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!
Linguistik Aktuell/Linguistics Today 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.
Series Editor Werner Abraham
Elly v. Gelderen
University of Vienna
Arizona State University
Advisory Editorial Board Guglielmo Cinque
Ian Roberts
University of Venice
Cambridge University
Günther Grewendorf
Ken Safir
J.W. Goethe-University, Frankfurt
Rutgers University, New Brunswick NJ
Liliane Haegeman
Lisa deMena Travis
University of Lille, France
McGill University
Hubert Haider
Sten Vikner
University of Salzburg
University of Aarhus
Christer Platzack
C. Jan-Wouter Zwart
University of Lund
University of Groningen
Volume 64 Multiple Wh-Fronting Edited by Cedric Boeckx and Kleanthes K. Grohmann
Multiple Wh-Fronting Edited by
Cedric Boeckx Harvard University
Kleanthes K. Grohmann University of Cologne
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 Multiple Wh-fronting / edited by Cedric Boeckx and Kleanthes K. Grohmann. p. cm. (Linguistik Aktuell/Linguistics Today, issn 0166–0829 ; v. 64) Includes bibliographical references and index. 1. Extraction (Linguistics) I. Boeckx, Cedric. II. Grohmann, Kleanthes K. III. Linguistik aktuell ; Bd. 64. P158.17.M85 2003 415-dc21 isbn 90 272 2787 X (Eur.) / 1 58811 419 8 (US) (Hb; alk. paper)
We would like to thank the LA/LT-Series Editors, Werner Abraham and Elly van Gelderen, as well as Kees Vaes, the Editorial Director, for their interest in and support of this project. Thanks also to Elly van Gelderen for her detailed comments on the final version of the manuscript. We are indebted to Željko Boškovic´ and Norbert Hornstein for spurring our interest in multiple wh-fronting phenomena when we were students and, through their works, making sure that we remain interested. We also extend thanks to our wives, Youngmi and Joy, for providing much needed moral support. Last, but not least, we are most grateful to our contributors for their willingness to join in this project and for their patience during the editorial process.
List of contributors
Cedric Boeckx Harvard University Department of Linguistics Boylston Hall Cambridge, MA 02138 USA
Kleanthes K. Grohmann Universität zu Köln Englisches Seminar Albertus-Magnus-Platz 50923 Köln Germany
cboeckx@fas.harvard.edu
kleanthes@punksinscience.org
Željko Boškovic´ University of Connecticut Department of Linguistics U-1145 Storrs, CT 06269 USA
Youngmi Jeong University of Maryland Department of Linguistics 1401 Marie Mount Hall College Park, MD 20742 USA
boskovic@sp.uconn.edu
yjeong@wam.umd.edu
Molly Diesing Cornell University Department of Linguistics Morrill Hall Ithaca, NY 14853 USA
Anikó Lipták Leiden University ATW/ULCL P. O. Box 9515 2300 RA Leiden The Netherlands
md20@cornell.edu
a.liptak@let.leidenuniv.nl
Marcel den Dikken CUNY Graduate Center Linguistics Program 365 Fifth Avenue New York NY 10016 USA
Ahmad Lotfi American University of Hawaii & Azad University at Esfahan Khorasgan Esfahan Iran
mden-dikken@gc.cuny.edu
arlotfi@yahoo.com
Lara Reglero University of Connecticut Department of Linguistics U-1145 Storrs, CT 06269 USA
Sandra Stjepanovic´ University of West Virginia Department of Foreign Languages P. O. Box 6298 Morgantown, WV 26506 USA
laregle@hotmail.com
sastjepanovic@mail.wvu.edu
Joachim Sabel Université catholique de Louvain Département d’études germaniques Place Blaise Pascal 1 1348 Louvain-la-Neuve Belgium sabel@lige.ucl.ac.be
Introduction Cedric Boeckx and Kleanthes K. Grohmann Harvard University / University of Cologne
This volume explores the formation of multiple wh-questions across languages, focussing on the strategy of fronting wh-elements. All contributions examine these fronting operations from a cross-linguistic point of view, either by concentrating on a particular language or by comparing the behaviour of wh-fronting in more than one language. They consider the syntax, semantics, and phonology involved in multiple wh-questions, and as these three areas of language go often hand in hand with respect to the possibility of moving wh-elements, we are dealing with an interface phenomenon par excellence. Against the background of the Minimalist Program (Chomsky 1995, 2000, 2001a) that takes Logical Form (LF) and Phonological Form (PF) to be the sole levels of representation, the study of the LF-/PF-interfaces with the computational system becomes crucial. From a universal perspective, the semantics of (multiple) questions should be invariable. How come, then, that the syntax is highly variable, both across languages (e.g., Bulgarian [front all Wh’s] vs. English [front one Wh] vs. Chinese [front no Wh]) and within languages (such as French, where English- and Chinese-style patterns are combined in matrix clauses)? What is the motivation for movement of a wh-phrase in the first place, and how does this tie in with the variation observed? These questions will be addressed in detail in the chapters that follow. The following paragraphs offer a brief overview of the research on multiple wh-fronting, concentrating on wh-movement in paradigmatic languages like Bulgarian and Serbo-Croatian. Extensions and refinements of the patterns observed in those languages are a major ingredient of the contributions to this volume.
2
Cedric Boeckx and Kleanthes K. Grohmann
1.
A brief overview
Seminal work by Wachowicz (1974), Toman (1981), and especially Rudin (1988) brought to light instances of wh-movement that were to become of crucial theoretical importance in the Principles-and-Parameters framework, especially in the currently explored Minimalist Program (Chomsky 1995, 2000, 2001a). Work on wh-in situ in Chinese and Japanese (Huang 1982a,b, Lasnik and Saito 1984) had proven central in revealing the properties of the covert side of the computational component. More particularly, (1a–b) were shown to be two different surface manifestations of the same underlying universal principle that demands the creation of an operator-variable chain. (1) a. Who did you see? b. Ni kanijan-le shei? you see-asp who ‘Who did you see?’
(Chinese)
At first sight, Bulgarian (2) and English seem to be identical as far as wh-movement is concerned: the process takes place overtly. (2) a.
Kogo vižda John? who sees John ‘Who does John see?’ b. *John vižda kogo?
(Bulgarian)
However, when we turn to expressions containing multiple questions, a crucial difference emerges. All wh-phrases must be fronted in Bulgarian, whereas only one can move overtly in English. Witness the contrast between (3) and (4). (3) a.
Koj kogo vidjal? who whom saw ‘Who saw whom?’ b. *Koj vidjal kogo?
(Bulgarian)
(4) a. Who saw whom? b. *Who whom saw?
In-situ languages like Japanese allow multiple questions, with all wh-phrases insitu (but see Jeong’s contribution for a different view). (5) Taroo-ga dare-ni nani-o ageta no? Taroo-nom who-dat what-acc gave q ‘Who did Taroo give what?’
(Japanese)
Introduction
(3)–(5) reveal an interesting parallelism between Bulgarian-type and Japanesetype languages, viz. all wh-phrases behave alike: if one fronts, all front and if one remains in-situ, all remain in-situ. English thus appears to be ‘special’ from this point of view. Work on wh-in situ (both cross-linguistic investigations and semantic analyses of multiple wh-questions in English such as Higginbotham and May’s seminal 1981 paper) had led to the conclusion (already hinted at in Chomsky 1973) that all wh-words move to Comp by LF. While it was assumed that at most one wh-expression moved overtly, the Slavic data point to the fact that the requirement that all wh-phrases be in C may actually be fulfilled by Surface Structure (SS)/Spell-Out. This led Pesetsky (1987: 117) to claim that a wh-question in Slavic “seems to wear its LF [representation] on its sleeves”. We believe the Slavic data would not have attracted so much attention if (3) were the only peculiarity. A parameter forcing overt movement of all wh-phrases would have been very easy to set up. But as Rudin (1988) convincingly shows, this is not the whole story. If it is true that Slavic language are ‘Multiple Fronting’ languages (i.e. all wh-phrases must front overtly), there are crucial differences among them. Among the most salient: Bulgarian appears to impose a rigid order among wh-phrases; roughly, the order must reflect the underlying (‘Deep Structure’) hierarchy, as shown in (6). In Serbo-Croatian, the surface word order is free, as (7) illustrates. (Notice the presence of the auxiliary in between the two wh-words in Serbo-Croatian to which we will return presently.) (6) a.
Koj kogo vidjal? who whom saw ‘Who saw whom?’ b. *Kogo koj vidjal? c. Kogo kakvo e pital Ivan? who what is asked Ivan ‘Who did Ivan asked what?’ d. *Kakvo kogo e pital Ivan?
(Bulgarian)
(7) a.
(Serbo-Croatian)
Ko je koga vidio? who is whom seen ‘Who saw whom?’ b. Koga je ko vidio? c. Šta je kome Ivan dao? what is whom Ivan given ‘What did Ivan give to whom?’ d. Kome je šta Ivan dao?
3
4
Cedric Boeckx and Kleanthes K. Grohmann
Rudin concluded form (6)–(7) that only Bulgarian wh-movement obeys the Superiority Condition (8), from Chomsky (1973: 246). A classic example of Superiority is given in (9). (8) Superiority Condition a. No rule can involve X,Y in the structure … X … [… Z … WYV …] … where the rule applies ambiguously to Z and Y, and Z is superior to Y. b. The category A is superior to the category B if every major category dominating A dominates B as well but not conversely. (9) a. Who bought what? b. *What did who buy?
From this, Rudin claimed that wh-movement in Bulgarian was uniform movement to Comp (SpecCP): she assumed that the first wh-phrase moves to SpecCP (substitution), subsequent wh-movement is (right) adjunction to SpecCP, whereas wh-movement in the other multiple fronting languages consists in moving the first wh-phrase to SpecCP, and adjoining all the other wh-phrases to IP. This distinction enabled Rudin to account for other differences between the two language groups. For instance, she shows that the CP-IP split straightforwardly derives the fact that in long distance wh-movement (movement from Comp to Comp), all wh-phrases move to the highest Comp in Bulgarian, while only one wh-phrase moves in the other language group. The CP-IP divide also accounts for the impossibility of inserting overt material between the first and the second wh-phrase in Bulgarian, while it is possible in an ‘IP-language’ like Serbo-Croatian (wh-movement targeting two projections in this case). Empirical evidence comes from clitic insertion, insertion of adverbs and parentheticals, and other domains (but see Boškovic´’s recent work for a reconsideration of the evidence, such as Boškovic´ 2002, this volume). We illustrate the difference here with parentheticals. (10) Koj, spored tebe, kakvo pie? who according to.you what drinks ‘Who, according to you, drinks what?’
(Bulgarian)
(11) Koj, po tebi, šta pije? who for you what drinks ‘Who, according to you, drinks what?’
(Serbo-Croatian)
The facts just reviewed seem to point to major differences between Bulgarian and Serbo-Croatian. However, Boškovic´ (1997 et seq.) has shown that Rudin’s characterization is both too weak and too strong. First, Superiority effects
Introduction
disappear in Bulgarian once more than two wh-phrases are fronted. In this case, the first instance of movement obeys Superiority, but all subsequent movements are free from it. Consider (12). (12) a.
b. c. d. e. f.
Koj kogo kakvo e pital? who whom what is asked ‘Who asked whom what?’ Koj kakvo kogo e pital? *Kogo kakvo koj e pital? *Kakvo kogo koj e pital? *Kakvo koj kogo e pital? *Kogo koj kakvo e pital?
Second, Boškovic´ has convincingly argued that Serbo-Croatian and languages patterning like it are subject to the Superiority Condition in well-defined contexts. Thus, even though Superiority seems to be inoperative in shortdistance wh-movement (wh-movement within one clause), it clearly constrains long distance wh-movement (movement crossing more than one C-projection). Witness (13): (13) a.
Ko si koga turdio da je istukao? who are whom claimed that is beaten ‘Who did you claim beat whom?’ b. *Koga si ko turdio da je istukao?
(Serbo-Croatian)
Having clarified the empirical picture, Boškovic´ suggests that wh-movement in Bulgarian is an instance of substitution into SpecCP accompanied by right adjunction to CP (cf. Rudin 1988). As for Serbo-Croatian (and languages patterning like it), Boškovic´ contends that wh-movement is adjunction to a focus position below CP, which explains the absence of Superiority effects. In those cases where Superiority is operative, he assumes that one wh-phrase moves into SpecCP. The evidence he provides in favor of this position is twofold (see Boškovic´’s own work for extensive discussion). First, Boškovic´ notes that Superiority effects are detectable in Serbo-Croatian in those cases where French disallows wh-in situ. As is well known, French exhibits a mixed pattern. Wh-phrases can optionally remain in situ in ‘matrix’ questions. In ‘long distance’ cases, the in-situ strategy is no longer available. This leads Boškovic´ (1997, 2002) to conclude that the C-projection can optionally be present in ‘matrix’ questions (Boškovic´ 2002 shows how this option follows from the assumptions developed in Chomsky 1995). When C is projected, wh-movement
5
6
Cedric Boeckx and Kleanthes K. Grohmann
must take place; in the long-distance cases, C is obligatorily projected, it follows that wh-movement must take place. Given that, Boškovic´ claims that in the cases where Superiority is active in Serbo-Croatian, C is targeted for movement. In ‘short’/matrix contexts, C is optionally targeted, which allows for the free order among wh-phrases. The second piece of evidence Boškovic´ adduces in support of his analysis comes from the facts discussed by Stjepanovic´ (1998), who shows that SerboCroatian possesses a distinct structural position for (contrastively) focused phrases. It is a long-standing claim in the literature that wh-phrases are inherently focused. Therefore, it is natural to conclude that if wh-phrases front, but do not move to C, they are hosted by this focus position. Interestingly, Stjepanovic´ shows that focused non-wh-phrases cannot remain in situ in SerboCroatian, which points to the fact that the relevant (focus) feature is strong. In Serbo-Croatian, focusing is not restricted to one element. All focused phrases must move. This seems to provide a straightforward explanation for why all wh-phrases have to move overtly in the language. To explain the difference between Serbo-Croatian and Bulgarian, Boškovic´ conjectures that Bulgarian has no focus projection distinct from C, which accounts for obligatory movement to C in the language. This accounts has many appealing aspects. For one thing, it makes Serbo-Croatian (and languages patterning like it) well-behaved with respect to Superiority. For another, it derives the Bulgarian/Serbo-Croatian contrast from independent processes in the languages. Boškovic´’s characterization has received much support, and several papers included in the present volume (in particular Stjepanovic´’s) further strengthen his original insight. However, Boškovic´’s account is not without problems. To mention but a few: it is a well-established fact in the literature that languages allow for only one focus per clause (see the collection of papers in É. Kiss 1995 for review, also É. Kiss 1998 for a discussion of the relevant notion of ‘focus’) — so how is this reconcilable with multiple wh-fronting/focusing? Next, Rudin (1988) considers, but rejects, the focus movement analysis on the basis of Bulgarian. She shows that Bulgarian too has a specific, pre-CP position for non-wh-focused phrases. Under Boškovic´’s account, we would predict that wh-movement would not target C in Bulgarian, which is clearly not the case (note that Boškovic´’s claim that C is the focus position in Bulgarian is unmotivated). Also, it is worth mentioning that Hungarian, a multiple wh-fronting language which patterns like Serbo-Croatian in not targeting C, behaves differently as far as nonwh-focused phrases are concerned. Thus, Brody (1990) shows that it is possible
Introduction
to focus more than one element in Hungarian, but it is impossible to front them all. Rather, only one can be fronted, other foci remaining in situ.
2. Multiple wh-fronting and minimalism Although the basic typology of multiple wh-fronting was achieved in the mideighties in work by Catherine Rudin (Rudin 1985, 1988), the theoretical interest in the topic really grew out of a series of work in the mid-nineties that adopted a minimalist approach. (For a useful summary of the possible approaches within an early GB-framework that rested on the Comp Substitution Universal — the assumption that only one element can occupy the Comp position; cf. Bresnan’s (1970) Complementizer Attraction Universal — see Toman 1981.) We see at least four reasons for the marriage of multiple wh-fronting and minimalism to have been a successful one. First, the latent (Anti-)Superiority effects discovered by Boškovic´ (1998a, 2001) has helped sharpen the economy flavor of Chomsky’s original condition, in addition to offering a privileged perspective on the driving force for wh-movement, well summarized in Stjepanoviç’s contribution to this volume. In addition, the investigation of multiple specifiers (Koizumi 1995, Richards 1997) appears to be tailored for multiple wh-fronting (but see Grohmann 2000, this volume for contrary views). It is no surprise, therefore, that the technology used for multiple wh-fronting gave rise to important discoveries in other contexts, such as Nissenbaum’s (2000) study of multiple parasitic gaps, or Brown’s (in press) study of negative concord (multiple negation), and perhaps even instances of multiple agreement, as Boeckx suggests in his contribution. Second, the data from multiple wh-languages have enabled us to locate precisely the locus of featural illegitimacy giving rise to multiple wh-fronting, and the parameter(s) that underlie the phenomenon. A very good example of this is den Dikken’s analysis included in this work. Third, recent developments in the study of the Comp-area (more generally, the ‘cartographic’ approach to clausal architecture), proposing an elaborate CP-field (Rizzi 1997; see also the collection of articles in Belletti 2003, Rizzi 2003), should also connect in interesting ways with the general phenomenon of multiple wh-fronting. This is particularly clear in cases where it has been claimed that discourse notions such as focus are the driving force for multiple (wh-)movement — and in some instances even topic. Grohmann, Jeong, and Reglero explore such connections in their papers.
7
8
Cedric Boeckx and Kleanthes K. Grohmann
Fourth, multiple wh-fronting should prove an ideal testing ground regarding the nature of the interfaces. Recently, work by Boškovic´ (2001, 2002) suggests that restrictions on multiple wh-fronting provide important insights into the organization of the Phonological Component, in particular providing evidence for a PF-oriented mechanism of copy-pronunciation. Further investigations of interface issues should help us test the idea first formulated in Calabrese (1983) that the absence of multiple questions in some languages (such as Italian) is due to PF factors (especially, prosodic structure), something that Grohmann picks up in his contribution as well. At the LF interface, we ought to mention the influential proposal that takes Superiority as an interpretive condition. This was first proposed by Chierchia (1991, 1993), and is explored in Comorovski (1996), Hornstein (1995), Williams (1994, 2003), and also in Haider (2000). The essence of this view of Superiority is that pair-list readings in multiple interrogatives are mediated by (i.e. arise as a subspecies of) functional reading of the wh-phrase. This line of research is motivated by the fact that although Superiority has been seen as but one instance of a more general genus, viz. Relativized Minimality effects, it is surprising that it holds only in the case of multiple questions. To the best of our knowledge, no other instance of multiplicity in the A¢ domain, such as multiple topics or multiple (non-wh) foci appear to be subject to it. So far as we can tell, this does not follow in a non-stipulative way from the current understanding of Superiority as the result of derivational economy of the type we outlined above. Sadly, we could not incorporate any contribution defending this view in the present volume on time, but we urge the reader to consult Haider’s paper in Lutz, Müller and von Stechow (2000), which we suggest below is a companion volume to the present one.
3. Organization The present volume deals with a variety of topics. Several authors deal with the formalization of wh-fronting proper (Boeckx, den Dikken, Diesing, Grohmann, Jeong). Others present in-depth analyses of multiple wh-questions in the “traditional” multiple wh-fronting languages (Boškovic´, Stjepanovic´) or extend current views to less studied languages (Diesing, Lipták, Lotfi, Reglero, Sabel) and, in the case of Grohmann, even to German, yielding novel characterizations of a typology on the placement of wh-phrases. It should be clear that all those papers are interrelated, and their contributions are best appreciated when put together.
Introduction
The present volume may be seen as a companion to Lutz, Müller, and von Stechow (2000). While the latter focuses on the presence of ‘expletive’ wh-phrases alongside genuine wh-phrases (“wh-scope marking”) and examines their fronting strategies, the present study concentrates on the intricacies of multiple fronting of genuine wh-phrases. Since the collection of papers contained therein and the contributions to the present volume deal explicitly with the respective phenomena in a number of different, typologically unrelated languages, both perspectives are bound to help refine our understanding of the typology of wh-questions. To close this introduction, we provide brief summaries of the papers contained herein. 3.1 Boeckx: “Symmetries and asymmetries in multiple checking” This paper seeks to refine the formal aspect of multiple wh-fronting by refining Multiple Agree, recently advocated by Hiraiwa (2001, 2002) and Chomsky (2001b). Boeckx shows that an investigation of the minutiae of Agree can lead to a deeper understanding of symmetries and asymmetries in instances of Multiple Agree such as multiple wh-fronting, without abandoning important guiding ideas of current syntactic theorizing such as local economy, intervention effects, parallelism among features, and so on. Crucial is the distinction between Match and Agree, and the idea that the computational system is derivational (which allows for simultaneous operations, operations taking precedence over others etc.). 3.2 Boškovic´: “On wh-islands and obligatory wh-movement contexts in South Slavic” Željko Boškovic´ revisits Rudin’s (1988) claim that multiple wh-movement languages like Bulgarian allow extraction of a wh-phrase out of wh-islands, while those that pattern differently with respect to the landing site of fronted wh-phrases, such as Serbo-Croatian, do not. Taking into account extraction of wh-adjuncts, Boškovic´ shows that this description is not quite accurate. He then presents an extended set of data and concludes, in conformity with his earlier work (cf. Boškovic´ 1997–2002 cited in the references below), that there are two patterns of multiple wh-fronting, one where all wh-phrases are located in SpecCP (Bulgarian) and one where at most one wh-phrase sits up there (Serbo-Croatian).
9
10
Cedric Boeckx and Kleanthes K. Grohmann
3.3 Den Dikken: “On the morphosyntax of wh-movement” The central aim of Marcel den Dikken’s paper is to develop a general theory of wh-constructions which (i) allows us to accurately distinguish between wh-phrases as question-word phrases, echo-question phrases and indefinites in terms of a feature matrix involving the features [±Wh] and [±Focus], (ii) makes it understandable why wh-fronting targets SpecCP, SpecTopP or SpecFocP depending on context, (iii) accounts for root/non-root asymmetries with respect to the landing-site of wh-fronting, and (iv) explains the differences and similarities between [+Wh] foci and [−Wh] foci. Although den Dikken does not directly deal with patterns of multiple wh-fronting, his featural analysis of wh-phrases is bound to serve as an important tool in the domain of multiple wh-questions. 3.4 Diesing: “On the nature of multiple fronting in Yiddish” In her contribution, Molly Diesing concentrates on the issues of Superiority and landing sites in multiple wh-fronting in Yiddish. Yiddish is not only the only Germanic language which permits multiple wh-fronting (though see Grohmann’s paper for a different analysis of the situation in German), it is also a language which allows two strategies for forming multiple questions: multiple fronting and single fronting. Diesing shows that these two options differ in their syntactic properties, most notably regarding Superiority effects, which seem to be linked to multiple fronting only (see also Hoge 2000). She aims to provide a syntactic explanation of these facts by assuming that some sort of multiple wh-fronting takes place. Diesing also investigates dialectal variation with respect to multiple fronting and Yiddish. She shows that multiple wh-fronting in Yiddish is a genuine instance of multiple wh-fronting (viz. movement of all wh-phrases to CP). At the same time, Yiddish also provides evidence for fronting of a wh-phrase to a position below CP in some instances of single fronting. 3.5 Grohmann: “German is a multiple wh-fronting language!” Kleanthes Grohmann revisits Boškovic´’s (1998b) study on the availability of single pair/pair list readings in multiple wh-questions with special attention paid to how German might fit into the picture. It turns out that despite initial appearances, German behaves very much like the “traditional” multiple wh-fronting languages. Hence Grohmann argues for a novel analysis of German
Introduction
multiple wh-questions that involves overt movement into the Comp-domain of all wh-elements. He offers a number of empirical facts about the interplay of wh-phrases and apparent intervenors, some of which hitherto unknown. Further evidence comes from standard A¢-diagnostics and finer consideration of the contextual discourse licensing multiple wh-questions. In conclusion, Grohmann proposes a multiple wh-fronting analysis for German which differs from languages such as Bulgarian in not being operator A¢-movement, but topic-driven. Here he sketches some similarities between German and Italian, a language that lacks multiple wh-questions altogether. 3.6 Jeong: “Deriving Anti-Superiority effects: Multiple wh-questions in Japanese and Korean” Youngmi Jeong’s paper seeks to obtain a deeper understanding of two conditions on multiple questions in Japanese and Korean: so-called Anti-Superiority and the additional wh-effect. The author offers an alternative account of both Anti-Superiority and its relaxation by combining the effects of three independently motivated proposals: Kayne’s (1994) LCA-based claim that head-final languages are in fact head-initial, Rizzi’s (1997) articulated left periphery of the clause, and Pesetsky’s (2000) construal of the multiple wh-fronting requirement. Unlike previous accounts of Anti-Superiority, no appeal to notions like the ECP or government is necessary. 3.7 Lipták: “Conjoined questions in Hungarian” In her paper, Anikó Lipták investigates a little-discussed source of multiple questions: conjoined multiple questions. Such questions come in two varieties in Hungarian. In one we find conjoined wh-items in a preverbal position. In the other variety, we find one wh-item in preverbal position and another one (or other ones) in postverbal position, introduced by a coordinator element. The author shows that the two types of conjoined questions show different syntactic behavior in many respects, and argues that the syntactic differences have to be recognized by assigning distinct structural representations for the two types. The new analyses for these question types will also provide evidence for a binary-branching analysis of coordination.
11
12
Cedric Boeckx and Kleanthes K. Grohmann
3.8 “Lotfi: Persian wh-riddles” Ahmad Lotfi examines multiple wh-questions in Persian. Beyond a presentation of the relevant facts, he also picks up on recent treatments of single pair/pair list interpretations in multiple questions by Boškovic´ (1998b). A look at the Persian data allows a first characterization in that Persian is a wh-in situ language that also has a question particle (which is rarely used in wh-questions, but obligatory in yes/no-interrogatives). However, Persian has a productive scrambling property and Lotfi finds instances in which more than one wh-phrase may undergo fronting (subject to Superiority). It is this property which is directly relevant to the concerns of this volume, and on which Lotfi concentrates. The upshot of Lotfi’s treatment is that Persian multiple wh-questions are not as different as a first glance might suggest, and that they may be integrated nicely into recent approaches on the topic. 3.9 Reglero: “Non-wh-fronting in Basque” Basque multiple wh-questions have been looked at very rarely. Lara Reglero enriches the literature on the topic by offering a detailed overview of the data and the (apparent) two different strategies, fronting all wh-phrases or moving only one. Her analysis takes the work of Željko Boškovic´ (e.g. Boškovic´ 1999) as its starting point, and of Ortiz de Urbina (1995, 1999), who shows that wh-phrases in Basque behave very similarly to elements bearing focus. Reglero proposes that the driving force behind wh-fronting in Basque is focus. She offers an analysis for the singular wh-movement cases arguing that, contrary to appearances, both wh-phrases move overtly in the syntax. 3.10 Sabel: “Malagasy as an optional multiple wh-fronting language” This paper analyzes wh-question formation in Malagasy, an optional wh-in situ language that displays partial and full wh-movement as well. In addition, it is an optional multiple wh-fronting language. Sabel discusses the restrictions for multiple wh-in situ and wh-ex situ wh-questions and argues that the varieties of multiple wh-questions in Malagasy can be explained in terms of a feature checking analysis settled within the Minimalist Program.
Introduction
3.11 Stjepanovic´: “Multiple wh-fronting in Serbo-Croatian matrix questions and the matrix sluicing construction” Sandra Stjepanovic´ examines two problems with respect to multiple wh-fronting in Serbo-Croatian. One problem has to do with the positions to which wh-phrases are moving and the driving force behind this movement. An attempt to find an answer to this question will reveal that there is a parallelism between wh-phrases and identificationally focused material (in the sense of É. Kiss 1998) with respect to the positions they occupy in the sentence. The other problem is the presence of Superiority effects in the sentences involving sluicing and multiple wh-phrase remnants. The author shows that sluicing may be used as a probe into the mechanisms of multiple wh-fronting and the emergence of Superiority.
References Belletti, A. (ed). 2003. Structures and Beyond: The Cartography of Syntactic Structures, Vol. 2. Oxford: Oxford University Press. Boškovic´, Ž. 1997. “Superiority effects with multiple wh-fronting in Serbo-Croatian.” Lingua 102: 1–20. Boškovic´, Ž. 1998a. “Multiple wh-fronting and economy of derivation.” In Proceedings of the Sixteenth West Coast Conference on Formal Linguistics, E. Curtis, J. Lyle and G. Webster (eds), 49–63. Stanford, Calif.: CSLI. Boškovic´, Ž. 1998b. “On the interpretation of multiple questions.” In A Celebration: Essays for Noam Chomsky’s 70th Birthday, J. Fodor, S.J. Keyser and A. Brand (eds). Cambridge, Mass.: MIT Press. [Online publication — available for download at http://cognet.mit.edu/ Books/celebration/essays/Boeckx.html (sic); revised version to appear in Linguistic Variation Yearbook 1.] Boškovic´, Ž. 1999. “On multiple feature checking: Multiple wh-fronting and multiple head movement.” In Working Minimalism, S. D. Epstein and N. Hornstein (eds), 159–187. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 2001. On the Syntax–Phonology Interface: Cliticization and Related Phenomena. Amsterdam: Elsevier. Boškovic´, Ž. 2002. “On multiple wh-fronting.” Linguistic Inquiry 33: 351–383. Boškovic´, Ž. This volume. “On wh-islands and obligatory wh-movement contexts in South Slavic.” 27–50. Bresnan, J. 1970. “On complementizers: Toward a syntactic theory of complement types.” Foundations of Language 6: 297–321. Brody, M. 1990. Some Remarks on the Focus Field in Hungarian.” UCL Working Papers in Linguistics 2, 200–225. London: University College of London (Department of Linguistics and Phonetics). Brown, S. In press. “Negation in Russian and Attract All-F.” In Negation in Slavic, S. Brown and A. Przepiórkowski (eds). Bloomington, Ind.: Slavica Publishers.
13
14
Cedric Boeckx and Kleanthes K. Grohmann
Calabrese, A. 1984. “Multiple questions and focus in Italian.” In Sentential Complementation, W. de Geest and Y. Putseys (eds), 67–74. Dordrecht: Foris. Chierchia, G. 1991. Functional WH and weak crossover. In Proceedings of the Tenth West Coast Conference on Formal Linguistics, J. Bates (ed), 44–55. Stanford, Calif.: CLSI. Chierchia, G. 1993. Questions with quantifiers. Natural Language Semantics 1: 181–234. Chomsky, N. 1973. “Conditions on Transformations.” In A Festschrift for Morris Halle, S. R. Anderson and P. Kiparsky (eds), 232–286. New York: Holt, Rinehart, & Winston. Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. 2000. “Minimalist inquiries: The framework.” In Step by Step. Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 89–155. Cambridge, Mass.: MIT Press. Chomsky, N. 2001a. “Derivation by phase.” In Ken Hale: A Life in Language, M. Kenstowicz (ed), 1–52. Cambridge, Mass.: MIT Press. Chomsky, N. 2001b. “Beyond explanatory adequacy.” MIT Occasional Papers in Linguistics 20, 1–28. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). Reprinted 2003 in Structures and Beyond: The Cartography of Syntactic Structures, Vol. 2, A. Belletti (ed). Oxford: Oxford University Press. Comorovski, I. 1996. Interrogative Phrases and the Syntax–Semantics Interface. Dordrecht: Kluwer. Grewendorf, G. 2001. “Multiple wh-fronting.” Linguistic Inquiry 32: 87–122. Grohmann, K. K. 2000. Prolific peripheries: A radical view from the left. Doctoral dissertation, University of Maryland, College Park. Grohmann, K. K. This volume. “German is a multiple wh-fronting language!” 99–130. Haider, H. 2000. “Towards a superior account of Superiority.” In Wh-Scope Marking, U. Lutz, G. Müller and A. von Stechow (eds), 231–248. Amsterdam: John Benjamins. Higginbotham, J. and May, R. 1981. “Questions, quantifiers, and crossing.” The Linguistic Review 1: 41–80. Hiraiwa, K. 2001. “Multiple Agree and the defective intervention constraint in Japanese.” In MIT Working Papers in Linguistics 40: Proceedings of HUMIT 2000, O. Matushansky, A. Costa, J. Martin-Gonzalez, L. Nathan and A. Szczegielniak (eds), 67–80. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). Hiraiwa, K. 2002. Multiple Agree. Ms., Massachusetts Institute of Technology, Cambridge. Hoge, K. 2000. Superiority. Doctoral dissertation, University of Oxford. Hornstein, N. 1995. Logical Form. From GB to Minimalism. Oxford: Blackwell. Huang, C.-T. J. 1982a. Logical relations in Chinese and the theory of grammar. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge Huang, C.-T. J. 1982b. “Move Wh in a language without wh-movement.” The Linguistic Review 1: 369–416. Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. É. Kiss, K. (ed). 1995. Discourse Configurational Languages. Oxford: Oxford University Press. É. Kiss, K. 1998. “Identificational Focus versus Information Focus.” Language 74: 245–273. Koizumi, M. 1995. Phrase structure in minimalist syntax. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Lasnik, H. and Saito, M. 1984. “On the nature of proper government.” Linguistic Inquiry 15: 235–289.
Introduction
Lutz, U., Müller, G. and von Stechow, A. (eds). 2000. Wh-Scope Marking. Amsterdam: John Benjamins. Nissenbaum, J. 2000. Investigations of covert phrase movement. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Ortiz de Urbina, J. 1995. “Residual verb second and verb first in Basque.” In Discourse Configurational Languages, K. É. Kiss (ed), 99–121. Oxford: Oxford University Press. Ortiz de Urbina, J. 1999. “Focus in Basque.” In The Grammar of Focus, G. Rebuschi and L. Tuller (eds), 311–333. Amsterdam: John Benjamins. Pesetsky, D. 1987. “Wh-in situ: Movement and Unselective Binding.” In The Representation of (In)definiteness, A. G. B. ter Meulen and E. Reuland (eds), 98–129. Cambridge, Mass.: MIT Press. Pesetsky, D. 2000. Phrasal Movement and Its Kin. Cambridge, Mass.: MIT Press. Richards, N. 1997. What moves where when in which language? Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Rizzi, L. 1997. “The Fine Structure of the Left Periphery.” In Elements of Grammar: Handbook of Generative Syntax, L. Haegeman (ed), 281–337. Dordrecht: Kluwer. Rizzi, L. (ed). 2003. The Structure of CP and IP: The Cartography of Syntactic Structures, Vol. 3. Oxford: Oxford University Press. Rudin, C. 1985. Multiple wh-movement and the Superiority Condition. Paper presented at the 59th Annual Meeting of the Linguistic Society of America, Minneapolis, Minn. [January 1985] Rudin, C. 1988. “On multiple questions and multiple wh-fronting.” Natural Language and Linguistic Theory 6: 445–501. Stjepanovic´, S. 1998. Short distance movement of wh-phrases in Serbo-Croatian matrix clauses. Paper presented at the Comparative Slavic Morphosyntax Workshop, Spencer, Ind. [June 1998] Stjepanovic´, S. This volume. “Multiple wh-fronting in Serbo-Croatian matrix questions and the matrix sluicing construction.” 255–284. Toman, J. 1981. “Aspects of multiple wh-movement in Polish and Czech.” In Levels of Syntactic Representation, R. May and J. Koster (eds), 293–302. Dordrecht: Foris. Wachowicz, K. 1974. On the syntax and semantics of multiple questions. Doctoral dissertation, University of Texas, Austin. Williams, E. 1994. Thematic Structure in Syntax. Cambridge, Mass.: MIT Press. Williams, E. 2003. Representation Theory. Cambridge, Mass.: MIT Press.
15
Symmetries and asymmetries in multiple checking* Cedric Boeckx Harvard University
1.
Introduction
The present paper seeks to refine the formal aspect of multiple wh-fronting by refining the concept of Multiple Agree, recently advocated by Hiraiwa (2001, 2002), and Chomsky (2001b). I will mostly be dealing with the pattern of multiple wh-fronting found in Bulgarian. The language may be characterized abstractly as requiring movement of all wh-phrases (which I will call the symmetric part of the process), while at the same time favoring movement of the first wh-phrase, by imposing a Superiority condition on it (call this the asymmetric part of the process). Examples in (1) illustrate both aspects of the process. (The sentences in (1) show that all logically possible combinations of fronted wh-phrases are unacceptable unless the first wh-element of the sequence is the highest element before wh-fronting.) (1) a.
b. c. d. e. f. g. h. i.
Koj kogo kakvo e pital? who whom what is asked ‘Who asked whom what?’ Koj kakvo kogo e pital? *Kogo kakvo koj e pital? *Kakvo kogo koj e pital? *Kakvo koj kogo e pital? *Kogo koj kakvo e pital? *Koj e pital kogo kakvo? *Kogo e pital koj kakvo? *Kakvo e pital koj kogo?
There is reason to believe that the Bulgarian pattern is more basic than the one found in Serbo-Croatian (among other languages). As Boškovic´ has shown in
18
Cedric Boeckx
a series of work (Boškovic´ 1997b, 1998, 1999, 2002), Serbo-Croatian appears to lack the asymmetric aspect of multiple wh-fronting. Witness the absence of Superiority in (2).1 (2) a.
Ko šta gdje kupuje? who what where buys ‘Who buys what where?’ b. Ko gdje šta kupuje? c. Šta ko gdje kupuje? d. Šta gdje ko kupuje? e. Gdje ko šta kupuje? f. Gdje šta ko kupuje?
However, Boškovic´ has conclusively argued that the lack of Superiority is only true in matrix questions in which there is no overt evidence for a complementizer. Elsewhere (e.g., in embedded questions, long-distance questions, correlatives, matrix questions with overt complementizers), Serbo-Croatian is like Bulgarian, as (3) shows, using long-distance questions. (3) a.
Ko si koga turdio da je istukao? who are whom claimed that is beaten ‘Who did you claim beat whom?’ b. *Koga si ko turdio da je istukao?
To situate the investigation within current syntactic theorizing, I will take as my starting point Boškovic´’s (1999) characterization of Bulgarian. Boškovic´ argues that two features are checked against C0 in the process of multiple wh-fronting.2 Asymmetry arises from the checking of a [Wh]-feature, which can only be checked once (in Boškovic´’s terms, it has an [Attract-1] property). Superiority follows, as checking of the [Wh]-feature must be by the closest matching element (wh-phrase). Symmetry emerges from the checking of a [Focus]-feature,3 which unlike the [Wh]-feature, can be checked multiply (it has an [Attract-all] property). Because all features must be checked in the most economical way (a mainstay of the minimalist program), the order of elements checking the [Focus] feature does not matter (i.e. no Superiority effect arises), as the same number of nodes will ultimately be crossed (notice the global aspect of economy calculation here). In what follows I will preserve the gist of Boškovic´’s analysis, in particular the idea that two features ([Wh] and [Focus]) are involved in the process of multiple wh-fronting. But I want to embed the checking of those features within a different framework, so as to eliminate what I take to be suboptimal aspects of Boškovic´’s
Symmetries and asymmetries in multiple checking
analysis such as ‘Attract-1/Attract-All’ and global economy. This will require a detour into the nature of multiple checking in the A-domain, which is better understood, as it has been subject to extensive investigation in recent years.
2. Multiple Agree in the A-domain Consider the case of a ditransitive predicate, which for the purpose of discussion, takes two structurally Case-marked NP-objects. For illustration, I will use Japanese examples. (For equivalent constructions in a wide-range of languages, see Boeckx 2002.) Japanese double objects have an interesting property that has been documented in much detail in Boeckx and Niinuma (in press), and Boeckx and Jeong (2002). The property pertains to object honorification, which I take to be an instance of agreement (see the above references for extensive discussion of this issue). In the first modern study of object honorification, Harada (1976: 530) proposes the following rule called Object Honorific Marking: (3) Object Honorific Marking Mark the predicate as [Object Honorification] when an SSS (a person who is socially superior to the speaker) is included in a. the indirect object, if the predicate is ditransitive, or b. the direct object, if the predicate is transitive.
The relevant examples appear in (4–5). In (4), the verb is transitive, and it agrees with the direct object in honorification. In (5), we have a ditransitive predicate, and the verb agrees in honorification with the indirect object. (4) Taro-ga Tanaka sensei-o o-tasuke-si-ta. Taro-nom Tanaka teacher-acc help-oh-past ‘Taro helped Professor Tanaka.’ (5) Hanako-ga Tanaka Sensei-ni Mary-o go-syookai-si-ta. Hanako-nom Tanaka teacher-dat Mary-acc introduce-oh-past ‘Hanako introduced Mary to Professor Tanaka.’
However, Boeckx and Niinuma (in press) observe that (3) has to be refined in light of cases like (6). (6) *Hanako-ga Mary-ni Tanaka Sensei-o go-syookai-si-ta. Hanako-nom Mary-dat Tanaka teacher-acc introduce-oh-past ‘Hanako introduced Professor Tanaka to Mary.’
19
20
Cedric Boeckx
Here, the predicate is ditransitive, as in (5), but this time the NP bearing the relevant feature to trigger honorification functions is the direct object. In such a case, object honorification (i.e. agreement between the verb and the direct object) is impossible. If an object honorific marker surfaces on the verb, we obtain the odd interpretation that Taro respects Mary. What this amounts to is that the honorific marker in a ditransitive predicate can only associate with the indirect object, not the direct object. (7) summarizes the distribution of the object honorific marker in Japanese in light of (6). (7) Object Honorification Agreement Generalization Mark the predicate as [Object Honorification] when an SSS (a person who is socially superior to speaker) is a. a Dative argument b. the accusative object, if the predicate does not take a dative argument
Boeckx and Niinuma, and Boeckx and Jeong, argue that (6) constitutes a case of intervention (minimality). Since switching the surface order of the direct object and of the indirect object does not affect honorification marking (see (8)), Boeckx and Niinuma, and Boeckx and Jeong, claim that it must be the case that honorific agreement takes place prior to word order permutation. (8) *Hanako-ga Tanaka Sensei-o Mary-ni go-syookai-si-ta. Hanako-nom Tanaka teacher-acc Mary-dat introduce-oh-past ‘Hanako introduced Professor Tanaka to Mary.’
They take this to mean that agreement takes place under Chomsky’s derivational version of Agree, which applies as soon as the Probe (in our case, v) is introduced into the derivation. To capture the relevant intervention effect, they take the dative element to c-command the accusative element upon First Merge. If the reverse were a possible base order, the accusative element would have a chance of being closer to the functional head triggering agreement (say, v), and there would be no defective intervention. The relevant structure is provided in (9). (9) [vP v [VP IO [V′ DO V]]] X Agree
The above authors leave open the question of Case assignment in ditransitives. What their analyses show is that the direct object is inaccessible to v if a dative
Symmetries and asymmetries in multiple checking
element is present as far as φ-feature checking is concerned. But the direct object gets accusative (by assumption, structural) Case,4 which must come from v (Chomsky 1995).5 The paradox, then, is that the direct object is both accessible and inaccessible to v for checking purposes. More precisely, it is accessible for Case, but not for φ-feature checking. The problem is far from trivial, if we take seriously Chomsky’s (2000) claim that Case and φ-features are two sides of the same coin. But the same state of affairs is arrived at on independent grounds by Nomura (2002) and Hiraiwa (2002). Analyzing constructions from a wide range of languages, they show that Case is often assigned to several elements by one and the same Probe (a process they formalize in terms of Multiple Agree, on which see next section), but φ-feature checking is always asymmetric (it always takes place with the closest element). The situation is very reminiscent of the Bulgarian pattern of multiple wh-fronting. Assuming that overt displacement is a matter of EPP-satisfaction (Chomsky 2001a,b), one can characterize Bulgarian C0 as a probe that Agrees multiply for [Focus] and just once for [Wh]. At least two questions arise at this stage. First, how can an element both asymmetrically and symmetrically probe within the same domain? Second, why is it [Focus]/[Case] that can be probed multiply, and [Wh]/[φ] once, and not the other way round? The question is especially pressing, since [Case]/[φ] and [Focus]/[Wh] are taken to be reflexes of the same feature. This paper aims to answer those questions.
3. Toward a solution Let me start by saying what I take Chomsky’s statement to mean that [Case] and [φ] (and similarly for [Focus]/[Wh]) to be two sides of the same coin. First, I do not assume that they are the same feature, but instead entertain the possibility that they are features that have the same function. In particular, I take φfeatures to be interpretable on the Goal (NP) but uninterpretable on the Probe (T/v). The reverse situation obtains for Case: it is an interpretable feature on the Probe (corresponding to the Tense/Aspect feature), but uninterpretable on the Goal (NP).6 I assume that the presence of an uninterpretable feature on an element is forced to ensure ‘syntactibility’ (which expresses the same intuition as Chomsky’s 2000 notion of ‘visibility’): to ensure that the element can be combined (i.e. can participate in concatenation).
21
22
Cedric Boeckx
I further assume, following Chomsky (2000, 2001a,b), that the Agree relation is a composite operation consisting of a Matching procedure and a Valuation (agreement proper) procedure.7 Match is, by definition, symmetric: α and β match. By contrast, and quite crucially, Value is asymmetric: α values β. At first, it appears that the preceding paragraph offers a solution to our symmetry/asymmetry issue. All we have to do is pair the symmetric feature checking ([Case]/[Focus]) with Match and the asymmetric feature checking ([φ]/[Wh]) with Value. Although I will ultimately adopt this conclusion, it cannot be as straightforward, for plainly [Case]/[Focus] is valued on the Goal. Why isn’t valuation asymmetric in this case? I would like to propose that the answer to this question lies in the fact that in the case of symmetric feature checking, valuation proceeds from the Probe down to the Goal. I propose that the valuation (asymmetry) here piggybacks on (asymmetric) c-command (understood in terms of derivational Merge, in the spirit of Epstein 1999). By contrast, asymmetric valuation for [φ]/[Wh] cannot piggyback on asymmetric c-command. Instead they capitalize on another aspect of derivational order: closest/most recently merged element. The minimality effect follows. Summing up this section, asymmetric/symmetric checking is not due to an intrinsic difference between [Case]/[Focus] and [φ]/[Wh], but instead is a result of where they are placed.8,9 The interpretable feature located on the Probe can agree multiply, as the asymmetry is located elsewhere (the Probe c-commands the Goal). By contrast, the uninterpretable feature on the Probe can only be valued once, as the asymmetry arises via Closest (matching element). There is one more issue we have to touch on. As stated above, I follow Chomsky in taking Value to depend on Match. As Youngmi Jeong observes (p.c.), two situations could be envisaged here: the Probe matches (and enters into valuation with) the first Goal, and then matches the second Goal (without valuation from the Goal), as shown in (10). (10) Probe . . . Goal1 . . .
Goal2
1(Match) 2(Value) 3(Match)
Or else, the Probe matches both Goals as part of the same (simultaneous) operation. Then, valuation takes place (from the closest Goal), as schematized in (11). (11) Probe . . . Goal1 . . .
Goal2 1(Match)
2(Value)
Symmetries and asymmetries in multiple checking
(11) appears to be the only tenable option. If Valuation took place right after the first Matching element is found, the Probe would cease to be a Probe (its uninterpretable feature would be valued, hence invisible/inactive), and no subsequent matching/detection would be necessary (hence, by economy, should be ruled out). Returning now to the process of multiple wh-movement proper, we have to add that, from the discussion above, multiplicity as a result of Match cannot be parameterized (contra Fukui 1999): all languages match multiply (down to the next Probe).10 What is parameterized is whether languages associate that matching with overt displacement.11 When they do, the EPP-requirement is tied to the matching interpretable feature on the Probe. This converges with Boeckx’s (2001, 2002, 2003) claim that Move may take place under Match, independently of Agree. In fact, Serbo-Croatian matrix questions with no Superiority effect (see (2)) illustrate this perfectly: no valuation from the Goal takes place, but Match and Move happen.
4. Conclusion In this study I have argued that an investigation of the minutiae of Agree can lead to a deeper understanding of symmetries and asymmetries in instances of multiple agree such as multiple wh-fronting, without abandoning important guiding ideas of current syntactic theorizing such as local economy, intervention effects, parallelism among features, etc. Crucial was the distinction between Match and Agree, and the idea that the computational system is derivational (which allows for simultaneous operations, operations taking precedence over others etc.).
Notes * Thanks are due to the participants in my seminar at the University of Maryland in the Fall of 2002, to Kleanthes Grohmann, and to Norbert Hornstein for reactions. Special thanks to Youngmi Jeong for helping me understand the material presented here, for encouragement and much more. 1. A third pattern exists, exemplified by Basque, where Superiority regulates not only the first instance of wh-movement, but all of them, as shown in (i). For a treatment of that aspect of Basque multiple wh-questions, see Jeong (2002). (i) a.
Nork nori zer eman dio? who-erg whom-dat what-abs give aux ‘Who gave what to whom?’
23
24
Cedric Boeckx
b. *Nork zer nori eman dio? c. *Nori nork zer eman dio? d. *Nori zer nork eman dio? 2. Boškovic´ (2002) has argued that scrambling may be another source of wh-fronting in some cases. See also Grohmann (2000, this volume). 3. Boškovic´ here relies on Stjepanovic´’s (1998, 1999) finding that (non-wh) (contrastively) focused phrases in Serbo-Croatian must move overtly (just like wh-phrases) to positions that seem to be identical to the landing sites of wh-phrases. Elsewhere, Rizzi (1997) has argued that the Focus Phrase is the landing site of wh-movement. Numerous subsequent studies confirmed the isomorphism between focus-movement and wh-movement (with some exceptions, such as Lee 1999, to which I come back briefly below). 4. The possibility of honorification in simple transitive contexts such as (i) suggests that accusative Case is structural in Japanese (inherently Case-marked elements rarely show overt signs of agreement). (i) Taro-ga Tanaka sensei-o o-tasuke-si-ta/tasuke-ta. Taro-nom Tanaka teacher-acc help-oh-past/help-past ‘Taro helped Professor Tanaka.’ 5. I am assuming that v also assigns Case to the indirect (dative) object. Which Case value surfaces on which NP is arguably a case of morphology (late insertion; see Harley 1995, developing ideas of Alec Marantz). 6. My solution is in some sense related to Pesetsky and Torrego (2001, 2002) that Case is an uninterpretable T-feature. For me, Case is an uninterpretable feature corresponding to (but not necessarily identical to) T/Asp. I thus depart from Chomsky’s (2000) claim that feature matching is an instance of feature identity. I would like to claim that matching is in terms of featural compatibility, restricting featural identity to instances of adjunction. See Boeckx (in progress) for full discussion. 7. Note, incidentally, that in Chomsky’s (2001a,b) framework, Valuation cannot be a morphological operation, as valuation inactivates lexical items for the purposes of narrow syntax. 8. Abstractly, this is reminiscent of Boškovic´’s (1998) use of both (Probe-oriented) Attract and (Goal-oriented) Greed in an earlier version of his 1999 Attract-1/Attract all account sketched in Section 1. (Interestingly, Boškovic´ 1997a also argues that Case-checking, as opposed to φfeature checking, necessitates a notion like Greed). The present solution captures the same intuition, but without the computational complexity/globality associated with Greed. 9. The differences between wh-fronting and focus-movement noted by Lee (1999) may follow from this extrinsic difference. 10. This is reminiscent of Manzini and Roussou’s (2000) Scopal Agree, which lets Agree take place down to the next Attractor. Boeckx (in progress) explores the possibility of recasting control (as raising, see Hornstein 1999) in terms of multiple theta matching. Other areas of multiplicity, such as parasitic gap licensing, may also be fruitfully recast in those terms. 11. If they associate the EPP with Valuation, then only one overt movement process will take place.
Symmetries and asymmetries in multiple checking
References Boeckx, C. 2001. Mechanisms of chain formation. Doctoral dissertation, University of Connecticut, Storrs. Boeckx, C. 2002. Case and Agreement systems. Ms., University of Maryland, College Park. Boeckx, C. 2003. Islands and chains. Amsterdam: John Benjamins. Boeckx, C. In progress. Bare syntax. Ms., University of Maryland, College Park. Boeckx, C. and Niinuma, F. In press. “Conditions on agreement in Japanese.” Natural Language and Linguistic Theory. Boeckx, C. and Jeong, Y. 2002. The fine structure of syntactic intervention. Ms., University of Maryland, College Park. Boškovic´, Ž. 1997a. The Syntax of Nonfinite Complementation: An Economy Approach. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 1997b. “Superiority effects with multiple wh-fronting in Serbo-Croatian.” Lingua 102: 1–20. Boškovic´, Ž. 1998. “Multiple wh-fronting and economy of derivation.” In Proceedings of the Sixteenth West Coast Conference on Formal Linguistics, E. Curtis, J. Lyle and G. Webster (eds), 49–63. Stanford, Calif.: CSLI. Boškovic´, Ž. 1999. “On multiple feature checking: Multiple wh-fronting and multiple headmovement.” In Working Minimalism, S. Epstein and N. Hornstein (eds), 159–187. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 2002. “On multiple wh-fronting.” Linguistic Inquiry 33: 351–383. Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. 2000. “Minimalist inquiries: The framework.” In Step by Step. Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 89–155. Cambridge, Mass.: MIT Press. Chomsky, N. 2001a. “Derivation by phase.” In Ken Hale: A Life in Language, M. Kenstowicz (ed), 1–52. Cambridge, Mass.: MIT Press. Chomsky, N. 2001b. “Beyond explanatory adequacy.” MIT Occasional Papers in Linguistics 20, 1–28. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). Reprinted 2003 in Structures and Beyond: The Cartography of Syntactic Structures, Vol. 2, A. Belletti (ed). Oxford: Oxford University Press. Fukui, N. 1999. “The multiplicity parameter.” Glot International 4.9/10: 26–27. Grohmann, K. K. 2000. Prolific peripheries: A radical view from the left. Doctoral dissertation, University of Maryland, College Park. Grohmann, K. K. This volume. Harada, S-I. 1976. “Honorifics.” In Syntax and Semantics 5, M. Shibatani (ed), 499–561. New-York: Academic Press. Harley, H. 1995. Subjects, events, and licensing. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Hiraiwa, K. 2001. “Multiple Agree and the defective intervention constraint in Japanese.” In MIT Working Papers in Linguistics 40: Proceedings of HUMIT 2000, O. Matushansky, A. Costa, J. Martin-Gonzalez, L. Nathan and A. Szczegielniak (eds), 67–80. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT).
25
26
Cedric Boeckx
Hiraiwa, K. 2002. Multiple Agree. Ms., Massachusetts Institute of Technology, Cambridge. Hornstein, N. 1999. “Control and movement.” Linguistic Inquiry 30: 69–96. Jeong, Y. 2002. Basque multiple wh-fronting and iterative Superiority. Ms., University of Maryland, College Park. Lee, F. 1999. Antisymmetry and the syntax of San Luca Quiavini Zapotec. Doctoral dissertation, University of California, Los Angeles. Manzini, M. R. and Roussou, A. 2000. “A minimalist theory of A-movement and control.” Lingua 110: 409–447. Nomura, M. 2002. Multiple Case marking via Multiple Match. Ms., University of Connecticut, Storrs. Pesetsky, D. and Torrego, E. 2001. “T-to-C movement: Causes and consequences.” In Ken Hale: A Life in Language, M. Kenstowicz (ed), 355–426. Cambridge, Mass.: MIT Press. Pesetsky, D. and Torrego, E. 2002. Tense, Case, and the nature of syntactic categories. Ms., Massachusetts Institute of Technology, Cambridge and University of Massachusetts, Boston. Rizzi, L. 1997. “The fine structure of the left periphery.” In Elements of Grammar, L. Haegeman (ed), 281–337. Dordrecht: Kluwer. Stjepanovic´, S. 1998. Short distance movement of wh-phrases in Serbo-Croatian matrix clauses. Paper presented at the Workshop on Comparative Slavic Morphosyntax, Spencer, Ind. [June 1998] Stjepanovic´, S. 1999. What do second position cliticization, scrambling, and multiple wh-fronting have in common? Doctoral dissertation, University of Connecticut, Storrs.
On wh-islands and obligatory wh-movement contexts in South Slavic* Željko Boškovic´ University of Connecticut
The goal of this paper is to examine wh-islands effects in Bulgarian and SerboCroatian (SC) and constructions in which SC must have wh-movement. The starting point will be Rudin’s (1988) seminal paper on the structure of multiple wh-fronting (MWF) constructions in Bulgarian and SC. In Section 1 of the paper I briefly summarize Rudin’s (1988) analysis as well as a modification of Rudin’s analysis of SC from my previous work (see Boškovic´ 1997b, 1999, 2002a). In Section 2 I turn my attention to wh-islands in Bulgarian and SC. Although I will argue against Rudin’s analysis of wh-islands in the languages in question, which has served as a springboard for much recent work (see, for example, Richards 1997), I will end up endorsing the gist of Rudin’s analysis of MWF, namely, that there are two basic patterns of MWF constructions — one where all wh-phrases are located in SpecCP and one where at most one wh-phrase is located in SpecCP. However, the distribution of the two patterns will be shown to be quite different from that put forward by Rudin. In particular, the former pattern will be shown to be available in SC in certain contexts, contrary to what Rudin argues. All the differences between Bulgarian and SC MWF constructions will be traced to a single difference in the lexical properties of the interrogative C, more precisely, the PF affix status of the Bulgarian interrogative C. Furthermore, it will be shown that wh-islands do not provide evidence for the dichotomy of MWF constructions, as argued in Rudin (1988). In this respect, I will demonstrate that the resistance of Bulgarian to the wh-Island Constraint should not be related to the possibility of MWF in this language. More generally, I will show that the possibility of MWF itself cannot provide an escape hatch from wh-island effects, contrary to what is standardly assumed. The claim will be situated in a broader linguistic context and shown
28
Željko Boškovic´
to receive ample empirical and theoretical support from a variety of sources, including quantifier raising and Superiority.
1.
Structure of multiple wh-fronting constructions: Bulgarian vs. Serbo-Croatian
Rudin (1988) argues that in spite of the superficial similarity, Bulgarian and SC MWF constructions in (1)–(2) have very different structures. (1) Koj kakvo kupuva? who what buys ‘Who buys what?’ (2) Ko šta kupuje? who what buys ‘Who buys what?’
(Bulgarian)
(SC)
Rudin argues that Bulgarian and SC MWF constructions differ in two respects: only in Bulgarian fronted wh-phrases form a constituent and only Bulgarian allows more than one wh-phrase to be located in SpecCP. Rudin assigns the structures in (3) to (1)–(2). (3) a. [CP [SpecCP [SpecCP Koj] kakvo] [C¢ kupuva]]? b. [CP Ko [C¢ [IP šta [IP kupuje]]]]?
In Boškovic´ (1997b, 1999, 2002a) I argue that there is even a deeper difference between Bulgarian and SC MWF constructions. In particular, I argue that SC questions like (2) do not have to involve wh-movement at all, i.e. both wh-phrases can be located lower than the CP projection. This claim can be easily incorporated into Rudin’s analysis by pushing the first wh-phrase in (3b) a notch lower, i.e. by adjoining it to IP instead of moving it to SpecCP. I show in Boškovic´ (1997b, 1999, 2002a) that assuming that Bulgarian must, and SC does not have to, involve overt wh-movement to SpecCP can help us account for the well-known fact, noted by Rudin, that Bulgarian and SC MWF constructions like (1) differ with respect to possibilities for ordering of fronted wh-phrases. While in SC (2) and (4) the fronted wh-phrases are freely ordered, in Bulgarian (1) and (5), the nominative wh-phrase has to precede the accusative wh-phrase, which has been successfully analyzed in the literature in terms of Superiority.1
On wh-islands and obligatory wh-movement contexts in South Slavic
(4) Šta ko kupuje? what who buys ‘Who buys what?’ (5) *Kakvo koj kupuva? what who buys ‘Who buys what?’
(SC)
(Bulgarian)
Given the claim that Bulgarian (1) and (5) but not SC (2) and (4) must involve wh-movement, which I take to be movement motivated by checking the +wh-feature of C, the seemingly different behavior of wh-movement in the two languages with respect to Superiority can be easily explained. Since the SC questions in (2)/(4) do not have to involve wh-movement, they do not exhibit Superiority effects. Since the Bulgarian questions in (1)/(5) must involve wh-movement they exhibit Superiority effects. Under this analysis, wh-movement in Bulgarian and SC is well behaved with respect to Superiority — whenever wh-movement takes place we get Superiority effects. (See Boškovic´ 1999 for details of the analysis. See also Boškovic´ 2002a for discussion of the nature of wh-fronting in Slavic that does not involve wh-movement. I argue that this movement involves focalization, i.e. it is an instance of focus movement. Note that the second wh-phrase in Bulgarian (1) and (5) also undergoes focus movement. Only the first wh-phrase undergoes wh-movement, i.e. checks the strong +wh-feature of C.) The above analysis is confirmed by the selectivity of Superiority effects in Bulgarian. In Boškovic´ (1997a) I show that the highest wh-phrase prior to wh-fronting must move first to SpecCP in Bulgarian, the order of movement of other wh-phrases to SpecCP being free. The data illustrating this are given in (6)–(9). (6)–(7) show that the indirect object must move before the direct object when it is the highest wh-phrase before wh-fronting, as in (6), but not when it is not, as in (7), where the highest wh-phrase is koj. (Recall that the linear order of wh-phrases indicates the order of movement.) Additional data making the same point are given in (8)–(9). The accusative wh-phrase must move to SpecCP before the adjunct wh-phrase when it is the highest wh-phrase prior to wh-movement, as in (8), but not when a higher wh-phrase is present, as in (9).2 (6) a.
Kogo kakvo e pital Ivan? whom what is asked Ivan ‘Who did Ivan ask what?’ b. ?*Kakvo kogo e pital Ivan?
29
30
Željko Boškovic´
(7) a.
Koj kogo kakvo e pital? who whom what is asked ‘Who asked who what?’ b. Koj kakvo kogo e pital?
(8) a.
Kogo kak e tselunal Ivan? whom how is kissed Ivan ‘How did Ivan kiss whom?’ ? b. *Kak kogo e tselunal Ivan?
(9) a.
Koj kogo kak e tselunal? who whom how is kissed ‘Who kissed whom how?’ b. Koj kak kogo e tselunal?
Notice the parallelism between the wh-phrases in SC (2) and (4) and non-initial fronted wh-phrases in Bulgarian with respect to Superiority, more precisely, the lack of Superiority effects. The parallelism confirms the above analysis, where movement of the first wh-phrase in Bulgarian differs from the movement of the second and the third wh-phrase, which are in turn the same as the movement of all the wh-phrases in SC (2)/(4). In other words, since the second and third movements in Bulgarian (6)–(9) and both movements in SC (2)/(4) are the same (neither has to involve checking the strong +wh-feature of C; as shown in Boškovic´ 2002a, they can both be pure focus movement), it is not surprising that they behave in the same way with respect to Superiority, differing in this respect from the first movement in (6)–(9), which has to involve checking the strong +wh-feature of C. The data show that only the wh-phrase that checks the strong +wh-feature of C (which means only one wh-phrase) is subject to Superiority, wh-phrases undergoing pure focus movement being insensitive to it. Boškovic´ (1999) gives an economy-based explanation of this fact applicable to both SC and Bulgarian that is based on certain differences in formal properties of focus and wh-movement.3 Based on the above data, I conclude in Boškovic´ (1999, 2002a) that the distribution of Superiority effects in Bulgarian and SC provides evidence that in contrast to Bulgarian questions, SC questions like (2) and (4) do not have to involve overt wh-movement at all. For another argument to this effect concerning the interpretation of multiple questions in Bulgarian and SC, the reader is referred to Boškovic´ (1999, 2002a). In Boškovic´ (2002b) (see also Boškovic´ 2001a,b), I attribute the difference between Bulgarian and SC with respect to the obligatoriness of wh-movement
On wh-islands and obligatory wh-movement contexts in South Slavic
to the timing of interrogative C-insertion in Bulgarian and SC: interrogative C, whose presence triggers immediate wh-movement, must be inserted in overt syntax in Bulgarian, but not in SC, where it can be inserted in LF, hence wh-movement must take place overtly in Bulgarian, but not in SC.4 Why is there a difference in the timing of C-insertion between the two languages? I attribute it to a PF requirement on the interrogative C which is present in Bulgarian, but lacking in SC. In particular, I suggest that interrogative C is a PF verbal affix in Bulgarian, but not in SC. As a result, interrogative C must be inserted into the structure in overt syntax in Bulgarian, but not necessarily in SC. If interrogative C were to be inserted into the structure in LF in Bulgarian, the PF requirement could not be satisfied and the derivation would crash.5 Independent evidence for the difference between Bulgarian and SC is provided by the fact that Inversion is obligatory in Bulgarian, but not in SC questions. More precisely, the fact that interrogative C must be adjacent to a verb in PF in Bulgarian, but not in SC indicates that the C is a verbal affix in Bulgarian, but not in SC.6 (10) a. *Kakvo toj dade na Petko? what he gave to Petko ‘What did he give to Petko?’ b. Kakvo dade toj na Petko? c. Šta on dade Ivanu? what he gave Ivan ‘What did he give to Ivan?’
(Bulgarian)
(SC)
We thus have a uniform account of the different behavior of Bulgarian and SC with respect to Superiority and Inversion. As discussed in Boškovic´ (1997b, 2002a), under the above analysis SC actually can have overt wh-movement in constructions like (2). Whether or not wh-movement takes place overtly in (2) depends on whether or not interrogative C is inserted overtly — if it is, wh-movement takes place overtly; if it is not, it does not. Below, I discuss some constructions in which wh-movement indeed takes place overtly in SC.7 Having discussed my modification of Rudin’s analysis of SC which allows for the possibility of not having a wh-phrase in SpecCP in overt syntax in SC questions (in fact, not having a CP in overt syntax at all), I return now to Rudin (1988). Rudin gives several arguments supporting her proposal for the structure of MWF constructions in Bulgarian and SC, which posits a difference between the two languages with respect to the constituency of fronted wh-phrases and
31
32
Željko Boškovic´
the ability to multiply fill an interrogative SpecCP. Concerning the constituency of fronted wh-phrases, Rudin (1988) claims that SC allows parentheticals to intervene between fronted wh-phrases, which is not possible in Bulgarian. (11) Ko, po tebi, šta kupuje? who according.to you what buys ‘Who, according to you, is buying what?’ tebe, kakvo kupuva? (12) ?*Koj, spored who according.to you what buys
(SC)
(Bulgarian)
Rudin argues that the impenetrability of fronted wh-phrases in Bulgarian indicates that they form a constituent. She interprets the possibility of lexical material occurring between fronted wh-phrases in SC as indicating that fronted wh-phrases do not form a constituent in this language.8 It is worth noting, however, that (12) could also be accounted for under the multiple-specifiers analysis of Bulgarian MWF, proposed in Koizumi (1994) and further developed in Richards (1997) and Pesetsky (2000). On this analysis, fronted wh-phrases in Bulgarian are all located in SpecCP, as in Rudin’s analysis. However, they do not form a constituent in that position, contra Rudin (1988). Rather, each wh-phrase is located in a distinct Spec. Under this analysis (12) can be ruled out due to a feature clash: a [−Wh] element is located in an interrogative [+Wh] projection. From this perspective, (12) would be interpreted as evidence that all Bulgarian wh-phrases are located in interrogative SpecCP, not necessarily that they form a constituent in that position.
2. Wh-islands I now turn to an argument for Rudin’s structure for Bulgarian and SC questions which has incited a great deal of follow-up work (see especially Richards 1997) and which is also the topic of this paper. The argument concerns extraction out of wh-islands. Rudin claims that Bulgarian allows, and SC disallows, extraction out of wh-islands based on constructions like (13). (13) a.
Vidjah edna kniga, kojatoi se ˇcudja koj znae koj prodava ti. saw-1s a book which refl wonder-1s who knows who sells ‘I saw a book which I wonder who knows who sells.’ (Bulgarian) b. *Vidio sam knjigu kojui se pitam ko zna ko prodaje ti. seen am book which refl wonder-1s who knows who sells ‘I saw a book which I wonder who knows who sells.’ (SC)
On wh-islands and obligatory wh-movement contexts in South Slavic
Rudin interprets the data in (13) as indicating that, in contrast to SC, Bulgarian allows more than one wh-phrase to be located in SpecCP in overt syntax. As a result, kojato in the Bulgarian example can escape the wh-Island Constraint by moving through the embedded SpecCPs, occupied by koj. Since SC does not allow more than one wh-phrase in SpecCP overtly, the escape hatch from the wh-Island Constraint is not available in SC.9 The relevant facts are, however, more complex than (13) indicates. In particular, the status of Bulgarian with respect to the wh-Island Constraint is not completely clear. Rudin herself notes that, in contrast to relativization, Bulgarian exhibits wh-island effects in questions. Rudin’s example in (14) illustrates this. Rudin also observes that (15), containing a D-linked wh-phrase, contrasts with (14). Based on this, Rudin concludes that questioning out of a wh-island in Bulgarian is allowed with D-linked, but not with non-D-linked wh-phrases.10 (14) *Kakvoi se ˇcudiš koj znae koj prodava ti? what refl wonder-2s who knows who sells ‘What do you wonder who knows who sells?’ koj znae koj prodava ti? (15) ?Koja ot tezi knigii se ˇcudiš which of these books refl wonder-2s who knows who sells ‘Which of these books do you wonder who knows who sells?’
The literature on wh-islands in Bulgarian generally focuses on argument extraction and ignores adjunct extraction.11 The facts concerning adjunct extraction flatly contradict the claim that Bulgarian is not sensitive to the wh-Island Constraint. As shown in (16), extraction of adjuncts out of wh-islands leads to full unacceptability regardless of whether we are dealing with relativization or questioning. D-linking is also irrelevant. (16) a. *Pricˇinata, poradi kojatoi [Ivan znae dali Boris e zaminal ti] reason-def for which [Ivan knows whether Boris is left ‘the reason for which Ivan knows whether Boris left’ b. *Zašto/ poradi kakva pricˇinai znae [dali Boris e zaminal ti]? why for which reason knows [whether Boris is left ‘Why/for which reason does he know whether Boris left?’ c. cf. Zašto/ poradi kakva pricˇinai misliš [cˇe Boris e zaminal ti]? cf. why for which reason think-2s [that Boris is left ‘Why/for which reason do you think that Boris left?’
33
34
Željko Boškovic´
These facts indicate that wh-islands are islands in Bulgarian. Consequently, any analysis that completely voids Bulgarian of the wh-island effect must be on the wrong track. Note also that Swedish, a language that does not allow MWF, behaves in the same way as Bulgarian with respect to wh-islands. Thus, argument extraction out of wh-islands in Swedish is possible with relativization and D-linking questions, but not with non-D-linking questions, as observed in Comorovski (1996). With adjuncts, extraction out of wh-islands is never possible, which is generally ignored in the literature.12 (17) a. *Vad frågade Jan vem som skrev? what asked John who that wrote ‘What did John ask who wrote?’ (Maling 1978) b. Det är melodin, som Jan frågade vem som skrev. this is melody-def that John asked who that wrote ‘This is the melody that John asked who wrote.’ (Maling 1978) c. Vilken film var det du gärna ville veta vem som hade which film was it you gladly wanted know-inf who that had regisserat? directed ‘Which film did you want to know who had directed?’ (Engdahl 1986) d. *Varför/av vilket skäl undrar han [vem som lagade bilen t]? why /for which reason wonder-3s he [who that fixed car-def ‘Why/for which reason does he wonder who fixed the car?’ e. *orsaken varför han undrar [vem som lagade bilen t] reason-def why he wonder-3s [who that fixed car-def ‘the reason why he wonders who fixed the car’
The fact that Bulgarian, a MWF language, and Swedish, a non-MWF language, exhibit the same behavior with respect to wh-islands indicates that an analysis that crucially relates the possibility of extraction out of wh-islands in certain contexts in Bulgarian to the possibility of MWF is on the wrong track. SC confirms this conclusion. As noted in Boškovic´ (1997b, 2002a), SC questions must involve wh-movement in certain contexts. This happens in the contexts in which LF C-insertion, which is a prerequisite for the no-overtwh-movement derivation, is blocked.13 The contexts in question include embedded questions, where LF C-insertion is blocked because it would involve lexical insertion in the middle of the tree (Merger is allowed to take place only at the root of the tree, i.e. it must expand the tree) and questions involving the
On wh-islands and obligatory wh-movement contexts in South Slavic
phonologically overt complementizer li, which, being phonologically realized, obviously must enter the structure overtly. In Boškovic´ (1997b, 2002a) I give two additional contexts in which wh-movement must take place in SC, namely, long-distance and topicalization questions, and Stjepanovic´ (1999a,b) provides another context of this type involving sluicing (see these works for explanation why the LF C-insertion derivation is blocked in these contexts). All the contexts in question exhibit Superiority effects, as expected given that they involve overt wh-movement. In this respect they contrast with constructions like (2) and (4), where nothing blocks LF C-insertion, hence overt wh-movement is not forced. I give here examples from Boškovic´ (2002a) concerning embedded, longdistance, li, and topicalization questions.14 (18) a.
