University of Southern California Dissertations and Theses

Asymmetry of scope taking in wh -questions

(USC Thesis Other)

Asymmetry of scope taking in wh -questions

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ASYMMETRY OF SCOPE TAKING IN WH-QUESTIONS Copyright 2002 by Young-Sik Choi A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (LINGUISTICS) August 2002 Young-Sik Choi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3094316 UMI UMI Microform 3094316 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90089-1695 This dissertation, written by _______Y o u n g -S ik C h o i under the direction of h i s dissertation committee, and approved by all its members, has been presented to and accepted by the Director of Graduate and Professional Programs, in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY ■ ........ • ' Director Date Aug u s t f t , 7007 Dissertation Committee ChaT, MC Uc Reproduced with permission of the copyright owner. Further reproduction prohibited without permission Dedication To Jin-Hee Kim, Bomin Choi, and Ines J. Choi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Acknowledgments iii It has been a long journey but was w orth taking. Like my predecessors it is time to say thanks to those people for w hat they did for what I am. Among others, I would like to express my special thanks to my advisor, Jean-Roger Vergnaud, without whose constant encouragement and support, I would not have had the luxury o f writing this acknowledgement. A teacher like Jean-Roger is indeed a rare find, a hum an being with a warm personality and a scholar with a heraldic vision on linguistics as a cognitive science. I am proud that I am among the few selected to be his student. The many meetings at his home and elsewhere around, which mostly lasted more than several hours, will always remain a fond memory o f my life in this part o f the world. His patience in listening to my unorganized thoughts and trying to make sense out o f them was admirable. I would also like to thank the rest o f my dissertation committee, M aria Luisa Zubizarreta and Mario Saltareli, for their excellent service. M aria Luisa has served on every committee, including screening and qualification committees, that I have had to form during my studies at USC. If one is asked who is the m ost notorious in the department for sharp and penetrating comments, he or she will name her without much hesitation. It was always a pleasure to visit M ario’s office for an appointment, occasionally extended to an extra coffee session somewhere on campus. My thanks also go to Joseph Aoun, Barry Schein, and Audrey Li, who served on my qualification committee, and whose warm encouragement and guidance I will cherish in my heart. Hagit Borer and Hajime Hoji also deserve my thanks for many appointments with me when I first came to USC, and it was my honor to have Hagit on my screening committee. I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iv also would like to thank Jim Higginbotham, Barry Schein, and Philippe Schlenker for offering the many semantics classes that I should say were pretty much enjoyable. Rachel W alker and Dani Byrd also deserve my thanks for their excellent phonology and phonetics classes. Also I w ould like to thank Paul Hagstrom for commenting on the draft o f this dissertation and Nathan Klinedinst for commenting on and proofreading it, both out o f good friendship. Any shortcomings in my dissertation are solely due to the lack o f my own imagination. I also would like to thank Uffe Bergeton and Carolina Gonzalez who came to my defense, taking their valuable time. Uffe Bergeton and Chris Long deserve my special thanks. We talked a lot about things about our life, linguistic and non-linguistic. Their friendship made my life here much more tolerable. I also appreciate USC for the Predoctoral M erit Fellowship and Graduate Supplemental Award that enabled me to concentrate on my studies without worrying about the mundane but important aspects o f life. My thanks also go to the staff members in the department, Linda Culver, Patricia M urdoch and Karm a Dolma. They are always available for the kind o f help a graduate student needs in the department. To people on the other side o f the world, I thank my teacher Kiyong Lee in Korea University for guiding me into the wonderful world o f linguistics. I also thank my other teachers in Korea University, Jongkeon Kim, O k Song, Yongjae Lee, Kyungja Park, Kunjong Lee, and Keeho Kim, for their teaching and guidance. I also thank my teachers in Seoul National University, Chungm in Lee and Dong-W hee Yang, for their encouragement and support when I was most in need. Soon-Se Hong deserves my special thanks. He has never failed to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V have a firm belief that someday I can prevail. I hope that I will m eet his expectation someday. W hen it comes to my own family, I w ould first and foremost like to thank my wife, Jin-Hee Kim, and our two children, Bomin Choi and Ines J. Choi. My wife has made my life most enjoyable, with all the emotional support and willing companionship one can hope to get throughout the journey. We have had every moment together, either bitter or sweet, offering shoulders to each other. I hope that someday I can return to her what she has done for me. I also want to thank our two children for the many nights o f waiting for daddy’s returning home from school. I am glad I can come home early today. Finally, I also should mention my parents-in-law, Chwun-Kwon Kim and Ki-Soon Hwang, who have never failed to render their warm encouragement. I am sure my parents Soon-Cheol Choi and Hang-A Jang would certainly be glad to know that what I have done is the result of the cooperative effort o f our family, if they ever know. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table of Contents Dedication.....................................................................................................................................................ii Acknowledgments..................................................................................................................................... iii Abstract........................................................................................................................................................ viii CHAPTER 1 .Introduction.......................................................................................................................... 1 CHAPTER 2. Asymmetry o f Scope Taking in W h-Questions 1 .Introduction....................................................................................................................................... 13 2. W h-words as Indefinites..............................................................................................................23 3. The Interpretation o f W h-Questions.........................................................................................39 4. QP and W h-word Scope Interaction.........................................................................................41 4.1. Asymmetry o f Scope................................................................................................................ 43 4.1.1. Simplex W h-Question.......................................................................................................... 43 4.1.1.1. Uniform LF W h-movement A pproach.........................................................................55 4. 1.1.2.Uniform W h-in-situ A pproach........................................................................................77 4.1.1.3. Proposal.................................................................................................................................80 4.2. Symmetry o f Scope.................................................................................................................108 4.2.1. Scrambled W h-question..................................................................................................... 109 4.2.1.1 .Lack o f Superiority Effect............................................................................................... 121 4.2.1.2.Lack o f Reconstruction....................................................................................................123 4.2.1.2.1 .Lack o f Bound Variable Interpretation......................................................................124 4.2.1.2.2.Lack o f Binding Condition C E ffects........................................................................ 128 4.2.1.2.3. Lack o f Scope reconstruction o f way (w hy)........................................................... 128 4.2.2. Complex W h-Questions....................................................................................................132 5. Correlation.....................................................................................................................................137 5.1. W h-questions with an N PI..................................................................................................... 137 6. Functional A nsw er..................................................................................................................... 144 7. Conclusion......................................................................................................................................152 CHAPTER 3. Asymmetry o f Scope in Japanese and Chinese W h-questions 1. Introduction.................................................................................................................................. 154 2. W h-words as Indefinites...........................................................................................................155 3. The Representation o f W h-questions.....................................................................................159 4. QP and W h-word Scope Interaction....................................................................................... 160 4.1. The Position o f QM in Japanese and Chinese.................................................................. 161 4.2. Japanese......................................................................................................................................165 4.3.Chines e ........................................................................................................................................ 179 5. Conclusion.....................................................................................................................................187 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vii CHAPTER 4. Asymmetry o f Locality Effects in Islands 1. Introduction.......................................................................................................................................188 2.Locality Effects in Islands........................................................................................................... 189 2.1.Korean...........................................................................................................................................193 2.1.1 .Complex N oun Phrase Island........................................................................................... 193 2.1.2.Adjunct Island.............................................................................................................................196 2.1.3.Sentential Subject Island.......................................................................................................198 2.2. Japanese and Chinese...............................................................................................................200 2.2.1.Complex Noun Phrase Island.............................................................................................. 201 2.2.2.Adjunct Islands........................................................................................................................207 2.2.3. Sentential Subject Island......................................................................................................210 3. Pied-piping A pproach..................................................................................................................214 4.Tsai (1994).......................................................................................................................................224 5. Account........................................................................................................................................... 226 6. Conclusion......................................................................................................................................234 CHAPTER 5. Symmetry o f Locality in W h-Island Constructions 1. Introduction....................................................................................................................................... 236 2. W H -island.........................................................................................................................................237 2.1.Korea n ............................................................................................................................................ 243 2.2. Chinese and Japanese................................................................................................................261 3. Analysis............................................................................................................................................. 282 4. Conclusion........................................................................................................................................ 297 References.................................................................................................................................................300 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Abstract viii A growing body o f research has suggested that wh-words in Korean type languages, which normally do not have overt wh-movement, cannot be treated on a par with wh-words in English, since wh-words in languages o f the former type can have different interpretations depending on the context in which they occur. In this thesis, I claim that in Korean type languages wh-words corresponding to why in English stand apart from the rest o f wh-words in that the former are not an indefinite in the sense o f Lewis (1975) and Heim (1982), while the latter are. I will call the former propositional adjunct wh-words and the latter nonpropositional wh-words, since the former can typically quantify over propositions, contrary to the latter. I further claim that the difference between the two classes o f wh-words in their status as indefinites is directly associated with the asymmetric scope taking between them in wh-questions. Nonpropositional wh-words are unselectively bound by the question morpheme, which acts as a wh-operator, marking their scope, while the propositional adjunct wh-word, not being an indefinite, should undergo movement at LF into Spec o f CP via spec- head agreement to mark its scope thereby, driven by the need for proper interpretation. The asymmetry o f scope taking between the two classes o f wh-words in Korean type languages thus has an interesting implication for two important phenomena I will discuss in this thesis: scope and locality in wh-questions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 CHAPTER 1 Introduction There has been a growing body o f research suggesting that wh-words in Korean type languages are different from wh-words in English in that they are indefinites (Nishiguachi 1990, Cheng 1991, Li 1992, and Tsai 1994, Lin 1996, among others). I claim in this thesis that the adjunct wh-words in Korean type languages corresponding to why in English stand apart from the rest o f wh-words that quantify over individuals, measure, means and manner in that the former are not indefinites in the sense o f Lewis (1975) and Heim (1982), while the latter are. I will call the former propositional adjunct wh-words and the latter nonpropositional wh-words, since the former can typically quantify over propositions, in contrast to the latter. I will also claim that the difference between the two classes o f wh-words regarding their status as indefinites is directly associated with the asymmetry o f scope taking between them in wh-questions: a nonpropositional wh-word is unselectively bound by the question morpheme that serves as a wh-operator and marks its scope, while the propositional adjunct wh-word, not being an indefinite, undergoes movement at LF into Spec o f CP to mark its scope via spec-head agreement with the QM, driven by the need for proper interpretation. The asymmetry o f scope taking between the two types o f wh-words has an interesting implication for scope and locality in wh-questions in Korean type languages, which will be the main topic o f this thesis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 The status o f nonpropositional wh-words as indefinites, as opposed to the propositional adjunct wh-words, is attested by two important properties o f indefinites, observed in Lewis (1975) and Heim (1982). One is that an indefinite does not have any quantificational force of its own, its quantificational force instead being essentially determined by an external quantificational element. The other is that the indefinite is not subject to standard syntactic islands in taking scope. These are illustrated by the following examples in English: (1) a. If a m an owns a donkey he always beats it. b. Sometimes, if a cat falls from the fifth floor, it survives. (Heim 1982: 123) The quantificational force o f the indefinites in (1) is determined by what Lewis (1975) calls adverbs o f quantification in the main clause, as illustrated by the following paraphrases of the examples in (1): (2) a. For every man and every donkey such that the former owns the latter, he beats it. b. Some cats that fall from the fifth floor survive. (Heim 1982: 127) According to Lewis (1975), adverbial quantifiers like always and sometimes in (1) take the if- clause as restriction and the consequent as nuclear scope, and they are treated as unselective binders, in the sense that they can bind more than one variable, the former interpreted as in (3 a), and the latter as in (3b). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 (3) a. V<Dis true iff <bis true under every admissible assignment o f values to all variables free in O. b. 3 0 is true iff O is true under some admissible assignment o f values to all variables free in < D . (Lewis 1975: 7) Thus when the interpretation rules (3a) and (3b) are applied to the sentences in (la ) and (lb ), respectively, they will be interpreted as in (4): (4) a. Always x y ((x is a man and y is a donkey and x owns y) x beats y)) b. Sometimes x ((x is a cat and x falls from the fifth floor) x will survive)) The informal logical notations in (4) clearly show that the indefinites can take scope out of the syntactic island, namely, the adjunct island. W hen it comes to wh-words in Korean type languages, nonpropositional wh-words exhibit precisely these two properties, in other words, lack o f their own quantificational force and the ability to scope out o f standard islands, as illustrated by the Korean example below in (5) involving adverbs o f quantification construction. (5) a. [C p N w urka o-myen] (proj) kkok wuli-lul pangmwunhanta. who-NOM com e-if always us-ACC visit ‘for every x, x an individual, if x comes, x visits us.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. [ c p LA-eyse enu chinkwiii-ka o-myen] (proO kkok wuli-lul pangmwunhanta. LA-from which friend-NOM com e-if always us-ACC visit ‘for every x, x a friend, if x comes from LA, x visits us.’ c. [cpJohnj-i eti-lo oychwulha-myen] (pro;) kkok nusskey tolaonta. J-NOM where-to go out-if always late returns ‘for every x, x a place, if John goes out to x, he returns home late (from x).’ d. [ c p Johnj -i encey o-myen] (pro;) kkok wuli-ul pangmwunhanta. J-NOM when com e-if always us-ACC visit ‘for every x, x a time, if John comes at x, he visits us (at x).’ e. [ c p M aryj -nun cheycwung-i caki,-uy cengsang cheycwung-pota M -TOP weight-Nom her-POSS normal weight-than elma te naka-myen ] (pros) kkok diet-lul hanta. how much more w eigh-if always diet-ACC do ‘for every x, x a weight, if M ary weighs x more than normal weight, she goes on a diet (for x).’ f. [C p Johnri ettehkey wuli cipey o-myen] (prod kkok senmwul-ul kacikoonta. J-NOM how our house-to com e-if always gift-ACC bring ‘for every x, x a means, if John comes to our place by x, he brings a gift (by x).’ g. [cp Nay-ka ettehkey John-eytayhay malha-myen] I-NOM how J-about talk-if M ary-nun kkok hwalul naykonhanta. M -TOP always get angry ‘for every x, x a manner, if I talk about John in x, Mary gets angry (with x).’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 As shown by the informal logical notations in (5), the quantificational force o f the nonpropositional wh-word is determined by the external quantificational element kkok (always), which conveys universal quantificational force and the wh-word can take scope out o f the syntactic island, i.e., adjunct island. M eanwhile, the propositional adjunct wh-word way (why) in analogous construction in (6) cannot be construed in an analogous way: (6)*[John-i way o-myen] kkok wuli-lul pangmwunhanta. J-NOM why com e-if always us-ACC visit ‘for every x, x a reason if John comes for x, John visits us (for x).’ Given the two properties o f an indefinite discussed above, one can establish that the propositional adjunct wh-word way (why) is not an indefinite, while nonpropositional wh- words are. Our proposal is reminiscent o f the proposal o f Tsai (1994), according to which nominal wh- words, but not nonnominal wh-words, are indefinites in Chinese. Tsai claims that in Chinese argument wh-words in the sense o f Huang (1982), the adjunct wh-words zenmeyang (how) under means and resultative construal, and weishenme (why) under purpose construal are nominal, while these adjunct wh-words are nonnominal under manner construal and reason construal, respectively. However, his dichotomy for indefmiteness along nominal vs. nonnominal wh-words seems to be flawed, given Korean example with an adjunct wh-word having a manner construal in (5g), for example, which still admits an indefinite construal as shown by the informal logical notation, contrary to the prediction his proposal makes. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 The different status o f wh-words as indefinites has a very interesting implication when it comes to the interpretation o f wh-words as wh-interrogatives. The notable difference between English and Korean type languages regarding wh-questions is that in English wh- words are overtly raised to sentence-initial position to m ark their scope, while in Korean type languages they remain in situ, with the question morpheme marking their scope, as illustrated by the English and Korean examples below in (7-8), respectively. 1 (7) a. Who did John meet? b. W hy did John come? (8) a. John-i nwukwu-lul mannass-ni? J-NOM whom-ACC met-QM ‘W ho did John m eet?’ ‘Did John m eet anyone?’ b. John-i way wass-ni? J-NOM why came-QM ‘Why did John com e?’ # ‘Did John come for any reason?’ 1 The question morpheme ni as in (8) is rather an informal one, widely used in everyday conversation in standard Korean dialect. Korean has a polite and formal question morpheme, ka, which is adopted when the hearer is older or higher in social position than the speaker. When ka is used, an honorific morpheme is also added before the question morpheme. The particular choice of the question morpheme, however, does not affect our discussion here and thus I will use the informal question morpheme ni throughout this paper for the simplicity’s sake. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 In Korean, which is a typical instance o f a head-final language with an in-situ wh-word, as illustrated in (8), the question morpheme ni in (8a) with a nonpropositional wh-word, unlike the one in (8b) with a propositional adjunct wh-word, can be construed either as a yes-no question morpheme or as a wh-question morpheme, and thus prosody is crucial for the disambiguation o f the sentence in (8a). Rising intonation at the end o f the sentence on the question morpheme gives rise to a yes-no question while falling intonation on it yields a wh- question. One may thus posit two different question morphemes in terms o f [Q, WH] feature specification, a question morpheme with a [+Q, +W H], giving rise to a wh-question and a question morpheme with a [+Q, -W H], giving rise to a yes-no question. Following the intuitions in the literature (Chomsky 1962, 1964: 38, Katz and Postal 1964: 93, Klima 1964: 253, K uroda 1969: 266ff, Vergnaud and Zubizarreta 2002, among others), according to which a wh-word is viewed as a combination o f wh-operator and indefinite some {one, thing...) in English, I suggest that the question morpheme (QM , henceforth) with the [+Q, +WH] feature specification in Korean type languages is equivalent to a wh-operator and that the nonpropositional wh-word corresponds to the indefinite some {one, thing...). I thus suggest that in the sense o f Heim (1982), the QM as a wh-operator unselectively binds and thus marks the scope o f the nonpropositional wh-word, which acts as a variable (see Baker 1970). W hen it comes to a propositional adjunct wh-word as in (8b), I suggest that it should undergo movement into Spec o f CP at LF via spec-head agreement with the QM, driven by the need for proper interpretation by forming an operator variable chain, since it does not contain indefinite some part to be bound by the wh-operator. Thus when it comes to the propositional Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. adjunct wh-word, the QM does not behave as a wh-operator, but just as a position holder, such that the propositional adjunct wh-word, which enters into spec-head agreement with it, checks o ff its wh-feature and marks its scope therein at LF. I will show that the different scope taking properties o f the two types of wh-words in wh-questions can provide a natural account for the two important phenom ena I investigate in this thesis: scope and locality in wh-questions in Korean type languages. The present research into scope phenomena in wh-questions in Korean reveals two interesting facts in Chapter 2. W hen it comes to a simplex wh-question i.e., a mono clausal wh-question, there exists an interesting asymmetry o f scope between the two types o f wh- words: In simplex wh-questions, nonpropositional wh-words cannot take wide scope over a matrix subject QP, while propositional adjunct wh-word way (why) can. When simplex questions are embedded to form complex wh-questions, i.e., bi-clausal wh- questions where an embedded subject QP is followed by a wh-word, the asymmetry between the two types o f wh-words disappears, with both types o f wh-words being able to take scope over the embedded subject QP. This interesting asymmetry and symmetry o f scope taking in wh-questions between the two classes o f wh-words comes out only with the introduction of QPs whose meaning can only be captured using higher order logic, such as QPs corresponding to most NP in English, and modified QPs o f various semantic types such as monotone-decreasing, m onotone-increasing or non-monotonic. To the extent that one confines his research to wh-questions with standard ‘first order logic’ QPs such as someone and everyone, as has typically been the case in the Korean literature, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 this important asymmetry and symmetry of scope between the two types of wh-words I observe in wh-questions in Korean does not show up on the surface. It will be shown that any theory (e.g. Hoji 1985, Kim 1990, Aoun and Li 1993b), which fails to recognize the asymmetric scope taking properties o f the two types o f wh-words described in this thesis is bound to fail to give a full account o f the scope interaction between wh-words and QPs. The scope pattern in other wh-in-situ languages such as Japanese and Chinese, investigated in Chapter 3, further confirms the interesting symmetry and asymmetry o f scope between the two types o f wh-words: In a simplex wh-question, a nonpropositional wh-word cannot take wide scope over a subject QP, while a propositional adjunct wh-word can. In a complex wh- question, where an embedded subject QP is followed by a wh-word, both types o f wh-words can take scope over the embedded subject QP. This observation, if correct, has an important implication for the proposals in the literature, namely, the standard observations in the literature (see Huang 1982, Hoji 1985, Aoun and Li 1989, 1993abc), according to which in simplex wh-questions wh-words can invariably take scope over subject QPs in Chinese, but not in Japanese and Korean, are not correct. Chapter 4 is devoted to asymmetry o f locality effects when it comes to wh-questions involving standard syntactic islands such as Complex Noun Phrase Island construction (Ross 1967), Sentential Subject Island construction (Ross 1967), and Adjunct Island construction (Huang 1982). The present thesis will show that the proposed dichotomy o f wh-words regarding indefmiteness, o f propositional adjunct wh-words on the one hand and nonpropositional wh-words on the other is further confirmed in these constructions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 The important proposals in the literature, such as the ECP approach o f Huang (1982), unselective binding approach o f Tsai (1994), and pied-piping approach o f Nishigauchi (1990) will be reviewed, and potential problems with them will be raised. The problem in Huang and Tsai has to do primarily with the proposed dichotomy o f wh-words along the dimension o f argument vs. adjunct, and nominal vs. nonnominal, respectively. Huang claims that an adjunct wh-word cannot scope out o f a syntactic island due to ECP violation and Tsai claims that a nonnominal wh-word cannot scope out o f it since it cannot be construed via unselective binding. The dichotomy as proposed by them is flawed, given that an adjunct wh- word under manner construal, which cannot be either an argument wh-word in Huang (1982) or a nominal wh-word in Tsai (1994) can still scope out o f a syntactic island in Korean type languages, as illustrated by the Korean example in (9). (9) Ne-nun [M ary-ka ne-eytayhay ettehkey malha-myen] cohkeyss-ni? you-TOP M -NOM you-about how talk-if will be good-QM ‘W hat is the manner x such that if M ary talks about you in x you will be happy?’ Nishigauchi’s pied-piping approach is also problematic in that, among other problems, the mechanism runs into a fatal problem o f yielding the wrong semantics, as noted in von Stechow (1996). Moreover, his approach also invites a nontrivial ontological question: If wh- words are indefinites in general in Korean type languages as he claims, why then can they not take scope out o f syntactic islands without movement? Does that mean the grammar has another type o f indefinite that is subject to island constraints? Our proposal for the asymmetric scope taking properties o f propositional adjunct wh-words and nonpropositional wh-words can account for the asymmetry o f locality effects between them without running Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 into the problem o f wrong semantics in Nishigauchi and without necessitating two types o f indefinites in the grammatical ontology. Chapter 5 deals with a surprising symmetry o f scope in W h-island constructions (Chomsky 1962, Ross 1967) in Korean type languages. I observe that both propositional adjunct wh- words and nonpropositional wh-words do exhibit locality symmetrically in that they do not exhibit the W h-island effects when scoping out o f the island. This symmetry is quite surprising, given the sharp asymmetry o f locality between the two types o f wh-words when it comes to islands such as Complex Noun Phrase Islands, Adjunct Islands, and Subject Islands, as discussed in Chapter 4. This thesis, however, shows that the proposed different ways in which propositional and nonpropositional wh-words take scope are both compatible with their escaping W h-islands, thus accounting for the empirical fact that no asymmetry exists between the two types o f wh-words with respect to W h-island locality effects, and at the same time maintaining the asymmetry with respect to non-W h-island locality derived in Chapter 4. The present observation for the symmetry o f locality in W h-island constructions in Korean type languages, i.e., the lack o f W h-island effects, if correct, crucially undermines the observations in Huang (1982) and Nishigauchi (1990), according to which W h-island effects exist in these languages. To the extent that the observations I make with respect to scope and locality in wh-questions in Korean type languages, are correct, the current proposal that there exist two independent scope taking strategies for in-situ wh-words seems to be defendable: unselective binding by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 the QM o f nonpropositional wh-words at LF, and movement o f propositional adjunct wh- words into Spec o f CP via spec-head agreement with the QM at LF, driven by the need for proper interpretation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 CHAPTER2 Asymmetry of Scope Taking in Wh-Questions 1. Introduction Two approaches have been taken for the scope taking o f wh-words in Korean type languages. One is that they move at LF to mark their scope (Huang 1982, Hoji 1985, Kim 1990, Aoun and Li 1993 a, among others); the other is that wh-words are interpreted in situ (Aoun and Li 1993b). Purely for notational convenience, I will call the two approaches the uniform LF wh-movement approach and the uniform wh-in-situ approach, respectively. The current research into Korean QP and wh-word scope interaction, however, leads one to believe that neither o f the two approaches can give an adequate account for the observed asymmetry and symmetry o f scope between the adjunct wh-word way (why) and the rest of wh-words in Korean. For convenience sake, I will call the former a propositional adjunct wh- word and the latter nonpropositional wh-words, as the former can typically quantify over propositions, unlike the latter. Among other observations I make in this chapter, I observe that simplex wh-questions, i.e., monoclausal wh-questions, with a modified subject QP followed by a nonpropositional wh-word, as in (1), are marginal, admitting only a functional answer, while simplex questions with the same QP followed by the propositional adjunct wh- word way (why) as in (2) are grammatical, admitting a single answer. (l)a.?* T w u salam isang-i nwukwu-lul chotayhayss-ni? two man more-NOM whom-ACC invited-QM ‘Who did more than two men invite?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 b. ?*Twu salam isang-i eti-se tampay-lul phiwess-ni? two man more-NOM where- at cigarette-ACC smoked-QM ‘W here did more than two men smoke?’ c. ?*Twu salam isang-i encey tampay-lul phiwess-ni? two man more-NOM when cigarette-ACC smoked-QM ‘W hen did more than two men smoke?’ d. ?*Twu salam isang-i ettehkey wass-ni? two man more-NOM how came-QM ‘W hat is the means x such that more than two men came by x?’ e. ?*Twu salam isang-i M ary-eytayhay ettehkey malhayss-ni? two man more-NOM M -about how talked-QM ‘W hat is the manner x such that more than two men talked about M ary in x?’ (2) Twu salam isang-i John-ul way chotayhayss-ni? two man more-NOM J-ACC why invited-QM ‘Why did more than two men invite John?’ Given the standard assumption in the literature that a single answer signals wide scope o f the wh-word over the QP (M ay 1985, among others), the scope fact as we saw in simplex wh- questions such as (1 -2) thus seems to exhibit an interesting asymmetry o f scope between the two types o f wh-words in simplex wh-questions: The propositional adjunct wh-word in (2) can take scope over the subject QP while the nonpropositional wh-words in (1) cannot. This asymmetric scope taking o f the two types o f wh-words in simplex wh-questions has an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 important implication for the scope o f the two types o f wh-words in simplex wh-questions with a universal subject QP as in (3-4). (3) a. M otun salam-i nw ukwu-lul chotayhayss-ni? every man-NOM whom-ACC invited-QM ‘W ho did everyone invite?’ b. Motun salam-i eti-se tampay-lul phiwess-ni? every man-NOM where-at cigarette-ACC smoked-QM ‘W here did everyone smoke?’ c. M otun salam-i encey tampay-lul phiwess-ni? every man-NOM when cigarette-ACC smoked-QM ‘W hen did everyone smoke?’ d. Motun salam-i ettehkey wass-ni? every man-NOM how came-QM ‘W hat is the means x such that everyone came by x?’ e. M otun salam-i M ary-eytayhay ettehkey malhayss-ni? every man-NOM M -about how talked-QM ‘W hat is the manner x such that everyone talked about Mary in x?’ (4) Motun salam-i Tom-ul w ay chotayhayss-ni? every man-NOM T-ACC why invited-QM ‘W hy did everyone invite Tom ?