3, 42, 44–50, 53–54 Contrastive topic Alternative questions, 85, 88, 96 CT-congruence, 76, 77, 80, 82 CT-marking, 73, 76–77, 79, 80–82 CT-value, 76–77, 79, 82 Prosody, 100–104 Conventional meaning see Implicatures Conversational implicature see Implicatures Copy and Delete, 3, 9, 23–24, 107, 116, 125 see Merge, Movement copies CP domain, 4, 34–38, 60–64, 129 CT-congruence see Contrastive topic CT-marking see Contrastive topic CT-value see Contrastive topic Czech National Corpus, 1ff, 5 D Demonstrative pronouns see Pronouns Discourse trees, 74 Dutch, 2, 41, 119, 120, 129
E Economy, 7, 36, 108ff, 113, 115, 116 Economy of pronunciation, 113, 115, 116 Embedded clauses, 23, 25, 57–58, 62–63 Engdahl, E., 116 English, 2, 8, 11, 12, 15, 30, 32, 33, 55, 67–71, 73, 82, 87, 88, 91, 97, 100, 101, 105, 114, 119, 121, 123 EPP see Features F Familiarity, 78, 84, 89, 91, 96, 108 Features Contrastive topic, 33–34, 107–108, 115 EPP, 8, 9, 14, 18, 19, 20–22, 24, 26, 34, 35, 107, 108, 111, 113, 114 Interpretable, 23–24, 34, 107 Topic, 107–108 Uninterpretable, 9, 23–25, 34, 107–109 Wh features, 35–37 Féry, C., 127 Finnish, 69 Focus Discourse function, 16–17, 68–71, 73–81 Syntax, 20, 27, 36 Value, 70, 71, 76, 79, 80, 86, 87 French, 11 Frey, W., 42, 43, 65, 68, 70ff, 78, 97, 120, 127 Fried, M., 10, 64 Functional Generative Description, 3–4
Index G Gender, 5 Grammatical, 43, 50–53, 66, 118, 120, 127 Natural, 50–52, 118, 120 German, 2, 3, 9, 31, 39, 41–43, 53, 65, 68, 92, 97, 105, 119, 120, 123, 125–127, 129 Givón, T., 93 Goldberg, L., 14 Gregory, M., 5, 67, 68, 77, 88, 91–93, 96–97 Grohmann, K., 3, 39, 41–43, 53, 68, 97, 109, 120, 125, 126–127, 131 Gundel, J., 67, 88, 119 H Hajičová, E., 3–4, 68, 69, 71–73, 77, 78ff Hebrew, 110–112 Hungarian, 69 I Implicatures Conventional meaning, 32, 79–81, 86–88 Conversational, 79–81, 86–88 Inflectional domain, 20, 23–27, 47, 129 Informativity, 74, 75, 81 Island sensitivity, 23, 56, 58–60 Complex NP, 23, 30, 59 Wh island, 23, 25, 30, 58, 59 Italian, 2 J Jackendoff, R., 68–69, 77, 100 Japanese, 119 Junghanns, U., 16 K Koopman, H., 116 Kresin, S., 119 L Lambrecht, K., 68, 84, 91, 106 Landau, I., 5, 99, 110–113, 114 Landman, M., 123 Lasnik, H., 30–32 Lenertová, D., 4, 7, 10, 16, 24, 34
Long distance dependencies, 23, 27, 29, 30, 56 ‘Loose Aboutness’ Left Dislocation 64ff M McCloskey, J., 14, 28ff, 63, 116 Merge, 21, 108 Remerge, 9, 25 Michaelis, L., 5, 67, 68, 78, 88, 91–93, 96, 97 Middlefield, 9, 11, 13, 16, 18, 26, 31, 38, 42, 65, 114 Mikkelsen, L., 121 Minimalist Program, 23, 110 Modals, 12–13, 14–16 Movement chain see Chain formation Movement copies, 112–114, 116, 118, 120, 125
Null, 10, 96, 119 Personal, 29, 39ff, 42, 118–120 Possessive, 28–29, 45–49 Reflexive, 10, 28–29, 45–50, 53–55 Resumptive, 100–105, 118–119, 120–125 Q Quantifier binding, 29, 45–48 Question Under Discussion, 75
O Operators, 23–24, 27, 30, 32–34, 107, 115 see Chain formation Open proposition, 32, 69–71, 81, 90, 95