[Ko koga voli], taj o njemu i govori. [who whom loves that-one about him even talks ‘Everyone talks about the person they love.’ ? b. *[Koga ko voli], taj o njemu/o njemu taj i govori.
(19) a. ?Ko koga tvrdiš da je istukao? who whom claim-2s that is beaten ‘Who do you claim beat whom?’ b. *Koga ko tvrdiš da je istukao? (20) a. (?)Ima ko šta da ti proda. has who what prt you sells ‘There is someone who can sell you something.’ b. *Ima šta ko da ti proda. (21) a.
Ko li koga voli? who q whom loves ‘Who on earth loves whom?’ b. *Koga li ko voli?
(22) a.
Tom ˇcovjeku, ko je šta poklonio? that man who is what bestowed ‘On that man, who bestowed what?’ b. ??Tom ˇcovjeku, šta je ko poklonio?
It is worth noting that even in the contexts where overt C-insertion and wh-movement must take place, Inversion is still not required, as expected given that interrogative C is not a PF affix in SC. As a result, its presence in the structure does not require Inversion.
35
36
Željko Boškovic´
(23) Tom ˇcovjeku, šta Jovan poklanja? that man what John bestows ‘On that man, what is John bestowing?’
Notice also that the contexts in question exhibit selective Superiority effects, just like Bulgarian questions. Recall that, as demonstrated in (6)–(9), only the first wh-phrase in Bulgarian questions is subject to Superiority effects, i.e. the highest wh-phrase prior to wh-fronting must move first to SpecCP in Bulgarian, the order of movement of other wh-phrases to SpecCP being free. Interestingly, as discussed in Boškovic´ (2002a), the contexts in which SC must have wh-movement also display selective Superiority effects: the highest wh-phrase prior to movement is first in the linear order, the order of other wh-phrases being free. This is illustrated with respect to existential embedded questions in (24) (see also Stjepanovic´ 1999a,b for sluicing). (24) a. ?Ima kome kako da pomogne. has whom how prt helps ‘S(he) has someone to help somehow.’ b. *Ima kako kome da pomogne. c. ?Ima ko kako kome da pomogne. has who how whom prt helps ‘There is someone who can somehow help somebody.’
The parallelism suggests that in the contexts in question, SC questions have the same structure as Bulgarian questions. In Boškovic´ (2002a) I show that this is indeed the case: as in Bulgarian, in the contexts in question all fronted wh-phrases are located in SpecCP. The claim is based on the fact that, like Bulgarian wh-phrases, SC wh-phrases in the contexts in question cannot be broken by parentheticals, which according to Rudin’s criteria indicates that they are all located in SpecCP. ((25) should be compared with (11).)15 (25) a. *Ko, po tebi, koga vjeruju da tucˇe? who according.to you who believe-3p that beats ‘Who, according to you, they believe beats who? ?? b. Tom ˇcovjeku, ko, po tebi, šta poklanja? that-dat man-dat who-nom according.to you what bestows ‘On that man, who, according to you, bestows what?’ c. *Ima ko, po tebi, šta da mu proda. has who according.to you what prt him sells ‘There is someone who, according to you, can sell him something.’
On wh-islands and obligatory wh-movement contexts in South Slavic
d. *Ko, po tebi, koga voli, taj o njemu i govori. who according.to you whom loves that-one about him even talks ‘According to you, everyone talks about the person they love.’ e. *Ko, po tebi, koga li voli? who according.to you whom q loves ‘Who on earth loves whom, according to you?’
It follows then that SC allows more than one wh-phrase in SpecCP in the contexts in question. Significantly, even the contexts in question exhibit wh-island effects. This is illustrated by (13b) and the examples in (26). (Given the above discussion, all the wh-phrases in (13b) and (26), which involve embedded and longdistance questions, move to SpecCP. Furthermore, as discussed above, interrogative SpecCPs in (26) and (13b) can be multiply filled.) (26) a. *Koju od tih knjigai se pitaš ko li prodaje ti? which of these books refl wonder-2s who q sells ‘Which of these books do you wonder who on earth sells?’ b. *Koje knjigei ima ko da ti proda ti? which books has who prt you sells ‘Which books is there someone who can sell to you?’
Notice that, like (13b), (26a–b) are acceptable in Bulgarian: (27) a.
Koja knigai ima koj da ti prodade ti? which book has who prt you sells ‘Which book is there someone who can sell to you?’ b. Koja ot tezi knigii se ˇcudiš koj li prodava ti? which of these books refl wonder-2s who q sells ‘Which of these books do you wonder who on earth sells.’
The SC data indicate that, contrary to what is standardly assumed (see, for example, Comorovski 1986, Rudin 1988, Koizumi 1995, and Richards 1997), the possibility of having more than one wh-phrase in SpecCP at S-Structure does not make possible derivations in which a wh-phrase moves through a filled SpecCP to escape wh-island effects. This is not surprising in light of the data discussed in Epstein (1992) and Boškovic´ (1997c), which indicate that once an operator moves to an A¢-position in which it can establish an operator (Op)-variable relation, it cannot undergo further A¢-movement. In fact, in Boškovic´ (1997c) I adopt the condition in (28), which would freeze a wh-phrase entering an interrogative SpecCP in the interrogative SpecCP, thus preventing it from passing through it (for much relevant discussion, see also Epstein 1992 and references therein).
37
38
Željko Boškovic´
(28) Op in Op-variable chains cannot undergo further movement.
The well-known ban on Quantifier Raising (QR) topicalized quantifiers illustrates the effect of (28). Lasnik and Uriagereka (1988) observe that every problem cannot have scope over someone in (29b) even for the speakers for whom it can scope over someone in (29a). (29) a. Someone thinks that Mary solved every problem. b. Someone thinks that every problem, Mary solved.
Every problem presumably scopes over someone in (29a) as a result of QR into the matrix clause. Given this, (29b) indicates that topicalization has a freezing effect on QR (i.e. a topicalized element cannot undergo QR), which follows from (28) (for much relevant discussion, see also Epstein 1992). SC constructions involving overt wh-movement provide more evidence for the necessity of (28). Recall that in the contexts in which SC must have wh-movement, all SC wh-phrases move to SpecCP. One wh-phrase undergoes wh-movement, checking the strong +wh-feature of C, and other wh-phrases undergo focus movement, which means that C can license wh-phrases for focus in SC, just as in Bulgarian. Wh-movement is sensitive to Superiority, while focus movement is not. As a result, the highest wh-phrase must move first to SpecCP, the order of movement of other wh-phrases being free. Recall, however, that SC also has the possibility of licensing wh-phrases for focus in a lower position. This possibility is, for example, realized in (2) and (4), which do not have to involve overt wh-movement at all (see Boškovic´ 1997c and Stjepanovic´ 1998, 1999a for the precise landing site of the wh-phrases in these constructions). The question now arises why SC wh-phrases in questions that must involve overt wh-movement cannot first undergo focus movement to the focus position below C, which would be followed by wh-movement of one wh-phrase to SpecCP. In the current system, we want to rule out this derivation for two reasons. First, given that focus movement is not sensitive to Superiority, wh-phrases could be freely ordered in the focus projection. It appears then that we would have no way of ensuring the existence of Superiority effects in the contexts in question.16 In other words, in the derivation in question, focus movement would provide an escape hatch from Superiority effects even in the contexts where wh-movement must take place in SC, which, as shown above, do display Superiority effects. Second, under this derivation it would be difficult to account for the fact that it is more difficult to separate SC wh-phrases by a parenthetical in the contexts that involve wh-movement than in those that do not
On wh-islands and obligatory wh-movement contexts in South Slavic
(cf. the contrast between (11) and (25)). I conclude, therefore, that the derivation on which focus movement feeds wh-movement needs to be blocked. As observed in Boškovic´ (1997c), (28) in fact straightforwardly blocks the derivation in question. Under this derivation, a wh-phrase first undergoes focus movement, which I assume is A¢-movement that creates an Op-variable chain. The wh-phrase then undergoes wh-movement, in violation of (28).17 By ruling out the possibility of focus movement feeding wh-movement, (28) ensures the desired result: although in principle SC wh-phrases can be checked for focus either in SpecCP or in a position lower than C, the latter option is blocked in constructions involving wh-movement, where C enters the structure overtly, triggering overt wh-movement. The upshot of the analysis is that whenever interrogative C is present in the structure it must be the focus licenser. Licensing wh-phrases for focus in a lower position in the presence of interrogative C violates (28).18 We thus account for the switch to the Bulgarian paradigm in constructions in which SC must have wh-movement.19 We also explain why SC and Bulgarian appear to differ with respect to which elements license wh-phrases for focus. Using a focus licenser lower than C in Bulgarian questions will never give a legitimate result (putting aside the potentially relevant context from note 18), since it would invariably result in a violation of (28). In SC this is a possibility in questions that do not involve overt wh-movement. In fact, there is evidence that SC and Bulgarian do not differ in this respect. In Boškovic´ (1997b, 2002a) I show that even echo wh-phrases must undergo focus movement in the languages under consideration, (30a–b) being unacceptable even on the echo reading of what.20 (30) a. ?*Ivan kupuje ŠTA? Ivan buys what ‘Ivan buys WHAT?’ ? b. *Ivan e kupil KAKVO? Ivan is bought what ‘Ivan bought WHAT?’
(SC)
(Bulgarian)
As discussed in Boškovic´ (2002a), in echo questions interrogative C does not have to be present in the structure. Given this, if Bulgarian in principle had the possibility of licensing wh-phrases for focus in a position lower than CP we would expect the possibility to be taken advantage of in echo questions. Since wh-movement does not have to take place in such constructions, licensing of a wh-phrase for focus in a lower position would not lead to a violation of (28).
39
40
Željko Boškovic´
The possibility is indeed realized, as shown in (31), where the fronted echo wh-phrase is clearly located lower than SpecCP. (31) Ti misliš ˇce KAKVO e kupil Petko? you think that what is bought Petko ‘You think that Petko bought WHAT?’
I conclude therefore that, like SC, Bulgarian has the option of licensing wh-phrases for focus in a position lower than SpecCP. In fact, all the differences between Bulgarian and SC MWF constructions noted by Rudin (1988) now follow from a single lexical difference between Bulgarian and SC interrogative C — the C is an affix in Bulgarian, but not in SC — which has independent motivation (cf. the Inversion data in (10)). Notice also that the analysis is in line with the recent research attempt to attribute all cross-linguistic variation to lexical properties. Returning now to (28), in Boškovic´ (1997c) I show that (28) is also responsible for the ungrammaticality of constructions like *What do you wonder John bought (when). Chomsky (1995) argues that features that have semantic import (interpretable features) are ‘unaffected’ by checking. They can undergo checking both more than once and less than once. According to Chomsky, the +wh-feature of wh-phrases is an interpretable feature. Therefore, it can enter multiple checking. Given this, consider the derivation in (32). What first moves to the lower SpecCP, checking the strong +wh-feature of the embedded C. It then moves to the matrix SpecCP, checking the strong +wh-feature of its head. (32) *Whati do you wonder [CP ti C [IP John bought ti (when)]]
It is not clear how (32) can be ruled out (note that when can be interpreted in the embedded SpecCP, perhaps through unselective binding). In fact, it seems to be well-formed syntactically.21 Its ungrammaticality can then be taken to indicate that a wh-phrase cannot pass through an interrogative SpecCP even when that SpecCP is empty, which follows from (28). (What in (32) undergoes A¢-movement after moving to a position where it can establish an operatorvariable relation.) Rudin’s way of voiding the wh-island effect in Bulgarian is abstractly very similar to (32), the only difference between (32) and the Bulgarian cases being that in the Bulgarian cases the SpecCP through which the wh-phrase passes is already filled, which is irrelevant given that by hypothesis Bulgarian allows more than one wh-phrase in SpecCP. I conclude therefore that the possibility of an interrogative wh-phrase passing through an interrogative SpecCP needs to be ruled out. We then need
On wh-islands and obligatory wh-movement contexts in South Slavic
a new analysis of the (selective) lack of wh-island effects in Bulgarian which will not rely on passing through a filled SpecCP, i.e. on the possibility of multiplyfilled SpecCPs at S-Structure. As noted above, the fact that Swedish, a nonMWF language, behaves like Bulgarian with respect to wh-islands leads to the same conclusion. The fact that SC exhibits wh-island effects in the contexts in which it allows multiply-filled SpecCPs in overt syntax confirms that the possibility of MWF itself (more precisely, multiply-filled SpecCPs) cannot provide an escape hatch from the wh-Island Constraint. Concerning the structure of MWF constructions, I conclude that the different behavior of Bulgarian and SC with respect to wh-islands is not related to the possibility of locating more than one wh-phrase in interrogative SpecCP at S-Structure and therefore does not provide evidence for the structures Rudin proposes for SC and Bulgarian questions. The cause of this difference remains to be determined. It is, however, worth emphasizing that removing the wh-island argument does not invalidate Rudin’s analysis of MWF. Thus, eliminating the wh-island argument does not affect the impenetrability-ofwh-phrases argument, which provides evidence for Rudin’s claim that all Bulgarian wh-phrases are located in interrogative SpecCP. As for SC, in certain contexts all SC wh-phrases remain below CP, which requires modifying Rudin’s structure in (3b) by pushing even the first wh-phrase below CP. (The CP is actually not even present in the contexts in question.) There are, however, contexts in which wh-movement is forced in SC. In these contexts, SC exhibits the Bulgarian pattern, with all fronted wh-phrases located in SpecCP. The different behavior of Bulgarian and SC with respect to the obligatoriness of overt wh-movement as well as the availability of the pattern in which fronted wh-phrases are located lower than SpecCP follows from a single difference in the lexical specification of the interrogative C, the element being a PF affix in Bulgarian but not in SC, in line with the current view of cross-linguistic variation as resulting from lexical differences.
Notes * Parts of the discussion of wh-islands from this paper were presented at the Comparative Slavic Morphosyntax Workshop (Spencer, Indiana), the 5th European Summer School in Generative Grammar (Debrecen), Harvard University, Princeton University, Université de Paris 8, University of Maryland, Universidade Estadual de Campinas, University of the Basque Country in Vitoria-Gasteiz, ZAS Berlin, and the University of Connecticut. I thank all of these audiences for thought-provoking questions. I also thank Iliyana Krapova,
41
42
Željko Boškovic´
Mariana Lambova, Roumyana Pancheva, Christer Platzack, Penka Stateva, and Sandra Stjepanovic´ for help with judgments and Kleanthes Grohmann for editorial assistance. 1. See Rudin (1988), Boškovic´ (1997a, 1998b, 1999, 2002a), Richards (1997, 1998), and Pesetsky (2000), among others. One argument that the fixed order of the wh-phrases in (1)/(5) is a result of Superiority concerns the fact that (5) improves with D-linked (i-a) and echo wh-phrases (i-b). (KOJ in (i-b) is an echo wh-phrase.) The same happens with Superiority violations in English (ii). Notice that all the above-mentioned authors argue that the wh-phrase that is first in the linear order in Bulgarian questions is the one that moves first, in accordance with Superiority. The second wh-phrase either right-adjoins to the first wh-phrase, located in SpecCP, as in Rudin (1988), or moves to a lower SpecCP (the first wh-phrase is located in the higher SpecCP), as in Richards (1997) and Pesetsky (2000). The multiple specifiers analysis was originally proposed in Koizumi (1994). For another approach to Bulgarian MWF, see Kim (1997), Grewendorf (2001), and Sabel (2001, this volume). (Notice also that SC exhibits Superiority effects in certain contexts, which are discussed in Section 2.) (i) a. ?Koja kniga koj ˇcovek kupuva? which book which man buys ‘Which book is which man buying?’ b. ?Kakvo KOJ kupuva? (ii) a. Who bought what? b. *What did who buy? c. Which book did which man buy? d. What did WHO buy? 2. As argued in Boškovic´ (1997a), the object wh-phrase checks the strong +wh-feature of C in (8) rather than the adjunct because it moves to its Case-checking position prior to wh-movement, thus ending up higher than the adjunct prior to wh-movement. Notice also that (i) shows that (6)–(9) do not exhibit the same phenomenon as English (ii), noted in Kayne (1984), where addition of a lower wh-phrase rescues a Superiority violation (see Pesetsky 2000 for a recent discussion of this effect.) (i) a. *Kogo koj kak e tselunal? b. *Kogo koj kakvo e pital? (ii) a. *What did who buy? b. (?)What did who buy where? 3. The gist of the analysis is the following: When it comes to wh-movement (recall that I take wh-movement to be movement motivated by checking the +wh-feature of C), only one wh-phrase needs to move, checking the strong +wh-feature of C. In order to check the feature in the most economical way, i.e. through the shortest movement possible, it is always the highest wh-phrase that moves to check the +wh-feature C. (Note that I assume that movement to SpecCP triggers Spec–Head agreement with C, checking its +wh-feature. This means that with respect to wh-movement, the highest wh-phrase always must move first; otherwise, the +wh-feature will not be checked in the most economical way.) With focus movement, we are dealing with multiple movement to the same position. (All wh-phrases undergo this movement in the languages in question.) Regardless of the order of movements, the same number of nodes will always be crossed, hence no order is preferred by Economy.
On wh-islands and obligatory wh-movement contexts in South Slavic
(In Boškovic´ 1998b I state the focus requirement as an inadequacy of wh-phrases, i.e. I assume that wh-phrases have a strong focus feature, which needs to be checked overtly. In Boškovic´ 1999, on the other hand, I state the focus requirement as an inadequacy of the target of movement, giving the target head the specification Attract-all for focus, which is satisfied by attracting all focalized elements. Under both of these analyses, all orders of movement of wh-phrases are equally economical when it comes to satisfying the focus requirement.) 4. The underlying assumptions here are that +wh C is strong in both Bulgarian and SC and that strength is defined as in Chomsky (1995), namely as something that has to be eliminated from the structure through checking as soon as it enters the structure. 5. More precisely, the presence of phonological information in LF would cause a crash. (The same would happen if, for example, Mary were to be inserted into the structure in LF.) If Bulgarian interrogative C (or Mary for that matter) is inserted into the structure overtly, the phonological information from its lexical entry is stripped off when the structure is sent to PF, so that it does not enter LF. 6. Izvorski (1993) shows convincingly that the verb in Bulgarian (10b) is not located in C. Rather, it is located lower in the structure. Thus, Izvorski observes that if Bulgarian were to have I-to-C movement in questions, (i-b) should be acceptable, just like its English counterpart What had Maria forgotten about. (Notice that the auxiliary, which Izvorski assumes is located in I and therefore should be affected by I-to-C movement, is not a proclitic on the verb, like some other auxiliary forms in Bulgarian.) Also, if the verb in Bulgarian questions were located in C (which means that a subject following it could be located in SpecIP), the adverb in (ii-b) should have both the low, manner reading, and the high, subject-oriented adverb reading, just like the adverb in (ii-a) and English constructions of this type. (Izvorski gives What did John carefully read?, where the adverb can have either the manner or the subject-oriented adverb reading.) (i) a.
Maria beše zabravila za sreštata. Maria was forgotten about meeting-def ‘Maria had forgotten about the meeting.’ b. *Za kakvo beše Maria zabravila? about what was Maria forgotten ‘About what had Maria forgotten?’ c. Za kakvo beše zabravila Maria? d. *Za kakvo Maria beše zabravila?
(ii) a.
Petko pravilno otgovori na vu˘prosa im. Petko correctly answered to question-def they-dat ‘Petko did the right thing when he answered their question.’ ‘Petko gave a correct answer to their question.’ b. Na kakvo otgovori Petko pravilno? to what answered Petko correctly *‘What was Petko right to answer?’ ‘What did Petko give a correct answer to?’
43
44
Željko Boškovic´
Given that the verb is lower than C, the simple structural explanation of the ungrammaticality of (10a) that would rely on V-to-C movement, namely, there is not enough space to place the subject between the wh-phrase in SpecCP and the verb, is not available. In Boškovic´ (2001a,b, 2002b) I apply Chomsky’s (1957) affix hopping analysis to (10a–b). The C affix undergoes affix hopping to the verb, which is only possible when the subject is not phonologically realized in SpecIP, PF adjacency being a prerequisite for affix hopping. (I actually argue that the subject does move to SpecIP. However, a lower copy of the subject is pronounced to make affix hopping possible (note that Bulgarian does not have anything like do-support), in line with Franks’s (1998) proposal, discussed below, that a lower copy of a non-trivial chain can be pronounced iff this is necessary to satisfy a PF requirement (see also Bobaljik 1995, Hiramatsu 2000, Pesetsky 1997, 1998, and Boškovic´ 2001a, 2002a).) 7. The LF C-insertion analysis is applied to French in Boškovic´ (1998a, 2000b). LF C-insertion results in wh-in situ, i.e. lack of overt wh-movement, in French. As in SC, overt C-insertion triggers overt wh-movement. It turns out that LF C-insertion is blocked in the same contexts in French as in SC, which results in an interesting correlation between the possibility of wh-in situ in French and the lack of Superiority effects in SC (they both depend on the availability of the LF C-insertion derivation). It is worth noting here that in Boškovic´ (2000b) I argue that the different behavior of English and French with respect to the obligatoriness of wh-movement should be accounted for in the same way as the different behavior of Bulgarian and SC with respect to the phenomenon. As in Bulgarian, interrogative C is a PF affix in English, hence it must be inserted into the structure overtly. Overt C-insertion triggers overt wh-movement. As in SC, interrogative C is not a PF affix in French, hence it does not have to be inserted overtly. As discussed above, when the C is not inserted overtly wh-movement does not take place overtly. The analysis is supported by the fact that, like Bulgarian and SC, English and French differ with respect to Inversion. As in Bulgarian, Inversion must take place in English questions like (i) (i.e. C must be V-adjacent, see Boškovic´ 2000b for explanation why Inversion does not take place in English embedded questions). French patterns with SC in the relevant respect, as expected (see (ii)). The LF C-insertion analysis thus provides us with a uniform account of the different behavior of Bulgarian and English on one hand, and SC and French on the other hand, with respect to Inversion and wh-movement/Superiority. (i) a. Who have you seen? b. *Who you have seen? (ii) Qui tu as vu? who you have seen ‘Who have you seen?’ 8. The Bulgarian constructions improve when the intervening material is contrastively focused, the relevance of which is clear under the focus movement analysis of MWF argued for in Boškovic´ (2002a). (For much relevant discussion, see also Lambova 2002.) Rudin also observes that SC and Bulgarian MWF constructions differ with respect to the possibility of splitting fronted wh-phrases by adverbs and clitics. However, the clitic data involve an interfering factor. While SC clitics are second position (2P) clitics, whose only requirement is that they occur in the 2P of their intonational (I)-phrase, Bulgarian clitics are verbal clitics, which must be adjacent to the verb (see Boškovic´ 2001a and references therein.)
On wh-islands and obligatory wh-movement contexts in South Slavic
Notice also that, as discussed in Boškovic´ (2001a) and note 14, in some cases a 2P clitic may be pronounced between 2 wh-phrases in SC as a result of a PF re-ordering mechanism (more precisely, due to pronunciation of a lower copy of one of the wh-phrases), although in the syntax it follows them. The interfering factors invalidate the clitic intervention test. 9. Rudin claims that Bulgarian and SC also differ with respect to the possibility of multiple extraction of wh-phrases out of declarative clauses. However, I and other speakers I have consulted find SC (i) acceptable. (i) Ko šta želite da vam kupi? who what want-2p that you buys ‘Who do you want to buy you what?’ 10. It is worth noting here that for some (though not all) speakers, questions involving extraction of a D-linked wh-phrase out of a wh-island are worse if they involve extraction across another D-linked wh-phrase, as in (i). (Thanks are due to Cedric Boeckx for suggesting checking extraction out of D-linked wh-islands.) (i)
??(?)Koja
ot tezi knigii se ˇcudiš koj ˇcovek znae koj ucˇitel prodava t i ? which of these books refl wonder-2s which man knows which teacher sells ‘Which of these books do you wonder which man knows which teacher sells?’
11. Note that in the current theoretical system, traditional Subjacency and ECP violations (the former arising with extraction of arguments and the latter with extraction of adjuncts out of islands) are treated in essentially the same way (see Chomsky and Lasnik 1993, Takahashi 1994, Boškovic´ and Lasnik 1999, among others). 12. As in Bulgarian, where extraction out of a wh-island is possible in Swedish, it can take place out of more than one wh-island. Notice also that adjuncts can be extracted longdistance out of declarative complements in Swedish. Significantly, D-linked questions also exhibit wh-island effects if the wh-island itself contains a D-linked wh-phrase in SpecCP, another parallelism with Bulgarian (see note 10). (i) ??Vilken film var det du gärna ville veta vilken skådespelare som hade which film was it you gladly wanted know-inf which actor that had regisserat? directed ‘Which film did you want to know which actor had directed?’ 13. Recall that the +wh C in SC has a strong +wh-feature. As a result, it triggers wh-movement as soon as it enters the structure. If it enters the structure overtly, it triggers overt wh-movement. 14. I ignore the irrelevant echo-question reading. Note that I do not give here indirect questions as examples of embedded questions because such questions involve an interfering factor. Indirect questions formally do not differ at all from matrix questions in SC. As a result, there is always a danger that they could be analyzed as matrix questions, with the superficial matrix clause treated as an adsentential. The problem does not arise with correlative constructions like (18) and existential constructions like (20), which also contain embedded questions (see Izvorski 1996, 1998). However, I show in Boškovic´ (1997b) that when the interfering factor noted above is controlled for, indirect questions also exhibit
45
46
Željko Boškovic´
Superiority effects. Notice also that the S-Structure of (21a) can actually be Ko koga li voli in light of the discussion of (25) below. Given that li is a 2P clitic and that the 2P clitic effect is a PF effect, as shown in Boškovic´ (2001a), we are forced to pronounce a lower copy of the wh-phrase immediately preceding li (namely koga) to satisfy the 2P requirement on li under Franks’s (1998) approach to the pronunciation of non-trivial chains, where we are allowed to pronounce a lower copy of a non-trivial chain iff this is necessary to satisfy a PF requirement. (See Boškovic´ 2001a for discussion of the 2P effect in multiple questions under Franks’s approach to the pronunciation of non-trivial chains. Notice that the algorithm for determining which copy of a non-trivial chain to pronounce from Boškovic´ 2001a, 2002a (see also Boškovic´ and Franks 2002) ensures that koga rather than ko is pronounced in a lower position in (21a), whose S-Structure is given in (i). Under the algorithm, it will always be the wh-phrase that immediately precedes li that is pronounced in a lower position, so that (21b) is ruled out due to a violation of Superiority — the accusative wh-phrase moves to SpecCP before the nominative wh-phrase.) (i) Ko koga li ko koga voli… who whom q who whom loves The above remarks concerning (21a) also extend to (22a), where both wh-phrases may precede at S-Structure the auxiliary je, a 2P clitic like li. (Note that the topicalized element is parsed as a separate I-phrase, which means that it is irrelevant to the 2P requirement on je. As discussed in note 15, SC 2P clitics must be second within their I-phrase.) (ii) …ko šta je ko šta… 15. As in Bulgarian, contrastively focusing the parenthetical generally leads to improvement. Note, however, that the correlative example (25d) may be irrelevant, since the wh-clause of the correlative does not tolerate the parenthetical in question regardless of its position. However, other material cannot intervene between the wh-phrases of the wh-clause either. Note also that (25e) does not violate the 2P requirement, so that lower pronunciation of the second wh-phrase is not allowed, in contrast to (21a). As discussed in Boškovic´ (2000a, 2001a), the proper formulation of the 2P effect is that SC clitics must be second within their I-phrase, not clause. As a result, since parentheticals are followed by an I-phrase boundary, a clitic can be hosted by an element that immediately follows a parenthetical even when this would prevent it from being second within its clause. (The element immediately following a parenthetical is always the initial element of an I-phrase.) This is illustrated by (i-a), which contrasts with (i-b), where the clitic ´ce is located in the third position of its I-phrase. The 2P requirement is satisfied in (25e) in the same way as in (i-a). (i) a.
Znacˇi da, kao što rekoh, oni ´ce sutra doc´i. means that as said-1s they will tomorrow arrive ‘It means that, as I said, they will arrive tomorrow.’ b. *Znacˇi da oni ´ce sutra doc´i.
16. See, however, Lambova’s (2002) discussion of Bulgarian. 17. Grohmann’s (2000) approach to anti-locality, which bans movement that is too local (see Grohmann 2000 for precise definitions), may also rule out the derivation in question, at least in some cases.
On wh-islands and obligatory wh-movement contexts in South Slavic
18. There is actually one exception, regarding SC constructions like (i). (i) Ko tvrdiš da je šta kupio? who claim-2s that is what bought ‘Who do you claim bought what?’ Boškovic´ (1997c) and Stjepanovic´ (1998, 1999a) show that in constructions in which SC does not have obligatory overt wh-movement, there are still two focus licensing positions, one above and one below sentential adverbs. (Both are still lower than C. Note that whatever element licenses focus we need to assume either that the element does not have to be present in the structure or that it is only optionally taken from the lexicon with a focus feature; this is necessary since it is not the case that every sentence must contain a wh-phrase or a contrastively focused non-wh-phrase.) If only the lower but not the higher focus position is activated in the embedded clause of (i), the sentence can be derived without violating (28). Šta can be focus-licensed in the lower focus position. Since this position is located below the embedded clause SpecIP, the embedded clause subject still has to be focus-licensed in the matrix clause, more precisely, matrix SpecCP. That šta in (i) is indeed located in the lower focus position is confirmed by (ii)–(iii), which show that šta is located below sentential adverbs. (Recall that the lower focus position is located below sentential adverbs, and the higher focus position above them.) (ii) a. ?Ko tvrdiš da je vjerovatno šta kupio? who claim-2s that is probably what bought ‘Who do you claim probably bought what?’ b. *Ko tvrdiš da je šta vjerovatno kupio? (iii) a. ?Ko tvrdiš da je mudro šta kupio? who claim-2s that is wisely what bought ‘Who do you claim that it was wise of to buy what?’ ‘Who do you claim bought what in a wise manner?’ b. Ko tvrdiš da je šta mudro kupio? *‘Who do you claim that it was wise of to buy what?’ ‘Who do you claim bought what in a wise manner?’ It is worth noting here that there is some variation with respect to the availability of the low focus position below sentential adverbs in Slavic. Thus, while this position is available in SC, as demonstrated in Boškovic´ (1997c) and Stjepanovic´ (1998, 1999a), it is not available in Russian, as shown in Stepanov (1998). 19. Note that the appeal to acyclic adjunction made in Boškovic´ (1997c) to account for the SC constructions in question is no longer necessary. 20. I am considering only the reading on which the echo question asks for repetition of what the questioner has not heard. Echo questions in situ are acceptable on the reading on which they express surprise. As discussed in Boškovic´ (2002a), this can be straightforwardly accounted for under the focus movement analysis of wh-fronting in Slavic since the value of the echo wh-phrase is fully known to the speaker, as well as the hearer, on the surprise reading, but not on the request for repetition reading. Hence, the wh-phrase has to undergo focus movement only on the latter reading. (Note that focus represents new information.)
47
48
Željko Boškovic´
21. The problem arises in Chomsky’s (1995) system. Chomsky (2000: 123) does offer a technical way out of the problem based on the proposal that only an element with an uninterpretable feature can undergo movement. Chomsky makes the proposal in order to deal with certain facts regarding A-movement (p. 128–129), but he (p. 128) does suggest its extension to A¢-movement (without offering empirical motivation for this move though). Returning to (32), the uninterpretable feature that makes what visible for movement is presumably deleted in the embedded SpecCP. However, Boškovic´ (2002c) and especially Saito (2000) show that, apart from its stipulatory nature, the visibility approach faces a number of very serious problems. They argue that it can be, and should be, eliminated from the grammar, which is also what I assume here. It is worth noting, however, that the visibility account would also extend to Bulgarian wh-islands, preventing Bulgarian wh-phrases from passing through filled SpecCPs, in accordance with what is argued for here. (In fact, (28) might be deducible under the visibility approach, if the same uninterpretable feature, say the +Q feature, is involved in all operator movement.)
References Bobaljik, J. 1995. Morphosyntax: The syntax of verbal inflection. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Boškovic´, Ž. 1997a. “On certain violations of the Superiority Condition, AgrO, and economy of derivation.” Journal of Linguistics 33: 227–254. Boškovic´, Ž. 1997b. “Superiority effects with multiple wh-fronting in Serbo-Croatian.” Lingua 102: 1–20. Boškovic´, Ž. 1997c. “Fronting wh-phrases in Serbo-Croatian.” In Formal Approaches to Slavic Linguistics: The Indiana Meeting, 1996, M. Lindseth and S. Franks (eds), 86–107. Ann Arbor, Mich.: Michigan Slavic Publications. Boškovic´, Ž. 1998a. “LF movement and the Minimalist Program.” In Proceedings of NELS 28, P. N. Tamanji and K. Kusumoto, (eds), 43–57. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Boškovic´, Ž. 1998b. “Multiple wh-fronting and economy of derivation.” In Proceedings of the Sixteenth West Coast Conference on Formal Linguistics, E. Curtis, J. Lyle and G. Webster (eds), 49–63. Stanford, Calif.: CSLI. Boškovic´, Ž. 1999. “On multiple feature-checking: Multiple wh-fronting and multiple headmovement.” In Working Minimalism, S. Epstein and N. Hornstein (eds), 159–187. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 2000a. “Second position cliticisation: Syntax and/or phonology?” In Clitic Phenomena in European Languages, F. Beukema and M. den Dikken (eds), 71–119. Amsterdam: John Benjamins. Boškovic´, Ž. 2000b. “Sometimes in [Spec, CP], sometimes in situ.” In Step by Step: Essays on Minimalism in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 53–87. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 2001a. On the Nature of the Syntax–Phonology Interface: Cliticization and Related Phenomena. Amsterdam: Elsevier.
On wh-islands and obligatory wh-movement contexts in South Slavic
Boškovic´, Ž. 2001b. “PF Merger in Scandinavian: Stylistic Fronting and Object Shift.” Working Papers in Scandinavian Syntax 68: 75–115. Boškovic´, Ž. 2002a. “On multiple wh-fronting.” Linguistic Inquiry 33: 351–383. Boškovic´, Ž. 2002b. “On certain differences between Bulgarian and Serbo-Croatian C(P).” Balkanistica 15: 35–48. Boškovic´, Ž. 2002c. “A-movement and the EPP.” Syntax 5: 167–218. Boškovic´, Ž. and Franks, S. 2002. “Phonology-syntax interactions in South Slavic.” Balkanistica 15: 49–74. Boškovic´, Ž. and Lasnik, H. 1999. “How strict is the cycle?” Linguistic Inquiry 30: 691–703. Chomsky, N. 1957. Syntactic Structures. The Hague: Mouton. Chomsky, N. 1995. “Categories and transformations.” In N. Chomsky, The Minimalist Program, 219–394. Cambridge, Mass.: MIT Press. Chomsky, N. 2000. “Minimalist inquiries.” In Step by Step: Essays on Minimalism in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 89–155. Cambridge, Mass.: MIT Press. Chomsky, N. and Lasnik, H. 1993. “The theory of principles and parameters.” In Syntax: An International Handbook of Contemporary Research, Vol. 1, J. Jacobs, A. von Stechow, W. Sternefeld and T. Vennemann (eds), 506–569. Berlin: Walter de Gruyter. Comorovski, I. 1986. “Multiple wh-movement in Romanian.” Linguistic Inquiry 17: 171–177. Comorovski, I. 1996. Interrogative Phrases and the Syntax–Semantics Interface. Dordrecht: Kluwer. Engdahl, E. 1986. Constituent Questions. Dordrecht: Reidel. Epstein, S. D. 1992. “Derivational constraints on A¢-chain formation.” Linguistic Inquiry 23: 235–259. Franks, S. 1998. Clitics in Slavic. Paper presented at the Comparative Slavic Morphosyntax Workshop, Spencer, Ind. [June 1998] Grewendorf, G. 2001. “Multiple wh-fronting.” Linguistic Inquiry 32: 87–122. Grohmann, K. K. 2000. Prolific peripheries: A radical view from the left. Doctoral dissertation, University of Maryland, College Park. Hiramatsu, K. 2000. Accessing linguistic competence: Evidence from children’s and adults’ acceptability judgments. Doctoral dissertation, University of Connecticut, Storrs. Izvorski, R. 1993. On wh-movement and focus movement in Bulgarian. Paper presented at ConSOLE 2, University of Tübingen. [December 1993] Izvorski, R. 1996. “The syntax and semantics of correlative proforms.” In Proceedings of NELS 26, K. Kusumoto, (ed), 133–147. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Izvorski, R. 1998. “Non-indicative wh-complements of possessive and existential predicates.” In Proceedings of NELS 28, P. N. Tamanji and K. Kusumoto (eds), 159–173. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Kayne, R. 1984. Connectedness and Binary Branching. Dordrecht: Foris. Kim, J.-S. 1997. Syntactic focus movement and ellipsis: A minimalist approach. Doctoral dissertation, University of Connecticut, Storrs. Koizumi, M. 1994. “Layered specifiers.” In Proceedings of NELS 24, M. Gonzalez, (ed), 255–269. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts).
49
50
Željko Boškovic´
Koizumi, M. 1995. Phrase structure in minimalist syntax. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Lambova, M. 2002. “On A¢-movements in Bulgarian and their interaction.” The Linguistic Review 18: 327–374. Lasnik, H. and Uriagereka, J. 1988. A Course in GB Syntax: Lectures on Binding and Empty Categories. Cambridge, Mass.: MIT Press. Maling, J. 1978. “An asymmetry with respect to wh-islands.” Linguistic Inquiry 9: 75–89. Pesetsky, D. 1997. “Optimality theory and syntax: Movement and pronunciation.” In Optimality Theory: An Overview, D. Archangeli and T. Langendoen (eds), 134–170. Oxford: Blackwell. Pesetsky, D. 1998. “Some optimality principles of sentence pronunciation.” In Is the Best Good Enough?, P. Barbosa, D. Fox, P. Hagstrom, M. McGinnis and D. Pesetsky (eds), 337–383. Cambridge, Mass.: MIT Press and MITWPL. Pesetsky, D. 2000. Phrasal Movement and Its Kin. Cambridge, Mass.: MIT Press. Richards, N. 1997. What moves where when in which language? Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Richards, N. 1998. “Shortest moves to (anti-)Superiority.” In Proceedings of the Sixteenth West Coast Conference on Formal Linguistics, E. Curtis, J. Lyle and G. Webster (eds), 335–349. Stanford, Calif.: CSLI. Rudin, C. 1988. “On multiple questions and multiple wh-fronting.” Natural Language and Linguistic Theory 6: 445–501. Sabel, J. 2001. “Deriving multiple head and phrasal movement: The cluster hypothesis.” Linguistic Inquiry 32: 532–547. Sabel, J. This volume. “Malagasy as an optional multiple wh-fronting language.” 229–254. Saito, M. 2000. Scrambling in the Minimalist Program. Ms., Nanzan University. Stepanov, A. 1998. “On wh-fronting in Russian.” In Proceedings of NELS 28, P. N. Tamanji and K. Kusumoto (eds), 453–467. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Stjepanovic´, S. 1998. Short distance movement of wh-phrases in Serbo-Croatian matrix clauses. Paper presented at the Comparative Slavic Morphosyntax Workshop, Spencer, Ind. [June 1998] Stjepanovic´, S. 1999a. “Multiple sluicing and Superiority in Serbo-Croatian.” In Proceedings of NELS 29, P. Tamanji, M. Hirotani and N. Hall (eds), 145–159. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Stjepanovic´, S. 1999b.What do second position cliticization, scrambling, and multiple wh fronting have in common? Doctoral dissertation, University of Connecticut, Storrs. Takahashi, D. 1994. Minimality of movement. Doctoral dissertation, University of Connecticut, Storrs.
On the nature of multiple fronting in Yiddish* Molly Diesing Cornell University
1.
What is so special about Yiddish multiple wh-fronting?
Since the work of Rudin (1988), researchers have distinguished two major types of multiple wh-fronting languages. The first type, exemplified by Bulgarian and Romanian, involves moving all the wh-phrases to CP, exploiting either adjunction (as in Rudin’s original proposal), or multiple specifiers (as in subsequent work — for example Richards 1997 and Pesetsky 2000). In the second type, which includes Polish and Czech (as well as Serbo-Croatian, see Boškovic´ 1999), only a single wh-phrase may move to CP, while the rest move to IP, again by either adjunction or multiple specifiers.1 A number of diagnostics for distinguishing the two types have been proposed, including (see Richards 2001 for a more thorough overview): (1) Type 1: Bulgarian, Romanian a. shows Superiority effects b. escapes wh-islands2 c. no local (A-movement) scrambling (2) Type 2: Serbo-Croatian (Polish, Czech) a. no Superiority effects in short movement, but shows Superiority in long movement b. wh-islands in effect c. has local scrambling
In this paper I will be focusing mainly on the issues of Superiority and landing sites in multiple wh-fronting in Yiddish. Explanations for the distinctions among multiple wh-fronting languages have also taken somewhat different forms, including recasting the CP/IP distinction in terms of an enriched peripheral structure in the clause, along the
52
Molly Diesing
lines of Rizzi (1997). An example of this is the “prolific periphery” of Grohmann (2000), which distinguishes Ω- and Φ-domains as landing sites for multiple fronting, and thereby delineating the various multiple fronting types. Yet another tactic is to investigate the possibility that not all wh-fronting is in fact wh-movement. In a series of papers, Boškovic´ (1998a-b, 1999, 2000) thus distinguishes wh-fronting from focus movement. An interesting consequence of Boškovic´’s work is the claim that multiple fronting in the sense of multiple wh-movement actually does not exist. Boškovic´ (2000) proposes that multiple wh-fronting languages be categorized in the same way that single-fronting languages are: (3) Boškovic´’s Typology a. Type 1: Always move one wh-phrase (English) b. Type 2: Never move any wh-phrase (Chinese) c. Type 3: Sometimes move one wh-phrase, sometimes move none (French)
On this view, in the multiple fronting languages the additional instances of fronting are taken to be focus movement. Thus, when augmented by focus movement, the multiple fronting languages can be classified as follows: Bulgarian becomes a language of Type 1, Russian is supposedly of Type 2, and SerboCroatian is of Type 3. It is in light of these recent proposals that I wish to investigate multiple fronting in Yiddish. Among the West Germanic languages, Yiddish poses something of a typological puzzle. For not only is Yiddish the only Germanic language which permits multiple fronting of wh-phrases, it is also a language which allows two strategies for forming multiple questions: multiple fronting (like Bulgarian) and single fronting (like German). Interestingly, these two options differ in their syntactic properties, most notably with regard to the so-called Superiority effect. In particular, Superiority seems to be linked to multiple fronting (see also Hoge 2000). In what follows I examine the question of where Yiddish fits in the overall typology of multiple wh-fronting, taking as a starting point the assumption that some form of Rudin’s CP vs. IP landing site typology is correct (though this position will be refined somewhat in my final analysis). To give an overview, Sections 2 and 3 provide a preliminary presentation of the relevant data. Section 4 focuses on the Superiority effect in Yiddish, while Section 5 aims to provide a syntactic explanation of the Superiority facts, assuming that some sort of multiple wh-fronting takes place. The final section investigates dialectal variation with respect to multiple fronting and Yiddish. I
On the nature of multiple fronting in Yiddish
also examine the Yiddish data in the context of Boškovic´’s proposals regarding the taxonomy of multiple fronting, and I show that multiple fronting in Yiddish is indeed a genuine instance of multiple wh-fronting (in the sense of moving multiple wh-phrases to CP, rather than to specifiers of a Focus projection related to CP), in contrast to what Boškovic´ has claimed for other multiple fronting languages. At the same time, Yiddish also provides evidence for fronting of a wh-phrase to a position below CP in some cases of single fronting.
2. A first look at the data As mentioned above, the multiple question construction in Yiddish (in particular, the Southeastern dialect) can take multiple forms. The first possibility consists of fronting of a single wh-phrase to the position immediately preceding the inflected verb (here it is important to bear in mind that Yiddish is a verbsecond language): (4) a.
Ver vet vuhin geyn mit aykh? who will where go with you? ‘Who will go where with you?’ b. Ver hot vos gekoyft? who has what bought ‘Who bought what?’ c. Ver geyt vuhin? who goes where ‘Who is going where?’
Examples (4a–b) illustrate another property of the single fronting variant of a multiple question; the “wh-in situ” appears immediately to the left of the (nonfinite) main verb (when there is one), despite the fact that the normal VP order is VO. In earlier work, I suggested that an explanation for this lies in the fact that Yiddish has a preverbal focus position below the subject position (Diesing 1997), which can (and even must) host a single in situ wh-phrase. This of course involves a somewhat extended meaning of the term “in situ”: I use it here to mean a wh-phrase which has not been moved to a specifier position above the subject (that is, a non-fronted wh-phrase). This focus position will be discussed further in Sections 5 and 6 below. The role of the focus position in multiple questions of course raises the question of what happens when there are more than two wh-phrases, since the
53
54
Molly Diesing
focus slot can host only one phrase. Interestingly, multiple questions in Yiddish seem to be restricted to binary questions. My Yiddish consultants simply do not like multiple questions with more than two wh-phrases; they tended to conjoin wh-phrases in excess of two. Kerstin Hoge (p.c.) reports similar judgments from her consultants. An explanation for this observation will also be deferred to Section 6. The second option for forming multiple questions involves multiple fronting of the wh-phrases; in these cases the fronted wh-phrases all precede the position of the inflected verb: (5) a.
Ver vuhin vet geyn mit aykh? who where will go with you ‘Who will go where with you?’ (Mark 1978: 380) b. Ver vos hot gekoyft? who what has bought ‘Who bought what?’ c. Ver vuhin geyt? who where goes ‘Who is going where?’ (Zaretski 1929)
As far as I have been able to determine, the two orders in these examples are synonymous. Neither option is associated with any special meaning. In particular, the wh-in situ in the single-fronting cases need not be D-linked (in the sense of Pesetsky 1987 — the potential answers to these questions are not necessarily limited to sets of entities previously defined in the discourse), in contrast with languages like Bulgarian (Pesetsky 2000). It appears that multiple fronting in Yiddish is genuinely optional in these cases. An obvious initial question that arises in consideration of multiple fronting in Yiddish is that of where the multiply-fronted wh-phrases move to. Is the landing site CP (as argued for Bulgarian by Rudin and others), or some lower functional projection such as IP (as claimed for Polish)? Evidence bearing on this question can be found by examining word order in embedded questions. In the dialect of Yiddish under consideration, both single fronting (6a) and multiple fronting (6b–c) are possible in an embedded question. (6) a.
Hot zi ni(sh)t gekent farshteyn [ver es shlogt zikh has she not been-able understand [who expl hits self mit vemen]. with whom ‘(So) she couldn’t understand who was fighting with whom.’
On the nature of multiple fronting in Yiddish
b. Hot zi ni(sh)t gekent farshteyn [ver mit vemen es shlogt zikh]. (Jacobs et. al. 1994: 414) c. Lomir geyn, ver vuhin es geyt. let’s go who where expl goes ‘Let’s go where(ever) who(ever) is going.’ (Zaretski 1929)
Each of these examples involves extraction from the subject position, leaving a subject gap. Interestingly, an expletive es ‘it’ is required in these cases (Birnbaum 1978, Travis 1984, Diesing 1990). The expletive does not appear if a nonsubject has been topicalized (in (7), the adverb frier ‘earlier’, in (8) haynt ‘today’) has been topicalized, occupying the SpecIP position (see Diesing 1990 for arguments justifying this placement of non-subject topics): (7) Zi iz gekumen zen [ver frier vet kontshen]. she is come see [who earlier will finish ‘She has come to see who would finish earlier.’ (Diesing 1990: 50) (8) Lomir geyn, ver vuhin haynt geyt. let’s go who where today goes ‘Let’s go where(ever) who(ever) is going today.’
One can characterize the data as follows: the expletive does not appear when there is no gap in SpecIP, but when such a gap exists, the expletive is obligatory. If the fronted wh-phrases were able to land in SpecIP (as in Rudin’s analysis of Polish), this obligatory appearance of the expletive in the examples in (7) and (8) would be totally unexpected, since there would be no gap. In other words, the appearance of the expletive in the multiple fronting cases demonstrates that the multiply fronted wh-phrases are not adjoined to IP, but rather are attached higher up, at the CP level, as in Bulgarian. Furthermore, the fronted wh-phrases form a unit which cannot be broken up (i.e. by a parenthetical), a property which Yiddish shares with Bulgarian (Rudin 1988): (9) *Ver, nokh dayn meynung, vuhin vet geyn? who after your opinion where will go
This is consistent with the conclusion that Yiddish multiple fronting involves movement to CP. Before proceeding to the next section, in which I discuss some restrictions on multiple fronting, there are a few additional empirical facts worth noting. These pertain to the fact that there is some dialectal variation in Yiddish with respect to multiple fronting. The observations noted above — where multiple fronting is possible, but not obligatory in both direct and indirect questions —
55
56
Molly Diesing
hold for Southeastern Yiddish only. For many speakers of Central Yiddish, multiple fronting is only possible in indirect questions. In Northeastern Yiddish multiple fronting is not possible at all. I will return to these dialectal differences below in Section 6.
3. Some mysterious restrictions on multiple fronting In this section I discuss two situations in which multiple fronting cannot occur in Yiddish. In both the ban on multiple fronting is linked with the type of wh-phrases involved in the multiple question. In the first of these, discourselinking emerges as a factor which blocks multiple fronting. This was noted by Hoge (2000), who observed that “inherently D-linked” wh-phrases (such as “which N”, “whose N”, see the discussion in Comorovski 1996) are only grammatical in single-fronting questions (shown here in the (a) examples below), the multiple fronting counterparts are ungrammatical ((b) and (c)): (10) a.
Voser farshlog hot ver kritikirt? which proposal has who criticized ‘Who criticized which proposal?’ (Hoge 2000: 216) b. *Ver voser farshlog hot kritikirt? c. *Voser farshlog ver hot kritikirt?
(11) a.
Vemens hitl hot ver avekgeganvet? whose hat has who away.stolen ‘Who stole whose cap?’ b. *Ver vemens hitl hot avekgeganvet? c. *Vemens hitl ver hot avekgeganvet?
Constraints of a similar nature can be seen in other languages. In some multiple fronting languages there is in fact at least a preference for leaving D-linked wh-phrases in situ (Pesetsky 1987: 128, note 32). But those examples (from Polish and Romanian) all involve two D-linked wh-phrases, one of which is fronted. As far as I know, the Yiddish restriction is the only such case which applies to questions containing a single D-linked wh-phrase.3 A fact worth noting is that the single-fronted versions do not show Superiority effects. I will return to this later, and will show that the ban on D-linked wh-phrases actually follows from the analysis of the Superiority facts that I develop. A second class of wh-phrases which cannot appear in multiple fronting constructions is that of the “true adjuncts” — such as ‘why’ and ‘how’ (in
On the nature of multiple fronting in Yiddish
distinction from “quasi-argumental” adjuncts such as “where” and “when”, I will return to this difference later in the paper). While these can appear in single fronting multiple questions (without showing Superiority effects), they cannot front in a multiple-fronting question: (12) a.
Vi azoy hot ver gezungen? how has who sung ‘Who sang how?’ b. Ver hot vi azoy gezungen? c. *Ver vi azoy hot gezungen? d. *Vi azoy ver hot gezungen?
(13) a.
Farvos hot ver ongeklungen der mamen? why has who called the mother ‘Who called the mother why?’ b. Ver hot farvos ongeklungen der mamen? c. *Ver farvos hot ongeklungen der mamen? d. *Farvos ver hot ongeklungen der mamen?
While a similar claim has been made with regard to Bulgarian (see Rudin 1988: 476, fn. 21), Boškovic´ (1998a) shows quite conclusively that adjuncts can in fact appear in multiple fronting questions, unlike Yiddish. This prohibition on adjuncts is a rather striking restriction, given that ‘why’ and ‘how’ resist being left in situ in multiple questions in a number of other languages, such as English (see Reinhart 1998, Hornstein 1995, among others). I will offer an explanation of this restriction below, when I discuss the analysis of the so-called Superiority effects in Yiddish. Before proceeding to the discussion of the analysis of multiple fronting, however, I will first examine more closely the behavior of the two multiple question strategies with respect to Superiority.
4. The two question strategies and Superiority One of the most striking properties of multiple questions in Yiddish is that the two types of multiple questions (single fronting and multiple fronting) show rather different behavior with respect to Superiority (see also Hoge 2000). In one case there is a seeming immunity to Superiority: multiple questions formed by fronting a single wh-phrase do not show Superiority effects, patterning in this regard like German (Haider 1986). For example, extracting a subject over an object is grammatical (as is the reverse pattern):
57
58
Molly Diesing
(14) a.
Veri hot ti vos gekoyft? who has what bought ‘Who bought what?’ b. Vosi hot ver ti gekoyft?
Superiority effects are also not seen in interactions between object wh-phrases with wh-adverbials: (15) a.
Vos hot Maks vi azoy/farvos geshpilt? what has Max how /why played ‘How/why did Max play what?’ (Hoge 2000) b. Vi azoy/farvos hot Maks vos geshpilt?
Multiple questions involving a subject wh-phrase and a wh-adverbial also fail to show any Superiority effects: (16) a.
Ver hot vi azoy/farvos geshpilt pyane? who has how /why played piano ‘Who played the piano how/why?’ (Hoge 2000) b. Vi azoy/farvos hot ver geshpilt pyane?
There are also no “pure” Superiority effects. (17) a.
Vemen hot der lerer t vos geheysn leyenen? whom has the teacher what ordered read ‘Who did the teacher tell to read what?’ (Hoge 2000) b. Vos hot der lerer vemen geheysn leyenen t?
In all of the examples given above, a pair-list reading results regardless of which wh-phrase moves first (as is the case in German). In other words, the Superiority-violating order of wh-movement does not induce a single-pair reading, as has been claimed to be the case for English multiple questions involving D-linked wh-phrases and other languages (Barss 2000). Boškovic´ (1998b) links the loss of the single-pair possibility to movement of a wh-phrase to SpecCP, somewhat as a diagnostic indicator of true wh-movement, as opposed to mere wh-fronting. In the analysis in the final section of this paper I will claim that this view must be revised slightly to account for the semantic behavior (e.g. the pair-list reading) in matrix single fronting questions in Yiddish, though I believe that the necessary revision is still consistent with Boškovic´’s original idea. Turning now to the multiple fronting cases, these clearly do show Superiority effects, patterning very much like Bulgarian. Thus, we see that while the subject wh-phrase can precede the direct object wh-phrase in the initial group of wh-phrases, the reverse order is strongly ungrammatical:
On the nature of multiple fronting in Yiddish
(18) a.
Ver vos hot gekoyft? who what has bought ‘Who bought what?’ b. *Vos ver hot gekoyft? c. Ikh veys nit ver vos es hot gekoyft. I know not who what expl has bought. ‘I don’t know who bought what.’ d. *Ikh veys nit vos ver es hot gekoyft.
Similar contrasts are seen in multiple questions involving a subject and an indirect object. The Superiority effect is in evidence regardless of whether the indirect object was marked simply with the dative case (a–b examples), or with a preposition (c–d examples): (19) a.
Ver vemen hot gegebn khanike-gelt? who whom has given Hanukkah-money ‘Who gave whom Hanukkah money?’ (Hoge 2000) b. *Vemen ver hot gegebn khanike-gelt? c. Ver tsu vemen hot geshikt leshonetoyves? who to whom has sent Rosh Hashanah cards ‘Who sent Rosh Hashanah cards to whom?’ d. *Tsu vemen ver hot geshikt leshonetoyves?
Finally, questioning of double objects shows that a dative-marked indirect object is superior to the direct object: (20) a.
Vemen vos hot Maks gegebn? whom what has Max given ‘What did Max give (to) whom?’ b. *vos vemen hot Maks gegebn?
The prepositional dative, on the other hand, shows the reverse Superiority relation, the direct object being superior (as is consistent with the results of Barss and Lasnik 1986): (21) a.
Vos tsu vemen hot Maks geshikt? what to whom has Max sent ‘What did Max send to whom?’ b. *Tsu vemen vos hot Maks geshikt?
Briefly summarizing the facts presented so far, I have shown that Yiddish has both single and multiple wh-fronting in both main and embedded clauses. Where both
59
60
Molly Diesing
options exist, they appear to be more or less interchangeable, in that neither is associated with any special meaning. I also claimed, on the basis of the word order in embedded questions, that multiply fronted wh-phrases are in SpecCP — or some position higher than topicalized phrases. This gives Yiddish a preliminary classification as a “Bulgarian-style” multiple fronting language. There are, however some restrictions on multiple fronting. Wh-adverbials and inherently D-linked wh-phrases cannot occur in multiple-fronting constructions. Also, multiple fronting shows Superiority effects, while single fronting does not. Interestingly, pair-list readings result in both types of questions in all cases. Finally, although D-linking is generally considered to “override” Superiority in some way (Pesetsky 1987), it clearly does not do so in Yiddish multiple fronting constructions, since D-linked wh-phrases are not even allowed. In the next section I take up the question of how to explain the peculiarities of the syntax of Superiority in Yiddish.
5. On the asymmetry of Superiority in Yiddish Clearly the most striking fact about Superiority in Yiddish is its asymmetric nature. Superiority effects seem to only appear with multiple fronting. Single fronting questions are mysteriously immune. This then is the main question to be answered: why does Yiddish show Superiority effects only with multiple fronting? Before embarking on the search for an answer, it is useful to review some current views on just what the Superiority effect is. In many recent works (for example Kitahara 1993, Boškovic´ 1997, Grohmann 1997a, and also Richards 1997, 2001), Superiority is characterized as a sub-case of the “Attract Closest” condition (Chomsky 1995), an economy condition which restricts phrasal movement more generally:4 (22) Attract Closest α can attract K only if there is no legitimate operation Move β targeting K, where β is closer to K.
Applied to a multiple question involving a subject wh-phrase, this principle yields only one legitimate derivation. The subject wh-phrase is the one closest to CP; therefore, as required by Attract Closest, it moves first. The remaining wh-phrase(s) move afterwards and “tuck-in” under the first-moved wh-phrase (an idea due to Richards 1997), as this “tucking-in” movement yields the shortest moves for the subsequent wh-phrases. While the “Attract Closest”
On the nature of multiple fronting in Yiddish
approach appears to be perfectly compatible with the Yiddish multiple fronting cases, it is less obvious that it can work for the single fronting cases, where Superiority effects are absent. Richards (1997: 90ff.) also provides an account of the absence of Superiority effects in German, based on the fact that German allows scrambling (a similar suggestion is made in Grohmann 1997b). In this analysis, the wh-phrases undergo scrambling prior to wh-movement (perhaps adjoining to VP, or landing in the specifier of an inflectional head, here indicated as ?P), and then the wh-phrases (or features, in a single-fronting language) undergo movement to check wh features. The scrambling has the effect of bringing a “lower” wh-phrase closer to CP prior to wh-movement. This effectively reorders the Superiority relations, or even voids them by rendering the scrambled phrases “equidistant” (but see Hoge 2000 for criticisms of this approach). Since Yiddish does allow scrambling (Diesing 1997), it is tempting to apply a solution of this sort to Yiddish single fronting. The example below illustrates a derivation exploiting scrambling as an “escape” mechanism: (23) a.
Vos hot ver gekoyft? what has who bought ‘What did who buy?’ b. [CP vosj hot [?P tj [VP ver tj gekoyft]]]
Crucially, this strategy exploits the fact that scrambling in Yiddish itself does not show Superiority effects (Diesing 1997). The end result is that scrambling moves the object wh-phrase to a position where it is now “superior” to the subject wh-phrase, which allows subsequent wh-movement to proceed without a “real” Superiority violation. In other words, Superiority gets pre-empted by scrambling. But it is clear that this proposal is not without problems when applied to the full array of Yiddish data. The main difficulty is that in the multiple fronting cases, the pre-wh-movement scrambling must be blocked somehow, so that the Superiority effects will surface. It’s hard to imagine any economy condition (global or otherwise) that would permit scrambling in single-fronting cases but not in multiple fronting examples. Another concern with this line of analysis is whether the scrambling of wh-phrases is itself sufficiently well-motivated. I will return to this point later.5 Another analysis which attempts to account for “asymmetric Superiority” of a somewhat different nature is Boškovic´ (1999). Boškovic´ takes as a starting point the idea that not all wh-fronting is wh-movement. In this way he explains
61
62
Molly Diesing
the fact that Superiority seems to apply only to the first phrase moved in Bulgarian, subsequent wh-phrases can move in any order.6 Thus in a Bulgarian multiple questions, the first wh-phrase moves by wh-movement (moving to SpecCP). This movement is subject to Attract Closest, and thus shows Superiority effects. Subsequent wh-phrases undergo focus movement, which is not subject to Superiority. This raises the question of why there should be such a difference between wh-movement and focus movement. To explain this, Boškovic´ exploits the idea that heads can differ in how many times they attract a given feature (along the lines of a suggestion made in Chomsky 1995). When this idea is implemented, heads come in different “flavors”, depending on how many times they can attract a given feature. An Attract 1F head checks a feature only once, an Attract 2F head can check a feature twice, and an Attract AllF feature will attract all instances of the given feature. Boškovic´ claims that the [+Wh] head is of the Attract 1F type, and the [+Focus] head is Attract AllF. Furthermore, Attract AllF will be satisfied regardless of the order in which the attracted phrases move. In other words, movement to an Attract AllF head will not show any Superiority effects. It follows from this account that Bulgarian exhibits Superiority effects in multiple questions with two wh-phrases, since the Attract 1F [Wh] must attract the closest wh-phrase. With more than two wh-phrases, the relative order of the wh-phrases below the topmost one is expected to be free. This is in fact the case in Bulgarian. But can the Boškovic´ explanation be applied to Yiddish? The answer appears to be both yes and no. Portions of the analysis apply, but they do not yield an account of the asymmetric Superiority effects seen in Yiddish. In showing this, I address first the initial question of whether Yiddish shows any evidence of focus movement. Here the answer is yes, as shown by the examples below: Ver vet vuhin geyn mit aykh? who will where go with you? ‘Who will go where with you?’ b. Ver hot vos gekoyft? who has what bought ‘Who bought what?’
(24) a.
As I noted earlier, the unmoved object wh-phrase appears to the left of the verb, despite the fact that the normal VP order in Yiddish is VO. As mentioned above, in Diesing (1997) I proposed an explanation of this fact in terms of a preverbal focus position, which can (and must) host wh-phrases which have not been fronted.
On the nature of multiple fronting in Yiddish
It seems fairly straightforward that wh-phrases (which semantically function as focus) should move to a focus slot, just as appears to be the case in Bulgarian. But there are some features of this “focus movement” that make it look very different from the Bulgarian focus movement. First, the focus position is immediately before the verb. It is really not high up enough in the tree to be “standing in” for multiple fronting as focus movement does in Bulgarian: Ver vet haynt vuhin geyn mit aykh? who will today where go with you? ‘Who will go where with you today?’ b. Ver hot nekhtn vos gekoyft? who has yesterday what bought ‘Who bought what?’ c. Maks hot nekhtn a bukh geleyent. Max has yesterday a book read ‘Max read a BOOK yesterday.’
(25) a.
As these examples show, the focus position is to the left of the verb, but to the right of adverb. As the example (25c) demonstrates, contrastively focused nonwh-phrases can move there too, given the right context. It is important to bear in mind that this leftward movement to the “focus slot” is not the same as scrambling. The essential features of scrambling in Yiddish are that it moves phrases past the adverb, and does not require heavy contrastive focus, but does require definiteness/specificity: (26) a. *Maks hot a bukh mistome /nekhtn geleyent. Max has a book probably/yesterday read ‘Max has probably/yesterday read a book.’ b. Maks hot dos bukh mistome /nekhtn geleyent. Max has the book probably/yesterday read ‘Max has probably/yesterday read the book.’
None of these properties hold of movement to the preverbal focus slot. Another prominent property of the preverbal focus position is that it is limited to only one element: (27) a. *Nekhtn hot Maks nit dem yingl dos bukh gegebn. yesterday had Max not the boy the book given ‘Yesterday, Max didn’t give the boy the book.’ b. *Nekhtn hot Maks nit ken yingl ken bukh gegebn. yesterday had Max not no boy no book given ‘Yesterday, Max didn’t give no boy no book.’
63
64
Molly Diesing
While this property makes it clear that the Yiddish focus movement is not really equivalent to the focus movement Boškovic´ has exploited in his analysis of Bulgarian multiple wh-fronting, it does provide an explanation for one of the odd facts about Yiddish multiple questions I noted earlier: the fact that they are limited to two wh-phrases. If in fact VP-internal wh-phrases must move to the focus position at some point in the derivation, and that position can host only one element, then there can be only two wh-phrases total in any sentence.7 This property of Yiddish is somewhat reminiscent of the explanation offered by Calabrese (1984) for the absence of multiple questions of any kind in Italian. Calabrese claims that Italian allows only one phrase to be in focus, and can therefore allow only one wh-phrase (since wh-phrases are inherently focused). In Yiddish, however, it is clear that such a “one only” restriction applies only to in situ wh-phrases. It is possible that the two sets of data — Italian and Yiddish — could be brought under one explanatory umbrella with a more nuanced view of the notion of “focus” (such as Rooth 1992 or É. Kiss 1998, to offer two rather different alternatives). For now, I will leave this as a matter for future research.8 In any case, it will be hard to exploit the Yiddish preverbal focus position in explaining the multiple fronting cases in Yiddish — at the very least the [Focus] head can only be Attract 1F, in Boškovic´’s terms. While in Yiddish this may not be a problem (given that we only see two wh-phrases in multiple questions), the focus head is also situated too low in the tree to serve as a plausible landing site for fronted wh-phrases. In short, Yiddish is not like Bulgarian after all. At this point I will shift my focus to consideration of another question, that of what exactly enables multiple fronting. Here there are a couple of options. One is based on the idea that CPs can come in two varieties: [±multiple specifiers]. A [+multiple] CP allows multiple instances of phrasal movement, yielding multiple wh-movement. A [−multiple] specifier allows only one phrase to move overtly, the non-fronted wh-phrases check features by feature movement or by the operation of Agree (Chomsky 2000). (This is clearly related to the Attract 1F and Attract AllF idea discussed above.) In an analysis along these lines Yiddish chooses between two types of CPs: [+multiple specifiers] and [−multiple specifiers] (cf. Richards 1997 and Pesetsky 2000). Another alternative is to have multiple landing sites. In this case the “fine structure” of the Yiddish periphery would contain two inflectional heads, one which allows multiple specifiers and the other which doesn’t (an analysis of multiple fronting in Slavic of this type can be found in Grohmann 2000, following work on the structure of the clausal periphery by Rizzi 1997, this idea also has a certain kinship with the Boškovic´ proposal discussed earlier).
On the nature of multiple fronting in Yiddish
In the end, the analysis of Yiddish will require an amalgamation of both of these approaches. I will start by assuming that Yiddish multiple wh-fronting does in fact involve movement to a [+multiple] CP. At this point, the analysis is simply a minor modification of that in Pesetsky (2000) for Slavic multiple fronting. Further modification will be needed to account for the data regarding Superiority in Yiddish. To derive the Yiddish Superiority facts, I add an additional constraint on the [+multiple] CPs: (28) Yiddish Multiple Fronting Constraint In Yiddish, movement to a [+multiple] CP must originate from an A-position.
In conjunction with the Attract Closest constraint, this will in fact enforce the Superiority constraint in the multiple fronting cases, since this blocks scrambling before multiple wh-movement, which would void Superiority effects if it took place, as shown in the discussion above (assuming that this scrambling is in fact movement to an A¢ position, for arguments to this effect see Diesing 1997).9 At first blush, the Yiddish Multiple Fronting Constraint may seem somewhat ad hoc, but a closer look reveals that it can also explain the other peculiar properties of Yiddish multiple fronting constructions. First, it also prevents adjuncts from participating in multiple-fronting constructions, explaining the otherwise mysterious ban on ‘why’ and ‘how’. This now follows from the simple fact that such adjuncts do not appear in A-positions, and thus are blocked by the Yiddish Multiple Fronting Constraint from moving to a [+multiple] CP. They are therefore excluded from multiple fronting constructions.10 Furthermore, the Yiddish Multiple Fronting Constraint provides a possible explanation for the fact that D-linked wh-phrases cannot participate in multiple-fronting constructions. In this case, I exploit the idea that inherently D-linked whs are in some sense [+specific] (they involve reference to a set of individuals previously established in the discourse, see É. Kiss 1993 and Comorovski 1996), and must scramble out of VP (as argued in Diesing 1997). As a consequence of such scrambling, they move to an A¢ position before wh-movement takes place, and consequently are ineligible to move to a [+multiple] CP. At this point, a syntactic analysis based on the preceding discussion is conceivable. In Diesing (2001) I assumed that Yiddish allows both [+multiple] and [−multiple] CP-specifiers, and that Yiddish also allows scrambling. I also made the assumption that Yiddish is subject to a “multiple fronting constraint”
65
66
Molly Diesing
in that movement to a [+multiple] CP-specifier must be from an A-position, while movement to a [−multiple] CP-specifier is unrestricted. This led to the result that scrambling can mask Superiority effects, so that they are not seen in single fronting. Finally, while a focus position does play a role in Yiddish questions, it is clearly not the same as claimed by Boškovic´ for Bulgarian and Serbo-Croatian. Such a proposal raises a number of questions regarding the nature of Superiority and the multiple fronting constraint. Specifically, what is the connection between the two concepts? Can Superiority be simply reduced to a requirement that the fronted wh-phrases and the A-positions have to be isomorphic? Or is the Yiddish Superiority effect seen in multiple fronting just a “pseudo-Superiority effect” induced by the Yiddish Multiple Fronting Constraint, with English Superiority effects being something different? In other words, how general is the Yiddish Multiple Fronting Constraint? These questions are necessarily quite speculative, but in the next section I examine some further data from dialectal variation within Yiddish and also single fronting in embedded clauses that may shed some light on these issues.