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 The simplex wh-questions in (3-4) with a universal QP followed by a wh-word are grammatical, and the wh-questions in (3) admit a pair list answer, a single answer and a functional answer, while the one in (4) admits a single answer. Before we further proceed, I will briefly discuss what are pair list answers, single answers and functional answers to wh- questions. I will take the English examples below in (5-6) for the expository purpose. (5) Who did everyone invite? (6) a. John invited Mary, Tom invited Bill and Jack invited Brown. b. Jane c. His mother The answers as listed in (6) are all possible answers to the wh-question above in (5). The answer in (6a) is called a pair list answer in that the answer consists o f list of inviters and invitees. The answer in (6b) is called a single answer in that a particular individual is picked for the answer. The answer in (6c) is called a functional answer in that the answer is a function, mapping individuals into their m other.1 Turning back to the Korean examples above, given the asymmetry o f scope taking between the two types o f wh-words in simplex wh-questions that we observe in (1-2), one is naturally led to raise the question: W hat is the source o f the single answer reading in (3) and in (4)? I will crucially suggest that the single answer in (3) is actually a special case o f the pair list 1 Some researchers call the answer in (6b) an individual answer (Groenendijik and Stokhof 1984, among others). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 answer, contrary to the one in (4), which derives from the wide scope o f the propositional adjunct wh-word over the QP. I also observe that the asymmetry o f scope between the two types o f wh-words in interacting with a modified subject QP in simplex wh-questions as in (1-2) virtually disappears in other environments. A complex wh-question, i.e. biclausal question, with a modified subject QP in the embedded clause followed by a wh-word as in (7-8) is a typical case o f such an environment: (7) a. (?)Ne-nun [cptwu salam isang-i nwukwu-lul chotayhayssta-ko] sayngkakha-ni? you-TOP two man more-NOM whom-ACC invited-COMP think-QM ‘Who do you think more than two men invited?’ b. (?)Ne-nun [cptwu salam isang-i eti-se tampay-lul phiwessta-ko] you-TOP two man more-NOM where-at cigarette-ACC smoked-COMP sayngkakha-ni? think-QM ‘W hat is the place x such that you think more than two men smoked at x?’ c. (?)Ne-nun [cptwu salam isang-i encey tampay-lul phiwessta-ko] you-TOP two man more-NOM when cigarette-ACC smoked-COMP sayngkakha-ni? think-QM ‘W hat is the time x such that you think more than two men smoked at x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 d. (?)Ne-nun [C p twu salam isang-i ettehkey wassta-ko] sayngkakha-ni? you-TOP two man more-NOM how came-COMP think-QM ‘W hat is the means x such that you think more than two men came by x?’ e. (?)Ne-nun [C p twu salam isang-i M ary-eytayhay ettehkey malhayssta-ko] you-TOP two man more-NOM M -about how talked-COMP sayngkakha-ni? think-QM ‘W hat is the manner x such that you think more than two men talked about Mary in x?’ (8) Ne-nun [cp twu salam isang-i Tom-ul way chotayhayssta-ko] sayngkakha-ni? you-TOP two man more-NOM T-ACC why invited-COMP think-QM ‘W hat is the reason x such that you think more than two men invited Tom for x?’ The complex wh-questions in (7-8) with a modified subject QP in the embedded clause are all grammatical irrespective o f what type o f wh-word follows the QP. The questions in (7) can admit a single answer and a functional answer, and the question in (8) admits a single answer. Given that a single answer signals the wide scope o f the wh-word over the QP, the scope facts in (7-8) thus suggests that both types o f wh-words can take scope over the QP, hence their wide scopal behavior over the QP in this context is symmetrical. The asymmetry and symmetry o f scope between way (why) and the rest o f wh-words is further confirmed in wh-questions with a negative polarity item (NPI, henceforth), as in (9- 12). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 (9) a. ?*Amwuto nwukwu-lul chotayhaci an hayss-ni? anyone whom-ACC invite NOT did-QM ‘Who did no one invite?’ b. ?*Amwuto eti-se tampay-lul pwiwuci an hayss-ni? anyone where-at cigarette-ACC smoke NOT did-QM ‘W here did no one smoke?’ c. ?*Amwuto encey tampay-lul pwiwuci an hayss-ni? anyone when cigarette-ACC smoke NOT did-QM ‘W hen did no one smoke?’ d. ?*Amwuto ettehkey oci an hayss-ni? anyone how come NOT did-QM ‘W hat is the means x such that no one came by x?’ e. ?*Amwuto M ary-eytayhay ettehkey malhaci an hayss-ni? anyone M -about how talk NOT did-QM ‘W hat is the manner x such that no one talked about M ary in x?’ (lO )A m w uto way Tom-ul chotayhaci an hayss-ni? anyone why T-ACC invite NOT did-QM ‘Why did no one invite Tom ?’ Simplex wh-questions with an NPI in the subject position followed by a nonpropositional wh-word, as in (9), are marginal in grammaticality, allowing a functional answer only, while simplex wh-questions with the same NPI followed by a prepositional adjunct wh-word way Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 (why) in (10) are grammatical, admitting a single answer. This means that the nonpropositional wh-word cannot take scope over the NPI in (9), while the prepositional adjunct wh-word in (10) can, attesting again asymmetry o f scope between the two types o f wh-words in simplex wh-questions. Meanwhile, the complex wh-questions below in (11 -12) with an NPI in the embedded subject position are all grammatical irrespective o f what type o f wh-word follows the NPI. The questions in (11) can admit a single answer and a functional answer, and the question in (12) admits a single answer. Given that a single answer signals the wide scope o f the wh-word over the QP, the scope facts in (11-12) again suggests that both types o f wh-words can take scope over the NPI, hence their wide scopal behavior over the NPI in this context is symmetrical. (11) a. (?)Ne-nun [cpamwuto nwukwu-lul chotayhaci an hayssta-ko] sayngkakha-ni? you-TOP anyone whom-ACC invite NOT did-COM P think-QM ‘Who do you think no one invited?’ b. (?)Ne-nun [ c p amwuto eti-se tampay-lul phiwuci an hayssta-ko] you-TOP anyone where-at cigarette-ACC smoke NOT did-COMP sayngkakha-ni? think-QM ‘W hat is the place x such that you think no one smoked at x?’ c. (?)Ne-nun [ c p amwuto encey tampay-lul phiwuci an hayssta-ko] you-TOP anyone when cigarette-ACC smoke NOT did-COMP sayngkakha-ni? think-QM ‘W hat is the time x such that you think no one smoked at x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 d. (?)?Ne-nun [C p amwuto ettehkey oci an hayssta-ko] sayngkakha-ni? you-TOP anyone how come NOT did-COMP think-QM ‘W hat is the means x such that you think no one came by x?’ e. (?)Ne-nun [C P amwuto M ary-eytayhay ettehkey malhaci an hayssta-ko] you-TOP anyone M -about how talk NOT did-COM P sayngkakha-ni? think-QM ‘What is the manner x such that you think no one talked about M ary in x?’ (12) Ne-nun [ c p amwuto Tom-ul way chotayhaci an hayssta-ko] sayngkakha-ni? you-TOP anyone T-ACC why invite NOT did-COM P think-QM ‘W hat is the reason x such that you think no one invited Tom for x?’ As one may notice, the grammaticality o f a given wh-question is closely tied to its answerability. Given the marginality o f the sentences in (1) and (9) with only a functional answer, which is not as informative as either a pair list answer or a single answer, one may suggest that answerability should count as an evaluation measure for the acceptability o f a wh-question as w e ll.2 1 will thus make the proposal in (13): 2 Paul Hagstrom (personal communication) points out that the functional answer may be more informative in that it gives one a way to find out the answer for any given first member of a pair while the pair list answer gives one just the pairings. Although it is a trickey issue, functional answer is not as informative as a single answer and a pir list answer in that it does not specifiy individuals in question. Thus the hearer does not have to be fully informed of the relevant individuals’s names when he answers a question by giving a functional answer ‘his mother’ for example, while he should be fully informed of those individual’s names, when giving a pair list answer to the same question. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 (13) W h-questions should be answerable with at least a pair list answer or a single answer. Thus wh-questions, which cannot be answered in at least one of the two ways are marginal in grammaticality under normal situation.3 Given the relative insignificance o f the availability o f a functional answer to the acceptability o f a wh-question, as shown in (1) and (9), I will primarily focus on pair list answers and single answers throughout this chapter, although I will discuss the grammatical status o f wh-questions which admit only a functional answer in section 6. One may dismiss the proposal in (13) as rather absurd, considering that wh- questions in English normally admit at least one o f the two readings as shown below in (14). (14) a. Who did everyone invite? b. W ho did more than two men invite? The question in (14a) admits pair list answer, single answer and functional answer, while the one in (14b) admits a single answer and a functional answer. The sentences in (14) are of course all grammatical. However, when it comes to wh-questions in Korean where wh- questions with a modified subject QP as in (1) or an NPI subject as in (9) admit only a functional answer and are marginal, the proposal in (13) merits a serious consideration. Then an important question immediately arises: Why do the two types o f wh-words in Korean, in other words, propositional adjunct wh-word way (why) and nonpropositional wh- words exhibit this important asymmetry in means o f scope taking in wh-questions? To 3 The grammaticality of a sentence with only a functional answer will improve if the functional answer is the only answer expected in a particular context. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 answer this important question and thus to lay the groundwork for the discussion o f scope in wh-questions later on, I will turn in the next section to an investigation o f the nature o f wh- words in Korean. 2. Wh-words as Indefinites The wh-words in Korean are listed in (15) (also see Suh 1990:27, Kim 1990:212, Chung 1995:140). a. nwukwu ‘w ho’ b. mwues ‘w hat’ c. enu N ‘w hich N ’ d. eti ‘w here’ e. encey ‘w hen’ f. elma ‘how m uch’ g. ettehkey ‘how ’ h. way ‘w hy’ The nonpropositional wh-words in (15a-g), as opposed to the propositional adjunct wh-word way (why) in (15h), can typically receive a different interpretation depending on the context in which they appear. In what follows, I will illustrate the point by considering these contexts. Consider questions in (16-17), which are typical contexts licensing an NPI. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (16) a. John-i nwukwu-lul mannass-ni? J-NOM whom-ACC met-QM ‘Who did John m eet?’ ‘Did John meet anyone?’ b. John-i enu haksayng-ul chotayhayss-ni? J-NOM w hich student-ACC invited-QM ‘W hich student did John invite?’ ‘Did John invite any student?’ c. John-i eti-lo oychwulhayss-ni? J-NOM where-to went out-QM ‘W here did John go out?’ ‘Did John go out anyw here?’ d. John-i encey M ary-ul mannass-ni? J-NOM when M -ACC met-QM ‘W hen did John m eet M ary?’ ‘Did John meet M ary any tim e?’ e. John-i cheycwung-i yocum elma nuless-ni? J-NOM weight-NOM lately how much gained-QM ‘How much weight has John gained lately?’ ‘Has John gained any weight lately?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 f. John-i ettehkey LA-ey kass-ni? J-NOM how LA-to went-QM ‘How did John go to LA ?’ ‘Did John go to LA by any m eans?’ g. John-i M ary-etayhay ettehkey malhayss-ni? J-NOM M -about how said-QM ‘How did John say about M ary?’ ‘Did John talk in any manner about M ary?’ (17) John-i way wass-ni? J-NOM why came-QM ‘Why did John com e?’ # ‘Did John come for any reason?’ As shown by the questions in (16-17), Korean is a head final language with in situ wh- words. In modern standard Korean, the question morpheme ni in (16) can be construed either as a yes-no question morpheme or as a wh-question morpheme: Prosody is crucial for disambiguating. Thus, rising intonation at the end o f the sentence on the question morpheme gives rise to a yes-no question, while falling intonation on it yields a wh-question. One may thus posit two different question morphemes in terms o f [Q, WH] feature specification such that question morpheme with [+Q, +WH] is a wh-question morpheme while the one with [+Q, -WH] is a yes-no question morpheme in (16).4 As indicated by the glosses in (16), 4 In some Kyengsang dialects, especially those spoken in the areas around the cities of Masan and Kimhay the question morpheme for the wh-question and yes-no question remain morphologically Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 nonpropositional wh-words quantifying over individuals (16a-d), measure (16e), means (16f) and manner (16g) can have a negative polarity item (NPI,henceforth) reading, in addition to a wh-interrogative reading, contrary to the propositional adjunct wh-word in (17), which admits a wh-interrogative reading only. In the context o f negation, another typical licensing context for NPIs, nonpropositional wh- words quantifying over individuals (18a-d), measure (18e), means (18f) and manner (18g) have an NPI reading, again in contrast to the propositional adjunct wh-word, which cannot, as shown by the ungrammaticality o f (19). (18) a. John-i ecey nwukwu-lul manaci an hayssta. J-NOM yesterday who-ACC meet NOT did ‘John did not m eet anyone yesterday.’ b. ?John-i ecey enu chinkwu-lul mannaci an hayssta. J-NOM yesterday which friend-ACC meet NOT did ‘John did not meet any friend yesterday.’ c. John-i ecey eti-lo oychwulhaci an hayssta. J-NOM yesterday where-to go out NOT did John did not go out anywhere yesterday.’ d. Na-nun nayil-cwung encey Mary-lul mannaci mos hakessta. I-TOP tomorrow-during when M-ACC meet NOT will I will not be able to m eet M ary any time tomorrow. ’ distinct. Ka and na stand for the yes-no question morpheme while ko and no stand for a wh-question morpheme (see Suh 1987: 2ff, among others). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 e. John-i cheycwung-i elma an nulessta. John-NOM weight-NOM how m uch NOT gained ‘John did not gain any significant weight.’ f. John-i swuep-ey tocehi ettehkey cesikan-ey ol swu ep-essta. J-NOM class-to at all how time-in come can NOT-was ‘John could not come to class on time by any m eans.’ g. John-i tocehi ettehkey ku os-ul ipul swu ep-essta. J-NOM at all how the clothes-ACC wear can NOT-was ‘John could not wear the clothes in any w ay.’ (19) * John-i ecey Mary-lul way manaci an hayssta. J-NOM yesterday M-ACC why m eet NOT did ‘John did not m eet M ary yesterday for any reason.’ W hen it comes to indicatives that do not license NPIs, as in (20-21), nonpropositional wh- words quantifying over individuals (20a-d), measure (20e), means (20f) and manner (20g) are interpreted existentially, while the propositional adjunct wh-word cannot be construed in an analogous way as shown by the ungrammaticality o f (21). (20) a. John-i nwukwu-lul mannassta. J-NOM whom-ACC met ‘John met som eone.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 b. John-i enu chinkwu-lul manassta. J-NOM which friend-ACC met ‘John met a certain friend.’ c. John-i encey LA-ey kal yecengita. J-NOM when LA-to go plan ‘John plans to go to LA sometime.’ d. John-i cinan cwumal-ey eti-lo J-NOM last weekend-over where ‘John went somewhere last weekend.’ e. John-i cheycwung-i elma J-NOM weight-NOM how much ‘John gained some weight. ’ f. John-i ettehkey swipkey ku elyewun mwuncey-lul J-NOM how easily that difficult problem-ACC ‘John solved that difficult problem easily somehow.’ g. ?John-i M ary-eytayhay ettehkey malhayssta. J-NOM M -about how said ‘John talked about M ary in some w ay.’ (21) * John-i w ay wassta. J-NOM why came ‘John came for some reason.’ kassta. went nulessta. gained phwulessta. solved Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 Based on the initial paradigm in (16-21), which shows the interpretation o f nonpropositional wh-words, contrary to that o f the propositional adjunct wh-word, is determined by the particular context in which they appear, one may initially suggest that nonpropositional wh- words are indefinites in the sense o f Lewis (1975) and Heim (1982), with the question morpheme w ith [+Q, -WH] and negation being responsible for the NPI reading in (16) and (18), respectively, and existential closure (Heim 1982) for the existential reading in (20). I will turn to the interpretation o f wh-words as wh-interrogatives in section 3. Before further examining the status o f wh-words as indefinites in Korean, I will briefly review the discussion o f indefinites in Heim (1982). Heim (1982), following Lewis (1975), suggests that indefinites do not have an inherent quantificational force but serves as variables in the logical representation, bound by another quantificational element in a larger domain, enabling them to take scope out o f syntactic islands. The following examples in (22) illustrate these two aspects o f the behavior o f indefinites: (22) a. If a man owns a donkey he always beats it. b. Sometimes, if a cat falls from the fifth floor, it survives. (Heim 1982: 123) The quantificational force o f the indefinites a man, a donkey and a cat in (22) is determined by the adverbial quantifiers always and sometimes that can bind more than one variable, given the interpretation rule for the adverbs o f quantification such as always and sometimes as in (23a) and (23b), respectively.5 5 According to Lewis (1975), the unselective binding as in (23) is distinguished from the selective binding, which is defined as below. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 (23) a.VOis true iff O is true under every admissible assignment o f values to all variables free in O. b. 3d>is true iff O is true under some admissible assignment o f values to all variables free in O. (Lewis 1975: 7) Moreover, as shown by the informal logical notations in (24), the indefinites can take scope out o f the syntactic islands in which they are embedded - a fact which is captured in the Heim-Lewis approach in which adverbs o f quantification and conditionals are mapped onto a quantification structure for interpretation.6 (24) a. If a man owns a donkey he always beats it. ‘for every x ,y, x an individual, y a donkey if x owns y, x beats y .’ b. Sometimes, if a cat falls from the fifth floor, it survives. ‘for some x, x a cat if x falls from the fifth floor, x survives.’ W ith the two important characteristics of indefinites in mind, namely, lack o f inherent quantificational force and the ability to scope out o f syntactic islands, let us turn to the following to (also) construction, illustrated below in (25-26) and conditionals with an a.Vxd> is true under any admissible assignment f of values to all variables free in O except x, iff for every admissible value of x, < t> is true under the assignment of that value to x together with the assignment f of values to the other variables free in d> . b. 3x4> is true under any admissible assignment f of values to all variables free in < t > except x, iff for some admissible value of x, < t> is true under the assignment of that value to x together with the assignment / of values to the other variables free in < f > (Lewis 1975: 6). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 adverbial quantifier, illustrated below in (27-28), which provide an ideal test for the status of wh-words as indefinites: (25) a. Na-nun [Np [cp nwu-ka ssun] chayk]-to ilkkocipta. I-TOP who-NOM wrote book -TO want to read ‘for every x, y, x a person, y a book x wrote, I want to read y.’ b. Na-nun [N P [cp enu enehakca-ka ssun] chayk]-to ilkkocipta. 1-TOP which linguist-NOM wrote book -TO want to read ‘for every x, y, x a linguist, y a book x wrote, I want to read y .’ c. Na-nun [N P [cP John-i eti-se ssun] chayk]-to ilkkocipta. I-TOP J-NOM where-at wrote book -TO want to read ‘for every x, y, x a place, y a book John wrote at x, I want to read y.’ d. Na-nun [N P [cp John-i encey ssun] chayk]-to ilkkocipta. I-TOP J-NOM when wrote book -TO want to read ‘for every x, y, x a time, y a book John wrote at x, I want to read y.’ e. Na-nun [N P [C P kakyek-i elma nakanun] chayk]-to sakosipta. I-TOP price-NOM how much cost book-TO want to buy ‘for every x, y, x a price, y a book which costs x, I want to buy y .’ f- [np [cp John-i ettehkey kulin] kulim]-to pissata. J-NOM how painted painting -TO is expensive ‘for every x, y, x a means, y a painting John painted by x, y is expensive.’ 6 See Geach (1962) for the original discussion of a donkey sentence. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 g- [np [cp M ary-ka ettehkey ipnun] os]-to cal ewullye pointa. M-NOM how wear clothes-TO well seems to match ‘for every x, y, x a manner, y a clothes M ary wears in x, y suits her w ell.’ (26) * [N p [C p John-i way ssun] chayk]-to chwulphantoyessta. J-NOM why wrote book-TO was published ‘for every x, y, x a reason, y a book John wrote for x, y was published.’ (27) a. [cp Nwuj-ka o-myen] ( pro;) kkok wuli-lul pangmwunhanta. who-NOM com e-if always us-ACC visit ‘for every x, x an individual, if x comes, x visits us.’ b. [cp LA-eyse enu chinkwuj-ka o-myen] (pro*) kkok wuli-lul pangmwunhanta. LA-from which friend-NOM com e-if always us-ACC visit ‘for every x, x a friend, if x comes from LA, x visits us.’ c. [cpJohnj-i eti-lo oychwulha-myen] (proO kkok nusskey tolaonta. J-NOM where-to go out-if always late returns ‘for every x, x a place, if John goes out to x, he returns home late (from x).’ d. [C p Johnj -i encey o-myen] (pro,) kkok wuli-ul pangmwunhanta. J-NOM when com e-if always us-ACC visit ‘for every x, x a time, if John comes at x, he visits us (at x).’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 e. [C p Maryj -nun cheycwung-i cakij-uy cengsang cheycwung-pota M -TOP weight-Nom her-POSS normal weight-than elma te naka-myen ] (pro;) kkok diet-lul hanta. how much more w eigh-if always diet-ACC do ‘for every x, x a weight, if M ary weighs x more than normal weight, she goes on a diet (for x).’ f. [C p Johnj-i ettehkey wuli cipey o-myen] J-NOM how our house-to com e-if (proj) kkok senmwul-ul kacikoonta. always gift-ACC bring ‘for every x, x a means, if John comes to our place by x, he brings a gift (by x).’ g. [ c p Nay-ka ettehkey John-eytayhay malha-myen] I-NOM how J-about talk-if M ary-nun kkok hwalul naykonhanta. M-TOP always get angry ‘for every x, x a manner, if I talk about John in x, Mary gets angry (with x).’ (28) * [ c p John,-i way o-myen] (proi) kkok wuli-lul pangmwunhanta. J-NOM why com e-if always us-ACC visit ‘for every x, x a reason, if John comes for x, he visits us (for x).’ Note that the interpretation o f the nonpropositional wh-words in (25) as universal quantifiers is determined by to (also) that is attached to the head NP o f the relative clause and carries a universal quantificational force. Likewise, their interpretation in (27) as universal Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 quantifiers is determined by the adverbial quantifier kkok (always) in the main clause, which conveys a universal quantificational force. The informal logical notations in (25) and (27) also indicate that these wh-words indeed can take scope out of the syntactic islands, complex noun phrases and adjunct clauses, respectively. Thus one may suggest that the nonpropositional wh-words in (25) and (27) are indefinites unselectively bound by to (also) and kkok (always), respectively, assuming of course the movement o f the adverbial quantifier into a position from which it can bind them at LF. 7 Meanwhile, the propositional adjunct wh-word cannot be construed in an analogous way in (26) and (28), which are simply ungram m atical.8 Interestingly, when to (also) is replaced with nka, which carries existential quantificational force, nonpropositional wh-words are construed as existentially below in (29), as opposed to the propositional adjunct wh-word that cannot be interpreted in an analogous way as shown by the ungrammaticality o f (30) b elo w .9 (29) a. [N P [cp Nwu-ka ssun] chayk]-nka-ka chwulphantoyessta. who-NOM wrote book-NKA-NOM was published ‘for some x, y, x a person, y a book x wrote, y was published.’ 7 The anaphoric expressions that are not overtly manifested in the main clause in the examples in (27c- g) are ku kos-ulopwuthe (from that place), ku lemyen (then), ku ttaymwuney (because of that), kulentaylo ( by that means) and ku cemeytayhay (in that manner), respectively. 8 Incidentally, the nonpropositional wh-words in (25) and (27) seem to have an NPI reading too, which is not unexpected, given that these wh-words appear in the relative clause (25) and the //^clause (27) that is the restriction of to (also) and the adverbial quantifier kkok (always), respectively, which both carry universal quantificational force, and that the restriction of a universal quantifier is a context that typically licenses NPIs (see Ladusaw 1980 among others). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 b. [np [cp Enu enehakca-ka ssun] chayk]-nka-ka chwulphantoyessta. which linguist-NOM wrote book-NKA-NOM was published ‘for some x, y, x a linguist, y a book x wrote, y was published.’ c. [np [cp John-i eti-ese ssun] chayk]-nka-ka chwulphantoyessta. J-NOM where-at wrote book -NKA-NOM was published ‘for some x, y, x a place, y a book John wrote at x, y was published.’ d. [np [cp John-i encey ssun] chayk]-nka-ka chwulphantoyessta. J-NOM when wrote book-NKA-NOM was published ‘for some x, y, x a time, y a book John wrote at x, y was published.’ c. [N P [ c p Cheycwung-i elma nakanun] kolay]-nka-ka capheyssta. weight-NOM how much amount whale-NKA-NOM was caught ‘for some x, y, x a weight, y a whale that weighs x, y was caught.’ f-?[ n p [ c p John-i ettehkey ponen] phyenci]-nka-ka pwunsiltoyessta. J-NOM how sent mail-NKA-NOM was lost ‘for some x, y, x a means, y a mail John sent by x, y was lost.’ g- [np [cp M ary-ka ettehkey taniten] hwoysa]-nka-ka M-NOM how used to work company-NKA-NOM pwutokanassta. was bankrupt ‘for some x ,y ,x a manner, y a company M ary used to work for in x, y was bankrupt.’ 9 A high front vowel [i] should be inserted before nka as in (29-30) when the preceding word ends Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 (30) * [ np [cp John-i w ay ssun] chayk]-nka-ka chwulphantoyessta. J-NOM why wrote book-NKA-NOM was published ‘for some x, y, x a reason, y a book John wrote for x, y was published.’ Likewise, when the adverbial quantifier kkok (always) above in (27-28) is replaced with kakkum (sometimes) below as in (31-32), nonpropositional wh-words are construed as existentially, in contrast to the propositional adjunct wh-word, which cannot be interpreted in an analogous way as shown by the ungrammaticality o f (32). (31) a. [cp Nwuj-ka o-myen] (proj) kakkum wuli-lul pangmwunhanta. who-NOM com e-if sometimes us-ACC visit ‘for some x, x an individual, if x comes, x visits us.’ b. [C p LA-eyse enu chinkw urka o-myen ] (proj) kakkum wuli-lul pangmunhanta. LA-from which friend-NOM com e-if sometimes us-ACC visit ‘for some x, x a friend, if x comes from LA, x visits us.’ c. [ c p John-i encey o-myen ] (pro,) kakkum wuli-lul pangmwunhanta. J-NOM when com e-if sometimes us-ACC visit ‘for some x, x a time ,if John comes at x, John visits us (at x).’ d. [ c p John,-i eti-lo oychwulha-myen ] (proO kakkum nusskey tolaonta. J-NOM where-to go out-if sometimes late returns ‘for some x, x a place, if John goes out to x, he returns home late (from x).’ with a consonant. I will not notate it for simplicity’s sake. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 e. [C p Maryi-nun cheycwung-i cakij-uy cengsang chaycwung-pota M-TOP weight-NOM her-POSS normal weight-than elma te naka-myen] (proj) kakkum diet-lul hanta. how much more w eigh-if sometimes diet-ACC do ‘for some x, x a weight, if M ary weighs x more than normal weight, she goes on a diet (for x).’ f. [cpJohnj-i ettehkey wuli cip-ey o-myen] (proj) kakkum senmwul-ul J-NOM how our house-to com e-if sometimes gift-ACC kacikoonta. bring ‘for some x, x a means if John comes to our place by x, he brings a gift (by x).’ g. [cp Nay-ka ettehkey John-eytayhay malha-myen] I-NOM how J-about talk-if M ary-nun kakkum hwalul naykonhanta. M -TOP sometimes get angry ‘for some x, x a manner, if I talk about John in x, Mary gets angry (with x).’ (32) * [cpJohnj-i way o-m yen] (pro;) kakkum wuli-lul pangmwunhanta. J-NOM why com e-if sometimes us-ACC visit ‘for some x, x a reason, if John comes for x, he visits us (for x).’ This state o f affairs again suggests that contrary to the propositional adjunct wh-word, the nonpropositional wh-words are indefinites, with their quantificational force being Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 determined by nka in (29) and the adverbial quantifier kakkum (sometimes) in (31), which unselectively bind them at LF, respectively. Moreover, the informal logical notations in (29) and (31) clearly indicate that nonpropositional wh-words can take scope out o f syntactic islands, complex noun pharses and adjunct clauses, respectively. The data in (25-32) thus enable us to reach the conclusion that nonpropositional wh-words are indefinites, as opposed to the propositional adjunct wh-word way (why) in that nonpropositional wh-words do not have their own quantificational force with their quantificational force determined by an external quantificational element, and that they can take scope out o f syntactic islands. Before closing this section, it should be noted that our proposal for the different status o f wh- words in Korean as indefinites along the dimension o f nonpropositional wh-words on the one hand and the propositional adjunct wh-word on the other is reminiscent o f the proposal o f Tsai (1994), according to which nominal wh-words, but not nonnominal wh-words, are indefinites in Chinese. The dichotomy in Tsai (1994), when extended to Korean, is flawed, however, since the adjunct wh-word ettehkey (how) with a manner construal, which should be a nonnominal wh-word in his system still behaves as an indefinite as we saw thus far in this section, which indeed suggest that the relevant dimention should be nonpropositional wh-words vs. propositional adjunct w h-w ords.1 0 1 0 The eccentric behavior of the propositional adjunct wh-word corresponding to why in English is wellknown crosslinguistically. Please refer to Bromberger (1992) for the discussion of why in English, among others. Jean-Roger vergnaud (personal communication) also notes that French wh-word corresponding to why in English behaves differently from the other wh-words, when it comes to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 3. The Interpretation of Wh-Questions Given the difference between the propositional adjunct wh-word way (why) and the nonpropositional wh-words regarding their status as an indefinite as shown in section 2, one may wonder what implication this difference may have for the interpretation o f wh-words as wh-interrogatives. Consider the following wh-questions in English in (33) and Korean in (16a) and (17), repeated in (34a) and (34b), respectively: (33) a. Who did John meet? b. W hy did John come? (34) a. John-i nwukwu-lul mannass-ni? J-NOM whom-ACC met-QM ‘W ho did John m eet?’ ‘Did John meet anyone?’ b. John-i way wass-ni? J-NOM why came-QM ‘W hy did John com e?’ #Did John come for any reason?’ As shown in section 2, the sentence in (34a) in contrast to the one in (34b), which is construed as a wh-question only, can be construed as either a yes-no question or a wh- stylistic inversion. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 question. In what follows, I will focus on wh-questions only. The most obvious difference between English and Korean when it comes to the interpretation o f wh-questions like (33-34) is that wh-words in English are preposed into sentence initial position, this surface position marking their scope, while they remain in situ in Korean, with the question morpheme marking their scope. Following the intuitions in the literature (Chomsky 1964:38, Katz and Postal 1964:93, Klima 1964:253, Kuroda 1969: 266ff, and Vergnaud and Zubizarreta 2002, among others), according to which wh-words in English are viewed as a combination o f a wh-operator and indefinite some (one, thing...), I suggest that in (34a) with the [+Q,+WH] question morpheme ni, ni and the wh-word correspond to the wh-operator and the indefinite someone, respectively. 1 1 Hereafter, the ‘Q M ’ should be understood as referring to the QM with the feature specification [+Q,+W H], unless specified otherwise. I thus suggest that QM as a wh- operator unselectively binds and thus marks the scope o f the nonpropositional wh-word at LF (cf. Baker 1970), adopting the minimalist tenet that LF is the only level for semantic interpretation (Chomsky 1995). When it comes to the propositional adjunct wh-word in (34b), I suggest that it should undergo movement into Spec o f CP at LF via spec head agreement with the QM, driven by the need for proper interpretation by forming an operator variable chain, since it is not an indefinite. Thus the QM in (34b) in contrast to the one in (34a) does not behave as a wh- 1 1 For the opposite view, see Hagstrom (1998) where the indefinite is viewed as a combination of the wh-word and something else. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 operator but rather just as a position holder triggering movement o f the propositional adjunct wh-word to check off its wh-feature and thereby mark its scope at LF. To summarize, when it comes to the interpretation o f wh-words in Korean as wh- interrogatives, the grammar has two different mechanisms: One is that nonpropositional wh- words are unselectively bound by the QM which marks their scope at LF and the other is that the propositional adjunct wh-word way (why) should undergo movement at LF into Spec of CP thereby marking its scope under spec head agreement with the QM by forming an operator variable chain to enable proper interpretation. W ith the different scope taking mechanisms o f the two types o f wh-words in mind, let us now turn to QP and wh-word scope interaction in Korean wh-questions. 4. QP and Wh-word Scope Interaction Departing from the standard practice in Korean literature where standard ‘first order predicate logic Q Ps’ such as a universal QP and an existential QP are mainly discussed (Joo 1989, Kim 1990, Suh 1990), I will also include QPs such as most NP and modified QPs whose meaning cannot be captured in first order logic, which will reveal the asymmetric means o f scope taking associated with the two types o f wh-words: The unselective binding o f the nonpropositional wh-words by the QM, which marks their scope at LF and the movement o f the propositional adjunct wh-word into Spec o f CP at LF to mark its scope. When it comes to the choice o f the universal quantifier corresponding to English everyone, motun salam and nwukwuna seem to be in competition in the literature. The former is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 adopted in Suh (1990), while the latter is adopted in Joo (1989) and Kim (1990). I will opt for the former, however, since the latter has a free choice any interpretation which is not available for its English counterpart as noted in Suh (1990). As for modified QPs, I will use twu salam isang (more than two men), se salam miman (less than three men) and ttak twu salam (exactly two men), which are monotone-increasing, monotone-decreasing and non monotonic QPs, respectively. 1 2 1 3 It should be noted that when it comes to QP and wh-word scope interaction, I will limit my discussion to non-discourse-linked wh-words. Thus enu NP (which NP) in the paradigm in (15) in section 2 will not be included in the discussion o f QP and wh-scope interaction. The measure phrase elma (how much) will not be included either, since when it comes to modified QP and wh-word scope interaction as shown in (35), the ungrammaticality o f the sentence may be independent o f the QP and wh-word scope interaction, given the semantics o f modified QPs. (35) *Twu salam isang-i chaycwung-i elm a naka-ni? two man more-NOM weight-NOM how much weigh-QM ‘How much do more than two men w eigh?’ 1 2 It should be noted that the semantics of twu salam isang is different from more than two men in English, since isang (more) means ‘the same or lager in number or amount’. However, I will translate it as meaning more than two men, since however one interprets it, the QP is monotone-increasing, which is the relevant property for our purpose here. 1 3 See the discussion of the various QP types within the framework of generalized quantifier theory in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 Even under the ideal situation in which every individual in the discourse weighs the same amount, given the semantics o f the monotone-increasing quantifier, twu salam isang (more than two men), the hearer o f the question in (35), will not be able to come up with any informative answer. The wh-words I will use for QP and wh-word scope interaction in the subsequent sections thus will include all o f the wh-words in (15) in section 2 except for enu NP (which NP) and measure phrase elma (how much). 4.1. Asymmetry of Scope W ith the asymmetric scope taking o f the two types o f wh-words in mind, i.e., unselective binding o f the nonpropositional wh-words by the QM that marks their scope, vs. the LF movement o f the propositional adjunct wh-word way (why) into Spec o f CP, driven by the need for proper interpretation, let us turn to QP and wh-word scope interaction in Korean simplex wh-questions first. By ‘simplex wh-questions’, I mean mono-clausal wh-questions. 4.1.1. Simplex Wh-question Let us suppose throughout this chapter a context in which the speaker and the hearer are chatting over John, Mary, Jack, Peter, Barry, Jim and Joseph, who were at the end of Barwise and Cooper (1981), Szabolcsi (1997a), and Heim and Kratzer (1998), among others. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 semester party to which each o f them was supposed to invite someone. Let us further suppose that the wh-word is not discourse-linked (see Pesetsky 2000: 16 for the relevant definition). The speaker could not join the party for some reason and thus asks the questions in (36-38) to the hearer who has come right from the party. (36) a. ?