R Reconstruction, 3, 29, 31, 44–50, 53–55 Reflexive pronouns see Pronouns Reinhart, T., 23, 67, 88 Relevance, 74–75 Resumption, 1–3, 39–41, 43–44, 62, 64–65, 109–120, 125–127 see Pronouns, Resumptive Richards, N., 25–26, 30, 31ff Riemsdijk, H., 2, 41, 64ff, 99, 129 Rizzi, L., 4, 7, 30, 126 Roberts, C., 68, 69, 75 Romanian, 25 Rooth, M., 76 Ross, J., 2, 3, 99, 131 Rudin, C., 25–26 Russian, 119, 126
P P-recoverability, 112–113, 115, 116 Partial answers, 73–75, 84, 85 Pesetsky, D., 109, 110 PF, 109, 110–116 Phi-features, 117–118, 125, 127 Polish, 25, 123 Possessive pronouns see Pronouns Praat, 101 Prince, E., 70–71, 77, 78ff, 91, 106 Prolific domain, 125–127 ‘Pronounce highest’ 114 Pronouns, 118–120 Bound, 29, 31–32, 47–48 Clitics, 10–11, 15ff, 35–37, 40-41, 60ff, 104 Demonstrative, 39, 65, 118–121 Logophoric, 26, 45, 46
S Scrambling, 9, 31, 36, 114 A-scrambling, 31ff Semantic type, 121 Serbo-Croatian, 25, 30 Sets Alternative set, 71, 75–77, 78 Cognitively salient set, 67, 68, 70–71, 77–78, 84, 89, 96, 106 Sgall, P., 3–4, 22, 56ff, 68, 69, 71–73, 77, 78ff Sortal key, 77–78, 82, 84, 85 Spanish, 15ff, 119 Spell out, 99, 109, 110–114, 116–118, 120, 121, 125, 126–128, 131 Partial spell out, 116–117 Stalnaker, R., 75 Steedman, M., 4, 68, 71
N Null Complement Anaphora, 15ff Nunes, J., 109
Index Stowell, T., 30–32 Superiority, 25–26 T Tectogrammatical trees, 3 Topic Continuing, 93 Persistance, 93–95, 119 Promotion, 67–68, 88–96, 105–107 Topic Focus Articulation, 3, 71, 101
Topicalization, 17, 18, 21ff, 27–35, 68–73, 84, 97, 100, 105–107, 110–113 V Vat, J., 3, 41, 42, 99 Verb raising, 22 Veselá, K., 3, 17, 100–101, 104 Veselovská, L., 10, 11, 12ff, 21ff, 83ff vP domain, 7–8, 11–16, 25–27, 31–32, 35–37
VP ellipsis, 14–16 VP fronting, 110–112 W Ward, G., 68, 77, 91 Weak Crossover, 30–32 Wh-Movement, 23–27, 30–32, 34–37, 57, 114 Z Zwart, J., 120 Zwarts, F., 2, 41, 99, 129
Linguistik Aktuell/Linguistics Today A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 140 Roehrs, Dorian: Demonstratives and Definite Articles as Nominal Auxiliaries. Expected March 2009 139 Hicks, Glyn: The Derivation of Anaphoric Relations. xii, 306 pp. + index. Expected February 2009 138 Siddiqi, Daniel: Syntax within the Word. Economy, allomorphy, and argument selection in distributed morphology. xii, 136 pp. + index. Expected February 2009 137 Pfau, Roland: Grammar as Processor. A Distributed Morphology account of spontaneous speech errors. xiii, 368 pp. + index. Expected January 2009 136 Kandybowicz, Jason: The Grammar of Repetition. Nupe grammar at the syntax–phonology interface. 2008. xiii, 168 pp. 135 Lewis, William D., Simin Karimi, Heidi Harley and Scott O. Farrar (eds.): Time and Again. Theoretical Perspectives on Formal Linguistics. In honor of D. Terrence Langendoen. xiv, 261 pp. + index. Expected December 2008 134 Armon-Lotem, Sharon, Gabi Danon and Susan Rothstein (eds.): Current Issues in Generative Hebrew Linguistics. vi, 388 pp. + index. Expected December 2008 133 MacDonald, Jonathan E.: The Syntactic Nature of Inner Aspect. A minimalist perspective. xv, 234 pp. + index. Expected November 2008 132 Biberauer, Theresa (ed.): The Limits of Syntactic Variation. 