6. Towards a final analysis of Yiddish multiple questions As I noted above in Section 2, there are three major dialect divisions within Yiddish with respect to multiple fronting. For the speakers of the dialect I have been discussing so far (Southeastern Yiddish) multiple fronting is possible, but not obligatory, in both direct and indirect questions. There are also speakers for whom multiple fronting is only possible in indirect questions (perhaps coterritorial with Central Yiddish), and a third group of speakers for whom multiple fronting is not possible at all (many speakers of Northeastern Yiddish fall into this category). In what follows, I will label the three dialects by the rough geographic associations (corresponding to the three main dialects recognized in the Yiddish literature, see Katz 1983), though it is by no means clear that all speakers of a given geographic category will share the same judgments regarding multiple fronting. As an initial attempt at characterizing the dialectal variation, it is possible to represent the differences in terms of the availability of the [+multiple] CP option: (29) a. Southeastern Yiddish: freely selects [+multiple] or [−multiple] CP b. Central Yiddish: allows [+multiple] CP only in embedded contexts c. Northeastern Yiddish: only has [−multiple] CP
On the nature of multiple fronting in Yiddish
I will instead consider an alternative, which is more in line with my earlier work on Yiddish word order (Diesing 1990), in which matrix single fronting is not due to wh-movement to SpecCP, but to SpecIP, while in embedded clauses wh-phrases always move to CP. In my earlier paper this claim was justified by certain word order differences between matrix and embedded questions — specifically, that the latter allows co-occurring topicalization while the former does not. In addition to being consistent with that earlier result, an advantage of taking this approach here is that we don’t necessarily have to resort to the masking effects of scrambling to explain the absence of Superiority effects. Having a distinct landing site for matrix single fronting allows us to consider another possibility — that fronting to SpecIP is not subject to Superiority, perhaps as a consequence of the somewhat mixed nature of the position, hosting both A and A¢ elements (Diesing 1990). Assuming this to be the case, here’s the alternative dialect characterization spelled out: (30) a.
Southeastern Yiddish: freely allows [+multiple] CPs in main and embedded clauses, matrix single fronting results from movement to ([−multiple]) IP b. Central Yiddish: allows [+multiple] CP only in embedded contexts, single fronting to IP c. Northeastern: only has [−multiple] CP in embedded clauses
Within a theory of feature checking, there is still the question of how the wh-features get checked in a matrix question, since the wh-phrase does not move to CP in the syntax. One possibility is simply that wh-features can also be checked as a consequence of movement to the SpecIP position. But while it’s clear that some feature associated with SpecIP is motivating the movement of the wh-phrase, it is not at all clear that it must (or even can) be the wh-features. In a sense, allowing wh-features to be checked in IP would amount to simply relabeling CP as IP — a rather strange (and perhaps trivial) move. I will therefore assume that the wh-phrase moves to check some discourse feature in IP.11 But then how do the wh-features get checked? An alternative is to follow the lines of a proposal made by Boškovic´ (2000) for Serbo-Croatian and French, and say that a phonologically null C is inserted at LF. Such LF-insertion will be dependent on a wh-phrase having moved to SpecIP, or else we would see cases of well-formed matrix questions with no overtly moved wh-phrases, as in French. This restriction can be derived in terms of relativized minimality, providing that topicalization to SpecIP is A¢ movement (as claimed in Diesing 1990). If feature movement is regarded as head movement, the head of IP will
67
68
Molly Diesing
be an A¢ head intervening, if LF feature checking is accomplished by covert phrasal movement, the SpecIP will be an A¢ intervener. Under either view of LF movement, the contents of IP will be the closest candidates for attraction to C, and therefore LF insertion of a [+Wh] will only be possible when a wh-phrase has moved to SpecIP.12 One obstacle to this tactic appears to be the fact that single fronting (to IP) is allowed in embedded clauses in Southeastern Yiddish, but a crucial feature of Boškovic´’s analysis is the rather natural assumption that inserting C at LF can only happen at the top of the tree, since any other insertion would be countercyclic. What’s needed at this point is a closer examination of single fronting in embedded contexts. The earlier example, repeated here (as (31a)), actually shows a case of single fronting in an embedded clause which obeys Superiority, while its multiple fronting counterpart in (31b) is also grammatical. The crucial question is whether embedded single fronting must obey Superiority. Example (31c) below shows an example of single fronting which does not obey Superiority, and as indicated, consultants find this sentence considerably less acceptable that the Superiority-compliant (31a). Hot zi ni(sh)t gekent farshteyn [ver es shlogt zikh has she not been.able understand [who exp hits self mit vemen]. with whom ‘(So) she couldn’t understand who was fighting with whom.’ b. Hot zi ni(sh)t gekent farshteyn [ver mit vemen es shlogt zikh]. c. ??Hot zi ni(sh)t gekent farshteyn [mit vemen ver shlogt zikh]. d. *Hot zi ni(sh)t gekent farshteyn [mit vemen ver es shlogt zikh].
(31) a.
However, (31c) is not nearly as bad as violating Superiority in a multiple fronting context, as in (31d). In these examples it is possible that the Superiority violating single fronting sentence is simply too similar to its multiple fronting counterpart (differing only in that the latter has the expletive es ‘it’ in SpecIP). Examples which do not have a subject wh-phrase allow us to control for this factor: (32) a.
Ikh veys vemen Maks hot vos geshikt. I know whom Max has what sent ‘I know whom Max sent what.’ b. ??Ikh veys vos Maks hot vemen geshikt? c. Ikh veys vemen vos Maks hot geshikt? d. *Ikh veys vos vemen Maks hot geshikt?
On the nature of multiple fronting in Yiddish
Example (32a) shows embedded single fronting which obeys Superiority. As expected, this sentence is grammatical. The Superiority-violating variant in (32b), however, is significantly degraded, though again it appears that it is not judged to be quite as bad as the Superiority-violating multiple fronting case in (32d). Note that unlike the case with multiple fronting, adjuncts are not ruled out, and a Superiority effect is seen: (33) a.
Ikh veys farvos Maks hot vos gekoyft. I know why Max has what bought ‘I know why Max bought what.’ ?? b. Ikh veys vos Maks hot farvos gekoyft.
While these data are not crystal-clear, I will tentatively conclude that there is after all a Superiority effect in embedded single fronting, which is distinct from the Multiple Fronting Constraint in that wh-phrases can be fronted from nonA-positions so long as Superiority is respected. If this is the case, then Yiddish may in fact share some properties with Serbo-Croatian. Boškovic´ (1998b, 2000) argues that Serbo-Croatian only shows Superiority effects with movement to SpecCP, which occurs only in embedded clauses — the first wh-phrase moves to SpecCP, subsequent wh-phrases move to an Attract AllF focus position. Matrix questions simply involve multiple fronting to a focus position, with CP not being involved until it is inserted at LF. Boškovic´ further suggests this type of interpretation can serve as a diagnostic of “true” wh-movement to CP. Movement to SpecCP will result in only a pair-list interpretation, while movement to the focus position results in a singlepair interpretation being possible. Boškovic´ derives this contrast utilizing the semantics for wh-in situ in Hagstrom (1998). On this analysis, single-pair answers result when the Q-morpheme (an existential quantifier over choice functions) required for an interrogative interpretation is generated above the wh-phrases, so that the wh-phrases are within the scope of the choice function: (34) C [Q [Wh1 … Verb … Wh2]]
Both wh-phrases are within the domain of the choice function. Wh2 will yield a set of properties (corresponding to the verb taking a possible object), and this set of properties is then applied to each member of the set associated with Wh1. If the verb in (34) is “likes”, we end up with a set of propositions of the form: (35) {X likes a, X likes b, … Y likes a, Y likes b, …}
69
70
Molly Diesing
According to Hagstrom, where the meaning of a question is a set of propositions, a single answer results. A pair-list answer results when one wh-phrase is outside of the scope of the Q morpheme. Boškovic´ claims this happens with “true” wh-movement to C. If the Q morpheme is generated above both wh-phrases, its presence should block wh-movement by some version of relativized minimality. Thus, when a wh-phrase is moved to C, the Q-morpheme is merged with the lowest wh-phrase: (36) Wh1 C [t Verb Wh2+Q]
Interpretively, here the choice function takes Wh2 (a set of individuals) as its argument, and returns an individual which serves as the object to the verb. Wh1, is not subject to the choice function and therefore has a set of individuals as its value. The parts of the meaning are combined by flexible functional application to yield a set of sets of propositions (for a more detailed description of the semantic composition, see Hagstrom 1998) — a set for each value in the set denoted by Wh1. Assuming Wh1 to be “who”, Wh2 to be “who”, and the verb to be “likes”, we end up with: (37) {{A likes f1 (WHO), A likes f2(WHO), …} {B likes f1 (WHO), likes f2(WHO), …}, …}
This reduces to a set of questions: {Who does A like?, Who does B like?, …}, which in turn corresponds to a pair-list answer in Hagstrom’s system. Returning now to Yiddish, I have claimed above that the Superiorityavoiding cases of wh-movement result from movement to some position below SpecCP. What is interesting is that these questions still have the properties of “true questions” — e.g. the pair-list interpretation. However, this is not necessarily inconsistent with Boškovic´’s (1998b) analysis of the loss if the pairlist reading in Serbo-Croatian; all that is necessary is to assume that even in the case of wh-fronting to SpecIP the Q-morpheme must attach to the lower wh-phrase.13 Movement to [+multiple] CP, on the other hand is really multiple wh-movement (not merely wh-fronting) and must obey the Yiddish Multiple Fronting Constraint. This stricter-than-Superiority restriction explains the strange behavior of adjuncts and D-linked wh-phrases. Movement to a [−multiple] CP (as in embedded questions) appears to obey Superiority, but since adjuncts and D-linked wh-phrases are not ruled out in these cases, it appears that “Superiority” is in fact a distinct, weaker constraint than the Yiddish Multiple Fronting Constraint. If this is true, this is in agreement with certain
On the nature of multiple fronting in Yiddish
aspects of Boškovic´’s analysis of Serbo-Croatian, namely that movement to CP obeys Superiority in some form, either “traditional” Superiority or the stricter Multiple Fronting Constraint. Deriving the other dialects is straightforward. Central Yiddish allows both types of CP, [±multiple], in embedded clauses, but only inserts C at LF in matrix clauses. This yields multiple fronting only in embedded clauses. Finally, Northeastern Yiddish does not permit the [+multiple] CP at all. As far as the overall typology of multiple fronting is concerned, Yiddish is a hybrid of sorts. It certainly does not fit neatly into a binary typology of the sort based on Rudin (1988), simply because more than one multiple question strategy exists. It does support the view that differences in wh-movement can correspond to differing properties of CP, as espoused by Pesetsky (2000) and others, but the different patterns of Superiority judgments also indicate a need for different landing sites for wh-fronting, as argued by Grohmann (2000) as well as Boškovic´ in several papers. Interestingly, Haider (2000) also reaches the conclusion that a uniform, or mono-causal, approach to Superiority effects is untenable, based on crosslinguistic differences between German and English. His account includes structural factors as well as conditions on the syntax/semantics interface. The relevant structural constraint involves the position of the in situ wh-phrase: it must be minimally included by a lexical projection in order to be c-commanded by a non-superior wh-element. This constraint could explain the contrast between matrix single fronting and multiple fronting in Yiddish — if one construed “in situ” to include any wh-phrase to the right of the topmost one. Thus, in a multiple fronting sentence, the right hand wh-phrase is necessarily not included in a lexical projection, hence it is subject to Superiority. Apart from whether this extension of Haider’s idea is really valid, this still leaves the Superiority effects that surface in embedded single fronting cases a mystery. Overall, I do believe that the Yiddish data support an analysis which focuses on the landing site of the moved wh-phrases as the crucial parametric determinant of Superiority, rather than the position of an in situ wh-phrase. Putting these observations all together, the Yiddish picture looks likes this: (38) a.
[+multiple] CP: multiple wh-movement, subject to the Multiple Fronting Constraint b. [−multiple] CP: single wh-movement, subject to Superiority c. no CP: single wh-fronting to IP, not subject to Superiority
71
72
Molly Diesing
While this may look like a “have your cake and eat it too” sort of solution, in that it exploits both the [±multiple] feature checking and distinct landing site strategies, it’s not clear that this possibility is necessarily unexpected or ruled out under any of the earlier accounts, and this solution has the virtue of providing an explanation of the dialectal variation in a relatively elegant manner.
Notes * Portions of the research of which this paper forms a part were presented at the SULA Conference at the University of Massachusetts at Amherst in May 2001, and in a colloquium at the University of Connecticut in April 2002. I am grateful for the comments and suggestions I received from these audiences, as well as conversations and correspondence with Chris Collins, Heidi Harley, Andrew Carnie, and Kerstin Hoge, as well as comments from Kleanthes Grohmann and Cedric Boeckx. Finally, I’d also like to thank my consultants, Aaron Lerner, Rose Rosen, and Harry Zuckman (sadly, all now deceased). 1. Russian actually constitutes a third type which never shows Superiority effects (see Stepanov 1998). 2. The validity of the wh-islands diagnostic is not clear. In particular, Boškovic´ (1998a) shows that adjuncts are prohibited from moving out of islands even in Bulgarian. In any case, the island diagnostic does not play any role in this paper. See also Boškovic´’s contribution to this volume for more on the issue of islands in multiple-fronting. 3. Irene Heim (p.c.) has raised the question of whether this is in fact an effect of D-linking, or whether perhaps “heaviness” may in fact be at stake. Testing heavier wh-phrases which are not clearly D-linked yields somewhat unclear results: multiple fronting constructions containing wh-phrases such as vifl bikher “how many books” were accepted by some speakers, while wh-phrases such as vos far bikher “what kind of books” were less acceptable. Interestingly, one speaker (brought to my attention by Kerstin Hoge) has the exact opposite judgments. In any case, the clearest cases of ungrammaticality are those involving D-linked wh-phrases. 4. Of course, there are alternative explanations. See for example Hornstein (1995), Reinhart (1998), Hoge (2000), Grewendorf (2001), and Sabel (2001, this volume) for some of the possibilities. Hoge also provides a very insightful critique of the “attract closest” approach. 5. Grewendorf (2001) offers an account of the absence of Superiority effects in German based on the notion of “clustering” — wh-phrases adjoin to form a cluster and then move together (see also Sabel 2001, this volume). Hoge (2000) considers whether this approach may be fruitfully applied to Yiddish, but concludes (rightly, I believe) that it fails to account for the full range of restrictions on multiple-fronting in Yiddish. In particular, the absence of subject-adverbial and object-adverbial Superiority effects remains mysterious. 6. Richards (1997, 2001) analyzes these facts in terms of his “Principle of Minimal Compliance” (PMC) — Superiority only needs to be “satisfied” once, and thus applies only to the first phrase moved. I will not attempt to address the issue of which of these approaches is
On the nature of multiple fronting in Yiddish
correct. Note also that these effects cannot be tested in Yiddish, since multiple questions are limited to two wh-phrases. 7. Note that it is essential that this also apply to fronted wh-phrases that originate VP-internally, as the restriction to two wh-phrases also applies to multiple fronting, as well as to single fronting cases where an object wh-phrase has fronted but a subject wh-phrase has not. 8. I thank Kleanthes Grohmann for drawing this possibility to my attention. 9. Essential to this line of explanation is that the focus movement discussed above not be a case of A¢ movement, since this position also plays a limiting role in multiple fronting, which is also limited to two wh-phrases. More research is needed here, but this “focus movement” is in fact quite restricted, and does not appear to display any of the traditional properties of A¢ movement (see Diesing 1997 for further discussion). In any case, this shorter leftward movement may well correspond to cases of “short scrambling” which have in fact been claimed to be A-movement, in contrast to A¢, longer-distance scrambling (see vanden Wyngaerd 1989, Mahajan 1990, Bailyn 2002, among others). 10. This requires a different status for adjuncts such as “when” and “where”, since they do undergo multiple fronting (see example (5)), but there are numerous antecedents for making a distinction between “true” adjuncts (like “why” and “how”) and “quasi-argumental” adjuncts (“where” and “when”) (see for example Rizzi 1992, or Hornstein 1995). I will simply regard “when” and “where” as being closer to argument status, while “why” and “how” are purely adjuncts. 11. I will leave the exact nature of this feature vague. Grohmann (1998) has argued for associating wh-phrases with some sort of “topicality” or old information in certain cases (namely those where discourse-linking seems to be involved. A similar tactic is taken by Reglero (this volume) as well. This will not obviously carry over to the Yiddish data, since discourse-linking is not necessarily involved. Nor is it obvious that discourse-linking suffices to make an element “topic” (this of course depends on the precise notion of “topic”, see Prince 1999 for a discussion of the relevant issues). In any case, it seems more likely the feature is in some sense focus-related, though clearly distinct from the contrastive focus associated with the internal focus position discussed above in Section 5. As in the case of the focus position discussed earlier, here too a more fine-grained view of the discourse-related concepts such as “topic” and “focus” promises to yield a more satisfying explanation, though that is well beyond the scope of this paper. 12. Recent work (such as Rizzi 2003) has suggested that Relativized Minimality is sensitive features of “similar type”, not just A vs. A¢ status. As far as I can tell, the analysis proposed here could be translated into that sort of approach, with the relevant features falling into Rizzi’s “quantificational” class. 13. I thank Željko Boškovic´ for pointing this out to me.
73
74
Molly Diesing
References Bailyn, J. 2002. “On scrambling: A reply to Boškovic´ and Takahashi.” Linguistic Inquiry 32: 635–58. Barss, A. and Lasnik, H. 1986. “A note on anaphora and double objects.” Linguistic Inquiry 17: 347–354. Barss, A. 2000. “Minimalism and asymmetric interpretation.” In Step by Step. Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 31–52. Cambridge, Mass.: MIT Press. Birnbaum, S. 1978. Yiddish: A Survey and a Grammar. Toronto: University of Toronto Press. Boškovic´, Ž. 1997. “On certain violations of the Superiority condition AgrO, and economy of derivation.” Journal of Linguistics 33: 227–254. Boškovic´, Ž. 1998a. Wh-phrases and Wh-movement in Slavic. Ms., University of Connecticut, Storrs. Boškovic´, Ž. 1998b. “On the interpretation of multiple questions.” In A Celebration: Essays for Noam Chomsky’s 70th Birthday, J. Fodor, S. J. Keyser and A. Brand (eds). Cambridge, Mass.: MIT Press. [Online publication — available for download at http:// cognet.mit.edu/Books/celebration/essays/Boeckx.html (sic); revised version to appear in Linguistic Variation Yearbook 1.] Boškovic´, Ž. 1999. “On multiple feature checking: multiple wh-fronting and multiple head movement.” In Working Minimalism, S. D. Epstein and N. Hornstein (eds), 157–187. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 2000. “Sometimes in [Spec, CP], sometimes in situ.” In Step by Step. Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 53–87. Cambridge, Mass.: MIT Press. Boškovic´, Ž. This volume. “On wh-islands and obligatory wh-movement contexts in South Slavic.” 21–50. Calabrese, A. 1984. “Multiple questions and focus in Italian.” In Sentential Complementation, W. de Geest and Y. Putseys (eds), 67–74. Dordrecht: Foris. Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. 2000. “Minimalist inquiries: The framework.” In Step by Step. Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 89–155. Cambridge, Mass.: MIT Press. Comorovski, I. 1996. Interrogative Phrases and the Syntax–Semantics Interface. Dordrecht: Kluwer. Diesing, M. 1990. “Verb movement and the subject position in Yiddish.” Natural Language and Linguistic Theory 8: 41–79. Diesing, M. 1997. “Yiddish VP order and the typology of object movement in Germanic.” Natural Language and Linguistic Theory 15: 369–427. Diesing, M. 2001. Multiple questions in and about Yiddish. In The Proceedings of SULA: The Semantics of Under-Represented Languages in the Americas, J.-Y. Kim and A. Werle (eds), 29–37. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Grewendorf, G. 2001. “Multiple wh-fronting.” Linguistic Inquiry 32: 87–122.
On the nature of multiple fronting in Yiddish
Grohmann, K. K. 1997a. “German Superiority.” Groninger Arbeiten zur germanistischen Linguistik 40: 97–107. Grohmann, K. K. 1997b. German Superiority: A discussion of the cross-linguistic phenomenon and an approach to solve the German paradox. Paper presented at the 13th Comparative Germanic Syntax Workshop, Cornell University, Ithaca, N. Y. [July 1997] Grohmann, K. K. 1998. “Syntactic inquiries into discourse restrictions on multiple interrogatives.” Groninger Arbeiten zur germanistischen Linguistik 42: 1–60. Grohmann, K. K. 2000. Prolific peripheries: A radical view from the left. Doctoral dissertation, University of Maryland, College Park. Hagstrom, P. 1998. Decomposing questions. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Haider, H. 1986. “Affect α: A reply to Lasnik and Saito, ‘On the nature of proper government’.” Linguistic Inquiry 17: 113–125. Haider, H. 2000. “Towards a superior account of Superiority.” In Wh-Scope Marking, U. Lutz, G. Müller and A. von Stechow (eds), 231–248. Amsterdam: John Benjamins. Hoge, K. 2000. Superiority. Doctoral dissertation, University of Oxford. Hornstein, N. 1995. Logical Form: From GB to Minimalism. Oxford: Blackwell. Jacobs, N., Prince, E. and van der Auwera, J. 1994. “Yiddish.” In The Germanic Languages, E. König and J. van der Auwera (eds), 388–419. London: Routledge. Katz, D. 1983. Grammar of the Yiddish Language. London: Duckworth. É. Kiss, K. 1993. “Wh-movement and specificity.” Natural Language and Linguistic Theory 11: 85–120. É. Kiss, K. 1998. “Identificational focus versus information focus.” Language 74: 245–273. Kitahara, H. 1993. “Deducing Superiority effects from the shortest chain requirement.” In Harvard Working Papers in Linguistics 3, H. Thráinsson, S. Epstein and S. Kuno (eds), 109–119. Cambridge, Mass.: Department of Linguistics, Harvard University. Mahajan, A. 1990. On the A-/A¢-distinction. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Mark, Y. 1978. Gramatik fun der yidisher shprakh. New York: YIVO Institute for Jewish Research. Pesetsky, D. 1987. “Wh-in situ: movement and unselective binding.” In The Representation of (In)definiteness, E. Reuland and A. ter Meulen (eds), 98–129. Cambridge, Mass.: MIT Press. Pesetsky, D. 2000. Phrasal Movement and Its Kin. Cambridge, Mass.: MIT Press. Prince, E. 1999. “How not to mark topics: ‘topicalization’ in English and Yiddish.” Ms. University of Pennsylvania, Philadelphia. Reglero, L. This volume. “Non-wh-fronting in Basque.” 187–227. Reinhart, T. 1998. “Wh-in situ in the framework of the Minimalist Program.” Natural Language Semantics 6: 29–56. Richards, N. 1997. What moves where when in which language? Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Richards, N. 2001. Movement in Language. Oxford: Oxford University Press. Rizzi, L. 1992. Relativized Minimality. Cambridge, Mass.: MIT Press. Rizzi, L. 1997. “The fine structure of the left periphery.” In Elements of Grammar: Handbook of Generative Syntax, L. Haegeman (ed), 281–337. Dordrecht: Kluwer. Rizzi, L. 2003. “Locality and left periphery.” In Structure and Beyond: The Cartography of Syntactic Structures, Vol. 2, A. Belletti (ed). Oxford: Oxford University Press.
75
76
Molly Diesing
Rooth, M. 1992. “A theory of focus interpretation.” Natural Language Semantics 1: 75–116. Rudin, C. 1988. “On multiple questions and multiple WH fronting.” Natural Language and Linguistic Theory 6: 445–502. Sabel, J. 2001. Deriving multiple head and phrasal movement: The cluster hypothesis. Linguistic Inquiry 32: 532–547. Sabel, J. This volume. “Malagasy as an optional multiple wh-fronting language.” 229–254. Stepanov, A. 1998. “On wh-fronting in Russian.” In Proceedings of NELS 28, P. N. Tamanji and K. Kusumoto (eds), 453–467. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Travis, L. deMena. 1984. Parameters and the effects of word order variation. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. vanden Wyngaerd, G. 1989. “Object shift as an A-movement rule.” In MIT Working Papers in Linguistics 11: Proceedings of the Student Conference in Linguistics, P. Branigan, J. Gaulding, M. Kubo and K. Murasugi (eds), 256–271. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). Zaretski, A. 1929. Yidishe gramatik. Vilna: Vilner Farlag.
On the morphosyntax of wh-movement Marcel den Dikken The Graduate Center of The City University of New York
The central aim of this paper is to develop a general theory of wh-constructions which (i) allows us to accurately distinguish between wh-phrases as questionword phrases, echo-question phrases and indefinites in terms of a feature matrix involving the features [±Wh] and [±Focus], (ii) makes it understandable why wh-fronting targets SpecCP, SpecQP or SpecFocP depending on context, (iii) accounts for root/non-root asymmetries with respect to the landing-site of wh-fronting, and (iv) explains the differences and similarities between [+Wh] foci and [−Wh] foci.1 English, Dutch and Hungarian will serve as the key languages of reference, with English being identified as a language in which wh-fronting targets SpecCP in embedded clauses but SpecFocP in root wh-questions, driven throughout by a strong [+Wh] feature in Comp which, in agreement with an overarching hypothesis about the nature of strong features (due to Chomsky 1995: Ch. 4), is checked via overt-syntactic category movement unless [+Wh] is a feature of the root node.
1.
Five questions about wh-constituents, their features and their movement
1.1 Wh-fronting and the question/relative clause dichotomy Let me start by making some preliminary observations about wh-fronting in Hungarian (cf. also Benincà 2001 for discussion of Italian, and Boškovic´ 2002 for Slavic). Hungarian fronts its wh-phrases to the focus position in all questions (whether root or embedded; cf. Brody 1995 and references cited there). The focus position is relatively low in the structure, below topics and the complementizer hogy, as shown in (1). That wh-phrases front to SpecFocP in
78
Marcel den Dikken
Hungarian is evident not just from their placement vis-à-vis topics and complementizers, but also from their complementarity with fronted non-wh-foci in the same simple clause: (2a,b) are both ungrammatical; in general, tautoclausal multiple focus movement is impossible (cf. É. Kiss 1987). (1) a.
[TopP Marinak [FocP mit adott [TP János…]]]? Mari-dat what-acc gave János ‘What did János give to Mari?’ b. Kiváncsi vagyok [CP hogy [TopP Marinak [FocP mit I.wonder that Mari-dat what-acc adott [TP János…]]]] gave János ‘I wonder what János gave to Mari.’
(2) a. *Mit MARINAK adott János? what-acc Mari-dat(foc) gave János ‘What did János give to MARI?’ b. *MARINAK mit adott János?
But a focus inside a relative clause is perfectly legitimate, as shown in (3) (adapted from Lipták 2001: 97). (3) Bejöhet [DP az [CP aki [FocP PÉTERT hívta meg]]] may.come.in that who Péter-acc invited preverb ‘Those who invited Peter (as opposed to someone else) may come in.’
So wh-relative pronouns, unlike wh-question words, do not front to SpecFocP — not surprisingly, given their information-structural profile. But what wh-relative pronouns and wh-question words have in common is their possession of the morphological feature [+Wh]. It is this [+Wh] feature that is apparently the driving force for wh-fronting in relative clauses in Hungarian. But this [+Wh] feature ostensibly does not attract wh-constituents up to the C-domain in Hungarian questions. This raises our first question about wh-fronting: (Q1) What distinguishes relatives from questions such that wh is attracted to SpecCP overtly?
1.2 Wh-types and the distribution of the morphological feature [+Wh] Further questions quickly arise as we contemplate the notion that it is the morphological feature [+Wh] that drives wh-fronting. The morphological feature [+Wh] is common to the entire class of wh-constituents, simply as a
On the morphosyntax of wh-movement
reflex of their wh-morphology. This class also includes wh-constituents used as indefinites, and wh-phrases in echo questions. Dutch provides a particularly good illustration of all these types: (4) a.
b.
c.
d.
e.
f.
g.
Wat is er gebeurd? what is there happened ‘What happened?’ Er is wat gebeurd. there is what happened ‘Something happened.’ Is er wat gebeurd? is there what happened ‘Did something happen?’ Er is WAT gebeurd?! there is what happened ‘WHAT happened?!’ Wat er is gebeurd, is onduidelijk. what there is happened is unclear ‘What happened is unclear.’ Wanneer is er wat/WAT gebeurd? when is there what happened ‘When did something/WHAT happen?’ *‘What happened when?’ Gisteren is er *wat/WAT gebeurd? yesterday is there what happened ‘Yesterday, WHAT happened?’
All of these examples feature the word wat, the Dutch cognate of English what. In Dutch (as in German, but unlike in English), this wh-word doubles as an indefinite pronoun, meaning something like ‘something’ (cf. e.g. Postma 1994 and Bennis 1995 for discussion). But regardless of whether it is semantically interpreted as a question word or as an indefinite pronoun, its morphological composition is invariant. Put differently, if (as is standardly assumed — plausibly, in the light of its morphological composition) wat qua question word has the morphological feature [+Wh], then wat qua indefinite pronoun has this feature as well. And since it is specifically the morphological features of linguistic elements (and only those features) which are the driving force for movement (cf. Chomsky 1995: 197: “all morphological features must be checked somewhere, for convergence”), we now face the following question:
79
80
Marcel den Dikken
(Q2) Why do wh-indefinites and echo-wh’s fail to undergo wh-movement?
The paradigm in (4) tells us that it is not the case that morphologically [+Wh]marked constituents must wh-front — not even in simple questions (cf. (4c–d)). Nor is it the case that wh-fronting, whenever it does take place, categorically affects the closest morphologically [+Wh]-marked constituent — (4f), while (predictably) impossible on a multiple question interpretation (‘What happened when?’), is perfect on a reading in which wat is interpreted as an indefinite or an echo-question word. This leads us to a third question: (Q3) What counts as the closest possessor of [+Wh] when it comes to wh-fronting?
Skeptics of a [+Wh]-based approach to wh-movement might conclude at this point that this is sufficient evidence to give up on that kind of analysis altogether, and instead blame wh-fronting on a ‘question feature’ in C. But apart from the fact that such a feature would not in any obvious sense qualify as a morphological feature, an analysis of wh-movement along these lines would fail to countenance the fact that wh-fronting does not just take place in questions — relative clauses feature it, too (cf. (3) and (4e)). Moreover, the position to the left of the finite verb in a question is strictly reserved for constituents that are morphologically [+Wh]-marked (cf. (4g) vs (4f)). So there is good evidence that the [+Wh] feature plays a key role in wh-movement, but a sweeping statement to the effect that C attracts the closest [+Wh]-marked constituent to its specifier would be much too crude. 1.3 Structural complexity and inertia What we need to find is an intelligent way of preventing [+Wh] indefinite pronouns and echo-question words from being attracted by the [+Wh] feature of the probe. To say, as does Lipták (2001), that the [+Wh] feature is not a feature of the wh-word itself but is introduced by a Q-marker added onto wh-words used as question words, seems to me to be getting things backwards: plainly, the [+Wh]-morphology is a part of the wh-word itself, not of the Q-marker (which has no observable morphology at all); moreover, it is unclear how such an approach would make the desired distinction between ‘true’ question words and echo-wh’s. A more plausible approach would be to say that the syntactic structure of indefinite ‘what’ and echo-‘what’ is complex, such that the [+Wh] feature of
On the morphosyntax of wh-movement
the wh-word is too deeply embedded in them to be visible by an outside head. More technically, that feature would be embedded inside the DP phase, not visible ‘on the edge’. For indefinite ‘what’ such is altogether reasonable, in the light of the overt complexity of German (5a) and Hungarian (5b) (cf. Rudin 1988 on Bulgarian, Citko 1998 on Polish, and Scott 2002 on Russian; Scott notes that the non-wh-part of indefinitely used wh-pronouns is optionally omitted in spoken Russian): (5) a.
If for echo-‘what’ we can likewise uphold a syntactically complex representation in which ‘what’ is embedded (which I will show in the next section is indeed the case; cf. the [DP Q [who]] substructure in (7c), below), we get the desired result that the [+Wh] feature of the probe is perfectly capable of attracting a [+Wh] question word across a [+Wh]-marked indefinite pronoun or echo-‘what’: though wat in (4f) is closer to the probe than wanneer ‘when’, and wat possesses a [+Wh] feature, it nonetheless fails to interfere with the attraction of wanneer by the probe, simply because of the fact that wat’s [+Wh] feature is too deeply buried inside the complex noun phrase of which it is a part for the probe to be able to see it. This, in essence, is the answer to question (Q2). And the answer to (Q3) is obviously related to it: the closest attractee is the closest wh-phrase that has a [+Wh] feature which C can ‘see’. 1.4 Echoes and multiple questions In den Dikken and Giannakidou (2002), a syntactically complex analysis of echo-wh’s is defended on the basis of the fact that multiple wh-questions which normally receive a regular pair-list interpretation can only get a single-pair echoic reading when the fronted wh-constituent is a wh-the-hell expression: (6) a.
Who is in love with who? [pair-list or single-pair echo interpretation] b. ?Who the hell is in love with who? [single-pair echo interpretation only (marginal)]
81
82
Marcel den Dikken
The essence of the account of (6b) outlined in den Dikken and Giannakidou (2002) is this. Who the hell is a polarity item, which in (6b) needs to be licensed under c-command by the Q-operator in the matrix C. Such will succeed only if who the hell raises to a position lower than SpecCP — in the light of Hungarian (1), SpecFocP would be an obvious choice. But (2) tells us that wh-fronting to SpecFocP is incompatible with focus on some other constituent in the same clause. Now, in situ wh-constituents are always focused, hence need to be licensed in SpecFocP. So no matter what we do in (6b) to try to get the pair-list reading, things will go awry: if we take the in situ Wh to SpecFocP, we have no choice but to take who the hell to SpecCP, where it will not be able to be licensed by Q in C (cf. (7a));2 but if we take who the hell to SpecFocP and accomplish its licensing as a PI, we will not be able to license the focus feature of the in situ wh-constituent (cf. (7b)). This is why (6b) lacks a garden-variety pair-list interpretation. (7) a. *[CP [who the hell]i [CQ (…) [FocP [with who]j [Foc [TP ti is in love tj]]]]] fi wh-the-hell is not c-commanded by CQ, hence not licensable qua polarity item b. *[CP CQ (…) [FocP [who the hell]i [Foc [TP ti is in love with who]]]] fi in situ who is not licensable as a focus c. [CP CQ (…) [FocP [[who the hell]i [with [DP Q [who]]]j] [Foc [TP ti is in love tj]]]] fi single-pair echo reading
An echoic single-pair reading is nonetheless available for (6b), and here is how it is obtained. The idea is that echo-wh’s have a syntactically complex representation featuring a Q-morpheme and a wh-phrase. It is the Q-morpheme which is solely responsible for the fact that simple echo questions like John said WHAT? receive a question interpretation: in simple echo questions such as this, there is no Q-operator up in C (cf. the ban on PI licensing in simple echo questions: *John said WHAT to anyone?),3 so the entire burden of procuring an interrogative interpretation rests on the shoulders of the echo-wh itself. Now, the idea developed in den Dikken and Giannakidou (2002) is that a wh-phrase equipped with this Q-morpheme can team up with another wh-phrase in SpecFocP and form a single pair of wh-expressions together with it (cf. (7c)) — the result being a single-pair echo question, precisely the kind of creature that (6b) instantiates. In questions of the type in (6b), then, both wh-phrases are in SpecFocP — neither is fronted to SpecCP. On the non-echoic pair-list interpretation of (6a),
On the morphosyntax of wh-movement
on the other hand, the in situ Wh is uniquely associated with focus, the superior wh-phrase raising higher. So here we have yet another question: (Q4) Why do question-wh’s front to SpecFocP in some contexts and to a higher position in others, within the same language and within the same type of clause?
1.5 A root/non-root asymmetry We can compound the conundrum by drawing attention to root/non-root asymmetries that asserts themselves in the domain of wh-fronting in English. Thus, den Dikken and Giannakidou (2002) note that, while (6b) permits only a single-pair echoic reading, embedded (8b) is perfectly susceptible of a pair-list interpretation. (8) a. I {am wondering/would like to know} who is in love with who. b. I {am wondering/would like to know} who the hell is in love with who.
Of the two structures in (7a) and (7b), the latter will be ungrammatical in any context due to the fact that the in situ Wh cannot be associated to focus, as it should; but apparently, (7a), while bad in root contexts, converges in embedded ones. Den Dikken and Giannakidou’s (2002) account of this is that, in (8b), wh-the-hell is not dependent on the Q-operator in the embedded C-head at all: it can be licensed as a polarity item by a non-veridical licenser in the matrix clause (see the original paper for details). The fact, then, that wh-the-hell can raise to SpecCP only in embedded contexts falls out from its status as a polarity item. A second root/non-root contrast in the realm of wh-questions, unrelated to polarity properties, concerns the placement of fronted wh-constituents vis-à-vis fronted topics. As Pesetsky (1989) notes, topics (whose position is relatively constant cross-linguistically)4 precede wh-fronted constituents in root questions but follow them in embedded ones:5 (9) a. b.
?To Mary, what should we give? ?I don’t know what to Mary, we should give.
If we do indeed assume that to Mary is in the same position in (9a,b), then these examples tell us that the landing-site of wh-movement in English is different in root and embedded clauses, which leads us to formulate the following question: (Q5) Why do question-wh’s front to SpecCP in embedded wh-questions but to SpecFocP (modulo pair-list wh-questions like (6a)) in root wh-questions, in some languages?
83
84
Marcel den Dikken
Question (Q5) completes the checklist of fundamental questions about wh-constructions that I will address in this paper. 1.6 The feature composition of wh-constituents One thing we have established in the discussion so far is that a syntax of wh-movement which puts the blame on the morphological feature [+Wh] can be made to ‘overlook’ indefinite and echoic bearers of the feature such that it successfully picks out precisely the desired victims, while a syntax of wh-movement phrased in terms of a C-head with a ‘question feature’ in it will not only fail to be consistent with the minimalist dictum that only morphological features drive movement but will also leave relative clauses by the wayside. For the feature composition of wh-constituents, this discussion leads to the following typology: (10) a. b. c. d.
regular question words echo-question words indefinite wh-words relative wh-words
(single questions) [+Wh] not attractable [+Wh] not attractable
Of these four wh-word types, only (10a,b) are attractable to the focus position,6 and only (10a,d) are attractable by a [+Wh] feature in C. This already tells us why echo-wh’s and indefinite Wh’s will not undergo wh-fronting (in the strict sense of that term), and why they will not ‘count’ as closer potential attractees/ goals for attraction to [+Wh] C when they intervene between it and a questionwh lower down. But it does not tell us yet why wh-fronting of relative Wh’s always targets SpecCP, cross-linguistically; and it also does not have anything to say so far about the questions raised with respect to language-particular and clause-particular peculiarities of movement operations affecting regular question words. 1.7 Road map In what follows, the first thing I will do is argue that in all (non-echo) questions in all languages, there is a [+Wh] feature up in C. This is the topic of Section 2. In Section 3, I will subsequently address the root/non-root asymmetry seen in Section 1.5, from the perspective of the question of the strength of the [+Wh] feature in C and what its strength means when it comes to overt featurechecking movement. Section 4 then addresses the question of why there is no
On the morphosyntax of wh-movement
language variation when it comes to the movement of wh-constituents in relative clauses. And in Section 5 I finally turn to the question of whether the wh-phrase of an embedded single wh-question travels through SpecFocP or instead goes straight to SpecCP, advancing a (somewhat tentative) argument in favour of the latter hypothesis.
2. Attraction to Focus and [+Wh] Comp The C-head of a ‘true’ (i.e., non-echoic) wh-question has a [+Wh] feature. That [+Wh] feature does not end up attracting the wh-constituent up to its specifier position in English root single wh-questions (cf. (9a)), though it does in embedded wh-questions as well as in pair-list multiple wh-questions (cf. Section 3); similarly, in Hungarian wh-questions in general the [+Wh] feature in C fails to attract wh-constituents (cf. (1)–(2)). But there is evidence nonetheless that there is indeed a [+Wh] feature up in C in such questions, and that that feature does indeed get checked against the [+Wh] feature of the question word, via feature movement at LF. The evidence in question comes from an interesting intervention effect in Hungarian wh-questions, recently discussed in Lipták (2001), from which source all the Hungarian data in Section 2.1 are taken. After going through Lipták’s argument based on Hungarian, I will then proceed, in Sections 2.2 to 2.5, to discussing English attraction to focus against the background of the results of Section 2.1. 2.1 Covert feature movement to [+Wh] Comp — An intervention effect There is little doubt that wh-phrases front to SpecFocP in Hungarian wh-questions. They end up in precisely the same position which non-wh foci occupy, and they are in complementary distribution with non-wh foci (cf. (2)), in keeping with the general fact that multiple focus movement is impossible (cf. É. Kiss 1987). However, there is one peculiar way in which wh-foci behave differently from non-wh foci: while the latter are perfectly happy with a quantificational adverb to their left (as in (11a), where the adverb scopes over the focus), wh-foci cannot be preceded by such adverbs (cf. (12a)). (11) a.
Mindig PÉTERT hívtam meg. always Péter-acc invited pv ‘At all times, I invited PÉTER’.
(Hungarian)
85
86
Marcel den Dikken
b. PÉTERT hívtam meg mindig. Péter-acc invited pv always ‘Péter was the only one I invited all the time.’ (12) a. *Mindig kit hívtál meg? always who-acc invited pv b. Kit hívtál meg mindig? who-acc invited pv always ‘Who did you invite all the time?’
To account for the deviance of (12a), Lipták (2001: 81) offers the interesting proposal that the construction instantiates an intervention effect of the type discussed by Beck (1996), and more recently — and, crucially, with reference to LF feature movement — by Pesetsky (2000) as well. In particular, Lipták postulates a structure of the type in (13) for (12a). (13) *[CP C[+Wh] [DistP mindig [Dist [FocP kit[+Wh,+foc] [Foc[+foc] hívtál [TP pro … meg …]]]]]]
Here mindig is sitting in the specifier position of a Distributive Phrase (for distributive quantifiers; cf. Beghelli and Stowell 1997), harmfully intervening between the [+Wh] feature in the probe, C, and its target, the [+Wh] feature of the wh-phrase in SpecFocP. We know from the literature that intervention effects are specific to movement dependencies: the Agree relationship between C[+Wh] and kit[+Wh] is not obstructed, either by a phase boundary or by a closer match. So we conclude that [+Wh] C actually attracts the wh-constituent in SpecFocP. It does not seem to attract the wh-constituent in overt syntax, however: there is no category movement of the wh-phrase beyond SpecFocP. We are dealing, therefore, with a case of covert feature movement, instigated by C’s desire to get its uninterpretable [+Wh] feature checked. Feature movement will succeed if and only if there are no harmful interveners on the trajectory between the probe and the goal; since the quantificational adverb mindig is such a harmful intervener, however, feature movement fails, the uninterpretable [+Wh] feature of C cannot be checked, and the derivation crashes, as desired. The ungrammaticality of (12a), so construed, thus presents evidence for the presence of a [+Wh] feature in C in ‘true’ wh-questions — a feature which must be able to attract the [+Wh] feature of the closest wh-constituent. In (13) this fails, which is why (12a) is ungrammatical.
On the morphosyntax of wh-movement
2.2 Attraction to focus in English We can exploit Lipták’s account of the ungrammaticality of (12a) to account for the word order facts of English constructions featuring attraction to focus. Recall from the above (also cf. den Dikken and Giannakidou 2002) that there is evidence that English wh-phrases front to SpecFocP in root single wh-questions like (9a). Attraction to focus is evidently overt here: the wh-constituent ends up in the left periphery of the clause. For English non-wh foci, Kayne (1998) has argued recently that displacement to SpecFocP is overt as well. Yet, non-wh foci do not, as a rule, end up in a left-peripheral position in the sentence — in (14a), it looks like the focused noun phrase is in situ. On the basis of such contrasts as the one between (14a) and (14b), however, Kayne argues that the focused constituent has indeed moved in overt syntax, the fact that it does not surface at the left edge of the sentence being due, according to Kayne, to overt remnant movement of the TP across the landing-site of the focused constituent, to the specifier position of what he calls ‘WP’. Kayne’s derivation of (14a) is illustrated in (15). (14) a. I spoke only to JOHN. b. *I spoke to only JOHN. (15) a.
[FocP [PP to JOHN]i [Foc¢ only [TP I spoke ti]]] fi remnant TP fronting+raising of only fi b. [WP [TP I spoke ti]j [W¢ onlyk [FocP [PP to JOHN]i [Foc¢ tk [TP tj]]]]]
The contrast between (14a) and (14b) will now follow on the assumption that movement to SpecFocP cannot strand a preposition — (14b) could only be obtained by raising the focused noun phrase to SpecFocP on its own, leaving the preposition to behind, which would then be taken along to a position to the left of the constituent in SpecFocP as a result of remnant movement of the TP (I spoke to). Kayne shows that a variety of restrictions on the placement of focus particles like only can be accounted for on the assumption that focus movement happens in the overt syntax, its word order effects largely being undone by fronting of the remnant TP to a position to the left of the focus position. So let us take Kayne’s analysis of English focus to support the idea that focus movement is overt in this language. Then the hypothesis that wh-phrases undergo overt-syntactic focus movement to SpecFocP in English root questions becomes a subpart of a general approach to focus in terms of overt displacement.
87
88
Marcel den Dikken
2.3 No remnant TP fronting in ‘true’ wh-questions — Another intervention effect But of course there is a difference between (14a) and a wh-question like who did John kiss? — the focus in (14a), despite the fact that it is overtly fronted to SpecFocP, still ends up in clause-final position, due to the fact that both IP and only (Foc0) raise across it. Applying the same operations in a situation in which our focus is a wh-phrase will not deliver a regular wh-question but an echo question like John kissed WHO? instead. In a ‘true’ wh-question, remnant TP-fronting to SpecWP is apparently impossible. Taking the discussion in Section 2.1 as my cue, I would like to argue that this is so because attraction of the [+Wh] feature of the wh-constituent in SpecFocP by [+Wh] in C would fail in a situation in which the remnant TP fronts to SpecWP, across the landing-site of the focused wh-phrase: (16) *[CP C[+Wh] [WP TPj [W [FocP wh-phrase[+Wh,+foc] [Foc[+foc] [TP tj]]]]]]
The configuration in (16) is exactly like the one in (13): once again, something intervenes between the probe and the goal; and that something (i.e., TP) is once again quantificational (on the assumption that tense is quantificational), hence counts as a harmful intervener. Thus, (16) poses the same kind of intervention problem that we found in (13). We now have an account of the fact that, even though the remnant TP usually raises across a fronted focus in English, it cannot do so when the focus is a wh-phrase in a ‘true’ question. In echo questions (John kissed WHO?), by contrast, the TP can raise across the echo-wh in SpecFocP — no intervention effect interfering with the checking of C’s [+Wh] feature presents itself because in echo questions, C does not possess a [+Wh] feature. The [+Wh] feature of C is the prerogative of ‘true’ questions. This follows on the assumption that the possession of a [+Wh] feature is actually a property of the Q-operator on C: there is no Q-operator on C in echo questions (cf. Section 1.4, above). 2.4 No ban on remnant TP fronting in echo and multiple questions In multiple wh-questions, we also see no ban on remnant TP fronting to SpecWP: the in situ wh-phrase in (6a) (repeated here) is the focus of the construction, but it does not show up at the left periphery of the sentence; it behaves just like a non-wh ‘in situ’ focus such as the one seen in (14a). A partial derivation of (6a) thus reads as in (17), which is parallel to (15).7
On the morphosyntax of wh-movement
(6) a.
Who is in love with who?
(17) a.
[FocP [with who]i [Foc¢ Ø [IP who is in love ti]]] fi remnant TP fronting+Foc-raising fi b. [WP [TP who is in love ti]j [W¢ Øk [FocP [with who]i [Foc¢ tk [IP tj]]]]]
There would be an ‘intervention effect’ for the checking of the [+Wh] feature of C (which is introduced outside WP) here, as in (16), if the in situ wh-phrase (i.e., the one in SpecFocP) were the only thing that could check this [+Wh] feature. But of course, in a multiple wh-question there will always be at least one wh-phrase in addition to the one in SpecFocP which possesses a [+Wh] feature, and which is structurally closer to the wh-phrase raised to SpecFocP: in (6a)/(17), the pertinent wh-phrase is the subject-wh, sitting in SpecTP in (17). It is this wh-phrase which will eventually take care of the checking of C’s [+Wh] feature. The wh-phrase in SpecFocP (which, alongside its focus feature, does of course possess the morphological feature [+Wh] as well) does not get its [+Wh] feature checked against the [+Wh] feature of C at all — it couldn’t, for exactly the same reason that checking the [+Wh] features of C and the focused wh-phrase fails in (14). One thing that the discussion up to this point allows us to establish, then, is that it is not the [+Wh] features of individual wh-phrases that are in need of checking; on the contrary, it is the features of functional heads that drive checking. This is precisely what the ‘suicidal Greed’ approach of Chomsky (2000) leads us to expect. What we have here, then, is an indication that ‘Greed’ is false, and ‘suicidal Greed’ (or ‘attract’) is correct. 2.5 Overt attraction to Comp The derivation in (17), for the sentence in (6a), is only partial; beyond this point, the ‘superior’ wh-phrase will still need to get the [+Wh] feature of Comp checked. No harmful intervener finds itself between the two, however, so the checking of the [+Wh] feature of Comp will proceed unobstructed by an intervention effect. But note that the possessor of the [+Wh] feature which will be checked against that of C finds itself in the head of the specifier of the specifier of WP. Feature movement from that position is arguably illegitimate — while there are reported cases in the literature of subextraction from a specifier via phrasal movement (cf. e.g. Torrego’s 1985 Spanish examples, quoted in Chomsky 1986:45; or Who did you put [[a picture of t] in the drawer]?), established specimens of head movement like noun incorporation never raise the head of the specifier of another specifier into a higher head. Though actually presenting
89
90
Marcel den Dikken
a theory that explains this contrast between phrasal and head movement is by no means a trivial exercise, the empirical facts lead us to conclude that in contexts of the type in (17b), only the wh-phrase in SpecTP as a whole will be attractable by [+Wh] in C; attraction of just the feature fails as a result of the fact that the feature is too deeply embedded, not attractable to the probe. This in turn implies that, regardless of the ‘strength’ properties of the [+Wh] feature up in C, the ‘superior’ wh-phrase in a pair-list multiple wh-question must undergo phrasal movement to a position outside WP. Two options now remain, in principle: (i) the wh-phrase can raise directly to SpecCP and check C’s [+Wh] feature under Spec–Head agreement, or (ii) the wh-phrase can raise to the specifier position of a Quantifier Phrase (‘QP’) between C and WP, with Wh’s [+Wh] subsequently being attracted up to C under covert feature movement. The discussion in É. Kiss (1993) favours (ii), with the focused wh-phrase raising to SpecFocP and the other wh(’s) raising to (iterative) SpecQP (cf. Hungarian (18)). É. Kiss argues that, since they scope over another operator (the Wh in SpecFocP) and bind a variable in the latter’s scope, the Wh’s in SpecQP must have specific reference and be D-linked. It is this which explains why, in Hungarian multiple wh-questions, aggressively non-D-linked and robustly nonspecific wh-the-hell is necessarily the last one in the sequence of wh-constituents (cf. den Dikken and Giannakidou 2002: 53): it cannot move to SpecQP. (18) Kiváncsi vagyok [CP hogy [QP ki [FocP mit [IP … vett …]]]] I.wonder that who what-acc bought ‘I wonder who bought what.’
For an English root multiple wh-question such as (6a), it looks like the ‘superior’ Wh likewise fronts to SpecQP rather than all the way to SpecCP. We can take the placement of the ‘superior’ Wh vis-à-vis topics as our cue here: as Pesetsky (1989) points out, in a root single wh-question, the wh-phrase follows the topic (cf. (9a)); likewise, in a root multiple wh-question the topic precedes the ‘superior’ wh-phrase: (19) a. ?[TopP To Mary, [QP who should give what]]? b. *[CP Who [TopP to Mary, should give what]]?
The result in (19a) is strained, but it is still markedly better than (19b), which would result if the ‘superior’ Wh fronted overtly to SpecCP in a root multiple wh-question in English. We conclude, therefore, that option (ii) is correct, for Hungarian as well as English. For (6b), which features a wh-the-hell ‘superior’ wh-phrase, option (ii) is
On the morphosyntax of wh-movement
ruled out on account of the fact that wh-the-hell, due to its quantificational properties, does not qualify as a candidate for raising to SpecQP. So only option (i) would remain; but of course, a derivation along these lines will crash as well, since raising wh-the-hell to the SpecCP position of the root clause will take it outside the c-command domain of its licenser, the Q-operator in C. Tertium non datur — in particular, it is impossible to leave the wh-phrase in situ in SpecTP and to have its [+Wh] attracted by [+Wh] C at LF. If such were legitimate, we would fail to explain the unavailability of a pair-list interpretation for (6b); but as we have just seen, there are in fact good empirical reasons to believe that such is not legitimate. The account thus ensures that in English root multiple wh-questions there will always be overt fronting of the ‘superior’ wh-phrase to a position outside WP (the landing-site of the fronted remnant TP). We have derived this result from a theory centred around the checking of the [+Wh] feature of Comp — a theory which answers (Q4), accommodates the unavailability of a pair-list answer for (6b), explains why remnant IP fronting to Kayne’s SpecWP is possible in echo and multiple questions but not in single ‘true’ wh-questions featuring wh-fronting to SpecFocP, and derives the ungrammaticality of Hungarian (12a).
3. The root/non-root asymmetry revisited We have found that there is good evidence that there is a [+Wh] feature on C that checks a matching feature against a wh-question word. We have also found that wh-question words systematically check that feature covertly in English root single questions. So far, then, all the evidence suggests that, just as in Hungarian, the [+Wh] feature of English C is weak. But in English embedded clauses, wh-fronting proceeds all the way up to SpecCP — the word order contrast in (9a,b) is a particularly clear indicator to that effect. So now we made our way back to question (Q5): Why is there this asymmetry between root and embedded questions when it comes to the point in the derivation (pre- or post-Spell-Out) at which the [+Wh] feature of interrogative C is checked? More specifically, why does the [+Wh] feature of C drive overtsyntactic wh-movement to SpecCP in English only in non-root contexts, not when it is a feature of the root node? The answer to this question is actually quite simple if we base ourselves on Chomsky’s (1995: 234) characterization of the nature of strong features. For
91
92
Marcel den Dikken
Chomsky, the hallmark of a strong feature is the following (where ‘D’ is ‘derivation’, and ‘α’ is a category with a strong feature):8 (20) “D is canceled if α is in a category not headed by α.”
A violation of (20) can never arise in cases in which the strong feature is on the root node itself: there will never be any category not headed by α in such cases. The idea is that the strong features of a head H must be checked prior to the point in the derivation at which H gets included in a larger structure of which H is no longer the head; but of course, if H is the head of the root node, its projection will never be dominated by any larger superstructure, and nothing will ever force even the strong features of H to be checked in the overt syntax. Thus, the features of the root node are not active at all in the overt syntax — equivalently, therefore, the account can be phrased, if one should so prefer, in terms of covert Merge of the root CP (cf. Boškovic´ 2002: 362 for an analysis along these lines). Since overt displacement is never permissible unless forced (‘least effort’), Chomsky’s outlook on strong features (or its recasting in terms of LF-Merge at the root) guarantees straightforwardly that no feature of the root C will ever be checked via overt displacement. For English, then, what we can say is that the [+Wh] feature of C is strong (throughout), which results in overt-syntactic wh-movement in all contexts except the root CP: (21) a. The [+Wh] feature of English interrogative C is strong. b. Overt category movement to SpecCP is forced by (20) in all English interrogative CPs, except in root CPs, where (20) is inoperative.
For Hungarian (and Chinese etc.), the situation is different: the [+Wh] feature on C there is weak, hence we get no wh-displacement in overt syntax in any wh-question. All of this is precisely the desired result. Empirically, it gives us all the facts (including the otherwise elusive distribution of wh-the-hell in English, as discussed in detail in den Dikken and Giannakidou 2002), and theoretically, it confirms the approach to strong features built on (20).
4. Why wh-fronting is always overt in relative clauses While English is different from Hungarian, and Hungarian in turn is different from Chinese (and Japanese, Korean), when it comes to the displacement of
On the morphosyntax of wh-movement
wh-constituents in wh-questions, it looks like all languages that have wh-operators in relative clauses must displace them to SpecCP in the overt syntax (cf. Huang 1982 for Chinese, Watanabe 1991, Ishii 1991 for Japanese, Lipták 2001 for Hungarian) — regardless of the strength or weakness of the [+Wh] feature in C. This leads us back to our very first question, (Q1). Chomsky and Lasnik (1993; see Chomsky 1995: 71) suggest that the root of this contrast between questions and relative clauses may be the fact that relative clauses are predicates of the noun phrases they are construed with. That, combined with Williams’ (1980) conclusion that predication has to be established at S-structure, will then give us the desired result that, even in languages whose [+Wh] feature in C is weak, we still get wh-fronting in relative clauses. But of course we are still begging the question of why predication should need to be established ‘at S-structure’. Elaborating on Chomsky and Lasnik’s suggestion, I follow the line of thought emanating from Browning (1987) and Mulder and den Dikken (1992), inter alia, and say that the minimal maximal projection including the landingsite and trace of an unbound operator qualifies as a predicate. In order for the relative clause (CP) to be able to function as a predicate, therefore, it must feature operator movement to SpecCP in the overt syntax — LF feature movement could not accomplish this, since feature movement, by its very nature, is never operator movement. As a result, even in languages in which the [+Wh] feature of Comp is weak (Chinese, Japanese, Hungarian), the movement operation that checks this feature still targets SpecCP in overt syntax in relative clauses, driven by the need to establish a predicate. The need to establish a predicate is responsible for the overtness of wh-movement in relative clauses. But the movement per se is still to be blamed on the morphological [+Wh] feature of C — after all, all movement must be feature-driven; the locus of movement may be determined by a variety of factors (strength, legibility conditions), but the trigger for movement is systematically a morphological feature. So relative clauses are well-behaved: the only surprise is the overtness of wh-movement, which turns out to be for reasons having nothing to do with feature strength.
5. Embedded wh-questions and Focus movement While this takes care of the core of wh-movement, there is a little bit more to be said about English embedded wh-questions. I will address these points briefly here,
93
94
Marcel den Dikken
for the sake of completeness. While the discussion in this section is rather more tentative than the foregoing, nothing of what is said here in any way threatens the results of the preceding discussion. We have argued that in English root wh-questions, wh-movement targets SpecFocP and goes no further. In embedded questions, on the other hand, the wh-phrase raises to SpecCP — as forced by the strength of the [+Wh] feature in C, in tandem with (20). Now, does that wh-phrase travel through SpecFocP, or does it go straight to SpecCP? Here we address some facts which bear on that question. Let us start out by assuming that Negative Inversion, as illustrated in (22), is movement to SpecFocP. Just like wh-fronting to SpecFocP in root clauses, it triggers subject–auxiliary inversion. But Negative Inversion is much more radical in that respect than wh-fronting: it systematically gives rise to subject– aux inversion, no matter whether it takes place in root or embedded contexts. Wh-fronting, on the other hand, leads to inversion of subject and finite auxiliary only in root questions: (23b) contrasts sharply with (22b). (22) a. Under no circumstances {would he/*he would} do that. b. John said that under no circumstances {would he/*he would} do that. (23) a. What {would he/*he would} do? b. I wonder what {*would he/he would} do.
This suggests that wh-fronting in embedded clauses does not proceed through SpecFocP. On the other hand, the incompatibility of wh-fronting and Negative Inversion in the same clause, in root and embedded contexts alike (and regardless of the relative order of the wh- and negative constituents), suggests that wh-fronting in embedded clauses does proceed through SpecFocP:9 (24) a. *{What, under no circumstances} should he do? b. *John wondered {what, under no circumstances} should he do.
These facts generate the following questions: (25) a.
How do we prevent subject–aux inversion in embedded wh-questions, if the wh-phrase travels through SpecFocP? b. If we deny that the wh-phrase travels through SpecFocP, how do we capture (24b)?
The (tentative) answer that I propose here runs as follows. The key hypothesis is that English interrogative C has a strong head-attracting feature that (in embedded clauses) attracts the Foc-head overtly, prior to Foc doing anything itself. C’s attracting Foc takes Foc’s strong head-attracting feature up to C.
On the morphosyntax of wh-movement
Foc’s own (strong) head-attracting feature can be checked there against C’s head-attracting feature, as a result of which Foc’s head-attracting feature is effectively satisfied. Consequently, once Foc has raised to C, it will no longer attract an auxiliary up to it. This accounts for the non-occurrence of subject– aux inversion (cf. (23b)). C’s attracting Foc also takes Foc’s strong [+foc] feature (demanding an overt [+foc] bearing specifier) up to C. Upon Foc-to-C movement, then, the complex C+Foc-head will demand an overt specifier (specifically, a [+Wh, +Foc] phrase), which raises to SpecCP directly, without a stopover in SpecFocP. In fact, due to Foc’s raising to C, SpecFocP is not projected at all; as a consequence, Negative Inversion will be impossible in embedded wh-questions, and (24b) is accounted for. The reader will notice a direct parallel here between the discussion here and Chomsky’s (1995: Ch. 3) and Bobaljik and Jonas’ (1996) account of nominativechecking by the subject in English: T raises overtly to AgrS, and as a result, all of T’s features are checked in the checking domain of AgrS; no SpecTP is ever projected, and constructions which depend on the presence of SpecTP (transitive expletive constructions, in particular) cannot arise. For me, the nonprojection of SpecFocP in embedded clauses takes care of the incompatibility of wh-fronting and Negative Inversion in these contexts; of course, the fact that Negative Inversion and wh-fronting cannot be combined in root questions either is straightforward as well: the two target the same position, SpecFocP, which is non-recursive. If this (tentative) answer to (25) stands up to scrutiny, we have a comprehensive account of the key facts of English wh-movement. As far as I can see, there are two main questions left to be addressed by the account sketched in the previous paragraph — a theoretical one and an empirical one. The latter concerns Kayne’s analysis of constructions like (14), in terms of overt movement to SpecFocP: wh-fronting and only/even-focus are not mutually exclusive in English, unlike in Hungarian. The former concerns the key hypothesis that C can attract Foc up to it prior to Foc becoming active itself. This is clearly countercyclic, in precisely the same way that Chomsky’s (1995: Ch. 3) and Bobaljik and Jonas’ (1996) account of nominative-checking in English was countercyclic. I will not speculate here on the proper perspective on the cycle, leaving this for future research.
95
96
Marcel den Dikken
Notes 1. The ideas reported in what follows arose out of my joint work with Anastasia Giannakidou on the syntax and semantics of ‘aggressively non-D-linked’ wh-phrases (such as what the hell); cf. den Dikken and Giannakidou (2002). While she disagrees with many of the ingredients of the specific proposals put forward in this paper, her feedback on this material is gratefully acknowledged. My indebtedness to Anikó Lipták’s (2001) work on wh-constructions is also very evident to the eye. 2. The structure in (7a) is an interim representation; it will be developed further in Sections 2.4 and 2.5, below. For our present purposes it will suffice. 3. Also note the well-known fact that echo questions do not outwardly behave like questions at all; thus, they cannot satisfy a matrix predicate’s selectional requirement for an interrogative complement (*I wonder John kissed WHO?). 4. There might well be multiple topic positions in the left periphery (cf. Rizzi 1997 and subsequent work); but my assumption is that whatever position a language exploits for its topics, it will be the same position in root and embedded clauses alike. 5. Neither sentence in (9) is impeccable; Emonds (1976) originally starred any combination of topicalisation and wh-fronting in root clauses (cf. also Baltin 1985:155). But for speakers who accept tautoclausal topicalisation and wh-fronting, the sentences in (9) are clearly superior to their counterparts with the reverse relative order of the topic and the wh-constituent. 6. Regular question words do not have to be attracted to Focus: though they can be specified as [+focus], they do not have to be so specified. See the discussion below of multiple wh-questions below. 7. Two questions arise at this point which need to be briefly addressed. (Thanks to Youngmi Jeong for bringing them up.) One concerns the movement of what to SpecFocP, crossing who — why is this not a case of harmful intervention? The assumption here is that movement to an operator position (SpecFocP) is obstructed only by intervening operators; who, which is still in situ at the time what raises to SpecFocP, is not an operator (operators are in AN-positions), hence does not count as a harmful intervener. Nor is who a closer potential attractee: it lacks the feature [+Focus], so the closest match for the [+Focus] feature in Foc is indeed what, as desired. The second question that arises is why movement of the remnant TP to SpecWP, across the focused wh-constituent in SpecFocP, is not blocked by an intervention effect. This time we cannot say that what in SpecFocP is not an operator: it plainly is. But apparently, movement to SpecWP is altogether oblivious to intervening quantificational elements. One way of understanding this is to make the assumption that movement to SpecWP is of a different nature than movement to, for instance, SpecFocP or SpecCP: it is not operator movement but something else. (Precisely what it is will remain unclear absent a better understanding of the nature of ‘W’; but that movement to the Kaynean SpecWP is not operator movement is clear enough: all sorts of things may raise to Kayne’s SpecWP, regardless of whether they are operators or not.) The fact that the presence of a tensed TP in SpecWP leads to harmful intervention in (16), above, does not contradict what I just said: the raised TP is not itself an operator but it does contain a quantificational component (a tense operator), and it is this quantificational component inside TP, not TP itself, that is the harmful intervener in (16).
On the morphosyntax of wh-movement
8. For those readers who prefer the use of ‘EPP features’ to that of ‘strong features’, the ensuing discussion is straightforwardly ‘recodable’ by interpreting α as a category with an ‘EPP feature’. 9. Culicover (1993) has pointed out that wh-extraction from a clause featuring Negative Inversion is grammatical (and actually lifts the that-trace effect): (i) a. b.
Leslie is the person who I said that under no circumstances would run for president. Who did John say that under no circumstances would run for president?
Since (i-a) (Culicover’s own example) is a case of relativization, the wh-phrase here is not a focus (cf. (10d), above), so this is not a particularly interesting case. More interesting is (i-b) (our variation on Culicover’s theme): equally grammatical, and much better than (24b). The key to understanding this contrast is the fact that there is no successive-cyclic focus movement — a constituent X is always the focus of at most one single clause, never of several clauses at the same time (Hungarian is a perfect exemplification of this; see Lipták 2001 for discussion). So the FocP where who checks its [+Foc]-feature is in the matrix clause, and we get no competition for the SpecFocP position in the embedded clause in (i-b) (unlike in (24b)).
References Baltin, M. 1985. Toward a Theory of Movement Rules. New York: Garland. Beck, S. 1996. “Quantified structures as barriers for LF-movement.” Natural Language Semantics 4: 1–56. Beghelli, F. and Stowell, T. 1997. “Distributivity and negation: The syntax of each and every.” In Ways of Scope Taking, A. Szabolcsi (ed), 71–107. Dordrecht: Kluwer. Benincà, P. 2001. “The position of topic and focus in the left periphery.” In Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi, G. Cinque and G. Salvi (eds), 39–66. Amsterdam: Elsevier. Bennis, H. 1995. “The meaning of structure: The wat voor construction revisited.” In Linguistics in the Netherlands 1995, M. den Dikken and K. Hengeveld (eds), 25–36. Amsterdam: John Benjamins. Bobaljik, J. and Jonas, D. 1996. “Subject positions and the roles of TP.” Linguistic Inquiry 27: 195–236. Boškovic´, Ž. 2002. “On multiple wh-fronting.” Linguistic Inquiry 33: 351–383. Brody, M. 1995. “Focus and checking theory.” In Approaches to Hungarian V, I. Kenesei (ed). Szeged: JATE Press. Browning, M. 1987. Null operator constructions. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. [Appeared 1991. New York: Garland.] Chomsky, N. 1986. Barriers. Cambridge, Mass.: MIT Press. Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. 2000. “Minimalist inquiries.” In Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 89–155. Cambridge, Mass.: MIT Press.
97
98
Marcel den Dikken
Chomsky, N. and Lasnik, H. 1993. “The theory of principles and parameters.” In Syntax: An International Handbook of Contemporary Research, Vol. 1, J. Jacobs, A. von Stechow, W. Sternefeld and T. Vennemann (eds), 506–569. Berlin: Walter de Gruyter. Citko, B. 1998. “On multiple wh-movement in Slavic.” In Formal Approaches to Slavic Linguistics: The Connecticut Meeting, 1997, Ž. Boškovic´, S. Franks and W. Snyder (eds), 97–113. Ann Arbor, Mich.: Michigan Slavic Publications. Culicover, P. 1993. “Evidence against ECP accounts of the that-t effect.” Linguistic Inquiry 24: 557–561. den Dikken, M. and Giannakidou, A. 2002. “From hell to polarity: ‘Aggressively nonD-linked’ wh-phrases as polarity items.” Linguistic Inquiry 33: 31–61. Emonds, J. 1976. A Transformational Approach to English Syntax. New York: Academic Press. Huang, C.-T. J. 1982. Logical relations in Chinese and the theory of grammar. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Ishii, Y. 1991. Operators and empty categories in Japanese. Doctoral dissertation, University of Connecticut, Storrs. Kayne, R. 1998. “Overt vs. covert movement.” Syntax 1: 128–191. É. Kiss, K. 1987. Configurationality in Hungarian. Dordrecht: Reidel. É. Kiss, K. 1993. Wh-movement and specificity. Natural Language and Linguistic Theory 11: 83–120. Lipták, A. 2001. On the syntax of wh-items in Hungarian. Doctoral dissertation, of Leiden University. Mulder, R. and den Dikken, M. 1992. “Tough parasitic gaps.” In Proceedings of NELS 22, K. Broderick, (ed), 303–317. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Pesetsky, D. 1989. Language-particular processes and the Earliness Principle. Paper presented at GLOW 12, Utrecht University. [April 1989] Pesetsky, D. 2000. Phrasal Movement and Its Kin. Cambridge, Mass.: MIT Press. Postma, G. 1994. “The indefinite reading of WH.” In Linguistics in the Netherlands 1994, R. Bok-Bennema and C. Cremers (eds), 187–198. Amsterdam: John Benjamins. Rizzi, L. 1997. “The fine structure of the left periphery.” In Elements of Grammar: Handbook of Generative Syntax, L. Haegeman (ed), 281–337. Dordrecht: Kluwer. Rudin, C. 1988. “On multiple questions and multiple wh-fronting.” Natural Language and Linguistic Theory 6: 445–501. Scott, T. 2002. CP-Topic parameter and Russian WH. Ms., State University of New York, Stony Brook. Torrego, E. 1985. On empty categories in nominals. Ms., University of Massachusetts, Boston. Watanabe, A. 1991. Wh-in situ, subjacency, and chain formation. Ms., Massachusetts Institute of Technology, Cambridge. Williams, E. 1980. “Predication.” Linguistic Inquiry 11: 203–238.
German is a multiple wh-fronting language!* Kleanthes K. Grohmann University of Cologne
1.
Introduction
In this paper, I am going to explore properties of multiple wh-question formation in a variety of languages, concentrating on syntactic and interpretive facts from or relevant to German. I will propose two theses which might at first sound both unexpected and contradictory. First I will argue that German is like Bulgarian, in that in multiple wh-questions all wh-phrases (henceforth, WHs) move overtly into a finer articulated CP (the Comp-layer of the clause). Then I suggest that German is like Italian, in that it does not form true multiple constituent questions to begin with; displacement of WHs must thus have a different trigger. Putting both theses together, however, it becomes clear that the obligatory multiple wh-fronting argued for here is of a different nature than standardly assumed for languages with multiple wh-movement. Relevant for the first claim are cross-linguistic patterns of single-pair and pair-list readings of all WHs in multiple wh-questions.1 As the discussion in Section 2 will show, the typological characterization of the types of wh-displacement (which is not necessarily wh-movement) in multiple questions I suggest and the terminology I employ are the following: (1) Typology of wh-displacement (i) Multiple wh-movement languages A¢-move all WHs in the overt component. (ii) Singular wh-movement languages A¢-move one WH in the overt component and might A-move others. (iii) Zero wh-movement languages don’t A¢-move any WHs in the overt component, but may A-move them.