*Twu salam isang-i nwukwu-lul chotayhayss-ni? two man more-NOM whom-ACC invited-QM ‘Who did more than two men invite?’ b. ?*Twu salam isang-i eti-se tampay-lul phiwess-ni? two man more-NOM where-at cigarette-ACC smoked-QM ‘W here did more than two men smoke?’ c. ?*Twu salam isang-i encey tampay-lul phiwess-ni? two m an more-NOM when cigarette-ACC smoked-QM ‘W hen did more than two men smoke?’ d. ?*Twu salam isang-i ettehkey wass-ni? two man more-NOM how came-QM ‘W hat is the means x such that more than two m en came by x?’ e. ?*Twu salam isang-i M ary-eytayhay ettehkey malhayss-ni? two man more-NOM M -about how talk-QM ‘W hat is the manner x such that more than two men talked about Mary in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 (37) a. ?*Se salam miman-i nwukwu-lul chotayhayss-ni? three man less-NOM whom-ACC invited-QM ‘Who did less than three men invite?’ b. ?*Se salam miman-i eti-se tampay-lul phiwess-ni? three man less -NOM where-at cigarette-ACC smoked-QM ‘W here did less than three men smoke?’ c. ?*Se salam miman-i encey tampay-lul phiwess-ni? three man less -NOM when cigarette-ACC smoked-QM ‘W hen did less than three men smoke?’ d. ?*Se salam miman-i ettehkey wass-ni? three man less-NOM how came-QM ‘W hat is the means x such that less than three men came by x ?’ e. ?*Se salam miman-i M ary-eytayhay ettehkey malhayss-ni? three man less-NOM M -about how talked-QM ‘W hat is the manner x such that less than three men talked about M ary in x?’ (38) a. ?*Ttak twu salam-i nwukwu-lul chotayhayss-ni? exactly two man-NOM whom-ACC invited-QM ‘W ho did exactly two men invite?’ b. ?*Ttak twu salam-i eti-se tampay-lul phiwess-ni? exactly two man-NOM where-at cigarette-ACC smoked-QM ‘W here did exactly two men smoke?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 c. ?*Ttak twu salam-i encey tampay-lul phiwess-ni? exactly two man-NOM when cigarette-ACC smoked-QM ‘W hen did exactly two men smoke?’ d. ?*Ttak twu salam-i ettehkey wass-ni? exactly two man-NOM how came-QM ‘W hat is the means x such that exactly two men came by x?’ e. ?*Ttak twu salam-i M ary-eytayhay ettehkey malhayss-ni? exactly two man-NOM M -about how talk-QM ‘W hat is the manner x such that exactly two men talked about M ary in x?’ The sentences in (36-38) are marginal in grammaticality, admitting only a functional answer. One may wonder whether the marginality o f these sentences is independent o f the fact that a modified subject QP is followed by a nonpropositional wh-word. The grammaticality o f the sentences below in (39) identical to the sentences in (36) except that the wh-words have been replaced with a non-wh-expression suggests that the marginality o f (36-38) is indeed due to the fact that a modified subject QP is followed by a nonpropositional wh-w ord.1 4 (39) a. Twu salam isang-i Sue-lul chotayhayss-ni? two m an more-NOM S-ACC invited-QM ‘Did more than two men invite Sue?’ 1 4 It should be noted that with an NPI reading of the nonpropositional wh-word, the sentences with a modified subject QP in (36-38) are in general grammatical as a yes-no question. Thus the sentences in (36abc and e), for example, are grammatical as yes-no questions, while the sentence in (36d), on the other hand, still sounds marginal even as yes-no question, maybe due to the lack of an appropriate context. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 b. Twu salam isang-i pakey-se tampay-lul phiwess-ni? two man more-NOM outside-at cigarette-ACC smoked-QM ‘Did more than two men smoke outside?’ c. Twu salam isang-i siksa-hwuey tampay-lul phiwess-ni? two man more-NOM meal-after cigarette-ACC smoked-QM ‘Did more than two men smoke after the m eal?’ d. Twu salam isang-i cha-lo wass-ni? two man more-NOM car-by came-QM ‘Did more than two men come by car?’ e. Twu salam isang-i M ary-eytayhay coh-key malhayss-ni? two man more-NOM M -about nicely talked-QM ‘Did more than two men talk nicely about M ary?’ Interestingly, simplex wh-questions with a modified subject QP followed by the propositional adjunct wh-word way (why) are grammatical, admitting a single answer, as shown in (40). chotayhayss-ni? invited-QM (40) a. Twu salam isang-i Tom-ul way two man more-NOM T-ACC why ‘W hy did more than two men invite Tom ?’ b. ?Se salam miman-i Tom-ul way chotayhayss-ni? three man less-NOM T-ACC why invited-QM ‘W hy did less than three men invite Tom ?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 c. Ttak twu salam-i Tom-ul w ay chotayhayss-ni? exactly two man-NOM T-ACC why invited-QM ‘W hy did exactly two men invite Tom ?’ Thus the two types o f wh-words exhibit an interesting asymmetry o f scope in simplex wh- questions regarding the wide scope o f the wh-word over the QP in simplex wh-questions as shown in (36-38) vs. (40): Simplex wh-questions with a modified subject QP followed by a nonpropositional wh-w ord are marginal, admitting only a functional answer, while simplex wh-questions with the same QP followed by the propositional adjunct wh-word way (why) are grammatical, admitting a single answer. This means that in simplex wh-questions, nonpropositional wh-words cannot take scope over a QP while the propositional adjunct wh- word way (why) can, given the assumption that a single answer signals the wide scope o f the wh-word over the QP. Next suppose the speaker asks the following questions in (41-45) to the hearer: (41) a. M otun salam-i nwukwu-lul chotayhayss-ni? every man-NOM whom-ACC invited-QM ‘W ho did everyone invite?’ b. M otun salam-i eti-se tampay-lul phiwess-ni? every man-NOM where-at cigarette-ACC smoked-QM ‘W here did everyone smoke?’ c. M otun salam-i encey tampay-lul phiwess-ni? every man-NOM when cigarette-ACC smoked-QM ‘W hen did everyone smoke?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 d. M otun salam-i ettehkey wass-ni? every man-NOM how came-QM ‘W hat is the means x such that everyone came by x?’ e. M otun salam-i M ary-eytayhay ettehkey malhayss-ni? every man-NOM M -about how talked-QM ‘W hat is the manner x such that everyone talked about Mary in x?’ (42) a. Ku twu salam-i nwukwu-lul chotayhayss-ni? the two man-NOM whom-ACC invited-QM ‘W ho did the two men invite?’ b. Ku twu salam-i eti-se tampay-lul phiwess-ni? the two man-NOM where-at cigarette-ACC smoked-QM ‘W here did the two men smoke?’ c. K u tw u salam-i encey tampay-lul phiwess-ni? the two man-NOM when cigarette-ACC smoked-QM ‘W hen did the two men smoke?’ d. K u tw u salam-i ettehkey wass-ni? the two man-NOM how came-QM ‘W hat is the means x such that the two men came by x?’ e. Ku twu salam-i M ary-eytayhay ettehkey malhayss-ni? the two man-NOM M -about how talked-QM ‘W hat is the manner x such that the two m en talked about M ary in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 (43) a. John-kwa M ary-ka nwukwu-lul chotayhayss-ni? J-and M -NOM whom-ACC invited-QM ‘Who did John and M ary invite?’ b. John-kwa M ary-ka eti-se tampay-lul phiwess-in? J-and M -NOM where-at cigarette-ACC smoked-QM ‘W here did John and M ary smoke?’ c. John-kwa M ary-ka encey tampay-lul phiwess-in? J-and M -NOM when cigarette-ACC smoked-QM ‘W hen did John and M ary smoke?’ d. John-kwa M ary-ka ettehkey wass-ni? J-and M -NOM how came-QM ‘W hat is the means such that John and Mary came by x?’ e. John-kwa M ary-ka Tom-eytayhay ettehkey malhayss-ni? J-and M -NOM T-about how talked-QM ‘W hat is the manner x such that John and Mary talked about Tom in x?’ (44) a. Taypwupwun-uy salam-i nwukwu-lul chotayhayss-ni? most-POSS man-NOM whom-ACC invited-QM ‘Who did most o f the men invite?’ b. Taypwupwun-uy salam-i eti-se tampay-lul phiwess-ni? most-POSS man-NOM where-at cigarette-ACC smoked-QM ‘W here did most o f the men smoke?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 c. Taypwupwun-uy salam-i encey tampay-lul phiwess-ni? most-POSS man-NOM when cigarette-ACC smoked-QM ‘W hen did most o f the men smoke?’ d. Taypwupwun-uy salam-i ettehkey wass-ni? most-POSS man-NOM how came-QM ‘W hat is the means x such that most o f the men came by x?’ e. Taypwupwun-uy salam-i M ary-eytayhay ettehkey malhayss-ni? most-POSS man-NOM M -about how talked-QM ‘W hat is the manner x such that most o f the men talked about M ary in x?’ (45) a. M otun salam-i Tom-ul way chotayhayss-ni? every man-NOM T-ACC why invited-QM ‘W hy did every man invite Tom ?’ b. Ku twu salam-i Tom-ul way chotayhayss-ni? the two man-NOM T-ACC why invited-QM ‘Why did the two men invite Tom ?’ c. John-kwa M ary-ka Tom-ul way chotayhayss-ni? J-and M -NOM T-ACC why invited-QM ‘W hy did John and M ary invite Tom?’ d. Taypwupwun-uy salam-i Tom-ul way chotayhayss-ni? M ost-POSS man-NOM T-ACC why invited-QM ‘W hy did most o f the men invite Tom?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 The simplex wh-questions in (41-45) with a definite subject QP are all grammatical and the sentences in (41-44) with a nonpropositional wh-word admit pair list answer, single answer, and functional answer, while the one in (45) admits only a single answer. Thus the hearer, faced with the question in (41a) for example, can felicitously answer the question by listing pairs below as in (46) or by giving a single answer as in (47) if it happened to be the case that everyone invited the same person, or by giving a functional answer as in (48). (46) John-un Bomin-lul, M ary-nun Ines-lul, Joseph-un Jinhee-lul, Barry-nun Sue-lul J-TOP B-ACC, M -TOP I-ACC J-TOP J-ACC B-TOP S-ACC Jim-un Fred-lul, Peter-nun Carol-lul, kuliko Jack-un Tom-ul chotayhayssta. J-TOP F-ACC P-TOP C-ACC and J-TOP T-ACC invited ‘John invited Bomin, M ary invited Ines, Joseph invited Jinhee, Barry invited Sue, Jim invited Fred, Peter invited Carol, and Jack invited Tom .’ (47) M otun salam-i Bomin-ul chotayhayssta. every man-NOM B-ACC invited ‘Everyone invited Bom in.’ (48) M otun salam-i caki emeni-lul chotayhass-ta. every man-NOM his mother-ACC invited ‘Everyone invited his m other.’ Although the pair list answer for the sentences in (44) is a choice pair list answer in that the pair list answer is essentially from the hearer’s own choice o f a majority set (see Groenendijk Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 and Stokhof 1984, Szabolcsi 1997b), as compared with the bona fide pair list answer in (41- 43), I will refer to both types o f pair list answer as a pair list answer without making further distinction between the two, since the distinction is immaterial for the discussion in this thesis. Now given the asymmetry o f scope between the two types o f wh-words as shown in simplex wh-questions with a modified subject QP in (36-38) and (40), one may wonder what the source o f the single answer in (41-45) is. I suggest that the single answer reading o f the sentences in (41-44) with a nonpropositional wh-word, contrary to the one o f the sentences in (45) with a propositional adjunct wh-word which involves wide scope o f the wh-word over the QP, is a special case o f a pair list answer as will be discussed in detail in section 4.1.1.3. To summarize, simplex wh-questions with a modified subject QP followed by a nonpropositional wh-word are marginal, admitting neither a pair list answer nor a single answer but only a functional answer, while simplex wh-questions with the same QP followed by the propositional adjunct wh-word way (why) are grammatical, admitting a single answer, attesting asymmetry o f scope taking between the two types of wh-words when it comes to wide scope construal o f the wh-word over the QP. Meanwhile, simplex wh-questions with a definite subject QP followed by a wh-word are all grammatical. Such simplex wh-questions with a nonpropositional wh-word admit pair list answer, single answer and functional answer, while the same questions with a propositional adjunct wh-word admit only a single answer. The single answer for simplex wh-questions with a definite subject QP followed by a nonpropositional wh-word will be argued to be a special case o f a pair list answer, as Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 opposed to the single answer in simplex wh-questions with a definite QP followed by the propositional adjunct wh-word way (why), which involves the wide scope o f the wh-word over the QP. The scope facts regarding the simplex wh-questions in (36-38), (40) and (41- 45) are summarized below in (49-50).1 5 PL, S and F stand for pair list answer, single answer and functional answer, respectively. (49) QP-NOM . . .nonpropositional w h-w ord... V PL S F QP- NOM Definite QP a. motun salam ‘everyone’ b. John-kwa M ary ‘John and M ary’ c. ku twu salam ‘the two m en’ d. taypwupwun-uy salam ‘most o f the m en’ OK OK OK QP- NOM M odified QP a. twu salam isang ‘more than two m en’ b. ttak twu salam ‘exactly two m en’ c. se salam miman ‘less than three m en’ * * OK (50) QP-NOM ...W H Y ... V PL S F QP- NOM Definite QP a. motun salam ‘everyone’ b. John-kwa M ary ‘John and M ary’ c. ku twu salam ‘the two m en’ d. taypwupwun-uy salam ‘most o f the m en’ * OK * QP- NOM M odified QP a. twu salam isang ‘more than two m en’ b. ttak twu salam ‘exactly two m en’ c. se salam miman ‘less than three m en’ * OK * At this point, before presenting our own analysis o f the scope facts as summarized in (49-50) above, I will briefly digress to review the proposals in the literature: uniform LF wh- 151 will call the subject QP types in (41-45) definite QPs following the tradition in the literature (see Milsark 1975, Larson and Segal 1995). Also see the related discussion of the definite QPs within the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 movement approaches by Hoji (1985) and Kim (1990), and uniform wh-in-situ approach by Aoun and Li (1993b) to see how they account for the scope facts in (49-50). 4.1.1.1. Uniform LF Wh-movement Approach In this section, I will review uniform LF wh-movement approach as in Hoji (1985) and Kim (1990). Hoji claims that wh-words in Japanese undergo movement at LF to COMP (Spec o f CP in current terms), following H uang’s (1982) proposal of a uniform LF wh-movement hypothesis. Kim (1990) on the other hand proposes that wh-words in Korean type languages are quantifiers and so should undergo standard quantifier raising either into IP or VP at LF, following M ay (1985). Although the specific assumptions they are making about the particular mode o f wh-movement are different, they converge on that in situ wh-words somehow undergo movement at LF. Let us start our discussion with Hoji (1985), first. Hoji observes that a sentence in which a subject QP precedes a wh-word is ungrammatical, while with the reversed word order the sentence is grammatical, as shown in (51). (51) a. D are-ka sake ka b iiru -o nom im asitaka who-NOM sake or beer-ACC drank QM ‘W ho drank sake or beer?’ b. * John ka Bill-ga nani-o nomimasita ka John or Bill-NOM what-ACC drank QM ‘W hat did John or Bill drink?’ (Hoji 1985: 267-268) framework of generalized quantification theory (Barwise and Cooper 1981, among others). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 Based on the grammaticality contrast as in (51) among other examples, Hoji (1985) proposes the following LF-filter constraint in (52), assuming that a wh-word undergoes movement into COMP (Spec o f CP in current terms) at LF, following Huang (1982): (52) At LF, *QPi, QPj, tj f where each member c-commands the member to its right (Hoji 1985: 248). 1617 The constraint in (52) accounts for the grammaticality in (51), given the LF representations in (53). (53) a. [s> WH-NOMi [ s QP-ACC , [ s t, [V P V tj ]]]] b. [s’ WH-ACCj [ s QP-NOM j [ s tj [V P V tj ]]]] The LF representation in (53b) for the sentence in (51b) violates the constraint in (52), and hence is ruled out as ungrammatical, while the LF representation in (53 a) for the sentence in (51a) conforms to the constraint in (52), and hence is ruled in as grammatical. Apparently, 1 6 Hoji (1985) assumes the notion of c-command proposed by Reinhart (1976: 32), which is defined as below. Node A c(constituent)-commands node B if neither A nor B dominates the other and the first branching node which dominates A dominates B. 1 7 The constraint in (52) is essentially a restatement of a proposal in Reinhart (1976) as cited below. ‘A logical structure in which a quantifier binding a variable x has wide scope over a quantifier binding a (distinct) variable y is a possible interpretation for a given sentence S just in case in the surface structure of S the quantified expression corresponding to y is in the c-command domain of the quantified expression corresponding to x.’ (Reinhart 1976: 191) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 the same contrast in grammaticality as in Japanese in (51) also holds in Korean, as illustrated in (54). (54) a. N wu-ka John-ina Mary-lul chotayhayss-ni? who-NOM J-or M-ACC invited-QM ‘Who invited John or M ary?’ b. *?John-ina M ary-ka nwukwu-lul chotayhayss-ni? J-or M -NOM whom-ACC invited-QM ‘W ho did John or M ary invite?’ As for the question in (54a), one can answer it by giving a single answer, while one cannot do so for the question in (54b). Based on the contrast in grammaticality in (54), one may adopt H oji’s LF filter constraint for Korean. Now consider the examples in (41-44) as repeated in (55-58): (55) a. M otun salam-i nwukwu-lul chotayhayss-ni? every man-NOM whom-ACC invited-QM ‘W ho did everyone invite?’ b. M otun salam-i eti-se tampay-lul phiwess-ni? every man-NOM where-at cigarette-ACC smoked-QM ‘W here did everyone smoke?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 c. M otun salam-i encey tampay-lul phiwess-ni? every man-NOM when cigarette-ACC smoked-QM ‘W hen did everyone smoke?’ d. M otun salam-i ettehkey wass-ni? every man-NOM how came-QM ‘W hat is the means x such that everyone came by x?’ e. M otun salam-i M ary-eytayhay ettehkey malhayss-ni? every man-NOM M -about how talked-QM ‘W hat is the manner x such that everyone talked about Mary in x?’ (56) a. Ku twu salam-i nwukwu-lul chotayhayss-ni? the two man-NOM whom-ACC invited-QM ‘Who did the two men invite?’ b. Ku twu salam-i eti-se tampay-lul phiwess-ni? the two man-NOM where-at cigarette-ACC smoked-QM ‘W here did the two men smoke?’ c. K u tw u salam-i encey tampay-lul phiwess-ni? the two man-NOM when cigarette-ACC smoked-QM ‘W hen did the two men smoke?’ d. Ku twu salam-i ettehkey wass-ni? the two man-NOM how came-QM ‘W hat is the means x such that the two men came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 e. Ku twu salam-i M ary-eytayhay ettehkey malhayss-ni? the two man-NOM M -about how talked-QM ‘W hat is the manner x such that the two men talked about M ary in x?’ (57) a. John-kwa M ary-ka nwukwu-lul chotayhayss-ni? J-and M -NOM whom-ACC invited-QM ‘Who did John and Mary invite?’ b. John-kwa M ary-ka eti-se tampay-lul phiwess-in? J-and M -NOM where-at cigarette-ACC smoked-QM ‘W here did John and M ary smoke?’ c. John-kwa M ary-ka encey tampay-lul phiwess-in? J-and M -NOM when cigarette-ACC smoked-QM ‘W hen did John and M ary smoke?’ d. John-kwa M ary-ka ettehkey wass-ni? J-and M -NOM how came-QM ‘W hat is the means such that John and Mary came by x?’ e. John-kwa M ary-ka Tom-eytayhay ettehkey malhayss-ni? J-and M -NOM T-about how talk-QM ‘W hat is the manner x such that John and Mary talked about Tom in x?’ (58) a. Taypwupwun-uy salam-i nwukwu-lul chotayhayss-ni? most-POSS man-NOM whom-ACC invited-QM ‘W ho did most o f the men invite?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 b. Taypwupwun-uy salam-i eti-se tampay-lul phiwess-ni? most-POSS man-NOM where-at cigarette-ACC smoked-QM ‘W here did most o f the men smoke?’ c. Taypwupwun-uy salam-i encey tampay-lul phiwess-ni? most-POSS man-NOM when cigarette-ACC smoked-QM ‘W hen did most o f the men smoke?’ d. Taypwupwun-uy salam-i ettehkey wass-ni? most-POSS man-NOM how came-QM ‘W hat is the means x such that most o f the men came by x?’ e. Taypwupwun-uy salam-i M ary-eytayhay ettehkey malhayss-ni? most-POSS man-NOM M -about how talked-QM ‘W hat is the manner x such that most o f the men talked about M ary in x?’ As mentioned already in 4.1.1, the above wh-questions in (55-58) with a definite subject QP followed by a wh-word are all grammatical, admitting pair list answer, single answer and functional answer. The LF structure o f these sentences will be the one in (59) under H oji’s system. (59) [s’ WH-wordj [ s QP-NOM , [ s t, [V P V tj ]]]] Throughout this chapter, I will notate the LF representation for Korean sentences as if Korean is head-initial for the reader’s convenience, unless specified otherwise. The LF structure in (59), however, violates the LF filter in (52). Thus the grammaticality o f the sentences in (55-58) necessarily invites a stipulation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 Hoji maintains that the grammaticality o f the sentence as in (60) in Japanese is due to the specific group construal o f the subject QP. (60) ??Daremo-ga nani-o kaimasita ka? everyone-NOM what-ACC bought-QM ‘W hat did everyone buy?’ (Hoji 1985: 270) Thus for a wh-question such as (60), which has a sequence o f a subject QP preceding a wh- word at S structure to be grammatical, the subject QP should necessarily be construed as a specific group expression. He presents the following example in (61) to show that a specific group interpretation o f the subject QP is indeed what is responsible for the improved grammaticality o f the sentence in (60): (61) *Daremo-ga sorezore nani-o kaimasita ka? everyone-NOM each what-ACC bought-QM ‘W hat did everyone each buy?’ (Hoji 1985: 270) The contrast in grammaticality between (60) and (61) is attributed to the fact that the subject QP with sorezore (each) in (61) is forced to be interpreted as a quantificational expression, while the same subject without sorezore (each) in (60) can be interpreted as referring to a specific group. As Suh (1990) observes, the Korean example in (62), which corresponds to the Japanese example in (60) is perfect, allowing both a pair list and a single answer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 (62) M otun salam-i mwues-ul sass-ni? every man-NOM what-ACC bought-QM ‘W hat did everyone buy?’ (Suh 1990: 195) Joo (1989) also observes that a sentence like the one in (62) admits a pair list answer and a single answer, although the universal QP adopted by Joo is nwukwuna (everyone) (see Joo 1989: 186). I further observe that even when a quantificational interpretation o f the subject QP is forced by kakkak (each), as in (63), the sentence is still grammatical, yielding a pair list answer. (63) M otun salam-i kakkak mwues-ul sass-ni? every man-NOM each what-ACC bought-QM ‘W hat did everyone each buy?’ Thus the example in (63) clearly suggests that the grammaticality o f the sentences in (55-58) has nothing to do with a nonquantificational specific group interpretation o f the subject QP. After all the judgm ent o f Hoji with regard to the Japanese sentence in (60) is not uncontroversial, a point which I will return to in Chapter 3. Now given the availability o f the pair list reading in (55-58), one is led to believe that the constraint in (52) cannot be a correct account o f QP and wh-word scope interaction for these sentences. These sentences are wrongly predicted by the constraint (52) in conjunction with H oji’s assumption o f uniform obligatory LF wh-movement to have only a single answer, since a nonquantificational group interpretation for the subject QP is the only option to save Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 the constraint (52). The problems in H oji’s system do not end here. W hen one considers wh- questions with a modified QP, the LF filter constraint in (52) poses a serious problem to accounting for the contrast in grammaticality between the sentences with a modified QP followed by a nonpropositional wh-word and the sentences with the same QP followed by a propositional adjunct wh-word. For this, let us consider (36) and (40) as repeated in (64-65). (64) a. ?*Twu salam isang-i nwukwu-lul chotayhayss-ni? two man more-NOM whom-ACC invited-QM ‘Who did more than two men invite?’ b. ?*Twu salam isang-i eti-se tampay-lul phiwess-ni? two man more-NOM where-at cigarette-ACC smoked-QM ‘W here did more than two men smoke?’ c. ?*Twu salam isang-i encey tampay-lul phiwess-ni? two man more-NOM when cigarette-ACC smoked-QM ‘W hen did more than two men smoke?’ d. ?*Twu salam isang-i ettehkey wass-ni? two m an more-NOM how came-QM ‘W hat is the means x such that more than two men came by x?’ e. ?*Twu salam isang-i M ary-eytayhay ettehkey malhayss-ni? two man more-NOM M -about how talk-QM ‘W hat is the manner x such that more than two men talked about M ary in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 (65) a. Twu salam isang-i Tom-ul w ay chotayhayss-ni? two man more-NOM T-ACC why invited-QM ‘W hy did more than two men invite Tom ?’ b. ?Se salam miman-i Tom-ul way chotayhayss-ni? three man less-NOM T-ACC why invited-QM ‘W hy did less than three men invite Tom ?’ c. Ttak twu salam-i Tom-ul w ay chotayhayss-ni? exactly two man-NOM T-ACC why invited-QM ‘W hy did exactly two men invite Tom ?’ As mentioned in section 4.1.1, the sentences in (64) are marginal in grammaticality, admitting neither a pair list answer nor a single answer but only a functional answer, while the sentences in (65) are grammatical with a single answer, hence exhibiting asymmetry of scope when it comes to wide scope construal o f the wh-word over the subject QP, given the assumption in the literature (M ay 1985, among others) that the presence o f a single answer signals the wide scope o f the wh-word over the subject QP. The LF representation for the above sentences in (64) and (65) will be the one in (66) in H oji’s system. (66) [s> WH-wordi [s QP-NOM, [s tj [ V r t, ]]]] The above LF representation for the sentences in (64-65), however, violates the LF filter constraint in (52). Thus, in H oji’s system, the status o f the sentences is predicted to be the same, i.e., ungrammatical, contrary to the fact. One may attempt postulating a specific group interpretation o f the modified QPs in (65) to account for the grammaticality while Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 maintaining the LF filter constraint. This is not a plausible option, given that m odified QPs normally resist specific interpretation (Reinhart 1997, Beghelli and Stowell 1997) as in (67). (67) a. Five students left shortly after the exam started. They could not understand the questions. b. M ore than four students left shortly after the exam started. They could not understand the questions. (Reinhart 1997: 385) In a situation, where 10 students actually left, the pronoun in (67a) can refer to just five students while the pronoun in (67b) cannot refer to just any particular group o f students numbering greater than 4. So in case only five students out o f 10 who left did not understand the questions, (67b) is false while (67a) is true, according to Reinhart (1997). Thus to the extent that modified QPs cannot easily be interpreted as a specific group denoting expressions, one cannot attribute the grammaticality o f the sentences in (65) to the availability o f a specific group interpretation o f the subject QP. For argument’s sake, let us suppose that the modified QPs in (65) somehow can have a specific group interpretation, allowing the wide scope o f the wh-word over the QP, hence yielding a single answer. However, to the extent that a specific group interpretation o f the modified QPs in (65) renders these sentences acceptable, it should be the case that the sentences in (64) with the same modified subject QPs should be on a par with the sentences in (65) in terms of grammaticality, allowing a wide scope o f the wh-word over the QP, hence a single answer. However, this is not the case. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 The data in (64-65) thus further suggest that positing the availability o f a specific group interpretation cannot save the LF filter which is based on the uniform wh-movement hypothesis, since doing so would leave the asymmetry o f scope between the two types o f wh-words in (64-65) unaccounted for. It may be instructive at this point to review the motivations for the LF uniform wh-movement hypothesis, upon which H oji’s (1985) LF filter constraint is crucially based. The hypothesis for uniform LF wh-movement is grounded mainly in three considerations originally from Huang (1982): selectional restriction, contrast in locality effects o f wh-words involving syntactic islands, and scope o f wh-words. An argument based on selectional restriction comes from the paradigm in (68). (68) a. Zhangsan wen wo [ cp she* mai-le shu] Zhangsan ask me who buy-ASP book ‘Zhangsan asked who bought books.’ b. Zhangsan xiangxin [cp shei mai-le shu]? Zhangsan believe who buy-ASP book ‘Who does Zhangsan believe bought books?’ (Huang 1982: 254) Verbs like ask take a wh-question as a complement cross-linguistically while verbs like believe take a proposition. In Chinese as shown in (68), both types o f matrix verb apparently seem to take a wh-clause, however. According to Huang, this is not a problem, once the abstract wh-movement at LF is assumed such that the selectional restriction is made in terms o f the matrix verb and the wh-word in COMP (Huang 1982: 258). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Another argument for uniform LF wh-movement comes from the contrast in locality effect between the adjunct wh-word and the argument wh-word as in (69) involving a syntactic island. (69) a. [ n p [ cp shei xie ] de shu ] zui youqu? who write DE book most interesting ‘Books who wrote are most interesting?’ b.*[Np[cpta weishenme xie] de shu ] zui youqu? he why write DE book most interesting ‘Books that he wrote why are most interesting?’ (Huang 1982: 526) The contrast in locality between the adjunct wh-word and the argument wh-word in (69) receives a straightforward account, given the assumption that ECP but not subjacency is the valid locality constraint at LF. 1 8 The argument wh-word is lexically governed by its verb in (69a), satisfying the ECP, hence immune to locality, while the adjunct wh-word in (69b) cannot be antecedent governed due to the intervening maximal projections (S’ and NP in Huang’s system), violating ECP, hence exhibiting locality effect, which leads to the contrast in grammaticality. 1 8 Huang (1982: 471) adopts the notion of proper government of Aoun and Sportiche (1981), given below. A properly governs B if and only if A governs B and (a) A is a lexical category, or (b) A is co-indexed with B. Please note that proper government is a subcase of government, and that government does not obtain when a maximal projection boundary intervenes between the governor and the govemee in his system. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 W ide scope o f wh-words over the subject QPs as in (70) is also suggested as an additional argument for the movement o f wh-word at LF. ‘W hat did everybody buy?’ (Huang 1982: 263) The sentence in (70) can have a single answer that signals the wide scope o f the wh-word over the subject QP and thus indicates the movement o f the in situ wh-word at LF into Spec o f CP at LF, where it can c-command the subject QP, according to Huang (1982). I find, however, that none o f the three arguments for LF wh-movement are convincing. First, when it comes to selectional restriction in Japanese and Korean, which are also wh-in situ languages like Chinese, the requirement is met by the question morpheme as noted by Kim (70) mei-ge-ren dou mai-le sheme? every-CL -person all buy-ASP what (1990). (71) a. John-un [cp M ary-ka mues-ul sat-nya-ko] muletta. J-TOP M -NOM what-ACC bought-QM -COM P asked ‘John asked what M ary bought.’ b. John-wa [ cp M ary-ga nani-o katta ka-to] tazuneta. John-TOP M -NOM what-ACC bought-QM -COM P asked ‘John asked what M ary bought.’ Kim (1990: 227) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 (72) a. John-un [ c p M ary-ka m ues-ul satta-ko] sayngkakha-ni? J-TOP M -NOM what-ACC bought-COM P think-QM ‘W hat does John think M ary bought?’ b. John-wa [cp M ary-ga nani-o katta-to] omotteru-no? J-TOP M -NOM what-ACC bought-COM P think-QM ‘W hat does John think M ary bought?’ As shown in (71-72), the existence o f the question morpheme in the embedded clause indicates whether an embedded clause with a wh-word is a wh-clause or not. Next, the contrast in locality effects between the sentences in (69) can also be accounted for independently, if the argument wh-words as in (69a) can stay in situ, while the adjunct wh- words as in (69b) are forced to undergo movement, hence leading to a subjacency (Chomsky 1986) violation. The scope o f the wh-word in (70) will be discussed in Chapter 3, where I will try to show that the single answers available in sentences like the one in (70) do not necessarily signal wide scope o f the wh-word over the subject QP. In fact, it will be claimed that Chinese nonpropositional wh-words in simplex wh-questions cannot take scope over subject QPs, contrary to H uang’s claim. It thus seems that none o f the arguments for the uniform LF wh-movement hypothesis are conclusive. In fact, the asymmetric scope taking between the propositional adjunct wh-word and nonpropositional wh-words in simplex wh-questions argued for in Korean in section 3, which is directly associated with their different status as an indefinite, seriously undermines the validity for the uniform LF wh-movement hypothesis at least when it comes to Korean. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 Moreover, I will show in the next chapter that uniform LF wh-movement hypothesis cannot be maintained in Japanese and Chinese either, given considerations for scope in wh- questions in Japanese and Chinese. As shown thus far, H oji’s LF filter cannot account for the scope facts in sentences with a definite subject QP as in (55-58) and in sentences with a modified subject QP as in (64-65). The resort on specificity to avoid the LF filter constraint cannot be o f any help. It gains nothing towards accounting for the asymmetry o f scope in (64-65) and the availability o f the pair list answer in (55-58). Moreover, the uniform LF wh-movement hypothesis, which his proposal is crucially based on is not convincing at all, as discussed above. Let us turn to Kim (1990) next. Kim (1990) argues that wh-words in Korean are quantifiers, which uniformly undergo quantifier raising at LF either into IP or VP via quantifier raising as in May (1985), thus diverging from Huang (1982) and Hoji (1985) who propose that wh- words undergo wh-movement into COM P at LF. To the extent that his quantifier raising approach assumes movement for wh-words, it may be called a uniform LF wh-movement approach. Kim (1990) proposes the topic parameter in (73) for Korean, maintaining that Korean is a topic prom inent language, which forces the subject QP to move into topic position at S structure i.e., Spec o f CP in the default case. (73) Topic Parameter Every matrix clause in Chinese type languages has a topic position that must be filled overtly at S-structure. (Kim 1990: 125) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 The movement o f the subject into Spec o f CP at S structure enforced by the language specific parameter in (73) thus leads the chain scope as stated in (74) to be established at S structure.1 9 (74) chain scope ( or c-scope) = df a set o f nodes (i) that are c-commanded by an operator a, and (ii) that dominate at least one member of the chain o f a (Kim 1990: 24) He further proposes a well-formedness condition on chain scope as in (75). (75) W ell-formedness condition on chain scope. The relative scope relation o f a and (3 is established at the level where either cx or |3 acquires its chain scope. (Kim 1990: 27) Kim further proposes that if the chain scope o f a quantifier A dominates that o f another quantifier B (or simply B), A has scope over B. Otherwise there is scope ambiguity. Given K im ’s proposal, the S structure representation for the sentences in (55-58) will look like the one below in (76). (76) [ c p QP-NOMi [ i p t, [V P W H-word ]]] 1 9 It is not clear whether Kim (1990) intends the notion of c-command of Reinhart (1976: 32) or that of May (1985: 34). Either option may work for him since QR always targets a maximal projection. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 The QP in the subject position should undergo movement into Spec o f CP at S structure to satisfy the Topic Parameter in (73). This means that the relative scope o f the subject QP and the wh-word is determined at S structure, given the well-formedness condition on chain scope in (75). According to the definition o f chain scope in (74), the chain scope o f the subject QP in (76) is {IP} and it dominates the wh-word, leading to wide scope o f the subject QP over the wh-word. Given his crucial assumption that narrow scope o f a wh-word with respect to a QP yields ungrammaticality (Kim 1990: 242), the grammaticality o f the sentences in (55-58) needs an additional stipulation akin to that of Hoji (1985), i.e., a specific group interpretation o f the subject QP (Kim 1990: 2 4 0 ff).2 0 To the extent that he relies on a specific group interpretation to account for the grammaticality o f the sentences with a definite subject QP as in (55-58), Kim seems to encounter essentially the same empirical problem as in Hoji (1985) to accounting for the availability o f the pair list answer in these sentences. The contrast in grammaticality as in (64-65) also poses a nontrivial empirical problem to Kim. For this, let us consider the examples, as repeated in (77-78). (77) a. ?*Twu salam isang-i nwukwu-lul chotayhayss-ni? two m an more-NOM whom-ACC invited-QM ‘W ho did more than two men invite?’ 