2008. vii, 521 pp. 131 De Cat, Cécile and Katherine Demuth (eds.): The Bantu–Romance Connection. A comparative investigation of verbal agreement, DPs, and information structure. 2008. xix, 355 pp. 130 Kallulli, Dalina and Liliane Tasmowski (eds.): Clitic Doubling in the Balkan Languages. 2008. ix, 442 pp. 129 Sturgeon, Anne: The Left Periphery. The interaction of syntax, pragmatics and prosody in Czech. 2008. xi, 143 pp. 128 Taleghani, Azita H.: Modality, Aspect and Negation in Persian. 2008. ix, 183 pp. 127 Durrleman-Tame, Stephanie: The Syntax of Jamaican Creole. A cartographic perspective. 2008. xii, 190 pp. 126 Schäfer, Florian: The Syntax of (Anti-)Causatives. External arguments in change-of-state contexts. 2008. xi, 324 pp. 125 Rothstein, Björn: The Perfect Time Span. On the present perfect in German, Swedish and English. 2008. xi, 171 pp. 124 Ihsane, Tabea: The Layered DP. Form and meaning of French indefinites. 2008. ix, 260 pp. 123 Stoyanova, Marina: Unique Focus. Languages without multiple wh-questions. 2008. xi, 184 pp. 122 Oosterhof, Albert: The Semantics of Generics in Dutch and Related Languages. 2008. xviii, 286 pp. 121 Tungseth, Mai Ellin: Verbal Prepositions and Argument Structure. Path, place and possession in Norwegian. 2008. ix, 187 pp. 120 Asbury, Anna, Jakub Dotlačil, Berit Gehrke and Rick Nouwen (eds.): Syntax and Semantics of Spatial P. 2008. vi, 416 pp. 119 Fortuny, Jordi: The Emergence of Order in Syntax. 2008. viii, 211 pp. 118 Jäger, Agnes: History of German Negation. 2008. ix, 350 pp. 117 Haugen, Jason D.: Morphology at the Interfaces. Reduplication and Noun Incorporation in Uto-Aztecan. 2008. xv, 257 pp. 116 Endo, Yoshio: Locality and Information Structure. A cartographic approach to Japanese. 2007. x, 235 pp. 115 Putnam, Michael T.: Scrambling and the Survive Principle. 2007. x, 216 pp. 114 Lee-Schoenfeld, Vera: Beyond Coherence. The syntax of opacity in German. 2007. viii, 206 pp. 113 Eythórsson, Thórhallur (ed.): Grammatical Change and Linguistic Theory. The Rosendal papers. 2008. vi, 441 pp. 112 Axel, Katrin: Studies on Old High German Syntax. Left sentence periphery, verb placement and verbsecond. 2007. xii, 364 pp. 111 Eguren, Luis and Olga Fernández Soriano (eds.): Coreference, Modality, and Focus. Studies on the syntax–semantics interface. 2007. xii, 239 pp.
110 Rothstein, Susan (ed.): Theoretical and Crosslinguistic Approaches to the Semantics of Aspect. 2008. viii, 453 pp. 109 Chocano, Gema: Narrow Syntax and Phonological Form. Scrambling in the Germanic languages. 2007. x, 333 pp. 108 Reuland, Eric, Tanmoy Bhattacharya and Giorgos Spathas (eds.): Argument Structure. 2007. xviii, 243 pp. 107 Corver, Norbert and Jairo Nunes (eds.): The Copy Theory of Movement. 2007. vi, 388 pp. 106 Dehé, Nicole and Yordanka Kavalova (eds.): Parentheticals. 2007. xii, 314 pp. 105 Haumann, Dagmar: Adverb Licensing and Clause Structure in English. 2007. ix, 438 pp. 104 Jeong, Youngmi: Applicatives. Structure and interpretation from a minimalist perspective. 2007. vii, 144 pp. 103 Wurff, Wim van der (ed.): Imperative Clauses in Generative Grammar. Studies in honour of Frits Beukema. 