As will become clear presently, I take wh-displacement to be different from wh-movement. The latter term will be reserved for movement of a WH to check
100 Kleanthes K. Grohmann
some wh-related property — for example, an arbitrary feature [Wh] or a focusfeature [Foc] (see the discussion of the “Hagstrom-Boškovic´ approach” in Section 3); in other words, operator A¢-movement into Comp. wh-displacement simply denotes movement of a WH for other reasons, such as wh-scrambling as observed in Japanese or the fronting of WHs in languages like Serbo-Croatian (see Sections 2 and 3 for clarification); this movement is arguably not of the same type as the former, but rather A-movement within Infl. In Section 3, I will present an adaptation of two recent approaches to (the syntax and semantics of) multiple wh-questions and illustrate the main ideas from that and the previous section. As I implement Boškovic´’s (1998) version of Hagstrom’s (1998) original theory into the current framework, I term this the Hagstrom-Boškovic´ approach. With the interpretive facts as initial motivation to hypothesize multiple wh-fronting in German and a theory of wh-question licensing at our disposal, I will investigate this claim further in Section 4. Here I will present evidence for multiple A¢-displacement of WHs in German, which will come from apparent intervention effects (cf. Beck 1996, Pesetsky 2000) and an extension thereof, the behaviour of adverbs, binding effects, and finally, discourse requirements (from Grohmann 1998 and subsequent work). To complete the initial promise, I provide a brief discussion of and comparison with relevant facts from Italian in Section 5 to demonstrate that German only looks like Bulgarian on the surface. The truth is somewhat more complicated, with the result that German is a language that fronts all WHs into Comp without wh- or operator A¢-moving them. Or, to put it differently, German does not make available true multiple wh-questions, just like Standard Italian (Rizzi 1978, Calabrese 1984). Section 6 summarizes this paper.
2. Interpretation and typology of multiple wh-questions This section presents the lay of the land. I will introduce some general points about multiple wh-questions across languages and discuss interpretive issues. Here I will be concerned with the availability of single-pair and pair-list readings in multiple wh-questions by basically dissecting the findings and explanations of Boškovic´ (1998), who in turn extends Hagstrom’s (1998) proposal. These two sources form the base of what I dub the Hagstrom-Boškovic´ approach (see Section 3). To be as close to this line of reasoning as possible, I will have to consider the data available as briefly and limited as Boškovic´ does
German is a multiple wh-fronting language!
(see also Boškovic´ 2002). This and the following section can thus be taken as a slightly extended exegesis of his interpretation of the facts and the relevance of Hagstrom’s work. 2.1 Single-pair and pair-list readings As noted first, to my knowledge, by Wachowicz (1974), multiple wh-questions are often ambiguous between a single-pair (SP) and a pair-list (PL) reading. Thus the question in (2) may be answered in two ways, identifying a singular referent for each WH (pairing each WH with a single answer, SP) or giving an exhaustive set of referents (matching a list of wh-answer pairs, PL), where the felicity of either answer depends on the context: (2) Q: Who bought what? A: Mary bought a sweater. (SP) A¢: Mary bought a sweater, Jane bought shoes, Anne bought a skirt… (PL)
However, while contexts may be used to “disambiguate” the appropriate answer, the availability of SP- or PL-interpretations differs across languages. What follows is a cursory look at the data (as alluded to above, mostly taken from Boškovic´ 1998).2 In a scenario like the one in (3), for example, where the context ensures that each WH is paired with a single referent, Japanese allows, but English disallows, the SP-reading; in other words, the multiple wh-question is felicitous only in Japanese:3 (3) Scenario I: John is in a department store and off in the distance sees somebody buying an article of clothing, but he does not see who it is, and neither does he see exactly what is being bought. He goes to a shop assistant and asks: a. #Who bought what? (English: *SP) b. Dare-ga nani-o katta no? (Japanese: SP) who-nom what-acc bought q ‘Who bought what?’
Following Boškovic´’s discussion (see among others also Hornstein 1995, Comorovski 1996), we could entertain the following generalization: (4) Descriptive generalization Japanese allows, but English disallows, SP-readings in multiple wh-questions.
101
102 Kleanthes K. Grohmann
One difference between Japanese and English generally agreed upon is that only the latter exhibits overt wh-movement, viz. operator A¢-movement into Comp of one WH. I will thus refine the following working hypothesis (from Boškovic´ 1998): (5) Working hypothesis (to be revised) Overt wh-movement forces PL-interpretation.
2.2 A cross-linguistic perspective If we consider the same type of scenario, but look at French, a language which allows two strategies to form wh-question, overt wh-movement (“French I”) and wh-in situ (“French II”), we find the following pattern: (6) Scenario I¢: Mary sees John down the corridor. Not wearing her glasses, she notices that John gives something to someone but can’t figure out what the item nor who the person is. So she asks her friend to find out: (French I: *SP) a. #Qu’ a-t-il donné à qui? what has-he given to whom ‘What did he give to whom?’ b. Il a donné quoi à qui? (French II: SP) he has given what to whom ‘What did he give to whom?’
French I seems to pattern in the relevant respect with English, French II with Japanese. Looking at languages in which all WHs are obligatorily fronted (such as Bulgarian, Serbo-Croatian, Romanian, Polish, Russian etc.; cf. Rudin 1988 for the original and, among many others, Boškovic´ 2002 for more recent discussion), we find the following, different patterns: (7) Scenario I (cf. (3)) a. #Koj kakvo e kupil? b. *Kakvo koj e kupil? what who is bought ‘Who bought what?’ (8) Scenario I (cf. (3)) a. Ko je šta kupio? b. Šta je ko kupio? what is who bought ‘Who bought what?’
(Bulgarian: *SP)
(Serbo-Croatian: SP) (Serbo-Croatian: SP)
German is a multiple wh-fronting language! 103
Bulgarian (and also Romanian, which I will briefly address below), strictly observing the Superiority Condition (Chomsky 1973), disallows (7b) on purely grammatical grounds, unlike Serbo-Croatian. But the question that arises is why Serbo-Croatian4 should allow an SP-reading in both cases. 2.3 Three types of Superiority Boškovic´ (1998) formalize three types of Superiority effects to further describe cross-linguistic variation in multiple wh-question formation and interpretation.5 In some languages, a bare, non-D(iscourse)-linked WH may not be fronted over a structurally higher one, the standard Superiority effect of note 5 (see Section 5 for more on D-linking); these languages are sensitive to “Syntactic Superiority”. In others, this restriction does not hold: syntactically, any WH may front over another (“Apparent Superiority”). In languages that are sensitive to a third type of Superiority, called “Interpretive Superiority” by Boškovic´, the PL-reading gets lost in those cases where a lower WH (WH2) is fronted over an originally higher one (WH1). This three-way split of Superiority effects is listed in (9), and I will briefly discuss each one in turn.6 (9) Types of Superiority (i) Syntactic Superiority: *non-D-linked WH2 > non-D-linked WH1 (such as English, Bulgarian, Romanian, Chinese, French) (ii) Apparent Superiority: (any) WH2 > (any) WH1 (such as Japanese, German, Serbo-Croatian) (iii) Interpretive Superiority: loss of PL-reading when WH2 > WH1 (such as English, Japanese, Serbo-Croatian)
Checking with the languages just considered, the effects of this tripartite formulation of Superiority seem to be borne out. The next scenario forces a pair-list answer; in a given language, the question may or may not be felicitous, just along the lines of (9i–iii): (10) Scenario II: Mary sees John finishing off his daily dealings. She sees a bunch of people that she knows walk away and John stacking his leftover merchandise, so she asks him: a. Dare-ga nani-o katta no? (Japanese: PL) b. #Nani-o dare-ga katta no? (Japanese: *PL) what-acc who-nom bought q ‘Who bought what?’
104 Kleanthes K. Grohmann
In Japanese, a strict wh-in situ language, the PL-reading gets destroyed if the lower WH2 moves over WH1 (presumably, some sort of “wh-scrambling”). Next, consider languages in which all wh-phrases are obligatorily fronted: (11) Scenario II (cf. (10)) a. Koj kakvo e kupil? b. *Kakvo koj e kupil? what who is bought ‘Who bought what?’ (12) Scenario II (cf. (10)) a. Ko je šta kupio? d. #Šta je ko kupio? what is who bought ‘Who bought what?’
(Bulgarian: PL)
(Serbo-Croatian: PL) (Serbo-Croatian: *PL)
While in a PL-enforcing context WH2 may precede WH1 in Serbo-Croatian syntactically, the desired reading is not available, i.e. the PL-interpretation is lost. This is an instance of Interpretive Superiority. In Bulgarian, this is not even an option. This separates the two languages already, supporting the conclusion that in multiple wh-fronting language the WHs need not all move the same way (cf. Rudin 1988, Boškovic´ 2002 and references cited). Leaving the “why” for later, this immediately suggests that with simple WHs, Bulgarian and English fit the bill projected by (9a), while Serbo-Croatian seems to fit (9c). Interestingly, some questions in English do so too, namely those that contain two D-linked WHs: (13) Scenario II (cf. (10)) a. Who bought what? b. Which guy bought which drug? c. #Which drug did which guy buy?
(English: PL) (English: PL) (English: *PL)
Thus English mulitiple D-linked wh-questions are subject to Interpretive Superiority, just as all multiple wh-questions in Serbo-Croatian are. It is well-known that D-linking allows for more freedom regarding (Syntactic) Superiority violations (Pesetsky 1987). As Citko and Grohmann (2001) note, Bulgarian D-linked multiple wh-questions do allow movement of either WH. Furthermore, if a D-linked WH2 moves over WH1, the preferred reading is a PL-reading (Roumyana Izvorski, p.c.); the same applies to Romanian (Ileana Comorovski, p.c.): it is thus not Interpretive Superiority that plays a role in these languages, but Apparent Superiority. The relevant examples are given in (14):
German is a multiple wh-fronting language! 105
(14) [along some variation of Scenario II] (cf. (10)) a. Koga kniga koj covek e kupil? which book which person is bought ‘Which book did which person buy?’ b. La care cînd te ai gîndit? of which when you have thought ‘Which one have you thought of when?’
(Bulgarian: PL)
(Romanian: PL)
This suggests that whatever the exact landing site of WHs in Bulgarian (D-linked or not), it is different from the one(s) in Serbo-Croatian in licensing a PL-interpretation and yet, conforming to the original working hypothesis in (5), destroying an SP-reading. As I will explore in the next section, the landing site is indeed the crucial difference. I will tie it in with a larger re-interpretation of A- vs. A¢-movement. These patterns, coupled with that re-interpretation (Section 3), will allow us to restate our initial working hypothesis slightly: (15) Working hypothesis (extension) (i) A¢-movement of any WH forces PL-interpretation regardless of Superiority. (ii) A-movement of all WHs showing Apparent Superiority destroys PL reading (leading to Interpretive Superiority). (iii) A-movement of all WHs without Apparent Superiority allows SP-reading.
This extension, in turn, applied to the data presented above, implies a cataloguing such as that in Table 1 (basically adopted from Boškovic´ 1998).7 Note that one consequence of this tripartition is that languages like SerboTable 1.Quasi-tripartition If a language has:
It allows readings:
Zero wh-movement languages (Japanese, Serbo-Croatian, French II…)
Singular wh-movement languages (English, French I, German, …)
WH1 > WH2: PL WH2 > WH1: SP (such as D-linking)
Multiple wh-movement languages (Bulgarian, Romanian…)
WH1 > WH2: PL WH2 > WH1: PL (even D-linking)
But:
German: WH2 > WH1: PL (even D-linking)
106 Kleanthes K. Grohmann
Croatian — such as Polish or Russian, i.e. those multiple wh-fronting languages that arguably target a position lower than Comp8 — are now characterized as “Zero wh-movement” languages. This does not mean that they leave all WHs in situ proper; it just says that no WH undergoes movement to check a wh-feature (or whatever the trigger for wh-movement into Comp). So far I have not looked at German; its position in Table 1 is inferred from Boškovic´’s discussion (in line with Hagstrom’s suggestion that German employs “Q-stranding”, which I address in Section 3.4 below). But it turns out that at closer inspection, this is not correct. 2.4 Oddball German The third column in Table 1 makes the table look pretty ugly, so in the following I will clean it up. Beyond cosmetic-surgical aspirations, I intend to motivate this clean-up empirically by showing that German patterns with the Multiple wh-movement language Bulgarian in making only PL-readings available, not with the Singular wh-movement language English (the standard assumption; cf. Grewendorf 2002: 244, for example). First, looking at the context that forces a PL-reading, German patterns like Bulgarian or Romanian (Scenario II is repeated for convenience from (10) above): (16) Scenario II: Mary sees John finishing off his daily dealings. She sees a bunch of people that she knows walk away and John stacking his leftover merchandise, so she asks him: a. Wer hat was gekauft? (German: PL) b. Was hat wer gekauft? (German: PL) what has who bought ‘Who bought what?’
Second, in a context that forces an SP-reading, such as Scenario I (repeated from (3)), German does actually not pattern with Japanese or French II (compare with (3) and (6b), respectively) — the syntactically well-formed question does not allow for SP-interpretation: (17) Scenario I: John is in a store and off in the distance sees somebody buying an article of clothing, but he does not see who it is, and neither does he see exactly what is being bought. He goes to a shop assistant and asks: a. #Wer hat was gekauft? (German: *SP)
German is a multiple wh-fronting language! 107
b. #Was hat wer gekauft? what has who bought ‘Who bought what?’ c. #Welcher Kunde hat welches Produkt gekauft? d. #Welches Produkt hat welcher Kunde gekauft? which product has which customer bought ‘Which customer bought which product?’
(German: *SP)
(German: *SP) (German: *SP)
I take these data to suggest — admittedly, on the limited grounds considered here, i.e. little more than an exegesis of Boškovic´ (1998) — a cleaner table: Table 2.An improved tripartition If a language has:
It allows readings:
Zero wh-movement languages (Japanese, Chinese, Serbo-Croatian, French II…)
WH1 > WH2: PL/SP WH2 > WH1: SP
Singular wh-movement languages (English, French I…)
WH1 > WH2: PL WH2 > WH1: SP
Multiple wh-movement languages (Bulgarian, Romanian, German…)
WH1 > WH2: PL WH2 > WH1: PL
In the next section I will present the Hagstrom-Boškovic´ approach to questions and explore the initial working hypothesis (revised from (5) in (18)), before addressing the proposal that German might actually be a multiple wh-movement language, as Table 2 suggests. The full characterization of the syntactic formation of a multiple wh-question and the resulting interpretive possibilities is captured in (18). I will return to the issue of what it means to be a “wh-movement” language and what other options could look like later on. (18) Working hypothesis (revised) (i) Movement of any WH targeting the Comp-layer forces PL-interpretation. (ii) Movement of any WH inside the Infl-layer allows SP-interpretation. (iii) a. Movement of any WH within the Infl-layer forces SP if WH2 > WH1. b. Movement of any WH within the Infl-layer allows PL if WH1 > WH2.
108 Kleanthes K. Grohmann
3. The Hagstrom-Boškovic´ approach to questions After presenting one possible argument in favour of overt multiple wh-movement in German from typological considerations, I follow further Boškovic´’s (1998) extension of the syntax and semantics of questions by Hagstrom (1998), the Hagstrom-Boškovic´ approach, the theoretical underpinnings of which I will sketch in a concise manner in this section. For more details on the semantics, see the original sources. The syntactic implementation also draws from the framework outlined in Grohmann (2000). 3.1 A semantics for multiple wh-questions Let’s start with the basic semantics assumed. In the Hagstrom-Boškovic´ approach, questions are licensed by a question particle [Q] (which may or may not be overtly realized), rather than by movement of a WH. Displacement of WHs is driven by other properties of the grammar (such as focus; see e.g. Horvath 1986, Stjepanovic´ 1995, Sabel 2000, Boškovic´ 2002). [Q] always moves to C to type the clause interrogative (cf. Cheng 1991), and it may do so from one of two positions: a high position, c-commanding all WHs, or a low position (where it is adjoined to the lowest WH). The trace left behind by [Q]-movement to C serves as a choice function variable driving the semantics of questions (Hagstrom 1998). What Boškovic´ (1998) finds is that languages fall into two categories, whether they allow SP-readings in a multiple questions or not: [Q] moves from some clause-internal position to C0 Æ low [Q]: pair-list b. [Q] is generated above both WHs and moves to C0 Æ high [Q]: single-pair
(19) a.
A single question is a set of propositions, which can be responded to either by choosing one proposition from the set or by denying the presupposition that there is an answer. (20) Single Question Recognition (Hagstrom 1998: 148) If the semantic value of an utterance is of type ·ptÒ (a set of propositions), then the utterance is a (single) question. To respond: (a) one proposition from the set is selected, or (b) the presupposition (that there is an answer) is denied.
German is a multiple wh-fronting language! 109
By extension, the rule “Multiple Question Recognition” identifies a (pairlist) multiple question as a set of questions, which can also be responded to in two ways: by selecting one proposition from the set A for each member set A or by denying the presupposition that there is an answer in A. [Q] is an existential quantifier over a choice function. (21) Multiple Question Recognition (Hagstrom 1998: 148) If the semantic value of an utterance is of type ·pt, tÒ (a set of questions), then the utterance is a (pair-list multiple) question. To respond: For each member set A, (a) one proposition from the set A is selected, or (b) the presupposition (that there is an answer in A) is denied. (22) Choice Function (Hagstrom 1998: 130) A function f is a choice function if it applies to a (non-empty) set and yields a member of that set.
The semantic value of a WH is a set of individuals. Merging with WH2 and moving on to C, the choice function variable left behind by [Q] takes WH2 as its argument and returns an individual, which is taken as an argument by the verb. The value of the higher WH1 is a set, as it is outside the scope of the choice function variable. Further composition of WH1 yields a set of sets of propositions, one set of propositions for each value x in the set of WH1. Hagstrom (1998: 142ff.) calls this composition Flexible Function Application, a repair strategy for semantic type mismatches. It applies when a function receives a set of arguments instead of a single argument, in which case it is applied to each argument in the set of arguments. The result of this function application yields a set. By the Multiple Question Recognition rule, an appropriate answer to a question such as (23a), where [Q] is merged low (attached on WH2), provides an answer to each of the constituent questions of the set {What did A buy?, What did B buy?, …} or, more formally: (23) a. Who bought what? b. {{A bought f1 (WHAT), A bought f2 (WHAT), …}, {B bought f1 (WHAT), B bought f2 (WHAT), …}, …}
Merging [Q] low thus yields a set of questions — but merging [Q] high yields a set of propositions. Both WHs are contained within the argument of the choice function variable. Composition of WH2 (qua Flexible Function Application) returns a set of properties which, once applied to the set WH1, yields one
110
Kleanthes K. Grohmann
proposition for each possible subject with each possible object, namely (24). So the Single Question Recognition rule gives an SP answer. (24) {A bought a, A bought b, …, B bought a, B bought b, …}, (where WH1 = {A, B, …} and WH2 = {a, b, …})
If the choice function variable left behind by [Q]-movement to C scopes over WH2 only (“low [Q]”), we get a PL-interpretation; if it scopes over both WHs (“high [Q]”), we get an SP-interpretation (in the following, strikethrough indicates a lower copy or trace and bold strikethrough the position of the choice function variable, i.e. the original position of [Q]): (25) Choice Function Variable and Question Interpretation a. [Q]-C ... WH1 ... WH2-[Q] ... b. [Q]-C ... [Q] ... WH1 ... WH2
(low [Q]: PL) (high [Q]: SP)
3.2 The syntax of multiple wh-questions Let us now turn to the syntax of the Hagstrom-Boškovic´ approach and identify the structural positions of high [Q], low [Q], and other important players. I assume an articulated Comp-layer, slightly modifying Rizzi’s (1997) structure, where F0 generates “high [Q]”: (26) CP > TopP* > FocP > TopP* > FP
Applying Hagstrom’s analysis, the generalization emerges that syntactic movement of a WH over high [Q] destroys the SP-reading and only allows a PL-interpretation (cf. (18)): (27) Availability of SP-/PL-interpretation (i) Wh-in situ languages always allow SP-interpretation (Japanese, Chinese, French II — and, under this criterion, Serbo-Croatian); (ii) Singular wh-movement languages allow SP-interpretation only with D- linked WHs (which arguably do not have to move so high — English, French I); (iii) Multiple wh-movement languages never allow SP-interpretation (assumption: all WHs move into Comp-layer — Bulgarian, Romanian).
German is a multiple wh-fronting language!
According to the tripartition from Table 2, German falls into category (27iii). Let’s see how. The two options — wh-scrambling within the Infl-layer (WH2 > WH1) or retaining base order (WH1 > WH2) — are available to wh-in situ languages, which explains why both SP- and PL-readings are felicitous in either context (Section 2). With high [Q], the choice function computed over the variable ranges over all WHs in situ (yielding SP), whereas with low [Q] it ranges only over WH2 (yielding PL). I suggest (29) to be the rough and relevant structures for a multiple wh-question in Chinese, such as (28): (28) Shei mai-le shenme? who buy-asp what ‘Who bought what?’ (29) a.
b. [CP [Q]-C ... [FP [Q]-F [TP WH1 T ... [vP WH1 ... WH2 ...]]]]
(SP)
The same strategies apply to other Zero wh-movement languages as well: (30) a.
Dare-ga nani-o katta no? who-nom what-acc bought q ‘Who bought what?’ b. Ko je šta kupio? who is what bought ‘Who bought what?’
(31) a.
(Japanese)
(Serbo-Croatian)
[CP [Q]-C [TP WH1 T [AgrOP WH2-[Q] AgrO [vP WH1 v WH2-[Q]]]]] (PL)
b. [CP [Q]-C [FP [Q]-F [TP WH1 T [AgrOP WH2 AgrO [vP ...]]]]
(SP)
3.3 Wh-scrambling Wh-scrambling (reordering of WHs) in Zero wh-movement languages stays within the Infl-layer; for simplicity’s sake, TP for wh-subjects and AgrOP for wh-objects.9 Thus the relevant derivations underlying (32a) and (32b) would be (33a) and (33b), respectively:
111
112
Kleanthes K. Grohmann
(32) a.
Nani-o dare-ga katta no? what-acc who-nom bought q ‘Who bought what?’ b. Šta je ko kupio? what is who bought ‘Who bought what?’
(33) a.
(Japanese)
(Serbo-Croatian)
[CP [Q]-C [AgrOP WH2-[Q] AgrO [TP WH1 T [vP WH1 v WH2-[Q]]]]] (SP)
b. [CP [Q]-C [FP [Q]-F [AgrOP WH2 AgrO [TP WH1 T [vP … ]]]]]
(SP)
Given that a bare WH2 may never move over a bare WH1 in Singular wh-movement languages, the only interpretation we can get here is PL. Regardless of whether WH2 stays really in situ at the point of Spell Out or whether it moves into the Infl-layer overtly, movement of low [Q] to C leaves behind the choice function variable which invariably only has WH2 in its scope and thus takes WH2 as its argument: (34) Who bought what? (35) a.
[CP [Q]-C [FocP WH1 Foc [TP WH1 T [vP WH1 v WH2-[Q]]]]]
(PL)
b. [CP [Q]-C [FocP WH1 Foc [FP [Q]-F [TP WH1 T [vP WH1 v WH2]]]]] (PL)
3.4 A note on “Q-Stranding” Under the assumption that English and German only differ in that the latter allows WH2 to move over WH1 (presumably targeting FocP too), the resulting configuration would look as follows, where, Hagstrom (1998) argues, [Q]-movement is unaffected by wh-movement in that WH2 and [Q] move separately (“Q-stranding”): (36) [CP [Q]-C [FocP WH2 Foc [TP WH1 T [AgrOP WH2-[Q] AgrO [vP …]]]]
(PL)
Q-stranding seems like an ad hoc mechanism, most likely applicable to German only. Moreover, it’s not quite clear what [Q] looks like, where it’s inserted, how it moves etc.10 Thus, if we could find a way to not invoke this
German is a multiple wh-fronting language!
strategy, we would save Table 2 and (18). This would imply that the syntactic operations in multiple wh-questions in German would look more like those in Multiple wh-movement languages, namely that all WHs undergo movement into the Comp-layer of the clause. Let’s look at what the Hagstrom-Boškovic´ approach has to say for these. One assumption I am making for the time being is that FocP and FP are the landing sites for multiply fronted WHs (where F hosts high [Q]; cf. Cardinaletti and Roberts 1991, Uriagereka 1995, Rizzi 1997, Roussou 1998, and many more for a low Comp-position). This assumption would assign the structures in (38) for (37): (37) a.
Cine ce a vazut? who what has seen ‘Who saw what?’ b. Koj kogo vižda? who whom sees ‘Who sees whom?’
The rough structures in (38) apply to all relevant questions in Multiple wh-movement languages: WH2 > WH1 is not attested, as the languages in question are sensitive to Syntactic Superiority — though not German, the “new unlikely candidate”. In the following, I will explore a finer characterization of the three different wh-displacement types in general and, more narrowly, the relevance of the notion “Multiple wh-movement language” for German. It will turn out that by understanding wh-movement to denote a displacement operation of a WH into the Comp-layer, an articulated CP, we can analyse German as involving multiple wh-movement. This claim warrants some reconditioning which will be done in Section 5. The upshot will be that while patterning with Bulgarian in moving all WHs into CP, German differs in the nature of the movement operation.
4. Some evidence in favour of multiple wh-fronting in German If all WHs in German were fronted the same way as in “regular” Multiple wh-movement languages (e.g. Bulgarian or Romanian), it should have syntactic
113
114
Kleanthes K. Grohmann
consequences or reflexes. In particular such an approach would make certain predictions, e.g. low [Q] is higher than high [Q], only topics may appear in between the WHs etc. The target structure would be this: (39)
CP C¢ [Q]-C0
FocP WH1 V-...-Foc0
Foc¢ TopP* Top¢ Top0
FP WH2
F¢ (low [Q]) F0
TP (high [Q])
A simple application of the above discussion would yield this rough structure: (40) [CP [Q]-C [FocP wer hat-Foc [FP was [Q]-F [TP wer … was … gekauft]]]] who has what bought ‘Who bought what?’
Let’s see whether we can find independent evidence for such a derivation. 4.1 “Intervention effects” One type of argument in favour of multiple wh-fronting comes from so-called “intervention effects”. Beck (1996) observes that certain quantificational elements may not appear in between two WHs, but may well follow them. She postulates a barrier created by these “intervening” quantifiers, on the assumption that all WHs must move at LF. This assumption is not shared here; see Hornstein (1995, 2001), Sabel (1998), Simpson (2000) and others for doubts on covert wh-movement. And indeed, it will be shown that the “intervention” phenomenon is more general,
German is a multiple wh-fronting language!
namely that non-topicalizable elements cannot sit in TopP, ruling out the ungrammatical cases as desired. Many elements can occur between two WHs as well as follow them: (41) a. Wer hat Maria/seine Freundin/einen Porsche (gestern) wo geküßt? b. Wer hat wo Maria/seine Freundin /einen Porsche (gestern) who has where Maria /his girlfriend/a Porsche (yesterday geküßt? kissed ‘Who kissed Maria/his girlfriend/a Porsche (yesterday) where?’
This is not so for all quantificational phrases, though (cf. Beck 1996): (42) a.
Wer hat alle Bücher wo gekauft? who has all books where bought ‘Who bought all (the) books where?’ b. *Wer hat kein Buch wo gekauft? who has no book where bought ‘Who bought no book where?’
Universal quantifiers seem to be able to appear in between the two WHs, while negative ones do not. If all WHs really must end up in CP at one point (through wh-movement at LF, for example), Beck’s barrier-approach would constitute one possibility. Note that nothing rules out the co-occurrence of two WHs and a negative quantifier: if the quantifier follows all WHs, the question is well-formed. (43) a.
Wer hat gestern wo alle Bücher gekauft? who has yesterday where all books bought ‘Who bought all books where yesterday?’ b. Wer hat gestern wo kein Buch gekauft? who has yesterday where no book bought ‘Who bought no book where yesterday?
However, not only negative quantifiers are prohibited from “intervening”. And the class of possible interveners is not restricted to universal quantifiers either. (44) a.
Wer hat viele /die meisten/mehr als drei Bücher wo gekauft? who has many/the most /more than three books where bought ‘Who bought many/the most/more than three books where?’
115
116
Kleanthes K. Grohmann
b. *Wer hat wenige/höchstens drei /weniger als drei Bücher who has few /at-most three/fewer than three books wo gekauft? where bought ‘Who bought few/at most three/fewer than three books where?’
And again, if the quantifiers follow the WHs, everything is alright: (45) a.
Wer hat wo viele /die meisten/mehr als drei Bücher gekauft? who has where many/the most /more than three books bought ‘Who bought many/the most/more than three books where?’ b. Wer hat wo wenige/höchstens drei /weniger als drei who has where few /at-most three/fewer than three Bücher gekauft? books bought ‘Who bought few/at most three/fewer than three books where?’
These contrasts allow for the following generalization (from Grohmann 1998; cf. Beck 1996: 30 in passing, Citko and Grohmann 2000 for discussion, and Boeckx 1999 for French): (46) Quantifier-intervention generalization Monotone increasing quantifiers may appear in between two wh-phrases, while monotone decreasing quantifiers may not appear in between two wh-phrases.
We can now restate the “intervention effect”. Given the articulated Comp-layer and the purported landing sites for A¢-moved WHs in (39), any material intervening between the two WHs must target TopP. An explanation for the ungrammaticality of WH > Qdecr > WH is that decreasing quantifiers cannot be topicalized. 4.2 Non-topicalizability There is independent evidence for the assumption that (monotone) decreasing quantifiers are “non-topicalizable” (i.e. that they may not be topicalized/move to SpecTopP):11 (47) a.
VIELE Bücher hat Peter gestern gelesen (, nicht ALLE). many books has Peter yesterday read not all ‘MANY books Peter read yesterday (, not ALL).’
German is a multiple wh-fronting language!
b. Viele Bücher hat Peter gestern gelesen. (Er ist fleißig.) many books has Peter yesterday read (he is industrious ‘Many books, Peter read yesterday. (He is a hard worker.)’ (48) a.
WENIGE Bücher hat Peter gestern gelesen (, nicht VIELE). few books has Peter yesterday read not many ‘FEW books Peter read yesterday (, not MANY).’ b. *Wenige Bücher hat Peter gestern gelesen. (Er ist faul.) few books has Peter yesterday read (he is lazy *‘Few books, Peter read yesterday. (He is lazy.)’
In this context, consider the following questions: (49) a. b. c. d.
What happened? What did Peter do yesterday? Did Peter read few/many books yesterday? How many books did Peter read yesterday?
For a felicitous answer to (49a–b) in German, the new information appears in a default focus position, where the type of focus is “information focus” (Kiss 1998) and the “default” position is rather low in the structure (see e.g. Abraham 1995, Meinunger 2000). Considering (49b), the relevant one for us, note that neither construction in (47) or (48) can be used; the only word order of a felicitous answer is in (50): (50) a.
Peter hat gestern viele Bücher gelesen. Peter has yesterday many books read ‘Peter read many books yesterday.’ b. Peter hat gestern wenige Bücher gelesen. Peter has yesterday few books read ‘Peter read few books yesterday.’
However, (47a) and (48a) can be used as felicitous answers to questions like (49c–d), where the answer contains a contrastive element as part of the new information. (47b) and (48b) are simple topic structures: there is no contrastive stress, and the fronted elements can only pick up a referent already established in the discourse; both are infelicitous.12 4.3 Intervening adverbs Certain adverbs may not “intervene” in between two WHs either (Citko and Grohmann 2001), and it can be shown that these too are non-topicalizable.
117
118
Kleanthes K. Grohmann
Obviously, this could hardly be captured under a barrier-approach, even assuming LF-movement. As it turns out, however, non-topicalizable adverbs may be fronted if they are contrastively stressed. In these cases we would expect them also to be legitimate in between two WHs. Take, for example, manner adverbs, which arguably sit low in the clause structure. These must follow the direct object, unless it is focused: (51) a.
Peter hat das Buch gerne /komplett gelesen. Peter has the book with.pleasure/completely read ‘Peter read the book with pleasure/completely.’ b. Peter hat gerne/komplett das BUCH gelesen.
In combination with multiple WHs, these adverbs may not appear in between the two: (52) a.
Wer hat was gerne /komplett gelesen? who has what with.pleasure/completely read ‘Who read what with pleasure/completely?’ b. *Wer hat gerne/komplett was gelesen? c. Was hat wer gerne/komplett gelesen? d. *Was hat gerne/komplett wer gelesen?
Moreover, such adverbs may not be topicalized, in contrast to sentence adverbs like gestern ‘yesterday’ and epistemic modal adverbs such as wahrscheinlich ‘probably’: (53) a. *Gerne /Komplett hat der Peter das Buch gelesen. with.pleasure/completely has the Peter the book read *‘With pleasure/Completely, Peter read the book.’ b. Wahrscheinlich/Gestern hat der Peter das Buch gelesen. probably /yesterday has the Peter the book read ‘Probably/Yesterday, Peter read the book.’
But again, they may be fronted when focalized. In this case, they may also appear in between the two WHs (see note 12): (54) a.
GERNE/KOMPLETT hat der Peter das Buch gelesen. ‘WITH PLEASURE/COMPLETELY, Peter read the book.’ b. Wer hat GERNE/KOMPLETT was gelesen? ‘Who read what PLEASURE/COMPLETELY?’
The same contrasts can also be found in embedded contexts.
German is a multiple wh-fronting language!
4.4 Non-operator A¢-properties If all WHs do indeed (A¢-)move overtly into an articulated Comp-layer in German, we would expect this operation to have syntactic consequences that go beyond cooccurrence or “intervention” effects, such as A¢-binding.13 Now, as it turns out, such effects are notoriously difficult to show in German. But interestingly, we can find some — and while these don’t necessarily suggest A¢-movement proper, such as wh-movement in English, they do suggest some sort of movement, even (and especially) of the lower WH, or at least are not in conflict with the current proposal. I will turn to a finer understanding of this in the next section. Take parasitic gaps, for example, which are licensed by overtly A¢-moved elements only — at least in English (see Chomsky 1982, Engdahl 1983, and much subsequent work).14 As it turns out, the lower WH in German may license a parasitic gap PG, thus at first glance suggesting overt A¢-movement: (55) a.
Wer hat weni ohne PGi zu kennen eingeladen? who has who without to know invited ‘Who invited whom without knowing *(him)?’ b. Wer hat wasi ohne PGi gelesen zu haben rezensiert? who has what without read to have reviewed ‘Who reviewed what without reading *(it)?’
In order to go through as a bona fide argument in favour of overt displacement of the lower WH into the Comp-layer, it has to be shown that German does indeed have parasitic gaps on a par with English.15 But since WHs don’t scramble in German (viz. Japanese “wh-scrambling”), the landing site of the lower WH cannot be the “typical” scrambling position. In effect, the licensing of parasitic gaps by the lower WH can indeed be integrated into the analysis that the landing site is within the Comp-layer, and even as things stand now concerning the status of parasitic gaps in German, data such as (55) don’t provide a counter-argument to my proposal. Weak crossover effects are obviated by A¢-elements (Chomsky 1977 and much subsequent work). In fact, these elements must be in a non-operator A¢-position (in English). Again, the lower WH in German acts accordingly which is predicted if it is indeed A¢-moved:16 (56) a.
Was hat wemi seinei Mutter gegeben? what has who his mother given ‘What did hisi mother give to who*i/k?’ (cf. ‘Whoi did hisi mother give what?’)
119
120 Kleanthes K. Grohmann
b. Wann hat weni seini Vater geschlagen? when has who his father beaten ‘When did hisi father beat who*i/k?’
While these data might be interesting in and of themselves, and possibly serve as further corroborating evidence for the hypothesis that German is a multiple wh-fronting language like Bulgarian, they might also suggest an alternative implementation of this hypothesis. Both WHs may be fronted into an articulated CP-layer — but maybe not to the positions indicated in (39), clear A¢-operator positions. Maybe these fronted WHs target A¢-positions, but don’t bear operator-status (as the current section suggested). Next I will explore where this suggestion could lead to.
5. The Italian caveat: No “real” multiple wh-questions in German?! The non-traditional proposal that all WHs in German move very high in the clause structure is driven by typological and interpretive considerations. It is further empirically supported by the interaction with non-topicalizable material and other syntactic effects, given reasonable assumptions on wh-question formation in general. At the same time, one difference between German and the “regular” Multiple wh-movement languages (like Bulgarian or Romanian) relates to the fact that the quantificational character of WHs in multiple questions is restricted by the discourse, an observation which could be linked to the absence of multiple question formation in Italian, tying it in with the syntactic derivations proposed. The ensuing discussion will thus follow up on the remarkable result from the previous section that although German fronts all WHs into Comp, they don’t seem to have the same quantificational A¢-status as a singularly fronted WH in, say, English. 5.1 Italian In passing, Rizzi (1978) mentions that Italian does not allow multiple wh-questions at all. (57) a. *Chi ha scritto che cosa? who has written what ‘Who wrote what?’
(Italian)
German is a multiple wh-fronting language!
b. *Chi è partito quando? who is left when ‘Who left when?’
Calabrese (1984, 1992) offers an explanation for this “odd” ban: a multiple wh-question, if not used as an echo or quiz question, asks for “new information” regarding all WHs. New information in a declarative sentence is construed with non-default sentential stress — or focus. The type of focus relevant for a felicitous reply to a question is information (or presentational) focus, not identificational (or, as I use it, contrastive) focus (see the discussion in Kiss 1998, for example). In Italian, the position for information focus is unique and so, argues Calabrese, it follows that multiple wh-questions are ill-formed: the multiple information requested cannot be realized in the answer. (58) a. *MARIO ha scritto una LETTERA. Mario has written a letter ‘MARIO wrote a LETTER.’ b. *FRANCO è partito alle CINQUE. Franco is left at five ‘FRANCO left at FIVE O’CLOCK.’
(Italian)
This state of affairs is independent of an SP- or PL-reading, as Calabrese shows: (59) #MARIO ha scritto una LETTERA, FRANCO un ARTICOLO, CARLO un LIBRO… (Italian) ‘MARIO has written a LETTER, FRANCO an ARTICLE, CARLO a BOOK…’ (Calabrese 1984: 67)
The intended interpretation of (59) is not contrastive, but information focus (cf. ‘What did people write, and who did it?’), hence the hash mark. (Presumably, an utterance like (59) is felicitous and well-formed in a context that focuses contrast.) As it turns out, we can detect a similar effect in German: the analogue of (59) can only be used contrastively as well. 5.2 Discourse restrictions After the evidence suggested above, a further argument for multiple wh-displacement into the Comp-layer in German comes from discourse restrictions. For example, compare the following two discourse contexts and the (in)felicity of a multiple wh-language between English and German:
121
122 Kleanthes K. Grohmann
(60) Context I: A man comes to a newsstand and just sees three people leaving in different directions, each fiddling with their purchases. Asking the newsagent whether he sold anything interesting to those three people, the newsagent replies: “I can’t believe it! I just sold an Anarchist newspaper, a Fascist magazine and the Christian church news.” a. Who bought what? b. Wer hat was gekauft? who has what bought ‘Who bought what?’ (61) Context II: A jeweler comes home to his wife for lunch and says excited: “I had a great morning! I sold a platinum watch, a gold necklace and a titanium wedding band.” a. Who bought what? b. #Wer hat was gekauft? who has what bought ‘Who bought what?’
As can be seen, German only allows for a felicitous binary wh-question if the set of individuals is part of the common ground between speaker and hearer, introduced in the discourse, where (61a) is a perfectly reasonable question and a potential answer would be something like (62):17 (62) A businessman bought the platinum watch, an old lady bought the gold necklace and a young groom bought the titanium wedding band.
This felicity condition is captured by Discourse Restricted Quantification: (63) Discourse Restricted Quantification (DRQ; from Grohmann 2000: 269) Questions involving two wh-expressions are well-formed if the value of both wh-expressions is determined by the context; determination of values is satisfied by providing a set of at least two possible referents in the discourse.
In other words, DRQ forces a certain salience of the referents to all WHs in a multiple wh-question. As the above contrasts show, it strictly applies to German, but not to English. 5.3 Absence of multiple information questions What DRQ essentially says is that all WHs in a German multiple wh-question must be D-linked (Pesetsky 1987). While a WH in an information question
German is a multiple wh-fronting language!
represents a novel set of individuals (“new information”), the set of individuals represented by D-linked WHs is to some degree known (“old information”). The term DRQ can be used to distinguish a condition of D-linking on all WHs (in German), while D-linking simply applies to those elements commonly identified as such (e.g. which-phrases). It seems that German and Italian pattern alike. In a sense, then, German multiple wh-questions are not really questions. However, German has the option of marking contrastiveness in two positions: by fronting the relevant element or by assigning it heavy stress in low position. Thus, if a question is formed that asks for contrastive information (in a syntactically well-formed way), a possible answer is also well-formed syntactically. However, German and Italian apparently differ with respect to D-linking: (64) *Quale ragazza ha dato un bacio a quale ragazzo? which girl has given a kiss to which boy ‘Which girl kissed which boy?’ (Calabrese 1984: 67)
(Italian)
I leave this discrepancy for future research. But note that if a multiple question with two bare WHs is D-linked, it might provide us with a clue as to why Syntactic Superiority may be violated in German: the two wh-elements are actually D-linked, which — for reasons that are beyond the scope of this discussion — may circumvent the Superiority Condition (see Boeckx and Grohmann, to appear for suggestions). 5.4 More on “D-linking” German requires all DRQed/“D-linked” WHs to move into the Comp-layer and as a result always force a PL-reading, where the individuals of each set are contrasted with one another. A clue comes from pairing the order of WHs in the question and the order of the values for each WH in the answer: (65) a. Wer hat was gekauft? b. Was hat wer gekauft? ‘Who bought what?’ (66) a. Der Peter hat das Buch gekauft, der Martin hat das Fahrrad gekauft … b. Das Buch hat der Peter gekauft, das Fahrrad hat der Martin gekauft … ‘Peter bought the book, Martin bought the bicycle…’
If the order of WHs in the question is WH1 > WH2, the appropriate answer follows the same order for the values of each WH. However, if the order in the
123
124 Kleanthes K. Grohmann
question is WH2 > WH1, the order in the answer is analogous (see also Kiss 1993: 103f., fn. 6 for Hungarian). If the first position in a declarative matrix clause can be a contrastive position, and if contrastive focus can also be assigned in situ, it follows that not only a PL-reading is available, but also the two positions are made use of in the same way. Here we find a contrast with English. Consider the following: (67) a. Which man killed which Kennedy? b. #Which Kennedy did which man kill?
(English: PL) (English: *PL)
Following Barss’ (1990) observation that D-linked wh-questions in English which front WH2 over WH1 can only have an SP-reading, (67b) is predicted to be ill-formed (cf. Barss 1999). Incidentally, (67) is a good example to show that the PL-reading is not available in these contexts. Not surprisingly, the analogues of (67) are both well-formed in German: (68) a.
Welcher Mann hat welchen Kennedy ermordet? which man has which Kennedy killed ‘Which man killed which Kennedy?’ b. Welchen Kennedy hat welcher Mann ermordet? which Kennedy has which man killed #‘Which Kennedy did which man kill?’
(German: PL)
(German: PL)
However, the only felicitous replies for the a- and b-cases are the following: (69) a.
Lee Harvey Oswald hat John ermordet und Sirhan Bishara Sirhan Robert. ‘Lee Harvey Oswald killed John and Sirhan Bishara Sirhan Robert.’ b. John hat Lee Harvey Oswald ermordet und Robert Sirhan Bishara Sirhan. lit. ‘John, Lee Harvey Oswald killed and Robert, Sirhan Bishara Sirhan.’
5.5 Final comparison In sum, while German seems to pattern with Bulgarian in multiple wh-question formation in that both move all WHs into the Comp-layer, it also resembles Italian in not forming an information multiple question, but necessarily D-links all WHs. Unlike English, however, (at least bare) D-linked WHs must be licensed in the Comp-layer, a fact we might want to explain by the prolific nature of the left periphery in German, in that it allows much more discourserelated movement into the articulated Comp-layer. Fronted WHs in Bulgarian may not be split up, and topics precede (rather than follow).
German is a multiple wh-fronting language!
(70) a. Ivan vcˇera kakvo kupi? b. Vcˇera Ivan kakvo kupi? yesterday Ivan what bought ‘What did Ivan buy yesterday?’ (Richards 1997: 111) (71) a. *Koj vcˇera kakvo kupi? b. Vcˇera koj kakvo kupi? yesterday who what bought ‘Who bought what yesterday?’
The same effects can be found in Romanian (Ileana Comorovski, p.c.). We will thus have to say something more about the landing sites, and what’s going on in the Comp-layer(s).
6. Conclusion I first proposed a typological tripartition of wh-movement into Zero, Singular and Multiple wh-movement languages, based on the (non-)availability of single-pair readings. I suggested that German is like Bulgarian in that neither allows such a reading which seems to be a general property of languages in which all WHs move overtly into the Comp-layer. I then substantiated this claim with syntactic evidence of two types: intervention effects with quantifiers and adverbs in between the two WHs and binding effects with parasitic gaps and in weak crossover configurations. I finally suggested that German is like Italian in not forming “real” (= information) multiple wh-questions to begin with on the basis of Discourse Restricted Quantification, where all (bare) WHs in German multiple questions are essentially D-linked, and the availability of high and low focus positions plays a role to distinguish the two. I leave further issues for future research (such as a finer exploration of the different projections in the Comp-layer, the role of discourse-driven displacement, cross-linguistic variation, and certainly a better-informed view of focus regarding its syntactic and semantic licensing through stress, interpretation, positions, movement etc.). If on the right track, this investigation sheds new light on a typology of (multiple) wh-question formation on the one hand and the syntax of D-linked WHs, still a rather understudied property of grammar, on the other. Especially when we consider the discourse properties observed in German — and possibly linked to Italian, but also applied to D-linking in English — the unorthodox view that German is a multiple wh-fronting language should not sound so strange anymore, which I would take as a positive result.
125
126 Kleanthes K. Grohmann
Notes * This article is a slightly revised version of Grohmann (2002), itself an adaptation of Grohmann (2000: Ch. 5). It benefited from additional presentations at the Workshop on the Syntax–Semantics Interface in the CP-Domain (ZAS Berlin, March 6–8, 2002), the Linguistischer Arbeitskreis (University of Cologne, December 4, 2002), and the workshop “On wh-Movement” (Leiden University Centre for Linguistics and Utrecht Institute of Linguistics, December 11–13, 2002), whose audiences I am very grateful to for valuable feedback. My gratitude extends to the audiences of various other presentations relating to multiple wh-questions in German over the past five years or so. I am particularly grateful to Cedric Boeckx, Željko Boškovic´, Lisa Cheng, Robert Kemp, Howard Lasnik, Horst Lohnstein, Øystein Nilsen, Marga Reis, Luigi Rizzi, Joachim Sabel, and Peter Svenonius for constructive pointers and discussion from which the present version gained a lot. 1. For purposes of presentation, I concentrate exclusively on binary wh-questions. At this point, I have to leave open how, or even whether, the claims carry over to questions with more than two WHs. It would not be at all surprising if they didn’t (at least not without further refinements), given, for example, the well-known “additional wh-effect”, first noted for English by Kayne (1984). 2. See also Dayal (2002) for some relevant discussion, especially in long-distance contexts. 3. Throughout, infelicitous questions will be marked with the hash mark ‘#’, while unavailability of a particular reading receives the standard marker for ungrammaticality, the star ‘*’. 4. And by extension, presumably the other multiple wh-fronting languages that don’t target Comp as well, pending a more careful investigation of the facts in Czech, Polish, Russian etc. 5. “Superiority” refers to movement of a lower WH2 over a higher WH1, where the variations discussed in the text all draw from the original descriptive Superiority Condition (Chomsky 1973: 246): (i) a. b.
No rule can involve X, Y in the structure … X … [α … Z … __ WYV] … where the rule applies ambiguously to Z and Y, and Z is superior to Y. The category A is ‘superior’ to the category B if every major category dominating A dominates B as well but not conversely.
In this work, I will not be concerned with theoretical means to capture Superiority. But note that the tripartite line pursued by Boškovic´ (1998) is nothing but a differentiated line of explanation also explicit in other approaches. Haider (2000: 247), for example, also denies Superiority the status of a “uniform grammatical property” and links Superiority effects to the interplay of syntax, semantics, parsing and discourse properties of a language. This not only fits with Boškovic´’s approach, but also with the reasoning developed in Section 5 below, when I address Superiority issues in German (and English). 6. Among the limited array of languages listed in (9), I only mention those for which I offer at least one piece of data in this paper. Note that this list by far not exhaustive. 7. Regarding the languages cited, see note 6. Other wh-in situ languages arguably include Hindi, for example, singular wh-movement languages Swedish, to name just two. It would
German is a multiple wh-fronting language! 127
be expected that the multiple wh-fronting languages Czech, Polish or Russian classify like Serbo-Croatian (and Japanese!) rather than Bulgarian. 8. I abstract away from those cases, in which Serbo-Croatian does in fact show Superiority effects (of the standard sort, viz. “Syntactic Superiority”). As Boškovic´ (1997, 2002) shows, these occur, making Serbo-Croatian look like Bulgarian, when there is overt evidence for C. 9. In this context, TP and AgrOP are simply mnemonics; a scrambling analysis doesn’t concern me here, but in line with Grohmann (2000, 2003), I want to avoid XP-movement/ adjunction within this layer. 10. See Grohmann (2000) for some discussion beyond Hagstrom (1998). 11. There seems to be disagreement among speakers as to whether universal (monotone increasing) quantifiers may indeed be topicalized, or whether fronting into the first position of a matrix clause is necessarily focus-driven. The data presented in the text reflect my own judgements and at least some speakers’ intuitions. The observation that monotone increasing quantifiers may, but monotone decreasing ones may not topicalize was to my knowledge first made explicit by Citko and Grohmann (2000, 2001). Interestingly, Grewendorf (2002: 77–79) comes to virtually the same conclusion. 12. In this connection it may be interesting to note that once contrastively stressed, the quantifiers identified by Beck as “interveners” — monotone decreasing quantifiers — become suddenly good (even clearer with the focus particles sogar/nur): (i) a.
b.
Wer hat (sogar) VIELE Bücher wo gekauft? who has (even many books where bought ‘Who bought (even) MANY books where?’ Wer hat (nur) WENIGE Bücher wo gekauft? who has (only few books where bought ‘Who bought (only) FEW books where?’
These questions carry quite a different meaning, however, and I will this leave them aside. 13. I am grateful to Joachim Sabel for bringing these diagnostics to my attention. 14. Nissenbaum (2000) suggests a revision of Engdahl’s (1985) original claim that wh-in situ cannot license parasitic gaps: multiple parasitic-gap licensing is possible in English, and it bears on overt A¢-movement of the so-called “in situ” WH (which seems to be “D-linked” by necessity, a property I return to in the next section). 15. This is not at all clear. See, among many others, Koster (1987) on parasitic gaps in SOV languages, Bennis and Hoekstra (1984) for Dutch, Felix (1985) and Webelhuth (1992) for German. Grewendorf (2002: 234) also employs the idea that German wh-movement licenses parasitic gaps. 16. And again, it is not clear that German shows real weak crossover effects as English; see Haider (1993) for critical inspection, and also Grohmann (2000, 2003), Grewendorf (2002: 74–80, 234–236, 296–299) and Sabel (2002) for discussion. I assume here that obviation of weak crossover effects is indeed a property of non-operator A¢-moved elements, though not necessarily exclusively. 17. In fact, as Peter Svenonius points out, this state of affairs carries over to English. While in a situation that conforms to DRQ (see below), such as (60), even the structure violating
128 Kleanthes K. Grohmann
Syntactic Superiority is well-formed, it is not so in (61). In other words, English speakers may employ What did who buy? in a context that identifies potential referents of both WHs. This further confirms Haider’s (2000) intuition that Superiority is not a uniform condition on the grammar, evident in much current research, but rather a complex epiphenomenon of the interplay of various grammatical properties. More has to be said which ones it actually concerns. Haider concentrates on syntax–semantics/parsing, Boškovic´ (1998) on syntactic word order and interpretive constraints, while I investigate the role of discourse proper on the syntax (viz. D-linking/DRQ).
References Abraham, W. 1995. Deutsche Syntax im Sprachenvergleich. Grundlegung einer typologischen Syntax des Deutschen. Tübingen: Gunter Narr. Barss, A. 1990. Optional movement, absorption, and the interpretation of wh-in situ. Paper presented at NELS 21, Université de Québec à Montreal. [October 1990] Barss, A. 1999. “Minimalism and asymmetric wh-interpretation.” In Step by Step. Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 31–52. Cambridge, Mass.: MIT Press. Beck, S. 1996. “Quantified structures as barriers for LF-movement.” Natural Language Semantics 4: 1–56. Bennis, H. and Hoekstra, T. 1984. “Gaps and parasitic gaps.” The Linguistic Review 4: 29–89. Boeckx, C. 1999. “Decomposing French questions.” In Proceedings of the 23rd Annual Penn Linguistics Colloquium. University of Pennsylvania Working Papers in Linguistics 6.1, J. Alexander, N.-R. Han and M. Minnick Fox (eds), 69–80. Philadelphia, Penn.: Department of Linguistics, University of Pennsylvania. Boeckx, C. and Grohmann, K. K. To appear. “Sub-Move: Towards a unified account of scrambling and D-linking.” In Peripheries, D. Adger, C. de Cat and G. Tsoulas (eds). Dordrecht: Kluwer. Boškovic´, Ž. 1997. “Superiority effects with multiple wh-fronting in Serbo-Croatian.” Lingua 102: 1–20. Boškovic´, Ž. 1998. “On the interpretation of multiple questions.” In A Celebration: Essays for Noam Chomsky’s 70th Birthday, J. Fodor, S. J. Keyser and A. Brand (eds). Cambridge, Mass.: MIT Press. [Online publication — available for download at http:// cognet.mit.edu/Books/celebration/essays/Boeckx.html (sic); revised version to appear in Linguistic Variation Yearbook 1.] Boškovic´, Ž. 2002. “On multiple wh-fronting.” Linguistic Inquiry 33: 351–383. Calabrese, A. 1984. “Multiple questions and focus in Italian.” In Sentential Complementation, W. de Geest and Y. Putseys (eds), 67–74. Dordrecht: Foris. Calabrese, A. 1992. “Some remarks on focus and logical structure in Italian.” In Harvard Working Papers in Linguistics 1, S. Kuno and H. Thráinsson (eds), 91–127. Cambridge, Mass.: Department of Linguistics, Harvard University. Cardinaletti, A. and Roberts, I. 1991. Clause structure and X-second. Ms., University of Venice and University of Wales, Bangor.
German is a multiple wh-fronting language! 129
Cheng, L. L.-S. 1991. On the typology of wh-questions. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Chomsky, N. 1973. “Conditions on transformations.” In A Festschrift for Morris Halle, S. R. Anderson and P. Kiparsky (eds), 232–286. New York: Holt, Rinehart, & Winston. Chomsky, N. 1977. Essays on Form and Interpretation. Amsterdam: Elsevier North-Holland. Chomsky, N. 1982. Some Concepts and Consequences of the Theory of Government and Binding. Cambridge, Mass.: MIT Press. Citko, B. and Grohmann, K. K. 2000. A new argument in favour of a syntactic focus projection. Paper presented at the Focus Workshop of GLOW 23, University of Deusto, Bilbao. [April 2000] Citko, B. and Grohmann, K. K. 2001. “The non-uniqueness of multiple wh-fronting: German = Bulgarian.” In Formal Approaches to Slavic Linguistics: The Indiana Meeting, 2000, S. Franks, T. King and M. Michael Yadroff (eds), 117–136. Ann Arbor, Mich.: Michigan Slavic Publications. Comorovski, I. 1996. Interrogative Phrases and the Syntax–Semantics Interface. Dordrecht: Kluwer. Dayal, V. 2002. “Single-pair vs. multiple-pair answers: Wh in-situ and scope.” Linguistic Inquiry 33: 512–520. Engdahl, E. 1983. “Parasitic gaps.” Linguistics and Philosophy 6: 5–34. Engdahl, E. 1985. “Parasitic gaps, resumptive pronouns, and subject extractions.” Linguistics 23: 3–44. Felix, S. 1985. “Parasitic gaps in German.” In Erklärende Syntax des Deutschen, W. Abraham (ed), 173–200. Tübingen: Narr. Grewendorf, G. 2002. Minimalistische Syntax. Tübingen: Francke. Grohmann, K. K. 1998. “Syntactic inquiries into discourse restrictions on multiple interrogatives.” Groninger Arbeiten zur germanistischen Linguistik 42, 1–60. Grohmann, K. K. 2000. Prolific peripheries: A radical view from the left. Doctoral dissertation, University of Maryland, College Park. Grohmann, K. K. 2002. “Multiple wh-fronting and the left periphery: German = Bulgarian + Italian.” In Georgetown University Working Papers in Theoretical Linguistics 2, S. Mauck and J. Mittelstaedt (eds), 83–115. Washington, D. C.: Department of Linguistics, Georgetown University. Grohmann, K. K. 2003. Prolific Domains. On the Anti-Locality of movement dependencies. To appear with Amsterdam: John Benjamins. Hagstrom, P. 1998. Decomposing questions. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Haider, H. 1993. Deutsche Syntax — generativ. Tübingen: Narr. Haider, H. 2000. “Towards a superior account of Superiority.” In Wh-Scope Marking, U. Lutz, G. Müller and A. von Stechow (eds), 231–248. Amsterdam: John Benjamins. Hornstein, N. 1995. Logical Form. From GB to Minimalism. Oxford: Blackwell. Hornstein, N. 2001. Move! A Minimalist Theory of Construal. Oxford: Blackwell. Horvath, J. 1986. FOCUS in the Theory of Grammar and the Syntax of Hungarian. Dordrecht: Foris. Kayne, R. 1984. Connectedness and Binary Branching. Dordrecht: Foris.
130 Kleanthes K. Grohmann
É. Kiss, K. 1993. “Wh-movement and specificity.” Natural Language and Linguistic Theory 11: 85–120. É. Kiss, K. 1998. “Identificational focus versus information focus.” Language 74: 245–273. Koster, J. 1987. Domains and Dynasties. Dordrecht: Foris. Meinunger, A. 2000. Syntactic Aspects of Topic and Comment. Amsterdam: John Benjamins. Nissenbaum, J. 2000. Investigations of covert phrase movement. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Pesetsky, D. 1987. “Wh-in situ: Movement and unselective binding.” In The Representation of (In)definiteness, A. G. B. ter Meulen and E. Reuland (eds), 98–129. Cambridge, Mass.: MIT Press. Richards, N. 1997. What moves where when in which language? Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Rizzi, L. 1978. “Violations of the wh-island constraint in Italian and the Subjacency Condition.” In Montreal Working Papers in Linguistics 11, C. Dubuisson, D. Lightfoot and Y. C. Morin (eds), 155–190. Montreal: L’association linguistique de Montreal. [Reprinted in Rizzi (1982), Italian Syntax. Dordrecht: Foris, 49–76.] Rizzi, L. 1997. “The fine structure of the left periphery.” In Elements of Grammar: Handbook of Generative Syntax, L. Haegeman (ed), 281–337. Dordrecht: Kluwer. Roussou, A. 1998. “Wh-interrogatives: From Classical Greek to Modern Greek.” In The Proceedings of the 12th International Symposium on Theoretical and Applied Linguistics, Vol. 1, S. Lambropoulou (ed), 109–126. Thessaloniki: Aristotle University. Rudin, C. 1988. “On multiple questions and multiple wh-fronting.” Natural Language and Linguistic Theory 6: 445–501. Sabel, J. 1998. Principles and parameters of wh-movement. Habilitationsschrift, University of Frankfurt. Sabel, J. 2000. “Partial wh-movement and the typology of wh-questions.” In Wh-Scope Marking, U. Lutz, G. Müller and A. von Stechow (eds), 409–446. Amsterdam: John Benjamins. Sabel, J. 2002. “Intermediate traces, reconstruction and locality effects.” In Theoretical Approaches to Universals, A. Alexiadou (ed), 259–313. Amsterdam: John Benjamins. Simpson, A. 2000. Wh-Movement and the Theory of Feature Checking. Amsterdam: John Benjamins. Stjepanovic´, S. 1995. Short distance movement of wh-phrases in Serbo-Croatian matrix clauses. Ms., University of Connecticut, Storrs. Uriagereka, J. 1995. “An F position in Western Romance.” In Discourse Configurational Languages, K. È. Kiss (ed), 153–175. Oxford: Oxford University Press. Wachowicz, K. 1974. On the syntax and semantics of multiple questions. Doctoral dissertation, University of Texas, Austin. Webelhuth, G. 1992. Principles and Parameters of Syntactic Saturation. Oxford: Oxford University Press.
Deriving Anti-Superiority effects Multiple wh-questions in Japanese and Korean* Youngmi Jeong University of Maryland
1.
Introduction
The present paper seeks to obtain a deeper understanding of two conditions on multiple questions in Japanese (and Korean): so-called Anti-Superiority and the additional wh-effect. Anti-Superiority refers to the phenomenon illustrated in (1).1 (1) a. *Naze nani-o anata-wa katta no? why what-acc you-top bought q ‘Why did you buy what?’ b. *Wae mwues-ul ne-nun sa-ess-ni? why what-acc you-top buy-past-q ‘Why did you buy what?’
(Japanese)
(Korean)
Descriptively, Anti-Superiority means that the adjunct wh-phrase naze ‘why’ cannot precede another wh-phrase in a multiple wh-question.2 Switching the order results in grammaticality, as the sentences in (2) show. (2) a.
Nani-o naze anata-wa katta no? what-acc why you-top bought q ‘Why did you buy what?’ b. Mwues-ul wae ne-nun sa-ess-ni? what-acc why you-top buy-past-q ‘Why did you buy what?’
As noted by Watanabe (1992) and Saito (1994), among others, Anti-Superiority is avoided if there is an additional wh-element, as in (3c). ((3a) and (3b) complete the paradigm, and, incidentally, show that the facts observed in (1) and (2) are independent of whether material precedes the wh-words; data from Japanese due to Saito 1994.)
132
Youngmi Jeong
(3) a.
John-ga nani-o naze katta no? John-nom what-acc why bought q ‘Why did John buy what?’ b. *John-ga naze nani-o katta no? John-nom why what-acc bought q *‘What did John buy why?’ c. Dare-ga naze nani-o katta no? who-nom why what-acc bought q *‘What did who buy why?’
Both Watanabe and Saito attribute the ungrammaticality of (1) to an ECP violation, and propose a rescuing device to account for (3c). Such accounts (the details of which will be ignored here) are not available in the minimalist program, as they rely on the notion of government, which minimalism seeks to abandon. In this study I offer an alternative account of both Anti-Superiority and its relaxation by combining the effects of three independently motivated proposals: Kayne’s (1994) LCA-based claim that head-final languages are in fact headinitial, Rizzi’s (1997) articulated left periphery of the clause, and Pesetsky’s (2000) construal of the multiple wh-fronting requirement.
2. Core assumptions In this section I lay out the three core assumptions I adopt to approach the phenomena described in Section 1. First, I adopt the essence of Kayne’s (1994) Antisymmetry proposal. In particular, I assume that (i) there is no rightward adjunction, only leftward adjunction is licit; (ii) head-final languages have the same basic configuration as head-initial languages: all of them start off as a universal SVO structure. Head-finality results from movement of the verb to some very high position, which Kayne identifies as C, followed by leftward movement of the IP to a position beyond the landing site of V. Kayne claims that IP moves to SpecCP, and thereby accounts for Bach’s (1971) observation that head-final languages lack overt wh-movement. Kayne’s reasoning is that if IP occupies SpecCP, there is no landing site left for wh-movement. Though compelling at the time it was first made, Kayne’s explanation for the lack of wh-movement in head-final languages must be revised in light of the evidence offered by Rizzi (1997) and many others following him in favor of an articulated CP-structure containing several projections. According to Rizzi, the
Deriving Anti-Superiority effects
traditional notion of CP is to be decomposed as in (4). (4)
ForceP TopP* FocusP TopP FinP IP
For the purposes of this paper I will assume a slightly different hierarchy, given in (5). (5)
ForceP TopP* FocusP XP FinP IP
The only amendment I make to (4) concerns the lower TopP. I will assume that this position is not a dedicated topic position (which has been argued independently, in Beninca and Poletto 2001, among others), but instead a position that hosts scrambled material. Because the featural nature of scrambling is still unclear, I dub this projection XP. We will see some evidence for its existence in the next section. Following Rizzi I assume that wh-phrases land into SpecFocP when they undergo wh-movement. Rizzi’s left periphery allows for wh-movement in head-final languages. The fact that wh-movement is not immediately detected suggests that it is covered up by remnant movement, which is what I will show in the next section, where we consider actual derivations. The third assumption I adopt is from Pesetsky (2000). Most researchers have assumed that in multiple wh-fronting languages, all wh-phrases must
133
134 Youngmi Jeong
check a common feature, capturing the idea that the multiple wh-fronting requirement means that all wh-phrases must undergo movement. Pesetsky proposes instead that all the multiple wh-fronting requirement means is that more than one wh-phrase must front, but not necessarily all of them. Put differently, at least two wh-phrases must front in multiple wh-fronting cases. Pesetsky adduces evidence from Bulgarian for his claim. According to him, sentences like (6) show that in situations where three wh-phrases may front, if only two front, and one remains in situ, no ungrammaticality results. (6) ?Koj kogo e pital kakvo? who whom is asked what ‘Who asked whom what?’
When combined with Rizzi’s split-CP, Pesetsky’s claim amounts to saying that in multiple wh-fronting contexts, FocusP hosts two wh-phrases.
3. Anti-Superiority In this section I show how the three assumptions made in Section 2 conspire to yield Anti-Superiority effects in Japanese and Korean. Let us first see how one would derive a single wh-question sentence like (7). [Throughout, English words are used in the tree representations for ease of exposition.] (7) John-ga naze kitta no? John-nom why came q ‘Why did John come?’
(7) is important for our argument because it involves naze ‘why’, which is the trouble maker. Following Rizzi (1990) and many others (see Boeckx 2001, Bromberger 1986, Law 1991, 1993, Starke 2001, and Uriagereka 1988), I assume, as I did in Jeong (2002), that “why”, in this case, naze, is base-generated in SpecFocP. The correct word order in (7) is obtained by moving the verb kita ‘came’ successive cyclically to X0, and then the remaining FinP/IP moves to SpecTopP via remnant movement (I see no evidence for distinguishing between FinP and IP in the cases discussed here, hence I will assume from now on that they are the same projection). The derivation is schematized in (8).
Deriving Anti-Superiority effects
(8)
ForceP TopP Top¢ FocP
Top why
Foc¢ XP
Foc 2
X¢ FinP/IP
X camei
X
John ti
1
Note, though, that (7) is not the only possible word order. (9) is also available in Japanese. (9) Naze John-ga kita no? why John-nom came q ‘Why did John come?’
To generate this sentence, two possibilities are available. Keeping the position of naze and of the verb constant (SpecFocP and head-adjoined to X0, respectively), we may position the remnant IP in SpecXP, as schematized in (10). Or else, we may claim that IP is in SpecTopP, as in (8), with the difference coming from movement (scrambling) of naze to SpecForceP, as shown in (11). At the moment I have no evidence forcing me to adopt one possibility over the other.
135
136 Youngmi Jeong
(10)
ForceP TopP Top¢ FocP
Top why
Foc¢ XP
Foc
X¢ 2
FinP/IP
X
John ti
X
camei 1
(11)
ForceP TopP Top¢ 3
FocP
Top why
Foc¢ XP
Foc 2
X¢ FinP/IP
X camei
X
John ti
1
With this much background about simple ‘why’-questions, let’s consider the Anti-Superiority phenomenon. Essentially, the question we will be asking is
Deriving Anti-Superiority effects
how naze must not be the first member of a wh-cluster. Given the assumptions made in Section 2, Anti-Superiority is captured in the following way. Naze ‘why’ is base-generated in SpecFocP. The second wh-phrase, nani-o ‘what-acc’, raises and adjoins to naze (in a Kaynean framework, multiple specifiers/adjunctions are unavailable, so adjunction to naze is virtually the only option in this situation). Movement of nani-o to naze meets Pesetsky’s requirement on multiple-wh fronting. At least two wh-phrases are in SpecFocP. Because Kayne’s LCA demands that adjunction always be to the left, and never to the right, the order ·nani-o nazeÒ is the only one available. The ungrammatical order *·naze nani-oÒ is simply underivable. The relevant example and the corresponding derivation are given in (12)–(13). (The position of John-ga is to be handled as discussed above. Nani-o naze John-ga katta no is also possible, as expected.) (12) John-ga nani-o naze katta no? John-nom what-acc why bought q ‘Why did John buy what?’ (13)
ForceP TopP Top¢ FocP
Top why whatj
Foc¢
why Foc
XP
3
X¢ FinP/IP
X
2
camei
X
John ti
1
The present approach also provides a natural account of the so-called ‘additional wh-effect’. By Pesetsky’s assumption, only two wh-phrases need to move to SpecFocP.3 I conclude that any subsequent movement is a matter of scrambling,
137
138
Youngmi Jeong
targeting SpecXP. Given that, consider the example in (3c), repeated here as (14). (14) Dare-ga naze nani-o katta no? who-nom why what-acc bought q *‘What did who buy why?’
The correct word order is obtained by scrambling4 nani-o to SpecXP and the moving verb (successive cyclically) to X0. The next step consists in adjoining dare-ga to naze. Again, the remnant IP, which is empty in this case, may be assumed to occupy SpecTopP. The tree in (15) provides a representation for (14). (15)
ForceP TopP Top¢ FocP
Top why whok
Foc¢
why Foc
XP
4
whatj
X¢
3
FinP/IP
X 2
boughti
X
tk ti tj
1
The same sentence may also be derived by leaving nani-o inside the IP remnant, and moving that remnant to SpecXP, as we did in earlier derivations, as schematized in (16). Summing up this section, we have seen that the three assumptions made in Section 2 conspire to yield a natural characterization of the Anti-Superiority phenomenon in Japanese (and Korean) and the relaxation of it known as the ‘additional wh-effect’ without appealing to the ECP or devices requiring the use of government.
Deriving Anti-Superiority effects 139
4. Conclusion In this paper I have shown how the interaction of three independently motivated ideas (Rizzi’s split CP; Kayne’s ban on rightward adjunction; and Pesetsky’s view on multiple wh-questions) provides a natural account of Anti-Superiority and the additional wh-effect in Japanese (and Korean), two phenomena which had proven recalcitrant to a minimalist treatment, as previous accounts relied on the ECP and the notion of government.
Notes * Part of this paper was presented at the University of North Carolina Linguistics Colloquium (2002) and LSK 2002 (Kyung Hee University, Seoul). I thank the audiences for their comments. Special thanks go to Cedric Boeckx for his indefatigable support and his effort to bring out the best in me. 1. Korean and Japanese diverge when it comes to multiple questions involving ‘how’, as the sentences in (i) shows. (i) a. *Do nani-o anata-wa shita no how what-acc you-top did q ‘How did you do what? b. ??Ettehkey mwues-ul ne-nun ha-ess-ni how what-acc you-top do-past-q ‘How did you do what?