2 0 Kim maintains that S structure can also be a relevant level of syntactic representation for semantic interpretation (1990: 27). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 b. ?*Twu salam isang-i eti-se tampay-lul phiwess-ni? two man more-NOM where-at cigarette-ACC smoked-QM ‘W here did more than two men smoke a cigarette?’ c. ?*Twu salam isang-i encey tampay-lul phiwess-ni? two man more-NOM when cigarette-ACC smoked-QM ‘W hen did more than two men smoke a cigarette?’ d. ?*Twu salam isang-i ettehkey wass-ni? two man more-NOM how came-QM ‘W hat is the means x such that more than two men came by x?’ e. ?*Twu salam isang-i M ary-eytayhay ettehkey malhayss-ni? two m an more-NOM M -about how talk-QM ‘W hat is the manner x such that more than two men talked about M ary in x?’ (78) a. Twu salam isang-i Tom-ul way chotayhayss-ni? two man more-NOM T-ACC why invited-QM ‘Why did more than two men invite Tom ?’ b. ?Se salam miman-i Tom-ul way chotayhayss-ni? three man less-NOM T-ACC why invited-QM ‘W hy did less than three men invite Tom ?’ c. Ttak twu salam-i Tom-ul way chotayhayss-ni? exactly two man-NOM T-ACC why invited-QM ‘Why did exactly two men invite Tom ?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 In Kim ’s system, the sentences in (77-78) will have the following S structure representation in (79): (79) [C P QPi-NOM [IP h [V P W H-word ]]] This will lead the chain scope o f the subject QP to be established at S structure, leading to wide scope for the subject QP. Thus the adjunct wh-word will end up having narrow scope with respect to the subject QP in (78), leading to ungrammaticality in his analysis, contrary to the facts. For argument’s sake, let us again suppose that the modified subject QPs in (78) can have a specific group interpretation to correctly account for their grammaticality, with a single answer. However, to the extent that specific group interpretation o f the modified QPs in (78) renders the sentences acceptable, it should be the case that the sentences in (77) with the same QP should be on a par with the sentences in (78) in terms o f grammaticality, with a single answer, which, however, is not the case. Kim ’s proposal for the wh-words in Korean type languages as quantifiers is not convincing, either. He claims that wh-words in Korean type languages are quantifiers since they do not undergo overt movement, given that quantifiers typically do not undergo quantifier-raising overtly. However, given that an indefinite does not undergo movement, his claim for wh- words as quantifiers solely based on their lack o f overt movement is too strong. Moreover, the examples in (27-28) in section 2 repeated in (80-81), where nonpropositional wh-words can take scope out o f the adjunct island, their quantificational force determined by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 the external quantificational element, kkok (always), in contrast to the propositional adjunct wh-word, clearly demonstrate that the former are indefinites while the latter is not. (80) a. [C p Nwui-ka o-myen] ( pro,) kkok wuli-lul pangmwunhanta. who-NOM com e-if always us-ACC visit ‘for every x, x an individual, if x comes, x visits us.’ b. [C p LA-eyse enu chinkwui-ka o-myen] (proO kkok wuli-lul pangmwunhanta. LA-from which friend-NOM com e-if always us-ACC visit ‘for every x, x a friend, if x comes from LA, x visits us.’ c. [C p John; -i eti-lo oychwulha-myen] ( proj) kkok nusskey tolaonta. J-NOM where-to go out-if always late returns ‘for every x, x a place, if John goes out to x, he returns home late (from x).’ d. [cp John; -i encey o-myen] (proO kkok wuli-ul pangmwunhanta. J-NOM when com e-if always us-ACC visit ‘for every x, x a time, if John comes at x, he visits us (at x).’ e. [cp Mary; -nun cheycwung-i cakip-uy cengsang cheycwung-pota M -TOP weight-Nom her-POSS normal weight-than elma te naka-myen ] (pro,) kkok diet-lul hanta. how m uch more w eigh-if always diet-ACC do ‘for every x, x a weight, if M ary weighs x more than normal weight, she goes on a diet (for x).’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 f. [ c p Johnri ettehkey uli cipey o-myen] J-NOM how our house-to com e-if (proj) kkok senmwul-ul kacikoonta. always gift-ACC bring ‘for every x, x a means, if John comes to our place by x, he brings a gift (by x).’ g. [ c p Nay-ka ettehkey John-eytayhay malha-myen] I-NOM how J-about talk-if M ary-nun kkok hwalul naykonhanta. M -TOP always get angry ‘for every x, x a manner, if I talk about John in x, Mary gets angry (with x).’ (81) *[cpJohni-i way o-myen] (proO kkok wuli-lul pangmwunhanta. To summarize, any proposal built on the assumption that wh-words in Korean type languages uniformly undergo movement at LF for scope taking either by wh-movement as in Hoji (1985), or quantifier raising as in Kim (1990), cannot be maintained, given the asymmetric scope taking strategy between the two types o f wh-words as argued in section 3. Since their proposals are crucially based on the wrong hypothesis that wh-words uniformly undergo movement at LF, it is only a natural consequence that they encounter empirical problems in accounting for the asymmetry o f scope in the simplex wh-questions with a modified J-NOM why com e-if always us-ACC visit ‘for every x, x a reason, if John comes for x, he visits us (for x).’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 subject QP as in (64-77) and (65=78) between the two types o f wh-words and the availability o f the pair list reading in simplex wh-questions with a definite QP as in (55-58). 4.1.1.2. Uniform Wh-in-situ Approach Aoun and Li (1993b) argue that wh-words in Chinese are indefinites and are interpreted in situ at LF without undergoing movement at all. They propose that the null question operator (QU) is base-generated in Spec o f QuP between IP and VP (Aoun and Li 1993b: 232) and that QU undergoes overt movement into Spec o f CP, marking the scope o f the in situ wh- word. Adopting their proposal, we may have the structure below in (83) for the sentences in (45) as repeated in (82): (82) a. Mo tun salam-i Tom-ul w ay chotayhayss-ni? every man-NOM T-ACC why invited-QM ‘W hy did everyone invite Tom ?’ b. Ku twu salam-i Tom-ul w ay chotayhayss-ni? the two man-NOM T-ACC why invited-QM ‘W hy did the two men invite Tom ?’ c. John-kwa M ary-ka Tom-ul way chotayhayss-ni? J-and M -NOM T-ACC why invited-QM ‘Why did John and Mary invite Tom ?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 d. Taypwupwun-uy salam-i Tom-ul w ay chotayhayss-ni? Most-POSS man-NOM T-ACC why invited-QM ‘Why did most o f the men invite Tom ?’ (83) [ c p QUj [IP QP-NOM, [IP t, [r [Q u P tj WH-wordj ]]]]] The above structure does not violate the Minimal Binding Requirement as stated below in (84), since although the subject QP is the closer A ’-binder for tj, it is not a potential binder, since the co-indexing o f tj with the subject QP would make tj A-bound by t„ leading to a binding Condition C violation. (84) The Minimal Binding Requirement A variable must be bound by the most local potential A ’-binder. (A qualifies as a potential A ’-binder for B iff A c-commands B and the co-indexing o f A and B would not violate the binding principles) (Aoun and Li 1993a: 11, Aoun and Li 1993b: 204) The Scope Principle below in (85) thus predicts the sentences in (82) to be ambiguous, with the subject QP and the wh-word taking scope over each other, since the QU in Spec o f CP c- commands the subject QP and the latter c-commands the trace o f the QU in Spec o f Q u P .2 1 2 1 In Aoun and Li (1993b), the trace of the question operator is viewed as an A’ element, (see Aoun and Li 1993b: 227 fn.25) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 (85) The Scope Principle An operator A may have scope over an operator B iff A c-commands B or an A ’ element in the chain headed by B. (Aoun and Li 1993a: 11, Aoun and Li 1993b: 204) This is not 'a correct prediction, however, since the sentences in (82) can have only a single answer as discussed in section 4.1.1. Also when questions with a modified QP as in (77) and (78) are considered, Scope Principle in (85) predicts these sentences to be ambiguous, with the subject QP and the wh-word taking scope over each other, since the QU in Spec CP c- commands the subject QP and the latter c-commands the trace o f the QU in Spec QuP, given an LF representation analogous to (83) for these sentences. This prediction is not correct either: Sentences with a non-propositional wh-word in (77) are marginal, admitting only a functional answer, while sentences with a propositional adjunct wh-word in (78) are grammatical, admitting a single answer. One may fix the problem with their system by assuming c-command to be a necessary condition for the determination o f relative scope between scope bearing elements to the effect that only certain types o f QPs, i.e., definite QPs can take wide scope over wh-words. That will do for the sentences in (78), correctly predicting the wide scope o f the wh-word, hence yielding a single answer. However, the problem in (77) remains, since even w ith the qualification on the QP type for taking scope over the wh-word, their system will still allow a wide scope of the wh-word over the subject QP, since the QU in Spec CP c-commands the subject QP, hence a single answer reading contrary to the fact. As we have seen thus far, neither the uniform LF wh-movement approach nor the uniform wh-in-situ approach seems to be able to give an adequate account for the scope facts in (49- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 50). Moreover, neither o f the two approaches can be maintained given the asymmetric scope taking between the propositional adjunct wh-word and the non-propositional wh-words, which thus prompts one to make an alternative proposal. 4.1.1.3. Proposal I have thus far tried to establish that neither the uniform LF wh-movement approaches nor the uniform wh-in-situ approach can adequately account for the scope pattern as summarized in (49-50). In this section, I will present an alternative account for the QP and wh-word scope interaction in wh-questions. I will assume the quantifier-raising hypothesis for QPs in wh-questions as in M ay (1977, 1985). However, unlike May, I will adopt the view that a QP is a second-order predicate o f semantic type « e ,t > t > and that quantifier raising (QR, henceforth) is essentially driven by the need to avoid type mismatch, following Pesetsky (1982), Heim and Kratzer (1998) and Fox (2000). Given this view, it follows that the subject QP, which is raised into Spec IP by the subject raising operation will not have to undergo QR, while an object QP is forced to undergo QR, assuming the standard VP internal subject hypothesis (Koopman and Sportiche 1985, 1988, Fukui and Speas 1986, Kuroda 1988, and Pollock 1989, among others). The subject QP in Spec IP can avoid a type mismatch given a ^.-abstraction operation, while the object QP cannot do so without being raised out o f VP. The question to be further asked is how far the object QP is raised by QR. Following Heim and Kratzer (1998), I will assume that QR targets a clause-denoting expression, i. e., o f semantic type <t >. Thus by QR, the object QP will adjoin to either VP or IP, which is o f Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 semantic type o f < t>, which is essentially the view also taken by May (1985). Now let us consider the following example in English: (86) a. Who bought everything for Max? b. W hat did everyone buy for Max? (May 1985: 38-39) The sentence in (86b) can be answered either by a single answer as ‘Everyone bought Max a Bosendorfer piano’ or by a pair list answer as ‘Mary bought Max a tie, Sally a sweater, and Harry a piano’, according to M ay (1985). M eanwhile (86a) can be answered only by a single answer as ‘Oscar bought everything for M ax’. It has been claimed by many researchers (Groenendijik and Stokhof 1984, M ay 1985, among others) that single answer signals the wide scope o f the wh-word over the QP while pair list answer signals the wide scope o f the latter over the former, i.e., quantifying into questions. May (1985), for example, accounts for the single answer and the pair list answer as in (86) directly from QP and wh-word scope interaction. The contrast in scopal possibilities in (86), according to May, is captured by the scope principle in (87) together with the ECP in (88). (87) Scope Principle M embers o f E-sequence are free to take on any type o f relative scope relation. A class o f occurrences o f operators \j/ is a E-sequence if and only if for any Oj Oj e \|/, O; governs Oj, where operator means phrases in A-bar position at LF. (May 1985: 34) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 (88) a governs (3 =df a c-commands [3 and there are no maximal projection boundaries between a and (3. (M ay 1985: 33) a c-commands (3 =df every maximal projection dominating a dominates (3, and a does not dominate (3. (M ay 1985: 34) To see how the scope ambiguity between the QP and the wh-word is captured in his system, consider the following LF representations below in (89) for the sentences above in (86): (89) a. [s> whoj [s h [ V p everythingj [V p bought tj for Max ]]]] b. [s’ whatj [s everyone; [s t, bought tj for M ax ]]] As for the LF representation in (89a), the wh-word antecedent-govems its trace in the subject position and the trace o f everything is lexically governed by the verb, hence it is a well formed representation in terms o f ECP, which forces the object QP to be adjoined to VP but not S, hence leading to the wide scope reading o f the wh-word. It should be noted that in his system it is stipulated that the subject and its trace should be adjacent to each other to satisfy antecedent government for the subject trace. Consider the LF representation in (89b) now. The trace o f the wh-word is lexically governed by the verb and the trace o f the subject is antecedent-governed by everyone, which is adjacent to it. Thus it is a well-formed representation in terms o f ECP (or more precisely, in terms o f the path containment condition, see Pesetsky 1982: 309). The scope principle in (87) and the related notions in (88) correctly predict the LF in (89b) for (86b) to be ambiguous, since no maximal projection Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 boundary intervenes between the wh-word and the subject QP. Please note that S is not a maximal projection in M ay’s system. Following Karttunen (1977), Engdahl (1986), and Chierchia (1991, 1993), I claim, however, that pair list answer is not derived from scope interaction between the QP and wh-word, i.e., quantifying into questions. Quantifying into questions is problematic technically, since quantification as a semantic operation is defined on propositions, not on questions, given the semantics o f a wh-question as sets o f propositions (Hamblin 1973: 48, and Karttunen 1977: 10), as pointed out by Chierchia (1991: 80). 2 2 Quantification into questions as the source of a pair list answer is problematic empirically too, given the availability o f the pair list answer as in (90). (90) Who do you think everyone saw at the rally? (May 1985:45) The above question in (90) admits a pair list answer as well as a single answer. If quantification into questions is the source o f a pair list answer, one needs to assume QR o f the universal QP out o f the embedded CP, which, however, is not plausible given that CP complement clause as in (90) never allows a universal QP such as everyone to scope out o f it (see Farkas 1981, Beghelli 1997 and Reinhart 1997, among others). 2 2 Paul Hagstrom (personal communication) points out that as far as one can repair the quantification operation to allow quantification ito questions (see Hagstrom 1998: 171 ff), the operation will not be a problem. However, quantification into questions still encounters empirical problems of the kind as in (90). Please refer to Engdahl (1986) and Chierchia (1993) for other arguments against quantifying into questions as the source of a pair list answer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 As an alternative, I claim that a pair list answer to a wh-question obtains when a definite QP c-commands and thus binds the indefinite some part o f the wh-word at LF, assuming wh- word consists o f a w/z-operator and indefinite some part, following the intuition in the literature (Chomsky 1964: 38, Katz and Postal 1964: 93, Klima 1964: 253, Kuroda 1969: 266ff, and Vergnaud and Zubizarreta 2002, among others). Throughout this thesis, I will opt for c-command in terms o f the first branching node in the sense o f Reinhart (1976: 32) in the current system, and suggest that c-command is a necessary condition for the determination o f the relative scope relations between scope bearing elements at LF, adopting the minimalist tenet that LF is the level o f representation for semantic interpretation following Chomsky (1995). To be specific, let us suppose that the wh-word who in (90), consisting o f a wh- operator and indefinite someone is decomposed into a wh-operator and someone at LF such that the latter is reconstructed (see Heim 1987). Let us assume that who has a DP structure where the trace o f wh-operator is sitting in Spec o f DP. The LF representation for (90) after reconstruction will be the following in our system :2 3 2 4 (91) [ cp whj [ip you think [Cp [ip everyone [ v p saw [Dp t, [ n p someone ]], at the rally ]]]]] Please note someone in (91) is an indefinite that does not have its own quantificational force. It needs to be bound by an external quantificational force for proper interpretation. Suppose someone is bound by the universal quantifier everyone as shown in (92). 2 3 To derive the structure in (91), one may need Copy and Move operation followed by the complimentary deletion in the sense of Chomsky (1995). 2 4 The index of the DP is the result of spec-head agreement in DP and subsequent percolation from the head of DP into its maximal projection. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 (92) [Cp wh; [ ip you think [ cp [ ip everyonej [ v p saw [ d p t, [N P someonej ]]i at the rally ]]]]] The indefinite someone thus picks up the definite quantificational force o f everyone by being bound by it. Since the indefinite someone now has a definite quantificational force, the individuals as replaced by the indefinite someone can be identified by the hearer o f the question in (90), hence the hearer may give an answer by listing pairs. The reader should recall that it is a standard observation in the literature that a definite subject QP as illustrated in (93) can induce a pair list answer, although the pair list answer in (93d) is different from the bona fide pair list answer available in (93abc) in that it is a pair list answer based on the hearer’s own choice o f a majority set (see Groenendijk and Stokhof 1984: 143, Chierchia 1991: 81, and Liu 1997: 143, among others). (93) a. Who did everyone invite? b. W ho did John and Bill invite? c. W ho did the two men invite? d. Who did most o f the men invite? Now one may wonder what blocks the pair list reading for the sentence in (86a) in the present system. Given that QR can target either VP or IP in the current system, one cannot exclude the object QP from adjoining to IP from which the QP can bind the indefinite some part o f the subject wh-word. For this, consider the following initial representation in (94) for the sentence in (86a), where QP has undergone QR to adjoin IP, since this option is allowed Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 6 in our system, assuming only A ’ movement reconstructs (Chomsky 1995, Lasnik 1999, and Jean-Roger Vergnaud, personal com m unication).2 5 (94) [C P wh; [IP everythingj [IP [D P tj [N P someone]]; [V P t, bought tj for Max ]]]] Please recall that someone is an indefinite whose quantificational force is determined by the external quantificational element. Thus depending on whether it is bound by the wh-operator or the QP, the initial representation in (94) will have the following LF representations in (95a) and (95b), respectively: (95) a. [C p whj [iP everythingj [iP [D P t, [N P someone;]]; [V p t, bought tj for Max ]]]] b. [ cp whj [ ip everythingj [iP [D p t, [N P someonej]]; [ v p tj bought tj for M ax ]]]] In (95a), the wh-operator binds the indefinite someone, thus the indefinite someone is interpreted as who. The wh-operator c-commands the subject QP, taking wide scope over it, hence yielding a single answer. Pair list answer is not a possibility, since the indefinite someone is not bound by the definite QP. How about the LF representation in (95b)? Since the universal QP, which is definite, binds the indefinite someone, one may expect a pair list answer in our system. The LF structure is illegitimate, however. The binding o f indefinite someone by the universal QP invariably leads to w eak crossover violation. Please recall 2 5 It has been a controversy whether A-movement reconstructs since May (1977: 189) introduced and discussed the ambiguity of the following sentence: Some politician is likely to address John’s constituency. See Chomsky (1995: 32615) and Lasnik (1999) for arguments against A-reconstruction. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 that an indefinite is treated as a variable in our system following Lewis (1975) and Heim (1992). Thus no pair list answer is available. Accordingly, the sentence in (86a) cannot have a pair list answer, since neither LF representation for it in (95) allows it. Our account is reminiscent o f Chierchia’s (1991, 1993) proposal that the variable o f a wh- word has a composite functional and referential indices and pair list answer obtains as a special case o f a functional answer. In Chierchia (1991, 1993), the LF representation o f the sentence in (86a) will be the following in (96): (96) [ C p who; [ ip [ n p t, [N P e j ] ] i [V P bought everythingj for Max]]] The movement o f everything to bind the empty pronominal with the index j for a pair list answer and functional answer triggers weak crossover violation. Our position that a wh-word in English consists o f a wh-operator and the indefinite some {one, thing...) directly accounts for Cheirchia’s intuition that the trace o f a wh-word has complex indices. In Chierchia’s proposal the coindexing between everything and the empty pronominal in (96) is crucially based on the proposal that everything is an argument o f the function o f the whole NP including the empty pronominal. This position leads to the conclusion that pair list answer is a special case o f a functional answer (also see Engdahl 1986 for the same position).2 6 2 6 The internal struature of a wh-trace has drawn much interest recently in the investigation of QP and wh-word scope interaction. Sloan (1991:233), for example, suggests that wh-word consists of a wh- operator and a null anaphoric pronoun. Sloan further argues that a pair list answer arises when a QP locally licenses the null anaphoric pronoun. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. There, however, exists a nontrivial difference between the current proposal and the one in Chierchia (1991, 1993). The obvious difference is in the nature o f the wh-trace. In Chierchia, it contains an empty pronominal. In the current proposal, it contains an indefinite. Thus in Chierchia, additional syntactic procedure o f absorption is required to disambiguate the LF representation for a pair list answer from the one for a functional answer. 2 7 In the present system, pair list answer simply obtains when the definite QP binds the indefinite some part o f a wh-word, thus the latter picking up the definite quantificational force o f the former. As the indefinite part o f a wh-word is definite now, the hearer can identify the individuals (or measure, means, manner) as replaced by the indefinite and thus pair list answer can be given. I will come back to the issue o f how a functional answer obtains in section 6. W ith our discussion o f QP and wh-scope interaction in English in mind, now let us turn to Korean QP and wh-scope interaction. Consider simplex wh-questions with a modified subject QP as in (36=64=77) and (37-38), which we repeat as (97-99). (97) a. ?*Twu salam isang-i nwukwu-lul chotayhayss-ni? two man more-NOM whom-ACC invited-QM ‘Who did more than two men invite?’ b. ?*Twu salam isang-i eti-se tampay-lul phiwess-ni? two man more-NOM where-at cigarette-ACC smoked-QM ‘W here did more than two men smoke?’ 2 7 Chierchia (1993: 210) adopts Aborption as defined below, essentially following Higginbotham and May (1981). [wh Ni [NPj S]] => [[wh N, NP,] S]] Please note that the Absorption Cheirchia adopts for the pair list answer is in a sense quantifying into Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 c. ?*Twu salam isang-i encey tampay-lul phiwess-ni? two m an more-NOM when cigarette-ACC smoked-QM ‘W hen did more than two men smoke?’ d. ?*Twu salam isang-i ettehkey wass-ni? two m an more-NOM how came-QM ‘W hat is the means x such that more than two men came by x?’ e. ?*Twu salam isang-i M ary-eytayhay ettehkey malhayss-ni? two man more-NOM M -about how talk-QM ‘W hat is the manner x such that more than two men talked about M ary in x?’ (98) a.?*Se salam miman-i nw ukwu-lul chotayhayss-ni? three man less-NOM whom-ACC invited-QM ‘W ho did less than three men invite?’ b.?*Se salam miman-i eti-se tampay-lul phiwess-ni? three man less -NOM where-at cigarette-ACC smoked-QM ‘W here did less than three men smoke?’ c.?*Se salam miman-i encey tampay-lul phiwess-ni? three man less -NOM when cigarette-ACC smoked-QM ‘W hen did less than three men smoke?’ d.?*Se salam miman-i ettehkey wass-ni? three man less-NOM how came-QM ‘W hat is the means x such that less than three men come by x?’ questions, although he claims that quantifying into questions cannot be the source of a pair list answer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 e.?*Se salam miman-i M ary-eytayhay ettehkey malhayss-ni? three man less-NOM M -about how talked-QM ‘W hat is the manner x such that less than three men talked about Mary in x ?’ (99) a.?*Ttak twu salam-i nwukwu-lul chotayhayss-ni? exactly two man-NOM whom-ACC invited-QM ‘W ho did exactly two men invite?’ b.?*Ttak twu salam-i eti-se tampay-lul phiwess-ni? exactly two man-NOM where-at cigarette-ACC smoked-QM ‘W here did exactly two men smoked?’ c.?*Ttak twu salam-i eti-se tampay-lul phiwess-ni? exactly two man-NOM where-at cigarette-ACC smoked-QM ‘W here did exactly two men smoked?’ d. ?*Ttak twu salam-i ettehkey wass-ni? exactly two man-NOM how came-QM ‘W hat is the means x such that exactly two men came by x?’ e. ?*Ttak twu salam-i M ary-eytayhay ettehkey malhayss-ni? exactly two man-NOM M -about how talk-QM ‘W hat is the manner x such that exactly two men talked about Mary in x?’ As noted throughout this chapter, the sentences in (97-99) are marginal in terms of grammaticality, admitting only a functional answer, sharply contrasting with the sentences in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 in (40=65=78) with the same modified subject QP followed by a propositional adjunct wh- word, which admit a single answer, as repeated in (100) (100) a. Twu salam isang-i Tom-ul w ay chotayhayss-ni? two man more-NOM T-ACC why invited-QM ‘W hy did more than two men invite Tom ?’ b. ?Se salam miman-i Tom-ul way chotayhayss-ni? three man less-NOM T-ACC why invited-QM ‘W hy did less than three men invite Tom ?’ c. Takk twu salam-i Tom-ul way chotayhayss-ni? exactly two man T-ACC why invited-QM ‘Why did exactly two men invite Tom ?’ Thus the propositional adjunct wh-word in (100) can take scope over the subject QP, in contrast to the nonpropositional wh-words in (97-99). W hat is the source o f this asymmetry o f scope between the two types o f wh-words regarding the wide scope behavior o f the wh- word over the QP, to begin with? At this point the reader should be reminded o f our claim in section 3 that in Korean nonpropositional wh-words are indefinites in the sense o f Lewis (1975) and Heim (1982), while the propositional adjunct wh-word way (why) is not and that this difference is directly associated with an asymmetry in the way they take scope in wh-questions. To be specific, following the intuitions in the literature (Chomsky 1964:38, Katz and Postal 1964:93, Klima Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 1964:253, K uroda 1969: 266ff, and Vergnaud and Zubizarreta 2002 among others), according to which a wh-word in English is viewed as a combination o f wh-operator and some (one, thing...), I suggested that QM in Korean corresponds to wh-operator in English, unselectively binding and thus marking the scope o f the nonpropositional wh-word at LF, which corresponds to some (one, thing...) in English. Meanwhile, I suggested that since the propositional adjunct wh-word is not an indefinite it should undergo movement into Spec o f CP at LF in order to be properly interpreted by forming an operator variable chain and entering into spec-head agreement with the QM. Thus QM in (100) as opposed to the one in (97-99) does not behave as a wh-operator, but just as a position holder such that the propositional adjunct wh-word checks o ff its wh-feature, marking its scope therby at LF. Let us see how our proposal for the asymmetric scope taking for the two types o f wh-words accounts for the scope fact in (97-100). Consider the examples in (97-99) with a nonpropositional wh-word, first. Since the QM marks the scope o f nonpropositional wh- words in Korean in the current proposal, the position o f the QM at LF bears a crucial importance in accounting for the QP and wh-word scope interaction in Korean. I will first try to locate where the QM is base-generated. The importance o f the QM as the scope marker for in-situ wh-words has seldom been discussed in Korean literature. It seems that Kim (1990) investigated the question o f where the QM is base-generated for the first time, although for a different purpose. Kim (1990:227ff) claims that the QM in Korean is base-generated in the head o f IP, mainly based on the fact that QM does not show a complementary distribution with the quotative marker, ko (that), which projects CP as shown in (101). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 (101) John-un M ary-ka mwus-ul sat-nya-ko muletta. J-TOP M -NOM what-ACC bought-QM -COM P asked ‘John asked what M ary bought.’ (Kim 1990:227) It should be also noted that the QM is not a frre standing morpheme nor can it be deleted, either in the embedded or in the matrix clause, which means that it is part o f the verbal morphology as shown in (102) (also in (101)). Korean verbal morphology including the QM is heavily agglutinating as shown in (102), consisting o f more than one bound morpheme. (102) a. John-i nwukw u-lul manna-ss-*(ni)? J-NOM whom-ACC meet-PAST-(QM) ‘W ho did John m eet?’ b. Na-nun [John-i nwukwu-lul manna-ss-nun-*(ci)] alko sipta I-TOP J-NOM whom-ACC meet-PAST-(QM ) want to know ‘I want to know who John m et.’ Given the recent proposals (Pollock 1989, Chomsky 1993, among others) viewing the IP system as an extension o f the VP system, with each inflectional morpheme o f the verb projecting a separate functional projection o f the IP system, it follows that the QM, which is part o f the verbal morphology, should not project in the CP domain in the sense o f Rizzi (2000). Since Korean is a strictly head-final language, the particular sequence o f the bound morphemes on the verb in (101-102) including the QM thus leads one to postulate the following clausal structure in (102), where the past tense morpheme and question morpheme Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 project TP and IP respectively, right on top o f VP under the strict projectionist hypothesis (Pollock 1989 and Chomsky 1993, also see Benmamoun 1992:8ff), according to which each bound morpheme o f the verb projects its own functional projection in the IP field: (103) [ ip [tp [v p v ] t] i] Thus based on K im ’s proposal mainly drawn from the lack of complimentary distribution of the QM and the quotative marker ko (that) in Korean and our discussion on the Korean verbal morphology, I will conclude that QM indeed projects IP .2 8 Now the LF structure for these sentences in (97-99) will be the following in (104) in the current system, depending on whether the QP or the QM binds the wh-word, (suppressing irrelevant d etails):2 9 2 8 The quoatative marker ko is an abbreviation of the free standing morpheme hako that is a complimentizer (see Suh 1994:1216). Although ko is apparently a bound morpheme on the verb, it is not part of the inflectional morphology on the verb since it is not obligatory as shown below. John-un [C p Mary-ka mwues-ul sat-nya-?(ko)] mwullet-ta, J-TOP M-NOM what-ACC bought-QM asked ‘John asked what Mary bought.’ 2 9 I suggest that nonpropositional wh-words in Korean in contrast to the propositional adjunct wh- word have a DP structure. This is mainly based on the following paradigm where the former can combine with the referential expression ku (the), which presumably is sitting in the Spec of DP (also see the paradigm in Chung 1996:251): (i) a. ku nwukwu ‘the who’ b. ku mwues ‘the what’ c. ku enu N ‘the which N’ d. ku eti ‘the where’ e. ku encey ‘the when’ f. ku elnia ‘the how much’ g. ku ettehkey ‘the how’ h. *ku way ‘the why’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 (104) a. [IP modified QP-NOM [r QM, [V i > [ d p [ np WH-word,]]]]] b. [ip modified QPj-NOM [r QM, [V P [Dp [ n p WH-wordj ]],]]] Consider the LF representation in (104a) for (97-99), first. The lack o f a single answer follows straightforwardly. The QM in the head o f IP binds the indefinite wh-word, thus the indefinite wh-word is interpreted as a wh-interrogative. The QM, however, fails to c- command the subject QP in Spec o f IP, thus cannot take scope over it, hence no single answer. Please recall that I am taking the position that only A ’-movement reconstructs, following Chomsky (1995) and Lasnik (1999). Thus the subject QP in Spec IP does not reconstruct into Spec o f VP such that it is c-commanded by the QM in the head o f IP at LF. Pair list answer is impossible, since the indefinite wh-word is bound by the QM. At this point please recall our proposal that pair list answer obtains when a definite QP binds the indefinite wh-word. Now consider the other LF representation in (104b) for (97-99). Single answer does not obtain. The QM in the head o f IP does not bind the indefinite wh-word, thus the indefinite wh-word is not interpreted as an interrogative. Pair list answer does not obtain, either: The modified QP that is not definite binds indefinite wh-word so the indefinite can never have a definite quantificational force. Thus the hearer o f these questions in (97-99) can never identify the individuals (measure, means, manner) as replaced by the indefinite. Thus no pair list answer obtains. Also see Watanabe (1992) for DP analysis of wh-words. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 After all, it is a standard observation in the literature that modified QPs in simplex wh- questions as in (105) do not induce a pair list answer (Groenendijk and Stokhof 1984: 181, Chierchia 1991: 81, Liu 1997 among others).3 0 (105) a. W ho did more than two students invite? b. Who did exactly three students invite? c. Who did less than three students invite? The grammaticality o f those sentences in (97-99) thus receives a straightforward account, since neither o f the two LF representations in (104) for these sentences yields a single answer or a pair list answer, given the current proposal in (13), repeated in (106), according to which a wh-question should be answerable at least by either a pair list answer or a single answ er.3 1 (106) W h-questions should be answerable with at least a pair list answer or a single answer. I will come back to the grammatical status o f wh-questions, which admit only a functional answer in section 6. The proposal in (106) above may sound rather absurd when one considers wh-questions in English in (93) and (105) repeated as (107-108). 3 0 See Engdahl (1980) and Higginbotham and May (1981), Groenendijk and Stokhof (1984), and Chierchia (1991,1993) among others, for the discussion of the pair list answer. 3 1 It should be noted that I am not claiming this is the only criterion relevant to the acceptability of a given wh-question. There is another independent criterion, which is purely syntactic. This will be the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 (107) a. Who did everyone invite? b. Who did John and Bill invite? c. W ho did the two men invite? d. Who did most o f the men invite? (108) a. W ho did more than two students invite? b. Who did exactly three students invite? c. Who did less than three students invite? The questions in (107) admit a pair list answer, a single answer and a functional answer, while the ones in (108) admit a single answer and a functional answer. However, when it comes wh-questions in Korean where wh-questions with a modified subject QP as in (97-99) are marginal , only admitting a functional answer, the proposal in (106) merits a serious consideration. Next let us turn to the wh-questions in (100) with a propositional adjunct wh- word. The LF structure for them will be the one in (109) under the present analysis. (109) [c p H w y , [c QM, [IP modified QP-NOM [r t, [V P 1 j ]]]]] Please note the QM in (109) is raised into the head o f CP from its base-generated position in the head o f IP at LF, driven by the need to enter into a spec-head agreement with the propositional adjunct wh-word in Spec of CP at LF. In (109), pair list answer is not a possibility to begin with, given that propositional adjunct wh-word way (why) cannot be decomposed into wh-operator and indefinite some part at LF. The adjunct wh-word way topic of our discussion in Chapters 4 and 5. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 (why) in Spec o f CP, however, c-commands the subject QP, thus the former can take scope over the latter, inducing a single answer, and the sentence is thus grammatical, given the current proposal in (106). W ith the asymmetric scope taking between the propositional adjunct wh-word and the nonpropositional wh-word in (97-99) and (100) in mind, let us turn to simplex wh-questions with definite QPs in (55-58) and (45=82=100), repeated in (110- 113) and (114), respectively. (110) a. M otun salam-i nwukwu-lul chotayhayss-ni? every man-NOM whom-ACC invited-QM ‘W ho did everyone invite?’ b. M otun salam-i eti-se tampay-lul phiwess-ni? every man-NOM where-at cigarette-ACC smoked-QM ‘W here did everyone smoke?’ c. M otun salam-i encey tampay-lul phiwess-ni? every man-NOM when cigarette-ACC smoked-QM ‘W hen did everyone smoke?’ d. M otun salam-i ettehkey wass-ni? every man-NOM how came-QM ‘W hat is the means x such that everyone came by x?’ e. M otun salam-i M ary-eytayhay ettehkey malhayss-ni? every man-NOM M -about how talked-QM ‘W hat is the manner x such that everyone talked about M ary in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Ill) a. Ku twu salam-i nw ukwu-lul chotayhayss-ni? the two man-NOM whom-ACC invited-QM ‘W ho did the two m en invite?’ b. Ku twu salam-i eti-se tampay-lul phiwess-ni? the two man-NOM where-at cigarette-ACC smoked-QM ‘W here did the two men smoke?’ c. Ku twu salam-i encey tampay-lul phiwess-ni? the two man-NOM when cigarette-ACC smoked-QM ‘W hen did the two men smoke?’ d. Ku twu salam-i ettehkey wass-ni? the two man-NOM how came-QM ‘W hat is the means x such that the two men came by x?’ e. Ku twu salam-i M ary-eytayhay ettehkey malhayss-ni? the two man-NOM M -about how talked-QM ‘W hat is the manner x such that the two men talked about M ary in x?’ (112) a. John-kwa M ary-ka nwukw u-lul chotayhayss-ni? J-and M -NOM whom-ACC invited-QM ‘W ho did John and M ary invite?’ b. J-kwa M ary-ka eti-se tampay-lul phiwess-in? J-and M -NOM where-at cigarette-ACC smoked-QM ‘W here did John and M ary smoke?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 c. John-kwa M ary-ka encey tampay-lul phiwess-in? Jo-and M -NOM when cigarette-ACC smoked-QM ‘W hen did John and M ary smoke?’ d. John-kwa M ary-ka ettehkey wass-ni? J-and M -NOM how came-QM ‘W hat is the means such that John and M ary came by x?’ e. John-kwa M ary-ka Tom-eytayhay ettehkey malhayss-ni? J-and M -NOM T-about how talk-QM ‘W hat is the manner x such that John and M ary talked about Tom in x?’ (113) a. Taypwupwun-uy salam-i nwukwu-lul chotayhayss-ni? most-POSS man-NOM whom-ACC invited-QM ‘Who did most o f the men invite?’ b. Taypwupwun-uy salam-i eti-se tampay-lul phiwess-ni? most-POSS man-NOM where-at cigarette-ACC smoked-QM ‘W here did most o f the men smoke?’ c. Taypwupwun-uy salam-i encey tampay-lul phiwess-ni? most-POSS man-NOM when cigarette-ACC smoked-QM ‘W hen did most o f the men smoke?’ d. Taypwupwun-uy salam-i ettehkey wass-ni? most-POSS man-NOM how came-QM ‘W hat is the means x such that most o f the men came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 e. Taypwupwun-uy salam-i M ary-eytayhay ettehkey malhayss-ni? most-POSS man-NOM M -about how talked-QM ‘W hat is the manner x such that most o f the men talked about Mary in x?’ (114) a. M otun salam-i Tom-ul way chotayhayss-ni? every man-NOM T-ACC why invited-QM ‘W hy did every one invite Tom ?’ b. Ku twu salam-i Tom-ul way chotayhayss-ni? the two man-NOM T-ACC why invited-QM ‘W hy did the two men invite Tom ?’ c. John-kwa M ary-ka Tom-ul way chotayhayss-ni? J-and M -NOM T-ACC why invited-QM ‘W hy did John and M ary invite Tom ?’ d. Taypwupwun-uy salam-i Tom-ul way chotayhayss-ni? most-POSS man-NOM T-ACC why invited-QM ‘W hy did most o f the men invite Tom ?’ As mentioned before, the simplex questions in (110-114) are grammatical and the questions in (110-113) admit a pair list answer, a single answer, and a functional answer, while the questions in (114) admit only a single answer, as summarized previously in (49-50). Two questions immediately arise with respect to the scope facts in (110-114). One is what the source o f the single answer in (110-113) is. Given the asymmetry o f scope taking Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 between the propositional adjunct wh-word and nonpropositional wh-words in simplex wh- questions as attested in simplex wh-questions with a modified subject QP in (97-99) and (100), I will crucially suggest that the single answer in simplex questions with a nonpropositional wh-word in (110-113) is actually a special case o f the pair list answer, contrary to the one in simplex questions with a propositional adjunct wh-word in (114), which involves wide scope o f the propositional adjunct wh-word over the QP. The other question is why the question in (114) with the propositional adjunct wh-word does not have a pair list answer, in contrast to the ones in (110-113). Let us turn to the first question. Since I suggested that the single answer for the questions in (110-113) is a special case o f a pair list answer, let us first discuss how the pair list answer in these questions is derived. Consider the LF structure for the sentences above in (110-113) that will be the following in (115) in the current system, (suppressing irrelevant details), depending on whether the QM or the QP binds the indefinite wh-word: (115) a. [IP QP-NOM [r QM, [V P [D P [N P WH-word,]]]]] b. [IP QPj-NOM [r QM; [V P [ dp [n p WH-word,]], ]]] In (115a), the QM binds the indefinite and thus the indefinite is construed as a wh- interrogative. The QM, however, cannot c-command and thus cannot take scope over the subject QP, hence no single answer. Pair list answer does not obtain either, since the indefinite wh-word is bound by the wh-operator not by the definite QP. Consider now the other LF representation in (1 15b). Single answer is not a possibility, since the QM does not bind the indefinite wh-word. How about pair list answer? The definite QP can c-command Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 and thus bind the indefinite wh-word, which renders the indefinite to become a definite such that the hearer upon hearing questions in (110-113) can identify the individuals (and measure, means, manner) as replaced by the indefinite, enabling him to give an answer by listing pairs. This accounts for the pair list answer for the sentences in (110-113), since one o f the two possible LF representations for these sentences yields a pair list answer. Now let us turn to the nature o f the single answer to wh-questions in (110-113). As we saw, none o f the LF representations in (115) for these sentences yield a single answer, but apparently these sentences can have it, as one can answer the sentence in (110a), for example, w ith (116). (116) M otun salam-i Bomin-ul chotayhayssta. every man-NOM B-ACC invited ‘Everyone invited Bom in’ Why is a single answer available in wh-questions in (110-113), given that the QM binding the indefinite wh-word cannot c-command the subject QP in the LF representations in (1 15a) and that c-command at LF is a necessary condition for determining the relative scope of scope bearing elements? One might propose the possibility o f QM -raising into the head o f CP in (115a). I will not opt for this, however, given that nothing is apparently forcing the QM to be raised into the head o f CP or any position, thus it should not be raised, given the notion o f economy in the sense o f Chomsky (1995). I will discuss instances where the QM is raised into the head o f CP later in this section. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 In fact we have a good reason to suspect that the single answer in (110-113) is not the wide scope of the wh-word but rather a special case o f the pair list answer, given the asymmetry o f scope between the two types o f wh-words as discussed in simplex wh-questions with a modified QP in (97-100). There it was shown that the propositional adjunct wh-word can take scope over the subject QP in simplex wh-questions while the nonpropositional wh-word cannot do so. Although traditionally the single answer has been viewed as signaling a wide scope construal o f the wh-word with respect to the QP in the subject position as we saw in (86) (M ay 1985, W illiams 1986, 1988 among others), this position is not conclusive: One may view the single answer as in (116) as the wide scope o f the wh-word over the subject QP as in (1 10a), but this is not clearly warranted, since the single answer in this case always entails the pair list answer. In other words, whenever the answer in (116) is a true answer to the question in (110a) under the wide scope reading o f the wh-word over the subject QP, it will also be a true answer under the wide scope reading o f the subject QP over the wh-word (also see Liu 1997: 143 for the related discussion). Alternatively, the single answer in (116) may be construed as describing an occasion where every individual accidentally invited the same individual whose name is Bomin. In other words, the answer in (116) for the question in (1 10a) may be viewed as a conjunction of three separate statements as in (117) supposing that Tom, Mary and Jack exhaust the set of men in the discourse context. (117) Tom invited Bomin, M ary invited Bomin, and Jack invited Bomin. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 I will thus claim the single answer in (116) actually is a special case o f a pair list answer to the question in (110a), for example, based on the discussion o f the asymmetry o f scope between the propositional adjunct wh-word and the nonpropositional wh-word as in (97- 100). To further strengthen our claim, let us consider the scope interaction between a nonpropositional wh-word and a subject QP as in (1 13a) with most NP, repeated in (118). (118) Taypwupwun-uy salam-i nwukwu-lul chotayhayss-ni? most-POSS man-NOM whom-ACC invited-QM ‘Who did most o f the men invite?’ The subject QP taypwupwunuy salam (most men) is free from the entailment problem inherent with the universal quantifier and thus enables us to decide the right truth condition for it. Let us suppose a context involving seven individuals A, B, C, D, E, F, and G. Let us further suppose that individuals A, B, C, D invited Prof. John and individuals C, D, E, F invited Prof. Mary. Given the above context, suppose the speaker asks the question in (118). The hearer o f the question, knowing precisely who invited who, cannot answer by saying ‘Prof. John and Prof. M ary’. 3 2 This means the set denoted by most men can never co-vary with the choice o f a professor, which means the wh-word in (118) cannot take scope over the subject QP. To the extent that the answer ‘Prof. John and Prof. M ary’ is a possibility to the above question in (118) in Korean, the context must be such that a majority set o f individuals, for example A, B, C, D invited both the two professors, as depicted in (119). 3 2 Barry Schein (personal communication) notes that this answer is possible in English. Paul Hagstrom (personal communication) also agrees with Barry Schein but notes that ‘Prof. John or Prof. Mary’ is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 Prof. John (119) A Prof. Mary Thus one can reach the conclusion that the apparent single answer to the questions in (110- 113) is rather a special case o f a pair list answer. This answers the question why a single answer is available for these sentences, although the QM in the LF in (115) for these sentences cannot c-command the subject QP, and as w e are assuming that c-command at LF is a necessary condition for determining the relative scope o f scope bearing elements. Now let us turn to the single answer to wh-questions in (114). I suggested that the single answer in this case is actually derived from wide scope o f the propositional adjunct wh- word over the subject QP, given the availability o f wide scope for the adjunct wh-word as shown in wh-questions in (99). Consider the LF representation for the sentences in (114): (120)[CpHwyj[c-QMi[n> QP-NOM [r tj [w tj ]]]]] From the above LF representation, the wide scope o f the propositional adjunct wh-word over the subject QP follows, since the former, in Spec o f CP at LF, can c-command the latter. For more natural in the above context. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 the skeptical mind, suppose as a relevant context the department orientation for the incoming class of graduate students A, B, C, D, E, F, and G among whom students A, B, C, D visited Prof. John and students C, D, E, F visited Prof. Mary. Given the above context, the speaker can ask the question in (121), and the hearer can answer by saying ‘because they are the most prominent and influential professors in the departm ent’ as a single answer. (121) Taypwupwun-uy haksayng-i w ay John-kwa Mary-lul pangmwunhayss-ni? most-POSS student-NOM why J-and M-ACC visited-QM ‘Why did most o f the students visit John and M ary?’ Now let us turn to the second question. Why is it that the sentences in (114) do not have a pair list answer, contra the sentences in (110-113) with the same definite subject QP? Given the standard observation in the literature that definite QPs can induce a pair list answer, the lack o f it in (114) may be rather unexpected. In what follows, I will try to give an account for this fact, which will make crucial reference to the nature o f the propositional adjunct wh- word. Please recall that pair list answer can obtain when the definite QP c-commands and binds the indefinite some part o f a wh-word in our analysis. Since the propositional adjunct wh-word is not an indefinite, it cannot be decomposed into wh-operator and indefinite some part at LF. Thus although the definite QP in the LF representation (120) for the sentences in (114) c-commands the trace o f the propositional adjunct wh-word, no pair list answer can obtain. Note that it is also the case that corresponding sentences in English do not readily induce a pair list answer.3 3 3 3 It seems that some native speakers still find an example like the one as in (122a) ambiguous. For this, see Aoun and Li (1993a: 151 ff). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 (122) a. W hy did everyone invite John? b. Why did John and M ary invite Tom? c. W hy did the two men invite John? d. W hy did most o f the men invite John? The lack o f a pair list answer in English in (122) also follows, once our proposal that the propositional adjunct wh-word in Korean cannot be an indefinite so that it cannot be decomposed into wh-operator and indefinite some part is extended to English propositional adjunct wh-words. To summarize, I tried to show that the scope facts listed in (49-50) can be best accounted for, by recognizing an asymmetry o f movement between the two types o f wh-words to the effect that the QM marks the scope o f nonpropositional wh-words in-situ, while the propositional adjunct wh-word undergoes movement into and takes scope from Spec o f CP at LF. 4. 2. Symmetry of Scope Suppose now a situation where the QM that marks the scope o f nonpropositional wh-word is also raised from its base-generated position into a position or is base-generated in a position from which it can c-command the subject QP, like the propositional adjunct wh-word at LF. Then what is expected is symmetry o f scope between the two types o f wh-words when it Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 comes to QP and wh-word scope interaction. Good candidates for this situation are the so- called ‘scram bled’ wh-question, and complex questions with an embedded subject QP and wh-words. 4. 2.1. Scrambled Wh-question The asymmetry o f scope between the propositional adjunct wh-word and the nonpropositional wh-word in simplex questions with modified subject QPs as observed in (97-99) on the one hand and (100) on the other, disappears when the word order between the QP and the wh-word is reversed as in (123-126). (123) a. Nwukwu-lul twu salam isang-i chotayhayss-ni? whom-ACC two man more-NOM invited-QM ‘Who did more than two men invite?’ b. Eti-se twu salam isang-i tampay-lul phiwess-ni? where-at two man more-NOM cigarette-ACC smoked-QM ‘W here did more than two men smoke?’ c. Encey twu salam isang -i tampay-lul phiwess-ni? when two men more-NOM cigarette-ACC smoked-QM ‘W hen did more than two men smoke?’ d. Ettehkey twu salam isang-i wass-ni? how two man more-NOM came-QM ‘W hat is the means x such that more than two men came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. e. Ettehkey twu salam isang-i M ary-eytayhay malhayss-ni? How two man more-NOM M -about talked-QM ‘W hat is the manner x such that more than two men talked about M ary in x? (124) a. Nwukwu-lul se salam m im an-i chotayhayss-ni? whom-ACC three man less -NOM invited-QM ‘Who did less than three men invite?’ b. Eti-se se salam miman-i tampay-lul phiwess-ni? where-at three man less-NOM cigarette-ACC smoked-QM ‘W here did less than three m en smoke?’ c. Encey se salam miman-i tampay-lul phiwess-ni? when three man less-NOM cigarette-ACC smoked-QM ‘W hen did less than three men smoke?’ d. Ettehkey se salam miman-i wass-ni? how more three man less-NOM came-QM ‘W hat is the means x such that less than three men came by x?’ e. Ettehkey se salam miman-i M ary-eytayhay malhayss-ni? How three man less-NOM M -about talked-QM ‘W hat is the manner x such that less than three men talked about M ary in x? (125) a. Nwukwu-lul ttak twu salam-i chotayhayss-ni? whom-ACC exactly two men-NOM invited-QM ‘W ho did more than two men invite?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I l l b. Eti-se ttak twu salam-i tampay-lul phiwess-ni? where-at exactly two man -NOM cigarette-ACC smoked-QM ‘W here did exactly two men smoke?’ c. Encey ttak twu salam-i tampay-lul phiwess-ni? when exactly two man -NOM cigarette-ACC smoked-QM ‘W hen did exactly two men smoke?’ d. Ettehkey ttak twu salam-i wass-ni? how exactly two man -NOM came-QM ‘W hat is the means x such that more than two men came by x?’ e. Ettehkey ttak twu salam-i M ary-eytayhay malhayss-ni? How exactly two man-NOM M -about talked-QM ‘W hat is the manner x such that exactly two men talked about M ary in x?’ (126) a. Way twu salam isang-i Tom-ul chotayhayss-ni? why two men more-NOM T-ACC invited-QM ‘Why did more than two men invite Tom ?’ b. Way se salam miman-i Tom-ul why three man less-NOM T-ACC ‘Why did less than three men invite Tom ?’ c. W ay ttak twu salam-i Tom-ul chotayhayss-ni? why exactly two man-NOM T-ACC invited-QM ‘Why did exactly two men invite Tom ?’ chotayhayss-ni? invited-QM Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 I will call these questions ‘scrambled’ simplex wh-questions for expository purpose. The sentences in (123-126) are all grammatical, admitting only a single answer. W hen it comes to ‘scrambled’ simplex wh-questions with a definite QP followed by a wh-word as in (127- 131), the same scope facts hold: These sentences admit only a single answer. (127) a. Nwukw u-lul motun salam-i whom-ACC every man-NOM ‘Who did everyone invite?’ b. Eti-se motun salam-i where-at every man-NOM ‘W here did everyone smoke?’ c. Encey motun salam-i when every man-NOM ‘W hen did everyone smoke?’ d. Ettehkey motun salam-i how every man-NOM ‘W hat is the means x such that everyone came by x?’ e. Ettehkey motun salam-i M ary-eytayhay malhayss-ni? how every man-NOM M -about talked-QM ‘W hat is the manner x such that everyone talked about Mary in x?’ chotayhayss-ni? invited-QM tampay-lul phiwess-ni? cigarette smoked-QM tampay-lul phiwess-ni? cigarette smoked-QM wass-ni? came-QM Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 (128) a. Nwukwu-lul ku twu salam-i chotayhayss-ni? whom-ACC the two man-NOM invited-QM ‘Who did the two men invite?’ b. Eti-se ku twu salam-i tampay-lul phiwess-ni? where-at the two man-NOM cigarette smoked-QM ‘W here did the two men smoke?’ c. Encey ku twu salam-i tampay-lul phiwess-ni? when the two man-NOM cigarette smoked-QM ‘W hen did the two men smoke?’ d. Ettehkey ku twu salam-i wass-ni? how the two man-NOM came-QM ‘W hat is the means x such that the two men came by x?’ e. Ettehkey ku twu salam-i M ary-eytayhay malhayss-ni? How the two man-NOM M -about talked-QM ‘W hat is the manner x such that the two men talked about M ary in x?’ (129) a. Nwukwu-lul John-kwa M ary-ka chotayhayss-ni? whom-ACC J-and M -NOM invited-QM ‘Who did John and M ary invite?’ b. Eti -se John-kwa M ary-ka tampay-lul phiwess-ni? where-at J-and M -NOM cigarette smoked-QM ‘W here did John and M ary smoke?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 c. Encey John-kwa M ary-ka tampay-lul phiwess-ni? when J-and M -NOM cigarette smoked-QM ‘W hen did John and M ary smoke?’ d. Ettehkey John-kwa M ary-ka wass-ni? how J-and M -NOM came-QM ‘W hat is the means x such that John and M ary came by x ?’ e. Ettehkey John-kwa Barry-ka M ary-eytayhay malhayss-ni? how J-and B-NOM M -about talked-QM ‘W hat is the manner x such that John and Barry talked about M ary in x?’ (130) a. Nwukwu-lul taypwupwun-uy salam -i chotayhayss-ni? whom-ACC most-POSS men-NOM invited-QM ‘W ho did most o f the men invite?’ b. Eti-se taypwupwun-uy salam -i tampay-lul phiwess-ni? where-at most-POSS men-NOM cigarette-ACC smoked-QM ‘W here did most o f the men smoke?’ c. Encey taypwupwun-uy salam -i tampay-lul phiwess-ni? when most-POSS men-NOM cigarette-ACC smoked-QM ‘W hen did most o f the men smoke?’ d. Ettehkey taypwupwun-uy salam -i wass-ni? how most-POSS men-NOM came-QM ‘W hat is the means x such that most o f the men came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 e. Ettehkey taypwupwun-uy salam -i M ary-eytayhay malhayss-ni? How most-POSS men-NOM M -about talked-QM ‘W hat is the manner x such that most o f the men talked about Mary in x?’ (131) a. Way motun salam-i Tom-ul chotayhayss-ni? why every man-NOM T-ACC invited-QM ‘W hy did everyone invite Tom ?’ b. Way ku twu salam-i Tom-ul chotayhayss-ni? why the two man-NOM T-ACC invited-QM ‘W hy did the two men invite Tom ?’ c. Way John-kwa M ary-ka Tom-ul chotayhayss-ni? why J-and M -NOM T-ACC invited-QM ‘W hy did John and M ary invite Tom ?’ d. Way taypwupwun-uy salam-i Tom-ul chotayhayss-ni? why most man-NOM T-ACC invited-QM ‘Why did most o f the men invite Tom ?’ The scope facts for ‘scrambled’ simplex wh-questions in (123-131) are summarized in (132- 133) below. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 (132) Nonpropositional w h-w ord.. .QP-NOM .. .V PL s F QP- NOM Definite QP a. motwun salam ‘everyone’ b. John-kwa M ary ‘John and M ary’ c. ku twu salam ‘the two m en’ d. taypwupun-uy salam ‘most o f the m en’ * OK * QP- NOM M odified QP a. twu salam isang ‘more than two m en’ b. ttak twu salam ‘exactly two m en’ c. se salam miman ‘less than three m en’ * OK * (133) W H Y ....Q P-N O M ... V PL S F QP- NOM (1) Definite QP a. motwun salam ‘everyone’ b. John-kwa M ary ‘John and M ary’ c. ku twu salam ‘the two m en’ d. taypwupwun-uy salam ‘most o f the m en’ * OK * QP- NOM (2) M odified QP a. twu salam isang ‘more than two m en’ b. ttak twu salam ‘exactly two m en’ c. se salam miman ‘less than three m en’ * OK * Simplex wh-questions in (123-131) was claimed to involve optional wh-fronting by some linguists (Choe 1994, among others). I, however, suggest that both the nonpropositional wh- words and the propositional adjunct wh-words are base-generated where they appear at S structure in these wh-questions, the supporting arguments for which I will present in section 4. 2.1.1 and 4.2.1.2. 3 4 I suggest that the position these fronted wh-words are taking at S structure is IP-adjoined position (cf. Chomsky 1977). I thus would like to attribute the symmetry o f scope between the two types o f wh-words as summarized in (132-133) to this very structural symmetry o f the two types of wh-words: both of them are base-generated in 3 4 It should be also mentioned that essentially the same idea has been proposed in the literature (Aoun and Li 1993: 181, Cheng 1997: 29, Ueyama 1998 among others). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 in IP-adjoined position, o f course, assuming a null operator movement for the proper interpretation o f the wh-words in sentence initial position. Given the current proposal for the base-generation o f wh-words in ‘scram bled’ wh-questions in IP-adjoined position as in (123-131), what follows is that the QM in the head o f IP should be raised at LF into a position, i.e., the head o f CP from which it can bind the nonpropositional wh-word in IP adjoined-position such that the indefinite wh-word can be properly interpreted. Likewise, the propositional adjunct wh-word way (why) in IP-adjoined position should be raised into Spec o f CP at LF to be properly interpreted by forming an operator variable chain, since it is not an indefinite. Thus the LF representations for the sentences in (123-131) will be the following in (134) under our analysis (irrelevant detail suppressed): (134) a. [C p [c’ QMj [n > [D p[N pWH-wordj ]] [IP QP-NOM [r t, [ V p ]]]]]] b. [ c p way$ [ e QM; [ip tj [ip QP-NOM [ p t; [ V p ]]]]]] The LF representation in (134a) is for ‘scram bled’ questions with nonpropositional wh- words in (123-125) and (127-130) while the LF representation in (134b) is for scrambled questions with the propositional adjunct wh-words in (126) and (131). The QM in the head o f CP unselectively binds the nonpropositional wh-word in IP adjoined position in (134a), thus it is interpreted as a wh-interrogative, while the propositional adjunct wh-word way (why) in Spec o f CP binds its trace in IP-adjoined position in (134b). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 Given the LF structures in (134), the wide scope o f the wh-word over the subject QP in the sentences in (123-131) simply follows, since the QM, which binds the indefinite wh-word c- commands the subject QP, and the propositional adjunct wh-word c-commands the QP, respectively. The lack o f a pair list answer in (134a) also follows, since the indefinite wh- word in IP-adjoined in (134a) is not bound by the QP. 3 5 Pair list answer in (134b) is simply out o f the question, given that the propositional adjunct wh-word cannot be an indefinite. To summarize, the scope symmetry between the two types o f wh-words in ‘scram bled’ simplex wh-questions as summarized in (132-133) follows from the fact that both are base generated in IP-adjoined position. Before closing this section, I will review the uniform LF wh-movement approach by Kim (1990) and the uniform wh-in-situ approach in Aoun and Li 1993b), seeing how the scope facts in (132-133) can be accounted with them. When it comes to Kim ’s (1990) uniform wh-movement approach on the ‘scram bled’ wh-questions, Kim may account for the lack o f a pair list answer in (123-126) with some additional stipulation on the QP types such that the pair list answer will be blocked for modified subject QPs. The real problem for Kim, however, is that his system may predict the sentences in (127-131) to be ambiguous, since he takes the position that the wh-word in sentence-initial position is in Spec o f CP position at S structure via topicalization. Given the S structure representation in (135) for the sentences in (127-131), the chain scope o f the fronted wh-word, {IP, VP} fails to dominate the subject QP, hence predicting ambiguity given the rest o f his proposal. The reader should be reminded o f his proposal given in section 4.1.1.1, which says that a given 3 5 Suppose a subject QP, i.e., a definite QP as in (134a) can be further raised into a position where it can bind the indefinite wh-word in IP-adjoined such that pair list answer can obtain. This movement Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 sentence is ambiguous when the chain scope o f one QP fails to dominate the chain scope of the other. (135) [C P WH-word; [IP QP-NOM [V P f ]]] Kim recognizes this as a problem in his system, arguing that some presupposition is implicated in the lack o f scope ambiguity in the sentences as in (127) (Kim 1990: 254). Next, let us turn to the uniform wh-in-situ approach o f Aoun and Li (1993b). W hen it comes to ‘scram bled’ wh-questions as in (123-126), the lack o f a pair list answer may be accounted for in their system by assuming c-command as a necessary condition such that a modified QP cannot take scope over the wh-word. The real problem is that when extended to the sentences as in (127-131), Aoun and Li (1993b) would predict them to be ambiguous, since the LF structures for them will be as in (136): (136) [cp QU j [ip WH-wordj [iP QP-NOM; [ip ti [r [q uP tj [vptj ]]]]]]] The above LF structure does not violate the Minimal Binding Requirement in (84) as repeated in (137), since although the subject QP is the closer A ’-binder for tj, it is not a potential binder, since the co-indexing o f tj (trace o f QU) with the subject QP will make it A- bound by t;, leading to condition C violation. of the subject QP for a pair list answer will trigger weakcross over violation and will be ruled out. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 (137) The Minimal Binding Requirement A variable must be bound by the most local potential A ’-binder. (A qualifies as a potential A ’-binder for B iff A c-commands B and the co-indexing o f A and B would not violate the binding principles) (Aoun and Li 1993a: 11, Aoun and Li 1993b:204) The scope principle in (85) repeated in (138) thus predicts the sentences in (127-131) to be ambiguous as to the relative scope o f the subject QP and wh-word, since the QU in Spec of CP c-commands the subject QP and the latter, which is a definite QP, in turn c-commands the trace o f QU in Spec o f QuP. This is not a correct prediction: these sentences have only a single answer, i.e., wide scope o f the wh-word over the subject QP. (138) The Scope Principle An operator A may have scope over an operator B iff A c-commands B or an A ’ element in the chain headed by B. (Aoun and Li 1993a: 11 Aoun and Li 1993b: 204) Thus none o f the two approaches can give a satisfactory account for the scope facts as summarized in (132-133). Please recall that I proposed to account for the scope facts in (132- 133) by assuming that the fronted wh-words as in (123-131) are base-generated as they appear at S structure. In the next sections, I will present the lack o f superiority effects and reconstruction effects in so-called ‘scrambled’ wh-questions in supports of the present proposal for the base-generation o f wh-words in sentence initial position in these constructions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 4.2.1.1. Lack of Superiority Effect Superiority has been viewed as a typical test for movement in the literature (Kuno and Robinson 1972, and Chomsky 1977, among others). It is well-known, however, that Korean does not show this effect as illustrated in (139). The lack o f the superiority effects as witnessed in (139a) sharply contrasts with its English counterpart in (140a). (139) a. Nwukwu-lul nwu-ka manness-ni? whom-ACC who-NOM met-QM ‘*W hom did who m eet?’ b. Nwu-ka nw ukwu-lul manness-ni? who-NOM whom-ACC met-QM ‘W ho met whom ?’ (140) a. *W hom did who meet? b. Who met whom? One way to account for the lack o f superiority effects in Korean is to maintain that the wh- phrase in sentence initial position has undergone A-movement. This move, however, is problematic given the fact that a wh-word preposed into sentence initial position across a CP boundary does not strongly show superiority effects either, as shown in (141). This fact cannot be accounted for by relying on A-movement o f the wh-word, given the assumption that A-movement cannot involve a CP boundary. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 (141) a. (?)?Nwukwu-lul [ cp ne-nun nwu-ka coahanun-ci ] a-ni? whom-ACC you-TOP who-NOM like-QM know-QM ‘*Who is the person x such that you know who likes x?’ b. Ne-nun [ C P nwu-ka nwukwu-lul coahanun-ci] a-ni? you-TOP who-NOM whom-ACC like-QM know-QM ‘Who is the person x such that you know who likes x?’ Moreover, it is not clear whether the (subtle) contrast in grammaticality in (141) can be taken as an argument for superiority effects in Korean since the deviant status is independent o f the constraint, given that scrambling across a clausal boundary normally reduces the grammaticality o f a sentence, no matter what was scrambled as shown below in (142) ( also see Nishigauchi 1990). (142) a. (?)?John-lul [ cp ne-nun nwu-ka coahanun-ci] a-ni? J-ACC you-TOP who-NOM like-QM know-QM ‘Do you know who likes John?’ b. Ne-nun [ C P nwu-ka John-lul coahanun-ci ] a-ni? you-TOP who-NOM J-ACC like-QM know-QM ‘Do you know who likes John?’ Our proposal for the base-generation o f the wh-word in IP-adjoined position in ‘scram bled’ wh-questions gives a natural account for the lack o f superiority effects in (1 3 9 a)3 6 Since 36 One may naturally wonder why the following sentences with the adjunct wh-word in sentence Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 the wh-word is base-generated in sentence initial position, superiority effects are predicted not to show up. On the other hand, since in English this strategy for wh-words is not available, the movement o f a wh-word is predicted correctly to always trigger superiority effects in (140a). 4.2.I.2. Lack of Reconstruction I claimed that wh-words in ‘scram bled’ wh-questions as in (123-131) are base-generated in IP-adjoined position. The question that arises immediately is whether the proposal is consistent with reconstruction effects. Because reconstruction has been viewed more or less as a reflex o f movement, if the ‘scram bled’ wh-questions show reconstruction effects, the current proposal may be considerably undermined. Although reconstruction phenomenon has not been discussed robustly in the Korean literature, below I will try to give some preliminary discussion concerning whether the ‘scram bled’ wh-questions as in (123-131) exhibit the reconstruction effects. initial position show a contrast in grammaticality, given our position that such wh-words are base generated in IP-adjoined position: (i) a.?*Way nwu-ka John-ul mannass-ni? why who-NOM J-ACC met-QM *Why did who meet John?’ b. Nwu-ka way John-ul mannass-ni? who-NOM why J-ACC met-QM *Who meet John why?’ The contrast in grammaticality apparently indicates that the adjunct wh-word may have been preposed into sentence initial position, triggering superiority effects and hence potentially undermining our position for base-generation of the wh-word in ‘scrambled’ wh-questions. I will address this in Chapter 5. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 For this purpose, I will consider three aspects o f reconstruction: reconstruction with respect to bound variable reading, reconstruction with respect to binding Condition C effects, and scope reconstruction. 4.2.1. 2.1. Lack of Bound Variable Interpretation I will consider reconstruction effects with respect to bound variable reading first. It has been common practice in the literature to use caki (he / she) in the discussion o f reconstruction effects (Cho 1994, among others). It should be noted, however, that caki is logophoric in that it has a strong subject orientation as observed by many researchers in the literature (O ’Grady 1992, among others). This point is illustrated in (143). (143) a. John.-i Tom ,-ekey caki; emma-eytayhay malhayssta. J-NOM Tom-DAT his mother-about told ‘John; told Tomj about his, m other.’ b. *Jobn, -i Tomj-ekey cakij emma-eytayhay malhayssta. J-NOM Tom-DAT his mother-about told ‘Johnjtold Tomj about hisj m other.’ Due to this factor, I will not use it for the present discussion. For the bound variable pronoun, I will use ku (he), which is not logophoric as shown below in (144). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 (144) a. Johnj-i Tom j-ekey kuj-uy emma-eytayhay malhayssta. J-NOM Tom-DAT his-POSS m other-about told ‘John; told Tomj about his, m other.’ b. Johtij-i Tomj-ekey kuj-uy emma-eytayhay malhayss-ta J-NOM Tom-DAT his mother-about told ‘Johriitold Tomj about hisj m other.’ It should be noted, however, that some native speakers o f Korean do not get a bound variable interpretation o f ku (he) as easily as do English speakers for its counterpart (see Hong 1985, among others). 3 7 I am sympathetic with those native speakers, since there clearly exists a difference between ku (he) in Korean and he in English with respect to the availability o f a bound variable interpretation. This point is illustrated in (145). (145) a.*Nwukwunaj [cp ku,-ka toktokhata-ko] sayngkakhanta. everyone he-NOM is intelligent-COM P think ‘Everyonej thinks hej is intelligent.’ (Hong 1985: 95) b. Everyonej thinks he; is intelligent. The sentence in (145a) in Korean, when construed as indicated in the gloss is ungrammatical, in contrast to the English example in (145b). However, when ku (he) occurs in a sentence 3 7 One may wonder why ku (he) in Korean does not have a bound variable interpretation as easily as its English counterpart. Our conjecture is that ku (he) is still more or less understood as a referring expression among many native speakers. Historically Korean has used ku + N as referring expression, but the introduction of ku (he) with meaning akin to he in English is fairly recent (See Kang 1988: 196). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 where a bound variable interpretation is pragmatically most natural, the reading can be obtained more or less, as shown in (146). (146) ?Onulpwulo, motun cito haksayng.-i ku,-uy cito kyoswu-lul cenghayssta. As o f today, every student-NOM his-POSS advisor -ACC decided ‘As o f today, every student; decided his; advisor.’ It seems that one can get a bound variable interpretation more or less, once ku (he) is part of an NP, which enters into a close relationship with the antecedent QP, namely the relationship o f an advisor and his student as in (146). 3 8 W ith our discussion o f ku (he) in mind, consider the following sentences in (147): (147) a. ?Motun cito kyoswu;-ka ku;-uy enu cito haksayng-lul chwuchenhayss-ni? every advisor-NOM his which advisee-ACC recommended-QM ‘W hich o f his; students did every advisor; recom m end?’ b.?*ku;-uy enu cito haksayng-lul motun cito kyoswuj-ka chwuchenhayss-ni? his which advisee-ACC every advisor-NOM recommended-QM ‘W hich o f his; students did every advisor; recom m end?’ The wh-phrase in (147) contains only a complement so that Lebeaux factor (Lebeaux 1988) regarding adjunct phrase will be controlled. In the example in (147a), ku (he) in the wh- phrase ku-uy enu cito haksayng (which o f his students) can be construed as a bound variable 3 8 Even when the antecedent and dependent term enter into a close relationship as in (146), some people still cannot get the bound variable interpretation (Chungmin Lee, personal communication). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 bound by motun cito kyoswu (every advisor). Meanwhile, ku (he) cannot be construed in an analogous way in (147b). One will encounter an immediate problem in accounting for the lack o f the reconstruction effect in (147b), once the wh-word in sentence initial position is viewed as an instance o f movement at S structure from its thematic base position since such movement should make it subject to reconstruction at LF.The lack o f reconstruction effects in (147b) can receive a natural account in our terms: Since the wh-word is base-generated in sentence-initial position, it cannot undergo reconstruction, and thus cannot have a bound variable construal at LF. W hen ku (he) and the antecedent QP are further separated by an embedded clause, the grammaticality o f (147b) further deteriorates as shown in (148), indicating no reconstruction effects for the ‘scram bled’ wh-questions. (148) a. ?Motun cito kyoswuj-ka [cp ne-nun every advisor-NOM you-TOP kuj-uy enu cito haksayng-lul chwuchenhayssta-ko] sayngkakha-ni? his which advisee-ACC recommended-COM P think-QM ‘W hich o f his, students do you think every advisor, recommended?’ b. *kui-uy enu cito haksayng-lul [cp ne-nun his which advisee-ACC you-TOP motun cito kyoswuj-ka chwuchenhayssta-ko] sayngkakha-ni? every advisor-NOM recommended-COM P think-QM ‘W hich o f his; students do you think every advisor; recommended?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 4.2.1.2.2. Lack of Binding Condition C Effects The acceptable status o f the sentences in (149) further suggests that the wh-phrase in sentence initial position including John is not reconstructed into its original thematic position at LF, as it does not give rise to binding Condition C effects (Chomsky 1981: 188) in contrast to the corresponding English examples in (150) (see van Riemsdijk and Williams 1981, and Lebeaux 1988 for English): (149) a. ?Johnj-uy enu sacin-ul [C p ne-nun kui-ka silehanta-ko] sayngkakha-ni? J-POSS which picture-ACC you-TOP he-NOM dislike-COM P think-QM *‘W hich picture o f John; do you think he; dislikes?’ b. ?John;-uy enu sacin-ul ne-nun ku;-ekey poyecwuess-ni? J-POSS which picture-ACC you-TOP he-DAT showed-QM *‘W hich picture o f John; did you show him,?’ (150) a. *‘W hich picture o f John; do you think he; dislikes?’ b. *‘W hich picture o f John; did you show him;?’ 4.2.1.2.3. Lack of Scope Reconstruction of Way (why) The Korean wh-question in (151) with the propositional adjunct wh-word way (why) does not have a reading where the adjunct wh-word is construed as having embedded scope, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 which sharply contrasts with the corresponding English example in (152), where the adjunct can be construed as having either embedded scope or matrix scope. (151) W ay ne-nun [ C pJohn-i hakkyo-ey kassta-ko] saynkakha-ni? why you-TOP John-NOM school-to went-COM P think-QM ‘What is the reason x such that for x you think John went to school?’ # ‘W hat is the reason x such that you think John went to school for x?’ (152) Why do you think John went to school? ‘W hat is the reason x such that for x you think John went to school?’ ‘W hat is the reason x such that you think John went to school for x?’ If the adjunct wh-word in sentence initial position in (151) in Korean is the result o f overt movement, i. e., from its base-position in the embedded clause, one cannot account for why the adjunct wh-word cannot have embedded scope, which is available in the corresponding English example in (152). Before closing the section, Let us consider the following examples in (153-154), which may pose a potential problem to our proposal that wh-words in ‘scrambled’ wh-questions are base-generated: (153) a. *Nwu-ka ne-nun [ n p [ cp ssun chayk-ul]] ilkess-ni? who-NOM you-TOP wrote book-ACC read-QM ‘Who is the person x such that you read a book x w rote?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 b. *Enu cakka-ka ne-nun [ N P [C p ssun chayk-ul]] ilkess-ni? which writer-NOM you-TOP wrote book-ACC read-QM ‘W hich writer x is such that you read a book x w rote?’ c. *Eti-se ne-nun [n p [ cp John-i ssun chayk-ul]] ilkess-ni? where-at you-TOP J-NOM wrote book-ACC read-QM ‘W hat is the place x such that you read a book John wrote at x?’ d. *Encey ne-nun [n p [cp John-i ssun chayk-ul]] ilkess-ni? when you-TOP J-NOM wrote book-ACC read-QM ‘W hat is the time x such that you read a book John wrote at x?’ e. *Cheychwung-i elma ne-nun [n p [cp nakanun salam-ul]] coaha-ni? weight-NOM how much you-TOP weigh man-ACC like-QM ‘W hat is the weight x such that you like a man who weighs x?’ f. *Ettehkey ne-nun [n p [cp John-i kulin kulim-ul]] coaha-ni? how you-TOP J-NOM painted painting-ACC like-QM ‘W hat is the means x such that you like pictures John drew by x?’ g. *Ettehkey ne-nun [n p [cp os-ul ipnun salam-ul]] coaha-ni? how you-TOP clothes-ACC get dressed man-ACC like-QM ‘W hat is the manner x such that you like a man who gets dressed in x?’ (154) *W ay ne-nun [ n p [ c p John-i ssun chayk-ul]] ilkess-ni? why you-TOP J-NOM wrote book-ACC read-QM ‘W hat is the reason x such that you read a book John wrote for x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 Given that subjacency (Chomsky 1986) is commonly viewed as a test for movement, the ungrammaticality o f these sentences involving a subjacency violation may lead one to believe that the wh-words have undergone overt movement. These examples, however, are not problematic in our system. Since ‘scrambled’ wh-questions involve a null operator movement at LF for the proper interpretation o f wh-words in our system, the ungrammaticality o f these examples can be accounted for, as the null operator movement 39 involves subjacency violation, crossing the complex noun phrase in (153-154). After all, the lack o f superiority and reconstruction effects regarding bound variable reading, binding Condition C effects and scope reconstruction effects o f the propositional adjunct wh- word in Korean which I have thus far presented, suggest that our position for the base- generation o f the wh-word in ‘scram bled’ wh-questions is essentially on the right track. Now let us turn back to our discussion of scope in wh-questions. Below I will consider complex wh-questions, i.e., biclausal wh-questions with an embedded subject QP followed by a wh-word. I will show that this context is another instance o f symmetry o f scope between nonpropositional wh-words and the propositional adjunct wh-word when it comes to wide scope behavior o f the wh-word over the QP. 3 9 As an alternative for the null operator mmovement at LF, one may opt for movement approach for the wh-words in scrambled wh-questions to account for the ungrammaticality of the sentences in (153- 154). This may give a simplified account for the ungrammaticality these sentences without stipulating null operator movement. With a movement approach, the lack of reconstruction effects in examples as in (148b) and (149a) may be accounted for by assuming initial A-movement into IP adjoined position followed by long distance A’-movement together with the assumption that only A’-movement reconstructs (Jean-Roger vergnaud, personal communication). With the movement approach, one, however, encounters a problem in accounting for the lack of an embedded scope as in (151). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 4.2.2. Complex Wh-Questions Given our proposal that the QM in the head o f IP marks the scope o f the nonpropositional wh-words while the propositional adjunct wh-word moves into Spec CP at LF to mark its scope, when it comes to complex wh-questions as in (155-158), one may expect symmetry o f scope between the two types o f wh-words, since both the QM in the head of the matrix IP and the propositional adjunct wh-word in Spec o f CP at LF can c- command the QP in embedded subject position. (155) a. (?)Ne-nun [cptwu salam isang-i nwukwu-lul chotayhayssta-ko] sayngkakha-ni? you-TOP two man more-NOM whom-ACC invited-COM P think-QM ‘Who do you think more than two men invited?’ b. (?)Ne-nun [cptwu salam isang-i eti-se tampay-lul phiwessta-ko] sayngkakha-ni? you-TOP two man more-NOM where-at cigarette-ACC smoked-COMP think-QM ‘W hat is the place x such that you think more than two men smoked at x?’ c. (?)Ne-nun [cp twu salam isang-i encey ttenassta-ko] sayngkakha-ni? you-TOP two man more-NOM when left-COM P think-QM ‘W hat is the time x such that you think more than two men left at x?’ d. (?)Ne-nun [cp twu salam isang-i ettehkey wassta-ko] sayngkakha-ni? you-TOP two man more-NOM how came-COM P think-QM ‘W hat is the means x such that you think more than two men came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 e. (?)Ne-nun [cptwu salam isang-i M ary-eytayhay ettehkey malhayssta-ko] you-TOP two man more-NOM M -about how talked-COM P sayngkakha-ni? think-QM ‘W hat is the manner x such that you think more than two men talked about Mary in x?’ (156) a. (?)Ne-nun [cpse salam miman-i nwukwu-lul chotayhayssta-ko] sayngkakha-ni? you-TOP three man less-NOM whom-ACC invited-COMP think-QM ‘Who do you think less than three men invited?’ b. (?)Ne-nun [ c p se salam miman-i eti-se tampay-lul phiwessta-ko] you-TOP three man less-NOM where-at cigarette-ACC smoked-COMP sayngkakha-ni? think-QM ‘W hat is the place x such that you think less than three men smoked at x?’ c. (?)Ne-nun [C pse salam miman-i encey tampay-lul phiwessta-ko ] you-TOP three man less-NOM when cigarette-ACC smoked-COM P sayngkakha-ni? think-QM ‘W hat is the time x such that you think less than three men smoked atx?’ d. (?)Ne-nun [c p se salam miman-i ettehkey wassta-ko] sayngkakha-ni? you-TOP three man less-NOM how came-COMP think-QM ‘W hat is the means x such that you think less than three men came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 e. (?)Ne-nun [cp se salam miman-i M ary-eytayhay ettehkey malhayssta-ko] you-TOP three man less-NOM M -about how talked-COMP sayngkakha-ni? think-QM ‘W hat is the manner x such that you think less than three men talked about Mary in x ?’ (157) a. (?)Ne-nun [C p ttak twu salam-i nwukwu-lul chotayhayssta-ko] sayngkakha-ni? you-TOP exactly two man-NOM whom-ACC invited-COMP think-QM ‘Who do you think exactly two men invited?’ b. (?)Ne-nun [C p ttak twu salam-i eti-se tampay-lul phiwessta-ko] you-TOP exactly two man -NOM where-at cigarette-ACC smoked-COM P sayngkakha-ni? think-QM ‘W hat is the place x such that you think exactly two men smoked at x?’ c. (?)Ne-nun [C P ttak twu salam-i encey tampay-lul phiwessta-ko ] you-TOP exactly two man -NOM when cigarette-ACC smoked-COMP sayngkakha-ni? think-QM ‘W hat is the time x such that you think exactly two men smoked at x?’ d. (?)Ne-nun [C P ttak twu salam-i ettehkey wassta-ko] sayngkakha-ni? you-TOP exactly two man-NOM how came-COM P think-QM ‘W hat is the means x such that you think exactly two m en came by xT Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 e. (?)Ne-nun [cp ttak twu salam-i M ary-eytayhay ettehkey malhayssta-ko] you-TOP exactly two man -NOM M -about how talked-COMP sayngkakha-ni? think-QM ‘W hat is the manner x such that you think exactly two men talked about Mary in x?’ (158) a. Ne-nun [cptwu salam isang-i Tom-ul way chotayhaessta-ko] sayngkakha-ni? you-TOP two man more-NOM T-ACC why invited-COM P think-QM ‘W hat is the reason x such that you think more than two men invited Tom for x?’ b. Ne-nun [C p se salam miman-i Tom-ul way chotayhayssta-ko] sayngkakha-ni? you-TOP three man less-NOM T-ACC why invited-COMP think-QM ‘W hat was the reason x such that you think less than three m en invited Tom for x?’ c. Ne-nun [cp ttak twu salam-i Tom-ul way chotayhayssta-ko] sayngkakha-ni? you-TOP exactly two man-NOM T-ACC why invited-COMP think-QM ‘W hat is the reason x such that you think exactly two men invited Tom for x?’ This prediction is borne out indeed: The sentences in (155-157) with an embedded modified QP followed by a nonpropositional wh-word are all grammatical, yielding a single answer as well as a functional answer, just like the sentences with an embedded modified subject QP interacting with a propositional adjunct wh-word in (158), which induces a single answer. This is another case o f symmetry o f scope between the two types o f wh-words when it comes to wide scope construal o f the wh-word over the QP. The LF structures for complex wh- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 questions in (155-157) with a nonpropositional wh-word will be the following in (159), depending on whether the QM or the QP binds the wh-word: (159) a. [IP NP-NOM [r QM, [V P [C P [ ip modified QP-NOM [V P [D P [ n p WH-wordi]]]]]]]] b. [ ip NP-NOM [r QMj [V P [ c p [ip modified QPj-NOM [V P [ d p [ n p WH-wordj]]i]]]]]] In the LF (159a), the QM in the matrix clause unselectively binds the indefinite wh-word in the embedded clause, thus the indefinite wh-word is interpreted as a wh-interrogative. The QM c-commands the embedded modified subject QP, thus having scope over it, yielding a single answer. Pair list answer is impossible, since the indefinite wh-word is bound by the QM. Consider the other LF in (159b). The indefinite wh-word, being bound by the modified QP can never be definite thus leading to the lack o f a pair list answer. Single answer does not obtain either, since the indefinite wh-word is bound by the QP. Although the LF in (159b) does not induce either a pair list answer or a single answer, the one in (159a) induces a single answer, accounting for the grammaticality o f these sentences, given the proposal in (106). The LF for the wh-questions in (158) with a propositional adjunct wh-word will be roughly the one in (160) in the present system: (160) [cpwoyi [ e QMj [iP NP-NOM [ptj [v p [cp [ ip modified QP-NOM [yp h ]]]]]]]] The wh-adjunct way (why) c-commands the embedded modified subject QP, thus having scope over the latter, yielding a single answer. Pair list answer is out o f the question to begin with, given that the adjunct wh-word is not an indefinite. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 5. Correlation I thus far claimed that the asymmetry and symmetry o f scope in Korean wh-questions between the propositional adjunct wh-word and the nonpropositional wh-word with respect to the wide scope o f the wh-word over the QP can receive a satisfactory account when the asymmetry o f scope taking between these wh-words is accepted. Evidence for an asymmetry in means o f scope taking o f the two types of wh-words can be found in another context: Wh- questions with an NPI. This further points to the present generalization that nonpropositional wh-words stay in situ unselectively bound by the QM, which marks their scope at LF, while the propositional adjunct wh-word undergoes movement at LF into Spec o f CP to mark its scope. 5.1 Wh-Questions with an NPI It has been noted in the literature (Beck and Kim 1997 among others) that sentences with a negative polarity item in the subject position with a wh-word following it are ungrammatical, exhibiting so called intervention effects as shown in (161). 40 (161) a. ?*Amwuto nw ukwu-lul chotayhaci an hayss-ni? anyone whom-ACC invite NOT did-QM ‘W ho did no one invite?’ 4 0 Beck and Kim (1997) discusses examples with an argument wh-word only, when it comes to wh- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 b. ?*Amwuto eti-se tampay-lul pwiwuci an hayss-ni? anyone where-at cigarette smoke NOT did-QM ‘W here did no one smoke?’ c. ?*Amwuto encey tampay-lul pwiwuci an hayss-ni? anyone when cigarette-ACC smoke NOT did-QM ‘W hen did no one smoke?’ d. ?*Amwuto ettehkey oci an hayss-ni? anyone how come NOT did-QM ‘W hat is the means x such that no one came by x?’ e.?*Amwuto M ary-eytayhay ettehkey malhaci an hayss-ni? anyone M -about how talk NOT did-QM ‘W hat is the manner x such that no one talked about M ary in x?’ The marginality o f the sentences can be accounted for in terms o f Negation Induced Barrier and Minimal Negative Structure Constraint in Beck and Kim (1997), according to which a negative quantifier cannot intervene between the wh-operator and its variable at LF, assuming wh-movement into Spec o f CP at LF. The definitions o f Negation Induced Barrier and Minimal Negative Structure Constraint are given below in (162) and (163), respectively. (162) Negation Induced barrier (NIB) The first node that dominates a negative quantifier, its restriction, and its nuclear scope is a negation induced barrier (Beck and Kim 1997: 355). questions with an NPI. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 (163) Minimal Negative Structure Constraint (MNSC) If an LF trace (3 is dominated by a NIB a , then the binder o f (3 must also be dominated by a (Beck and Kim 1997: 355). I observe that the account by Beck and Kim (1997) is inadequate, since the sentence as in (164) is grammatical, as originally observed by Cho (1998). 4 1 (164) Amwuto Tom-ul way mannaci an hayss-ni? anyone T-ACC why m eet NOT did-QM ‘W hy did no one m eet Tom ?’ According to Beck and Kim (1997), the sentence in (164) should be ungrammatical contrary to the fact, since the NPI in the subject position will intervene between the wh-operator and its variable at LF as shown in (165). (165) [C l» way, [1 P NPI t, ] ] The current proposal for the nonmovement o f the nonpropositional wh-word and LF movement o f the propositional adjunct wh-word way (why) can account for this interesting asymmetry o f grammaticality in (161) and (164). Before proceeding into the analysis of these data, let me present the scope fact in (161) and (164) first. The sentences in (164) 4 1 The sentences in (161 abc and e) are more or less acceptable under the negative polarity item reading of the wh-words. Although the wh-words in (161 d) can have a negative polarity item reading, the context simply makes the sentence sound very weird under the reading. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 admit only a functional answer, while the one in (164) admits a single answer, hence showing asymmetry o f scope when it comes to wide scope construal o f the wh-word over the QP. The LF representation for the sentences in (161) will be the one in (166a) and (166b), respectively, depending on whether the QM or the QP binds the wh-word. (166) a. [IP NPI [ r QM; [V P [Dp [ n p W H-word,]] ]]] b. [IP NPIj [ r QM, [V P [ d p [ n p WH-wordj]]i ]]] Consider the LF representation in (166a), first. The wh-word is acting as an indefinite bound by the QM in the head o f IP thus is construed as a wh-interrogative. The wide scope o f the wh-word over the QP in the subject position (NPI), i.e., the single answer is impossible, since the QM, a wh-operator cannot c-command the NPI in the subject position. The pair list answer is simply impossible, since the wh-word is bound by the QM. In the other LF representation in (166b), pair list answer is not possible since the QP (NPI) is not definite although the QP binds the indefinite wh-word. 42 Single answer is out o f the question since the QM does not bind the indefinite wh-word thus cannot be construed as a wh-interrogative. Since neither LF representations in (166) for the sentences in (161) can yield a pair list answer or a single answer, these sentences are correctly predicted to be marginal in grammaticality, given the proposal in (106). Turning to the LF representation below in (167) for the sentence in (164), the single answer obtains, since the propositional adjunct wh-word in Spec o f CP can c-command the subject QP (NPI), having scope over it, thus its 4 2 See Higginbotham (1991) for an alternative account for the lack of a pair list answer to a wh- question with the QP no one as a asubject. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 graimnaticality follows, again given the proposal in (106). Also note that pair list answer is out o f the question to begin with, since the propositional adjunct wh-word way (why) is not an indefinite. (167) [ cp wayi [ e QMj [ ip NPI [r tj t, ]]]] W hen the simplex wh-questions in (161) and (164) are embedded to form complex wh- questions, i.e., bi-clausal wh-questions where an embedded subject NPI is followed by a wh- word, the asymmetry between the two types o f wh-words disappears, with both types o f wh- words being able to take scope over the embedded subject QP. The sentences in (168) are all grammatical, yielding a single answer as well as a functional answer, like the sentence in (168) with an NPI in the embedded subject position followed by a propositional adjunct wh-word, which admits a single answer. Thus we have another case o f symmetry o f scope between the two types o f wh-words when it comes to wide scope construal o f the wh-word over the QP. 4 3 (168) a. (?)Ne-nun [cpamwuto nw ukwu-lul chotayhaci an hayssta-ko] sayngkakha-ni? you-TOP anyone whom-ACC invite NOT did-COM P think-QM ‘W ho do you think no one invited?’ 4 3 See Lee and Tomioka (2002) for the original observation of the amelioration of so called intervention effects of argument wh-words in the embedded contexts as in (168). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 b. (?)Ne-nun [cpamwuto eti-se tampay-lul phiwuci an hayssta-ko] you-TOP anyone where-at cigarette-ACC smoke NOT did-COM P sayngkakha-ni? think-QM ‘W hat is the place x such that you think no one smoked at x?’ c. (?)Ne-nun [cp amwuto encey tampay-lul phiwuci an hayssta-ko] you-TOP anyone when cigarette-ACC smoke NOT did-COM P sayngkakha-ni? think-QM ‘W hat is the time x such that you think no one smoked at x?’ d. (?)Ne-nun [ cp amwuto ettehkey oci an haysta-ko] sayngkakha-ni? you-TOP anyone how come NOT did-COM P think-QM ‘W hat is the means x such that you think no one came by x?’ e. (?) Ne-nun [ cp amwuto M ary-eytayhay ettehkey malhaci an hayssta-ko] you-TOP anyone M -about how talk NOT did-COMP sayngkakha-ni? think-QM ‘W hat is the manner x such that you think no one talked about M ary in x?’ (169) Ne-nun [cpam wuto Tom-ul way chotayhaci an hayssta-ko] sayngkakha-ni? you-TOP anyone T-ACC why invite NOT did-COM P think-QM ‘W hat is the reason x such that you think no one invited Tom for x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 The LF structures for the sentences in (168) will be the ones in (170a) and (170b), respectively in the present system, depending on whether the QM or the QP binds the indefinite wh-word: (170) a. [IP NP-NOM [ ,■ QM; [V P [cr [ir NPI [V P [dp [ np WH-wordi]]]]]]]] b. [IP NP-NOM [r QMi [V P [cp [ip NPIj [vp [dp [np WH-word,]],]]]]]] In (170a), the QM in the head o f matrix IP binds the indefinite wh-word and hence the indefinite wh-word is construed as a wh-interrogative. The QM c-commands the NPI in the embedded clause, thus having scope over the latter, yielding a single answer. Pair list answer is not a possibility, since the wh-word is bound by the QM. Consider another LF representation in (170b). The single answer cannot obtain since the indefinite wh-word is bound by the NPI in the embedded Spec of IP. Pair list answer is not possible either, since the NPI is not definite. Since the single answer obtains in one o f the LF representations for the sentences in (168), the grammaticality follows given the proposal in (106). Now the LF structure for the sentence in (169) will be roughly the one in (171). (171) [C P wayi [c - QMj [IP NP-NOM [r t, [v p [Cp [ip NPI [V P t, ] ] ] ] ] ] ] ] The adjunct wh-word way (why) in (171) c-commands the NPI in the embedded subject position, thus having scope over the latter, yielding a single answer. Pair list answer is simply out o f the question, given that the adjunct wh-word cannot be an indefinite. The grammaticality o f the sentence in (169) also follows, again given the proposal in (106). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 Before closing our discussion of QP and wh-word scope interaction, let me address functional answer, albeit briefly, which I have put aside thus far. I will discuss two issues. One is the grammaticality o f simplex wh-questions w ith a modified subject QP followed by a nonpropositional wh-word, which somehow admit only a functional answer. The other is how functional answer obtains with some theoretical implication regarding its status with respect to other readings. 6. Functional Answer Consider simplex questions with a modified subject QP followed by a nonpropositional wh- word in (36=64=77=97) and the same questions with an NPI subject followed by a nonpropositional wh-word in (161), as repeated in (172) and (173), respectively. (172) a.?*Twu salam isang-i nwukwu-lul chotayhayss-ni? two man more-NOM whom-ACC invited-QM ‘Who did more than two men invite?’ b.?*Twu salam isang-i eti-se tampay-lul phiwess-ni? two man more -NOM where cigarette-ACC smoked-QM ‘W here did more than two men smoke?’ c.?*Twu salam isang-i encey tampay-lul phiwess-ni? two man more -NOM when cigarette-ACC smoked-QM ‘W hen did more than two men smoke?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 d.?*Twu salam isang-i ettehkey wass-ni? two m an more-NOM how came-QM ‘W hat is the means x such that more than two men came by x?’ e.?*Twu salam isang-i M ary-eytayhay ettehkey malhayss-ni? two man more-NOM M -about how talked-QM ‘W hat is the manner x such that more than two men talked about Mary in x?’ (173) a.?*Amwuto nwukwu-lul chotayhaci an hayss-ni? anyone whom-ACC invite NOT did-QM ‘Who did no one invite?’ b.?*Amwuto eti-se tampay-lul pwiwuci an hayss-ni? anyone where-at cigarette smoke NOT did-QM ‘W here did no one smoke?’ c.?*Amwuto encey tampay-lul pwiwuci an hayss-ni? anyone when cigarette-ACC smoke NOT did-QM ‘W hen did no one smoke?’ d.?* Amwuto ettehkey o-ci an hays-ni? anyone how come NOT did-QM ‘W hat is the means x such that no one came by x?’ e.?*Amwuto M ary-eytayhay ettehkey malhaci an hayss-ni? anyone M -about how talk NOT did-QM ‘W hat is the manner x such that no one talked about M ary in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 M ost native speakers find the sentences in (172-173) marginal, but not completely ungrammatical. W hat keeps these sentences from being completely out? That is the first issue I need to address. I observe that the sentences in (172-173) can more or less admit a functional answer. 44 Thus, upon hearing a question in (172a), for example, one can answer by saying caki chinkwu (his/ her friend), if it is indeed the case that each individual invited his or her friend. I suggest that the availability o f a functional answer indeed has to do with the marginality o f those sentences in terms o f grammaticality, thus saving them from being completely ungrammatical. It is also instructive to note that unlike a single answer or a pair list answer, functional answer is not informative, thus the sentences, although allowing functional answer, remain marginal. 4 5 Next let us turn to the issue o f how a functional answer is derived. As discussed 4 4 I should note that the grammaticality of the sentences in (172-173) will improve further in a situation where a functional answer is expected to be the only informative answer. 4 5 Nam-kil Kim (personal communication) finds that the sentences in (172-173) are good. Why is it that some speakers find these sentences more or less acceptable in contrast to most native speakers? I suggest that it has to do with discourse-linking effect of nonpropositional wh-words in Korean since they can be typically discourse-linked as which phrase in English typically does. Thus, for example, in a situation where the speaker as a match maker tries to introduce someone to the hearer, the speaker can use nwukwu (whom) instead of enu salam (which person) as shown below: a. Khikhun salam-kwa ttwungttwunghan salam-cwung, ne-nun nwukwu-lul mannako sip-ni? tall man-and fat man-between, you-TOP whom-ACC want to see-QM ‘Between a tall man and a fact man, who do you want to meet? As a matter of fact, it is my intuition that in a situation in which the speaker and the hearer are talking about Bill, Sue, and Tom as the relevant individuals as replaced by the wh-word in (172a-173a), for example, certainly these questions sound more or less acceptable. Thus I can naturally give a single answer by naming one of the aforementioned three individuals. One can check this discourse-linking factor by asking whether he can answer these questions by giving an answer like ‘someone you don’t know’. This answer is virtually impossible to these questions in (172a-173a) even to native speakers who report the sentences in (172-173) are more or less acceptable (Nam-kil Kim, personal communication), while one can give this single answer to the corresponding English question below (Barry Schein, personal communication). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 in section 4, wh-questions with a QP followed by a nonpropositional wh-word can induce a functional answer, while wh-questions with a QP followed by a propositional adjunct wh- word cannot. Thus wh-questions in (172-173) with a modified QP or an NPI followed by a nonpropositional wh-word as well as the ones in (110-113) repeated in (174-177) with a definite QP followed by a nonpropositional wh-words all admit a functional answer, while the ones in (114) as repeated in (178) cannot. (174) a. M otun salam-i nwukwu-lul chotayhayss-ni? every man-NOM whom-ACC invited-QM ‘Who did everyone invite?’ b. M otun salam-i eti-se tampay-lul phiwess-ni? every man-NOM where-at cigarette-ACC smoked-QM ‘W here did everyone smoke?’ c. Motun salam-i encey tampay-lul phiwess-ni? every man-NOM when cigarette-ACC smoked-QM ‘W hen did everyone smoke?’ d. M otun salam-i ettehkey wass-ni? every man-NOM how came-QM ‘W hat is the means x such that everyone came by x ?’ b. Who did more than two men invite? Once the effect is properly controlled, I believe one can give a consistent judgment on these data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 e. M otun salam-i M ary-eytayhay ettehkey malhayss-ni? every man-NOM M -about how talked-QM ‘W hat is the manner x such that everyone talked about Mary in x ?’ (175) a. Ku twu salam-i nwukwu-lul chotayhayss-ni? the two man-NOM whom-ACC invited-QM ‘Who did the two men invite?’ b. K u tw u salam-i eti-se tampay-lul phiwess-ni? the two man-NOM where-at cigarette-ACC smoked-QM ‘W here did the two men smoke?’ c. Ku twu salam-i encey tampay-lul phiwess-ni? the two man-NOM when cigarette-ACC smoked-QM ‘W hen did the two men smoke?’ d. Ku twu salam-i ettehkey wass-ni? the two man-NOM how came-QM ‘W hat is the means x such that the two men came by x?’ e. Ku twu salam-i M ary-eytayhay ettehkey malhayss-ni? the two man-NOM M -about how talked-QM ‘W hat is the m anner x such that the two men talked about Mary in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 (176) a. John-kwa M ary-ka nwukwu-lul chotayhayss-ni? J-and M -NOM whom-ACC invited-QM ‘Who did John and M ary invite?’ b. John-kwa M ary-ka eti-se tampay-lul phiwess-in? J-and M -NOM where-at cigarette-ACC smoked-QM ‘W here did John and M ary smoke?’ c. John-kwa M ary-ka encey tampay-lul phiwess-in? J-and M -NOM when cigarette-ACC smoked-QM ‘W hen did John and M ary smoke?’ d. John-kwa M ary-ka ettehkey wass-ni? J-and M -NOM how came-QM ‘W hat is the means such that John and M ary came by x ?’ e. John-kwa M ary-ka Tom-eytayhay ettehkey malhayss-ni? J-and M -NOM T-about how talked-QM ‘W hat is the manner x such that John and M ary talked about Tom in x?’ (177) a. Taypwupwun-uy salam-i nwukwu-lul chotayhayss-ni? most-POSS man-NOM whom-ACC invited-QM ‘Who did most o f the m en invite?’ b. Taypwupwun-uy salam-i eti-se tampay-lul phiwess-ni? most-POSS man-NOM where-at cigarette-ACC smoked-QM ‘W here did most o f the men smoke?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 c. Taypwupwun-uy salam-i encey tampay-lul phiwess-ni? most-POSS man-NOM when cigarette-ACC smoked-QM ‘W hen did most o f the men smoke?’ d. Taypwupwun-uy salam-i ettehkey wass-ni? most-POSS man-NOM how came-QM ‘W hat is the means x such that most o f the men came by x?’ e. Taypwupwun-uy salam-i M ary-eytayhay ettehkey malhayss-ni? most-POSS man-NOM M -about how talked-QM ‘W hat is the manner x such that most o f the men talked about Mary in x?’ (178) a. M otun salam-i Tom-ul way chotayhayss-ni? every man-NOM T-ACC why invited-QM ‘W hy did every man invite Tom ?’ b. Ku twu salam-i Tom-ul way chotayhayss-ni? the two man-NOM T-ACC why invited-QM ‘W hy did the two men invite Tom ?’ c. John-kwa M ary-ka Tom-ul way chotayhayss-ni? J-and M -NOM T-ACC why invited-QM ‘Why did John and M ary invite Tom ?’ d. Taypwupwun-uy salam-i Tom-ul w ay chotayhayss-ni? most-POSS man-NOM T-ACC why invited-QM ‘W hy did most o f the m en invite Tom ?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 I will thus suggest that functional answer obtains when a QP binds the indefinite wh-w ord at LF. The LF for these sentneces in (172-173) and (174-177) will be roughly the following in (179) in our system, depending on whether the QM or the QP binds the indefinite wh-word: (179) a. [jpQPj [r QM, [V P V [D P[N P WH-wordi]] ]]] b. [IP QPj [r QM; [vp V [D P[N P WH-wordj]], ]]] Please note that what the sentence in (174a), for example, under the LF in (179a) is intended to ask is something like ‘what is the individual x such that everyone invited x?’, i.e. a single answer. Our intertest is thus the LF representation in (179b). W hat is the information solicited from the speaker with the sentence in (174a), for example, under this LF representation? Following Chierchia (1991, 1993), I will suggest that what is asked for under the LF (179b) is actually something like ‘which function / i s such that everyone x invited / x ) ? \ understanding the DP as a function taking the indefinite wh-word as the argument, which is in turn bound by the QP. Thus in our system, when a QP binds an indefinite wh- word a functional answer obtains. W hat is the mechanism that differentiates the functional reading from the pair list answer as in (179b)? Please recall Chierchia (1993), for example, adopts absorption in the sense of Higginbotham and M ay (1981) to differentiate the pair list reading from the functional reading (see fn. 27). In our system, pair list reading simply obtains when the definite QP binds the indefinite wh-word. As the indefinite wh-word becomes definite by picking up the definite quantificational force o f the binde, the hearer o f the question with a definite QP can Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 identify the individuals ( or measure, means, manner..) as replaced by the wh-word, hence being able to give a list answer. M eanwhile when a nondefinite QP binds the indefinite wh- word, pair list answer cannot obtain since the indefinite cannot have definite quantificational force such that the hearer o f a wh-question with a modified QP cannot identify the individuals (measure, means,manner) and thus cannot give list answers. Thus in our system pair list answer is a special case o f a functional answer, which conforms to the intution in Engdahl (1986) and Chierchia (1991, 1993). When it comes to the relation between the single answer and the functional answer, I suggest that functional answer and single answer is completely independent readings given the separate LF representations in (179). Thus one may easily imagine a situation where a functional answer obtains, while single answer does not, which is indeed attested by the examples in (172-173), which admits only a functional answer. The lack o f a functional answer in (178) follows from from the present proposal, since the propositional adjunct wh-word way (why) is not an indefinite, in contrast to nonpropositional wh-words. 7. Conclusion To conclude, I made some novel observations, namely that there exists an asymmetry and symmetry o f scope between the propositional adjunct wh-word way (why) and nonpropositional wh-words when it comes to the wide scope construal o f a wh-word over a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 QP in wh-questions in Korean: In simplex wh-questions, nonpropositional wh-words cannot take wide scope over a matrix subject QP, while propositional adjunct wh-word way (why) can. In complex wh-questions, where an embedded subject QP is followed by a wh-word, the asymmetry between the two types o f wh-words disappears, with both types o f wh-words being able to take scope over the embedded subject QP. The asymmetry and symmetry of scope between the two types o f wh-words I observe here cannot be fully accounted for under either a uniform LF wh-movement approach (Hoji 1985 and Kim 1990) or a uniform wh-in- situ approach (Aoun and Li 1993b). I tried to establish that the nonpropositional wh-word is indefinite in the sense o f Lewis (1975) and Heim (1982) using two criteria for an indefinite: the lack o f quantificational force and the availability to scope out o f standard syntactic islands. The asymmetry and symmetry of scope essentially follows once the asymmetric scope taking between the propositional adjunct wh-word and non-propositional wh-word in wh-question that is directly associated with their different status as an indefinite is accepted. Based on the marginality o f wh-questions that admit only a functional answer, I also proposed that answerability should count as an evaluation measure for the grammaticality of a wh-question too. In other words, wh-questions that cannot be answered by at least a single answer or a pair list answer are marginal in grammaticality. In the next chapter, I will further extend the asymmetric account o f scope taking for the two types o f wh-words as proposed in this chapter to other wh-in-situ languages such as Japanese and Chinese and try to show that these two languages in fact have the same asymmetric scope taking mechanisms as part o f their grammars. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 CHAPTER 3 Asymmetry of Scope in Japanese and Chinese Wh-questions 1. Introduction I proposed that nonpropositional wh-words quantifying over individuals, measure, means, or manner in Korean are indefinites, while the propositional adjunct wh-word way (why), which typically quantifies over propositions, is not, and that the scope o f the former is marked by the question morpheme (QM, henceforth), while the scope o f the latter is marked by its movement at LF into Spec o f CP. In this chapter, I will try to show how our proposal for asymmetric scope taking for the two types o f wh-words in Korean can extend to other wh-in- situ languages such as Japanese and Chinese. The generalization made in the literature is that normally wh-words cannot take scope over QPs in Japanese and Korean (see Hoji 1985, Kim 1990), while in Chinese they can (see Huang 1982, Aoun and Li 1989, 1993abc). W hen it comes to Korean, which is typologically akin to Japanese and Chinese, I showed that the generalization in the literature is manifested only in a limited way. One observation I made in Chapter 2 was that in simplex wh-questions, an important asymmetry o f scope emerges between nonpropositional wh-words and the propositional adjunct wh-word way (why): the former cannot take scope over a subject QP, while the latter can. However, in complex wh-questions with an embedded subject QP followed by a wh-word, no matter what type o f wh-word it is, the wh-word can take scope over the QP. The discussion in this chapter will reveal that Japanese and Chinese also pattern with Korean in exhibiting the same asymmetry and symmetry of scope between the two types o f wh-words in the aforementioned contexts. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 2. Wh-words as Indefinites Japanese and Chinese wh-words are listed below in (1) (also see Huang 1982: 241, W atanabe 1992: 48-49).1 Japanese Chinese a. dare shei ‘w ho’ b. dono N na-CL-N ‘which N ’ c. nani shenme ‘w hat’ d. doko nali ‘w here’ e. itsu shenmeshihou ‘w hen’ f. ikutsu ji-CL ‘how m uch’ g. dooyatte zenmeyang ‘how ’ h. doo zenmeyang ‘how ’ i. naze weishenme ‘w hy’ Please recall the two main criteria for an indefinite discussed in Chapter 2. One is its lack of inherent quantificational force. The other is its ability to scope out o f the syntactic islands. Thus one can say that if either o f the two properties is lacking, the item in question is not an indefinite. The paradigm in Japanese in (2-3) and Chinese in (4-5) below involving so-called mo and dou quantification construction (see Nishigauchi 1990 for Japanese and Tsai 1 In Japanese dooyatte (how) is an adjunct wh-word expressing ‘means’ and doo (how) is an adjunct wh-word convening ‘manner’. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 1994 for Chinese, among others), respectively, provides an ideal test to determine whether wh-words in these languages are indefinites or not. (2) Japanese a. [Dare-ga kaita tegami-ni]-mo onazi koto-ga kaiteatta. who-NOM wrote letter-in-M O same thing-NOM written ‘for every x, y, x a person, y a letter x wrote, the same thing was written in y.’ Nishigauchi (1990: 126) b. [Dono hito-ga kaita tegami-ni]-mo onazi koto-ga kaiteatta. which man-NOM wrote letter-in-M O same thing-NOM written. ‘for every x, y, x a person, y a letter x wrote, the same thing was written in y .’ c. [Kare-ga doko-de kaita tegami-ni]-mo onazi koto-ga kaiteatta. he-NOM where-at wrote letter-in-M O same thing-NOM written ‘for every x, y, x a place, y a letter he wrote at x, the same thing was written in y .’ d. [Kare-ga itsu kaita tegami-ni]-mo onazi koto-ga kaiteatta. he-NOM when wrote letter-in-MO same thing-NOM written. ‘for every x, y, x a time, y a letter he wrote at x, the same thing was written in y .’ e. [Taijuu-ga ikutsu aru hito-ni]-mo onazi seki-ga ataerareta. weight-NOM how much have person-to-M O same seat-NOM was given ‘for every x, y, x a weight, y a person who weighs x, the same seat was given to y .’ f. [Kare-ga dooyatte kaita e-ni]-mo onazi koto-ga kaiteatta. he-NOM how drew picture-in-M O same thing-NOM written. ‘for every x, y, x a means, y a picture he drew by x, the same thing was drawn in y .’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 g. [Sono kyooju-nituite doo itta gakusei-ni]-mo A-ga ataerareta. that professor-about how talk student-to-M O A-NOM was given ‘for every x, y, x a manner, y a student who was talking about that professor in x, A was given to y. ’ (3) Japanese * [Kare-ga naze kaita e-ni]-mo onazi koto-ga kaiteatta. he-NOM why wrote picture-in-M O same thing-NOM was written ‘for every x, y, x a reason, y a picture he drew for x, the same thing was drawn in y .’ (4) Chinese a. [shei xie de shu] dou hao. who write book all good ‘for every x, y, x a person, y a book x wrote, y is good.’ b. [na-ge-ren xie de shu] dou hao. which-CL-person write book all good ‘for every x , y, x a person, y a book x wrote, y is good.’ c. [John zai nali xie de shu] dou hao. John at where write book all good ‘for every x, y, x a place, y a book John wrote at x, y is good.’ d. [John shenmeshihou xie de shu] dou hao. John when write book all good ‘for every x, y, x a time, y a book John wrote at x, y is good.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 e. [zhong ji bang de ren] dou hao. amount how much person all good ‘for every x, y, x a weight, y a person who weighs x, y is good.’ f. [John zenmeyang hua de hua] dou hao. John how paint painting all good ‘for every x, y, x a means, y a painting John drew by x, y is good.’ g. [zenmeyang shuo wo de ren] dou hao. how say me person all good ‘for every x, y, x a manner, y a person who talks about me in x, y is good.’ (5) Chinese *[John weishenme xie de shu] dou hao. John why write book all good ‘for every x, y, x a reason, y a book John wrote for x, y is good.’ The data in (2) and (4) suggest that nonpropositional wh-words in Japanese and Chinese are indefinites: The quantificational force o f these wh-words are determined by an external quantificational element mo and dou, respectively, which carry a universal quantificational force. The informal logical notations indicate that these wh-words can take scope out o f a syntactic island, i.e., complex noun phrase. W hen it comes to the propositional adjunct wh- words in (3) and (5) in these languages, the analogous construal is simply impossible, leading to the opposite conclusion, namely, that they are not indefinites. W ith the different status of the two types o f wh-words as indefinites in mind, let us turn to the question o f how wh- questions are formed in Japanese and Chinese. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 3. The Representation of Wh-Questions It is well-known that wh-words in Japanese and Chinese normally do not undergo movement overtly, with their scope indicated by the QM, as shown in (6-7). (6) Japanese a. Kimi-wa [cp John-ga dare-o syootaisita-to] omotteru-no? you-TOP J-NOM whom-ACC invited-COMP think-QM ‘Who do you think John invited?’ b. Kimi-wa [cp John-ga naze Mary-o syootaisita-to] omotteru-no? you-TOP J-NOM why M-ACC invited-COMP think-QM ‘What is the reason x such that you think John invited Mary for x?’ (7) Chinese a. ni renwei [cp John yaoqing-le shei] (ne)? you think John invite-ASP whom (QM) ‘Who do you think John invited?’ b. ni renwei [ cp John weishenme yaoqing-le you think John why invite-ASP ‘W hat is the reason x such that you think John invited Mary for x?’ Again following the original intuitions in the literature (Chomsky 1964: 38, Katz and Postal 1964: 93, Klima 1964: 253, Kuroda 1969: 266ff), according to which a wh-word is a Mary] (ne)? Mary (QM) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 combination o f a wh-operator and indefinite some (one, thing...) in English, I suggest that like Korean, the question morphemes, no in Japanese and ne in Chinese (QM, henceforth) in (6a-7a) are equivalent to the wh-operator with [+Q, +WH] feature specification, and that nonpropositional wh-words correspond to the indefinite some part. I thus suggest that the QM as a wh-operator unselectively binds and thus marks the scope o f nonpropositional wh- words. W hen it comes to propositional adjunct wh-words, as in (6b-7b), I suggest they should undergo movement into Spec o f CP at LF to mark their scope via spec-head agreement with the QM with [+Q, +W H], driven by the need for proper interpretation, again like Korean. The QM in (6b-7b) does not have an operator force, but rather acts as a position holder. Hereafter ‘Q M ’ is used to refer to the QM with [+Q,+WH] feature specification exclusively, unless noted otherwise. To summarize, when it comes to the interpretation o f wh-words in Japanese and Chinese, I suggest that the grammar has two different mechanisms, just like Korean: nonpropositional wh-words are unselectively bound by the QM which marks their scope at LF, while the propositional adjunct wh-words are raised at LF into Spec of CP to mark their scope via spec-head agreement with the QM, driven by the need for proper interpretation 4. QP and Wh-word Scope Interaction With the two distinct scope taking mechanisms o f the two types o f wh-words in Japanese and Chinese in mind, now let us turn to Quantifier Phrase (QP, henceforth) and wh-word scope interaction. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 4.1. The Position of QM in Japanese and Chinese Since I suggested that the QM, as a wh-operator, marks the scope o f nonpropositional wh- words in Chinese and Japanese, it is important to determine its position before I proceed in the discussion o f QP and wh-word scope interaction. Consider Japanese question below in (8): (8) D are-ga ki-masita-*(ka)? who-NOM come-PAST-QM ‘Who cam e?’ Japanese has two types o f QMs, namely, no and ka. The former is informal QM and the latter is a formal one. The QM in Japanese is not a free standing morpheme nor can it be deleted normally (Nishigauchi 1985: 101) as indicated in (8) above.2 One may thus suggest that the QM in Japanese, either formal or informal, is part o f the verbal morphology. 3 Japanese verbal morphology is agglutinating, consisting o f more than one bound morpheme, as shown in (8), like Korean. Given the recent proposals (Pollock 1989, Chomsky 1993, among others) viewing the IP system as an extension o f the VP system, with each 2 When the QM is deleted, either formal or informal, the sentence should receive a rising intonation at the end of it. See Nishigauchi (1985: 101) and Lasnik and Saito (1992: 1 87, fn.8). 3 Since the QM can be deleted, albeit with a rising intonation at the end of the sentence, one may suggest that it is not part of the verbal morphology. However, to the extent that the QM can be deleted it may be that the rising intonation at the end of the sentence can compensate for the deletion of the QM not that the QM is not part of the verbal morphology. The QM in the embedded clause can never be deleted where rising intonation cannot be used as an option to compensate for the deletion of the QM, which indicates that the QM in Japanese is rather part of the verbal morphology. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 inflectional morpheme o f the verb projecting a separate functional projection o f the IP system, it follows that the QM, which is part o f the verbal morphology, should not project in the CP domain in the sense o f Rizzi (2000), which is comprised o f a focus projection, topic projection and complementizer projection. Given that Japanese is a strictly head-final language, the particular sequence o f the bound morphemes on the verb in (8) including the QM thus leads one to postulate the following clausal structure in (9), where the past tense morpheme and question morpheme project TP and IP respectively, right on top o f VP under the strict projectionist hypothesis (Pollock 1989 and Chomsky 1993, also see Benmamoun 1992:8ff), according to which each bound morpheme o f the verb projects its own functional projection in the IP field: (9) [ip [ tp [ v p v ] t] i] Our discussion o f the Japanese verbal morphology, which indicates that QM is base generated in the head o f IP can find additional support in Kim (1990) who originally makes the proposal that QM in Japanese is base-generated in the head o f IP, mainly based on the fact that the QM does not show a complementary distribution with the quotative marker, to (that), which is standardly assumed to project CP as shown in (10). (10) John-wa [ C p M ary-ga nani-o katta-ka-to] tazuneta. John-TOP M -NOM what-ACC bought-QM -COM P asked ‘John asked what M ary bought.’ (Kim 1990: 227) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 Thus based on our discussion on the Japanese verbal morphology and Kim ’s original proposal drawn from the lack o f complementary distribution o f the QM and the quotative marker to (that) in Japanese as in (10), I will conclude that QM in Japanese indeed projects IP, like Korean. As for Chinese, it has mostly been assumed in the literature that Chinese has a head initial structure at the CP level. However, Tang (1988) claims that Chinese is head final at the CP level, based on the sentence final occurrence o f the question particles, as cited in Lin (1992). (11) a. ta hui lai ma? he will come QP ‘W ill he com e?’ b. ni xihuan shei ne? you like whom QP ‘W hom do you like?’ (Lin 1992:309) According to Tang, the question particles ma and ne in (11), which are yes-no and wh- question morphemes, respectively, are interpreted in the head o f CP. M ore importantly, in Chinese the question particles, which appear at the end o f the sentence are free standing morphemes separate from the verb by the object as shown in (lib). Moreover, it can be freely deleted, as shown in (12). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 (12) a. ta hui lai? he will come ‘Will he com e?’ b. ni xihuan shei? you like whom ‘W hom do you like?’ The fact that the question particle in Chinese is a free standing morpheme standing apart from the verb as in (1 lb ) and is completely optional as in (12) strongly suggests that it is not part o f the verbal morphology and thus does not project a functional projection in IP system, given the strict projectionist view (Pollock 1989 and Chomsky 1993, among others) that verbal morphology projects functional projections in IP system. Thus, one may conclude that the QM in Chinese does not project a functional projection in IP system but somewhere in the CP field (Rizzi 2000), although where precisely it is generated cannot be determined yet, given the complex CP field as in Rizzi (2000), which I will return to in section 4. 3. Now, keeping in mind the unselective binding o f nonpropositional wh-words vs. LF movement o f propositional adjunct wh-words, and parametric variation regarding the position o f the QM, i.e., in the head o f IP in Japanese, and somewhere in the CP field in Chinese, let us turn to scope phenomena in Japanese and Chinese wh-questions. Throughout this chapter, let us suppose the situation where the speaker and the hearer are chatting over Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 John, Brown, Jack, Bill, Tom, Sue and Alec who were at the spring break party, to which each o f them was supposed to invite whoever they wished. 4.2. Japanese Consider the following simplex wh-questions w ith a modified subject QP: (13) a. ?*Hutari ijyou-no hito-ga dare-o syootaisita-no? two more-POSS man-NOM whom-ACC invited-QM ‘Who did more than two men invite?’ b. ?*Hutari ijyou-no hito-ga doko-de tabako-o sutta-no? two more-POSS man-NOM where-at tobacco-ACC smoked-QM ‘Where did more than two men smoke?’ c. ?*Hutari ijyou-no hito-ga itsu kaetta-no? two more-POSS man-NOM when left-QM ‘W hen did more than two men leave?’ d. ?*Hutari ijyou-no hito-ga dooyatte kitta-no? two more-POSS man-NOM how came-QM ‘W hat is the means x such that more than two m en came by x?’ e. ?* Hutari ijyou-no hito-ga Tom-nituite doo itta-no? two more -POSS man-NOM T-about how saying-QM ‘W hat is the manner x such that more than two men were talking about Tom in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 (14) ?Hutari ijyou-no hito-ga Tom-o naze syootaisita-no? two more -POSS man-NOM T-ACC why invited-QM ‘W hat is the reason x such that more than two men invited Tom for x?’ Like Korean, as discussed in Chapter 2, simplex questions with a modified subject QP co occurring with a nonpropositional wh-word, as in (13), are marginal, without admitting either a single answer or a pair list answer, but only a functional answer, while the sentence with the same QP co-occurring with the propositional adjunct wh-word, as in (14), is grammatical, admitting a single answer. The fact remains the same when the modified QP subjects in (13-14) are replaced with modified QPs o f other types, such as sanin ika-no hito (less than three men), or choudo sanin-no hito (exactly three men). The scope facts in (13- 14) thus shows that the two types o f wh-words show asymmetry o f scope regarding the wide scope construal o f the wh-word over the subject QP. The asymmetry o f scope illustrated in (13-14) in Japanese simplex questions can be accounted for, given the proposal that QM in Japanese is also base-generated in the head o f IP and that the QM as a wh-operator unselectively binds and marks the scope o f the nonpropositional wh-words, while the propositional adjunct wh-word should undergo movement at LF into Spec o f CP, via spec-head agreement with the QM, to be properly interpreted by forming an operator variable chain. Following the discussions of the structure of wh-words in English and Korean in Chapter 2, let us assume that Japanese nonpropositional wh-words also have a DP structure (cf. W atanabe 1992: 113). Thus the LF representations for the sentences in (13) will be (15a) and (15b), respectively with irrelevant details suppressed, depending on whether the QM or the QP binds the wh-word. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 (15) a. [IP QP-NOM [r QM; [V P V [D P[N P WH-word,]] ]]] b. [IP QPj-NOM [r QM, [V P V [ d p [ n p WH-word,]], ]]] I will provide the LF representations for Japanese and Chinese, using head initial language notation throughout this chapter, for the convenience o f the reader. In the LF (15a), the QM binds the indefinite wh-word and the indefinite wh-word is thus construed as a wh- interrogative. The QM, however, cannot c-command the QP in Spec o f IP thus cannot take scope over it, hence yielding no single answer. Pair list answer is impossible to begin with, given that the indefinite wh-word is bound by the QM. Consider the other LF in (15b). Single answer is out o f the question to begin with, since the indefinite wh-word is not bound by the QM. Pair list answer is impossible since the QP binding the indefinite wh-word is a modified QP, which cannot be definite. The marginality in (13) thus follows given the proposal in Chapter 2, as repeated in (16). (16) W h-questions should be answerable with at least a pair list answer or a single answer. Next consider the following LF representation in (17) for the sentence in (14) with a propositional adjunct wh-word, with irrelevant details suppressed: (17) [C P naze. [c QM, [IP QP-NOM [r t, [V P V t, ]]]]] The QM in (17) is raised into the head o f CP from its base-generated position in the head o f IP at LF, driven by the need to enter into a spec-head agreement with the propositional adjunct wh-word in Spec o f CP at LF. Since the propositional adjunct wh-word in Spec of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 168 CP can c-command the subject QP, a single answer obtains and the grammaticality o f (14) follows from the proposal in (16). The asymmetry o f scope between the two types o f wh- words shown in (13-14) has an important implication for the simplex wh-questions in (18) with a universal QP: (18) a. Daremo-ga dare-o syootaisita-no? everyone-NOM whom-ACC invited-QM ‘Who did everyone invite?’ b. Daremo-ga tabako-o doko-de sutta-no? everyone-NOM tobacco-ACC where-at smoked-QM ‘W here did everyone smoke?’ c. Daremo-ga itsu kaetta-no? everyone-NOM when left-QM ‘W hen did everyone leave?’ d. Daremo-ga dooyatte kita-no? everyone-NOM how came-QM ‘What is the means x such that everyone came by x ?’ e. Daremo-ga Tom-nituite doo itta-no? everyone-NOM T-about how saying-QM ‘W hat is the manner x such that everyone was talking about Tom in x?’ The sentences in (18) admit pair list answer, single answer and functional answer. The same is true when the universal QP is replaced with sono hutari-no hito (the two men) and John to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 169 Mary (John and M ary) (Hajime Ono and Naomi Harata, personal communication) although the Japanese informants note that the pair list answer may be difficult to get for some older generation people. The single answer, pair list answer and functional answer, for (18a) for example will be something like (19a), (19b), and (19c), respectively, assuming that John, Jack, Brown, Bill, and Sue exhaust the set o f people in the discourse context. (19) a. Mary-desu. M ary- is ‘It is M ary’ b. John-ga Ines-o, Jack-ga Jane-o, Brown-ga Janny-o, Bill-ga Mary-o J-NOM I-ACC J-NOM J-ACC B-NOM J-ACC B-NOM M-ACC sosite Sue-ga Jen-o syootaisita. and J-NOM J-ACC invited ‘John invited Ines, Jack invited Jane, Brown invited Janny, Bill invited M ary and Sue invited Jen.’ c. Zibun-no koibito-desu his-POSS girl friend is ‘It is his girl frien d / However, intuitions among native speakers regarding the data in (18) seem to vary. Hoji (1985: 270), for example, observes that the sentence similar to the one in (18a) admits only a single answer (also see Kim 1990: 240), reporting a judgm ent different from the one from the Japanese speakers I consulted. W hen it comes to the question o f why Japanese speakers vary in their judgm ents o f the availability o f a pair list answer for questions like (18a), for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170 example, I do not have any good answer at this point. But I would like to note that the following cleft construction (20) is a typical means o f answering a wh-question like the one in (18a) in Japanese (Naomi Harata, personal communication): (20) Daremo-ga syootaisita-no-wa M ary desu. Everyone-NOM invited-NOM N-TOP M ary is ‘It is M ary that everyone invited.’ Thus to the extent that the answer in (20) is chosen for the question in (18a), for example, a pair list answer is not a possibility, since the hearer upon hearing the question will interpret it as asking ‘who is the person x such that everyone together invited x?’. That is, for some speakers, (18a) might presuppose that there exists an individual that was invited by everyone, making a pair list answer o f course unavailable. W hen it comes to the nature o f a single answer to the question in (18a), for example, Hoji (1985) suggests that it is due to the specific group interpretation o f the subject QP, which is not a plausible proposal, however, since relying on specific group construal o f a QP will invite a nontrivial problem to accounting for the contrast in grammaticality as in (13-14) as I discussed in Chapter 2 regarding similar Korean examples. Since the example in (14) is grammatical, one needs to say that the modified subject QP is interpreted as referring to a specific group in H oji’s system. Then the question is why the sentences in (13) with the same QP are still marginal in grammaticality? Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 Given the scope facts in simplex wh-questions with a modified subject QP in (13-14), which suggest that nonpropositional wh-words cannot take scope over the modified subject QP while the propositional adjunct wh-word can do so, one may suggest that the single answer to the questions in (18) is actually a special case o f a pair list answer. The following sentence (21) actually confirms this: (21) Daibubun-no gakusei-ga dare-o hoomon shimasita-ka? most-POSS student-NOM whom-ACC visited-QM ‘Who did most o f the students visit?’ Interestingly, native speakers o f Japanese whom I consulted (Hajime Ono and Naomi Harata, and Yasuhiko M iura personal communication) observe that the sentence in (21) is quite grammatical (cf. different judgm ents for a similar example in Pesetsky 2000: 83). Let us suppose as a relevant context the department orientation for the new incoming class of graduate students, A, B, C, D, E, F, and G. Let us further suppose that students A, B, C, D visited Prof. John and students C, D, E, F visited Prof. Mary. Given the above context, suppose the speaker asks the question in (21). The hearer cannot answer the question by saying ‘Prof. John and Prof. M ary.’ This means that the set denoted by daibubun-no gakusei (most o f the students) can never co-vary with the choice o f a professor, which means that the wh-word in (21) cannot take scope over the subject QP. Now the prediction I make is that in complex wh-questions, where an embedded modified QP subject is followed by a wh-word, as in (22-23), the asymmetry o f scope between the two types o f wh-words will disappear, since the QM in the head of matrix IP at LF in (22), which Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 marks the scope o f the nonpropositional wh-word, can c-command the QP, as can the propositional adjunct wh-word in the Spec o f matrix CP at LF in (23). (22) a. (?)Kimi-wa [cphutari ijyou-no hito-ga dare-o syootaisita-to] omotteru-no? you-TOP two more-POSS man-NOM whom-ACC invited-COM P think-QM ‘W ho do you think more than two men invited?’ b. (?)Kimi-wa [cphutari ijyou-no hito-ga doko-de tabako-o you-TOP two more-POSS man-NOM where-at tobacco-ACC sutta-to] omotteru-no? smoked-COMP think-QM ‘W hat is the place x such that you think more than two men smoked at x?’ c. (?)Kimi-wa [cphutari ijyou-no hito-ga itsu kaetta-to] omotteru-no? you-TOP two more-POSS man-NOM when left-COM P think-QM ‘W hat is the time x such that you think more than two men left at x?’ d. (?) Kimi-wa [cphutari ijyou-no hito-ga dooyatte kitta-to] omotteru-no? you-TOP two more-POSS man-NOM how came-COM P think-QM ‘W hat is the means such that you think more than two men came by x?’ e. (?) Kim i-wa [C p hutari ijyou-no hito-ga Tom-nituite doo itta-to] you-TOP two more -POSS man-NOM T-about how talking-COMP omotteru-no? think-QM ‘W hat is the manner x such that you think more than two men were talking about Tom in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 (23) Kimi-wa [cphutari ijyou-no hito-ga Tom-o naze syootaisita-to] you-TOP two more-POSS man-NOM T-ACC why invited-COMP omotteru-no? think-QM ‘W hat is the reason x such that you think more than two men invited Tom for x?’ The prediction is indeed confirmed: complex wh-questions in (22-23) can admit a single answer, which means both types o f wh-words can take scope over the QP. The LF structures for the sentences in (22) will be roughly as the ones in (24a) and (24b), respectively, depending on whether the QP or the QM binds the indefinite wh-word. (24) a. [IP NP-TOP [r QM, [V P V [ c p [ i p QP-NOM [V P V [ d p [ n p WH-word,]] ]]]]]] b. [IP NP-TOP [r QM, [ V P V [ c p [ i p QPr NOM [vp V [D P[N P WH-wordj]]i ]]]]]] In the LF (24a), the QM in the matrix clause unselectively binds the indefinite wh-word in the embedded clause hence the indefinite wh-word is constm ed as a wh-interrogative. The QM c-commands the modified subject QP in the embedded clause, thus having scope over it, yielding a single answer. Pair list answer is impossible, since the wh-word is bound by the QM. Consider the other LF representation in (24b), where the modified QP c-commands and thus binds the wh-word. The wh-word, being bound by the modified QP can never be definite thus leading to the lack o f a pair list answer. Single answer is out o f the question to begin with, since the indefinite wh-w ord is not bound by the QM. Since one o f the two LF representations in (24) for the sentences in (22) yields a single answer, the grammaticality of these sentences follows, given the proposal in (16). The LF representation for the wh- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 question in (23) with a propositional adjunct wh-word will be roughly the one in (25) in the present system: (25) [C P naze, [c QM, [IP NP-TOP [r tj [vr V [C P [ ip QP-NOM [V P V t, ]]]]]]]] The QM in (25) is raised into the head o f CP from its base-generated position in the head o f IP at LF, driven by the need to enter into a spec-head agreement with the propositional adjunct wh-word in Spec o f CP at LF. The wh-adjunct naze (why) c-commands the embedded modified subject QP, thus having scope over the latter, yielding a single answer. The question for the pair list answer never arises, given that the adjunct wh-word can never be decomposed into wh-operator and indefinite and that a modified QP can never induce a pair list answer. The grammaticality again follows given the proposal in (16). The asymmetry and symmetry o f scope taking as exhibited by the two types o f wh-words in simplex questions in (13-14) and the complex questions in (22-23) can be further confirmed by wh-questions with a negative polarity item (NPI, henceforth). Consider the following simplex questions with an NPI in the matrix subject position as in (26-27) and complex questions with an NPI in the embedded subject position in (28-29): 4 (26) a. ?*Dare-mo dare-o syootasi-na-katta-ka? anyone whom-ACC invite-NOT-QM ‘Who did no one invite?’ 4 See Lee and Tomioka (2002) for the original observation of the amelioration of so called intervention effects of argument wh-words in the embedded contexts as in (28). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 175 b. ?*Dare-mo tabako-o doko-de suwa-na-katta-no? anyone-NOM tobacco-ACC where-at smoke-NOT-QM ‘W here did no one smoke?’ c. ?*Dare-mo itsu kaera-na-katta-no? anyone-NOM when-ACC leave-NOT-QM ‘W hen did no one leave?’ d. ?*Dare-mo dooyatte ko-na-katta-no? anyone-NOM how come-NOT-QM ‘W hat is the means such that no one came by x?’ e. ?*Dare-mo Tom-nituite doo ittei-na-katta-no? anyone-NOM T-ACC how saying-NOT-QM ‘W hat is the manner x such that no one was talking about Tom in x?’ (27) Dare-mo Tom-o naze syootaisi-na-katta-no? anyone T-ACC why invite-NOT-QM ‘Why did no one invite Tom ?’ (28) a. (?)Kimi-wa [cp dare-mo dare-o shootaisi-na-katta-to] omott-eru-no? you-TOP anyone whom-ACC invite-NOT-COM P think-QM ‘W ho do you think no one invited?’ b. (?)Kimi-wa [cpdare-mo doko-de tabako-o suwa-na-katta-to] omott-eru-no? you-TOP anyone where-at tobacco-ACC smoke-NOT-COMP think-QM ‘W hat is the place x such that you think no one smoked at x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 176 c. (?)Kimi-wa [ c p dare-mo itsu kaera-na-katta-to] omott-eru-no? you-TOP anyone when leave-NOT-COM P think-QM ‘W hat is the time x such that you think no one left at x?’ d. (?)Kimi-wa [ c p dare-mo dooyatte kona-katta-to] omott-eru-no? you-TOP anyone how come-NOT-COM P think-QM ‘W hat is the means x such that you think no one came by x?’ e. (?)?Kimi-wa [C p dare-mo Tom-nituite doo ittei-na-katta-to] omott-eru-no? you-TOP anyone Tom-about how saying-NOT-COM P think-QM ‘W hat is the manner x such that you think no one was talking about Tom in x?’ (29) Kimi-wa [C pdare-mo Tom-o naze maneka-na-katta-to ] omott-eru-no? you-TOP anyone T-ACC why invite-NOT-COM P think-QM ‘W hat is the reason x such that you think no one invited Tom for x?’ The sentences in (26) are marginal, only allowing a functional answer, while the sentence in (27) is grammatical, admitting a single answer. The contrast in the sentences in (26-27) disappears in the sentences (28-29), all o f which admit a single answer and hence are grammatical. The LF representation for the sentences in (26) will be the one in (30a) and (30b), respectively, depending on whether the QM or the QP binds the wh-word. (30) a. [IP NPI [ r QM, [V P V [D P[N P W H-word;]] ]]] b . [ i P NPIj [ [ > QM; [ v p V [ d p [ n p WH-wordj]]i ]]] Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 177 Consider the LF representation in (30a), first. The wh-word is acting as an indefinite bound by the QM in the head o f IP thus is construed as a wh-interrogative. The wide scope o f the wh-word over the NPI, i.e., the single answer is impossible, since the QM, a wh-operator cannot c-command the NPI in the subject position. The pair list answer is simply impossible, since the wh-word is bound by the QM. In the other LF representation in (30b), pair list answer is not possible, since the NPI is not definite although it binds the indefinite wh-word. Single answer is out o f the question, since the QM does not bind the indefinite wh-word thus cannot be construed as a wh-interrogative. Since neither LF representations in (30) for the sentences in (26) can yield a pair list answer or a single answer, these sentences are correctly predicted to be marginal in grammaticality, given the proposal in (16). Turning to the LF representation below in (31) for the sentence in (27), the single answer obtains, since the propositional adjunct wh-word in Spec of CP, which enters into a spec- head agreement w ith the QM raised into the head o f CP from its base position, can c- command and scope over the NPI, yielding a single answer, thus its grammaticality follows, again given the proposal in (16). Also note that pair list answer is out o f the question to begin with, since the adjunct wh-word naze (why) is not an indefinite. (31) [ cr naze, [c QMj [ IP NPI [r tj h ]]]] The LF structures for the sentences in (28) will be roughly the ones in (32a) and (32b), respectively in the present system, depending on whether the QM or the QP binds the indefinite wh-word: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 (32) a. [IP NP-TOP [r QM, [V P V [ c p [ ip NPI [V P V [D P [ N P WH-word,]]]]]]]] b. [iP NP-TOP QMi [V P V [ CP [iP NPI, [V P V [ d p [ n p WH-wordj]]i]]]]]] In (32a), the QM in the head o f matrix IP binds the indefinite wh-word and hence the indefinite wh-word is construed as a wh-interrogative. The QM c-commands the NPI in the embedded clause, thus having scope over the latter, yielding a single answer. Pair list answer is not a possibility, since the wh-word is bound by the QM. Consider now (32b). The single answer cannot obtain since the indefinite wh-word is not bound by the QM. Pair list answer is not possible since the NPI is not definite. Since the single answer obtains in one o f the LF representations in (32) for the sentences in (28), the grammaticality follows, given the proposal in (16). Now, the LF for (29) will be roughly the one in (33). (33) [ cp w q yi [c QM, [IP NP-TOP [r t, [V P V [C P [,P NPI [V P V t, ]]]]]]]] In (33), the QM is raised into the head o f CP at LF, driven by the need to enter into a spec- head agreement relation with the propositional adjunct wh-word in Spec o f CP. The propositional adjunct wh-word way (why) in (33) c-commands the NPI in the embedded Spec of IP, thus having scope over the latter, yielding a single answer. Pair list answer is simply out o f the question, given that the adjunct wh-word cannot be an indefinite. The grammaticality o f the sentence in (29) also follows, again given the proposal in (16). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 4.3. Chinese Consider the following simplex wh-questions in Chinese where a modified subject QP is followed by a wh-word: (34) a. ?*zhi shao you liang ge ren yaoqing-le shei? at least have two m en invite-ASP whom ‘Who did at least two men invite?’ b. ?*zhi shao you liang ge ren zai nali xi yan-le? at least have two men at where smoke-ASP ‘W here did at least two men smoke?’ c. ?*zhi shao you liang ge ren shenmeshihou zou-le? at least have two men when leave-ASP ‘W hen did at least two men leave?’ d. ?*zhi shao you liang ge ren zenmeyang lai-de? at least have two men how come-ASP ‘W hat is the means x such that at least two men came by x?’ e. ?*zhi shao you liang ge ren zenmeyang shuo Jack? at least have two men how say Jack ‘W hat is the manner x such that at least two men talked about Jack in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 180 (35) (?)?zhi shao you liang ge ren w eishenm e chi dao-le? at least have two men why late come-ASP ‘W hy did at least two men come late?’ As shown above, there exists a subtle contrast in grammaticality in simplex wh-questions: Sentences with a modified QP followed by a nonpropositional wh-word, as in (34), are very marginal, admitting neither a pair list answer nor a single answer but only a functional answer.5 The same holds when the modified QP zhi shao you liang ge ren (at least two men) is replaced with other modified QPs, such as bu chao guo san ge ren (less than three men) or zhenghao lian-ge ren (exactly two m en).6 Interestingly, however, the grammaticality o f the sentences, improve subtly when the modified subject QP co-occurs with a propositional wh- word, as in (35), admitting a single answer. The current proposal for asymmetric scope taking between the two types o f wh-words can be extended to account for the subtle but interesting asymmetry o f scope exhibited in (34-35) regarding the w ide scope construal o f a wh-word over a QP. In section 4. 