2007. viii, 352 pp. 102 Bayer, Josef, Tanmoy Bhattacharya and M.T. Hany Babu (eds.): Linguistic Theory and South Asian Languages. Essays in honour of K. A. Jayaseelan. 2007. x, 282 pp. 101 Karimi, Simin, Vida Samiian and Wendy K. Wilkins (eds.): Phrasal and Clausal Architecture. Syntactic derivation and interpretation. In honor of Joseph E. Emonds. 2007. vi, 424 pp. 100 Schwabe, Kerstin and Susanne Winkler (eds.): On Information Structure, Meaning and Form. Generalizations across languages. 2007. vii, 570 pp. 99 Martínez-Gil, Fernando and Sonia Colina (eds.): Optimality-Theoretic Studies in Spanish Phonology. 2007. viii, 564 pp. 98 Pires, Acrisio: The Minimalist Syntax of Defective Domains. Gerunds and infinitives. 2006. xiv, 188 pp. 97 Hartmann, Jutta M. and László Molnárfi (eds.): Comparative Studies in Germanic Syntax. From Afrikaans to Zurich German. 2006. vi, 332 pp. 96 Lyngfelt, Benjamin and Torgrim Solstad (eds.): Demoting the Agent. Passive, middle and other voice phenomena. 2006. x, 333 pp. 95 Vogeleer, Svetlana and Liliane Tasmowski (eds.): Non-definiteness and Plurality. 2006. vi, 358 pp. 94 Arche, María J.: Individuals in Time. Tense, aspect and the individual/stage distinction. 2006. xiv, 281 pp. 93 Progovac, Ljiljana, Kate Paesani, Eugenia Casielles and Ellen Barton (eds.): The Syntax of Nonsententials. Multidisciplinary perspectives. 2006. x, 372 pp. 92 Boeckx, Cedric (ed.): Agreement Systems. 2006. ix, 346 pp. 91 Boeckx, Cedric (ed.): Minimalist Essays. 2006. xvi, 399 pp. 90 Dalmi, Gréte: The Role of Agreement in Non-Finite Predication. 2005. xvi, 222 pp. 89 Velde, John R. te: Deriving Coordinate Symmetries. A phase-based approach integrating Select, Merge, Copy and Match. 2006. x, 385 pp. 88 Mohr, Sabine: Clausal Architecture and Subject Positions. Impersonal constructions in the Germanic languages. 2005. viii, 207 pp. 87 Julien, Marit: Nominal Phrases from a Scandinavian Perspective. 2005. xvi, 348 pp. 86 Costa, João and Maria Cristina Figueiredo Silva (eds.): Studies on Agreement. 2006. vi, 285 pp. 85 Mikkelsen, Line: Copular Clauses. Specification, predication and equation. 2005. viii, 210 pp. 84 Pafel, Jürgen: Quantifier Scope in German. 2006. xvi, 312 pp. 83 Schweikert, Walter: The Order of Prepositional Phrases in the Structure of the Clause. 2005. xii, 338 pp. 82 Quinn, Heidi: The Distribution of Pronoun Case Forms in English. 2005. xii, 409 pp. 81 FuSS, Eric: The Rise of Agreement. A formal approach to the syntax and grammaticalization of verbal inflection. 2005. xii, 336 pp. 80 Burkhardt, Petra: The Syntax–Discourse Interface. Representing and interpreting dependency. 2005. xii, 259 pp. 79 Schmid, Tanja: Infinitival Syntax. Infinitivus Pro Participio as a repair strategy. 2005. xiv, 251 pp. 78 Dikken, Marcel den and Christina Tortora (eds.): The Function of Function Words and Functional Categories. 2005. vii, 292 pp. 77 Öztürk, Balkız: Case, Referentiality and Phrase Structure. 2005. x, 268 pp. 76 Stavrou, Melita and Arhonto Terzi (eds.): Advances in Greek Generative Syntax. In honor of Dimitra Theophanopoulou-Kontou. 2005. viii, 366 pp.