(Japanese)
(Korean)
Whereas Anti-Superiority effects are found in Japanese, they are not (or at least, for many speakers, they are considerably weakened) in Korean. For a possible explanation, see Jeong (2002). 2. The term Anti-Superiority is standardly used to contrast the Japanese/Korean pattern discussed in this paper with the (opposite) English pattern (–why what/*what why) found in (i), traditionally taken to signal Superiority: (i) a. Why did John buy what? b. *What did John buy why? The phenomenon of Anti-Superiority and the additional wh-effect has been extensively discussed in the literature on Japanese. I will not attempt an exhaustive survey here, and I will draw mainly on Watanabe (1992) and Saito (1994). 3. It may be possible to derive Pesetsky’s conclusion by assuming, with Takahashi (1994), that adjunction to a moved element is prohibited (if adjunction took place, chain identity would be violated, as the lower copy of the moved element has nothing adjoined to it). Combined with Kayne’s ban on multiple specifiers/adjuncts, Takahashi’s claim derives the fact that a wh-cluster consists of at most two wh-phrase, with one phrase in SpecFocP and another adjoined to it. 4. A type of movement immune to Superiority, for unclear reasons.
140 Youngmi Jeong
References Bach, E. 1971. “Questions.” Linguistic Inquiry 2: 153–166. Beninca, P. and Poletto, C. 2001. Topic, focus, and V2: Defining the CP-sublayers. Ms., University of Padova. Boeckx, C. 2001. Mechanisms of chain formation. Doctoral dissertation, University of Connecticut, Storrs. Bromberger, S. 1986.What we don’t know when we don’t know why. Ms., Massachusetts Institute of Technology, Cambridge. Jeong, Y. 2002. “Wh-ladder.” In Proceedings of the LSK 2002 International Conference, Linguistic Society of Korea (eds), 215–221. Seoul: Thaehaksa. Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. Law, P. S. 1991. Effects of head movement on theories of subjacency and proper government. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Law, P. S. 1993. On the base position of wh-adjuncts and extraction. Paper presented at the 67th Annual Meeting of the Linguistic Society of America, Los Angeles. [January 1993] Pesetsky, D. 2000. Phrasal Movement and Its Kin. Cambridge, Mass.: MIT Press. Rizzi, L. 1997. “The fine structure of the left periphery.” In Elements of grammar, L. Haegeman (ed), 281–337. Dordrecht: Kluwer. Saito, M. 1994. “Additional wh-effects and the adjunction site theory.” Journal of East-Asian Linguistics 3: 195–240. Starke, M. 2001. Move dissolves into Merge. Doctoral dissertation, University of Geneva. Takahashi, D. 1994. Minimality of movement. Doctoral dissertation, University of Connecticut, Storrs. Uriagereka, J. 1988. On government. Doctoral dissertation, University of Connecticut, Storrs. Watanabe, A. 1992. “Subjacency and S-structure movement of wh-phrases.” Journal of EastAsian Linguistics 1: 255–291.
Conjoined questions in Hungarian* Anikó Lipták Leiden University
In this paper I discuss conjoined multiple questions in Hungarian, which come in two varieties. In one we find conjoined wh-items in a preverbal position. In the other variety, we find one wh-item in preverbal position and another one or other ones in postverbal position, introduced by a coordinator és ‘and’ element. While other, non-conjoined types of multiple questions have been studied in more detail (see for example Horvath 1986, É. Kiss 1993, Lipták 2001, Surányi 2002), conjoined questions did not receive much attention in the literature. This is presumably due to the fact that these questions easily lend themselves to a superficial analysis in terms of clausal conjunction, followed by ellipsis. In this paper I will show that the two types of conjoined questions show different syntactic behaviour in many respects. I will argue that the syntactic differences have to be recognized by assigning distinct structural representations for the two types, i.e. a unified analysis for both (as found in Bánréti 1992) is unsatisfactory. The new analyses for these question types will also provide evidence for a binary-branching analysis of coordination. The structure of this paper is as follows. To set the scene, Section 1 introduces multiple questions in Hungarian. Non-conjoined multiple questions will only be mentioned briefly, while conjoined questions will be exemplified in more detail. It will be shown that there are two syntactic types of conjoined questions in Hungarian, depending on whether the coordinator is found in a preverbal or postverbal position. Section 2 will challenge the previous analysis of conjoined questions, which assigned the same structure to both of these questions. Section 3 spells out the fine structure of postverbal conjoined questions, and Section 4 makes steps towards a new, adequate analysis for preverbal conjoined questions. Section 5 closes with a summary.
142 Anikó Lipták
1.
Multiple questions in Hungarian
1.1 Non-conjoined questions Hungarian has four types of multiple questions which can be distinguished on the basis of syntactic properties and on the basis of the kind of answer they trigger. For the purposes of the present paper, we can distinguish non-conjoined multiple questions and conjoined questions. Before spelling out the properties of conjoined questions, I briefly summarize what needs to be known about non-conjoined multiple questions. Non-conjoined questions come in two varieties. In one, all wh-items front. The other variety can only involve two wh-items, one of which fronts and the other stays in postverbal position. These are demonstrated in the following examples: (1) Ki kit látott? who whom saw-3sg ‘Who saw whom?’
(multiple fronting multiple question)
(2) Ki látott kit? who saw-3sg whom ‘Who saw whom?’
(single fronting multiple question)
Multiple fronting questions, like the one in (1), necessarily require a pair-list answer. The linearly non-last wh-item or items have to refer to D-linked sets of entities, i.e. they have to refer to a set that both speaker and listener know. According to analyses stemming from É. Kiss (1993), the pair list interpretation is due to the fact that the linearly last and non-last wh-items occupy different positions in the structure. More precisely, non-last wh-elements occupy SpecDistP (in the terminology of Beghelli and Stowell 1997) and act as universal quantifiers. The last wh-item, similar to any wh-item in single questions, sits in SpecFocP:1 (1¢) [QP Ki [FocP kit látott […]]]?
(multiple fronting)
According to an alternative analysis to multiple fronting questions (Surányi 2002), the non-last elements occupy outer specifiers of FocP and the multiple pair reading derives from absorption. Single fronting questions ((2) above) correspond to a single pair answer, unless they are formed with melyik ‘which’, in which case multiple pairs can constitute possible answers. These questions can only be used when (a) both wh-items are identical (though case-identity is not a requirement) and (b) both wh-items refer to a D-linked pair of individuals or two D-linked sets. Current
Conjoined questions in Hungarian 143
analyses (Horvath 1986, É. Kiss 1993, Lipták 2001, Surányi 2002) agree that the preverbal wh-item occupies the unique contrastive focus position in the left periphery, while the postverbal wh-item corresponds to a postverbal focused constituent. 1.2 Conjoined questions Conjoined questions come in two types in Hungarian.2 In one type, we find the coordinator element between two wh-items (3), in the other type between two clauses (4): (3) a.
Ki és mikor látta Marit? who-nom and when saw-3sg Mari-acc ‘Who saw Mari and when?’ b. Mikor és hol láttad Marit? when and where saw-2sg Mari-acc ‘When and where did you see Mari?’
(preverbal conjunction)
(4) a.
(postverbal conjunction)
Ki látta Marit és mikor? who-nom saw-3sg Mari-acc and when ‘Who saw Mari and when?’ b. Mikor láttad Marit és hol? when saw-2sg Mari-acc and where ‘When and where did you see Mari?’
From now on, I will refer to the first type (3a,b) as preverbal conjunction and the second one (4a,b) as postverbal conjunction. The most notable and theoretically relevant property of Hungarian conjoined questions is that they impose no restriction on the compatibility of wh-elements. In preverbal conjoined questions arguments can be conjoined with arguments or adjuncts, without any restriction on their categorial nature. Similarly, any adjunct can be combined with any other adjunct. Hungarian shares this property with Slavic languages as well as Rumanian (Comorovski 1998, Kazenin 2002): (5) Kto i zacˇem prixodil? who-nom and what-for came-3sg ‘Who came and for what?’
(Russian)
In languages like English or Dutch, however, these sentences are ungrammatical (Browne 1972). In contrast to preverbal conjunction, in Hungarian postverbal
144 Anikó Lipták
conjoined questions (similarly to Russian), the second conjunct can only be an optional wh-element, i.e. an adjunct: (6) a. *Mikor látta Marit és ki? when saw-3sg Mari-acc and who-nom ‘Who saw Mari and when?’ b. *Zacˇem prixodil i kto? what-for came-3sg and who-nom ‘Who came and for what?’
(Hungarian)
(Russian)
Given that categorial and functional identity is required in standard cases of coordination, these facts constitute problems for a syntactic analysis. We will return to them in the following sections. Before proceeding, note that although conjoined questions are most easily recognized from the presence of a coordinator element, the latter can be phonetically null as well. In this case, however, each wh-item has to be pronounced with equally heavy stress (8a) (note that the same lack of overt coordinators is less natural with postverbal conjoined questions, cf. (8b)): (7) a.
’Kit, ’mikor, ’hol láttál? who-acc when where saw-2sg ‘Whom did you see, where and when?’ b. ?’Kit láttál ’mikor, ’hol? who-acc saw-2sg when where ‘Whom did you see, where and when?’
As far as interpretation is concerned, conjoined questions are interpreted as two separate (conjoined) single questions. (3) and (4), for example, correspond to the logical formula in (8): (8) λp [⁄p ∫$x$t. x sees Mari at t]
That is, questions (3) and (4) above ask for identification of the person who saw Mari and the time when this happened, just like (9). (9) Ki látta Marit és mikor látta Marit? who-nom saw-3sg Mari-acc and when saw-3sg Mari-acc ‘Who saw Mari, and when did he see Mari?’
Answers to conjoined questions involve a single proposition, in which the constituents corresponding to the wh-items are focused:
Conjoined questions in Hungarian
(10) a.
PÉTER látta Marit, TEGNAP. Péter-nom saw-3sg Mari-acc yesterday ‘Péter saw Mari and it happened yesterday.’ b. PÉTER és TEGNAP látta Marit. Péter-nom and yesterday saw-3sg Mari-acc ‘Péter saw Mari and it happened yesterday.’
The variant in (10a) is the ordinary way of answering conjoined questions, while (10b) is clearly marked: it is used in situations where the answer is surprising or where the focused constituents get special emphasis for one reason or another. Note also that the answer to a conjoined question need not be picked from a specific set of individuals (these questions do not presuppose the existence of a specific, D-linked set). The two types of conjoined questions that Hungarian possesses have been analyzed as elliptical clausal coordinations (Bánréti 1992) in the following manner (∆ stands for elliptical material): [&P [FocP kinek ∆] és [FocP hogyan V…]] who-dat and how (preverbal conjunction: forward deletion) b. [&P [FocP kinek V…] és [FocP hogyan ∆]] who-dat and how (postverbal conjunction: backward deletion)
(11) a.
According to (11), conjoined wh-phrases in a sentence always result from the coordination of two clauses (FocPs to be precise), with optional but preferred ellipsis in either the first conjunct (in preverbal conjoined questions) or the second conjunct (in postverbal conjoined questions). This analysis places the two question types on parallel grounds, which is attractive. It suggests that except for the difference in the deletion sites the two questions do not differ in anything else. This conception, however, is wrong. First and foremost, it can be shown that preverbal and postverbal conjoined questions are different constructions with different syntactic structures: it is not the case that all instances of conjoined questions have a multiclausal structure. Rather, we can show on the basis of argument structure and agreement properties that only postverbal conjoined questions do. I will address these points in the next section.
145
146 Anikó Lipták
2. Differences between preverbal and postverbal conjoined questions In this section I will point to some important areas in which the two types of conjoined questions clearly show different properties. All these differences will be understood once we analyze preverbal conjoined questions as sentences involving clause internal coordination of wh-phrases, and postverbal conjoined questions as instances of clausal coordination accompanied by ellipsis. 2.1 Argument structure One clear difference between preverbal and postverbal structures can be found in the way arguments of the verb have to be realized. This can clearly be seen in cases where the wh-items correspond to arguments. As I have pointed out in the last section, the postverbal conjunct in conjoined questions cannot license an argument, while the second conjunct in a preverbal conjunction question can: (12) Ki és kit ölt meg? who-nom and who-acc killed-3sg pv ‘Who killed someone and who was it?’
(preverbal coordination)
(13) *Ki ölt meg és kit? who-nom killed-3sg pv and who-acc ‘Who killed someone and who was it?’
(postverbal coordination)
This is in sharp contrast to examples where one of the wh-items is an adjunct, as (3) and (4) above demonstrate. The generalization we can draw from these examples is that there is an argument/adjunct asymmetry at hand: postverbal conjoined questions can only contain adjunct wh-items in their postverbal position. The clear grammaticality contrast between (12) and (13) remains unaccounted for if we adopt the analysis in (11), where both types of questions are derived from the same underlying structure, and result in the same LF structures as well3 (basing ourselves on the commonly held assumption that elliptical sites are recovered at LF and thereby contribute to semantic interpretation). Bánréti’s analysis implies that the two types of conjoined questions are identical at LF: (14) a. [&P [FocP ki ölt meg] & [FocP kit ölt meg]] b. [&P [FocP ki ölt meg] & [FocP kit ölt meg]]
(= (12)) (= (13))
In both clauses in both types of questions, the verbs have a fully saturated argument structure as the Projection Principle requires, which must hold at all levels
Conjoined questions in Hungarian 147
of representation, including LF. If the pairs have the same LF structures, it is impossible to relate the ungrammaticality of postverbal conjoined questions to the fact that we are dealing with argumental wh-phrases, which cannot be missing from clauses according to the Projection Principle. Both (12) and (13) have the following representation after recovery of elided material: (12¢)/(13¢) [Ki ölt meg] és [kit ölt meg]? [who-nom killed-3sg pv and [who-acc killed-3sg pv
As it stands, (12/13) violates the Projection Principle, since the first clause only contains a subject, but no overt object. We cannot say that the object term is there, but it is not overt: no covert element can take the position of the object term. The covert object clearly cannot be a trace for lack of a binder. (15) a. *[Ki ölt meg ti] és [kiti ölt meg prosub]?
(t: unbound)
It cannot be a pro either because pro is only available for definite objects in Hungarian, i.e. known objects, and to have a known object in the first clause which is questioned in the second, is infelicitous:4 (15) b. *[Ki ölt meg proi] és [kiti ölt meg prosub]?
(proi: infelicitous)
A cataphoric construal of the missing object term is also hard to argue for, because this kind of cataphoric dependency does not exist elsewhere in the grammar. An overt cataphoric prononominal in the same position (16) is ungrammatical: (15) c. *[Ki ölte meg proi] és [kiti ölt meg prosub]? (16) *Ki ölte meg o˝t és kit ölt meg? who killed-3sg pv him/her and who killed-3sg pv ‘Who killed him/her and who was he/she, whom he killed?’
So a structural representation along the lines of (11) predicts that the sentences in (12/13) should be ungrammatical, because their first clause violates the Projection Principle. This predicts that no conjoined questions should exist with two arguments in them. However, only postverbal conjoined questions are ungrammatical under these circumstances. This proves that the representation in (12/13) is correct for postverbal conjoined questions, but is wrong for preverbal ones. Postverbal conjoined questions are indeed conjoined single questions with ellipsis in the second conjunct. Preverbal conjoined questions, however, are not conjoined clauses, but contain only one clause, in which all wh-phrases belong to one and
148 Anikó Lipták
the same verb. This way there is no violation of the Projection Principle, since all arguments are present in the clause. (17) [FocP [&P Ki és kit] ölt meg]?
The importance of facts like (12) and (13) has not been recognized so far in the literature on Hungarian, presumably because of prevailing studies on adjunct questions or verbs which can be interpreted both transitively and intransitively, like read. But even in the latter case, (18) has a different meaning from (19), as I indicate in the English translations: (18) Ki és mit olvasott? who-nom and what-acc read-3sg ‘Who has read something and what was it?’ (19) Ki olvasott és mit? who-nom read-3sg and what-acc ‘Who was engaged in reading and what was he reading?’
2.2 Agreement properties The other difference between preverbal and postverbal conjoined questions can be observed in the agreement properties of the base verb. The difference in agreement also suggests that in preverbal conjoined questions all wh-arguments belong to the same verb. Consider the following full (non-elided) clausal coordination: (20) Érdekel (hogy) mit csinálsz és hogyan csinál-od/*-sz. interest-3sg (that what-acc do-2sg.indef and how do-2sg.def/indef ‘I care about what you do and how.’
Definiteness agreement in the second conjunct is obligatory; given that the object is pro, definite agreement results. If (20) is turned into a preverbal conjoined structure by deleting the verb in the first conjunct, the result is expected to show the same agreement in the second conjunct, since that is not affected in any way by the ellipsis in the first. But this expectation is false, the grammatical sentence has the unexpected indefinite conjugation in the second conjunct: (21) Érdekel (hogy) mit és hogyan csinál-sz/*-od. interest-3sg (that what-acc and how do-2sg.indef/def ‘I care about what you do and how.’
Conjoined questions in Hungarian 149
This can only be the case if the verb in (21) agrees with mit ‘what’, which is an indefinite pronoun in Hungarian. This indicates that (21) is not derived by means of ellipsis, but rather, both wh-items are constituents of the same (and only) embedded clause, i.e. the verb has mit as its complement. 2.3 The structural representation of conjoined questions In the last two sections I have listed some arguments to the effect that preverbal conjoined questions are real multiple questions in the sense that we find more than one wh-phrase in a single clause in them. All conjoined wh-items are in one and the same clause. Postverbal conjoined questions on the other hand are multiple clausal structures, the result of clausal coordination followed by ellipsis. The right structural representations are given in (22): (22) a. Preverbal coordination: [FocP [wh1 & wh2] V] b. Postverbal coordination: [FocP wh1 V] & [FocP wh2 ∆]
Thus, in fact postverbal conjoined questions do not fall under multiple questions, since they involve multiple clauses with one single question in each. This means that these questions have the syntax of single questions, while preverbal conjoined questions clearly contain multiple instances of wh-items in the same clause. Interestingly, exactly the same conclusion is drawn about Russian conjoined questions by Kazenin (2002). It is important to note that this syntactic difference between preverbal and postverbal coordinations does not carry over to the interpretive level (which is why the syntactic difference is not so easily perceived). Both types of conjoined questions are the same semantically: they are used in the same situations, they elicit the same answers, i.e. they are functionally equivalent. Functionally speaking, they exist because other multiple questions cannot be used under circumstances where the speaker expects a single-pair answer: Hungarian multiple fronting questions can never be used when a single-pair answer is expected, and although single fronting multiple questions can, these are restricted to categorially identical wh-items (see Section 1 above). Thus, conjoined questions are the only means to ask a multiple question where the wh-items are not of the same category and where the speaker expects a singlepair answer. Note that when the same content can be expressed with a single fronting multiple question (because the wh-items are categorially identical), conjoined questions are strange (Kálmán et al. 2001):
150 Anikó Lipták
(23) a. ??Ki és kibe botlott bele? (conjoined question) who-nom and who-into bumped-3sg into ‘Who bumped into whom?’ b. Ki botlott bele kibe? (non-conjoined question)
The marginality of (23) can thus be given a functional explanation: the use of a coordinator element is marginal if the conjunction-less counterpart of the same sentence would be grammatical. This recalls the ‘Avoid Conjunction Principle’ (“if no conjunction marker is necessary, do not use any”) by Progovac (1999).
3. The fine structure of postverbal conjoined questions The previous section has shown that postverbal conjoined questions in Hungarian are best analyzed in terms of coordination of clausal material. To spell this out in some more detail, we have to find answer to several important questions. First of all, what kind of clausal category are we dealing with in these constructions? Second, how can we represent conjunction structurally? Third, what kind of ellipsis affects the second clause in these questions? I will answer these questions in turn, starting with the last one. It can be easily seen that the ellipsis found in postverbal conjoined questions is not a run-of-the-mill VP ellipsis. First of all, the missing sequence in these questions corresponds to a larger structure than VP: it actually corresponds to Foc¢. Recall that in sentences with focused constituents the verb raises up to Foc0. Under the common view that holds that ellipsis requires parallelism, this must mean that the structure affected by ellipsis in the second conjunct is a Foc¢ category as well. Also, postverbal conjoined questions clearly differ from VP-ellipsis in that in the latter but not in the former the two clauses can refer to two distinct events: (24) John met someone, but Peter did not.
This interpretation, however, is excluded in the case of conjoined questions: they always refer to the same event: (25) Ki látta Marit és mikor? who-nom saw-3sg Mari-acc and when ‘Who saw Mari and when was this?’ *‘Who saw Mari once and when did he see her another time?’
Conjoined questions in Hungarian
In this, a postverbal conjoined question patterns together with conjunction reduction, which is also impossible to interpret as referring to two distinct events: (26) Péter elment, mégpedig pénz nélkül. Péter left and money without ‘Péter left, and/moreover without money.’
(conjunction reduction)
This also shows (among other things) that postverbal conjoined questions are actually a case of conjunction reduction.5 Given the size of the elided material in these constructions it is not at all surprising that they denote one single event. It is clear that what gets affected by deletion in the second conjunct actually involves the event-argument as well (on a Davidsonian view, and assuming that the event argument is syntactically represented). The parallelism requirement explains why the event in the second conjunct necessarily has to be anaphoric. Let us proceed to finalize the structural representation of the conjunction and the conjoined items. At first blush it seems that the representation we have arrived at by arguing for a deletion process affecting Foc¢ in the second clause cannot be on the right track, for want of a licenser for the subject trace: (27) [FocP Kii [Foc¢ láttaj [VP ti tj Marit]]] who-nom saw-3sg Mari-acc és [FocP mikork [Foc¢ láttaj [VP ti tj Marit]] tk]? and when
As it stands, in this representation of (27) the second clause contains an unbound subject trace. This problem, however, vanishes once we take the view that specifiers c-command out of their maximal category. Evidence for this comes from the following cases of pronominal binding: (28) a.
Ki beszélt és melyik pro barátjáról? who-nom talked-3sg and which friend-poss.3sg-about ‘Who talked and about which friend of his?’ b. *Melyik pro barátja beszélt és kiro˝l? which friend-poss.3sg-nom talked-3sg and who-about ‘Who talked and about which friend of his?’
This shows on the one hand that the first wh-item can c-command out of its clause, but it also shows that there is an asymmetric relationship between the two conjuncts: the wh-item in the second conjunct does not c-command the first. This provides primary evidence for an X¢ theoretic approach to coordination
151
152
Anikó Lipták
(Johannessen 1998, Munn 1993), which predicts an asymmetric relation between conjuncts: the first conjunct c-commands the second, but not vice versa. These facts help to answer the questions concerning the representation of postverbal conjoined questions. The conjoined clauses in Hungarian postverbal conjoined questions are: i.
FocP constituents (otherwise, if they were larger, the wh-item in SpecFocP could not c-command out of them) ii. in an asymmetric relation (which provides evidence for the X¢ theoretic approach to coordination)
Adopting the view in which coordination is adjunction of a coordination phrase to a category (Munn 1993), we arrive at the following representation of postverbal conjoined questions: (29) [FocP [FocP Kii [Foc¢ láttaj [VP ti tj Marit]]] [&P és [FocP mikork [Foc¢ láttaj [VP ti tj Marit tk]]]]
The focused ki ‘who’ c-commands out of its FocP and thus it licenses its trace both in the first conjunct as well the second conjunct. Ellipsis affects the second conjunct under identity with the elements in the first one. The fact that the antecedent Foc¢ and the elided Foc¢ do not coincide completely is not unique to conjoined questions and conjunction reduction. As has been noticed by Chung, Ladusaw and McCloskey (1995), other instances of ellipsis, like IP-ellipsis also have this property: (30) John likes Mary. I don’t know why [John likes Mary t].
Chung, Ladusaw and McCloskey suggest a “sprouting” mechanism that adds the missing empty category to the recycled IP at LF. Summarizing, I have argued that postverbal conjoined questions should be assigned the representation in (29). They involve the conjunction of two clausal categories, Foc¢ to be precise, with deletion of phonetic material in the second conjunct. The arguments presented in this section provided evidence for an X¢ theoretic approach to coordination as well.
4. Towards the correct analysis of preverbal conjoined questions In Section 2 we established that preverbal conjoined questions involve coordination of wh-phrases. This claim faces a very serious problem, which Bánréti
Conjoined questions in Hungarian
(1992) did not encounter: how can categorially distinct items be conjoined? In most usual cases of coordination the category of the coordinates must be the same together with their theta-roles if they have any. This is the so-called “Law of Coordination of Likes” (Williams 1981): (31) *I helped Peter and quickly.
[NP & AP]
Before trying to tackle the problem preverbal conjoined questions present for this law, I will start out by showing that wh-items are not the only constituents that defy the Law of Coordination of Likes. Quantificational items also do. Some speculations as to how this is possible and what syntactic consequences this has, will be discussed in Section 4.2. 4.1 Clause internal coordination in preverbal conjoined questions: Parallels with quantifiers The Law of Coordination of Likes tolerates a great many exceptions even in English. One frequent example of unlike category coordination can be found when the coordinates have predicative function, well-described and analyzed in Sag et al. (1985): (32) Pat is either stupid or a liar.
[AP & NP]
The other type, more closely related to the Hungarian coordination facts, can be found in many languages including English: (33) John met Mary and in her house!
[NP & PP]
This example of unlike category coordination is licensed only if both coordinates are emphasized, and the sentence describes an unexpected state of affairs (indicated here by the exclamation mark). In a vein similar to the Hungarian examples, English can coordinate wh-items due to their “equal status […] as wh words” (Bolinger 1978: 139): (34) When and how did it happen?
It seems to be true across languages that emphatic operators are by and large conjoinable regardless of their different categorial status, although languages do differ as to what kind of quantifiers and environments are allowed. Hungarian is a language where all kinds of quantifiers can be conjoined within a clause, provided they are stressed (marked by ’):
153
154
Anikó Lipták
(35) Ide ’mindenki és ’mindig bejöhet. here everyone-nom and always pv-come-pot-3sg ‘Everyone can enter here and this holds for all times.’ (36) Ide ’bárki és ’bárkit meghívhat. here anyone-nom and anyone-acc pv-invite-pot-3sg ‘Anyone can invite anyone here.’ (37) Ide ’senki és ’semmikor nem jöhet be. here no-one-nom and never not come-pot-3sg pv ‘No-one can enter here and this holds for all times.’
Similar examples, involving negative polarity items, can be found in Russian and other Slavic languages (Kazenin 2002). The conclusion we can draw is that emphatic stress makes categorial differences invisible: the [&P [XP] & [YP]] pattern is available for clause-internal coordination as long as both XP and YP carry emphatic stress. One way of interpreting this pattern would be to say that for certain mechanisms emphatic quantifiers are recognized as items with a common ‘Q’ category/function,6 and therefore can be conjoined with any item with a similar ‘Q’ category/function. This is reminiscent in a way of what was observed in (32/33): there it was shown that unlike categories with the same predicative function can also coordinate. This shows that for coordination, it is not only the syntactic category of the coordinates that counts. 4.2 A possible derivation of preverbal conjoined questions The previous section has shown that clause-internal coordination of unlike categories is not restricted to wh-elements: the scope of elements that can occur in this construction is broader: it involves quantificational/emphatic elements in general, at least in Hungarian. To account for this pattern, however, is really difficult. There are many questions that one would need to answer before the final picture about these constructions can be pieced together. What is the fine structure of the coordination phrase? How is this formed? How are the traces of the coordinated items licensed? Unfortunately, I cannot provide definitive answers to all these questions, due to serious lack of understanding concerning coordination phrases in general. Pending further research in this area, here I only sketch the directions a future satisfactory analysis should take. As we will see, preverbal conjunctions have many contradictory properties.
Conjoined questions in Hungarian
Let us start with the representation of the coordination phrase. If the argumentation in the previous section is on the right track, we have evidence for an X¢ approach to coordination in Hungarian conjoined questions in particular, and hopefully to coordination in general. Adopting the structure I used in the previous section (Munn 1993), this means that preverbal questions have the following structure: (38) [FocP [whP kineki [&P és hogyanj] segítettél ti tj] who-dat and how helped-2sg
That is, the second wh-item is an adjoined coordination phrase (&P) to the first wh-item, and the whole complex (whose external distribution is that of a wh-item) occupies the SpecFocP position, as any single wh-item does in Hungarian. This analysis predicts that we find asymmetries between the conjoined wh-items. These asymmetries are indeed present, supporting the representation in (38): there is an asymmetric c-command relation between the two wh-items. The first, but not the second, c-commands the other: (39) a.
Ki és melyik pro barátjáról beszélt? who-nom and which friend-poss.3sg-about talked-3sg ‘Who talked and about which friend of his?’ b. *Melyik pro barátja és kiro˝l beszélt? which friend-poss.3sg-nom and who-about talked-3sg ‘Who talked and about which friend of his?’
This is exactly what we get with other quantifiers as well, further strengthening the parallelism found in Section 4.1 between the two: (40) a.
Senki és semelyik pro barátjáról nem beszélt. noone-nom and none friend-poss.3sg-about not talked-3sg ‘Noone talked about any friend of his.’ b. *Semelyik pro barátja és senkiro˝l nem beszélt. none friend-poss.3sg-nom and noone-about not talked-3sg ‘Noone talked about any friend of his.’
This basic asymmetry clearly refutes other, non-X¢ theoretic approaches to the coordination structures under study (such as the 3-dimensional representations in which coordinates are equal in structural terms, as in Goodall 1987, van Riemsdijk 1998). These approaches would not predict any asymmetries between the wh-items, contrary to fact.
155
156 Anikó Lipták
Given this, we can address the question of how the traces of the individual wh-items are licensed. It is easy to argue that the first wh-item c-commands out of the coordination complex, and as such it c-commands its trace lower in the structure. It is not clear, however, whether the second wh-item can do the same, although claims to this effect can be found in the literature. Zoerner (1995) argues that the following disjoint reference effect is to be explained as a Principle B violation, due to the fact that the second conjunct c-commands the pronominal: (41) *[Robin and Paulj] like himj.
It seems to me, however, that (41) is inconclusive to arrive at this conclusion, because it wrongly predicts that an anaphor in the place of the pronominal is grammatical: (42) *[Robin and Paulj] like himselfj.
So examples of A-binding seems to be unavailable to illustrate this point. Other examples, involving bound-pronominal interpretation, however, can illustrate this better: (43) [Standing on the stairs and with every boy present] the headmaster gave them their books.
A bound pronominal interpretation seems to be available in (43), which indicates that the universal quantifier has scope over the whole clause. Note that in (43) we are also dealing with unlike category coordination, just like in the case of conjoined questions. At this point it is difficult to make any claims as to how (43) is possible, whether it is due to feature percolation from the second conjunct to the whole coordination phrase or to some special mechanism. The behaviour of the two conjuncts is to some extent contradictory: on the one hand there is clear asymmetry between them (cf. 40), on the other, they are equal in scope taking. I leave the resolution of this contradiction for further research. Another problem we run into when analyzing preverbal conjoined questions is the formation of the coordinate structure itself. A representational approach to the coordination phrase would be more convenient, however, there is one aspect of these constructions, which would rather prefer a derivational account: superiority, another asymmetry between the wh-items. As Kálmán et al. (2001) also notes, there are ordering requirements between the conjoined wh-items in questions:7
Conjoined questions in Hungarian
(44) a.
Ki és kiro˝l beszélt? who-nom and who-about talked-3sg ‘Who talked and about whom?’ b. ???Kiro˝l és ki beszélt?
While superiority is very clearly present in Hungarian, lack of superiority effects has been reported about Russian (Kazenin 2002). And although I cannot contest Kazenin’s judgements about Russian, other Slavic languages pattern with Hungarian in that they do exhibit superiority in preverbal conjoined questions (Olga Tomicˇ, p.c.): (45) a.
Koj i kade odi? where and who-nom go-3sg ‘Who goes and where?’ b. ??Kade i koj odi?
(Macedonian)
These facts reinforce the view that the first conjunct is more distinguished than the second, and they prefer a derivational account. Superiority facts are currently handled in the derivational frame, and accounted for by “attract the highest”. It could be argued therefore that what gets attracted by the Foc0 head first is the first conjunct, and that the second conjunct, in the form of a &P adjunct, gets adjoined to it later, post-cyclically (recalling the spirit, but not the content of proposals like Fox 2002, Stepanov 2001 concerning the theoretical possibility of late adjunction). If this account is viable, it provides evidence for the X¢ theoretic and derivational approach to coordination.
5. Summary This paper made an attempt to achieve a better understanding of conjunction possibilities of Hungarian question words and quantifiers. It showed that there are two distinct patterns in which these items can take part, and that these two patterns cannot be lumped together under a uniform syntactic analysis. Although the definitive analysis of the preverbal conjunction pattern is still not in sight, we can be reasonably sure that it involves a single coordination phrase within one and the same clause. Postverbal conjunctions, however, necessarily have to receive a multiclausal analysis in which ellipsis is operative. As an auxiliary result of this investigation, the facts provided evidence for an X¢ theoretic approach to coordination.
157
158
Anikó Lipták
Notes * The production of this article was supported by a grant from the Netherlands Organization for Scientific Research (NWO). The gist of the ideas presented in Section 2 originate from Lipták (2000) and (2001). For discussions on the present topic I am indebted to Jeroen van Craenenbroeck, Marcel den Dikken, Johan Rooryck and Luis Vicente. All errors and shortcomings are mine. 1. Note that the position of the finite verb is Foc0 in sentences containing a focused element. 2. In this paper I do not consider questions with a disjunction between two wh-elements: (i) Kit vagy mit láttál? whom or what-acc saw-2sg ‘Who or what did you see?’ In these questions the conjoined wh-items always have to be categorially identical, which is not the case with questions where wh-items are coordinated by an ‘and’ coordinator. 3. Including the possibility that either type is derived from another. One could imagine that the postverbal pattern is derived via the extraposition of és kit — this, however, can be safely excluded not only because in this case, (13) would be expected to be grammatical, just like (12), but also because Hungarian does not allow for conjunct-extraposition in any context, evidenced by the following: (i) a.
Mari és Péter a birák. Mari and Péter the judge-pl ‘Mari and Péter are the judges.’ b. *Mari a birák és Péter. Mari the judge-pl and Péter
4. If (14b) were grammatical, it would occur with definite conjunction on the first verb due to the fact that object pro is a definite item. However, even definite conjugation on the verb does not make the sentence better: (i) *Ki ölte meg proobj és kit ölt meg prosub? who-nom killed-3sg.def pv and who-acc killed-3sg.indef pv ‘Who killed someone and who did he kill?’ 5. Kazenin (2002) treats postverbal conjoined questions as an instance of sluicing — a move I would not subscribe to. In some (albeit special) cases of sluicing anaphoricity of the events is not required: (i) I know John went to the party, but I do not know who else. 6. One might take recourse to an account where emphatic constituents instantiate a distinguished, e.g. EmphP category. This way, the categorial identity problem would be resolved. 7. Although the strength of the ordering restriction is dependent on stress assignment: heavily stressed wh-items can appear in any order.
Conjoined questions in Hungarian 159
References Bánréti, Z. 1992. “A mellérendelés.” In Strukturális magyar nyelvtan I. Mondattan, F. Kiefer (ed), 715–796. Budapest: Akadémiai Kiadó. Beghelli, F. and Stowell, T. 1997. “Distributivity and Negation.” In Ways of Scope Taking, A. Szabolcsi (ed), 71–107. Dordrecht, Reidel. Bolinger, D. 1978. “Asking more than one thing at a time.” In Questions, H. Hiz˙ (ed), 107–150. Dordrecht, Reidel. Browne, W. 1972. “Conjoined question words and the limitation on English surface structure.” Linguistic Inquiry 3: 223–226. Chung, S., Ladusaw, W. A. and McCloskey, J. 1995. “Sluicing and Logical Form.” Natural Language Semantics 3: 239–282. Comorovski, I. 1989. Discourse and the syntax of multiple constituent questions. Doctoral Dissertation, Cornell University, Ithaca. Fiengo, R. and May, R. 1994. Indices and Identity. Cambridge, Mass.: MIT Press. Fox, D. 2002. “Antecedent-Contained Deletion and the copy theory of movement.” Linguistic Inquiry 33: 63–96. Goodall, G. 1987. Parallel Structures in Syntax. Coordination, Causatives and Restructuring. Cambridge: Cambridge University Press. Horvath, J. 1986. Focus in the Theory of Grammar and the Structure of Hungarian. Dordrecht: Foris. Johannessen, J. 1998. Coordination. Oxford: Oxford University Press. Kálmán, L. (ed). 2001. Magyar leíró nyelvtan. Mondattan 1. Budapest: Tinta Könyvkiadó. Kazenin, I. K. 2002. On coordination of wh-phrases in Russian. Ms., University of Tübingen/ University of Moscow. É. Kiss, K. 1993. “Wh-movement and specificity.” Natural Language and Linguistic Theory 11: 83–120. Lipták, A. 2000. “Multiple relatives as relatives of questions.” In Approaches to Hungarian 7, G. Alberti and I. Kenesei (eds), 151–177. Szeged: JATE Press. Lipták, A. 2001. On the syntax of wh-items in Hungarian. Doctoral Dissertation, Leiden University. Munn, A. 1993. Topics in the syntax and semantics of coordinate phrases. Doctoral Dissertation, University of Maryland, College Park. Progovac, L. 1999. “Conjunction doubling and ‘Avoid Coordination Principle’.” In Topics in South Slavic Syntax and Semantics, M. Dimitrova-Vulchanova and L. Hellan (eds), 25–39. Amsterdam: John Benjamins. van Riemsdijk, H. 1998. “Trees and Scions — Science and trees.” In A Celebration: Essays for Noam Chomsky’s 70th Birthday, J. Fodor, S. J. Keyser and A. Brand (eds). Cambridge, Mass.: MIT Press. [Online publication — available for download at http:// mitpress.mit.edu/chomskydisc/riemsdyk.html.] Sag, I., Gazdar, G., Wasow, T. and Weisler, S. 1985. “Coordination and how to distinguish categories.” Natural Language and Linguistic Theory 3: 117–171. Stepanov, A. 2001. “Late adjunction and minimalist phrase structure.” Syntax 4: 94–125.
160 Anikó Lipták
Surányi, B. 2002. Multiple wh-fronting and wh-in-situ in Hungarian and choice functions. Papers presented at the Sixth International Conference on the Structure of Hungarian, Düsseldorf. [September 2002] Wilder, C. 1997. “Some properties of ellipsis in coordination.” In Studies on Universal Grammar and Typological Variation, A. Alexiadou and T. A. Hall (eds), 59–107. Amsterdam: John Benjamins. Williams, E. 1981. “Transformationless Grammar.” Linguistic Inquiry 17: 645–653. Zoerner, E. 1995. Coordination: The syntax of &P. Doctoral Dissertation, Cornell University, Ithaca.
Persian wh-riddles* Ahmad R. Lotfi Azad University at Esfahan It moveth and it moveth not at all, It is the farthest of the far, it is The nearest of the near, it is within, And yet it is without all that we know. Isha-Upanishad
In this paper I examine some data on Persian multiple wh-questions/fronting that seem to cause potential problems for some recent minimalist accounts of such phenomena (Boškovic´ 1998a, 1998b, 2000, Hagstrom 1998, and Pesetsky 2000 among others). The main claim here is that these inconsistencies are mere “apparent exceptions” to the standard minimalist accounts of multiple questions. The paper begins with a review of some major recent minimalist accounts of movement and multiple wh-questions. After a functional description of Persian (wh-)questions, three Persian ‘wh-riddles’ are introduced: (i) the language patterns with English rather than other wh-in situ languages in its single-pair (SP)/pair-list (PL) interpretation of multiple wh-questions, (ii) the language affords optional Q-markers which behave differently from those in other wh-in situ languages like Japanese and Sinhala, and (iii) optional overt wh-fronting in Persian shows ad hoc Superiority effects/violations. Some tentative answers to these riddles appear next. Here I try to show the importance of case requirements, scrambling, and also genuine Superiority effects in the ordering of Persian wh-phrases. Once the dust has settled, Persian multiple wh-questions do not seem to be much different from those in such betterstudied languages as English and Japanese.
1.
Where are we?
In Chomsky’s (2000, 2001) “probe-goal” model, “dislocation” is a property of human language that might be required by design specifications. Despite that,
162 Ahmad R. Lotfi
movement is afforded via some formal mechanism in which uninterpretability plays a central role: uninterpretable features of a probe seek the matching features of a local goal (the operation Agree) in order to be erased. Agree will be induced iff: (1) a. Matching is identity, b. Goal is in the domain of Probe D(P) which is the sister of P, and c. Locality reduces to the closest c-command. (Chomsky 2000: 122)
In this system, the wh-phrase (the goal) contains an interpretable feature [Q] that agrees with the uninterpretable [Q] feature of C (the probe). The goal also contains an uninterpretable [Wh] feature that is erased together with the [Q] feature of the probe once the goal has moved to SpecCP: (2)
CP C¢ ...
C[Q] Wh-[Q]
Languages differ from each other in that in some the movement of the wh-word is overt due to the presence of the EPP-feature (after the Extended Projection Principle) while in others it takes place covertly. Whatever the case, and contrary to his original position (Chomsky 1995) concerning covert movement, Chomsky (2000) assumes (in absence of evidence to the contrary, as he puts it) that features themselves cannot be attracted: it will be either the case that the wh-phrase moves prior to Spell-Out (as it does in English), or it moves at LF (like Japanese). In both cases, movement is phrasal. Languages with overt phrasal movement still differ systematically with respect to the number of wh-phrases that move overtly. In a Bulgarian-type language, the D-head of every wh-element contains an EPP-feature identical to the uninterpretable P- (peripheral) feature [Q]. As such, interrogative C and also wh-elements function as targets of wh-movement. It follows that in such multiple wh-fronting languages, wh-elements must be organized into a single wh-cluster (to undergo overt movement to SpecCP) while in English-type languages only one of the wh-phrases (usually the highest one) moves, and others remain in situ.1
Persian wh-riddles 163
Pesetsky (2000) re-analyzes Chomsky’s (1995) proposal that covert movement is not phrasal at all but a repair strategy intended to delete the uninterpretable feature F on the head K after moving another instance (usually interpretive) of F to K. According to Pesetsky, “feature movement” is not an alternative analysis of covert phrasal movement but a different syntactic operation. Then while the covert movement of the number feature of the associate DP to check that of T in a there-construction like There is a book on the table is a case of feature movement, that of a wh-in situ is phrasal. How should one account for the covert/overt distinction in wh-movement? In Pesetsky (1997, 2000), the distinction is not a matter of syntactic timing (as it is in a Y-model in which overtness/covertness of movement is a function of whether it takes place prior to or after Spell-Out) but that of phonological rules. “The trade-off comes in the phonological component, where the simple principle of Phonological Spellout in (13) [= “Pronounce only the highest position in a movement chain.”] is replaced by phonological principles of chain pronunciation that regulate the pronunciation of moved elements in a more complex manner” (Pesetsky 2000: 7). In traditionally termed “overt movement”, the moved element is pronounced in the final position while in “covert movement”, the moved element is pronounced in the target position. The differences are reduced to pronunciation rather than syntax. Pesetsky (2000: 8) proposes two different pronunciation rules within his Single-Output model to distinguish English- and Bulgarian-type languages. They are reproduced here as (3) and (4) respectively: (3) Pronunciation Rule (English) a. The first instance of wh-phrase movement to C is overt in that wh is pronounced in its new position and unpronounced in its trace positions. b. Secondary instances of wh-phrase movement to C are covert in that wh is pronounced in its trace position, and is unpronounced in its new position. (4) Pronunciation Rule (Bulgarian) All wh-phrase movement to C is overt, in that wh is pronounced in its new position, and unpronounced in its trace positions.
Pesetsky (2000) also proposes a cross-linguistic typology of wh-movement based on the specifier potential of C. In this typology, a language with Cm-spec is one whose C requires more than one wh-specifier attracted to SpecCP. In a language with C1-spec, on the other hand, only one wh-specifier is attracted.
164 Ahmad R. Lotfi
Finally, in a language with C0-spec, no wh-specifier of any sort is tolerated. Pesetsky expects Japanese and Korean C0-spec, like C1-spec and Cm-spec, to carry an uninterpretable wh-feature to be deleted in the course of the derivation. As no wh-specifier is tolerated in the CP of such languages, only wh-feature movement can erase the uninterpretable feature of C. Consequently, if a scope-bearer intervenes between all wh-phrases and C0-spec in such languages, an intervention effect will be inevitable. The violation will be eliminated then via scrambling: (5) Intervention effect with wh and negation/only (Japanese) a. *Hanako-sika nani-o yoma-nai no? Hanako-only what-acc read-neg q ‘What did only Hanako read?’ Scrambling eliminates the violation b. Nani-o Hanako-sika yoma-nai no? (Pesetsky 2000: 85)
As Boškovic´ (2000) points out, there are four distinct language types with respect to how they implement wh-movement in their multiple questions: (i) Slavic languages in which all wh-phrases are fronted, (ii) languages like English in which one and only one wh-phrase undergoes fronting, (iii) languages such as Japanese in which all wh-phrases remain in situ, and (iv) ‘mixed’ languages like French with properties of both English- and Japanese-type languages. Apparently, wh-in situ (in multiple questions) rightly merits the serious consideration it has been shown because the phenomenon recurs in three out of four language types (i–iv). Among the minimalist studies of the syntax and semantics of wh-in situ, Hagstrom’s (1998) seems to be of particular importance. Based on the data from the wh-in situ languages like Japanese, Premodern Japanese, Sinhala, and Okinawan, Hagstrom proposes that in wh-questions there is a morpheme Q base-generated as the sister of a wh-word. The morpheme moves to the clause periphery (via feature attraction) where it contributes an existential quantification to the semantics of the question. Hagstrom also tries to determine the launching site of Q in the multiple questions of such languages. His data suggest that the single-pair/pair-list reading of a multiple wh-question in such languages is a function of the launching site of Q. Hagstrom refers to this as “the Pairlist Anti-Superiority” according to which “in order to get a pair-list reading for a multiple-wh-question, one of the wh-words (has) to be outside the scope of the launching site of Q.” Accordingly, a pair-list question is actually a set of sets of questions realized as the “universal force” of the higher wh-word when Q moves
Persian wh-riddles 165
from the lower wh-word. The single-pair reading, on the other hand, arises “when Q is launched from outside both wh-words…” (Hagstrom 1998: 193), a position higher than both wh-phrases: (6) a.
Q ... tQ [... Wh-word1 ... Wh-word2 ...]
single-pair reading
b.
Q ... [... Wh-word1 ... tQ Wh-word2 ...]
pair-list reading (exhausts wh-word1)
(from Hagstrom 1998: 193)
Hagstrom’s analysis is in harmony with the common observation that singlepair questions are generally impossible in languages with overt movement while a multiple question in a wh-in situ language may have either a single-pair or a pair-list interpretation (Boškovic´ 1998). French also permits both interpretations depending on the use of the in situ or the wh-movement strategies: single-pair questions are possible in French only with the in situ strategy. “It is possible that the obligatoriness of syntactic movement of a wh-phrase to SpecCP for some reason forces the pair-list interpretation” (Boškovic´ 1998: 1). Then what happens in Japanese is due to the semantically motivated movement of Q while the overt wh-movement in English is motivated by a strictly formal syntactic requirement. Boškovic´ also observes some variation in multiple wh-fronting languages. Bulgarian (in which the overt movement of a wh-phrase to SpecCPs is obligatory) patterns with English, and Serbo-Croatian (no wh-phrase overt movement to interrogative SpecCPs) with Japanese. Grohmann (1999, 2000) extends Boškovic´’s (1998) adaptation of Hagstrom’s (1998) semantics to German. Both wh-elements in a German multiple question move overtly with one of WHs targeting SpecFocP and the other the lower projection FP. Then German patterns with Bulgarian rather than English with regard to multiple questions.
2. Mysterious Persian! Persian is a pro-drop, SOV (though widely scrambling), wh-in situ language whose yes/no-questions are formed with the addition of a Q-marker (aya in formal, and yani in informal Persian). However, a question can also occur without any Q-marker. In such cases, only the rising intonation signals Q:
166 Ahmad R. Lotfi
(7) Persian yes/no-questions a. Armin aks-e ye dinosaur-o be baba-sh dad. Armin picture-of one dinosaur-case to father-his gave-3s ‘Armin gave the picture of a dinosaur to his father.’ b. Aya/yani/Ø Armin aks-e ye dinosaur-o be baba-sh dad?
Persian wh-phrases in both single and multiple wh-questions usually remain in situ. Moreover, in Modern Persian, the Q-marker aya only rarely appears at the beginning of a wh-question (except in some formal texts or certain local varieties of the language like those spoken in the southern areas of Esfahan province). The Q-marker yani is quite acceptable in (Spoken) Standard Persian wh-questions. Despite that, even then yani does not occur in such questions very frequently. (8) Persian wh-questions a. Aya Sohrab be pedar che goft? q Sohrab to father what said ‘What did Sohrab tell his father?’ b. Aya pro che shenidid?2 q what heard-2p ‘What did you hear?’ c. Aya pro che xabari dari? q what news have-2s ‘Which news do you have?’ d. Yani/Ø Armin chi-o be baba-sh dad? q Armin what-case to father-his gave ‘What did Armin give to his father?’ e. Yani/Ø Armin chi-o be ki dad? q Armin what-case to who gave ‘What did Armin give to whom?’
(formal)
Although wh-fronting is readily available in Persian, it is not an obligatory syntactic operation. It primarily takes place for the sake of focusing. (9) Persian wh-fronting a. Yani/Ø una chi-o chera xaridan? (in situ) q they what-case why bought-3p ‘Why did they buy what?’ b. Yani/Ø chi-o una chera xaridan? (chi-o fronted) c. *Yani/Ø chera chi-o una xaridan? (chera and chi-o fronted) (well-formed as an echo question with no Q particle)
Persian wh-riddles 167
2.1 Riddle 1: Wh-in situ but not allowing both SP/PL readings As observed above, single-pair interpretation is repeatedly associated with wh-in situ. Significantly, the prediction is not empirically borne out against Persian data: Persian, a wh-in situ language with an optional Q-marker in the initial position to mark its interrogatives, seems to pattern with English rather than Japanese, Chinese, and Hindi. There is a very strong tendency in Persian to afford only pair-list interpretations in multiple wh-questions, which makes potential problems for both Hagstrom’s (1998) semantics of multiple questions and Boškovic´’s (1998) and Grohmann’s (1999, 2000) adaptations. In an attempt to empirically test the availability of SP/PL readings in Persian, a group of 40 adult native-speakers of (Esfahani) Persian studying at Azad University were asked to indicate on a five-point-scale how infelicitous a multiple question was in each of the two situations described below. In all cases, the Q-marker yani was employed to signal the question rather than aya because in informal Persian yani (or preferably no Q-marker at all) is used in wh-questions. Since multiple questions are rarely employed in written Persian, it was decided to assume a more conversational style in writing the items in question. Their ratings are tabulated for each case separately: (10) Situation I You are in a store and off in the distance see somebody buying an article of clothing, but you do not see who it is and exactly what is being bought. You go to the shop assistant and ask: A: Yani ki chi xarid? q who what bought ‘Who bought what?’ B: Ali ye pirhan xarid. Ali a shirt bought ‘Ali bought a shirt.’ Table 1. Scale of infelicity:
0
1
2
3
4
Number of participants who preferred each point on the scale: Number in percentages: Rating scores: Possible MAX: Possible MIN:
0
3
2
17
18
0% 0 160 0
7.5% 3
5% 4
42.5% 51
45% 72
Total:
130 (out of 160)
168 Ahmad R. Lotfi
(11) Situation II You are paying a social visit to a newly-wed couple in their apartment. While having a friendly conversation about their wedding presents, you ask about both what they got and who gave them each item: A: Yani ki chi ovord? q who what brought ‘Who gave you what?’ B: Ali ye sa’at ovord, Maryam ye angoshtar ovord, Mina ye Ali a clock brought Maryam a ring brought Mina a goldun ovord… vase brought ‘Ali gave us a clock, Maryam gave us a ring, Mina gave us a vase…’ Table 2. Scale of infelicity:
0
1
2
3
4
Number of participants who preferred each point on the scale: Number in percentages: Rating scores: Possible MAX: Possible MIN:
15
15
2
5
3
5% 4
12.5% 15
7.5% 12
Total:
46 (out of 160)
37.5% 37.5% 15 0 160 0
The ratings suggest that the PL-reading of a multiple question in Persian is about 2.83 times more felicitous than its SP-reading. Then: (12) SP-reading infelicitous in Persian3 a. #Ki chi /chi-o xarid? who what/what-case bought ‘Who bought what?’ b. Ki chi /chi-o ovord? who what/what-case brought ‘Who brought what?’
2.2 Riddle 2: Why Q-markers? Hagstrom (1998) argues convincingly that the occurrences of Japanese -ka in (13a) and (13b) are instances of one and the same fundamental morpheme (p. 15):
Persian wh-riddles 169
(13) Different instances of -ka morpheme (Japanese) a. John-ga nani-ka-o kaimasita. (an indefinite prior John-nom what-q-acc bought to Q-movement) ‘John bought something.’ b. John-ga nani-o kaimasita ka? (a question after Q-movement) John-nom what-acc bought q ‘What did John buy?’
As indefinites have no inherent quantificational force of their own (Hagstrom 1998: 131) but can “pick up” some from the environment, the Q morpheme contributes an existential quantification to the semantics of questions after its movement. In multiple wh-questions, a pair-list reading is made possible because Q is base-generated as the sister of the lower wh-word (e.g. Wh2) and as a result “[t]he wh-word outside the scope of Q has a set as its semantic value, and with the help of flexible functional application, yields a set of questions” (p. 169). He also observes that “multiple questions in Japanese can occur without a Q-marker and nevertheless be interpreted as questions. Flexible functional application predicts that such questions should exist, and moreover correctly predicts that they are limited to a “single-pair” reading” (p. 169). What is the position in which a Persian Q-marker (say yani) is optionally generated in a multiple wh-question like ‘yani Armin chio be ki dad’ (what did Armin give to whom)? There is no morphological evidence like that on Japanese -ka to suggest that Q is base-generated as a sister to any wh-word. This does not mean, however, that it is not. Then one can conceive of three hypothetical positions: (i) The Q-marker is generated where it is pronounced, i.e. there is no Q-movement at all, (ii) it is base-generated as a sister to Wh1 (chio), and (iii) it is base-generated as a sister to Wh2 (ki). If Hagstrom is right in his arguments, the only candidate position for the base generation of the Q-marker is Wh2, i.e. the lowest wh-word in the sentence; otherwise, the question could only have an SP-reading due to the structural height of the positions (i) and (ii) that will put both wh-words in the scope of the launching site of Q. On the other hand, as Hagstrom observes, the question must have an SP-reading whenever the Q-marker does not occur in the question. The prediction does not seem to be borne out by Persian data. Persian, just like Japanese, affords multiple questions without Q-markers. Contrary to Japanese (and also French wh-in situ), however, even in such cases, a single-pair reading is unavailable:
170 Ahmad R. Lotfi
(14) PL-reading in Persian wh-in situ questions with no Q-marker a. Ki chi goft? (pair-list answer only) who what said ‘Who said what?’ b. Armin koja chi xarid? (pair-list answer only) Armin where what bought ‘Where did Armin buy what?’ c. Armin kei koja raft? (pair-list answer only) Armin when where went ‘When did Armin go where?’ d. Armin kei ki-o did? (pair-list answer only) Armin when who-case saw ‘When did Armin see who?’
Apparently, the syntactico-semantic requirements of questions make impossible any Q-marker in Persian wh-questions. On the other hand, Persian Q-markers in multiple questions (though optional, even infrequent) are possible and do occur from time to time (see the description of Persian questions at the beginning of Section 2 for details). The impossible has been made possible! How? 2.3 Riddle 3: Ad hoc Superiority effects/violations Rudin (1988) had noticed that the order of wh-phrases in Bulgarian multiple questions corresponded to those in English: the leftmost wh-phrase in Bulgarian is the same that moves overtly in English. The next Bulgarian wh-phrase is the one that remains in situ in the other language. Pesetsky (2000) considers Chomsky’s (1995) “Attract Closest” (AC) to be a restatement of the Superiority Condition (Chomsky 1973) according to which no rule can involve X, Y in the structure … X … [α… Z … __ WYV…] where the rule applies ambiguously to Z and Y and Z is superior to (m-commands) Y. It follows that in a multiple wh-fronting language like Bulgarian, the leftmost (highest) wh-phrase moves first in the fulfillment of AC. The second wh-phrase, according to Richards (1997), “tucks in” below the first phrase making a lower specifier in CP. However, as Pesetsky (2000) observes, Superiority effects disappear with more than two wh-phrases: (15) Bulgarian: Wh1 Wh2 Wh3, or Wh1 Wh3 Wh2 a. Koi na kogo kakvo dade? who to whom what gave ‘Who gave what to whom?’
Persian wh-riddles
b. ?Koi kakvo na kogo dade?
For Richards (1997), the Principle of Minimal Compliance (PMC) is a general meta-constraint that explains such apparent Superiority violations (among other phenomena). Each operation pays some sort of “tax” that makes it possible for all other instances of that operation to ignore the relevant requirement. It follows that in Bulgarian multiple questions, both Wh1 Wh2 Wh3 and Wh1 Wh3 Wh2 are almost equally acceptable. Pesetsky (2000: 50) shows that “[t]he free choice of which wh-phrase moves overtly in [English] ternary questions with Wh1-in situ is … the same fact, in essence, as the free choice of which wh-phrase moves second in Bulgarian overt phrasal movement. […] We thus have support for the hypothesis that in apparent Superiority violations in English, the first instance of wh-movement is actually feature movement — specifically, feature movement that satisfies AC.” As observed earlier, Persian is a wh-in situ language with a basic SOV sentential word order. This means the morpho-syntactic requirements of a wh-phrase — whatever they are — are satisfied without a need for the phrase itself to move overtly from the position in which it is base-generated. Despite that, it is quite possible (and not very infrequent) to front one or more wh-phrases because of focusing or other discourse-related reasons. (16) Optional wh-fronting in Persian a. Armin koja raft? Armin where went ‘Where did Armin go?’ b. Koja Armin raft? (though Koja raft Armin? is more natural) c. Armin chi xarid? Armin what bought ‘What did Armin buy?’ d. Chi Armin xarid? (also Chi xarid Armin?) e. Un be ki shabih-e? s/he to who like-is ‘Who does s/he resemble?’ f. Be ki un shabihe? (also Be ki shabihe un?)
Persian multiple questions, too, reveal similar effects. One or more wh-phrases may be fronted, however:
171
172 Ahmad R. Lotfi
(17) Optional multiple wh-fronting in Persian a. Armin chi-o chera xarid? Armin what-case why bought ‘Why did Armin buy what?’ b. Chi-o Armin chera xarid? c. Chi-o chera Armin xarid? d. Tou ki-o koja didi? you who-case where met ‘Where did you meet whom?’ e. Ki-o tou koja didi? f. Ki-o koja tou didi?
Interestingly, although wh-fronting seems to be optional, once attempted certain Superiority constraints are inevitable: (18) Superiority effects in Persian optional multiple wh-fronting a. *Chera chi-o Armin xarid? (well-formed as an echo question) why what-case Armin bought ‘Why did Armin buy what?’ b. *Koja ki-o tou didi? (well-formed as an echo question) where who-case you met ‘Where did you meet whom?’
What will happen if the subject of the sentence is a wh-word? Is a subject wh-word displaced out of IP overtly (say to SpecCP)? The answer to this question seems to be negative. But even in absence of overt wh-subject movement, subject > object/non-subject Superiority effects still seem to be real. (19) With a wh-word as the subject a. Ki chi xarid? who what bought ‘Who bought what?’ b. *Chi ki xarid? c. Ki koja raft? who where went ‘Who went where?’ d. *Koja ki raft? e. Ki kei raft? who when went ‘Who went when?’ f. *Kei ki raft?
(well-formed as an echo question)
Persian wh-riddles
g. Ki koja un-o did? who where him/her met ‘Who met him where?’ h. *Koja ki un-o did?
So far not so bad. Real troubles begin when the direct object particle ra (usually cliticised as ro/-o) is attached to the end of the object wh-phrase. Apparently, specifying the accusative case of the object wh-phrase licenses certain Superiority violations when fronting wh-phrases. (20) Superiority violations with the Case-marker ra a. Ki chi-o xarid? who what-case bought ‘Who bought what?’ b. Chi-o ki xarid? c. Armin koja chi xarid? Armin where what bought ‘Where did Armin buy what?’ d. *Armin chi koja xarid? e. Armin chi-o koja xarid? f. Armin kei chi xarid? Armin when what bought ‘When did Armin buy what?’ g. *Armin chi kei xarid? h. Armin chi-o kei xarid?
Inconsistencies with other languages also show up when multiple wh-questions with more than two wh-phrases are formed. Contrary to Pesetsky’s observation, Persian Superiority effects are not violable even with more than two wh-words. Despite that, the case marker ra makes certain violations possible here, too. (21) Superiority effects in multiple questions with more than two wh-phrases a. Ki koja chi xarid? who where what bought ‘Who bought what where?” b. *Ki chi koja xarid? c. *Chi ki koja xarid? d. *Ki chi chera xarid? e. *Chi ki koja xarid? f. *Koja chi ki xarid?
173
174 Ahmad R. Lotfi
(22) Superiority violations that ra can eliminate a. Ki chi-o koja xarid? who what-case where bought ‘Who bought what where?” b. Chi-o ki koja xarid? c. Ki chi-o chera xarid? d. Chi-o ki chera xarid?
Just to add to the mystery, there seems to be a class of Superiority violations even ra cannot eliminate. (23) Superiority violations that ra cannot eliminate a. Armin chi-o chera xarid? Armin what-case why bought ‘Why did Armin buy what?’ b. Chi-o Armin chera xarid? c. *Chera Armin chi-o xarid? d. Chi-o chera Armin xarid? e. *Chera chi-o Armin xarid? f. Ki koja chi-o chera xarid? who where what-case why bought ‘Who bought what where?” g. *Ki chera koja chi-o xarid?
How can one explain such Superiority effects/violations? What is the role of ra in this respect? Is there any chance that these three “wh-riddles” are somehow related to each other? In Section 3, I’ll try to see if a unified account of these riddles with minimal revision of standard minimalist assumptions is possible.
3. Land ho! 3.1 The magic of ra! The syntax and semantics of the Persian morpheme ra (usually cliticised as -ro or -o) have been subject to remarkable controversy in Persian linguistics. Browne (1970) analyses the morpheme as a specificity marker. Karimi (1989, 1990) extends Browne’s analysis to ra as “specific-oblique marker”. Accordingly, a ra-marked NP is either definite-specific or indefinite-specific. She also analyses the morpheme as an oblique (= NOT nominative)-case marker.
Persian wh-riddles
Peterson (1974) and Windfuhr (1979, 1987) hypothesise some topic-marking function for it. Ghomeshi (1996, 1997), on the other hand, challenges all these analyses and claims that the morpheme is a phrasal affix heading a KP (Kase Phrase) that case-marks noun phrases adjoined to VP. Apparently, they all agree that direct objects marked with ra scramble (to the front of the sentence) rather freely. While these controversies are both conceptually and empirically important (and consequently impossible to remain impartial with regard to them), I’ll try to avoid the unnecessary controversies for the sake of the question under study here. I assume the following to be the minimal requirements for solving the wh-riddles presented in this paper: (24) Minimal assumptions required here on the syntax/semantics of ra a. Indefinite DPs can be ra-marked, too. b. Ra satisfies the case requirements of the DP it marks. c. Ra heads a phrase, say KP, with the DP it marks as the complement. d. Ra-marking facilitates scrambling. e. Ra marks (secondary) topics.
I adopt structural configurations of KP and DP similar to Ghomeshi’s (1996, 1997) with three modifications: (i) I ignore the distinction she makes between case assignment and case marking for the sake of convenience, (ii) contrary to Ghomeshi, I entertain the possibility here that the null determiner heading the complement DP can also be indefinite, and (iii) in support of Ghomeshi’s general approach, I also include the suffix -e (used in informal Persian to mark definiteness) in the tree: (25) Structural configurations of KP and DP a. KP with DP as the complement b. KP DP NP
K D
-ra
DPs DP NP
D -e/Ødef/-i/Øindef
-e/-i/Ødef
It follows that among four types of object DPs, the morpheme ra marks the first three, i.e. the definite determiner -e, the null definite, and the indefinite -i (ra-marking optional in the last case). I hypothesise that the fourth type, the null indefinite, is case-marked directly by the transitive verb. This correlates
175
176 Ahmad R. Lotfi
with the semantics of the null indefinite determiner that makes the DP the nonreferential nominal element of a phrasal verb: (26) Persian phrasal verbs with a null indefinite determiner heading the DP (non-referential DP; phrasal verb) a. Armin [[ghaza Øindef] xord]. Armin [[food ate ‘Armin ate.’ b. Armin [[ghaza -i] xord]. (non-referential DP; phrasal verb) Armin [[food indef ate ‘Armin ate some food.’ c. Armin [[ghaza -i] -ro] xord. (indefinite DP; simple verb) Armin [[food indef case ate ‘Armin ate some food.’ d. Armin [ghaza Ødef] -ro] xord. (definite DP; simple verb) Armin [food case ate ‘Armin ate the food.’ e. Armin [[ghaza -e] -ro] xord. (definite DP; simple verb) Armin [[food def case ate ‘Armin ate the food.’
What happens when a wh-word takes the position of the object DP? If Hagstrom (1998) is right in associating wh-words and indefinites, only sentences (26a–c) allow such assertion. The unacceptability of (27a) below supports Hagstrom’s arguments. I conclude that (27b) is the wh-inserted form of (26c) rather than (26d). Sentences (26a–b) both change to (27c) after the insertion of the wh-word. In (27b), ra case-marks the wh-phrase. In (27c), on the other hand, the transitive verb xordan ‘to eat’ does so. (27) Wh-questions with indefinite objects (Persian) a. *Armin ch-e (chi-e) ro xord? b. Armin chi-ro/-o xord? c. Armin chi xord?
In what follows, I argue that minimalist solutions to the Persian wh-riddles are conceivable in the light of the discussion above on the syntax and semantics of ra. 3.2 Q-Anti-Superiority Assuming Hagstrom’s (1998) syntax/semantics of Q’s in wh-in situ to be essentially true, I hypothesise that Persian wh-words, like Japanese ones, contain some morpheme Q (bearing the feature [Q]) that is base-generated as the sister to the
Persian wh-riddles 177
wh-word (under the node D in the trees in (25)). Likewise, Q contributes to the indefinite interpretation of the wh-word before Q-movement. The fact that all Persian wh-words can be paraphrased as phrasal Wh’s with a wh-element (either che or ke depending upon the value of the feature [Human] of the following nominal) followed by an indefinite pronoun (like chizi ‘something’, kasi ‘someone’, jaii ‘someplace’, etc.) supports this hypothesis: (28) Persian wh-words paraphraseable as phrasal ones a. Ki ketab xarid? who book bought ‘Who bought books?’ b. Che kas-i ketab xarid? which someone book bought ‘Who bought books?’ c. Armin chi xarid? Armin what bought ‘What did Armin buy?’ d. Armin che chiz-i xarid? Armin which something bought? ‘What did Armin buy?’ e. Armin koja raft? Armin where went ‘Where did Armin go?’ f. Armin che ja-i raft? Armin which place went ‘Where did Armin go?’ g. Armin kei raft? Armin when went? ‘When did Armin go?’ h. Armin che vaght-i raft? Armin which sometime went ‘When did Armin go?’ i. Armin chera raft? Armin why went ‘Why did Armin go?’ j. Armin be che elat-i raft? Armin for which reason went ‘Why did Armin go?’
178 Ahmad R. Lotfi
k. Armin chetor raft? Armin how went ‘How did Armin go?’ l. Armin be che shekl-i raft? Armin to which form went ‘How did Armin go?’ m. Ki chi xarid? who what bought ‘Who bought what?’ n. Che kasi-i che chiz-i xarid? which someone which something bought ‘Who bought what?’
Inspired by Hagstrom’s analysis of wh-markers, I also propose a principle of Q Anti-Superiority in Persian: (29) Q Anti-Superiority in Persian Q is base-generated as an indefinite morpheme that is sister to the lowest wh-word.
The (originally indefinite) Q morpheme moves (as feature movement) to the beginning of the question in the fulfillment of certain morphological requirements (whatever they are). Because of ‘Q Anti-Superiority’, Persian multiple wh-questions are necessarily interpreted as pair-list questions: at least one wh-word lies outside the scope of the launching site of Q. Q-movement as feature movement, on the other hand, accounts for the PL-reading of the question even when a Q morpheme does not overtly occur at the beginning of the question: (30) Q base-generated as the sister to the lowest wh-word a. Q ... [... ki (che kas-i) ... tQ koja (che jai-i) ...] b. q che kasi che jaii raft? ‘Who went where?’ (31) Derivation of (30b) Before introducing wh-morphemes into the derivation: kas-i jai-i raft. person-indef place-indef went ‘Someone went somewhere.’
pair-list reading
Persian wh-riddles 179
Step 1: Step 2: Pronunciation: Interpretation:
wh-morphemes are introduced in the derivation. CH-kas-i CH-jai-iQ raft Q is attracted to the clause periphery covertly. Qi [CH-kasi CH-jai-ti raft] che kasi che jaii raft? pair-list
(32) Tree diagram for (30b) CP ... wh-phraselowest che
DP NP
D
ja
-iQ
Questions with a single wh-word are derived similarly with Q as the (indefinite D) sister of the lowest wh. In questions with a single wh-word, Wh1 is obviously the lowest one, too. I hypothesise that even in multiple questions with a Q-marker, the feature [Q] is still attracted to CP covertly. The complementiser phrase, on the other hand, is here headed by aya (or yani) as the complementiser C, which also contains a feature [Q]. There is independent evidence from formal Persian that aya is a complementiser in Persian comparable to English whether. Contrary to whether, however, aya can also be used in matrix sentences: (33) Aya as a complementiser a. Madar mixahad bedanad aya/Ø tou berasti baradar-at mother wants know-subj whether you really brother-your ra didi. case met ‘Mother wants to know whether you really met your brother.’ b. Aya/Ø madar berasti in ra mixahad? q mother really this case wants ‘Does mother really want this?’
180 Ahmad R. Lotfi
(34) Aya/yani as C ‘probing’ the indefinite morpheme CP C¢ C
...
aya[Q]
DP -i[Q]
In questions with no Q-marker, the head C with the feature [Q] is still there without having a phonetic realization. 3.3 Scrambling, (multiple) wh-fronting and Superiority effects in Persian Scrambling is a common phenomenon in Persian. It enables Persian-speakers to afford the word orders SVO, OSV, and OVS (perhaps even VSO and VOS rather marginally) in addition to the basic SOV: (35) Scrambling in Persian a. Armin Elnaz-o did. Armin Elnaz-case saw ‘Armin saw Elnaz.’ b. Armin did Elnaz-o. c. Elnaz-o Armin did. d. Elnaz-o did Armin. e. ?Did Armin Elnaz-o. f. ?Did Elnaz-o Armin.
(SOV)
(SVO) (OSV) (OVS) (VSO) (VOS)
Interestingly, in sentences with an indefinite object DP, in which the casemarker ra can be dropped, scrambling is seriously constrained, especially in contexts where semantico-pragmatic clues cannot help the hearer to decide what the subject/object is: (36) Scrambling limited in absence of ra a. Sag-i gorbe-i did. dog-indef cat-indef saw ‘A dog saw a cat.’ b. Sag-i did gorbe-i.
(SOV)
(SVO)
Persian wh-riddles
c. d. e. f.
?Did sag-i gorbe-i.
*Gorbe-i did sag-i. *Gorbe-i sag-i did. *Did gorbe-i sag-i.
(VSO) (OVS) (OSV) (VOS)
Karimi (1999) shows that scrambling in Persian obeys movement conditions such as the MLC. She also observes that scrambling in Persian is not triggered by EPP but optionally by Topic and Focus (p. 4). “Persian does not exhibit obligatory wh-movement. These phrases, however, are subject to scrambling. […] More than one wh-phrase may undergo scrambling. […] [F]ocus is an instance of Move which is triggered by the feature Foc. The landing site of the moved element is the specifier of a functional projection whose head carries a comparable feature” (p. 6). Analyses of wh-fronting as movement for focusing reasons have also been proposed for Serbo-Croatian wh-phrases (Stjepanovic´ 1995) and Bulgarian (Boškovic´ 1998b). If my analysis of the Q-marker heading the CP to which [Q] moves ((33)–(34) above) is on the right track, this functional projection onto which focused wh-phrases land must be lower than CP. This suggests that Pesetsky’s (2000) solid analysis of multiple wh-fronting in terms of specifier potential of Cm-spec simply does not apply to Persian multiple wh-fronting. (37) Persian wh-phrases landing somewhere lower than SpecCP [CP yani/aya [ki-o [Armin koja did?]]] q [whom [Armin where met ‘Where did Armin meet whom?’
Concerning the syntax of the pair-list reading for multiple questions, Pesetsky (2000: 98) argues that such a reading is conditional upon the attraction of wh-features to the complementiser of the clause. If Persian wh-questions do not overtly move to a position as high as SpecCP, as the data above suggest, then Pesetsky’s condition on the availability of PL-reading must be satisfied covertly in Persian questions whether they remain in situ or scrambled to the left for the sake of focusing. This strongly suggests that, contrary to Bulgarian-type languages, multiple wh-fronting in Persian is not motivated by the feature [Wh]. Despite that, as shown in examples in Section 2, Persian multiple wh-questions show certain Superiority effects. I propose a descriptive rule in (38) below in order to account for these Superiority effects in Persian. (38) A descriptive rule of Superiority effects in Persian A wh-adjunct cannot cross a wh-object-argument.4
181
182 Ahmad R. Lotfi
The rule is silent on the issue of wh-arguments crossing wh-arguments and wh-adjuncts. I propose that if the Case Filter is satisfied for the wh-object, it can freely scramble in the fulfillment of focusing requirements of the sentence. Adjuncts, on the other hand, need no case marking in order to scramble, because they are not nominal in category. (39) Case Filter and moving wh-phrases a. A wh-object-argument can cross arguments/adjuncts as long as its case requirements are satisfied (by ra). b. Adjuncts need no case-marking to license their scrambling.
Why do wh-arguments and wh-adjuncts show different Superiority effects? In what follows, I propose a unified account of both wh-arguments and wh-adjuncts in which the differences between these two classes are reduced to the timing of scrambling: (40) A timing analysis of differences between wh-arguments and adjuncts in Persian A-scrambling Æ Q-movement Æ A¢-scrambling a. wh-arguments scramble before Q-movement. b. wh-adjuncts scramble after Q-movement.
In (41a) below, chi cannot scramble as it is case-marked by the transitive verb xaridan. The wh-adjunct koja in (41b), on the other hand, scrambles earlier than Q-movement. As a result, Q is generated next to the lowest wh-word chi. The PL-reading ensues. In (41c), the morpheme ra case-marks the wh-word chi. Because chi is an argument, it scrambles (if it moves at all: scrambling in Persian is optional) before Q-movement. Consequently, it will be the [Q] feature of koja that will move. This will block the A¢-scrambling of koja in (41d–e).5 In (41f), chi-o does not scramble; hence, [Q] will be generated in its sisterhood. Adjunct scrambling will not be blocked. (41) wh-/adjunct scrambling a. q ki koja chi[Q] xarid? who where what bought ‘Where did who buy what?’ b. q Koja ki chi[Q] xarid? c. q Chi-o ki koja[Q] xarid? d. *q Chi-o koja[Q] ki xarid?
(well-formed as an echo question)
Persian wh-riddles 183
e. *q Koja[Q] chi-o ki xarid? f. q Koja ki chi[Q]-o xarid?
(well-formed as an echo question)
The analysis of multiple wh-questions outlined throughout the paper allows us to explain the relevant data in a more principled way without unnecessary challenging of the findings from other languages like English, Bulgarian, and Japanese. Minimalist accounts of multiple wh-questions (Boškovic´ 1998, 2000, Hagstrom 1998, Grohmann 1999, 2000, and Pesetsky 2000 among others) are not in contradiction to Persian data anymore. Hagstrom’s syntax/semantics of wh-in situ and Pesetsky’s model of movement are both applied in the analysis of Persian multiple questions here. On the other hand, the analysis is compatible with the major findings of minimalist/P&P studies of scrambling and casemarking in Persian (Karimi 1989, 1990, 1999, Ghomeshi 1997, among others). A number of significant questions such as the possible weaknesses of minimalist theses of movement in general and those of the probe-goal system in particular, the nature of features [Q] and [Wh], un/interpretability, the proposed mechanisms of satisfying morphological requirements like checking, sharing and Agree, and the status of interface-levels with regard to wh-questions were all swept under the rug in order to keep the analysis as close to mainstream minimalism as possible without jeopardizing the author’s own brand of minimalist syntax. Despite that, fundamental questions concerning the status of language faculty and our understanding of it will inevitably recur over and over in any survey of any human language. Wisdom of Persian, like that of any other language, will shed more light on the dark corners of human language and cognition.
Notes * I would like to thank Kleanthes Grohmann for his careful reading of the first draft of this article, the stylistic corrections he made, and also his valuable comments on my analyses of the phenomena under study here. All shortcomings remain solely mine. 1. Other implementations are also conceivable, e.g. Boškovic´’s (1998b), in which no clustering of wh-phrases takes place. Instead, only Wh1 moves overtly with other wh-phrases moving for focusing purposes. My analysis of Persian data in 3.3 below is not incompatible with Boškovic´’s with the significant exception that in Persian even fronting Wh1 is focus movement rather than wh-movement to SpecCP. 2. Sentences (8b–c) are taken from Xanlari’s Persian Grammar (1976, 4th edition, Babak Publishing Company, Tehran, Iran, p. 107), which is now a classic. The work is written in Persian. The original sentences have got no pros.
184 Ahmad R. Lotfi
3. Unless necessitated by the ongoing discussion, Q-markers will be omitted throughout the reminder of the text for the reason of naturalness. 4. The technical implementation of this descriptive generalization is still another “Persian riddle” to me. Kleanthes Grohmann’s suggestion (p.c.) to study this under Lebeaux’s (1988) late insertion approach to adjuncts sounds very interesting. The other possibility is to think of such Persian wh-adjuncts as chera to be inserted within a wh-object-argument with the adjunct clustered to the object before the latter moves. The adjunct is then “frozen” inside the wh-object-argument (due to the fulfillment of its sharing requirements — see Lotfi 2000 for a discussion of feature sharing). It cannot move into FocP anymore unless together with the whole wh-object-argument (cf. Grewendorf ’s 2001 analysis of wh-clustering in German; also Sabel’s 2001 analysis of the supportive evidence for the Cluster Hypothesis from a number of languages, such as Bulgarian, Japanese, and Malagasy, as well as Sabel’s contribution to this volume). In any case, it does not cross the head of its “host” wh-object-argument. This can also explain why the ban only concerns wh-objects but not subjects. In either case, I find a technical analysis of the descriptive rule too tentative to elaborate on it here. 5. The unmarked wh-in situ word order in Persian seems to be Subj +Place/ Time+Obj+Verb: ki koja chi xarid? Once Obj is case-marked by ra, it can scramble to the left. This permits a wh-Reason to appear between Obj and the verb. Otherwise, the wh-word chera (Reason) cannot occur: (i) a.
Ki koja chi-o chera xarid? who where what-case why bought ‘Why did who buy what where?’ b. *Ki koja chera chi xarid? (also *Ki koja chera chi-o xarid?)
The timing analysis undergenerates chi-o chera una xaridan (lit. ‘what why they bought’), which is acceptable to some speakers. Apparently, chio has already crossed Subj. This must have frozen chera in place with Q low enough to ensure pair-list-only reading. Although an adjacency analysis of wh-phrases can account for the grammaticality of the sentence, it overgenerates *koja chi-o una xaridan. The timing analysis, on the other hand, rules out both *koja chi-o una xaridan and *chi-o koja una xaridan (both unacceptable to the author but acceptable to another informant) and generates chi-o una koja xaridan and koja una chi-o xaridan (acceptable to both of us) instead. One possibility is that chera is clustered to chi-o (cf. Grewendorf 2001, Sabel 2001, this volume) before the (optional) fronting of the whole cluster (A-scrambling): (ii) [chi-o chera]i [una ti xaridan]? [what-case why [they bought ‘Why did they buy what?’ This can also explain the availability of chi-o koja una xaridan/koja chi-o una xaridan to some Persian-speakers: for them, but not others, koja and chi-o form a cluster before scrambling. If the adjunct adjoins the argument, the cluster may move before Q-movement. On the other hand, if the argument adjoins the adjunct, it may be fronted after Q-movement.
Persian wh-riddles
References Boškovic´, Ž. 1998a. “On the interpretation of multiple questions.” In A Celebration: Essays for Noam Chomsky’s 70th Birthday, J. Fodor, S. J. Keyser and A. Brand (eds). Cambridge, Mass.: MIT Press. [Online publication — available for download at http:// cognet.mit.edu/Books/celebration/essays/Boeckx.html (sic); revised version to appear in Linguistic Variation Yearbook 1.] Boškovic´, Ž. 1998b. Wh-phrases and wh-movement in Slavic. Ms. University of Connecticut, Storss. Boškovic´, Ž. 2000. “Sometimes in [Spec, CP], sometimes in situ.” In Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels, J. Uriagereka (eds), 53–87. Cambridge, Mass.: MIT Press. Browne, W.1970. “More on definiteness markers: Integratives in Persian.” Linguistic Inquiry 1: 591–656. Chomsky, N. 1973. “Conditions on transformations.” In A Festschrift for Morris Halle, S. Anderson and P. Kiparsky (eds), 232–286. New York: Holt, Reinhart and Winston. Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. 2000. “Minimalist inquiries: The framework.” In Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels, J. Uriagereka (eds), 89–155. Cambridge, Mass.: MIT Press. Chomsky, N. 2001. “Derivation by phase.” In Ken Hale: A Life in Language, M. Kenstowicz (ed), 1–52. Cambridge, Mass.: MIT Press. Dabir-Moghaddam, M. 1992. “On the (in)dependence of syntax and pragmatics: Evidence from the postposition -ra in Persian.” In Cooperating with Written Texts, D. Stein (ed), 549–573. Berlin: Mouton de Gruyter. Ghomeshi, J. 1996. Projection and inflection: A study of Persian phrase structure. Doctoral dissertation, University of Toronto. Ghomeshi, J. 1997. “Topics in Persian VPs.” Lingua 102: 133–167. Grewendorf, G. 2001. “Multiple wh-fronting.” Linguistic Inquiry 32:87–122. Grohmann, K. K. 1999. German is a multiple wh-fronting language! Paper presented at the Colloque de Syntaxe et Sémantique à Paris 3, Paris. [October 1999] Grohmann, K. K. 2000. Prolific peripheries: A radical view from the left. Doctoral dissertation, University of Maryland, College Park. Hagstrom, P. 1998. Decomposing questions. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Karimi, S. 1989. Aspects of Persian syntax, specificity, and the theory of grammar. Doctoral dissertation, University of Washington, Seattle. Karimi, S. 1990. “Obliqueness, specificity, and discourse functions: Ra in Persian.” Linguistic Analysis 20:139–191. Karimi, S. 1999. “Is scrambling as strange as we think it is?” In MIT Working Papers in Linguistics 33. Papers on Morphology and Syntax, Cycle One, K. Arregi, B. Bruening, C. Krause and V. Lin (eds), 159–190. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT).
185
186 Ahmad R. Lotfi
Lebeaux, D. 1988. Language acquisition and the form of the grammar. Doctoral Dissertation, University of Massachusetts, Amherst. Lotfi, A. R. 2000. Semantico-Phonetic Form: A unitarianist grammar. Ms., Azad University at Esfahan. [http://www.geocities.com/arlotfi/lotfipage.html] Pesetsky, D. 1997. “Optimality theory and syntax: Movement and pronunciation.” In Optimality Theory: An Overview, D. Archangeli and D. Terence Langendoen (eds), 134–170. Oxford: Blackwell. Pesetsky, D. 2000. Phrasal Movement and Its Kin. Cambridge, Mass.: MIT Press. Peterson, D. 1974. Noun phrase specificity. Doctoral dissertation, University of Michigan, Ann Arbor. Richards, N. 1997. What moves where in which language? Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Rudin, C. 1988. “On multiple questions and multiple wh-fronting.” Natural Language and Linguistic Theory 6: 455–501. Sabel, J. 2001. Deriving multiple head and phrasal movement: The cluster hypothesis. Linguistic Inquiry 32: 532–547. Sabel, J. This volume. “Malagasy as an optional multiple wh-fronting language.” 229–254. Stjepanovic´, S. 1995. Short-distance movement of wh-phrase in Serbo-Croatian matrix clauses. Ms., University of Connecticut, Storrs. Windfuhr, G. L. 1979. Persian Grammar: History and State of Its Study. The Hague: Mouton de Gruyter. Windfuhr, G. L. 1987. “Persian.” In The World’s Major Languages, B. Comrie (ed), 523–546. London: Croom Helm.
Non-wh-fronting in Basque* Lara Reglero University of Connecticut
1.
Introduction
According to Boškovic´ (1999), all wh-phrases obligatorily move to the beginning of the sentence in Serbo-Croatian.1 Since the movement of one wh-phrase should be enough to check the strong +wh-feature of C, one can deduce that wh-phrases in Serbo-Croatian must be fronted overtly for independent reasons. Boškovic´ (1999) argues that Serbo-Croatian wh-phrases undergo focus movement to a position below C. He also shows that the following properties of Serbo-Croatian can be accounted for in a principled way under this analysis: lack of Superiority effects (i.e. free order of Serbo-Croatian wh-phrases) and the availability of single-pair answers. In this paper I will evaluate Boškovic´’s (1999) proposal with respect to question formation in Basque. The present study is the first attempt to offer an account of the mechanisms underlying multiple question formation in Basque. Previous research (Ortiz de Urbina (1995, 1999a)) has found that wh-words in Basque behave very similarly to elements bearing focus. However, there are no detailed studies dealing with multiple question formation in this language. In this paper I will offer an analysis of multiple questions in Basque which will help us gain a deeper insight into the nature of “wh”-movement in this language. The paper is organized as follows. First, I will introduce Boškovic´’s (1999) theory. This will be crucially important since I will use the tools provided by this author in trying to analyze the Basque data. Second, I will introduce the data to be accounted for. As shown below, Basque allows two strategies to ask a question such as Who bought what?. One possibility is that one wh-phrase fronts and the other stays in situ. The other possibility is that both wh-phrases move overtly to the front of the sentence. We will see that, independently of the strategy employed, Superiority effects always show up in this language. Third, I will offer some background regarding “wh”-movement in Basque. As will
188 Lara Reglero
become clear, the driving force behind wh-fronting in this language is focus. Fourth, I will give an analysis for the first strategy found in Basque. The basic idea is that, contrary to appearances, both wh-phrases move overtly in the syntax. Fifth, I will provide an analysis for the second strategy, that is, the Multiple wh-Fronting strategy. As shown below, wh-phrases raise to a syntactic position below C. Finally, I will offer a brief discussion of multiple questions in long-distance wh-movement contexts.
2. Theoretical background In this section I will discuss the three main ingredients of Boškovic´’s system: focus, Superiority and the interpretation of questions. 2.1 Focus In a series of papers (Boškovic´ 1997a, 1997c, 1998b, 1999, 2002), Boškovic´ argues that wh-phrases in Serbo-Croatian undergo focus movement.2 The main motivation for this approach comes from Stjepanovic´’s (1998) work. According to her, wh-phrases in Serbo-Croatian behave like contrastively focused nonwh-phrases in many respects.3 She interprets these similarities as an indication that wh-phrases in Serbo-Croatian are inherently focused and therefore undergo overt movement for focus purposes. This idea goes back to Horvath (1986). In her system, the movement of wh-phrases in certain languages is analyzed as an instance of focus movement. This line of research establishes a connection between the movement of wh-phrases and the movement of contrastively focused non-wh-phrases. In other words, if a language allows overt movement of contrastively focused non-wh-phrases, wh-phrases in that language front for focus reasons. It is important to emphasize that the notion of contrastive focus is crucial in this analysis. As É. Kiss (1998) points out, there are two different types of focus: identificational focus, which expresses exhaustive identification and occupies the specifier of a functional projection in focusmovement languages, and information focus, which conveys new, nonpresupposed information and is not associated with movement. The former type is also known as narrow or contrastive focus and the latter as wide or presentational focus. We are interested in the first notion. Boškovic´, building on Stjepanovic´’s (1998) work, adopts the idea that wh-fronting in Serbo-Croatian should be analyzed as focus movement. Before
Non-wh-fronting in Basque 189
doing that, he shows first that all wh-fronting in Serbo-Croatian is not wh-movement. For example, all wh-phrases in Serbo-Croatian must front. If we were dealing with wh-movement, the movement of one wh-phrase would suffice to check the +wh-feature of C. Since all wh-phrases must move overtly in SerboCroatian (see (1) below), there must be something else motivating the movement. (1) a.
Ko šta gdje kupuje? who what where buys ‘Who buys what where?’ b. *Ko kupuje šta gdje? c. *Ko šta kupuje gdje? d. *Ko gdje kupuje šta?
Another piece of evidence arguing for the fact that all wh-fronting in SerboCroatian is not wh-movement comes from echo questions. Wh-phrases in Serbo-Croatian cannot stay in situ in echo questions (cf. (2)). Given that these types of questions do not presumably involve checking of the +wh-feature, the obligatory fronting of wh-phrases must be motivated by something other than wh-movement.4 (2) ?*Jovan kupuje ŠTA? Jovan buys what ‘Jovan buys what?’
2.2 Superiority A curious property of focus fronting in Serbo-Croatian is that it does not exhibit Superiority effects. The grammaticality of the examples below shows that wh-phrases are freely ordered in this language: (3) Ko je šta kupio? who is what bought ‘Who bought what?’ (4) Šta je ko kupio?
Bulgarian, another language Boškovic´ (1999) analyzes in depth, differs from Serbo-Croatian in this respect. As shown in (5) and (6), Bulgarian conforms to the Superiority Condition.5
190 Lara Reglero
(5) Koj kakvo e kupil? who what is bought ‘Who bought what? (6) *Kakvo koj e kupil?
Boškovic´ (1999) accounts for the Bulgarian pattern in the following way. First, he argues that Bulgarian, in contrast to Serbo-Croatian, has wh-movement in these constructions. In order to capture this fact, Boškovic´ (1999) adopts the Economy account of Superiority which requires that the +wh-feature of C be checked in the most economical way (i.e. through the shortest movement possible). Under this account, the Nominative wh-phrase koj must move to SpecCP before the Accusative wh-phrase kakvo since this is the most economical way to check the +wh-feature of C.6 (6) is ruled out since the movement of kakvo is less economical, that is, it results in a longer link. In this respect, Bulgarian behaves exactly like English. To put it clearer, both languages exhibit Superiority effects. From the evidence collected so far, it seems that wh-movement is sensitive to Superiority but focus movement is not (cf. Boškovic´ 1999). The Bulgarian data raise several questions. If Bulgarian wh-fronting is an instance of wh-movement, why must all wh-phrases obligatorily move to the front of the sentence? (7) *Koj e kupil kakvo? who is bought what ‘Who bought what?
Boškovic´ (1999) takes this fact as evidence that focus movement is also involved in Bulgarian. More precisely, movement of one wh-phrase should suffice to check the strong +wh-feature of C. Wh-fronting of the remaining wh-phrases is an instance of focus movement. This analysis makes a prediction. If wh-movement, which is subject to the Superiority Condition, affects only one wh-phrase, then the movement of the highest wh-phrase would satisfy the requirement that the +wh-feature of C be checked in the most economical way. If focus movement, which is not subject to the Superiority Condition (cf. (3) and (4)), is responsible for the fronting of the rest of the wh-phrases, then one would expect these wh-phrases to be freely ordered. The prediction is borne out by the Bulgarian data in (10) and (11): (8) Kogo kakvo e pital Ivan? whom what is asked Ivan ‘Who did Ivan ask what?’
Non-wh-fronting in Basque
(9) *Kakvo kogo e pital Ivan? (10) Koj kogo kakvo e pital? who whom what is asked ‘Who asked whom what?’ (11) Koj kakvo kogo e pital?
As the examples above make clear, the Nominative wh-phrase koj moves first to check the strong +wh-feature of C. The second and third wh-phrases are freely ordered since they are subject only to focus movement. Boškovic´ (1999) accounts for these facts by modifying Chomsky’s (1995) Attract system. More specifically, Boškovic´ (1999) proposes that the attractor for wh-movement is an Attract-1F head (Attract-1F implies that the formal inadequacy of the attractor is overcome by attracting 1 feature F). This entails that given two potential attractees, the Attract-1F head will always attract the highest wh-phrase. Attract-1F heads give us Superiority effects. In contrast, Boškovic´ (1999) proposes that the attractor for focus movement in an Attractall-F head (an Attract-all-F head is a head that has a formal inadequacy that can only be overcome by attracting all features F). Superiority effects are not expected in this case since the order in which wh-phrases move to the relevant head yield equally economical derivations. To put it clearer, the derivation in which kogo (cf. (10)) moves first to the attracting head is equally economical as the derivation in which kakvo moves first (cf. (11)). Since the same number of nodes are crossed in both cases to satisfy the relevant Attract-All property, both derivations yield equally economical outputs. The Bulgarian pattern receives the following analysis in Boškovic´’s (1999) system. According to Boškovic´ (1999), C in Bulgarian has two features: Attract1F +wh-feature and Attract-all-F +focus feature. Once C enters the derivation, the movement of the highest wh-phrase satisfies the formal inadequacies of C. After the highest wh-phrase moves, the order of movement of the remaining wh-phrases is free since focus movement is not subject to Superiority. To summarize, wh-movement is subject to Superiority because the relevant head has an Attract-1F feature. Focus movement does not exhibit Superiority effects because focus movement has the Attract-all-F property. The discussion above makes a further prediction. Whenever Superiority effects show up, wh-movement is involved in the derivation. The data from Bulgarian and English support this claim. In those languages where the Superiority Condition is operative, wh-movement is taking place. Since Serbo-Croatian wh-phrases are freely ordered, wh-movement is not responsible for the overt
191
192 Lara Reglero
fronting of the wh-phrases. This state of affairs becomes relevant for certain constructions in Serbo-Croatian. As Boškovic´ (2002) shows, Serbo-Croatian exhibits Superiority effects in some contexts. More precisely, the Superiority Condition is operative in embedded, long-distance and overt C questions: (12) a.
[Ko koga voli], taj o njemu i govori. [who whom loves that-one about him even talks ‘Everyone talks about the person they love.’ b. ?*[Koga ko voli], taj o njemu/o njemu taj i govori.
(13) a. ?Ko koga tvrdiš da je istukao? who whom claim that is beaten ‘Who do you claim beat whom?’ b. *Koga ko tvrdiš da je istukao? (14) a.
Ko li koga voli? who C whom loves ‘Who on earth loves whom?’ b. *Koga li ko voli?
In order to account for these facts, Boškovic´ (2002) establishes a parallelism between Serbo-Croatian and French. The core idea of this proposal is that SerboCroatian exhibits Superiority effects in those contexts in which French must have wh-movement, which involve embedded, long-distance and overt C questions. To put it another way, Serbo-Croatian has real wh-movement when French has it. If this is so, then we have an explanation for the facts in (12) through (14). SerboCroatian wh-phrases are subject to ordering constraints in these examples because wh-movement, crucially not just pure focus movement, is taking place here.7 2.3 Interpretation of questions Whether wh-phrases undergo focus or wh-movement has consequences for the interpretation of questions. As Boškovic´ (1998a, 1999, 2002) points out, the English question in (15) obligatorily requires a pair-list answer: (15) Who bought what?
(15) cannot receive a single-pair answer. In other words, (15) cannot be felicitously uttered in the following situation: “John is in a store and off in the distance sees somebody buying an article of clothing, but does not see who it is and does not see what exactly the person is buying. He goes to the shopassistant and asks (15).”
Non-wh-fronting in Basque 193
Curiously, the Japanese and Chinese counterparts of (15) can receive not only a pair-list answer but also a single-pair answer. To put it clearer, the Japanese question in (16) can be asked in the situation described above. (16) Dare-ga nani-o katta no? who-nom what-acc bought q ‘Who bought what?’
One crucial difference between English and Japanese/Chinese is that wh-phrases in English move overtly to SpecCP. In contrast, wh-phrases in Japanese and Chinese stay in situ and consequently do not move overtly to SpecCP. Given this difference, it could be the case that filling SpecCP overtly forces the pair-list interpretation for some reason. As Boškovic´ (1998a, 1999, 2002) shows, French confirms this conjecture. French allows the in situ and the wh-movement strategy, as shown in (17) and (18). Crucially, single-pair answers are only allowed with the in situ strategy (cf. (17)). This state of affairs provides strong evidence for the claim that single-pair answers are only allowed when SpecCP is not filled in the overt syntax. (17) Il a donné quoi à qui? he has given what to who ‘What did he give to who?’ (18) Qu’a-t-il donné à qui?
If we now turn to the interpretation of multiple questions in Bulgarian and Serbo-Croatian, we expect the following: Bulgarian, a language in which SpecCP is filled overtly, should only allow a pair-list answer. In this respect, Bulgarian would pattern with English. In contrast, Serbo-Croatian, a language in which wh-phrases do not move to SpecCP overtly, should allow single-pair answers (note that the claim that wh-phrases in Serbo-Croatian do not move to SpecCP overtly goes against Rudin’s (1988) original proposal in which the first wh-phrase in Serbo-Croatian wh-questions is located in SpecCP). SerboCroatian would pattern with Japanese and Chinese in the relevant respect. The predictions are borne out. As Boškovic´ shows, the Serbo-Croatian multiple question in (3) (repeated as (19)) allows both a pair-list and a single-pair answer. In contrast, the Bulgarian counterpart in (5) (repeated as (20)) requires a pair-list answer: (19) Ko je šta kupio? who is what bought ‘Who bought what?’
194 Lara Reglero
(20) Koj kakvo e kupil? who what is bought ‘Who bought what?
This analysis makes the following prediction: when wh-movement takes place overtly in Serbo-Croatian, single-pair answers should be disallowed since SpecCP would be filled in the overt syntax. The topic construction is one such context. As Boškovic´ (2002) shows, such constructions in Serbo-Croatian exhibit Superiority effects (cf. (21)), which we have taken to be a diagnostic that wh-movement has taken place (see Boškovic´ (2002)) for an explanation why wh-movement takes place in this construction). If wh-movement is indeed taking place here, single-pair answers should be disallowed in (21a). The prediction is borne out. (21a) requires a pair-list answer. (21) a.
Tom ˇcoveku, ko je šta poklonio? that man who is what bestowed ‘To that man, who bestowed what?’ b. ??Tom ˇcoveku, šta je ko poklonio?
In conclusion, whenever wh-movement takes place, that is, whenever SpecCP is filled overtly, single-pair answers are disallowed. If only focus movement is involved, single-pair answers are allowed since the specifier of CP is not filled overtly in these cases. In this section I have summarized the core ideas of Boškovic´’s system. In the next section I introduce the data to be accounted for. In my attempt to offer an analysis, I will use the tools provided by Boškovic´ and I will evaluate whether his system can capture the Basque data.
3. The data Basque has two different strategies to ask a question such as Who bought what?. One possibility is that one wh-phrase is fronted and the other stays in situ, as in (22). The other possibility is that both wh-phrases move overtly to the beginning of the sentence, as in (23):8,9 (22) Nork erosi du zer? who-erg buy aux what-abs ‘Who bought what?’ (23) Nork zer erosi du?
Non-wh-fronting in Basque 195
Wh-phrases exhibit Superiority effects, as shown below:10,11 (24) *Zer erosi du nork? what-abs buy aux who-erg *‘What did who buy?’ (25) *Zer nork erosi du?
4. Focus in Basque In this section we will see that Basque wh-fronting is an instance of focus movement. In this respect, Basque exhibits the same behavior as Serbo-Croatian. Before going into the details of the account, I will present some basic facts about word order in Basque which will be crucially important for the analysis I propose in this paper. 4.1 Word order Basque is an SOV language (de Rijk 1969, Eguzkitza 1986, Ortiz de Urbina 1989 among others) with very flexible word order. All the word orders in (26) through (31) are allowed but only (26) is pragmatically neutral. (26) Jonek Miren ikusi zuen. Jon-erg Miren-abs see aux ‘Jon saw Miren.’ (27) Miren ikusi zuen Jonek. (28) Jonek ikusi zuen Miren. (29) Miren Jonek ikusi zuen. (30) Ikusi zuen Jonek Miren. (31) Ikusi zuen Miren Jonek.
(Ortiz de Urbina 1995)
The sentence in (26) would be the most natural way to answer a question such as What happened?. This implies that in a sentence where everything is new information the most appropriate word order is SOV. Similarly, in sentences with two objects the pragmatically neutral word order is S IO DO V, as in (32): (32) Jonek Mireni muxu bat eman zion Jon-erg Miren-dat kiss one give aux ‘Jon kissed Mary.’
196 Lara Reglero
In (27) and (28) the element to the left of the verb is interpreted as focus. Miren in (27) and Jonek in (28) bear contrastive stress and receive a focus interpretation. As Elordieta (2001) points out, postverbal elements in these types of sentences express given information. (29) exhibits a slightly different pattern. As in (27) and (28), the element preceding the verb, that is Jonek, is interpreted as focus. Miren in this sentence is separated from the focalized Jonek by a pause and receives a topic interpretation.12 Finally, the verb itself is focalized in (30) and (31). Note that the auxiliary zuen cannot be initial in (26) through (32). 4.2 Focus in Basque As mentioned above, focalized elements land in a position immediately adjacent to the verb. This observation has a long tradition which goes back at least to Altube (1929). More recently, de Rijk (1978), Eguzkitza (1986), Martin Callejo (1984) and Ortiz de Urbina (1989, 1995, 1999a) among others have investigated this phenomenon in depth (Basque grammarians call the focus position galdegaia, which literally means ‘the subject of question’ (galde ‘ask’, gaia ‘subject’)). Ortiz de Urbina (1999a), for example, has paid special attention to the similarities between focus movement and wh-movement. These similarities, to be explained below, led him to conclude that foci and wh-words undergo the same type of movement and that both types of elements land in the same position; namely, SpecFocP (see Ortiz de Urbina 1999b for the latter conclusion). Ortiz de Urbina (1999a) points out the following distributional similarities between foci and wh-words. To begin with, both occur in clause initial position and are immediately followed by the verb.13 Compare (28) with (33) below: (33) Nork ikusi zuen Miren? who-erg see aux Miren-abs ‘Who saw Miren?’
The ungrammaticality of (34) and (35) shows the adjacency between foci/ wh-words and the verb:14 (34) *JONEK Miren ikusi zuen. Jon-erg Miren-abs see aux ‘It is Jon that saw Miren.’ (35) *Nork Miren ikusi zuen? who-erg Miren-abs see aux ‘Who saw Miren?’
Non-wh-fronting in Basque 197
Second, both foci and wh-words are optionally preceded by topics. Compare (36) with (29) above: (36) Miren, nork ikusi zuen? Miren-abs who-erg see aux ‘As for Miren, who saw her?’
As Ortiz de Urbina (1999b) points out, a sentence may contain multiple topics. These topics are freely permutable and receive a listing intonation. (37) Atzo /Mirenekin /Jon nora /ETXERA joan zen. yesterday/Miren.with/Jon where/home.to go aux ‘Where did John go yesterday with Miren?’ ‘Yesterday, with Miren, John went HOME.’
Third, foci, like wh-words, can undergo cyclic movement with bridge verbs: (38) JONEK uste dut [t esan du-ela Mikelek [t idatzi du-ela eskutitza]. Jon-erg think aux say aux-comp Mikel-erg write aux-c letter ‘It is Jon that I think Mikel has said has written the letter.’ (39) Nork uste duzu [t esan du-ela Mikelek [t idatzi du-ela who think aux say aux-comp Mikel-erg write aux-comp eskutitza]]? letter ‘Who do you think Mikel has said has written the letter?’
Fourth, focal operators and interrogative operators behave in the same way in pied-piping configurations. Interrogative operator features can percolate in these configurations. Similarly, focal operators can also percolate in the same types of constructions. (40) [JONEN lagunek] idatzi zuten eskutitza. [Jon’s friends-erg write aux letter-abs ‘JON’s friends wrote the letter.’ (41) *[JONEN lagunek eskutitza] idatzi zuten. (42) [Noren lagunek] idatzi zuten eskutitza? [whose friends-erg write aux letter-abs ‘Whose friends wrote the letter?’ (43) *[Noren lagunek eskutitza] idatzi zuten?
A parallel behavior is also observed with clausal pied-piping constructions:
198 Lara Reglero
(44) [JONEK idatzi du-ela liburua] esan du Peiok [Jon-erg write aux-comp book-abs say aux Peio-erg ‘Peio said that JON wrote the book.’ (‘That JON wrote the book has Peio said.’) (45) [Nork idatzi du-ela liburua] esan du Peiok? [who-erg write aux-comp book-abs say aux Peio-erg ‘Who wrote the book has Peio said?’ (46) [[JONEK idatzi du-en] liburuak] izan ditu salmenta onak. [[Jon-erg write aux-comp books-abs be aux sale good ‘The book that JON has written sold well.’ (47) [[Nork idatzi du-en] liburuak] izan ditu salmenta onak? [[who-erg write aux-comp books-abs be aux sale good ‘The book that who wrote had good sales?’
The ungrammaticality of (48) suggests that focus and wh-elements compete for the same position. This pattern is observed both in matrix (cf. (48)) and in embedded sentences (cf. (49)): (48) a. *Nork MIREN ikusi du antzokian? who-erg Miren-abs see aux theater-at ‘Who saw MIREN at the theater?’ b. *MIREN nork ikusi du antzokian? (49) Galdetu didate (*JONEK) zer (*JONEK) erosi duen. ask aux (*Jon-erg what-abs (*Jon-erg buy aux-comp ‘They have asked me what JOHN bought.’
From the discussion above, we can draw the following conclusions: foci and wh-words undergo the same type of movement and occupy the same structural position; namely, the position immediately adjacent to the verb. Given the parallel behavior between foci and wh-words, it seems reasonable to conclude that wh-fronting in Basque is not the outcome of wh-movement but of focus movement (Ortiz de Urbina 1999a, Eguzkitza 1986 etc).15 In this respect, wh-words in Basque exhibit a parallel behavior to those in Serbo-Croatian. Further evidence supporting the idea that wh-words in Basque are inherently focused comes from echo questions. Similarly to Serbo-Croatian, wh-words in Basque move overtly to the beginning of the sentence even in questions where the +wh-feature is not present. Given the fact that wh-words in Basque move overtly even in questions with an echo interpretation, we can conclude that the fronting of wh-phrases in Basque is independent of the +wh-feature.
Non-wh-fronting in Basque 199
(50) below is a representative example: (50) A: Zugandik atera dira kontu zikin guzti horiek. you-from come aux stories dirty all those ‘All those dirty stories have come from you.’ B: a. Nigandik ZER atera dela? me-from what-abs come aux-comp ‘(That) what has come from me?’ b. *Nigandik atera dela ZER? (Etxepare & Ortiz de Urbina, in press)
Given the previous empirical evidence, I will assume in the remainder of this paper that wh-phrases in Basque move overtly to check a focus feature.
5. The wh-in situ strategy With all the background provided in Sections 2 and 4, we are now ready to discuss the Basque data from Section 3. I will first provide an analysis for the Basque sentences in which one wh-phrase moves to the beginning of the sentence and the other wh-phrase stays in situ. A representative example is (22), repeated as (51): (51) Nork erosi du zer? who-erg buy aux what-abs ‘Who bought what?’
I will argue that in sentences such as (51) the fronted wh-phrase is focalized whereas the in situ wh-element is D-linked. My last claim, that is, that in situ wh-elements are D-linked, is reminiscent of the behavior of in situ wh-phrases in Slavic. As Boškovic´ (2002) shows, non-D-linked wh-phrases in SerboCroatian must be fronted obligatorily. In contrast, D-linked wh-phrases can remain in situ, as in (52): (52) Ko je kupio koju knjigu? who is bought which book ‘Who bought which book?’
In this respect, Basque and Slavic exhibit a parallel behavior, as will be shown below. In order to support my claims empirically, I will consider Pesetsky’s (1987) the hell test. As Pesetsky (1987) points out, there are elements which are
200 Lara Reglero
good candidates for “aggressively non-D-linked” wh-phrases. In English examples such as (54), there is a conflict between aggressively D-linked which and aggressively non-D-linked the hell. The conflict does not arise in (53) since the nature of what is different from that of which: (53) What the hell book did you read that in? (54) *Which the hell book did you read that in?
To put it clearer, the hell is a wh-modifier which expresses complete ignorance. As den Dikken and Giannakidou (2001, 2002) indicate, when the hell combines with a wh-word, the speaker does not know what the value of the wh-word will be. Since the range of reference of D-linked wh-phrases is discourse-given, any attempt to attach an element which stands for ‘non-givenness’ (i.e. the hell) to an element which is inherently D-linked, will result in an ungrammatical sentence (cf. (54)). (53) is perfectly grammatical because what is not inherently D-linked and is therefore free to combine with the “aggressively non-D-linked” the hell. This type of test helps us differentiate between those wh-phrases which are D-linked from those which are not. Now, I will apply the same line of reasoning to some Basque examples: (55) Nor arraiok erosi du zer? who hell-erg buy aux what-abs ‘Who the hell bought what?’ (56) *Nork erosi du zer arraio? who-erg buy aux what hell-abs ‘Who bought what the hell?’
The grammaticality of (55) indicates that nork is not D-linked in this sentence since it is able to combine with the “aggressively non-D-linked” arraiok (the exact translation of arraio is ‘lighting’. I will translate it as hell in the text for ease of exposure). In contrast, (56) is ungrammatical suggesting that the in situ wh-phrase zer must be D-linked. In summary, Pesetsky’s (1987) the hell test indicates that the fronted wh-phrase in (51) is non-D-linked whereas the in situ wh-phrase is D-linked. As Boškovic´ (2002) points out, only non-D-linked wh-phrases in Slavic undergo focus movement. This correlation is fairly intuitive in the sense that non-D-linked wh-phrases do not refer to previously mentioned or contextually salient referents. Focus movement implies the notion of ‘new information’, something ‘not given’. Therefore, the semantic nature of non-D-linked wh-phrases allows them to undergo focus movement. D-linked elements, on the
Non-wh-fronting in Basque 201
other hand, have very different semantics from non-D-linked wh-phrases. As Pesetsky (1987) points out, when a speaker asks a question like Which book did you read? the range of felicitous answers is limited by a set of books both speaker and hearer have in mind. Therefore, the reference of D-linked wh-phrases such as which man is discourse given. This ‘discourse givenness’ property is reminiscent of the notion ‘topic’. As mentioned in note 12, “the topic is said to be the introductory part of a sentence foregrounding known information” (É. Kiss 1981: 185). Topics refer to old, given information, something that is not knew. Given the close connection between D-linking and Topicalization, I will claim that being D-linked is equivalent to being topicalized. This idea is not new. For extensive discussion of the similarities between D-linking and Topicalization, I refer the reader to Grohmann (1998). The basic idea is that D-linking in the minimalist framework correlates with some feature. Since being D-linked implies being ‘given’ and since being a topic also implies being ‘given’, it is reasonable to conclude that the feature D-linking correlates with is a topic feature. This is Grohmann’s (1998) conclusion with respect to the behavior of multiple questions in German. So far I have provided evidence indicating that nork in (51) undergoes focus movement by virtue of its non-D-linked nature. Zer, in contrast, is D-linked or topicalized. Since the range of reference of D-linked elements is discourse given, elements such as zer in (51) are not inherently focused and therefore do not undergo focus movement. 5.1 Predictions My analysis makes the following predictions. First, if the in situ wh-phrase zer is D-linked or topicalized in (51), one would expect inherently D-linked wh-phrases to appear in that position. The prediction is borne out, as in (57): (57) Nork erosi zuen zein liburu? who-erg buy aux which book-abs ‘Who bought which book?’
Furthermore, if the Ergative wh-phrase nork in that very sentence occupies a position for focalized elements, then we should expect inherently D-linked wh-phrases to be disallowed in preverbal position. The prediction is apparently not borne out (cf. (58a)):
202 Lara Reglero
(58) a.
Zein ikaslek erosi zuen zer? which student-erg buy aux what-abs ‘Which student bought what?’ b. *Zer erosi zuen zein ikaslek?
This problem does not only arise in multiple questions. In a simpler sentence such as (59a), the position corresponding to focalized elements is occupied by a D-linked wh-phrase: (59) a.
Zein liburu erosi du Jonek? which book-abs buy aux Jon-erg ‘Which book did Jon buy?’ b. *Jonek erosi du zein liburu?
From the evidence presented in the sentences above, it seems that inherently D-linked phrases are allowed in sentence initial position, contrary to our predictions. However, the facts are a bit more intricate than (58a) and (59a) show. Etxepare and Ortiz de Urbina (in press) notice that, in some circumstances, parentheticals can interrupt the sequence wh-phrase Verb. A representative example is given below: (60) Zein idazle, gaurko edo denbora bateko, iruditzen zaizu gidaririk which writer-abs today.from or time other seem aux guide zuzenena hitz kontuan? best word in-terms ‘Which writer, from today or from other times, seems to you the best guide in terms of words?’
What is curious in this example is that the wh-phrase which is separated from the verb is inherently D-linked. A question arises at this point: would we obtain the same grammaticality judgment if we substituted zein ikasle for a nonD-linked wh-phrase like nor? As (61) shows, this possibility is not allowed in Basque (note that from now on, I will assume non-D-linked contexts for noninherently-D-linked wh-phrases): (61) *Nor, gaurko edo denbora bateko, iruditzen zaizu gidaririk who-abs today-from or time other seem aux guide zuzenena hitz kontuan? best word in-terms ‘Who, from today or from other times, seems to you the best guide in terms of words?’
Non-wh-fronting in Basque 203
This discussion partially resembles some Bulgarian facts. As Boškovic´ (2002) shows, a parenthetical can more easily intervene between two fronted wh-phrases if the second one is D-linked. The contrast in (62) and (63) exemplifies this point. (62) *?Koj, spored tebe, kakvo e kupil? who according to.you what is bought ‘Who, according to you, bought what?’ tebe, koja kniga e kupil? (63) ?Koj, spored who according to.you which book is bought ‘Who, according to you, bought which book?
Given this state of affairs, Boškovic´ (2002) concludes that kakvo and koja kniga do not land in the same position. As he puts it, koja kniga in (63) does not undergo focus movement and lands in a position below CP. Basque behaves like Bulgarian in the sense that fronted D-linked wh-phrases seem to land in a different position from that of non-D-linked wh-phrases. If this were not the case, there should be no reason for different grammaticality judgments between the two types of wh-phrases. I would therefore like to suggest that inherently D-linked wh-phrases in Basque do not land in the same position as non-D-linked wh-phrases. Given the contrast in grammaticality between (60) and (61), I will assume that D-linked wh-phrases are higher than the position where non-D-linked wh-phrases move to. Presumably, non-D-linked wh-phrases land in SpecFocP (cf. Ortiz de Urbina 1999b) while inherently D-linked wh-phrases land in the specifier of a TopP, placed above FocP (cf. Rizzi (1997)). Leaving aside the precise nature of the positions where non-D-linked wh-phrases and inherently wh-phrases move to, the crucial point of my discussion is that the two types of wh-phrases land in different positions. My analysis makes another prediction. If the in situ position in sentences such as (51) is the position where D-linked or topicalized elements are located, then we should expect those elements which are incapable of receiving a D-linked interpretation to be disallowed in the in situ position. É. Kiss (1993) treats wh-phrases such as how and why differently from wh-phrases of the form which applicant (see Pesetsky (1987) for an earlier discussion on the behavior of why). According to her, how and why cannot be D-linked, or in her terminology, how and why are incapable of specificity. If this is correct, then we should expect how and why to be disallowed in the ‘in situ’ position. The prediction is borne out:
204 Lara Reglero
(64) *Nori esplikatu dio Jonek ariketa nola? who-dat explain aux Jon-erg exercise-abs how ‘To whom did Jon explain the exercise how?’ (65) *Nork lapurtu ditu bitxiak zergatik? who-erg steal aux jewels-abs why ‘Who stole the jewels why?’
Notice that nola and zergatik are allowed in preverbal position. This is expected since the preverbal position is the position where focalized elements move to. If nola and zergatik cannot be D-linked, then, we should expect those two wh-phrases to be allowed in a position where D-linking is not playing a role; namely, the preverbal position: (66) Nola esplikatu dio ariketa Jonek nori? how explain aux exercise-abs Jon-erg who-dat ‘How did Jon explain the exercise to whom?’ (67) ?Zergatik lapurtu ditu bitxiak nork? why steal aux jewels-abs who-erg ‘Why did who steal the jewels?’
5.2 Landing sites So far I have presented evidence that the preverbal wh-phrase is focalized whereas the in situ wh-phrase is D-linked or topicalized. A question arises at this point: where are the wh-phrases located in the structure? Before offering an answer to this question, let me show some data which will become relevant in our search for the landing positions of wh-phrases. At first sight, one would expect in situ wh-phrases to occupy the same position as their non-wh-counterparts. This conclusion is supported by sentences such as (68) and (69): (68) Nork erosi du zer? who-erg buy aux what-abs ‘Who bought what?’ (69) Nork erosi du liburua? who-erg buy aux book-abs ‘Who bought the book?’
Zer in (68) seems to occupy the same position as liburua in (69). However, this parallelism can no longer be maintained if more data are carefully examined.
Non-wh-fronting in Basque 205
(70) Nork eman dio muxua sutsuki Mireni? who-erg give aux kiss passionately Miren-dat ‘Who kissed Miren passionately?’ (71) a. *Nork eman dio muxua sutsuki nori? who-erg give aux kiss passionately who-dat ‘Who kissed whom passionately?’ b. Nork eman dio nori muxua sutsuki?
The data above suggest that “wh-phrases-in situ” are structurally higher than their non-wh counterparts. As (70) makes clear, the object Mireni can appear after the manner adverb sutsuki. In contrast, nori cannot appear after the adverb, as in (71a). The sentence is only grammatical when the wh-phrase appears higher in the structure (cf. (71b)) (nori can marginally appear after muxua in (71b). However, the sentence is clearly more degraded, ranging from ‘??’ to a ‘*’ depending on the speaker). Note that the same grammaticality judgment obtains if we substitute nori for an inherently D-linked wh-phrase such as zein neskari ‘to which girl’. This is expected since, under my analysis, in situ wh-phrases in Basque are D-linked. (72) a. *Nork eman dio muxua sutsuki zein neskari? who-erg give aux kiss passionately which girl-dat ‘Who kissed passionately which girl?’ b. Nork eman dio zein neskari muxua sutsuki?
In order to account for this pattern, I will suggest the following: wh-phrases “in situ” have a topic feature which must be checked overtly. This idea has its origins in Grohmann’s (1998) work. As I noted above, Grohmann (1998) suggests that wh-phrases in German are D-linked. This D-linking property is syntactically encoded on the wh-phrase by means of a topic feature which must be checked overtly in a spec-head relation. He proposes that wh-phrases in German, by virtue of being D-linked, undergo overt movement to the specifier of a TopP where the topic feature the wh-phrase carries can be formally checked. (71a) is then ungrammatical because the wh-phrase nori has failed to be attracted and therefore its topic feature remains unchecked. Once overt movement takes place, the output we obtain is grammatical (cf. (71b)). Since non-wh-objects do not have to check a topic feature, they can remain in their base position. Note that the non-wh-object Mireni in (70) can also appear before the adverb, as in (73):
206 Lara Reglero
(73) Nork eman dio Mireni muxua sutsuki? who-erg give aux Miren-dat kiss-abs passionately ‘Who kissed Miren passionately?’
The crucial difference between (73) and (71b) is that the movement of the wh-phrase in (71b) is obligatory. In contrast, Mireni undergoes it optionally. I will treat cases where the object moves overtly (cf. (73)) as instances of scrambling. According to Uriagereka (1999), there are no detailed analyses of scrambling in Basque. Given this state of affairs I will not discuss the exact mechanism which is prompting the movement of Mireni in (73). Suffice it to say that the movement of nori is obligatory whereas the movement of Mireni is optional (i.e. ‘scrambling’). Given this difference, I will assume that each movement is instantiated in a different way. At this point we are ready to investigate the landing positions of the wh-phrases in sentences such as (51). In order to do so, I will use data with adverbs. According to Watanabe (1993) and Boškovic´ (1997d), manner adverbs are VP adjoined and sentential adverbs are TP adjoined (see Jackendoff (1972) for an earlier version). Those adverbs which are ambiguous between the two interpretations allow both adjunction possibilities. Taking these facts into account let us investigate the position wh-phrases move to in the syntax. Example (74), where the sentential adverb atzo is placed after the verb, suggests that the preverbal wh-phrase and the verb must be higher than TP.16 (75) shows that the position of the adverb in (74) is the highest position the sentential adverb can be located in.17 (74) Nork eman zion atzo liburua Mireni? who-erg give aux yesterday book-abs Miren-dat ‘Who gave the book to Miren yesterday?’ (75) *Nork atzo eman dio liburua Mireni?
I will argue that the preverbal wh-phrase does not land in SpecCP. Following Ortiz de Urbina (1999b), I assume that the preverbal wh-phrase lands in the specifier of FocP. Further evidence for this idea comes from the interpretation of questions. As Boškovic´ (1999, 2002) shows (see the discussion above), singlepair answers are only allowed when SpecCP is not filled overtly by a wh-phrase. In Basque the most salient reading for a question such as (51) is a pair-list reading. However, one of my informants allows a single-pair answer for this question.18 I take this to suggest that wh-phrases do not move overtly to SpecCP in Basque. Note that Boškovic´ (2002) points out that not filling SpecCP overtly
Non-wh-fronting in Basque 207
is necessary but not sufficient for licensing single-pair answers. In other words, a language may not fill SpecCP overtly and still only allow pair-list answers (this follows under the analysis of the phenomenon under consideration presented in Boškovic´ (1998a)). Given this fact and given that one of my informants accepts single-pair answers, I will assume that preverbal wh-phrases in Basque do not move overtly to SpecCP but remain in a lower position, namely SpecFocP. Now let me offer some data which can help us figure out the position of the postverbal wh-phrase (from now on I will refer to ‘in situ’ wh-phrases as ‘postverbal wh-phrases’. Since we have found that in situ wh-phrases are not really in situ, I will change the terminology to make it more intuitive). As the data in (70) and (71) above made clear, the postverbal wh-phrase cannot stay inside the VP. If manner adverbs are adjoined to VP and nori cannot appear after the manner adverb sutsuki then we can safely conclude that nori must be higher than VP. The relevant example is repeated below: (76) *Nork eman dio muxua sutsuki nori? who-erg give aux kiss passionately who-dat ‘Who kissed whom passionately?
(= (71a))
Sentential adverbs indicate that the postverbal wh-phrase must be higher than TP. Atzo ‘yesterday’ in (77)–(78) is only acceptable when placed right after the postverbal wh-phrase. As (78) shows, zer cannot appear after the sentential adverb, indicating that zer cannot be lower than TP. The grammaticality of (77) gives further support to this idea.19 (77) ?Nork eman zion zer atzo Mireni? who-erg give aux what-abs yesterday Miren-dat ‘Who gave what to Miren yesterday?’ (78) ?*Nork eman zion atzo zer Mireni?
Data with ambiguous adverbs give further support to the claim that the postverbal wh-phrase must be higher than TP. In the examples below, the manner reading of zuhurki is possible in (80) but not in (79), indicating that nori must be higher than VP. Similarly, the sentential reading of zuhurki is allowed in (80) but not in (79). This state of affairs suggests that the postverbal wh-phrase nori is located above TP: (79) *Nork aholkatu zion zuhurki nori? who-erg advise aux wisely who-dat ‘Who advised wisely whom?’
208 Lara Reglero
(80) Nork aholkatu zion nori zuhurki?
Given the above discussion, I assume that the projection the postverbal wh-phrase lands in is TopP.20 5.3 Derivation In this section I will present the derivation for (51) and I will offer an explanation for the Superiority effects that this construction exhibits (cf. (24)). The derivation I am proposing in this paper is shown in (81):21,22 FocP
(81) nork who
Foc¢ Foc
AgrSP
erosi du tnork buy aux
AgrS¢ TopP
zer what
AgrS Top¢
AgrOP tzer
Top AgrO¢
VP NP tnork NP [+focus] tzer [+topic]
tv
tv AgrO
V¢
tv V tv
As the tree above shows, the derivation starts with the Ergative wh-phrase nork in the specifier of VP and the object wh-phrase zer as the complement of the verb erosi du. AgrOP enters the derivation and zer raises overtly to check its Absolutive Case.23 I will assume that the projection where zer lands is the next
Non-wh-fronting in Basque 209
projection that is inserted into the tree. The label of this projection is TopP since I am assuming that wh-phrases move overtly to check a topic feature. The crucial point is that when the head of that projection enters the derivation, it attracts the highest wh-phrase. In this case zer is higher than nork. Zer therefore moves to SpecTopP, checking a topic feature. Then, AgrS enters the tree and nork moves overtly for Case-checking purposes. Finally, FocP enters the derivation and nork, being the closest wh-phrase, is attracted. Superiority effects can be now easily accounted for. When the relevant attraction operation takes place, that is, when TopP enters the derivation, zer is higher in the structure. If we follow the Economy account of Superiority, moving nork over zer would result in a longer link.24 The relevant step is provided below: (82)
TopP Top¢ AgrOP zer
Top AgrO¢
AgrO
VP NP nork
tv
V¢ NP tzer
tv V tv
6. Multiple wh-fronting In this section I will investigate the properties of constructions where two wh-phrases have undergone movement to the beginning of the sentence. The relevant example is given in (83): (83) Nork zer erosi du? who-erg what-abs buy aux ‘Who bought what?’
(= (23))
210 Lara Reglero
I will argue that zer is focalized and nork is functioning as a topic. The claim that zer is focalized in (83) is not surprising given the evidence discussed above that those elements immediately to the left of the verb are focalized in Basque. As pointed out by Ortiz de Urbina (1999a), among others, any element preceding a wh-word functions as a topic in Basque. If liburua in (84) is undergoing topic movement, it is reasonable to say that nork in (83) is undergoing the same type of movement: (84) Liburua nork erosi du? book-abs who-erg buy aux ‘As for the book, who bought it?’
Moreover, nork behaves like a D-linked or topicalized element with respect to Pesetsky’s (1987) the hell test: (85) *Nor arraiok zer erosi du? who hell-erg what-abs buy aux ‘Who the hell bought what?’
In (85) the position nork occupies is the position of a D-linked or topicalized element. The sentence is ungrammatical because nork, being in a D-linked position, cannot combine with the aggressively non-D-linked arraio. We would now expect zer to be able to co-occur with arraio because zer is nonD-linked or focalized in my analysis. The prediction is borne out, as in (86): (86) Nork zer arraio erosi du? who-erg what-abs hell buy aux ‘Who bought what the hell?’
The grammaticality of (86) shows that zer can combine with the aggressively non-D-linked arraio suggesting that zer is non-D-linked in this example. 6.1 Predictions If nork is D-linked in (83) and zer is focalized, then we would expect inherently D-linked wh-phrases to be disallowed in the immediately preverbal position. In contrast, inherently D-linked elements should be allowed in the position nork occupies in the sentence under discussion. The predictions are borne out, as shown below:
Non-wh-fronting in Basque
(87) *?Nork zein liburu erosi zuen? who-erg which book-abs buy aux ‘Who bought which book?’ (88)(?)Zein ikaslek zer irakurri zuen? which student-erg what-abs read aux ‘Which student read what?’
(87) shows that inherently D-linked wh-phrases such as zein liburu are not allowed in the position where focalized elements land. This is expected given the incompatibility between D-linking and focus. The incompatibility does not arise in (88) because zein ikaslek, being inherently D-linked, can freely land in the position D-linked or topicalized elements move to. Example (88) is relevant regarding the discussion in Section 5.1. Recall that in that section I suggested that an example such as (58a) is grammatical because the preverbal inherently D-linked wh-phrase zein ikaslek does not land in the position corresponding to focalized elements but in a higher position in the structure, presumably in the specifier of a TopP. The ungrammaticality of (87) gives further support to this idea. In (87) there is a wh-phrase functioning as a topic and a second wh-phrase functioning as the focus of the sentence. Notice that inherently D-linked wh-phrases are allowed only in the position corresponding to the first wh-phrase; namely, nork (cf. (88)). If one tries to place them in the position corresponding to focalized elements, the output is ungrammatical, as in (87). Given these facts I will conclude that zein ikaslek in (89) lands in a position higher than the position occupied by focalized wh-phrases. (89) Zein ikaslek erosi zuen zer? which student-erg buy aux what-abs ‘Which student bought what?’
(= (58a))
My second prediction is as follows: those wh-phrases that were unable to stay in situ due to their impossibility to receive a D-linked interpretation should now be allowed to appear in the position immediately adjacent to the verb. The prediction is fulfilled as shown by the contrast in grammaticality between (90) and (91) and (92) and (93), respectively. (90) *Nori esplikatu dio Jonek ariketa nola? who-dat explain aux Jon-erg exercise how ‘Who did Jon explain the exercise how?’ (91) Nori nola esplikatu dio Jonek ariketa?
211
212 Lara Reglero
(92) *Nork lapurtu ditu bitxiak zergatik? who-erg steal aux jewels-abs why ‘Who stole the jewels why?’ (93) Nork zergatik lapurtu ditu bitxiak?
My analysis makes another prediction. Sentences such as (94)–(95) should be ruled out. In these constructions, elements which cannot receive a D-linked interpretation (i.e. nola ‘how’ and zergatik ‘why’) are placed in the position for D-linked elements. The ungrammaticality of the examples is thus expected under my analysis.25 (94) *Nola zer irakurri zuen Jonek? how what-abs read aux Jon-erg ‘How did John read what?’ (95) *Zergatik zer irakurri zuen Jonek? why what-abs read aux Jon-erg ‘Why did Jon read what?’
6.2 Derivation There are a number of questions that need to be answered before going into the details of the derivation of the Multiple wh-Fronting structure in (83). The first question I would like to raise is the following: are Topic and Focus licensed in the same projection? At first sight, it seems that they are licensed in different projections. This is so because the Topic–Subject–Verb order is possible (cf. (96)) whereas the order Focus–Subject–Verb is not, as shown in (97) (I assume that the subject is in SpecAgrSP): (96) Mireni, Jonek muxu bat eman zion. Miren-dat Jon-erg kiss one give aux ‘Miren, Jon kissed.’ (97) a. *MIRENI Jonek eman zion muxu bat. Miren-dat Jon-erg give aux kiss one ‘Jon has kissed MIREN.’ b. MIRENI eman zion Jonek muxu bat.
However, if Topic and Focus land in different projections, why can no material intervene between the two wh-phrases in (98)?
Non-wh-fronting in Basque
(98) *Nork horretaz/beraz /Joni zer erranen dio? who-erg on that /therefore/Jon-dat what-abs say aux ‘Who will say what to John/therefore/on that?’ (Etxepare & Ortiz de Urbina, in press)
Based on the ungrammaticality of (98), I assume that nork and zer are located in the same projection, with multiple specifier structures. At this point we are still left with some open questions: which head licenses the two features (i.e. Topic and Focus)? Why is the focused element lower in the structure? In order to answer the first question, let me offer a short digression. In examples such as (83), a single-pair answer is allowed. In Boškovic´’s (1999, 2002) system, the availability of single-pair answers is taken as evidence that SpecCP is not filled in the overt syntax. Since Basque allows single-pair answers in a multiple question such as (83), we can deduce that SpecCP is not filled overtly in Basque. Whatever head is licensing the Topic and Focus features, it must be below C. Data with sentential adverbs indicate that the head we are interested in must be at least higher than TP: (99) Nork zer eman zion atzo Mireni? who-erg what-abs give aux yesterday Miren-dat ‘Who gave what to Miren yesterday?’
It seems that whatever head is simultaneously licensing Topic and Focus must be below C but above TP. Furthermore, the head under discussion licenses both features at the same time. This point is important because Topic and Focus convey opposite communicative functions (i.e. Topic=old information and Focus=new information). Uriagereka (1995), Boeckx & Stjepanovic´ (1999) and Lambova (2001) have already dealt with this issue. Despite the differences, the three accounts basically argue for the presence of a functional projection capable of hosting discourse-related material. If this is possible, then there is no longer a problem in assuming that Topic and Focus can be licensed by the same head. Since Topic and Focus are both discourse-related, they can be hosted in the same projection. For the sake of exposition I will assume that both topicalized and focalized elements land in the specifier of ∆P (Lambova 2001). The projections suggested in Boeckx & Stjepanovic´ (1999) and Uriagereka (1995) are virtually equivalent to Lambova’s (2001) ∆P. The derivation I propose for (83) is as follows (I discuss the exact position of the verb below. At this point I place it in ∆ for ease of exposition):
213
214 Lara Reglero
(100)
DP D¢
nork who
D¢
zer what D
AgrSP AgrS¢
erosi du tnork buy aux AgrOP tzer
AgrS AgrO¢
VP NP
tv AgrO
V¢
tv
tnork NP
V
tzer
tv
The first steps of the derivation are similar to those in (81). We start with a VP which has nork as its specifier and which has zer as the complement of the verb erosi du. Nork is taken from the lexicon with a Topic feature. In contrast, zer is taken with a Focus feature. In the next two steps of the derivation, each wh-phrase raises overtly to check Case. In other words, once AgrOP enters the derivation, zer raises overtly to check Absolutive Case. Then, nork moves overtly to SpecAgrSP in order to get its Ergative Case checked. The relevant head enters the derivation now. Once ∆P shows up, the highest wh-phrase, that is nork, moves to the highest specifier and then the next wh-phrase, zer, moves to the lower specifier (note that I assume that TopP, which could be located below AgrSP, is only optionally present in the structure). If we attempt to move zer over nork, a Superiority violation would arise, as in (101): (101) *Zer nork erosi du? what-abs who-erg buy aux ‘What did who buy?’
(= (25))
What still needs to be explained is why we always end up with the order Topic Focus. In order to answer this question I will follow Lambova (2001),
Non-wh-fronting in Basque
who in turn follows Boškovic´ (2001), and suggest that the focus feature in the wh-phrase is a verbal PF affix and therefore must be adjacent to a verbal element, as shown in (97) above and (102) below.26 As a result, we always end up with the Topic–Focus–Verb order: (102) a. *Zer Mirenek eman zion Pellori? what-abs Miren-erg give aux Pello-dat ‘What did Miren give to Pello?’ b. Zer eman zion Mirenek Pellori?
Under this proposal, the adjacency between Focus and Verb is due to phonological rather than syntactic factors. Syntax can in principle yield either the Topic– Focus–Verb or the Focus–Topic–Verb order. However, if syntax derives the unacceptable order Focus–Topic–Verb, PF will rule out this derivation since the focus feature in the wh-phrase will fail to attach to its host; namely, the verb. As an alternative to the PF account, one could argue that a purely syntactic account can also account for the data in (102). Under a syntactic account, the verb eman zion raises to ∆ obligatorily. (102b) above is grammatical because V-to-∆ has taken place. (102a) is ruled out since the verb has not raised all the way to ∆. Both the PF account and the syntactic account discussed above account for the data in (102). However, there are additional data that show that the PF account is superior. Let us examine (96) and (97) (repeated as (103) and (104)) again in more detail. (103) Mireni, Jonek muxu bat eman zion. Miren-dat Jon-erg kiss one give aux ‘Miren, Jon kissed.’ (104) a. *MIRENI Jonek eman zion muxu bat. Miren-dat Jon-erg give aux kiss one ‘Jon has kissed MIREN.’ b. MIRENI eman zion Jonek muxu bat.
In (103) and (104a) the topic Mireni and the focus MIRENI are located in the same position; namely, Spec∆P. Under the syntactic account, the verb eman zion would raise to ∆ obligatorily. Consequently, there would be no extra space between the verb and the topic/focus. If this is so, the grammaticality of (103) remains unaccounted for. In other words, if Mireni is located in Spec∆P and the verb is in ∆, where is the extra material between the topic and verb located? It follows then that the verb is not located in the head of ∆P, otherwise (103) would be bad. More precisely, under the syntactic V to ∆ account of the
215
216 Lara Reglero
adjacency requirement, (103) should be bad for the same reason as (104a). The above data point to one direction: the focus–verb adjacency in Basque is not the result of V-to-∆ movement. The empirical evidence suggests that the adjacency requirement between focus and verb is the result of a PF process, as I have argued above. There is a problem with this account though. As I have explained above, it seems that contrastive focus forces V-adjacency. However, there are cases in Basque which suggest that V-adjacency forces contrastive focus. If (103) is slightly modified, as in (105), we obtain a different interpretation for the sentence. More precisely, when the verb is placed next to the subject, the subject has to be contrastively focused. (105) Mireni, JONEK eman zion muxu bat. Miren-dat Jon-erg give aux kiss one ‘Miren, JON kissed.’
In order to account for this state of affairs, I will assume that ∆ has a weak V feature. The verb moves overtly in those cases in which a PF violation is going to take place, that is, in those cases in which the affix is going to end up non-adjacent to the verb. To put it another way, the verb moves to ∆ overtly only if there is a focused element in Spec∆P so that it can support the PF verbal affix. Otherwise, the verb moves in LF in order not to violate Procrastinate. We thus have a mixed account between phonology and syntax. We have movement to ∆, but it is not obligatory. It is only forced when there is a verbal PF affix in Spec∆P. This account makes a prediction. The order Topic–Verb–Object should not be allowed, as in (106): (106) *Mireni eman zion liburua. Miren-dat give aux book ‘Miren, (Jon) gave the book.’
The reason why the topic cannot precede the verb is because there is no PF verbal affix involved in this case. There is then no reason for V-to-∆ to take place overtly. Hence the verb moves in LF in order not to violate Procrastinate.27 6.3 Three wh-phrases In Basque it is possible to front three wh-phrases, as in (109) (I am trying to avoid using an inanimate wh-phrase since some speakers are sensitive to animacy differences (cf. Billings and Rudin (1996)). The problem with the
Non-wh-fronting in Basque 217
paradigm below is that my informants do not have Superiority effects to start with, as shown in (107) and (108). Given this state of affairs, it is impossible to draw a definite conclusion with respect to Superiority with three wh-phrases in Basque.28,29 (107)(?)Nor nori aurkeztu zion Jonek? who-abs who-dat introduce aux Jon-erg ‘Who did Jon introduce to whom? (108)(?)Nori nor aurkeztu zion Jonek? (109) Nork nori nor aurkeztu zion? who-erg who-dat who-abs introduce aux ‘Who introduced who to whom?’ (110) Nork nor nori aurkeztu zion?
As I have shown, it is not clear what is going on with respect to Superiority in constructions with 3 wh-phrases. I will now examine the topic/focus status of wh-elements in constructions with 3 wh-phrases. In order to do that, I will use Pesetsky’s (1987) the hell test. The results of the test are as follows: the first two wh-phrases are D-linked/topicalized and the wh-phrase adjacent to the verb is focalized. (111) *Nor arraiok nori zer esan zion? who hell-erg who-dat what-abs say aux ‘Who the hell said what to whom?’ (112) *Nork nor arraiori zer esan zion? who-erg who hell-dat what-abs say aux ‘Who said what to who the hell?’ (113) Nork nori zer arraio esan zion? who-erg who-dat what hell-abs say aux ‘Who said what the hell to whom?’
Is there a way to account for the pattern Topic–Topic–Focus in Basque? In order to account for this pattern, I propose the following: in Basque there is a head with two features; namely, Attract-all-discourse and Attract-1Topic (I am assuming, following Boškovic´ (1999), that elements that are already located in discourse-related positions (e.g. lower SpecTopP) cannot be attracted by another discourse-related head). This system basically says that all discourserelated elements, that is, Topic and Focus, are attracted. Furthermore, the highest wh-phrase is attracted first. This analysis would give us the Bulgarian
218 Lara Reglero
pattern where the first wh-phrase is attracted first and the second and third wh-phrase are freely ordered (cf. (10) and (11)). This is of course an idealization of judgments. I am taking the Bulgarian pattern to be the representative one until clearer judgments emerge from the Basque data with three wh-phrases. The analysis proposed above, that is, that there is a head in Basque with both the Attract-all-discourse and the Attract-1-Topic properties thus ensures that the highest wh-phrase is attracted first. It also ensures that there is only one focus. The last claim follows from the adjacency requirement between the verbal PF affix and the verb. For example, in a Topic Focus Focus Verb order, the adjacency requirement between the focus feature of the second Focus and the Verb cannot be satisfied, hence this order is ruled out.
7. Long-distance contexts As discussed in Section 2.2., Serbo-Croatian must have wh-movement in those contexts where C is inserted overtly in the structure. If Basque is a SerboCroatian type language with respect to when it must have wh-movement we would expect Basque also to obligatorily have wh-movement when overt C is present in a sentence, as in (114).30 (114) Nork zer esan dute erosi duela? who-erg what-abs say aux bring aux-comp ‘Who did they say bought what?’
The data in (114) raises several questions though. To begin with, nork in (114) is D-linked, as shown in (115a). Furthermore, single-pair answers are allowed in (114) (recall that in Boškovic´’s system, single-pair answers are only allowed when wh-movement to SpecCP does not take place overtly): (115) a. *Nor arraiok zer esan dute erosi duela? who hell-erg what-abs say aux buy aux-comp ‘Who the hell did they say that bought what?’ b. ?Nork zer arraio esan dute erosi duela? who-erg what hell-abs say aux buy aux-comp ‘Who did they say that bought what the hell?’
This indicates that wh-movement is not taking place here. There are two possibilities to account for the data. It might be the case that -(e)la is not a C or, more likely, that Basque is like Russian.31 As Stepanov (1998) and Boškovic´
Non-wh-fronting in Basque 219
(2002) have argued, wh-movement does not take place in Russian even in those cases where C is inserted overtly in the structure.32 As expected, single-pair answers are allowed in Russian even in this type of constructions. If Basque is like Russian, we have an explanation for the fact that single-pair answers are still allowed even in the presence of an overt C.
8. Concluding remarks In this paper I have provided an analysis which accounts for the wh-movement strategies in Basque presented in Section 3. The main findings of this paper are as follows. First, in sentences such as (51) no wh-phrase remains in situ in the structure. Under my proposal, the postverbal wh-phrase raises overtly to check a topic feature. Second, I have offered an analysis for Multiple wh-fronting structures in which the wh-phrase closer to the verb is focalized and the rest of the wh-phrases are topicalized. In order to account for this pattern, I have proposed that Basque has a head with two features: Attract-all-discourse and Attract-1Topic. Under this analysis, all wh-phrases bearing focus or topic are moved to the front of the sentence. Furthermore, the highest wh-phrase must move first. The fact that only one wh-phrase is focalized follows from the adjacency requirement of a PF verbal affix. I have argued for a new paradigm of multiple question formation which has not been previously analyzed in the literature. I have shown that the paradigm can be successfully captured by Boškovic´’s Attract-all-F approach to multiple wh-fronting. More specifically, I have presented data from a language which can on the surface be treated as an Attract-all-Topic language. However, I have presented evidence that that is not the correct analysis for Basque. Instead, I have shown that the Basque data should be analyzed by positing a head with the Attract-all-discourse and the Attract-1Topic features. It is also worth noting that Basque is a multiple wh-fronting language which crucially differs from Slavic in that there is only one Focus. So far, we have the following multiple wh-fronting languages: Attract-all-Focus (Slavic) and Attract-all-Discourse (Basque). It remains to be seen whether there is a language where all and only topicalized wh-phrases are fronted to the beginning of the sentence.