75 Di Sciullo, Anna Maria (ed.): UG and External Systems. Language, brain and computation. 2005. xviii, 398 pp. 74 Heggie, Lorie and Francisco Ordóñez (eds.): Clitic and Affix Combinations. Theoretical perspectives. 2005. viii, 390 pp. 73 Carnie, Andrew, Heidi Harley and Sheila Ann Dooley (eds.): Verb First. On the syntax of verbinitial languages. 2005. xiv, 434 pp. 72 FuSS, Eric and Carola Trips (eds.): Diachronic Clues to Synchronic Grammar. 2004. viii, 228 pp. 71 Gelderen, Elly van: Grammaticalization as Economy. 2004. xvi, 320 pp. 70 Austin, Jennifer R., Stefan Engelberg and Gisa Rauh (eds.): Adverbials. The interplay between meaning, context, and syntactic structure. 2004. x, 346 pp. 69 Kiss, Katalin É. and Henk van Riemsdijk (eds.): Verb Clusters. A study of Hungarian, German and Dutch. 2004. vi, 514 pp. 68 Breul, Carsten: Focus Structure in Generative Grammar. An integrated syntactic, semantic and intonational approach. 2004. x, 432 pp. 67 Mišeska Tomić, Olga (ed.): Balkan Syntax and Semantics. 2004. xvi, 499 pp. 66 Grohmann, Kleanthes K.: Prolific Domains. On the Anti-Locality of movement dependencies. 2003. xvi, 372 pp. 65 Manninen, Satu Helena: Small Phrase Layers. A study of Finnish Manner Adverbials. 2003. xii, 275 pp. 64 Boeckx, Cedric and Kleanthes K. Grohmann (eds.): Multiple Wh-Fronting. 2003. x, 292 pp. 63 Boeckx, Cedric: Islands and Chains. Resumption as stranding. 2003. xii, 224 pp. 62 Carnie, Andrew, Heidi Harley and MaryAnn Willie (eds.): Formal Approaches to Function in Grammar. In honor of Eloise Jelinek. 2003. xii, 378 pp. 61 Schwabe, Kerstin and Susanne Winkler (eds.): The Interfaces. Deriving and interpreting omitted structures. 2003. vi, 403 pp. 60 Trips, Carola: From OV to VO in Early Middle English. 2002. xiv, 359 pp. 59 Dehé, Nicole: Particle Verbs in English. Syntax, information structure and intonation. 2002. xii, 305 pp. 58 Di Sciullo, Anna Maria (ed.): Asymmetry in Grammar. Volume 2: Morphology, phonology, acquisition. 2003. vi, 309 pp. 57 Di Sciullo, Anna Maria (ed.): Asymmetry in Grammar. Volume 1: Syntax and semantics. 2003. vi, 405 pp. 56 Coene, Martine and Yves D’hulst (eds.): From NP to DP. Volume 2: The expression of possession in noun phrases. 2003. x, 295 pp. 55 Coene, Martine and Yves D’hulst (eds.): From NP to DP. Volume 1: The syntax and semantics of noun phrases. 2003. vi, 362 pp. 54 Baptista, Marlyse: The Syntax of Cape Verdean Creole. The Sotavento varieties. 2003. xxii, 294 pp. (incl. CD-rom). 53 Zwart, C. Jan-Wouter and Werner Abraham (eds.): Studies in Comparative Germanic Syntax. Proceedings from the 15th Workshop on Comparative Germanic Syntax (Groningen, May 26–27, 2000). 2002. xiv, 407 pp. 52 Simon, Horst J. and Heike Wiese (eds.): Pronouns – Grammar and Representation. 2002. xii, 294 pp. 51 Gerlach, Birgit: Clitics between Syntax and Lexicon. 2002. xii, 282 pp. 50 Steinbach, Markus: Middle Voice. A comparative study in the syntax-semantics interface of German. 2002. xii, 340 pp. 49 Alexiadou, Artemis (ed.): Theoretical Approaches to Universals. 2002. viii, 319 pp. 48 Alexiadou, Artemis, Elena Anagnostopoulou, Sjef Barbiers and Hans-Martin Gärtner (eds.): Dimensions of Movement. From features to remnants. 2002. vi, 345 pp. 47 Barbiers, Sjef, Frits Beukema and Wim van der Wurff (eds.): Modality and its Interaction with the Verbal System. 2002. x, 290 pp. 46 Panagiotidis, E. Phoevos: Pronouns, Clitics and Empty Nouns. ‘Pronominality’ and licensing in syntax. 2002. x, 214 pp. 45 Abraham, Werner and C. Jan-Wouter Zwart (eds.): Issues in Formal German(ic) Typology. 2002. xviii, 336 pp. 44 Taylan, Eser Erguvanlı (ed.): The Verb in Turkish. 2002. xviii, 267 pp. 43 Featherston, Sam: Empty Categories in Sentence Processing. 2001. xvi, 279 pp.