220 Lara Reglero
Notes * My deepest thanks go to my advisor, Željko Boškovic´. Without his help, this paper would have never been possible. For helpful comments and discussion, I would like to thank Sigrid Beck, Cedric Boeckx, Andrea Calabrese, Kleanthes Grohmann, Howard Lasnik, Jon Ortiz de Urbina, Chris Wilder and audiences at the V LEHIA Workshop, the 6th Hispanic Linguistics Symposium and WECOL 2002. Thanks to my informants: Xabier Artiagoitia, Arantzazu Elordieta, Gorka Elordieta, Ricardo Etxepare, Elena García, Estíbaliz Izagirre, Zurine Lekuona and Itziar San Martín. Thanks to all of them for their comments and for their patience with the judgments. Last but not least, I would like to thank Daniel Solís for his invaluable support during the writing of this paper. This research has been funded by a scholarship from the Department of Education, Universities and Research of the Basque Government. 1. See Boškovic´ (2002) for some exceptions to this generalization. 2. Boškovic´ (1999) uses the term ‘non-wh-fronting’ to refer to the type of movement motivated independently of the +wh-feature. More specifically, he treats focus movement as an instance of ‘non-wh-fronting’. As the title of this paper suggests, I intend to investigate the phenomenon of non-wh-fronting in Basque. However, I am using the term ‘non-wh-fronting’ to refer not only to focus but also to topic movement of wh-phrases. 3. Her arguments come from the position that wh-words and contrastively focused phrases move to and also from the interpretation of adverbs. Since she found that both wh-phrases and contrastively focused non-wh-elements behave on equal grounds in both realms, she concluded that wh-phrases are inherently focused in Serbo-Croatian and consequently undergo the same type of movement contrastively focused non-wh-elements undergo. See Stjepanovic´ (1998) for the relevant tests. 4. The judgment in (2) holds for echo questions used to ask for repetition of information the person asking the question has not heard. (2) improves if the echo question is used to express surprise. This contrast follows from the nature of focus movement since, in contrast to the first reading, in the second reading the value of the wh-word is known to both the speaker and hearer. Hence, the wh-phrase is subject to focus movement only on the first reading. See Boškovic´ (2002) for details. 5. The Superiority Condition was originally formulated by Chomsky (1973). In order to account for contrasts such as (i) and (ii), Chomsky (1973) proposed the Condition in (iii): (i) Who bought what? (ii) *What did who buy? (iii) No rule can involve X, Y in the structure …X… […Z…WYV…] where the rule applies ambiguously to Z and Y, and Z is superior to Y. The category A is superior to the category B if every major category dominating A dominates B as well but not conversely. 6. Eventually, both wh-phrases land in SpecCP. The analysis proposed by Boškovic´ (1999) holds either if one adopts the rightward adjunction (cf. Rudin 1988) or the multiple specifier analysis of multiple wh-fronting in Bulgarian (cf. Richards 1997). Either way the first wh-phrase linearly moves first.
Non-wh-fronting in Basque 221
7. Boškovic´ (1997c) assumes that C-insertion triggers immediate wh-movement in SerboCroatian. C in (3) and (4) is phonologically null. Boškovic´ (1997c) assumes that it is merged at the root of the tree in LF. Since C does not enter the structure overtly, wh-movement does not take place overtly. In indirect questions C must be inserted overtly in order not to violate the requirement that Merger expand the tree. Example (14) above contains a phonologically realized C li. Given that li is phonologically realized, it follows that it must be introduced in overt syntax. The contexts where C must enter the structure overtly involve overt wh-movement under the assumption that C-insertion triggers immediate wh-movement (the assumption follows from Chomsky’s 1995 approach to strength). As a result, these contexts exhibit Superiority effects. 8. There are three main cases in Basque: the ergative -k, the absolutive ∆, and the dative -(r)i. I will use the following abbreviations: erg = Ergative, abs = Absolutive and dat = Dative. 9. In (22) nork is focalized. Zer in (23) bears focus and is separated by a short break from the first wh-word. I discuss this below. Note that some speakers disallow (23). 10. It is possible to front three wh-phrases, as in (i). I have not included the relevant data in the main text due to unclear judgments (when it comes to Superiority effects) and several interfering factors. For the relevant discussion on the data with three wh-phrases, see Section 6.3. (i)
Nork nori nor aurkeztu zion? who-erg who-dat who-abs introduce aux ‘Who introduced who to whom?’
11. According to Etxepare and Ortiz de Urbina (in press), (24) is grammatical. However, seven out of the eight informants I consulted about this sentence found (24) considerably more degraded than (22), so I will concentrate on this judgement only. 12. Syntactic tests point to the same direction. To begin with, topics denote old information. As É. Kiss (1981: 185) puts it, “the topic is said to be the introductory part of a sentence foregrounding known information”. In English, (i-a) cannot be the answer to (i-b): (i) a. What did Mary buy? b. *The house, Mary bought. Basque exhibits the same behavior, as in (ii): (ii) a.
Zer erosi du Mirenek? what-abs buy aux Miren-erg ‘What did Miren buy?’ b. *Etxea, Mirenek erosi du. house-abs Miren-erg buy aux ‘The house, Miren bought.’
Negative quantifiers cannot be topicalized, as in (iii): (iii)
*Nothing, Mary bought.
The counterpart of (iii) in Basque is also ruled out, as shown in (iv): (iv)
*?Ezer, Mirenek ez du erosi. nothing/anything Miren-erg neg aux buy ‘Nothing, Miren bought.’
222 Lara Reglero
Finally, idiom chunks generally resist topicalization, as in (v-b): (v) a. John kicked the bucket. b. *The bucket, John kicked. An idiom chunk such as (vi-a) cannot be topicalized in Basque, as shown in (vi-b): (vi) a.
Jonek adarra jo du. Jon-erg horn hit aux ‘Jon has hit the horn.’ (meaning ‘John was kidding.’) b. *Adarra, Jonek jo du.
13. Ortiz de Urbina (1993) provides an example containing a postverbal focalized element: (i) Etorri da AITA come aux father ‘It is the father that has arrived.’ According to Elordieta (2001), postverbal focus is interpreted as ‘focus of correction’. As she makes clear, ‘focus of correction’ has a different meaning and a different intonation pattern from that of preverbal focus. Therefore, I will disregard these examples since they seem to have different properties from the ones I am interested in. For further discussion of the difference between preverbal and postverbal focus, see de Rijk (1996) and Etxepare (1997). 14. As de Rijk (1978) points out, there are a few elements which can be placed between the focus/wh-word and the verb. For instance, ez ‘no’, omen ‘reportedly’, bide ‘apparently’, ote ‘by any chance’ and al (yes-no question marker) can interrupt the sequence focus/wh-word Verb. An example of this phenomenon is provided below: (i) Zergatik ote dago hainbeste tximeleta? why are so.many butterflies ‘Why are there so many butterflies? (Etxepare & Ortiz de Urbina, in press) What is responsible for the placement of these elements is beyond the scope of this paper. Presumably, these elements are clitics adjoined to the verb. Uriagereka (1992, 1999) also provides examples where the adjacency requirement is not respected: (ii) Nork ardoa edaten du? who-erg wine-abs drink aux ‘Who has drunk wine?’ (iii) Zergatik zaldunak herensugea hil zuen? why knight-erg dragon-abs kill aux ‘Why did the knight kill the dragon?’ The problem with these examples is that they are very marginal, especially the first one (Elordieta 2001). Given the exceptional character of these examples, I will disregard them for the purposes of my paper. 15. This conclusion is tentative. As I will show below, only some wh-words, that is, those wh-words adjacent to the verb, undergo focus movement in Basque. I will show that, when it comes to wh-fronting, there are other types of movements involved (i.e. topic movement), too.
Non-wh-fronting in Basque 223
16. I am taking atzo ‘yesterday’ as an example of a sentential adverb. I did not use the sentential adverb seguruenera ‘probably’, as in Watanabe (1993) and Boškovic´ (1997d), since there are interfering factors with it. To be more precise, (77)–(78) below are both bad with the sentential adverb seguruenera ‘probably’. 17. The sentential adverb can also appear in sentence initial position. In this example the adverb is interpreted as a topic and is presumably adjoined to some projection (to be defined) in the Left Periphery. I will disregard these types of examples since they do not affect the argument presented here. 18. I checked these data with three informants. Two of them only allowed pair-list answers and one allowed a single-pair answer. 19. One of my informants found (77) degraded. However, he found that (78) was worse than (77). Crucial for my argumentation is that there is a contrast in grammaticality between the two examples. 20. This idea receives further support from Stjepanovic´’s (1999) work. According to her, topics in Serbo-Croatian are licensed either in AgrSP or in a lower projection in the structure; namely, PredP. Under her analysis, PredP (located between T and AgrOP) is able to host topics. My TopP is very similar to Stjepanovic´’s (1999) PredP. Both projections are relatively low in the structure and are able to host topics. 21. Following Ortiz de Urbina (1999b), I will assume that FocP in Basque is left-headed despite the fact that Basque is an SOV language. See Ortiz de Urbina (1999b) for arguments in favor of this approach. 22. I have not included TP in the tree in (81) since it is not crucial for the analysis presented here. Furthermore, whether SpecTP is present in the structure varies from language to language (cf. Jonas and Bobaljik 1993). I assume that AgrSP is the locus of the EPP feature in Basque, with T raising to AgrS so that the EPP feature is checked in SpecAgrSP. Whether this is the correct analysis remains an open (but not a crucial) issue. 23. Boškovic´ (1997b) presents evidence that all wh-phrases that move to a position higher than their Case-checking position must pass through their Case-checking position before arriving at their final landing site. On the topic of A-movement prior to wh-movement, see also Grohmann (1997, 1998). 24. A question arises at this point: how about if nork is [+topic] and zer [+focus]? Could nork be attracted to SpecTopP across zer? The answer has to be no. It seems that the exact feature specification focus or topic does not really matter when it comes to locality. What really matters is either that focus and topic are both operator features or that focus and topic are both discourse related, hence they induce a blocking effect for each other. A similar point can be made with respect to topicalization out of wh-islands in English. As Boškovic´ (2000) points out, (i) is degraded: (i) ??(Peter thinks that) that booki you wonder wherej John put ti tj. The question that arises is the following: why should the [+Wh] feature be relevant in attracting topics? The Basque data raise a similar question. Like the Basque data in the text, (i) can be accounted for by appealing to a more inclusive +operator feature.
224 Lara Reglero
25. Superiority may be an interfering factor in (94)–(95) if, due to movement to its Casechecking position, the accusative wh-phrase is higher than the adjunct (see Boškovic´ 1997b for the relevant discussion). The Superiority problem should not arise in (i) and (ii) where we have the sequence nola-noiz ‘how-when’. Once we have controlled for the interfering factor in (94) to (95) it is clear that my prediction still holds. (i) is ungrammatical because nola ‘how’, being non-D-linked, cannot land in the position where D-linked elements land. (i) *Nola noiz konpondu zuen Jonek autoa? how when fix aux Jon-erg car-abs ‘How did Jon fix the car when?’ (ii) ??Noiz nola konpondu zuen Jonek autoa? 26. I am assuming the affix hopping analysis of affixation (cf. Chomsky 1957), in which an affix and its host must be linearly adjacent in PF. See in this respect Bobaljik (1995), Halle and Marantz (1993) and Lasnik (1995). 27. A question arises at this point. Given the badness of (106), why is (89) grammatical? Recall that I argued that the wh-phrase in (89) undergoes overt movement to TopP. Given the ungrammatical status of (106), I will assume that the wh-phrase in (91) undergoes wh-movement in this case with the verb moving to C. If this approach is correct, Basque would resemble Serbo-Croatian and French in having optional wh-movement. See Boškovic´ (2002) for relevant discussion of Serbo-Croatian and French. 28. One of my informants found (110) ungrammatical for an unclear reason. 29. See Reglero (2002) for fuller data with three wh-phrases in Basque. Unfortunately, the data presented in that paper are also inconclusive. 30. As Xabier Artiagotia (p.c.) points out to me, some speakers reject the Multiple wh-Fronting strategy in long-distance contexts even though they accept it in short-distance questions such as (23). 31. As Jon Ortiz de Urbina (p.c.) points out to me, in Basque there are no Complementizers with non-echo short-distance questions. This makes one wonder whether -(e)la in (114) is a Complementizer at all. 32. The difference between Serbo-Croatian and Russian is that Serbo-Croatian +wh-C has a strong +wh-feature, so it triggers wh-movement as soon as it enters the structure (more precisely, if it enters the structure overtly, it triggers overt wh-movement), while in Russian it is weak.
References Altube, S. 1929. Erderismos. Bilbao: Euskaltzaindia. [Reprinted in Altube, S. 1975. Orain Sorta, Bilbao: Indauchu Editorial.] Billings, L. and Rudin, C. 1996. “Optimality and Superiority: A new approach to overt multiple wh-ordering.” In Formal Approaches to Slavic Linguistics: The College Park Meeting, 1994, J. Toman (ed), 35–60. Ann Arbor, Mich.: Michigan Slavic Publications.
Non-wh-fronting in Basque 225
Bobaljik, J. 1995. Morphosyntax: The syntax of verbal inflection. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Boeckx, C. and S. Stjepanovic´. 2000. “The wh-clitic connection.” In Formal Approaches to Slavic Linguistics: The Philadelphia Meeting, 1999, T. H. King and I. Sekerina (eds), 22–40. Ann Arbor, Mich.: Michigan Slavic Publications. Boškovic´, Ž. 1997a. “Fronting wh-phrases in Serbo-Croatian.” In Formal Approaches to Slavic Linguistics: The Indiana Meeting, 1996, M. Lindseth and S. Franks (eds), 86–107. Ann Arbor, Mich.: Michigan Slavic Publications. Boškovic´, Ž. 1997b. “On certain violations of the Superiority Condition, AgrO, and economy of derivation.” Journal of Linguistics 33: 227–254. Boškovic´, Ž. 1997c. “Superiority effects with multiple wh-fronting in Serbo-Croatian.” Lingua 102: 1–20. Boškovic´, Ž. 1997d. The Syntax of Nonfinite Complementation: An Economy Approach. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 1998a. “On the interpretation of multiple questions.” In A Celebration: Essays for Noam Chomsky’s 70th Birthday, J. Fodor, S.J. Keyser and A. Brand (eds). Cambridge, Mass.: MIT Press. [Online publication — available for download at http://cognet.mit.edu/ Books/celebration/essays/Boeckx.html (sic); revised version to appear in Linguistic Variation Yearbook 1.] Boškovic´, Ž. 1998b. “Multiple wh-fronting and economy of derivation.” In Proceedings of the Sixteenth West Coast Conference on Formal Linguistics, E. Curtis, J. Lyle and G. Webster (eds), 49–63. Stanford, Calif.: CSLI. Boškovic´, Ž. 1999. “On multiple feature checking: Multiple wh-fronting and multiple headmovement.” In Working Minimalism, S. Epstein and N. Hornstein (eds), 159–187. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 2000. “Sometimes in [Spec, CP], sometimes in situ.” In Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, R. Martin, D. Michaels and J. Uriagereka (eds), 53–87. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 2001. On the Syntax–Phonology Interface: Cliticization and Related Phenomena. Amsterdam: Elsevier. Boškovic´, Ž. 2002. “On multiple wh-fronting.” Linguistic Inquiry 33: 351–383. Chomsky, N. 1957. Syntactic Structures. The Hague: Mouton. Chomsky, N. 1973. “Conditions on transformations.” In A Festschrift for Morris Halle, S. Anderson and P. Kiparsky (eds), 232–286. New York: Holt, Rinehart and Winston. Chomsky, N. 1995 “Categories and transformations.” In N. Chomsky, The Minimalist Program, 219–394. Cambridge, Mass.: MIT Press. den Dikken, M. and Giannakidou, A. 2001. “What the hell?!” In Proceedings of NELS 31, M. Kim and U. Strauss (eds), 163–182. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). den Dikken, M. and Giannakidou, A. 2002. “From hell to polarity: Aggressively nonD-linked wh-phrases as polarity items.” Linguistic Inquiry 33: 31–61. Eguzkitza, A. 1986. Topics on the syntax of Basque and Romance. Doctoral dissertation, University of California, Los Angeles. Elordieta, A. 2001. Verb movement and constituent permutation in Basque. Doctoral dissertation, HIL/University of Leiden.
226 Lara Reglero
Etxepare, R. 1997. “Two Types of Focus in Basque.” In Proceedings of the Fifteenth West Coast Conference on Formal Linguistics, B. Agbayani and S. Tang (eds), 113–128. Stanford, Calif.: CSLI. Etxepare, R. and Ortiz de Urbina, J. In press. “Focus.” In A Grammar of Basque, J. I. Hualde and J. Ortiz de Urbina (eds). Berlin: Mouton de Gruyter. Grohmann, K. K. 1997. “German Superiority.” In Groninger Arbeiten zur Germanistischen Linguistik 40: 97–107. Grohmann, K. K. 1998. “Syntactic inquiries into discourse restrictions on multiple interrogatives.” In Groninger Arbeiten zur germanistischen Linguistik 42: 1–60. Halle, M. and Marantz, A. 1993. “Distributed Morphology and the pieces of inflection.” In The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, K. Hale and S. J. Keyser (eds), 111–176. Cambridge, Mass.: MIT Press. Horvath, J. 1986. Focus in the Theory of Grammar and the Syntax of Hungarian. Dordrecht: Foris. Jackendoff, R. 1972. Semantic Interpretation in Generative Grammar. Cambridge, Mass.: MIT Press. Jonas, D. and Bobaljik, J. D. 1993. “Specs for subjects: The role of TP in Icelandic.” MIT Working Papers in Linguistics 18: Papers on Case and Agreement I, J. D. Bobaljik and C. Phillips (eds), 59–98. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). É. Kiss, K. 1981. “Structural relations in Hungarian, a “free” word order language.” Linguistic Inquiry 12: 185–213. É. Kiss, K. 1993. “Wh-movement and specificity.” Natural Language and Linguistic Theory 11: 85–120. É. Kiss, K. 1998. “Identificational focus versus information focus.” Language 74: 245–273. Lambova, M. 2001. “On A-bar movements in Bulgarian and their interaction.” The Linguistic Review 18: 327–374. Lasnik, H. 1995. “Verbal morphology: Syntactic Structures meets the Minimalist Program.” In Evolution and Revolution in Linguistic Theory: Essays in Honor of Carlos Otero, H. Campos and P. Kempchinsky (eds), 251–275. Washington, D. C.: Georgetown University Press. Martin-Callejo, E. M. 1984. Some aspects of Basque morphology and syntax. Doctoral Dissertation, University of Arizona, Tucson. Ortiz de Urbina, J. 1989. Parameters in the Grammar of Basque. Dordrecht: Foris. Ortiz de Urbina, J. 1993. “Feature percolation and clausal pied-piping.” In Generative Studies in Basque Linguistics, J. I. Hualde and J. Ortiz de Urbina (eds), 189–219. Amsterdam: John Benjamins. Ortiz de Urbina, J. 1995. “Residual verb second and verb first in Basque.” In Discourse Configurational Languages, K. É. Kiss (ed), 99–121. Oxford: Oxford University Press. Ortiz de Urbina, J. 1999a. “Focus in Basque.” In The Grammar of Focus, G. Rebuschi and L. Tuller (eds), 311–333. Amsterdam: John Benjamins. Ortiz de Urbina, J. 1999b. “Force phrases, focus phrases and left heads in Basque.” In Grammatical Analyses in Basque and Romance Linguistics. Papers in Honor of Mario Saltarelli, J. Franco, A. Landa and J. Martín (eds), 179–194. Amsterdam: John Benjamins. Pesetsky, D. 1987. “Wh-in situ: Movement an unselective binding.” In The Representation of (In)definiteness, E. J. Reuland and A. G. B. ter Meulen (eds), 98–129. Cambridge, Mass.: MIT Press.
Non-wh-fronting in Basque 227
Reglero, L. 2002. “Multiple questions in Basque.” Ms., University of Connecticut, Storrs. de Rijk, R. P. G. 1969. “Is Basque S. O. V.?” Fontes Linguae Vasconum 1, 319–351. [Reprinted in de Rijk, R. P. G. 1998. De Lingua Vasconum: Selected Writings [Supplements to ASJU XLIII], 13–38. Bilbao: EHU-UPV.] de Rijk, R. P. G. 1978. “Topic fronting, focus positioning and the nature of the verb phrase in Basque.” In Studies on Fronting, F. Jansen (ed), 81–112. Lisse: The Peter de Ridder Press. [Reprinted in de Rijk, R. P. G. 1998. De Lingua Vasconum: Selected Writings [Supplements to ASJU XLIII], 183–202. Bilbao: EHU-UPV.] de Rijk, R. P. G. 1996. “Focus and quasifocus on the Basque negative statements.” In Revista Internacional de Estudios Vascos 41: 63–76. Richards, N. 1997. What moves where when in which language? Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Rizzi, L. 1997. “The fine structure of the left periphery.” In Elements of Grammar: Handbook of Generative Syntax, L. Haegeman (ed), 281–337. Dordrecht: Kluwer. Rudin, C. 1988. “On multiple questions and multiple wh-fronting.” Natural Language and Linguistic Theory 6: 445–501. Stepanov, A. 1998. “On wh-fronting in Russian.” In Proceedings of NELS 28, P. N. Tamanji and K. Kusumoto (eds), 453–467. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Stjepanovic´, S. 1998. Short distance movement of wh-phrases in Serbo-Croatian matrix clauses. Paper presented at the Workshop on Comparative Slavic Morphosyntax, Spencer, Indiana. [June 1998] Stjepanovic´, S. 1999. What do second position cliticization, scrambling and multiple wh-fronting have in common? Doctoral dissertation, University of Connecticut, Storrs. Uriagereka, J. 1992. “The syntax of movement in Basque.” In Syntactic Theory and Basque Syntax [Supplements of ASJU XXVII], J. A. Lakarra and R. Gomez (eds), 417–446. Donostia-San Sebastián: Diputación de Gipuzkoa. Uriagereka, J. 1995. “Aspects of the syntax of clitic placement in Western Romance.” Linguistic Inquiry 26: 79–123. Uriagereka, J. 1999. “Minimal restrictions on Basque movements.” Natural Language and Linguistic Theory 17: 403–444. Watanabe, A. 1993. AGR-based Case Theory and its interaction with the A¢-system. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge.
Malagasy as an optional multiple wh-fronting language* Joachim Sabel Université catholique de Louvain
In this paper, I analyze (multiple) wh-questions in Malagasy. Malagasy is a wh-in situ language that allows for partial and full wh-movement as well. In addition, it is an optional multiple wh-fronting language. I will discuss the restrictions for single wh-in situ and wh-ex situ questions and for the varieties of multiple wh-questions as well. It will be shown that question formation in Malagasy can be explained in terms of a feature checking analysis settled within the Minimalist Program (Chomsky 1995, 2000). With respect to multiple wh-fronting construction, I will argue that multiple fronted elements form one constituent, a wh-cluster. The paper is organized as follows. In Section 1, I shortly describe the Malagasy sentence structure and voice system. Section 2 contains a discussion of the possible variants of single wh-questions, and Section 3 contains the discussion of multiple wh-questions. Section 4 is the summary.
1.
Sentence structure in Malagasy
Malagasy is a member of the Western Austronesian language family, spoken on the island of Madagascar. It is a verb-initial and subject-final language (VOS). In Malagasy, only a sentence-final derived grammatical subject can be questioned via extraction to sentence-initial position. In order to illustrate this fact, it is helpful to consider first the examples in (1). As shown in (1), in Malagasy we have different possibilities to express a sentence meaning such as, for example, the proposition the mother washes the clothes with soap. Semantically, the sentences in (1) are logically equivalent in that they have the same truth
230 Joachim Sabel
conditions. In each of the sentences a different argument occurs in the subject position at the right periphery of the sentence. Verb morphology indicates which θ-role is linked to the respective underlined (nominative) element in the sentence-final subject position. (In the following, the respective subject is always underlined.) In (1a), the agent is the grammatical subject and the verb shows agent topic marking. In (1b), the theme appears in the structural subject position and the verb indicates the theme θ-role. In (1c), the instrument is the subject. In this case the verb bears circumstantial topic morphology. In some analyses this sentence-final element is analyzed as a topic, in other analyses it is analyzed as the focused element.1 (1) a.
Manasa ny lamba amin’ ny savony ny reny. pres-at-wash the clothes with the soap the mother-nom ‘The mother washes the clothes with the soap.’ b. Sasan’ ny reny amin’ ny savony ny lamba. pres-tt(/pass)-wash the mother with the soap the clothes-nom ‘The mother washes the clothes with the soap.’ c. Anasan’ ny reny ny lamba ny savony. pres-ct-wash the mother the clothes the soap-nom ‘The mother washes the clothes with the soap.’
Several analyses have been proposed in order to account for the different word orders found in the examples (1). Guilfoyle at al. (1992) assume a passive analysis. Starting from an SVO base order, as shown in (2a), they argue that verbal morphology destroys the verb’s capacity to assign Case to the corresponding XP. Therefore this XP has to move to a Case position, i.e. the rightperipheral subject position. Keenan (2000) base-generates the different orders in (1); whatever argument represents the grammatical subject (XP) in the structure (2b) is base-generated in the SpecIP position. Another variant is the remnant movement analysis in (2c). The former analyses assume specifiers on the right, something that the (VP- or) IP-fronting analysis tries to avoid. Here VOS order is derived from an SVO base order by remnant (VP- or) IP-movement to the left of the previously topicalized XP (cf. Pensalfini 1995, among others). (2) a. [IP [I0 Vx] [I¢ Agent [VP tv Theme Instrument]] XP] b. [IP [Vx Theme Instrument Agent] XP] c. [CP [IP Vx Agent tv Theme Instrument] [ΣP XP tIP]] (XP = Agent, Theme, Goal, Instrument…)
Malagasy as an optional multiple wh-fronting language
A fourth analysis is based on the idea that the element in the right-peripheral position behaves like a topic in verb-second languages such as German or Icelandic (see Richards 2000).
2. Single wh-questions in Malagasy 2.1 The “Subjects-Only” restriction For present purposes, a commitment to one of these analyses is not crucial. Regardless of which analysis in (2) is adopted, it is important that only the sentence-final XP in SpecIP can be A¢-moved. This is exactly the element that the verb agrees with. In (3a) the verb agrees with the agent and the agent 0-role can be extracted. Other elements such as the theme and the instrument cannot be questioned. In (4) the theme is the derived grammatical subject and the only element that can be extracted, and in (5) it is the instrument. This condition on extraction is called the “subjects-only” restriction. (3) a.
Iza no manasa ny lamba amin’ ny savony t? who prt pres-at-wash the clothes with the soap ‘Who washes the clothes with soap?’ b. *Inona no manasa t amin’ ny savony ny reny? what prt pres-at-wash with the soap the mother ‘What does the mother wash with the soap?’ c. *Amin’ ny inona no manasa ny lamba t ny reny? with the what prt pres-at-wash the clothes the mother ‘With what does the mother wash the clothes?’
(4) a.
Inona no sasan’ ny reny amin’ ny savony t? what prt pres-tt-wash the mother with the soap ‘What does the mother wash with the soap?’ b. *Iza no sasana t amin’ ny savony ny lamba? who prt pres-tt-wash with the soap the clothes ‘Who washes the clothes with soap?’ c. *Amin’ ny inona no sasan’ ny reny t ny lamba? with the what prt pres-tt-wash the mother the clothes ‘With what does the mother wash the clothes?’
(5) a.
Inona no anasan’ ny reny ny lamba t? with.what prt pres-ct-wash the mother the clothes ‘With what does the mother wash the clothes?’
231
232 Joachim Sabel
b. *Iza no anasan t ny lamba ny savony? who prt pres-ct-wash with the soap the mother ‘Who washes the clothes with the soap?’ c. *Inona no anasan’ ny reny t ny savony? what prt pres-ct-wash the mother the soap ‘What does the mother wash with the soap?’
Note, that most adjuncts do not have to agree with the verb and become the subject before extraction takes place. As shown in (6), wh-extraction of adjuncts such as aiza ‘where’ is possible with all types of verbal forms: (6) a.
Aiza no manasa ny lamba amin’ ny savony ny reny? where prt pres-at-wash the clothes with the soap the mother b. Aiza no sasan’ ny reny amin’ ny savony ny lamba? where prt pres-tt-wash the mother with the soap the clothes c. Aiza no anasan’ ny reny ny lamba amin’ ny savony? where prt pres-ct-wash the mother the clothes with the soap ‘Where does the mother wash the clothes with the soap?’
To sum up, with respect to argument extraction only the right-peripheral structural subject may be extracted in Malagasy, in contrast to languages such as English, where argument extraction may proceed from the structural subject and object position (see Sabel 2002b for an attempt to derive the “subjects-only”restriction).2 Let us next turn to long extraction. 2.2 Full and Partial wh-movement In (7) the theme is questioned and overtly moved to SpecCP, i.e. to the position in which it takes scope. I call (3)–(6) and (7) the “full wh-movement” construction. (7) [CP1 Inona no heverin-dRabe [CP2 fa novidin-dRakoto t]]? what foc pres-tt-believe-Rabe that past-tt-buy-Rakoto ‘What does Rabe believe that Rakoto has bought?’
In the overt syntax, the position of wh-elements is not limited to positions in which they take scope. Besides full wh-movement, Malagasy allows for a second possibility of forming a wh-question, i.e. the so-called “partial wh-movement” construction in (8). Partial wh-movement consists of movement of a wh-phrase in an embedded [−Wh] SpecCP (assuming CP-recursion in (8); see the discussion below):
Malagasy as an optional multiple wh-fronting language 233
(8) [CP1 Heverin-dRabe [CP2 fa inona no novidin-dRakoto t]]? pres-tt-believe-Rabe that what foc past-tt-buy-Rakoto ‘What does Rabe believe that Rakoto has bought?’
In (8) we find some kind of “long distance linking” between the wh-element in SpecCP2 and the matrix SpecCP position. The wh-phrase in the lowest clause is interpreted in the SpecCP position of the highest clause. In other words, this construction is similar to the corresponding wh-questions in (7), which results from long wh-movement.3 2.3 Three restrictions for wh-in situ in Malagasy Malagasy allows for a third option of forming a wh-question. Wh-phrases may stay in their base-position in the overt syntax, as shown in (9). (9) [CP Heverin-dRabe [CP fa nividy inona Rakoto]]? pres-tt-believe-Rabe that past-at-buy what Rakoto ‘What does Rabe believe that Rakoto has bought?’
Three restrictions determine the possibility of wh-in situ in Malagasy (see Sabel 2002b for more detailed discussion). The first concerns the focus particle no. As shown in the examples already discussed, if the wh-phrase in Malagasy is moved to the left periphery of the sentence, the focus particle no appears immediately after the fronted element. If overt wh-movement applies, no is obligatory (10b). On the other hand, as can be seen from (10a), if the wh-phrase is in-situ, no cannot be present. Hence, the wh-phrase is moved if only if the marker no occurs.4 (10) a.
Mividy (*no) inona any amin’ ny magazay Rabe? pres-at-buy (*prt what there in the shop Rabe ‘What does Rabe buy in the shop?’ b. Inona *(no) vidin-dRabe any amin’ ny magazay? what (*prt pres-tt-buy-Rabe there in the shop ‘What does Rabe buy in the shop?’
A second restriction consists in the fact that wh-words may not remain in SpecIP, as shown in (11a) and (12a). In order to explain this fact, it is important to note that in Malagasy non-specific NPs may not appear in the sentence-final subject position, a language-particular restriction that also holds, for example, in Dutch. It is well known that wh-words such as what or who are difficult to be interpreted as specific in the absence of any context (see also Grohmann, this volume). Tentatively, I assume that Malagasy indefinites as well as Malagasy
234 Joachim Sabel
wh-words are inherently [−specific]. Then wh-words may not occur in SpecIP, as shown in (11a) and (12a). In (11a), the wh-object inona is in SpecIP. In this case, extraction to SpecCP must take place, as in (11b). The only non-wh-movement construction that is possible is (11c), where the verb agrees with the agent. (12) shows the same for a thematic subject: (11) a. *Novidin-dRabe (ny) inona? past-tt-buy-Rabe (the) what b. Inona no novidin-dRabe? what prt past-tt-buy-Rabe c. Nividy inona Rabe? past-at-buy what Rabe All three: ‘What has Rabe bought?’ (12) a. *Nividy ny vary iza? past-at-buy the rice who b. Iza no nividy ny vary? who prt past-at-buy the rice c. Novidin’ iza ny vary? past-tt-buy who the rice All three: ‘Who bought the rice?’
The third restriction for wh-in situ is well known from the languages of the world; it concerns non-referential adjuncts. In English, adverbs of time, direction and place behave like complements with respect to wh-movement and wh-in situ (cf. Who left when?/Who lives where?). These “referential” adjuncts or “quasi-arguments” behave similarly in Malagasy. Wh-in situ of adjuncts, such as aiza ‘where’ and oviana ‘when’, is possible (with all voices) as with arguments. This is illustrated in (13a–b) and (14a–b). Alternatively, these adjuncts may be fronted, as can be seen in (13c) and (14c): (13) a.
Nividy ny vary taiza Rabe? past-at-buy the rice past-where Rabe b. Novidin-dRabe taiza ny vary? past-tt-buy-Rabe past-where the rice c. Taiza no nividy ny vary Rabe? past-where foc past-at-buy the rice Rabe All three: ‘Where did Rabe buy the rice?’
(14) a.
Natory oviana ny reny? past-at-sleep when the mother ‘When did mother sleep?’
Malagasy as an optional multiple wh-fronting language 235
b. Niverina oviana ny reny? past-at-return when the mother ‘When did mother return?’ c. Oviana no niverina ny reny? when foc past-at-return the mother
In contrast to referential adjuncts, non-referential adjuncts do not occur in situ, as illustrated in (15) and (16). This holds for all voices, cf. (15a–c). In this respect, Malagasy conforms to a well-known cross-linguistic generalization.5 (15) a. *Nanasa ny lamba amin’ ny savony nahoana ny reny? past-at-wash the clothes with the soap why the mother ‘Why did the mother wash the clothes with soap?’ b. *Nosasan’ ny reny amin’ ny savony nahoana ny lamba? past-tt-wash the mother prep the soap why the clothes c. *Nanasan’ ny reny ny lamba nahoana ny savony? past-ct-wash the mother the clothes why the soap (16) *Mahandro ny vary (manao) ahoana ny reny? pres-at-cook the rice (how the mother ‘How does the mother cook the rice?’
The examples in (17) show extraction of the non-referential adjuncts (manao) ahoana ‘how’ and nahoana ‘why’: (17) a.
Ahoana no anasan’ ny reny ny lamba? how foc pres-ct-wash the mother the clothes ‘How does the mother wash the clothes?’ b. Nahoana no nanasa ny lamba amin’ ny savony ny reny? why foc past-at-wash the clothes with the soap the mother ‘Why did the mother wash the clothes with soap?’
To sum up, wh-in situ in Malagasy is constrained by (i) the presence of the focus particle no, by (ii) the prohibition of wh-elements occurring in the structural subject position, and by (iii) the referential/non-referential nature of adjuncts. Before I turn to the discussion of multiple wh-questions in Malagasy, I will outline an analysis for the different possibilities found with respect to the formation of single wh-questions in Malagasy in the following section. 2.4 Wh-movement as Focus movement For reasons outlined in Sabel (1998, 2000, 2001), I assume that the position of wh-words in wh-questions is universally determined by [Wh]- and [Focus]-
236 Joachim Sabel
features.6 To be more precise, I assume that wh-phrases have to check [+Focus]- as well as [+Wh]-features, although the only latter are always located in the position where the wh-phrase takes its scope. That wh-movement may be triggered by the need to check some [+Focus]features can be demonstrated with the following examples from Tuki and Malay (see Sabel 2000 for more extensive discussions and further arguments). Tuki is an SVO Bantu language, spoken in Cameroon (Biloa 1995). Note that in Tuki, as in Malagasy, wh-phrases may remain in situ (18a) or move to the left periphery of the sentence (18b). As can be seen in (18b), if wh-movement applies, a focusmarker occurs, as was observed already in Malagasy (SP = subject pronoun). (18) Tuki (Biloa 1995) a. Puta o-endam n(a) adongo ni? Puta sp-goes to village when ‘When does Puta go to the village?’ b. Ni owu Puta o-endam n(a) adongo t ? when foc Puta sp-goes to village ‘When does Puta go to the village?’
In (19b), derived from (19a), long distance wh-movement leaves agreeing focus markers in the embedded clause: (19) a.
Puta a-mu-dza ee vadzu va-mu-enda na ndzana. Puta sp-pl-say that children sp-pl-go to forest ‘Puta said that the children went to the forest.’ b. [CP Tane owu Puta a-mu-dza [CP ee (owu) vadzu va-mu-enda t]]? where foc Puta sp-pl-say that(foc children sp-pl-go ‘Where did Puta say that the children went?’
The same phenomenon as in (19b) can be observed in connection with wh-movement in Austronesian languages, for example in the SVO language Bahasa Indonesia/Malay (see Saddy 1990, Cole and Hermon 1995, 1998). Bahasa Indonesia/Malay is a wh-in situ language which also possesses the wh-ex situ strategy. As pointed out by Saddy (1990: 188) the difference between wh-in situ and wh-ex situ is reflected by absence or presence of overt focus morphology, i.e. the focus-marker yang. Consider the examples (20) which show that Bahasa Indonesia/Malay employs the in-situ as well as the ex-situ strategy for the construction of wh-questions as Tuki and Malagasy. The question word may appear in situ, as in (18a). Alternatively, the question word may appear in the clause-initial scopal position in which the wh-phrase is interpreted as in (18b) and (19b). In this case the focus-marker yang appears:
Malagasy as an optional multiple wh-fronting language 237
(20) Malay a. Sally men-cintai siapa? Sally trans-loves who ‘Who does Sally love?’ b. Siapa yang Sally cintai t? who foc Sally love ‘Who does Sally love?’
Now consider wh-extraction from embedded clauses. As in Tuki (19b), long wh-movement leaves overtly realized focus-markers in the embedded as well as in the matrix clause: (21) Malay [CP Siapa yang Bill harap [CP yang t akan membeli baju untuknya]]? who foc Bill hope foc will buy clothes for-him ‘Who does Bill hope will buy clothes for him?’
To conclude, there is overt morphological evidence for the fact that short as well as long wh-movement is in fact an instance of focus-movement. Languages such as Tuki and Bahasa Indonesia/Malay, in which a focus-marker is located on functional heads, provide evidence that checking of [+Focus]-features is involved if overt wh-movement takes place. Tuki and Malay provide evidence that functional heads, probably C0, I0, or Foc0, in intermediate clauses bear [+Focus]-features of some sort that might be argued to trigger successive-cyclic movement of wh-elements by forcing movement of wh-phrases into the specifier position of a [−Wh]-head. It is plausible to assume that wh-movement in languages such as English also applies in successive-cyclic manner for reasons having to do with [+Focus]-features, although in these languages the focusmarker (or [+Focus]-feature) is only covertly realized. Note that the idea to analyze wh-movement as an instance of focus-movement is not new. This analysis is sometimes traced back to the (semantic) fact that a wh-element is inherently a focus. For example, in a sentence such as I wonder what Susan said, the wh-word is the focus of the question/clause what Susan said since the wh-phrase designates what is not presupposed as known. This assumption can be found in Heny (1971), Rochemont (1978, 1986), Culicover and Rochemont (1983), Culicover and Wilkins (1984), Whitney (1984), Horvath (1986), Tuller (1986), and Bresnan and Mchombo (1987), among others. If we adopt the successive-cyclic movement approach in conjunction with the assumption that movement is triggered solely by feature-checking, we are forced to assume that movement through intermediate positions also applies to
238 Joachim Sabel
satisfy feature-checking. It should be obvious that the examples (19b) and (21) from Malay and Tuki provide the basis for an argument according to which successive-cyclic movement into intermediate positions of a [−Wh]-head is triggered by [+Focus]-features. A natural (technical) implementation of this idea could rely on the assumption that in the case of wh-movement, the embedded C-heads bear some [+Focus]-features that need to be checked: (22) [CP1 [+Wh/+Focus] wh-phrase … [CP2 [+Focus¢] t≤ …[CP3 [+Focus¢] t¢ [IP … t…]]]]
The [+Focus]-feature, like the [+Wh]-feature is assigned to the respective functional heads and wh-phrases in the numeration. If a [+Wh]-feature is realized in C0, a [+Focus]-feature always co-occurs in C0 and in further C0’s embedded under C0. In the following, I will demonstrate how the different variants of single wh-questions in Malagasy can be derived on the basis of this analysis. Overt wh-movement and wh-in situ can be analyzed as resulting from different mechanisms of feature-checking, i.e. as differing in the strength of the features of the functional heads that trigger wh-movement. Let us start with partial wh-movement in Malagasy, as illustrated in (8), repeated here as (23). I assume that in Malagasy, the [+Focus]-feature in C0, assuming CP-recursion (alternatively, this feature heads its own FocP embedded under CP in the sense of Rizzi 1997; see Sabel 2000: 432, fn. 12) is strong and the [+Wh]-feature in C0 is weak. (23) [CP1 Heverin-dRabe [CP2 fa inona no novidin-dRakoto t]]? pres-tt-believe-Rabe that what foc past-tt-buy-Rakoto ‘What does Rabe believe that Rakoto has bought?’
If a [+Wh]-feature is realized in C01, a [+Focus]-feature always co-occurs in C01 and in further C0’s embedded under C01. The wh-phrase obligatorily moves to SpecCP2 in (23) for the same reason that wh-expletive insertion applies, i.e. in order to check the strong [+Focus]-feature in C01. In (23), the empty wh-expletive (see note 3) checks the strong [+Focus]-feature in C01. The weak [+interpretable] [+Wh]-feature in C01 need not be checked via movement.7 Alternatively, if we start with a numeration that does not contain the wh-expletive, the wh-phrase in SpecCP2 moves up to the highest SpecCP position and checks the strong [+Focus]-features as a result of movement. This gives rise to the full wh-movement construction, as shown in (7), repeated here for convenience as (24): (24) [CP1 Inona no heverin-dRabe [CP2 fat¢ novidin-dRakoto t]]? what foc pres-tt-believe-Rabe that past-tt-buy Rakoto ‘What does Rabe believe that Rakoto has bought?’
Malagasy as an optional multiple wh-fronting language 239
I assume that the [+Wh]-feature in Malagasy is always weak and the [+Focus]feature is either strong or weak. The realization of the strong [+Focus]-feature in this language coincides with the appearance of no in the matrix sentence. Keenan (1976) argues that no is a focus-particle in Malagasy and Pensalfini (1995) assumes that no is located in C0 (see Chung 1998: Ch. 7 for a similar analysis of the analogous construction in Chamorro, another verb-initial Austronesian language). I conclude that wh-movement is only obligatory in Malagasy if the strong [+Focus]-feature is selected from the lexicon and no is realized. The [+Focus]-feature is not realized in the wh-in situ example of (9), repeated here as (25): (25) [CP Heverin-dRabe [CP fa nividy inona Rakoto]]? pres-tt-believe-Rabe that past-at-buy what Rakoto ‘What does Rabe believe that Rakoto has bought?’
The fact that the strong [+Focus]-feature may be realized or not has the consequence that Malagasy may have wh-in situ as well as overt wh-movement (see Sabel 2000, where a typology of wh-movement languages is derived on the basis of this analysis).
3. Multiple wh-questions in Malagasy So far we have seen that Malagasy allows for wh-words to occur in three positions in wh-questions. They occur in SpecCP where they are interpreted (full wh-movement), they are partially wh-moved, or they are in situ. Let us now consider multiple wh-questions. Three types of multiple questions exist in Malagasy. The first possibility is that one wh-phrase is moved to sentence-initial position and the other wh-phrase(s) rest in situ. A second possibility is that every wh-phrase remains in situ. These constructions will be discussed in Section 3.1. The third possibility, Malagasy also allows for multiple wh-fronting. The properties of this construction will be discussed in Section 3.2. 3.1 Multiple wh-questions in Malagasy with wh-elements in situ Multiple questions are possible as long as the three restrictions on the possible occurrences of wh-elements in situ and the ‘subjects-only’ restriction are respected. For example, given that a wh-phrase cannot stay in SpecIP, the examples in (26) with multiple wh-elements are ruled out. However, (26b) is
240 Joachim Sabel
perfect as a multiple question (without an echo reading) if wh-fronting of the wh-element in SpecIP, iza ‘who’, takes place (27a). (26) a. *Nividy ny vary taiza iza? past-at-buy the rice past-where who ‘Who bought the rice where?’ b. *Nividy inona iza? past-at-buy what who ‘Who bought what?’ inona t? Iza no nividy who foc past-at-buy what b. Inona no novidin’ iza t? what foc past-tt-buy who Both: ‘Who bought what?’
(27) a.
The examples in (27) show that in contrast to languages with typical subject/ object extraction asymmetries such as English, Malagasy does not show Superiority effects with wh-arguments. Superiority effects are also absent with long extraction, as illustrated in (28b): (28) a.
[CP Heverin’ iza [CP fa mividy inona Rakoto]]? pres-tt-believe who that pres-at-buy what Rakoto ‘Who thinks that Rakoto buys what?’ b. [CP Inona no heverin’ iza [CP fa vidin-dRakoto t]]? what foc believe who that pres-tt-buy-Rakoto ‘Who thinks that Rakoto buys what?’
As expected, non-referential adjuncts are possible in multiple wh-questions as long as they are located in SpecCP. This is illustrated in (29)–(30). (29) a.
Ahoana no andrahoan-dRakoto inona t? how foc pres-ct-cook-Rakoto what ‘How does Rakoto cook what?’ b. *Inona no andrahoina (manao) ahoana Rakoto t? what foc pres-tt-cook how Rakoto ‘What does Rakoto cook how?’
(30) a.
Nahoana no mividy inona ianao? why foc pres-at-buy what you ‘Why do you buy what?’ b. *Inona no vidinao nahoana t? what foc pres-tt-buy-you why ‘What do you buy why?’
Malagasy as an optional multiple wh-fronting language 241
Now consider (31). No ungrammaticality is observed in this construction, where the thematic wh-subject is in situ (in English, this construction is impossible). (31) Nahoana no vidin’ iza ny vary? why foc pres-tt-buy who the rice ‘Why does who buy the rice?’
The examples with single and multiple wh-questions show that wh-constructions in Malagasy are subject to different constraints than wh-constructions in languages such as English. Single and multiple wh-questions in Malagasy do not show the typical ECP-, Shortest Move- or MLC-effects. Question formation only obeys the ‘subjects-only’ restriction and the three additional factors that determine wh-in situ. I conclude that wh-in situ in Malagasy involves unselective binding (see Sabel 2002b for further discussion), as has been assumed to be generally the case in wh-in situ constructions (Chomsky 1995). Let us now turn to other variants of multiple wh-questions in Malagasy. Given that Malagasy is an optional wh-in situ language, it is expected that a second possibility for forming multiple questions exists, i.e. it should be possible to leave all wh-words in situ. As shown in (32), this prediction is borne out: (32) Anasan’ iza inona ny savony? pres-ct-wash who what the soap ‘Who washes what with the soap?’
3.2 Optional multiple wh-fronting in Malagasy As a third possibility, Malagasy allows for multiple wh-fronting. As shown in (33), the elements aiza ‘where’, ianao ‘you’, and ny vary ‘the rice’ may be moved together in sentence-initial position: (33) a.
Aiza iza no mividy ny vary? where who prt pres-at-buy the rice ‘Where does who buy the rice?’ b. Aiza (ny) inona no vidinao? where (the) what prt pres-tt-buy-you ‘Where do you buy what?’
The same is possible with other adjuncts, such as nahoana ‘why’, oviana ‘when’: (34) a.
Nahoana iza no mividy ny vary t t? why who prt pres-at-buy the rice
242 Joachim Sabel
b. Oviana (ny) inona no nosasan’ ny reny t amin when (the) what foc past-tt-wash the mother with
ny savony t? the soap Given that with respect to argument extraction only the structural subject may be extracted, multiple fronting of more than one argument is excluded in Malagasy. As shown in (35)–(36), two arguments cannot be extracted irrespective of the voice of the verb, and irrespective of the order of the fronted elements: (35) a. *Iza inona no mividy? who what prt pres-at-buy b. *Inona iza no mividy? what who prt pres-at-buy ‘What does who buy?’ (36) a. *Iza inona no vidin? who what prt pres-tt-buy b. *Inona iza no vidin? what who prt pres-tt-buy ‘What did who buy?’
Multiple wh-fronting has been analyzed as resulting from multiple wh-scrambling (see for example Takahashi 1993) or as movement to multiple specifiers (Richards 1997). In the following, I will argue that these analyses cannot be applied to Malagasy. Multiple fronting in Malagasy patterns with multiple wh-fronting in languages such as Bulgarian and Romanian, where all wh-phrases are located in one SpecCP (Rudin 1988). Firstly, multiple wh-fronting in Malagasy is not the result of (reconstructable) wh-scrambling as it is the case in languages such as for example Japanese, Persian or Hindi. Malagasy, in contrast to these languages, is not a scrambling language. The word order of subject, direct and indirect object (or obliques) is generally rigid (Keenan 1976:250) (although subject to a definiteness restriction). The examples (37b–c) that are derived from (37a) by scrambling are ungrammatical: (37) a.
Manasa ny lamba amin’ ny savony ny reny. pres-at-wash the clothes with the soap the mother-nom ‘The mother washes the clothes with the soap.’ b. *Manasa ny reny ny lamba amin’ ny savony. pres-at-wash the mother the clothes with the soap ‘The mother washes the clothes with the soap.’
Malagasy as an optional multiple wh-fronting language 243
c. *Manasa amin’ ny savony ny reny ny lamba. pres-at-wash with the soap the mother the clothes ‘The mother washes the clothes with the soap.’
To conclude, multiple wh-fronting in Malagasy is not the result of scrambling. Second, multiple wh-fronting is not derived from movement to multiple specifiers. The multiple specifier analysis falsely predicts that long movement across a short fronted element would be possible. The a- and b-examples show that long wh-movement across a partially moved wh-element is impossible, irrespective of the voice of the matrix verb. The c- and d-examples illustrate that the corresponding multiple question is possible if the wh-element that is long wh-moved in the a- and b-examples remains in situ: (38) a. *Oviana no mihevitra i Piera [CP fa iza no when foc pres-at-believe art Piera that who foc niverina t t]? past-at-return b. *Oviana no heverin i Piera [CP fa iza no when foc pres-tt-believe art Piera that who foc niverina t t]? past-at-return c. Mihevitra i Piera [CP fa iza no niverina oviana t]? pres-at-believe art Piera that who foc past-at-return when oviana t]? d. Heverin i Piera [CP fa iza no niverina pres-tt-believe art Piera that who foc past-at-return when All four: ‘When does P. believe that who did return?’ (39) a. *Aiza no mihevitra i Piera [CP fa inona no where foc pres-at-believe art Piera that what foc novidinao t t]? past-tt-buy-you b. *Aiza no heverin’ i Piera [CP fa inona no where foc pres-tt-believe art Piera that what foc novidinao t t]? past-tt-buy-you c. Mihevitra i Piera [CP fa inona no novidinao aizat]? pres-at-believe art Piera that what foc past-tt-buy-you where d. Heverin’ i Piera [CP fa inona no novidinao aiza t]? pres-tt-believe art Piera that what foc past-tt-buy-you where All four: ‘Where does Peter believe that you bought what?’
244 Joachim Sabel
If multiple fronted elements would land in different specifier positions, we would expect that long extraction as in (38a–b) and (39a–b) may proceed via a vacant landing site in the embedded clause. Hence, I conclude that multiple fronting in Malagasy is not derived by movement to multiple specifiers. In the remainder of this article, I will argue that multiple wh-elements in Malagasy undergo cluster formation and occupy one sentence-initial SpecCP position, similar to languages such as Bulgarian and Romanian. However, in contrast to languages such as Bulgarian and Romanian, Malagasy is an optional multiple wh-fronting language. The central claim of my analysis of single wh-questions in Malagasy already discussed in Section 2.4 is that wh-movement is triggered by the need to check strong [+Focus]-features. In the following, I will argue that this analysis provides a natural account of multiple wh-fronting in Malagasy, assuming that complex (multi-segmental) categories are built, i.e. “cluster formation” applies. I assume the Cluster-Hypothesis (CH)(Sabel 2001): (40) Cluster-Hypothesis A feature F which is attracted by K attracts a feature of the same type F.
(40) is only a descriptive generalization. In fact, (40) interacts with Attract F (see Chomsky 1995:Ch. 4): (41) Attract F K attracts F if F is the closest feature that can enter into a checking relation with a sublabel of K. (42) Closeness β is closer to K than α if β c-commands α.
Given the definitions in (41)–(42) and given a structure like […α…β…γ…] where α asymmetrically c-commands β and β asymmetrically c-commands γ, α cannot attract γ because β is closer to α. However, β can attract γ and α can attract the complex [β β γ] (or [β γ β]) if, as stated in (40), the feature F of β, which is attracted by α, may itself attract the analogous feature of γ. Multiple wh-fronting in Malagasy provides evidence for cluster formation as a result of (40). On the basis of the CH it can be argued that wh-elements in multiple wh-questions in Bulgarian, Romanian, and Malagasy attract wh-phrases overtly. The hypothesis that wh-clusters are formed in multiple questions, can be illustrated with examples from Bulgarian (43) (but the same holds for Romanian, see Comorovski 1986, 1989). As is well known, the characteristic property of these multiple fronting languages is that every wh-word has to be fronted in
Malagasy as an optional multiple wh-fronting language 245
multiple wh-questions. This can be seen from the contrast between (43a) and (43b) (Rudin 1988). (43) a.
Koj ku¯de misliš [CP t¢ ˇcet e otišu¯lt]? who where think-2sg that has gone ‘Who do you think (that) went where?’ b. *Koj misli [CP t¢ ˇcet e otišu¯l ku¯de]? who think-2sg that has gone where
According to Rudin (1988), all of the wh-phrases in multiple wh-constructions move to SpecCP in (43a) (see also Comorovski 1986, 1989, Boškovic´ 1997). The evidence for this analysis is traced back to the fact that, for example, in Bulgarian long extraction of all of the wh-phrases is possible (43a), in contrast to other multiple wh-fronting languages, such as Polish. As can be seen in example (44) from Polish, one of the wh-elements in (44) is moved to the embedded SpecCP position and the other is adjoined to the embedded IP, only the former wh-phrase may move from the embedded to the matrix SpecCP (see Rudin 1988, Sabel 2002a): (44) a.
Co Maria chce [CP t¢ ˙zeby [IP komu [IP Janek kupit t]]]? what Maria wants that to.whom Janek buy b. *Co komu Maria chce [CP z˙eby Janek kupit t]? what to.whom Maria wants that Janek buy ‘What does Maria want Janek to buy for whom?’
Turning to Malagasy, we observe that it behaves like Bulgarian and Romanian. Long extraction of multiple wh-elements is possible, as shown in (45a), derived from (45c). (45b) shows that long multiple movement also applies with the corresponding non-wh-phrases: (45) a.
[CP Amin’ ny inona ny inona [C¢ no [IP antenain-dRasoa [CP fa with the what the what foc pres-tt-hope-Rasoa that sasanao t t]]]]? pres-tt-wash-you ‘What hopes Rasoa that you washes with what?’ b. [CP Amin’ ity savony ity ny lamba [C¢ no [TP with this soap this the clothes foc antenain-dRasoa [CP fa sasanaot t]]]]. pres-tt-hope-Rasoa that pres-tt-wash-you ‘Rasoa hopes that you wash the clothes with soap.’
246 Joachim Sabel
c.
[CP Manantena Rasoa [CP fa manasa ny lamba amin’ pres-at-hope Rasoa that pres-at-wash the clothes with ity savony ity ianao]]. this soap this you ‘Rasoa hopes that you wash the clothes with soap.’
Finally, multiple wh-cluster fronting is subject to Superiority in Bulgarian (46)–(47) and Romanian, in contrast to multiple wh-fronting in languages such as Polish (but see Sabel 2001; fn. 4, Boškovic´ 2002, this volume, and the discussion around exx. (10)–(13) in Boeckx and Grohmann, this volume). (46) a.
Koj kogo t vidja t? who whom saw ‘Who saw whom?’ b. *Kogo koj t vidja t?
(47) a.
Koj kak t udari Ivant? who how hit Ivan ‘Who hit Ivan how?’ b. *Kak koj t udari Ivan t?
Malagasy patterns with Bulgarian, as shown in (48)–(49): (48) a.
Aiza iza no mividy ny vary? where who prt pres-at-buy the rice ‘Where does who buy the rice?’ ny vary? b. *Iza aiza no mividy who where prt pres-at-buy the rice ‘Where does who buy the rice?’
(49) a.
Aiza (ny) inona no vidinao? where (the) what prt pres-tt-buy-you ‘Where do you buy what?’ b. *Inona aiza no vidinao? what where prt pres-tt-buy-you ‘Where do you buy what?’
Superiority effects are also found with adjuncts such as nahoana ‘why’, oviana ‘when’: (50) a.
Nahoana iza no mividy ny vary? why who prt pres-at-buy the rice ‘Why buys who the rice?’
Malagasy as an optional multiple wh-fronting language 247
b. *Iza nahoana no mividy ny vary? who why prt pres-at-buy the rice (51) a.
Oviana (ny) inona no nosasan’ ny reny t amin ny savony? when the what foc past-tt-wash the mother with the soap ‘What washes the mother when with the soap?’ b. *Inona oviana no nosasan’ ny reny t amin ny savony? what when foc past-tt-wash the mother with the soap
A wh-phrase may attract other wh-phrases in Bulgarian (see also Kraskow 1990, Saito 1992, Abe 1993a, 1993b, Ackema and Neelemann 1998, Grewendorf and Sabel 1996, 1999, Sabel 2001 for similar suggestions). Examples such as (43a) and (45a-b) may plausibly be derived by the CH, i.e. by assuming that some feature F of koj ‘who’ and inona ‘what’ (and ny lamba) that is attracted by the same feature F in C0 attracts the similar feature F of ku¯de ‘where’ and amin’ny inona ‘with what’ (and aminm’ ity savony ity). A similar derivation with short extraction can take place when two wh-arguments are present, as in (46a), or as in (47a) with a non-referential adjunct and an argument. In (46), the relevant feature of koj ‘who’ (I leave open whether it is the [+Wh]- or [+Focus]-feature in Bulgarian) attracts the corresponding feature of kogo ‘whom’ and then the wh-cluster is attracted by the third occurrence of feature F in C0. Furthermore, in (47) the wh-adjunct is attracted by the wh-argument, then the so-formed wh-cluster is attracted by the relevant feature in C0. Assuming that the adjunct in this example is base-generated in a position lower than the argument only (47a) can be derived according to (40). In Malagasy, the feature that triggers wh-movement is the strong [+Focus]-feature. In (45a), (49a), and (51a) the strong [+Focus]-feature of inona, and in (48a) and (50a) the strong [+Focus]-feature of iza attracts the other wh-elements. Then the wh-cluster is attracted by the [+Focus]-feature in C0. The relevant part of the derivation of (43a), (46a), (47a) is depicted in (52a) (XP2= kogo in (46a) and kak in (47a)), and the important part of the derivation of (45a-b), (48a), (49a), (50a), (51a) is depicted in (52b). As can be seen from (52), the wh-cluster is formed in SpecIP (TP) before wh-movement to SpecCP applies.
248 Joachim Sabel
(52) a.
CP C¢ C
IP NP1
NP1
b.
IP XP2
... t2 ...
CP C¢ C
IP I¢ ... t2 ...
NP1 XP2
NP1
The following questions arise at this point: given the condition according to which adjunction is only possible to non-arguments for θ-theoretic reasons (see Chomsky 1986, 1995: 330), why is it possible to adjoin wh-words to wh-arguments? Secondly, does XP2 c-command its trace in (52), i.e. how is it possible that XP2 is attracted by the [+Focus]-feature of (DP1 or) NP1? Let us begin with the first question. Given the VP-internal subject hypothesis and the assumption that object and subject NPs have to raise out of their θpositions into positions for structural Case-assignment, i.e. SpecTP/IP for subjects and SpecvP in the case of objects, this problem does in fact not arise since NP1, the argument that represents the target of adjunction, is located in a non-θ-position. Concerning the second question, for reasons that will be discussed later, let us assume that the [+Focus]-feature is inherently associated with D0 in Malagasy and that it projects up to DP (this feature-percolation/projection has to be assumed to account for pied-piping possibilities anyway). Then D attracts via its projection path, i.e. DP is the attractor in examples with wh-clusters. Given that the higher DP c-commands the lower one, Attract F
Malagasy as an optional multiple wh-fronting language 249
may apply (this also opens up the possibility that certain non-DPs may function as attractors). Hence, the [+Focus]-feature that is realized on the maximal projection of D attracts another [+Focus]-feature and this gives rise to the phenomenon of XP-cluster formation in Malagasy. Note that the c-command situation is exactly as with head movement. No c-command problem arises if the immediately dominating segment of the cluster that is formed via adjunction does not block c-command (see Baker 1988: 449, fn. 10, Chomsky and Lasnik 1993: 522 for discussion). Note, furthermore, that adjunction to all wh-elements (adjuncts and arguments) in SpecCP is excluded if we assume the copy theory of movement (Chomsky 1995) and the Uniformity Condition on Chains (UC), i.e. ‘chains are uniform’. The UC allows that something is adjoined to the head of a trivial (one-membered) chain. On the other hand, given the copy theory of movement, the UC excludes adjunction to the head of a non-trivial chain, since this would render the head of the chain distinct from its other parts resulting in a non-uniform chain. Therefore, the UC excludes adjunction of wh-phrases to wh-phrases in SpecCP. On the other hand, A-movement is not analyzed as copy movement in Lasnik (1999) for reasons having mainly to do with binding theory. Thus, it follows that elements in A-positions are potential targets for cluster formation. In addition, no problem with the UC arises concerning head movement, as long as adjunction applies to the head of a trivial chain. To summarize, it follows that adjunction to an element in SpecCP is not permissible: such an adjunction would render the chain of this element non-uniform since its copy does not have an element adjoined to it. As far as the well-formed examples already discussed are concerned, the UC is respected. It is obvious that the formation of clusters is due to independent properties in the language in question. For example, the question arises as to what parametric property is responsible for the possibility of wh-cluster formation. Takahashi (1994) has argued that cluster formation is only possible in scrambling languages. Malagasy is a clear counterexample to this generalization. The question remains where else the locus of parameterization can be found. A promising alternative seems to be following (but I will not go into a detailed examination here). The difference between languages with respect to multiple wh-constructions is often argued to be due to different morphological properties of wh-phrases (see Cheng 1991, Watanabe 1991: 52ff., Grewendorf and Sabel 1996, 1999, Grewendorf 2001 for dicussion). Following Rudin (1988), I have argued that in overt multiple fronting languages such as Bulgarian and Malagasy, the fronted wh-elements constitute
250 Joachim Sabel
a complex wh-phrase in SpecCP, but contrary to Rudin, I assume that this complex has to be formed prior to wh-movement to SpecCP. Given this analysis, we correctly predict the properties of multiple wh-questions in languages such as Bulgarian, i.e. (i) fixed order of the fronted wh-elements, (ii) the fact that nothing may intervene between the fronted wh-phrases and in addition, (iii) that it is possible in these languages to have long distance movement of multiple wh-phrases as shown in (43), (45), in contrast to languages such as Polish, cf. (44). As discussed in Section 3.1, multiple questions in Malagasy may also take the form of single wh-question word movement. In this case, I assume that the strong [+Focus]-feature in D0 is not selected but only in C0. Finally, in order to derive multiple wh-questions in Malagasy in which every wh-word is in situ, I assume that the strong [+Focus]-feature is neither selected in D0 nor in C0.
4. Summary In this article, I have argued that wh-movement in Malagasy is triggered by optional selected strong [+Focus]-features. The three different wh-movement constructions (partial wh-movement, full wh-movement, and wh-in situ) have been shown to be derivable with this assumption. Single full wh-movement and partial wh-movement constructions are the result of selected strong [+Focus] features in C0 in Malagasy. Wh-in-situ constructions result from the use of weak [+Focus]-features in C0. I have further argued that this feature-based analysis provides a natural account for all variants of multiple wh-questions in Malagasy. For example, it accounts for the formation of complex categories i.e., for the process of “cluster formation” in Malagasy multiple (wh-) fronting constructions.
Notes * This paper presents parts of ongoing work on wh-questions in Malagasy. Special thanks to Hanitry-Al Gerull, Roger-Bruno Rabenilaina, and Elisabeth Ravaoarimalala, and to the editors Cedric Boeckx and Kleanthes K. Grohmann. 1. The paradigm in (1) is also sometimes called the “topic-” or “focus-” construction of the West Austronesian languages. In some analyses, the sentence-final element is analyzed as a topic, in others as the focused element. Sometimes the sentences in (1b–c) are called “passive” constructions, derived from the “active” construction in (1a). This passive however
Malagasy as an optional multiple wh-fronting language
differs from passive in languages such as English and German in so far as no Case absorption takes place. As can be seen from (1b–c), the thematic object in (1b) or the oblique argument in (1c) becomes the nominative subject. In contrast to normal passive constructions the agent ny reny ‘the mother’ does not appear in an optional by-phrase. I will be using the traditional terminology for the Malagasy voice system throughout this paper. The abbreviations used in the text are the following: at (Agent Topic Marker), tt (Theme Topic Marker), ct (Circumstantial Topic Marker), foc (Focus Particle), acc (Accusative), nom (Nominative), pres (Present), s (Singular), pass (Passive), prt (particle). 2. In Sabel (2002b) it is argued that in Malagasy-type languages, θ-roles are not linked to different structural positions with respect to a θ-hierachy and every argument may be basegenerated as a sister of the verb. Malagasy differs in this respect from English, where all θ-roles are linked to different structural positions in accordance with a θ-hierachy. From this difference, the following descriptive generalization emerges: if linking is not determined by a θ-hierachy in a language, then extraction is possible only from the complement position in this language. 3. I assume that an empty expletive, i.e. the equivalent of the wh-expletive was ‘what’ in German partial wh-movement constructions (se Sabel 2000), occupies SpecCP1 the scopal position of the ‘true’ wh-phrase in the embedded sentence. The overt wh-expletive is often realized as the equivalent of the bare accusative wh-phrase (for example was ‘what’ in German), although some Slavic languages such as, for example, Polish use jak ‘how’ (Willim 1989: 113ff.). Like expletives in A-chains, the wh-expletive need not be overtly realized. In some languages such as, for example, Albanian, Iraqi Arabic, Kikuyu, and Malay we find covert variants of wh-expletives like in Malagasy (see also Anyadi and Tamrazian 1993, Cole and Hermon 1998). 4. In examples such as Iza ianao? ‘Who are you?’, the particle is not realized because the wh-word is the nominal predicate in situ. 5. That non-referential wh-adjuncts are not licensed in situ is probably related to the fact that they lack a position for a variable, as has been pointed out by several authors (see Tsai 1994, Chomsky 1995: 386, note 65 among others). 6. However, wh-scrambling in languages such as Hindi, Persian, and Japanese is triggered by different factors; see Sabel (2003) for discussion of the features triggering wh-scrambling in Japanese. 7. I leave open whether this checking operation results from Move, or from Merge and Move. Concerning the latter, see Chomsky (1995: Ch. 4) for the possibility to check strong features in C0 via the operation Merge.
References Abe, J. 1993a. Binding conditions and scrambling without A/A¢ distinction. Doctoral dissertation, University of Connecticut, Storrs.
251
252 Joachim Sabel
Abe, J. 1993b. “The Economy Condition on dependency.” In Proceedings of the Tenth Eastern States Conference on Linguistics, A. Kathol and M. Bernstein, (eds), 1–12. Ithaca, NY: DMLL Publications. Ackema, P. and Neeleman, A. 1998. “Optimal questions.” Natural Language and Linguistic Theory 16: 443–490. Anyadi, S. and Tamrazian, A. 1993. “Wh-movement in Armenian and Ruhr German.” UCL Working Papers in Linguistics 5: 1–22. Baker, M. 1988. Incorporation. A Theory of Grammatical Function Changing. Chicago, Ill.: The University of Chicago Press. Biloa, E. 1995. Functional Categories and the Syntax of Focus in Tuki. Munich: Lincom. Boeckx, C. and K. K. Grohmann. This volume. “Introduction.” 1–15. Boškovic´, Ž. 1997. Multiple wh-fronting and economy of derivation. Ms., University of Connecticut, Storrs. Boškovic´, Ž. 2002. “On multiple wh-fronting.” Linguistic Inquiry 33: 351–383. Boškovic´, Ž. This volume. “On wh-islands and obligatory wh-movement contexts in South Slavic.” 27–50. Bresnan, J. and Mchombo, S. A. 1987. “Topic, pronoun, and agreement in Chichewa.” Language 63: 741–782. Cheng, L. L.-S. 1991. On the typology of wh-questions. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Chomsky, N. 1986. Barriers. Cambridge, Mass.: MIT Press. Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. 2000. “Minimalist Inquiries: The framework.” R. Martin, D. Michels and J. Uriagereka (eds), 89–155. Canbridge, Mass.: MIT Press. Chomsky, N. and Lasnik, H. 1993. “The theory of principles and parameters.” In Syntax: An International Handbook of Contemporary Research, Vol. 1, J. Jacobs, A. von Stechow, W. Sternefeld and T. Vennemann (eds), 506–569. Berlin: Walter de Gruyter. Chung, S. 1998. The Design of Agreement. Evidence from Chamorro. Chicago, Ill.: The University of Chicago Press. Cole, P. and Hermon, G. 1995. “Is wh-in-situ really in-situ? Evidence from Malay and Chinese.” In Proceedings of the Thirteenth West Coast Conference on Formal Linguistics, R. Aranovich, W. Byrne, S. Preuss and M. Senturia (eds), 189–204. Stanford, Calif.: CSLI. Cole, P. and Hermon, G. 1998. The typology of Wh-movement. Wh-questions in Malay. Ms., University of Delaware, Newark. Comorovski, I. 1986. “Multiple wh-movement in Romanian.” Linguistic Inquiry 17: 171–177. Comorovski, I. 1989. Discourse and the syntax of multiple constituent questions. Doctoral dissertation, Cornell University, Ithaca, NY. Culicover, P.W. and Rochemont, M. 1983. “Stress and focus in English.” Language 59: 123–165. Culicover P. W. and Wilkins, W. 1984. Locality in Linguistic Theory. New York: Academic Press. Grewendorf, G. 2001. “Multiple wh-fronting.” Linguistic Inquiry 32: 87–122. Grewendorf, G. and Sabel, J. 1996. “Multiple specifiers and the theory of adjunction: On scrambling in German and Japanese.” Sprachwissenschaft in Frankfurt. Arbeitspapier 16. Frankfurt am Main: Institut für Deutsche Sprache und Literatur II.
Malagasy as an optional multiple wh-fronting language 253
Grewendorf G. and Sabel, J. 1999. “Scrambling in German and Japanese: Adjunction versus multiple specifiers.” Natural Language and Linguistic Theory 16: 1–65. Grohmann, K. K. This volume. “German is a multiple wh-fronting language!” 99–130. Guilfoyle, E, Hung, H. and Travis, L. deMena. 1992. “Spec of IP and Spec of VP: Two subjects in Austronesian languages.” Natural Language and Linguistic Theory 10: 375–414. Heny, F. 1971. Focus and relativization in Bali. Ms., University of California, Los Angeles. Horvath, J. 1986. Focus in the Theory of Grammar and the Syntax of Hungarian. Dordrecht: Foris. Keenan, E. L. 1976. “Remarkable Subjects in Malagasy.” In Subject and Topic, C. Li (ed), 249–301. New York: Academic Press. Keenan, E. L. 2000. “Morphology is structure: A Malagasy test case.” In Formal Issues in Austronesian Linguistics, I. Paul, V. Phillips and L. deMena Travis (eds), 27–48. Dordrecht: Kluwer. Kraskow, T. E. 1990. “Discourse-linking and the wh-island extraction asymmetry.” In Proceedings of the Seventh Eastern States Conference on Linguistics, Y. No and M. Libucha, (eds), 154–165. Ithaca, NY: DMLL Publications. Lasnik, H. 1999. “Chains of Arguments.” In Working Minimalism, S. D. Epstein and N. Hornstein (eds), 188–215. Cambridge, Mass.: The MIT Press. Pensalfini, R. 1995. “Malagasy phrase structure and the LCA.” In Papers on Minimalist Syntax. MIT Working Papers in Linguistics 27, R. Pensalfini and H. Ura (eds), 209–221. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). Reinhart, T. 1995. “Interface Strategies.” OTS Working Papers. Utrecht: Utrecht Faculteit Letteren. Richards, N. 1997. What moves where when in which language? Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Richards, N. 2000. “Another look at Tagalog subjects.” In Formal Issues in Austronesian Linguistics, I. Paul, V. Phillips and L. deMena Travis (eds), 105–116. Dordrecht: Kluwer. Rizzi, L. 1997. “The fine structure of the left periphery.” In Elements of Grammar: Handbook of Generative Syntax, L. Haegeman (ed), 281–337. Dordrecht: Kluwer. Rochemont, M. 1978. A theory of stylistic rules in English. Doctoral dissertation, University of Massachusetts, Amherst. [Published by New York: Garland, 1985.] Rochemont, M. 1986. Focus in Generative Grammar. Amsterdam: John Benjamins. Rudin, C. 1988. “On multiple questions and multiple wh-fronting.” Natural Language and Linguistic Theory 6: 445–501. Sabel, J. 1998. Principles and parameters of wh-movement. Habilitationsschrift, University of Frankfurt. Sabel, J. 2000. “Partial wh-movement and the typology of wh-questions.” In Wh-Scope Marking, U. Lutz, G. Müller and A. von Stechow (eds), 409–446. Amsterdam: John Benjamins. Sabel, J. 2001. Deriving multiple head and phrasal movement: The cluster hypothesis. Linguistic Inquiry 32: 532–547. Sabel, J. 2002a. “Intermediate traces, reconstruction and locality effects.” In Theoretical Approaches to Universals, A. Alexiadou (ed), 259–313. Amsterdam: John Benjamins. Sabel, J. 2002b. “Wh-questions and extraction asymmetries in Malagasy.” In A. Rackowski and N. Richards (eds), 309–324, Proceedings of the 8th Conference of the Austronesian Formal Linguistics Association (AFLA VIII). MIT Working Papers in Linguistics 44. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT).
254 Joachim Sabel
Sabel, J. 2003. “Wh-questions in Japanese: Scrambling, reconstruction, and wh-movement.” Linguistic Analysis 31: 1–41. Saddy, D. 1990. “Wh-scope mechanisms in Bahasa Indonesia.” In L.-L. S. Cheng and H. Dermidache (eds), MIT Working Papers in Linguistics 15, 183–218. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). Saito, M. 1992. “Long distance scrambling in Japanese.” Journal of East Asian Linguistics 1: 69–118. Takahashi, D. 1993. “Movement of wh-phrases in Japanese.” Natural Language and Linguistic Theory 11: 655–678. Takahashi, D. 1994. “Sluicing in Japanese.” Journal of East Asian Linguistics 3: 265–300. Tsai, W.-T. D. 1994. On economizing the theory of A-bar dependencies. Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Tuller, L. 1986. Bijective relations in Universal Grammar and the syntax of Hausa. Doctoral dissertation, University of California, Los Angeles. Watanabe, A. 1991. “Wh-in-situ, Subjacency, and Chain Formation.” MIT Occasional Papers in Linguistics 2. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). Whitney, R. 1984. The syntax and interpretation of A-bar adjunction. Doctoral dissertation, University of Washington, Seattle. Willim, E. 1989. On Word Order: A Government-Binding Study of English and Polish. Kraków: Wydawnictwo Naukowe Uniwersytetu Jagiellonskiego.
Multiple wh-fronting in Serbo-Croatian matrix questions and the matrix sluicing construction Sandra Stjepanovic´ West Virginia University
1.
Introduction
In this paper I examine two problems with respect to multiple wh-fronting in Serbo-Croatian. One problem has to do with the positions to which wh-phrases are moving and the driving force behind this movement. As is well-known, in the majority of cases, wh-phrases in SC cannot stay in their in situ position, and have to be in some position preceding the verb, as illustrated in (1). (1) a. *Ko kupuje šta gdje? who buys what where ‘Who buys what where?’ b. *Ko šta kupuje gdje? c. Ko šta gdje kupuje?
Given this state of affairs, two question immediately arise: what exactly are the positions to which wh-phrases are moving and what the driving force behind this movement is. Rudin (1988) argues that the wh-phrase which is the first in the linear order moves to SpecCP, while others are adjoined to IP. The movement of the first wh-phrase can be taken to be an instance of familiar wh-movement to SpecCP for checking of a wh-feature in C. As far as the second wh-phrase is concerned, however, a question immediately arises as to what the driving force for the movement of this wh-phrase is. This question becomes even more interesting in the light of Boškovic´’s (1997b, 1998b) argument that in certain cases, even the first wh-phrase does not move to SpecCP overtly, although it is fronted. An attempt to find an answer to this question will reveal that there is a
256 Sandra Stjepanovic´
parallelism between wh-phrases and identificationally focused material (in the sense of É. Kiss 1998)1 with respect to the positions they occupy in the sentence. It will be shown that identificationally focused material can undergo overt movement to certain syntactic positions, as illustrated in (2).2 PETRA Marija voli. Petar-acc Marija-nom loves ‘Marija loves Petar.’ b. Marija PETRA voli. Marija-nom Petar-acc loves ‘Marija loves Petar.’
(2) a.
In these sentences, the focused element is in an immediately preverbal or sentence initial position, it bears an emphatic stress and it is necessarily the only focus of the sentence. I will first show that such sentences involve movement of identificationally focused elements to the projections where this focus is licensed. Then I will argue that multiple wh-fronting is a sub-case of this focus movement. In this sense, SC offers support to the often noted observation that if a language marks focus syntactically (i.e. if the focused material has to occur in (a) particular position(s) in the sentence), these positions also host wh-phrases. Some of the languages having this property are Somali, Chadic, Aghem, Basque, Hungarian, Omaha, Quetchua, Greek, and Finnish (see, among others, Horvath 1986, Rochemont 1986, and papers in É. Kiss 1995), and Romanian (Göbbel 1998). The other problem I will be concerned with is the presence of Superiority effects in the sentences involving sluicing and multiple wh-phrase remnants. I will call this type of sluicing multiple sluicing. It is a well-known fact about SC multiple wh-fronting that it does not exhibit Superiority effects in short distance matrix questions (Rudin 1988, Boškovic´ 1997b, 1998b), as (3) shows: (3) a.
Ko koga voli? who whom loves ‘Who loves whom?’ b. Koga ko voli?
However, this freedom of wh-phrase ordering is not manifested in the sentences such as (4b) and (4d), which involve multiple sluicing. (4) a.
Neko voli nekog. somebody loves somebody ‘Somebody loves somebody.’
Multiple wh-fronting in Serbo-Croatian matrix questions 257
b. Ko koga? who whom ‘Who (loves) whom?’ c. *Koga ko?
Examining the interaction between Superiority and multiple sluicing will reveal some interesting facts about this ellipsis phenomenon, and will shed more light on the nature of multiple wh-phrase fronting. 2. Multiple wh-fronting in SC Rudin (1988) shows that there are two types of multiple wh-fronting languages. One type is the Bulgarian type, which includes languages such as Bulgarian and Romanian. Rudin argues that in this type of languages all fronted wh-phrases are in SpecCP, forming a constituent, as in (5a). The other type of languages is the Serbo-Croatian type, which includes languages such as SC, Czech, and Polish. According to Rudin, in SC type of languages, the fronted wh-phrases do not form a constituent; only the first wh-phrase is located in SpecCP, while other fronted wh-phrases are adjoined to IP, as shown in (5b). (5) a.
[CP Koj kogo [IP vižda]] who whom sees ‘Who sees whom?’ b. [CP Ko [IP koga [vidi]]] who whom [sees ‘Who sees whom?’
(Bulgarian)
(SC)
One of Rudin’s arguments for this conclusion concerns the fact that non-wh material cannot split fronted wh-phrases in Bulgarian, while it can in SC, as shown in (6). (6) a.
Zavisi od tova, koj kogo prˇv e udaril. depends on it who whom first is hit ‘It depends on who whom hits first.’ b. *Zavisi od tova, koj prˇv kogo e udaril. c. Zavisi od toga ko koga prvi udari. d. Zavisi od toga ko prvi koga udari.
(Bulgarian)
(SC)
Another difference between the two types of languages observed by Rudin (1988) is that fronted wh-phrases are subject to strict ordering constraints in Bulgarian type, but not in SC type, as illustrated in (7):
258 Sandra Stjepanovic´
(7) a.
[CP Koj kogo [vižda]] who whom [sees ‘Who sees whom?’ b. *[CP Kogo koj [vižda]] c. [CP Ko [IP koga [vidi]]] who whom [sees ‘Who sees whom?’ d. [CP Koga [IP ko [vidi]]]
(Bulgarian)
(SC)
One question that immediately arises is why there are differences in constraints on linear ordering of wh-phrases between Bulgarian and SC types. As for Bulgarian type, Rudin (1988) and Boškovic´ (1997b, 1998b, 2002) argue that if adjunction to SpecCP in Bulgarian proceeds to the right, i.e. if the wh-phrase that is first in the linear order is the one that moves first to SpecCP, the strict ordering of fronted wh-phrases in Bulgarian follows from the Superiority Condition: the highest wh-phrase has to move first; otherwise, there is a Superiority effect. As for SC type, Rudin concludes that Superiority does not hold in SC by looking only at the examples of the type in (7c–d), i.e. short distance null C matrix questions, and offers an analysis in which SC type languages never yield Superiority effects. However, Boškovic´ (1997b, 1998b, 2002) shows that while it is true that in examples like (7c–d), SC does not show Superiority effects, in many other configurations Superiority effects do arise in SC. These configurations include embedded question contexts, long distance questions and matrix questions with overt C, as shown in (8). (8) Long-distance questions: a. ?Ko si koga tvrdio da je istukao? who are whom claim-2sg that is beaten ‘Who do you claim that beat whom?’ b. ?*Koga si ko tvrdio da je istukao? Embedded contexts:3 c. Ko koga voli, taj o njemu i govori. who whom loves that.one about him even talks ‘Everyone talks about the person they love.’ ? d. *Koga ko voli, taj o njemu/o njemu taj i govori. e. Ima ko šta da ti proda. has who what that you sells ‘There is someone who can sell you something.’ f. *Ima šta ko da ti proda.
(SC)
Multiple wh-fronting in Serbo-Croatian matrix questions 259
Root questions with overt C: g. Ko li šta kupuje? who C what buys ‘Who on earth buys what?’ h. *Šta li ko kupuje?
The ordering of fronted wh-phrases in SC long-distance questions, embedded questions and matrix questions with overt complementizers is not free. In these contexts, the highest wh-phrase has to appear first in the linear order, otherwise the sentence is bad, just as in Bulgarian. Boškovic´ argues that if, like in Bulgarian, the wh-phrase which is first in the linear order moves first, the ungrammaticality of (8b,d,f,h) is due to a violation of the Superiority Condition. SC is then not exempt from the Superiority Condition. As Boškovic´ points out, even if we did not have this empirical evidence, to claim that SC is exempt from the Superiority Condition would be conceptually problematic, since the Superiority Condition has recently been argued to follow from the Principles of Economy (Boškovic´ 1997a, Cheng and Demirdache 1990, Kitahara 1997) which are presumably universal, and therefore not a plausible locus of cross-linguistic variation. Superiority effects do not show up in SC short distance null C matrix questions (except in multiple sluicing contexts discussed below), while they do in a number of other contexts, including short distance overt C matrix questions, embedded contexts, and long-distance questions. Boškovic´ (1997b, 1998b) offers an account of these facts. He bases his account on an interesting parallelism between SC and French. French exhibits the same division between different types of questions as SC, but with respect to a somewhat different phenomenon. Exactly in those contexts in which SC exhibits Superiority effects, overt wh-movement is obligatory in French, while in those contexts in which SC does not exhibit Superiority effects, overt wh-movement does not need to take place in French (see Boškovic´ 1997b for the relevant data in French). The curious behavior of SC with respect to Superiority can then be explained if one assumes that SC is a French-type language with respect to when it must have overt wh-movement. Long-distance, embedded and overt C questions in SC then exhibit Superiority effects because in these contexts, overt wh-movement must take place in SC, just as in French. Short distance null C matrix questions in SC do not exhibit Superiority effects because, just like in French, these questions need not involve overt wh-movement. As a result, Boškovic´ argues, SC wh-movement is well-behaved with respect to Superiority: Whenever there is overt wh-movement in SC, Superiority is operative. The only difference between French and SC null C matrix questions is that in SC, wh-phrases still must front for some reason
260 Sandra Stjepanovic´
that is independent of checking the [+Wh] feature of C. If there is overt movement of wh-phrases, which is not driven by a [+Wh] feature of C, the question is what it is driven by, and to what positions the phrases are actually moving. In the following sections I will show that wh-phrases move to the same positions as phrases expressing identificational focus. In order to show this, I will first discuss sentences involving focus movement of non-wh elements.
3. Focus movement 3.1 Identificational vs. information focus In this section I will show that contrastively focused elements in SC undergo focus movement. First, let me discuss what I mean by contrastive focus. I will adopt É. Kiss’s (1998) conception of contrastive focus, which is subsumed under the notion of identificational focus. É. Kiss (1998) argues that a distinction should be made between ‘information’ focus (sometimes also called presentational focus or new information focus) and identificational focus. She argues that these two types of focus have different syntactic and semantic properties. Identificational focus has the following semantic-communicative function: (9) The function of identificational focus: An identificational focus represents a subset of the set of contextually or situationally given elements for which the predicate phrase can potentially hold; it is identified as the exhaustive subset of this set for which the predicate phrase actually holds. (É. Kiss 1998: 245)
According to É. Kiss (1998), semantically, the constituent expressing identificational focus represents the value of variable bound by an abstract operator expressing exhaustive identification. Syntactically, the constituent expressing identificational focus itself acts as an operator, moving into a scope position, the specifier of a functional projection, and binding a variable. Identificational focus thus expresses exhaustive identification. Furthermore, it does not have to be present in every sentence, but if it is present, it can trigger syntactic reordering of elements in the sentence. New information focus, on the other hand, merely conveys nonpresupposed information, without expressing exhaustive identification, it is present in every sentence, and does not trigger any syntactic reordering.
Multiple wh-fronting in Serbo-Croatian matrix questions 261
A difference between these two kinds of focus can be seen from their behavior in the test of exhaustive identification devised by Szabolcsi (1981). This test involves a pair of sentences in which the first sentence has a focus consisting of a coordinate DP, and the second sentence differs from the first one only in that one of the DPs in the coordinate phrase is dropped, as illustrated in (10)–(11) for Hungarian, from É. Kiss (1998): (10) a.
Mari EGY KALAPOT IS EGY KABÁTOT nézett ki magának. Mary a hat-acc and a coat-acc picked out herself.to ‘It was a hat and a coat that Mary picked for herself.’ b. Mari EGY KALAPOT nézett ki magának. Mary a hat-acc picked out herself.to ‘It was a hat that Mary picked for herself.’
(11) a.
Mari nézett ki magának EGY KALAPOT IS EGY KABÁTOT. Mary picked out herself.to a hat-acc and a coat-acc ‘Mary picked a hat and a coat for herself.’ b. Mari nézett ki magának EGY KALAPOT. Mary picked out herself.to a hat-acc ‘Mary picked a hat for herself.’
The examples in (10) involve focused elements in a preverbal position, while the examples in (11) involve postverbal focus. As É. Kiss (1998) points out, if the second sentence is not among the logical consequences of the first one, the focus expresses exhaustive identification. While the Hungarian sentence in (11b) is among logical consequences of the sentence in (11a), the sentence in (10b) is not a logical consequence of the sentence in (10a), on the contrary it contradicts (10a). The sentence in (10b) passes this test of exhaustivity while the sentence in (11b) does not. The examples in (11) involve information focus, while the examples in (10) involve identificational focus. É. Kiss (1998) claims that English it-cleft sentences show the same behavior as Hungarian sentences in (10), as the translations of (10a–b) show, which leads her to conclude that focus in these sentences is identificational focus. Identificational focus introduces an operator which changes the truth conditions of the sentence, which is not the case with information focus. É. Kiss (1998) also points out that there are two versions of identificational focus. Identificational focus can express contrast, which is identification with exclusion, or identification only. According to É. Kiss (1998), identificational focus expresses contrast, if it operates on a closed set of entities whose members are known to the participants of the discourse. In this case, the identification of
262 Sandra Stjepanovic´
a subset of a given set also identifies the contrasting complementary subset. However, identificational focus can also operate on an open set of entities, as in the Hungarian example in (12b), which is an answer to the question in (12a), from É. Kiss (1998). (12) a.
Ki írta a Háború és békét? who wrote the War and Peace ‘Who wrote War and Peace?’ b. A Háború és békét TOLSTSZTOJ írta. the War and Peace Tolstoy wrote ‘It was Tolstoy who wrote War and Peace.’
(12b), as an answer to the question in (12a), does not presuppose a closed set of persons who might have written War and Peace. As a consequence, according to É. Kiss, the identification of the subset for which predicate holds does not result in the delineation of a complementary subset with clearly identifiable elements. In this case, identificational focus is not contrastive. É. Kiss (1998) shows that in Hungarian, identificational focus is associated with the preverbal position, and that this focus movement of identificationally focused elements is obligatory in Hungarian. She also reports several other languages, in which identificational focus undergoes movement to a certain syntactic position, these languages being Romanian, as discussed in G‘bell (1998), Italian, Greek, discussed in Tsimpli (1994), Arabic, as discussed in Ouhalla (1994) and Finnish, discussed in Vilkuna (1995). She shows that these languages might be parameterized with respect to whether this focus is [±contrastive]. 3.2 Identificational focus in SC In this section I will show that SC is also one of the languages that require elements expressing identificational focus to move to a particular position in the sentence. In sentences such as those in (2) and repeated here as (13), focus expresses exhaustive identification. The interpretation of the sentence is such that Petar is the only one of a set of relevant persons that Marija loves. (13) PETRA Marija voli. Petar-acc Marija-nom loves ‘It is Petar that Marija loves.’
Multiple wh-fronting in Serbo-Croatian matrix questions 263
We can confirm that focus in cases such as these expresses exhaustive identification by applying Szabolcsi’s (1981) test illustrated in (10)–(11). Consider the pairs of SC sentences in (14) and (15). (14) a.
PETRA I IGORA Marija voli. Petar-acc and Igor-acc Marija loves ‘It is Petar and Igor that Marija loves.’ b. PETRA Marija voli. Petar-acc Marija loves ‘It is Petar that Marija loves.’
(15) a.
Marija voli PETRA I IGORA. Marija loves Petar-acc and Igor-acc ‘Marija loves Petar and Igor.’ b. Marija voli PETRA. Marija loves Petar-acc ‘Marija loves Petar.’
In (14), the sentence in (14b) is not among logical consequences of the sentence in (14a). In (15), on the other hand, the sentence in (15b) is among logical consequences of the sentence in (15a). This means that in sentences in (14), focus expresses exhaustive identification, which is not the case with sentences in (15). A similar situation also obtains in the pair of sentences in (16), which differ from the sentences in (14) in that the focused element is located in an immediately preverbal position, instead of in the sentence initial position as in (14). (16) a.
Marija PETRA I IGORA voli. Marija Petar-acc and Igor-acc loves ‘It is Petar that Marija loves.’ b. Marija PETRA voli. Marija Petar-acc loves ‘It is Petar that Marija loves.’
These sentences could be used in the same contexts as the sentences in (14). Just as in sentences in (14), in these cases too, the sentence in (16b) is not a logical consequence of the sentence in (16a), indicating that exhaustive identification is at play. It is obvious that in the examples (14) and (16), the focused elements do not occupy their base-generated position, which is one following the verb. This raises the question of what positions they actually occupy and how they get in these positions.
264 Sandra Stjepanovic´
There is evidence showing that the relevant positions are located above TP and between TP and VP. First of all, note that in sentences in (14), the focused element precedes the subject. In (16), however, it follows the subject. So, there might be two positions capable of hosting an identificationally focused element. Consider now the following data: (17) a.
Oni mudro JOVANA savjetuju. they wisely Jovan-acc advise ‘It is wise of them to advise Jovan.’ ‘They advise Jovan in a wise manner.’ b. Oni JOVANA mudro savjetuju. they Jovan-acc wisely advise *‘It is wise of them to advise Jovan.’ ‘They advise Jovan in a wise manner.’
(18) JOVANA oni mudro savjetuju. Jovan-acc they wisely advise ‘It is wise of them to advise Jovan.’ ‘They advise Jovan in a wise manner.’
According to Jakendoff (1972) some adverbs, such as wisely, are ambiguous between a subject-oriented and manner reading. On the latter reading, the adverb is a VP adverb, while on the former, it has a sentential reading. The adverb mudro ‘wisely’ exhibits an interesting behavior in the sentences in (17)–(18). In (17a), where the adverb precedes the focused element, both the sentential and manner readings are available, as indicated by English translations of the sentence. In (17b), where the focused element follows the adverb, on the other hand, the only possible reading of the adverb is the manner reading. (18) shows that when the focused element is sentence initial, both readings are available. The question is how do we interpret these facts. Stjepanovic´ (1998) takes these facts to show that there are two focus positions where identificationally focused material is licensed, by adopting Boškovic´’s (1997a) argument based on Watanabe (1993) that sentential adverbs are adjoined to TP, while VP adverbs are adjoined to VP. The contrast in the interpretation possibilities of adverbs in these two sentences shows that one focus position is below TP. In cases where the sentential reading of the adverb is available, it means that the adverb can be adjoined to TP, while in cases where it is not available, it means that the adverb cannot be adjoined to TP and has to be lower in the structure. Since in (17b), where the focused phrase precedes the adverb, the sentential reading of the adverb is not available, it means that the
Multiple wh-fronting in Serbo-Croatian matrix questions 265
adverb cannot be adjoined to TP, and it means that the focused phrase must be adjoined lower than TP, too. If the adverb precedes the focused phrase, as in (17a), both sentential and manner readings are available. This means that the adverb could be adjoined either to VP or TP, which means that the focused phrase is at least as high as the VP adjoined position. Stjepanovic´ (1998) takes the VP adjoined position to be one position where identificational focus is licensed. (18) shows that when a focused phrase is sentence initial, both readings of the adverb are available, which means that the focused phrase is higher than TP. Stjepanovic´ (1998) takes this position to be AgrSP position. Boškovic´ (1997b, 1998b) notes, however, that if one takes the lower focus position to be an AgrOP adjoined position (assuming that VP adverbs can also adjoin to AgrOP), then the job of focus licensing in SC can be reduced to a single category, AgrSP and AgrOP being the same category in two different positions.4 3.3 Multiple wh-fronting: A case of Focus movement We have seen above that SC is a type of multiple wh-fronting language in which the second phrase does not undergo overt wh-movement to SpecCP, although it fronts overtly. We have also seen that based on certain facts about Superiority in SC, Boškovic´ (1997b, 1998b) has argued that in null C short-distance matrix questions, even the first wh-phrase need not move to SpecCP, although it fronts overtly. The question is where they move and why. Note now that positions in which wh-phrases could be found are exactly those where identificationally focused elements are found, as illustrated in (19).5 (19) a.
Ko mudro koga savjetuje? who wisely whom advises ??‘Who is it wise of to advise whom?’ b. Ko koga mudro savjetuje? who whom wisely advises *‘Who is it wise of to advise whom?’ ‘Who advises whom in a wise manner?’ c. Koga ko mudro savjetuje? whom who wisely advises ??‘Who is it wise of to advise whom?’ ‘Who advises whom in a wise manner?
(SC)
In (19), exactly the same judgments obtain with respect to the interpretation of the adverb mudro ‘wisely’, as in the examples involving focused phrases. This
266 Sandra Stjepanovic´
means that wh-phrases could be occupying the same positions that identificationally focused phrases do. This is not implausible. Cross-linguistically it is well attested that interrogative wh-phrases share the syntactic behavior of focused phrases. In languages in which focus phrases move to a certain position, wh-phrases also move there. As shown in a number of articles in É. Kiss (1995), these languages are typologically as different as Somali, Chadic, Basque, Aghem, Hungarian, Haida, Omaha, Quetchua, Korean, Greek or Finnish. Stepanov (1998) argues that a similar situation obtains in Russian. Kidwai (1999) claims that the same situation holds in Hindi-Urdu. (20)–(21) shows examples from Hungarian and Aghem from Horvath (1986), which are all languages where focused and wh-phrases occupy the same syntactic positions. (20) a.
Attila melyik lanyt szereti legjobban? Attila which girl-acc likes best ‘Which girl does Attila like best?’ b. *Attila szereti legjobban melyik lanyt. c. Attila A FOLDRENGESTOL felt. Attila the earthquake.from feared ‘Attila was afraid of the earthquake.’ d. *Attila felt A FOLDRENGESTOL.
(21) a.
a mo nin ndugho ds p2 run who ‘Who ran?’ b. *ndugho mo nin (no) who p2 run (foc ‘Who ran?’ c. fil a mo zi AN SOM be-ko friends sm p2 eat in farm fufu ‘Friends eat fufu in the farm.’ d. *fil a mo zi be-ko AN SOM.
The overwhelming cross-linguistic evidence in this respect has led Horvath (1986) to express this wh-phrase — focus phrase connection in a form of a universal principle, given in (22). (22) The syntactic position(s) in which non-echo interrogative wh-phrases can appear in a language L will be identical to or be a proper subset of the position in which Focus constituents can appear in the language L.
Multiple wh-fronting in Serbo-Croatian matrix questions 267
The correspondence between Focus movement and the overt movement of wh-phrases leads Horvath to conclude that wh-phrases are inherently focused. A similar conclusion was reached by Rochemont (1986). If this is true, then we can assume that wh-phrases necessarily have a focus feature which is licensed in the positions in which non-wh focused phrases are licensed. To sum up, I have shown that identificational focus is licensed in particular syntactic positions in SC. These positions also host wh-phrases in SC, which suggests that the driving force behind multiple wh-fronting and identificational focus fronting is the same.
4. Multiple sluicing and Superiority We have seen that in SC, wh-phrases have to be fronted, but their fronting need not be due to checking of a [+Wh] feature in C. In original Rudin’s (1988) analysis, this was the case with the second wh-phrase. In Boškovic´’s (1997b, 1998a, 1998b, 2002) analysis based on the ambivalent behavior of Superiority, in those cases which should be Superiority violations, but they are not, none of the wh-phrases is undergoing overt wh-movement to SpecCP. According to Boškovic´, their overt fronting is due to some other reasons. In the preceding sections, I have shown that this non-wh-movement overt fronting of wh-phrases is due to focusing reasons. Boškovic´ takes this proposal and builds it into his analysis of the presence and absence of Superiority in SC. According to Boškovic´, if wh-phrases are undergoing only focus movement, then no Superiority effects are detected. In his system then, focus movement is not subject to Superiority. As it will be seen below, the lack of Superiority effects with focus movement is derived from Economy of Derivation due to formal properties of focus movement. At this point I would like to add another context in which Superiority effects seem to show up in SC, which is found in the examples in (23). This context is interesting, because it involves short distance null C matrix questions, which is exactly the context in which Superiority effects should not show up. However, as we can see in (23), movement of originally lower wh-phrase across a higher one results in degradation, suggesting that Superiority is at play. It would be interesting then to see why such examples as (23) exhibit restrictions on ordering of wh-phrases.
268 Sandra Stjepanovic´
(23) A: Neko je udario nekog. somebody is hit someone ‘Somebody hit someone.’ B: Ko koga? who whom ‘Who hit whom?’ ? B: *Koga ko? whom who ‘Who hit whom?’
I will argue that this context involves sluicing with multiple remnants, or multiple sluicing. Examining the interactions between Superiority and multiple sluicing will reveal some interesting facts about what component of grammar this ellipsis phenomenon takes place in, and will shed more light on the nature of wh-phrase focus fronting. 4.1 Multiple sluicing in SC Consider the following SC sentences. (24) A: Neko je nekad ovdje sakrio blago. somebody is somewhere here hidden treasure ‘Somebody hid the treasure here at some point in the past.’ B: Ko kad? who when ‘Who (hid the treasure) when?’ ? B: *Kad ko? (25) A: Neko je negdje sakrio blago. somebody is somewhere hidden treasure ‘Somebody hid the treasure somewhere.’ B: Ko gdje? who where ‘Who (hid the treasure) where?’ B: ?*Gdje ko? (26) A: Neko je nekoga sakrio ovdje. somebody is somebody hid here ‘Somebody hid somebody here.’
Multiple wh-fronting in Serbo-Croatian matrix questions 269
B: Ko koga? who whom ‘Who (hid) whom (here)?’ B: ?*Koga ko?
The Speaker B utterances in (24)–(26) are multiple matrix questions with a null C. All of them contain only wh-words, with the rest of the sentence material elided by some sort of ellipsis. On the face of it, the ellipsis process can be either gapping or multiple sluicing, which has been argued to exist, among others, in Japanese (Takahashi 1994), Korean (Kim 1998), and to some extent in English (Bolinger 1978, Merchant 1999, Richards 1997). I will show here that the ellipsis process in these examples is sluicing rather than gapping. Jackendoff (1971) and Takahashi (1994) point out that gapping in English is unacceptable with conjuncts other than and: (27) *Bill ate the peaches, but Harry the grapes.
The conjunction in (27) is but, and the sentence is degraded. SC also has a restriction on what conjunction can appear in unambiguously gapping constructions. The conjunction has to be a, the counterpart of English and. With ali ‘but’ the sentence is bad, as illustrated in (28). (28) a.
Ivan je pojeo jabuku, a Petar breskvu. Ivan is eaten apple and Petar peach ‘Ivan ate an apple, and Petar a peach.’ b. *Ivan je pojeo jabuku, ali Petar breskvu. Ivan is eaten apple but Petar peach ‘Ivan ate an apple, but Petar a peach.’
Sluicing is possible with ali ‘but’, as illustrated in (29). (29) Ivan je vidio nekoga, ali ne znam koga. Ivan is seen somebody but not know whom ‘Ivan saw somebody, but I don’t know whom.’
It is also possible to construct a parallel example to (29) with multiple remnants. Ali ‘but’ is still possible: (30) Neko je vidio nekog, ali ne znam ko koga. somebody is seen somebody but not know who whom ‘Somebody saw someone, but I don’t know who whom.’
270 Sandra Stjepanovic´
The example in (30) then seems to be an instance of multiple embedded sluicing and not gapping. In fact, embedded gapping is unacceptable (Lasnik and Saito 1992), while such sluicing is perfect, as illustrated in (31a) for gapping and in (31b) for sluicing. (31) a. *John likes Mary, and I think that Bill Jennifer, too. b. John likes somebody, but I don’t know who.
The same situation obtains in Serbo-Croatian. Gapping with subordination is not possible: (32) *Ivan je volio Mariju, a mislim da Goran Vesnu. Ivan is loved Marija and think that Goran Vesna ‘Ivan loved Marija, and I think that Goran loved Vesna.’
Sluicing with subordination, on the other hand is possible, as illustrated in (29). Furthermore, the example in (30) with multiple remnants is perfect, just like the sluicing example in (29) and unlike the gapping example in (32). Thus, the process of eliding all the sentence material except wh-phrases in (30) is not gapping. Lasnik (1999) shows that matrix sluicing is possible in English. (33) A: Mary loves somebody. B: Who?
Just as in English, sluicing is also allowed in matrix contexts in SC: (34) A: Marija je voljela nekog. Marija is loved somebody ‘Marija loved somebody.’ B: Koga? whom ‘Whom?’
If it is possible to have embedded sluicing with multiple remnants, then one would expect it to be possible to have matrix sluicing with multiple remnants. The Speaker B utterances in (24)–(26) seem to be exactly examples of sluicing with multiple remnants. Furthermore, as pointed out by Ross (1969), in single remnant sluicing, the remnant wh-phrase in the sluiced conjunct usually corresponds to an indefinite DP in the antecedent conjunct as in (31b), but it does not have to, for example, it does not correspond to anything visible on the surface in (35). (35) He is writing, but I don’t know what.
Multiple wh-fronting in Serbo-Croatian matrix questions 271
The same situation obtains with multiple remnants in SC: neither in embedded nor in matrix clauses do they need to have corresponding indefinite phrases in the antecedent. (36) a.
Marko piše, ali ne znam šta kome. Marko writes but not know what whom ‘Marko is writing, but I don’t know what he is writing to whom.’ A: Marko piše. Marko writes ‘Marko is writing.’ B: Šta kome? what whom ‘What (is he writing) whom?’
With gapping, the antecedents of remnants must be present overtly, as illustrated in (37). (37) a.
Marko piše pismo, a Petar pjesmu Marko writes letter-acc and Petar poem-acc ‘Marko is writing a letter, and Petar a song.’ b. *Marko piše, a Petar pjesmu Marko writes and Petar poem-acc ‘Marko is writing, and Petar a song.’
Given these facts, I conclude that examples in (24)–(26) are instances of multiple matrix sluicing, and not gapping. One curious thing about the multiple sluicing examples in (24)–(26) is that they exhibit strict ordering of wh-phrases. If the higher wh-phrase appears first, the sentence is good, as in (24a)–(26a), but if the lower wh-phrase appears first, the sentence is bad, as in (24b)–(26b). This is curious because if the Speaker B responds with full sentences without ellipsis, there are no constraints on linear ordering of wh-phrases, as shown in (38)–(40).6 (38) A: Neko je nekad ovdje sakrio blago. somebody is some.time.ago here hidden treasure ‘Somebody hid the treasure here at some point in the past.’ B: Ko je kad ovdje sakrio blago? who is when here hidden treasure ‘Who hid the treasure here when? B: Kad je ko ovdje sakrio blago?
272 Sandra Stjepanovic´
(39) A: Neko je negdje sakrio blago. somebody is somewhere hidden treasure ‘Somebody hid the treasure somewhere.’ B: Ko je gdje sakrio blago? who is where hidden treasure ‘Who hid the treasure where?’ B: Gdje je ko sakrio blago? (40) A: Neko je nekoga ovdje sakrio. somebody is somebody here hidden ‘Somebody hid somebody here.’ B: Ko je koga ovdje sakrio? who is whom here hidden ‘Who hid whom here?’ B: Koga je ko ovdje sakrio?
The constraint on linear ordering of wh-phrases in examples (24)–(26) is reminiscent of the Superiority Condition. If the linear ordering of the wh-phrases in these examples is constrained by some version of Superiority, then the question is why Superiority effects emerge in these matrix null C questions, when they do not normally do in other null C matrix questions. In order to give an answer to this question, I first have to examine current analyses of Superiority with multiple wh-fronting. 4.2 Superiority in SC The question is what is responsible for this ambivalent behavior of SC with respect to Superiority. There are at least two recent analyses in the literature attempting to offer an answer to this question. One of them is Boškovic´ (1997b, 1998a, 1998b), which has partially been given above, and the other is Richards (1997). As mentioned above, Boškovic´ draws a parallel between French and SC with respect to the contexts in which wh-movement takes place in these languages. This leads him to a conclusion that the curious behavior of SC with respect to Superiority can be explained if one assumes that SC is a French-type language with respect to when it must have overt wh-movement. Overt wh-movement is present in long-distance, embedded and overt C questions in SC just as in French. Short distance null C matrix questions in SC need not involve overt wh-movement, just as in French. Wh-phrases still front, though. On the question of motivation for this fronting in null C matrix question, Boškovic´ follows Stjepanovic´ (1998), who shows that in these questions wh-phrases
Multiple wh-fronting in Serbo-Croatian matrix questions 273
appear in the positions in which contrastively focused material occurs. Fronted wh-phrases that do not end up in SpecCP then undergo focus movement. The question that arises at this point is why wh-movement is obligatory in French and SC embedded, long distance and overt C matrix questions, unlike in null C matrix questions. Boškovic´ argues that a possible answer to this question lies in lexical insertion possibilities provided by the current minimalist framework (Chomsky 1995), and Chomsky’s (1995) definition of strong features. Boškovic´ argues that lexical insertion, or, more precisely Merger, can occur in LF under well-defined conditions: the element to be merged must be phonologically null since LF cannot deal with phonological features, and Merger must be at the top of the tree, since, by definition, Merger must expand the structure. Even an element with a strong feature can be inserted in LF, given Chomsky’s (1995) definition of strong features, where strong features are defined derivationally as objects that cannot be tolerated by the derivation and need to be eliminated immediately upon their introduction into the structure. So, according to Boškovic´, French and SC do not have obligatory overt wh-movement in null C matrix questions because a null C with a strong wh-feature, the trigger for wh-movement, can be inserted in LF here. In embedded, long distance and overt C matrix questions, LF C insertion is blocked (see Boškovic´ 1997b for discussion). C has to be present in the overt syntax, hence overt wh-movement is obligatory in this case. As a result, in such multiple questions Superiority effects show up if the wh-feature is not checked in the most economical way, given the Economy account of Superiority adopted by Boškovic´. The most economical way to check the [+Wh] feature is through the shortest movement possible, i.e. by moving the wh-phrase that is closest to C. The movement of a wh-phrase to SpecCP triggers Spec-head agreement with C, checking the wh-feature, so that the wh-phrase that moves there first necessarily checks it. In Boškovic´’s theory, overt wh-movement to SpecCP triggers Superiority effects, while focus movement does not. Boškovic´ argues that focus movement does not violate Superiority if (a) the focus feature attracting focus and wh-elements is an Attract All feature attracting all focus elements, and (b) the Economy account of Superiority is adopted, whereby every feature has to be checked in the most economical way, i.e. through the shortest movement possible. Consider how his system works.7 (41) wh-movement F wh-phrase1 +wh +wh strong weak
wh-phrase2 +wh weak
wh-phrase3 +wh weak
274 Sandra Stjepanovic´
So, according to Boškovic´ (1999), with wh-movement, the attractor is an Attract 1F(eature). This means that it attracts only one feature, which has to be checked in the most economical way, i.e. through the shortest movement possible. Here, the situation is the same as in languages like English, where the attractor for wh-movement ([+Wh] C) is clearly an Attract 1F head. Hence, if wh-phrase1 does not move first to check it, a Superiority effect will result. With focus movement, the Focus attractor is an Attract All feature. Since it is an Attract All feature, it attracts all focus feature bearing elements. As a result, no Superiority effects will be expected with Focus movement. The Attract All property is satisfied in the same way from the point of view of economy regardless of the order in which the wh-phrases move to the focused head. In (42), regardless, whether the wh-phrases move in 1–2–3, 1–3–2, 2–1–3, 2–3–1, 3–1–2, or 3–2–1 order, the same number of nodes will be crossed to satisfy the Attract All feature of the relevant head. Hence, the lack of Superiority with focus movement. (42) Focus movement F wh-phrase1 +Focus +Focus strong weak
wh-phrase2 +Focus weak
wh-phrase3 +Focus weak
So, for Boškovic´, the ambivalent behavior of SC with respect to Superiority is a result of the interaction of several aspects of grammar, including the Economy account of Superiority, lexical insertion possibilities and the nature of strong features. As far as Bulgarian type languages are concerned, which exhibit Superiority effects in all contexts, Boškovic´ argues that this is so because in these languages, C is lexically specified as a phonological affix, and it therefore must be inserted in the overt syntax. Recall that LF insertion of elements which are not phonologically null is not possible, since LF cannot deal with phonological information. So, a wh-phrase in Bulgarian always undergoes overt movement to SpecCP to check a strong wh-feature. Given the Economy account of Superiority, this will be the highest wh-phrase. As discussed above, however, Rudin (1988) shows that in Bulgarian multiple questions all wh-phrases are in SpecCP, not just the highest one. The question is why other phrases also move to SpecCP. Boškovic´ argues that the answer to this question lies in focus movement. Just like in SC, all wh-phrases in Bulgarian must undergo focus movement. The focus licenser in Bulgarian is C. So, the highest wh-phrase has to move first in order to satisfy the strong wh-feature of C, at the same time checking its own focus feature. Other wh-phrases are attracted by the Attract All focus, and thus
Multiple wh-fronting in Serbo-Croatian matrix questions 275
just as in SC, it does not trigger Superiority effects. As a result, in Bulgarian, the highest wh-phrase has to move first, and after that the order of movement of wh-phrases is free, as shown in (44). (43) shows that when only two phrases are present, if kak is the first in the linear order and kogo follows, Superiority effects arise. In (44), however, as long as the highest subject moves first, there are no ordering requirements on kak and kogo. (43) a.
Kogo kak e tselunal? who how is kissed ‘Who did he kiss how?’ b. *Kak kogo e tselunal? how whom is kissed ‘Who did he kiss how?’
(44) a.
Koj kogo kak e tselunal? who whom how is kissed ‘Who kissed whom how?’ b. Koj kak kogo e tselunal? who how whom is kissed ‘Who kissed whom how?’ c. *Kogo kak koj e tselunal? whom how who is kissed ‘Who kissed whom how?’
As mentioned above, an alternative analysis of the different behavior of SC and Bulgarian with respect to Superiority and the ambivalent behavior of SC in this respect is offered by Richards (1997). For Richards (1997), the difference between SC and Bulgarian with respect to Superiority lies in the interaction of several aspects of grammar, in particular the Principle of Minimal Compliance in (45), and a constraint on Attract, given in (46). The definition of Attract is given in (47). (45) Principle of Minimal Compliance (PMC) For any dependency D that obeys constraint C, any elements that are relevant for determining whether D obeys C can be ignored for the rest of the derivation for purposes of determining whether any other dependency D’ obeys C. (46) Shortest A pair P of elements {α, β} obeys Shortest iff there is no well-formed pair P’ which can be created by substituting γ for either α or β, and the set of nodes c-commanded by one element of P’ and dominating the other is
276 Sandra Stjepanovic´
smaller than the set of nodes c-commanded by one element of P and dominating the other. (47) Attract An attractor K attracts a feature F, creating a copy α¢ of an element α containing F, and Merging α¢ with K. The relations between α¢, K, and F must all obey Shortest.
Shortest constrains the relation between the attractor K and the attracted feature F, forcing the attractor to attract the nearest possible feature. This is what Richards calls Shortest Attract. Shortest also constrains the relation between F and the copy α¢ of α, requiring that movement be as short as possible. In this way, Shortest prevents movement of F past an attractor which could attract F, and also forces movement to be to the closest available landing site. This is what Richards calls Shortest Move. Richards argues that the interaction between PMC and Shortest, as well as the assumption that fronted wh-phrases occupy multiple specifiers of C, can account for the Superiority effects in Bulgarian. In the case of multiple wh-phrases, given Shortest, C first attracts the highest wh-phrase. At this point PMC renders the attractor C immune to Shortest, i.e. it turns off Shortest Attract. As a result, C can attract the leftover wh-phrases in any order. Furthermore, Richards argues that in the case of movement to multiple specifiers, an inner specifier is closer than an outer specifier. He also argues that although Shortest Attract is, Shortest Move is not affected by PMC. As a result, every subsequent movement of wh-phrases will be to an inner specifier. This is what Richards calls “tucking in”. So, in the case of wh-phrases in (48), C first attracts Wh1 and PMC turns off Shortest Attract. As a result, C can attract either Wh2 or Wh3. If at this point it attracts Wh2, Wh2 will move and tuck in, i.e. it will move to a lower specifier of C. Then Wh3 tucks into the lowest specifier of C. The resulting structure is given in (49). If, on the other hand, after attracting Wh1 first, C attracts Wh3 next, Wh3 will tuck into the lower specifier. After this C attracts Wh2, which moves to the lowest specifier. The resulting structure is given in (50). (48) C [Wh1 [Wh2 [Wh3 (49) [Wh1 [Wh2 [Wh3 [C]]]] (50) [Wh1 [Wh3 [Wh2 [C]]]]
As illustrated in (44), this is exactly the range of facts observed in Bulgarian. Richards’ analysis thus works well for Bulgarian. As we have seen above, SC exhibits different behavior with respect to Superiority than Bulgarian. Unlike
Multiple wh-fronting in Serbo-Croatian matrix questions 277
Bulgarian, SC lacks Superiority effects in short distance null C matrix questions, while in all other contexts it exhibits Superiority effects just like Bulgarian. Now, in order to explain why Superiority effects do not show up in SC short distance null C matrix questions, Richards argues that SC has a way of moving wh-phrases other than wh-movement to SpecCP. Local movement of wh-phrases is A-scrambling. In particular, Richards argues that SC allows arbitrarily many attractors within IP projections, which are responsible for scrambling wh-phrases. So, in case of two wh-phrases, as in (51), one possible derivation is when there are two such attractors (X and Y in (51)). The lower attractor Y attracts the higher Wh1. Now the higher attractor X must attract a wh-phrase and the only wh-phrase it can attract is the lower Wh2. (51) [CP C [XP X [YP Y [Wh1 Wh2]]]]
At this point C attracts Wh2, since it is the closest wh-phrase. This derivation, therefore, yields a sentence in which the originally lower wh-phrase moves to SpecCP without causing a Superiority effect, as in (7d). Given this mechanism, it is easy to think of a derivation where originally higher wh-phrase ends up in SpecCP, an expected result, as in (7c). Thus, in Richards’ theory, an escape hatch from Superiority in these examples is A-scrambling. In long distance questions, however, Richards argues that this escape hatch is not available, and that this is why multiple longdistance wh-fronting exhibits Superiority effects, as in (8b). Having outlined these analyses of Superiority with multiple fronting, let us go back to the SC examples in (24)–(26). One prediction of Boškovic´ ‘s analysis is that if in SC null C multiple matrix questions, which do not normally exhibit Superiority effects, a null C can be forced to be present overtly, the Bulgarian pattern should emerge, i.e. a Superiority effect should show up. I will show that this is true of the data in (24)–(26). Under Richard’s analysis, if all wh-phrases are required to ‘tuck in’ the same specifier, Superiority effects should emerge even in SC, unless they are A-scrambled first. 4.3 Multiple sluicing and Superiority: Sluicing as PF deletion Recall that I have argued that the data in (24)–(26) are instances of multiple sluicing. Sluicing with a single remnant is standardly analyzed as wh-movement followed by IP deletion (Ross 1969, Rosen 1976, Takahashi 1994, Lasnik 1999, Merchant 1999), or base-generated null IP licensed by a [+Wh] C agreeing with its specifier and filled with linguistic material by LF copying (Levin 1982,
278 Sandra Stjepanovic´
Chung, Ladusaw and McCloskey 1995, among others). So, both types of accounts agree that the remnant wh-phrase is in SpecCP. As far as multiple sluicing is concerned, there are analyses in which multiple remnants are also placed in SpecCP, such as Takahashi (1994). If we combine the proposal that wh-phrases in sluicing examples are in SpecCP with Boškovic´’s and Richard’s analysis of the ambivalent behavior of SC with respect to Superiority, then Superiority effects in multiple matrix sluicing do not come as a surprise. Recall that Boškovic´ (1997b, 1998a, 1998b) argues that the ambivalent behavior of SC with respect to Superiority effects is caused by the absence or presence of C in overt syntax. If C has to be present in overt syntax, Superiority effects show up (embedded, long-distance and overt C contexts). If it does not need to be present in overt syntax, i.e. if it can be inserted in LF (null C in matrix questions), no Superiority effects show up. Now, if wh-phrases in sentences undergoing sluicing are in SpecCP, then C must also be present in overt syntax in such sentences. The strong wh-feature it carries has to be eliminated in the most economical way. The most economical way is for it to be checked by the highest wh-phrase. This means that the highest wh-phrase has to move first. As far as the movement of the lower wh-phrase is concerned, recall that Boškovic´ claims that all wh-phrases in SC are attracted by a strong focus feature with Attract All property. Furthermore, Boškovic´ (1997b) shows that in the case of overt insertion of C in short distance matrix questions, C can act as a focus licensor for wh-phrases in SC. Given this, it is not implausible to claim that the lower wh-phrase in these examples moves to SpecCP to check its focus feature. This yields exactly the Bulgarian pattern discussed above. Under Richard’s analysis, since both wh-phrases are moving to the same specifier, on the face of it, they should be strictly ordered. Recall that in his system, when phrases are moving to the same specifier, the highest one moves first, and then the lower one ‘tucks’ in the specifier below the one where the highest phrase has moved. One caveat with Richard’s analysis, however, is the mechanism of arbitrarily many attractors in IP that is used to avoid Superiority effects in SC short distance matrix questions. As shown above, this mechanism of arbitrarily many attractors is able to scramble wh-phrases rendering their order opposite of the original order. C then attracts the closest wh-phrase, which due to scrambling may be the originally lower wh-phrase. Superiority effects are then voided. Notice now that in the sluicing examples in (24)–(26), which are short distance questions, the escape hatch from Superiority in the form of arbitrarily many attractors in IP projections is still available. Given this mechanism, nothing prevents these phrases from being first scrambled and then
Multiple wh-fronting in Serbo-Croatian matrix questions 279
attracted by C with the subsequent deletion of IP. As a result, Superiority effects should not show up, counter to fact. So, if the analysis of multiple sluicing I have presented here is right, these data argue against such a mechanism. Note now that given the Economy of Derivation account of Superiority, which Boškovic´ (1998a) argues is superior to alternative accounts based on multiple wh-fronting construction, and given SC data in (24)–(26), any account of sluicing as base-generated IP licensed by a [+Wh] C agreeing with the wh-phrases in its specifier (possibly followed by LF copying) cannot be maintained. Under this approach, wh-phrases are also base-generated in SpecCP, so any phrase could be base-generated first, checking the wh-feature of C. Superiority effects then should not show up. If wh-phrases, however, have to undergo overt movement, as in the wh-movement and PF deletion of IP approach, then Superiority effects are expected to emerge in case the highest wh-phrase does not move first to check the wh-feature. So far I have examined the behavior of SC multiple matrix sluicing with respect to only two remnant wh-phrases. I have shown that SC exhibits the Bulgarian pattern in this context with respect to Superiority. If SC follows the Bulgarian pattern in multiple matrix sluicing cases, then it should also behave like Bulgarian when more than two wh-phrases are involved. As shown in (44), if there are more than two wh-phrases in Bulgarian, Superiority cares only about the highest one, while it disregards other wh-phrases in the sentence. So, in a sentence with three wh-phrases, the highest wh-phrase must move first, and then the order of movement of the other two wh-phrases is free. As expected, SC behaves like Bulgarian in this respect: (52) a.
Ivan je nekog nekako prevario. Ivan is someone somehow cheated ‘Ivan cheated someone somehow. b. Koga kako? whom how ‘How (did Ivan cheat) whom?’ c. ?*Kako koga? how whom d. Neko je nekog nekako prevario. somebody is someone somehow cheated ‘Somebody cheated someone somehow.’ e. Ko koga kako? who whom how ‘How (did) who (cheat) whom?’
280 Sandra Stjepanovic´
f. Ko kako koga? g. *Kako ko koga? h. *Koga ko kako?
The contrast between (52b) and (52c) shows that prior to movement to SpecCP, kako ‘how’ starts lower in the structure than koga ‘whom’ (see Boškovic´ 1997d for an explanation). The acceptability of (52e) and (52f) shows that if the highest wh-phrase moves first, the order of other wh-phrases is free, while (52g) and (52h) show that we get unacceptable constructions if the highest wh-phrase does not move first. The interaction of multiple sluicing and Superiority in SC thus argues against the null IP and wh-phrase base-generation approach to sluicing, while they argue for the wh-movement followed by PF deletion approach to sluicing. Given that sluicing is an ellipsis phenomenon, then we have here an argument that ellipsis should be analyzed as a PF phenomenon.
5. Conclusion The examination of the syntax of multiple wh-fronting has shown that multiple wh-fronting, when it does not involve movement for checking of a [+Wh] feature in C, shares the syntactic behavior of identificational focus phrases. Multiple wh-fronting is then decomposable in a familiar movement for checking of a wh-feature in C and focus movement. Furthermore, the behavior of multiple wh-phrases with respect to Superiority in sluicing constructions has revealed that sluicing must be a PF phenomenon.
Notes 1. As will be discussed below, identificational focus is focus that expresses exhaustive identification and it subsumes contrastive focus. 2. Throughout the paper, identificationally focused material will be represented in capital letters and bold, while non-identificationally focused material will be represented in capital letters only. 3. Boškovic´ (1997b, 1998b) avoids giving indirect questions as examples of embedded questions because such questions involve an interfering factor. As Boškovic´ notes, indirect questions formally do not differ at all from matrix questions in SC. As a result, there is always a danger that they might be analyzed as matrix questions, with the superficial matrix clause
Multiple wh-fronting in Serbo-Croatian matrix questions 281
treated as an adsentential. Instead, Boškovic´ gives examples of correlative and existential constructions which, as shown by Izvorski (1996, 1998), also contain embedded questions. Boškovic´ does show that when this interfering factor in indirect questions is controlled for, true indirect questions in SC also exhibit Superiority effects. 4. See, however, Stjepanovic´ (1999) who provides evidence for the existence of phrases between AgrOP and T in SC, and argues that one of these phrases is reserved for hosting the discourse related elements. Furthermore, Stjepanovic´ (1999) shows that what traditionally are called VP adverbs could adjoin as high as this phrase, but crucially the highest position they can be found in is lower than the position in which sentential adverbs are found. See also Boeckx and Stjepanovic´ (2000) who argue for the existence of such a phrase. 5. Although better than (19b), (19a) and (19c) are somewhat degraded on the sentential adverb reading. This is due to the general incompatibility of sentential adverbs in questions. Who wisely advises whom is also somewhat degraded on the sentential reading of wisely. 6. One might suggest at this point that the order of wh-phrases in (22b)–(24b) is unacceptable because it does not follow the order of the indefinites in the antecedent sentence. However, this is not the case, as illustrated in (i) for (22a): (i) A:
Nekad je neko ovdje sakrio blago. some.time.ago is somebody here hidden treasure ‘Somebody hid the treasure here at some point in the past.’ B: ?Ko kad? who when ‘Who (hid the treasure) when?’ B: ?*Kad ko?
We can see that it is still better to have the higher wh-phrase first in the linear order, although even this response to the antecedent sentence of the speaker A is a bit unusual. The counterparts of (23a) and (24a) behave in the same way as the counterpart of (22a) in (i). It is worth checking whether the elliptical answers in the gapping pattern behave in the same way with respect to ordering. (ii) Ko je koga udario? who is whom hit ‘Who hit whom?’ (iii) a.
Marija Petra. Marija-nom Petar-acc ‘Marija Petar.’ b. ?*Petra Marija. Petar-acc Marija-nom ‘Marija Petar.’
While it is true that (iii-b) is an unnatural answer to (ii), this fact is not relevant to the examples in (22)–(24), since the non-elliptical source of (iii-b) has the same kind of degradation as an answer to (ii), which is not the case with corresponding wh-constructions (cf. (40)):
282 Sandra Stjepanovic´
(iv) a. #Petra je udarila Marija. Petar-acc is hit Marija-nom ‘Marija hit Petar.’ b. #Petra je Marija udarila. A degraded status of (iii-b) and (iv) as responses to (ii) may be due to constraints on the ways in which the information in a response to a question is organized, as discussed by Kuno (1982) and Kuno and Takami (1993). 7. Boškovic´ (1997b, 1998b) offers a slightly different account of these facts. Boškovic´ (1997b, 1998b) proposes that the strong focus feature resides not in the target of movement, but in the moving elements. See Boškovic´ (1997b, 1998b) for more discussion.
References Bolinger, D. 1978. “Asking more than one thing at a time.” In Questions, Hiz˙, H. (ed), 107–150. Dordrecht: Reidel. Boeckx, C. and Stjepanovic´, S. 2000. “The wh-clitic connection.” In Formal Approaches to Slavic Linguistics: The Philadelphia Meeting, 1999, T. H. King and I. Sekerina (eds), 22–40. Ann Arbor, Mich.: Michigan Slavic Publications. Boškovic´, Ž. 1997a. The Syntax of Non-finite Complementation: An Economy Approach. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 1997b. “Fronting wh-phrases in Serbo-Croatian.” In Formal Approaches to Slavic Linguistics: The Indiana Meeting, 1996, M. Lindseth and S. Franks (eds), 86–107. Ann Arbor, Mich.: Michigan Slavic publications. Boškovic´, Ž. 1997c. “Superiority effects with multiple wh-fronting in Serbo-Croatian.” Lingua 102: 1–20. Boškovic´, Ž. 1998a. “Multiple wh-fronting and economy of derivation.” In Proceedings of the Sixteenth West Coast Conference on Formal Linguistics, E. Curtis, J. Lyle and G. Webster (eds), 49–63. Stanford, Calif.: CSLI. Boškovic´, Ž. 1998b. Wh-movement and Wh-phrases in Slavic. Position paper presented at the Comparative Slavic Morphosyntax Workshop, Spencer, Ind. [June 1998] Boškovic´, Ž. 1999. “On multiple feature checking: Multiple wh-fronting and multiple head movement.” In Working Minimalism, S. D. Epstein and N. Hornstein (eds), 159–187. Cambridge, Mass.: MIT Press. Boškovic´, Ž. 2002. “On multiple wh-fronting.” Linguistic Inquiry 33: 351–383. Cheng L. and Demirdache, H. 1990. “Superiority Violations.” In MIT Working Papers in Linguistics 13: Papers on Wh-Movement, L. Cheng and H. Demirdache (eds), 27–46. Cambridge, Mass.: MITWPL (Department of Linguistics and Philosophy, MIT). Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Chung, S., Ladusaw, L. and McCloskey, J. 1995. “Sluicing and logical form.” Natural Language Semantics 3: 1–44. Göbbel, E. 1998. “Focus movement in Romanian.” In Proceedings of the Workshop on Focus. University of Massachusetts Occasional Papers in Linguistics 21, E. Benedicto, M. Romero
Multiple wh-fronting in Serbo-Croatian matrix questions 283
and S. Tomioka (eds), 83–99. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Horvath, J. 1986. Focus in the Theory of Grammar and the Syntax of Hungarian. Dordrecht: Foris. Izvorski, R. 1996. “The syntax and semantics of correlative proforms.” In Proceedings of NELS 26, K. Kusumoto (ed), 133–147. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Izvorski, R. 1998. “Non-indicative wh-complements of possessive and existential predicates.” In Proceedings of NELS 28, P. N. Tamanji and K. Kusumoto (eds), 159–173. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Jackendoff, R. 1971. “Gapping and related rules.” Linguistic Inquiry 2: 21–36. Jackendoff, R. 1972. Semantic Interpretation in Generative Grammar. Cambridge, Mass.: MIT Press. Kidwai, A. 1999. Word order and focus positions in Universal Grammar. Ms., Jawaharlal Nehru University. Kim, J.-S. 1998. Syntactic focus movement and ellipsis: A minimalist approach. Doctoral dissertation, University of Connecticut, Storrs. É. Kiss, K. (ed). 1995. Discourse Configurational Languages. Oxford: Oxford University Press. É. Kiss, K. 1998. “Identificational focus versus information focus.” Language 74: 245–273. Kitahara, H. 1997. Elementary Operations and Optimal Derivations. Cambridge, Mass.: MIT Press. Kuno, S. 1982. “The focus of the question and the focus of the answer.” In Papers from the Parasession on Nondeclaratives. 18th Regional Meeting of the Chicago Linguistic Society, R. Schneider, K. Tuite and M. Marks (eds), 134–157. Chicago, Ill.: Chicago Linguistic Society. Kuno, S. and K. Takami. 1993. Grammar and Discourse Principles: Functional Syntax and GB Theory. Chicago, Ill.: The University of Chicago Press. Lasnik, H. 1999. “On feature strength: Three minimalist approaches to overt movement.” Linguistic Inquiry 30: 197–217. Lasnik, H. and Saito, M. 1992. Move Alpha. Cambridge, Mass.: MIT Press. Levin, L. S. 1982. “Sluicing: a lexical interpretation procedure.” In The Mental Representation of Grammatical Relations, J. Bresnan (ed), 590–654. Cambridge, Mass.: MIT Press. Merchant, J. 1999. The syntax of silence: Sluicing, islands and identity in ellipsis. Doctoral dissertation, University of California, Santa Cruz. Ouhalla, J. 1994. “Focus in Standard Arabic.” Linguistics in Potsdam 1, 65–92, Potsdam: Department of Linguistics, University of Potsdam. Pesetsky, D. 1987. “Wh-in situ, movement and unselective binding.” In The Representation of (In)definites, E. Reuland and A. ter Meulen (eds), 98–129. Cambridge, Mass.: MIT Press. Richards, N. 1997. What moves where when in which language? Doctoral dissertation, Massachusetts Institute of Technology, Cambridge. Rochemont, M. 1986. Focus in Generative Grammar. Amsterdam: John Benjamins. Rosen, C. 1976. “Guess What About.” In Proceedings of NELS 6, A. Ford, J. Reighard and R. Singh, (eds), 205–211. Montreal: Montreal Working Papers in Linguistics. Ross, J. R. 1969. “Guess who?” In Papers from the 5th Regional Meeting of the Chicago Linguistic Society, R. Binnick, A. Davison, G. Green and J. Morgan (eds), 252–286. Chicago, Ill.: Chicago Linguistics Society.
284 Sandra Stjepanovic´
Rudin, C. 1988. “On multiple questions and multiple wh-fronting.” Natural Language and Linguistic Theory 6: 455–501. Szabolcsi, A. 1981. “The Semantics of topic–focus articulation. In Formal Methods in Study of Language, J. Groenendijk, T. Janssen and M. Stokhof (eds), 513–541. Amsterdam: Matematisch Centrum. Stepanov, A. 1998. “On wh-fronting in Russian.” In Proceedings of NELS 28, P. N. Tamanji and K. Kusumoto (eds), 453–467. Amherst, Mass.: GLSA (Department of Linguistics, University of Massachusetts). Stjepanovic´, S. 1998. Short distance movement of wh-phrases in Serbo-Croatian matrix clauses. Paper presented at the Comparative Slavic Morphosyntax Workshop, Spencer, Ind. [June 1998] Stjepanovic´, S. 1999. What do second position cliticization, scrambling, and multiple wh-fronting have in common? Doctoral dissertation, University of Connecticut, Storrs. Takahashi, D. 1994. “Sluicing in Japanese.” Journal of East Asian Linguistics 3: 265–300. Tsimpli, I. 1994. “Focusing in Modern Greek.” In Discourse Configurational Languages, K. É. Kiss (ed), 176–206. Oxford: Oxford University Press. Vilkuna, M. 1995. “Discourse configurationality in Finnish.” In Discourse Configurational Languages, K. É. Kiss (ed), 244–268. Oxford: Oxford University Press. Watanabe, A. 1993. Agr-based Case Theory and its interaction with the A¢-system. Doctoral dissertation. Massachusetts Institute of Technology, Cambridge.
In the series LINGUISTIK AKTUELL/LINGUISTICS TODAY (LA) the following titles have been published thus far, or are scheduled for publication: 1. KLAPPENBACH, Ruth (1911-1977): Studien zur Modernen Deutschen Lexikographie. Auswahl aus den Lexikographischen Arbeiten von Ruth Klappenbach, erweitert um drei Beiträge von Helene Malige-Klappenbach. 1980. 2. EHLICH, Konrad & Jochen REHBEIN: Augenkommunikation. Methodenreflexion und Beispielanalyse. 1982. 3. ABRAHAM, Werner (ed.): On the Formal Syntax of the Westgermania. Papers from the 3rd Groningen Grammar Talks (3e Groninger Grammatikgespräche), Groningen, January 1981. 1983. 4. ABRAHAM, Werner & Sjaak De MEIJ (eds): Topic, Focus and Configurationality. Papers from the 6th Groningen Grammar Talks, Groningen, 1984. 1986. 5. GREWENDORF, Günther and Wolfgang STERNEFELD (eds): Scrambling and Barriers. 1990. 6. BHATT, Christa, Elisabeth LÖBEL and Claudia SCHMIDT (eds): Syntactic Phrase Structure Phenomena in Noun Phrases and Sentences. 1989. 7. ÅFARLI, Tor A.: The Syntax of Norwegian Passive Constructions. 1992. 8. FANSELOW, Gisbert (ed.): The Parametrization of Universal Grammar. 1993. 9. GELDEREN, Elly van: The Rise of Functional Categories. 1993. 10. CINQUE, Guglielmo and Guiliana GIUSTI (eds): Advances in Roumanian Linguistics. 1995. 11. LUTZ, Uli and Jürgen PAFEL (eds): On Extraction and Extraposition in German. 1995. 12. ABRAHAM, W., S. EPSTEIN, H. THRÁINSSON and C.J.W. ZWART (eds): Minimal Ideas. Linguistic studies in the minimalist framework. 1996. 13. ALEXIADOU Artemis and T. Alan HALL (eds): Studies on Universal Grammar and Typological Variation. 1997. 14. ANAGNOSTOPOULOU, Elena, Henk VAN RIEMSDIJK and Frans ZWARTS (eds): Materials on Left Dislocation. 1997. 15. ROHRBACHER, Bernhard Wolfgang: Morphology-Driven Syntax. A theory of V to I raising and pro-drop. 1999. 16. LIU, FENG-HSI: Scope and Specificity. 1997. 17. BEERMAN, Dorothee, David LEBLANC and Henk van RIEMSDIJK (eds): Rightward Movement. 1997. 18. ALEXIADOU, Artemis: Adverb Placement. A case study in antisymmetric syntax. 1997. 19. JOSEFSSON, Gunlög: Minimal Words in a Minimal Syntax. Word formation in Swedish. 1998. 20. LAENZLINGER, Christopher: Comparative Studies in Word Order Variation. Adverbs, pronouns, and clause structure in Romance and Germanic. 1998. 21. KLEIN, Henny: Adverbs of Degree in Dutch and Related Languages. 1998. 22. ALEXIADOU, Artemis and Chris WILDER (eds): Possessors, Predicates and Movement in the Determiner Phrase. 1998. 23. GIANNAKIDOU, Anastasia: Polarity Sensitivity as (Non)Veridical Dependency. 1998. 24. REBUSCHI, Georges and Laurice TULLER (eds): The Grammar of Focus. 1999. 25. FELSER, Claudia: Verbal Complement Clauses. A minimalist study of direct perception constructions. 1999.
26. ACKEMA, Peter: Issues in Morphosyntax. 1999. ° 27. RUZICKA, Rudolf: Control in Grammar and Pragmatics. A cross-linguistic study. 1999. 28. HERMANS, Ben and Marc van OOSTENDORP (eds): The Derivational Residue in Phonological Optimality Theory. 1999. 29. MIYAMOTO, Tadao: The Light Verb Construction in Japanese. The role of the verbal noun. 1999. 30. BEUKEMA, Frits and Marcel den DIKKEN (eds): Clitic Phenomena in European Languages. 2000. 31. SVENONIUS, Peter (ed.): The Derivation of VO and OV. 2000. 32. ALEXIADOU, Artemis, Paul LAW, André MEINUNGER and Chris WILDER (eds): The Syntax of Relative Clauses. 2000. 33. PUSKÁS, Genoveva: Word Order in Hungarian. The syntax of È-positions. 2000. 34. REULAND, Eric (ed.): Arguments and Case. Explaining Burzio’s Generalization. 2000. 35. HRÓARSDÓTTIR, Thorbjörg. Word Order Change in Icelandic. From OV to VO. 2000. 36. GERLACH, Birgit and Janet GRIJZENHOUT (eds): Clitics in Phonology, Morphology and Syntax. 2000. 37. LUTZ, Uli, Gereon MÜLLER and Arnim von STECHOW (eds): Wh-Scope Marking. 2000. 38. MEINUNGER, André: Syntactic Aspects of Topic and Comment. 2000. 39. GELDEREN, Elly van: A History of English Reflexive Pronouns. Person, ‘‘Self’’, and Interpretability. 2000. 40. HOEKSEMA, Jack, Hotze RULLMANN, Victor SANCHEZ-VALENCIA and Ton van der WOUDEN (eds): Perspectives on Negation and Polarity Items. 2001. 41. ZELLER, Jochen : Particle Verbs and Local Domains. 2001. 42. ALEXIADOU, Artemis : Functional Structure in Nominals. Nominalization and ergativity. 2001. 43. FEATHERSTON, Sam: Empty Categories in Sentence Processing. 2001. 44. TAYLAN, Eser E. (ed.): The Verb in Turkish. 2002. 45. ABRAHAM, Werner and C. Jan-Wouter ZWART (eds): Issues in Formal German(ic) Typology. 2002. 46. PANAGIOTIDIS, Phoevos: Pronouns, Clitics and Empty Nouns. ‘Pronominality’ and licensing in syntax. 2002. 47. BARBIERS, Sjef, Frits BEUKEMA and Wim van der WURFF (eds): Modality and its Interaction with the Verbal System. 2002. 48. ALEXIADOU, Artemis, Elena ANAGNOSTOPOULOU, Sjef BARBIERS and HansMartin GAERTNER (eds): Dimensions of Movement. From features to remnants. 2002 49. ALEXIADOU, Artemis (ed.): Theoretical Approaches to Universals. 2002. 50. STEINBACH, Markus: Middle Voice. A comparative study in the syntax-semantics interface of German. 2002. 51. GERLACH, Birgit: Clitics between Syntax and Lexicon. 2002. 52. SIMON, Horst J. and Heike WIESE (eds): Pronouns – Grammar and Representation. 2002.
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)(Workshop). 2002. 54. BAPTISTA, Marlyse: The Syntax of Cape Verdean Creole. The Sotavento varieties. 2002. 55. COENE, M. and Yves D'HULST (eds): From NP to DP. Volume 1: The syntax and semantics of noun phrases. 2003. 56. COENE, M. and Yves D'HULST (eds.): From NP to DP. Volume 2: The expression of possession in noun phrases. 2003. 57. DI SCIULLO, Anna Maria (ed.): Asymmetry in Grammar. Volume 1: Syntax and semantics. 2003. 58. DI SCIULLO, Anna Maria (ed.): Asymmetry in Grammar. Volume 2: Morphology, phonology, acquisition. 2003. 59. DEHÉ, Nicole: Particle Verbs in English. Syntax, information structure and intonation. 2002. 60. TRIPS, Carola: From OV to VO in Early Middle English. 2002. 61. SCHWABE, Kerstin and Susanne WINKLER (eds.): The Interfaces. Deriving and interpreting omitted structures. 2003. 62. CARNIE, Andrew, Heidi HARLEY and Mary WILLIE (eds.): Formal Approaches to Function in Grammar. In honor of Eloise Jelinek. 2003. 63. BOECKX, Cedric: Islands and Chains. Resumption as stranding. 2003. 64. BOECKX, Cedric and Kleanthes K. GROHMANN (eds.): Multiple Wh-Fronting. 2003. 65. MANNINEN, Satu Helena: Small Phrase Layers. A study of Finnish Manner Adverbials. N.Y.P. 66. GROHMANN, Kleanthes K.: Prolific Domains. On the anti-locality of movement dependencies. N.Y.P.