42 Alexiadou, Artemis: Functional Structure in Nominals. Nominalization and ergativity. 2001. x, 233 pp. 41 Zeller, Jochen: Particle Verbs and Local Domains. 2001. xii, 325 pp. 40 Hoeksema, Jack, Hotze Rullmann, Víctor Sánchez-Valencia and Ton van der Wouden (eds.): Perspectives on Negation and Polarity Items. 2001. xii, 368 pp. 39 Gelderen, Elly van: A History of English Reflexive Pronouns. Person, Self, and Interpretability. 2000. xiv, 279 pp. 38 Meinunger, André: Syntactic Aspects of Topic and Comment. 2000. xii, 247 pp. 37 Lutz, Uli, Gereon Müller and Arnim von Stechow (eds.): Wh-Scope Marking. 2000. vi, 483 pp. 36 Gerlach, Birgit and Janet Grijzenhout (eds.): Clitics in Phonology, Morphology and Syntax. 2001. xii, 441 pp. 35 Hróarsdóttir, Thorbjörg: Word Order Change in Icelandic. From OV to VO. 2001. xiv, 385 pp. 34 Reuland, Eric (ed.): Arguments and Case. Explaining Burzio’s Generalization. 2000. xii, 255 pp. 33 Puskás, Genoveva: Word Order in Hungarian. The syntax of Ā-positions. 2000. xvi, 398 pp. 32 Alexiadou, Artemis, Paul Law, André Meinunger and Chris Wilder (eds.): The Syntax of Relative Clauses. 2000. vi, 397 pp. 31 Svenonius, Peter (ed.): The Derivation of VO and OV. 2000. vi, 372 pp. 30 Beukema, Frits and Marcel den Dikken (eds.): Clitic Phenomena in European Languages. 2000. x, 324 pp. 29 Miyamoto, Tadao: The Light Verb Construction in Japanese. The role of the verbal noun. 2000. xiv, 232 pp. 28 Hermans, Ben and Marc van Oostendorp (eds.): The Derivational Residue in Phonological Optimality Theory. 2000. viii, 322 pp. 27 Růžička, Rudolf: Control in Grammar and Pragmatics. A cross-linguistic study. 1999. x, 206 pp. 26 Ackema, Peter: Issues in Morphosyntax. 1999. viii, 310 pp. 25 Felser, Claudia: Verbal Complement Clauses. A minimalist study of direct perception constructions. 1999. xiv, 278 pp. 24 Rebuschi, Georges and Laurice Tuller (eds.): The Grammar of Focus. 1999. vi, 366 pp. 23 Giannakidou, Anastasia: Polarity Sensitivity as (Non)Veridical Dependency. 1998. xvi, 282 pp. 22 Alexiadou, Artemis and Chris Wilder (eds.): Possessors, Predicates and Movement in the Determiner Phrase. 1998. vi, 388 pp. 21 Klein, Henny: Adverbs of Degree in Dutch and Related Languages. 1998. x, 232 pp. 20 Laenzlinger, Christopher: Comparative Studies in Word Order Variation. Adverbs, pronouns, and clause structure in Romance and Germanic. 1998. x, 371 pp. 19 Josefsson, Gunlög: Minimal Words in a Minimal Syntax. Word formation in Swedish. 1998. ix, 199 pp. 18 Alexiadou, Artemis: Adverb Placement. A case study in antisymmetric syntax. 1997. x, 256 pp. 17 Beermann, Dorothee, David LeBlanc and Henk van Riemsdijk (eds.): Rightward Movement. 1997. vi, 410 pp. 16 Liu, Feng-hsi: Scope and Specificity. 1997. viii, 187 pp. 15 Rohrbacher, Bernhard Wolfgang: Morphology-Driven Syntax. A theory of V to I raising and prodrop. 1999. viii, 296 pp. 14 Anagnostopoulou, Elena, Henk van Riemsdijk and Frans Zwarts (eds.): Materials on Left Dislocation. 1997. viii, 349 pp. 13 Alexiadou, Artemis and T. Alan Hall (eds.): Studies on Universal Grammar and Typological Variation. 1997. viii, 252 pp. 12 Abraham, Werner, Samuel David Epstein, Höskuldur Thráinsson and C. Jan-Wouter Zwart (eds.): Minimal Ideas. Syntactic studies in the minimalist framework. 1996. xii, 364 pp. 11 Lutz, Uli and Jürgen Pafel (eds.): On Extraction and Extraposition in German. 1996. xii, 315 pp. 10 Cinque, Guglielmo and Giuliana Giusti (eds.): Advances in Roumanian Linguistics. 1995. xi, 172 pp. 9 Gelderen, Elly van: The Rise of Functional Categories. 1993. x, 224 pp. 8 Fanselow, Gisbert (ed.): The Parametrization of Universal Grammar. 1993. xvii, 232 pp. 7 Åfarlí, Tor A.: The Syntax of Norwegian Passive Constructions. 1992. xii, 177 pp. 6 Bhatt, Christa, Elisabeth Löbel and Claudia Maria Schmidt (eds.): Syntactic Phrase Structure Phenomena in Noun Phrases and Sentences. 1989. ix, 187 pp. 5 Grewendorf, Günther and Wolfgang Sternefeld (eds.): Scrambling and Barriers. 1990. vi, 442 pp.