1, the position o f the QM in Chinese was discussed, and I tentatively concluded that the QM is base-generated somewhere in the CP field, given that it is a free standing morpheme and that it can be freely 5 1 would like to thank the people who offered their intuitions regarding examples like the ones in (34). James Huang, Audrey Li, Feng hsi Liu, Bella Feng, Lingyun Ji, Lian-hee Wee, Liang Chen, among others. It should be noted that the judgments are not always clear cut. Audrey Li (personal communication) reports that sentences as in (34) are grammatical, and James Huang (personal communication) notes that they are slightly deviant. 6 It should be noted that most Chinese native speakers whom I consulted adopted a ‘scrambled’ wh- question when it comes to making the kind of inquiry intended as in (34) with a modified subject QP. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 deleted, in contrast w ith the QM in Japanese and Korean. I could not yet determine where in the CP field the QM appears, given the proposal in Rizzi (2000), according to which the CP field is comprised o f a focus projection, topic projection and complementizer projection. Suppose the subject QP in Chinese is also sitting in Spec IP like Korean and Japanese. Then the subject QPs in (34) will also be c-commanded by the QM, which is somewhere in the CP field, and as such the marginality o f the sentences w ould not be expected. Given the topic prominence character o f Chinese, let us instead suppose that the subject QPs in (34) should be in the Spec o f Topic Phrase (Spec of TOPP, henceforth), and that the QM is base generated somewhere in the CP field from which it cannot c-command the subjects, i.e., in the head o f TOPP. Then the marginality o f the sentences in (34) will follow. Following the discussions o f the structure o f wh-words in English and Korean in Chapter 2, let us assume that Chinese nonpropositional wh-words also have a DP structure. Since the QM in Chinese is optional in the simplex wh-questions, we need to posit an abstract QM. The LF representations for the sentences in (34) will be the ones in (36a) and (36b), respectively, with irrelevant details suppressed, depending on whether the QM or the QP binds the wh-word. (36) a. [ t o p p subject QP [ t o p ’ QMj [ v p V [ d p [ n p WH-wordj]] ]]] b. [ t o p p subject QPj [ t o p ’ QMj [ v p V [ D p [ n p WH-wordj]]j ]]] In the LF (36a), the QM binds the indefinite wh-word and the indefinite wh-word is construed as a wh-interrogative. The QM, however, cannot c-command the modified subject Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182 QP in Spec o f TOPP, thus cannot take scope over it, hence yielding no single answer. Pair list answer is impossible to begin with, given that the wh-word is bound by the QM. Consider the other LF in (36b). Single answer is out o f the question to begin with, since the indefinite wh-word is not bound by the QM. Pair list answer is impossible since the QP binding the wh-word is a modified QP, which cannot be definite. The marginality in (34) thus follows, given the proposal in (16). The improved grammaticality o f the sentence in (35) also follows, given the LF representation in (37), since the propositional adjunct wh-word in Spec o f CP, which enters into a spec-head agreement relation with the QM raised into the head o f CP from its base- position, will be able to c-command the modified subject QP in Spec o f TOPP. (37) [C P weishenme} [c> QM, [T O pp subject QP [top- t, [V P V ( ]]]]] Although our proposal for the Chinese QM as being base-generated in the head o f TOPP with the subject QP in its Spec position needs independent empirical support, the proposal at least enables us to correctly account for the asymmetry o f scope (hence grammaticality) in the simplex wh-questions in (34) and (35). The subtle asymmetry o f scope between the two types of wh-words in (34-35) again disappears in complex wh-questions with a modified subject QP in the embedded clause followed by a wh-word as in (38-39), although the contrast is very subtle. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 (38) a. (?)?ni renwei [zhi shao you liang ge ren yaoqing-le shei]? you think at least have two men invite-ASP whom ‘Who do you think at least two men invited?’ b. (?)?ni renwei [zhi shao you liang ge ren zai nali xi yan-le]? you think at least have two men at where smoke-ASP ‘Where do you think at least two men smoked?’ c. (?)?ni renwei [zhi shao you liang ge ren shenmeshihou zou-le] ? you think at least have two men when leave-ASP ‘W hen do you think at least two men left?’ d. (?)?ni renwei [zhi shao you liang ge ren zenmeyang lai-le]? you think at least have two men how come-ASP ‘W hat is the means such that you think at least two men came by x?’ e. (?)?ni renwei [zhi shao you liang ge ren zenmeyang shuo Jack]? you think at least have two men how say Jack ‘W hat is the manner x such that you think at least two men talked about Jack in x?’ (39) (?)?ni renwei [ zhi shao you liang ge ren weishenme chi lai-de]? you think at least have two men why late come-ASP ‘W hat is the reason x such that you think at least two men came late for x?’ Complex wh-questions in (38-39) can admit a single answer, which means both types o f wh- words can take scope over the Q P .7 It should be noted that we need to posit an abstract QM 71 should note that some native speakers I consulted report that the sentences in (38) are no better than Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 184 for the LF representations for the complex wh-questions in (38-39) since the QM is not realized phonologically in Chinese, unlike Japanese and Korean. The LF representations for the sentences in (38) will be the ones in (40a) and (40b), respectively depending on whether the QM or the QP binds the indefinite wh-word. (40) a. [ t o p p NP [ t o p ’ QMi [ v p V [ c p [ QP [ v p V [ d p [ n p WH-wordj]] ]]]]]] b. [ t o p p NP [ t o p 1 QM, [ v p V [ c p [ QPj [ v p V [ d p [ n p WFI-wordj]]i ]]]]]] In the LF (40a), the QM in the matrix clause binds the indefinite wh-w ord in the embedded clause hence the indefinite wh-word is construed as a wh-interrogative. The QM c- commands the embedded modified subject QP, thus having scope over it, yielding a single answer. Pair list answer is impossible, since the indefinite wh-word is bound by the QM. Consider the other LF representation in (40b), where the modified QP c-commands and thus binds the indefinite wh-word. The wh-word, being bound by the modified QP can never be definite thus leading to the lack o f a pair list answer. Single answer is not a possibility, since the indefinite wh-w ord is not bound by the QM. Since one of the two LF representations in (40) for the sentences in (38) can induce a single answer, the improved grammaticality of these sentences follow, given the proposal in (16). The LF representation for the wh-question in (39) with a propositional adjunct wh-word will be roughly the one in (41) in the present system: (41) [ c p weishenme, [c QMj [toppNP [T O P ’ tj [ v p [ c p [ QP [ v p h ]]]]]]]] the ones in (34). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185 The propositional adjunct wh-word, weishenme (why) in Spec o f CP, which enters a spec- head agreement with the QM raised into the head o f CP at LF, c-commands the modified subject QP in the embedded clause, thus having scope over the latter, yielding a single answer. The question for the pair list answer never arises given that the adjunct wh-word is not an indefinite. Given the proposal in (16), the grammaticality o f (39) also follows. Thus the scope pattern shown in (34-35) and (38-39) points to the same generalization at work in Korean and Japanese: Propositional adjunct wh-words and nonpropositional wh- words exhibit asymmetry o f scope in simplex questions, while the asymmetry in scope between them disappears in complex questions. In other words, a propositional adjunct wh-word can take scope over the subject QP, while a nonpropositional wh-word cannot, in simplex questions. In complex questions wh-words o f both types can take scope over the embedded QP. To the extent that the present observation on the asymmetry and symmetry o f scope in simplex and complex wh-questions in Chinese is correct, it seriously undermines the uniform LF wh-movement hypothesis o f Huang (1982), according to which wh-words in Chinese uniformly undergo movement into Spec o f CP at LF to take scope over the subject QP, thus predicting that the even sentences in (34) should all admit a single answer, contrary to the facts. Also uniform wh-in-situ approach (Aoun and Li 1993b) will likewise fail to account for these scope facts, since the operator in Spec o f CP in their analysis would c-command the subject QP, and thus such an account incorrectly gives wide scope to the wh-word, incorrectly predicting a single answer in the sentences in (34). If our proposed account for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 186 the scope facts exhibited in (34) is correct, it has additional implications for the scope facts o f sentences such as (42): (42) mei-ge-ren dou mai-le shenm e? every-CL-person all bought-ASP what ‘W hat did everyone buy?’ (Huang 1982: 263) The above sentence is grammatical, allowing a single answer, a pair list answer (Aoun and Li 1993a) and a functional answer (Audrey Li, personal communication) (cf. Huang 1982). The availability o f a single answer for the sentence in (42) with a universal subject QP has been taken as support for wide scope o f the wh-word over the subject QP, and hence the uniform LF wh-movement hypothesis (Huang 1982, Aoun and Li 1993 a). However, given the discussion above regarding the data in (34), the source o f the single answer in (42) should not be wide scope o f the wh-word over the subject QP, but should be a special case of the pair list answer. The following example (43) with daduoshu NP (most NP) as subject QP confirms our claim. (43) daduoshu xuesheng baifang-le shei? most students visited-ASP whom 'Who did most o f the students visit?' Let us again suppose as a relevant context the department orientation for the incoming class o f graduate students, A, B, C, D, E, F, and G. Let us further suppose that students A, B, C, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 187 D visited Prof. John and students C, D, E, F visited Prof. Mary. Given the above context, suppose the speaker asks the question in (43). The hearer cannot answer by saying ‘Prof. John and Prof. M ary’, according to the native speakers whom I consulted. This means that the set denoted by the subject QP in (43) can never co-vary with the choice of a professor, which means that the wh-word in (43) cannot take scope over the subject QP. 5. Conclusion To conclude, I observed that in Japanese and Chinese, just as in Korean, there exists asymmetry and symmetry o f scope between the propositional adjunct wh-words and nonpropositional wh-words when it comes to QP and wh-word scope interaction. In simplex wh-questions, a nonpropositional wh-word cannot take scope over the matrix subject QP, while the propositional wh-word can. Meanwhile, in complex wh-questions, the asymmetry o f scope essentially disappears, in that both types o f wh-words can take scope over the embedded subject QP. The asymmetry and symmetry o f scope can be accounted for in a natural way once it is accepted that the two types o f wh-words take scope in a fundamentally different way, i.e. , unselective biding vs. movement, and that the QM is base-generated in the head o f IP in Japanese, but in the head o f TOPP, with the subject in its Spec position in Chinese. Given the topic prominence o f Chinese the latter claim seems initially plausible, although more research is needed to determine whether it is in fact tenable. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 188 CHAPTER 4 Asymmetry of Locality Effects in Islands 1. Introduction This chapter will illustrate and account for the asymmetry between propositional adjunct wh- words and nonpropositional wh-words with respect to locality effects induced by scoping out o f standard syntactic islands. I will review the Complex Noun Phrase Island construction (Ross 1967), Sentential Subject Island construction (Ross 1967), and Adjunct Island construction (Huang 1982), and show that none o f the previous proposals in the literature, which I review in this chapter, i.e., ECP approach o f Huang (1982), unselective binding approach o f Tsai (1994) and pied-piping approach o f Nishigauchi (1990) can be the correct explanation for the asymmetry o f locality effects between the two classes o f wh-words in Korean type languages. I will show that the asymmetric scope taking mechanism o f nonpropositional wh-words and propositional adjunct wh-words, as proposed in Chapter 2, can give a straightforward account for the asymmetry o f locality effects between them to be observed in this chapter, assuming subjacency violation in the sense o f Chomsky (1986) directly leads to ungrammaticality: Nonpropositional wh-words, being indefinites, are interpreted in situ by being unselectively bound by the QM in the matrix clause, vacuously satisfying subjacency, while propositional adjunct wh-words, which are not indefinites, should undergo movement into Spec o f CP, driven by the need for proper interpretation, forming an operator variable chain, hence violating subjacency. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 189 2. Locality Effects in Islands It has been a standard observation that wh-movement is subject to locality, descriptive constraints such as the Complex Noun Phrase Island (Ross 1967), Sentential Subject Island (Ross 1967), and Adjunct Island (Huang 1982) being subsumed under the explanatory umbrella o f subjacency (see Chomsky 1986). The examples in (1) involve typical Complex Noun Phrase Island construction (strong CNPI, henceforth), while the ones in (2) involve appositive construction (weak CNPI, henceforth). (1) a. ?*W hat did John meet a man who fixed? ‘W hat is the thing x such that John met a man who fixed x?’ b. *How did John meet a man who fixed the car? ‘W hat is the means x such that John met a man who fixed the car by x?’ (Chomsky 1986: 35) (2) a. ??Whom did you hear a rum or that John met? ‘Who is the person x such that you heard a rumor that John met x?’ b. *How did you hear a rumor that John fixed the car? ‘W hat is the means x such that you heard a rumor that John fixed the car by x T The examples below in (3) involve Adjunct Island, while the ones in (4) involve Sentential Subject Island. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 190 (3) a. ?*W hat did you leave before fixing? ‘W hat is the thing x such that you left before fixing x?’ b. *How did you leave before fixing the car? ‘W hat is the means x such that you left before fixing the car by x? (Chomsky 1986: 32) (4) a. ?*W hat did that John fixed surprised you? ‘W hat is the thing x such that that John fixed x surprised you?’ b. *How did that John fixed the car surprised you? ‘W hat is the means x such that that John fixed the car by x surprised you?’ According to Chomsky (1986), the marginality o f the sentences in (la-4a) with argument wh- words essentially has to do with subjacency: At least one barrier is crossed during the derivation from its base position into its surface position. The ungrammaticality o f the sentences in (lb-4b) w ith adjunct wh-words is attributed to the incurring o f an ECP violation on top o f the subjacency violation: At some point o f the derivation at least one barrier intervenes between the adjunct wh-word and its trace, which disrupts local dependency (antecedent-govemment) between the adjunct wh-word and the trace. In a nutshell, subjacency violation, even involving one barrier by the adjunct wh-words, directly leads to ungrammaticality while the same violation by the argument wh-words induces marginality in grammaticality. W ith the effect o f subjacency violation regarding the adjunct wh-word in mind, let us turn to the locality effects o f wh-words in Korean type languages. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 191 Huang (1982) claims that wh-words in Chinese undergo movement at LF in a uniform manner. Drawing from data as in (5-6), he claims that subjacency cannot be a valid locality constraint for these wh-words, but that ECP is the valid constraint active at LF. His claim for ECP as the only valid constraint at LF is based on the following contrast in grammaticality in (5-6) in Chinese: (5) [shei xie de shu] zui youqu? who write book most interesting ‘Who is the person x such that books x wrote are most interesting?’ (Huang 1982: 526) (6) *[ta weishenme xie de shu] zui he why write book most ‘W hat is the reason x such that books he wrote (Huang 1982: 527) The sentence in (5) with an argument wh-word inside the strong CNPI is perfect while the one in (6) with an adjunct wh-word inside the same island is ungrammatical. If subjacency is the locality constraint active in Chinese, the grammaticality of the sentence in (5) should be on a par with the one for (la ) in English, as repeated in (7), which also involves a strong CNPI. (7) ?*W hat did John m eet a man who fixed? ‘W hat is the thing x such that John met a man who fixed x?’ youqu? interesting for x are most interesting?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 192 Huang thus claims that the contrast in grammaticality in (5-6) follows from ECP. To be specific, Huang adopts the disjunctive ECP by Aoun and Sportiche (1981), which says that nonpronominal empty categories should be properly governed either by lexical government or by antecedent government (also see Chomsky 1981, 1986).1 The trace o f the argument wh-word in its base-position in (5) is governed by the verb, hence ECP is satisfied. Meanwhile, the trace o f the adjunct wh-word in its base position in (6) is not lexically governed by the verb, and thus antecedent government should be satisfied, which, however, is impossible due to the maximal projections intervening between the adjunct wh-word in the matrix CP and its trace, hence ungrammaticality.2 Keeping in mind Huang’s ECP, which is based on the dichotomy o f argument wh-word and adjunct wh-word under the uniform LF wh-movement hypothesis, let us turn to Korean, first. 1 The disjunctive ECP by Aoun and Sportiche (1981) is as follows (Huang 1982:471): A properly governs B if and only if A governs B and (a) A is a lexical category, or (b) A is co-indexed with B. Please note that proper government is a subcase of government, and that government does not obtain when a maximal projection boundary intervenes between the governor and the govemee (also see Chapter 2. fii. 18). 2 Huang (1982) does not discuss the status of the intermediate trace (cf. Lasnik and Saito 1984). Given that wh-words undergo movement in a successive cyclic way into the final landing site, via comp-to comp movement in his system (Huang 1982: 450ff), the ungrammaticality of the sentence in (6) poses a problem to his account, since the trace is antecedent governed by the antecedent in the comp position and movement into matrix position from the embedded comp is innocuous due to the invalidity of subjacency at LF. The LF representation for (6) will be the following according to him (see Huang 1982: 494), using a head initial notation: [ s’ why [s V [ np [ s’ f i [ s h ]]]]] The problem may be resolved by assuming a null operator in the embedded comp position, as noted in Nishigauchi (1990: 45), such that the branching comp blocks antecedent government, assuming comp indexing mechanism (Aoun Homstein and Sportiche 1981). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 193 2.1. Korean Let us consider the paradigm involving the classical Complex Noun Phrase Island (Ross 1967) first. 2.1.1. Complex Noun Phrase Island The examples below in (8-9) involve strong CNPIs, while the examples in (10-11) involve weak CNPIs. (8) a. Ne-nun [nwu-ka ssun chayk-ul] ilkess-ni? you-TOP who-NOM wrote book-ACC read-QM ‘W hat is the person x such that you read a book x w rote?’ b. Ne-nun [enu cakka-ka ssun chayk-ul] ilkess-ni? you-TOP which writer-NOM wrote book-ACC read-QM ‘W hich writer x is such that you read a book x w rote?’ c. Ne-nun [John-i eti-se ssun chayk-ul] ilkess-ni? you-TOP J-NOM where-at wrote book-ACC read-QM ‘What is the place x such that you read a book John wrote at x?’ d. Ne-nun [John-i encey ssun chayk-ul] ilkess-ni? you-TOP J-NOM when wrote book-ACC read-QM ‘W hat is the time x such that you read a book John wrote at x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 194 e. Ne-nun [cheycwung-i elma nakanun salam-ul] coaha-ni? you-TOP weight-NOM how much weigh man-ACC like-QM ‘W hat is the weight x such that you like a man who weighs x?’ f. Ne-nun [John-i ettehkey kulin kulim-ul] coaha-ni? you-TOP J-NOM how painted painting-ACC like-QM ‘W hat is the means x such that you like pictures John drew by x?’ g. Ne-nun [os-ul ettehkey ipnun salam-ul] coaha-ni? you-TOP clothes-ACC how wear man-ACC like-QM ‘What is the manner x such that you like a man who gets dressed in x ?’ (9) *Ne-nun [John-i way ssun chayk-ul] ilkess-ni? you-TOP J-NOM why wrote book-ACC read-QM ‘W hat is the reason x such that you read a book John wrote for x?’ (10) a. Ne-nun [nwu-ka M ary-lul coahantanun somwun-ul] tuless-ni? you-TOP who-NOM M ary-ACC like rumor-ACC heard-QM ‘W hat is the person x such that you heard a rumor that x likes M ary?’ b. Ne-nun [enu kyoswu-ka M ary-lul coahantanun somwun-ul] tuless-ni? you-TOP which professor-NOM M-ACC likes rumor-ACC heard-QM ‘W hich professor x is such that you heard a rumor that x likes M ary?’ c. Ne-nun [John-i eti-se M ary-lul mannasstanun somwun-ul] tuless-ni? you-TOP J-NOM where-at M-ACC met rumor-ACC heard-QM ‘W hat is the place x such that you heard a rumor that John met M ary at x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 195 d. Ne-nun [John-i encey Mary-lul mannasstanun somwun-ul] tuless-ni? you-TOP J-NOM when M-ACC met rumor-ACC heard-QM ‘W hat is the time x such that you heard a rumor that John met Mary at x?’ e. Ne-nun [John-i cheycwung-i elma nakantanun somwun-ul] tuless-ni? you-TOP J-NOM weight-NOM how much weighs rumor-ACC heard-QM ‘W hat is the weight x such that you head a rumor that John weighs x?’ f. Ne-nun [John-i ettehkey USC-ey wasstanun somwun-ul] tuless-ni? you-TOP J-NOM how USC-at came rumor-ACC heard-QM ‘W hat is the means x such that you heard a rumor that John came to USC by x?’ g. Ne-nun [M ary-ka ne-eytayhay ettehkey malhakotanintanun somwun-ul] you-TOP M -NOM you-about how saying rumor-ACC tuless-ni? heard-QM ‘What is the manner x such that you heard a rumor that Mary is talking about you in x ?’ (11) *Ne-nun [John-i way USC-ey kasstanun somwun-ul] tuless-ni? you-TOP J -NOM why USC-to went rumor-ACC heard-QM ‘W hat is the reason x such that you heard a rumor that John went to USC for x?’ It should be noted that Huang means by argument wh-words, wh-words which can serve as an argument o f a lexical category such as V, P, N or A, while by adjunct wh-words he means wh-words which cannot do so (Huang 1982: 528-530). Thus nwukwm (who), mwues (what), enu-N (which N), and elma (how much) will be bona fide argument wh-words. W hen it Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 196 comes to wh-words such as eti (where) and encey (when), they can be viewed as argument wh-words since they can come together with a postposition, showing that they can be arguments o f a lexical category. 3 Ettehkey (how) and way (why) are genuine adjunct wh- words, since they cannot be arguments o f a lexical category, according to the argument and adjunct criterion in Huang (1982). The sentences in (9) and (11) with the adjunct wh-word way (why) are ungrammatical, as compared with the sentences in (8) and (10), with the rest o f the wh-words, which are grammatical. The contrast in grammaticality along the lines of adjunct wh-w ord way (why) and the rest o f wh-words may be rather surprising to the reader, who is likely to be fam iliar with the classical dichotomy presented in Huang (1982), according to which argument wh-words are immune to syntactic islands while adjunct wh- words, including how, are not. Given that ettehkey (how) either under means construal in (8f) and (lOf) or manner construal in (8g) and (lOg) is a bona fide adjunct wh-word under H uang’s definition o f argument/adjunct, the grammaticality of these sentences in (8f-g) and (lOf-g) poses a problem to H uang’s ECP account based on argument/adjunct dichotomy. With this problem in mind, let us continue on to examples involving Adjunct Islands. 2.1.2. Adjunct Island Consider the examples below in (12-13), which have to do with Adjunct Islands (Huang 1982): 3 The following sentence shows that encey (when) can be the argument of a postposition: Ne-nun encey-pwute Mary-ul mannass-ni? you-TOP when-since M-ACC dated-QM ‘Since when have you been dating Mary?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 197 (12) a. [Nwu-ka o-myen] chohkeyss-ni? who-NOM com e-if good-QM ‘Who is the person x such that if x comes it will be good?’ b. [Enu kyoswu-ka o-myen] cohkeyss-ni? which professor-NOM com e-if good-QM ‘W hich professor x is such that if x comes it will be good?’ c. [John-i M ary-lul eti-se manna-myen] cohkeyss-ni? J-NOM M ary-ACC where-at m eet-if good-QM ‘W hat is the place x such that if John meets Mary at x it will be good?’ d. [John-i M ary-lul encey manna-myen] cohkeyss-ni? J-NOM M ary-ACC when m eet-if good-QM ‘W hat is the time x such that if John meets M ary at x it will be good?’ e. [Nay-ka cheycwung-i elma naka-myen] cohkeyss-ni? I-NOM weight-NOM how m uch w eigh-if good-QM ‘W hat is the weight x such that if I weigh x it will be good?’ f. [John-i ettehkey o-myen] cohkeyss-ni? J-NOM how com e-if good-QM ‘W hat is the means x such that if John comes by x it will be good?’ g. [Nay-ka colepsik-ey os-ul ettehkey ipu-myen] cohkeyss-ni? I-NOM commencement-for clothes-ACC how w ear-if good-QM ‘W hat is the manner x such that if I get dressed in x for the commencement it will be good?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 198 (13) * [John-i w ay o-myen] cohkeyss-ni? J-NOM why com e-if good-QM ‘W hat is the reason x such that if John comes for x it will be good?’ As one can notice, the contrast in grammaticality in examples involving Adjunct Islands in (12-13) above is again along the lines o f way (why) on the one hand, and the rest o f wh- words on the other, posing a problem to H uang’s ECP account, which is based on the argument/adjunct dichotomy under the uniform LF wh-movement hypothesis. Lastly, we consider examples involving Sentential Subject Islands. 2.1.3. Sentential Subject Island Consider the examples in (14-15), involving Sentential Subject Islands. (14) a. [Nwu-ka onunkes-i] cohkeyss-ni? who-NOM come-NOM good-QM ‘Who is the person x such that that x comes is good?’ b. [Enu kyoswu-ka onunkes-i] cohkeyss-ni? which professor-NOM come-NOM good-QM ‘W hich professor x is such that that x comes is good?’ c. [John-i eti-lo onunkes-i] cohkeyss-ni? J-NOM where-to come-NOM good-QM ‘W hat is the place x such that that John comes to x is good?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 199 d. [John-i encey onunkes-i] cohkeyss-ni? J-NOM when come-NOM good-QM ‘W hat is the time x such that that John comes at x is good?’ e. [Nay-ka cheycwung-i elma nakanunkes-i] cohkeyss-ni? I-NOM weight-NOM how much weigh-NOM good-QM ‘W hat is the weight x such that that I weigh x is good?’ f. [John-i ettehkey onunkes-i] cohkeyss-ni? J-NOM how come-NOM good-QM ‘W hat is the means x such that that John comes by x is good?’ g. [Nay-ka colepsik-ey os-ul ettehkey ipnunkes-i] cohkeyss-ni? I-NOM commencement-at clothes-ACC how wear-NOM good-QM ‘W hat is the manner x such that that I get dressed in x for the commencement is good?’ (15) * [John-i way onunkes-i] cohkeyss-ni? J-NOM why come-NOM good-QM ‘W hat is the reason x such that John comes for x is good?’ As one can notice, the contrast in grammaticality in the examples in (14-15) involving Sentential Subject Islands is again along the dimension o f way (why) on the one hand, and the rest o f wh-words on the other, again posing a problem to Huang’s ECP account, which is based on the argument/adjunct wh-word dichotomy under the uniform LF wh-movement hypothesis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 200 To summarize, the Korean data that I have examined thus far with respect to CNPIs, Adjunct Islands and Sentential Subject Islands lead us to conclude that the classical ECP account based on adjunct and argument dichotomy under the uniform wh-movement hypothesis cannot be a correct one, at least when it comes to Korean. W hy are sentences with the adjunct wh-w ord ettehkey (how), w ith either a manner or means construal, still grammatical? Given the argument-hood criterion o f Huang (1982), ettehkey (how) should be genuine adjunct wh-word. Since it is an adjunct, it should antecedent govern its trace, which, however, is impossible due to the maximal projections intervening between them. Thus the prediction is that the grammatical status o f the sentences with the adjunct etteheky (how) embedded in an island should be on a par with that o f the corresponding sentences with the other adjunct wh-word way (why). This prediction, as we saw, is simply incorrect. Next let us turn to Japanese and Chinese to see whether wh-words in these languages exhibit any asymmetry o f locality effects in Complex Noun Phrase Islands, Sentential Subject Islands, and Adjunct islands along the lines suggested in Huang (1982), i.e., argument wh- words on the one hand and adjunct wh-words on the other. 2.2. Japanese and Chinese To begin with, let us consider Complex Noun Phrase Island constructions in Japanese and Chinese. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 201 2.2.1. Complex Noun Phrase Island The data in (16-19) involve strong CNPIs in Japanese and Chinese. The examples in (16-17) are from Japanese, and those in (18-19) from Chinese. (1 6 )Japanese a. Kimi-wa [dare-ga kaita hon-o] yomimasita-ka? you-TOP who-NOM wrote book-ACC read-QM ‘Who is the person x such that you read a book x w rote?’ (Nishigauchi (1990:48) b. Kimi-wa [dono hito-ga kaita hon-o] yomimasita-ka? you-TOP which man-NOM wrote book-ACC read-QM ‘W hich person x is such that you read a book x w rote?’ c. Kim i-wa [kare-ga doko-de kaita hon-o] yomimasita-ka? you-TOP he-NOM where at wrote book-ACC read-QM ‘W hat is the place x such that you read a book he wrote at x?’ d. Kimi-wa [kare-ga itsu kaita hon-o] yomimasita-ka? you-TOP he-NOM when wrote book-ACC read-QM ‘W hat is the time x such that you read a book he wrote at x?’ e. Kimi-wa [taijuu-ga ikutsu aru hito-o] sukidesu-ka? you-TOP weight-NOM how much have man-ACC like-QM ‘W hat is the weight x such that you like a person who weighs x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 202 f. Kimi-wa [kare-ga dooyatte kaita e-o] kaimasita-ka? you-TOP he-NOM how painted painting-ACC bought-QM ‘What is the means x such that you bought a picture he drew by x?’ g. Kim i-wa [kimi-nituite doo itteiru hito-o] sukidesu-ka? you-TOP you-about how say man-ACC like-QM ‘W hat is the manner x such that you like a person who talks about you in x?’ (17) Japanese * Kimi-wa [kare-ga naze kaita hon-o] yomimasita-ka? you-TOP he-NOM why wrote book-ACC read-QM ‘W hat is the reason x such that you read a book he wrote for x?’ (Nishgauchi 1990: 41) (18) Chinese a. ni bijiao xihuan [shei xie de shu]? you more like who write book ‘Who is the person x such that you like better a book x w rote?’ b. ni bijiao xihuan [na-ge-ren xie de shu]? you more like which person write book ‘W hich person x is such that you like better a book x w rote?’ c. ni bijiao xihuan [John zai nali xie de shu]? you more like John at where write book ‘W hat is the place x such that you like better a book John wrote at x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 203 d. ni bijiao xihuan [John shenmeshihou xie de shu]? you more like John when write book ‘W hat is the time x such that you like better a book John wrote at x?’ e. ni bijiao xihuan [zhong ji bang de ren]? you more like amount how much person ‘W hat is the amount x such that you like a person who weighs x? f. ni bijiao xihuan [John zenmeyang hua de hua]? you more like John how paint painting ‘W hat is the means x such that you like better a painting John painted by x?’ g. ni bijiao xihuan [zenmeyang shuo ni de ren]? you more like how say you person ‘W hat is the manner x such that you like better a person who talks about you in x?’ (19) Chinese *ni bijiao xihuan [John weishenme xie de shu]? you more like John why write book ‘W hat is the reason x such that you like better a book John wrote for x?’ The examples as illustrated below in (20-23) involve weak CNPIs in Japanese and Chinese. The examples in (20-21) are from Japanese, and the examples in (22-23) are from Chinese, respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 204 (20) Japanese a. Kimi-wa [dare-ga M ary-ni atta toyuu uwasa-o] kiita-no? you-TOP who-NOM M -DAT met rumor-ACC heard-QM ‘Who is the person x such that you heard a rumor that x met M ary?’ b. Kimi-wa [dono hito-ga M ary-ni atta toyuu uwasa-o] kiita-no? you-TOP which person-NOM M-DAT met rumor-ACC heard-QM ‘W hich person x is such that you heard a rumor that x met M ary?’ c. Kimi-wa [John-ga M ary-ni doko-de atta toyuu uwasa-o] kiita-no? you-TOP J-NOM M -DAT where-at met rumor-ACC heard-QM ‘W hat is the place x such that you heard a rumor that John met Mary at x?’ d. Kimi-wa [John-ga M ary-ni itsu atta toyuu uwasa-o] kiita-no? you-TOP J-NOM M -DAT when met rumor-ACC heard-QM ‘W hat is the time x such that you heard a rumor that John met Mary at x?’ e. Kimi-wa [M ary-ga taijuu-ka ikutsu aru toyuu uwasa-o] kiita-no? you-TOP M -NOM weight-NOM how much have rumor-ACC heard-QM ‘W hat is the weight x such that you heard a rumor that Mary weighs x?’ f. Kimi-wa [John-ga dooyatte USC-ni hairim asita toyuu uwasa-o] kiita-no? you-TOP J-NOM how USC-DAT entered rumor-ACC heard-QM ‘W hat is the means x such that you heard a rumor that John was admitted to USC by x?’ g. Kimi-wa [M ary-ga John-nituite doo hanosita toyuu uwasa-o] kiita-no? you-TOP M -NOM J-about how saying rumor-ACC heard-QM ‘W hat is the manner x such that you heard a rumor that Mary talks about John in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 205 (21) Japanese *Kimi-wa [John-ka M ary-ni naze atta toyuu uwasa-o] kiita-no? you-TOP J-NOM M-DAT why met rumor-ACC heard-QM ‘W hat is the reason x such that you heard a rumor that John met M ary for x?’ (22) Chinese a. ni tingshuo quo [shei lai de xinwen] you hear who come news ‘Who is the person x such that you heard a rumor that x com es?’ b. ni tingshuo quo [na-ge-ren lai de xinwen] you hear which person come news ‘W hich person x is such that you heard a rumor that x com es?’ c. ni tingshuo quo [John zai nali xuexi de xinwen] you hear John at where study news ‘W hat is the place x such that you heard a rumor that John studies at x?’ d. ni tingshuo quo [John shenmeshihou lai de xinwen] you hear John when come news ‘W hat is the time x such that you heard a rumor that John comes by x?’ e. ni tingshuo quo [John zong ji bang de shuofa] you hear John amount how much story ‘W hat is the weight x such that you heard a rumor that John weighs x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 206 f. ni tingshuo quo [John zenmeyang lai de xinwen] you hear John how come news ‘W hat is the means x such that you heard a rumor that John comes by x?’ g. ni tingshuo quo [John zenmeyang shuo ni de gushi] you hear John how say you rumor ‘W hat is the manner such that you heard a rumor that John talks about you in x?’ (23) *ni tingshuo quo [John weishenme lai de xinwen] you hear John why come news ‘W hat is the reason x such that you heard a rumor that John comes for x?’ According to H uang’s argument-hood criterion (1982: 528-530), shei (who), na-ge-N (which N), and ji-CL (how much) are bona fide argument wh-words in Chinese. Nali (where) and shenmeshihou (when) are argument wh-words since they can come together with a preposition. Zenmeyang (how) and weishenme (why) are genuine adjunct wh-words since they cannot be an argument o f a lexical category, according to Huang (1982). W hen it comes to Japanese, dare (who), dono-N (which N), and ikutsu (how much) are bona fide argument wh-words. W hen it comes to wh-words such as doko (where) and itsu (when), they can be viewed as argument wh-words since they can come together with a postposition, showing that they can be arguments o f a lexical category. 4 Dooyatte (how), doo (how), and 4 Itsu (when) can be the complement of a postposition in Japanese as shown below. Itsu-kara natu yasumi-ga hajimatta-no? when-from summer vacation-NOM start-QM ‘When does the summer vacation start?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 207 naze (why) are genuine adjunct wh-words, since they cannot be arguments o f a lexical category. As one can notice, the contrast in grammaticality o f (16-23), however, is essentially along the lines o f naze (why) in Japanese and weishenme (why) in Chinese and the rest o f wh-words in these languages, again posing a problem to H uang’s ECP account. Next, let us turn to Adjunct Islands. 2.2.2. Adjunct Island Consider the examples in (24-27), involving Adjunct Islands in Japanese and Chinese. The examples in (24-25) are from Japanese and those in (26-27) from Chinese. (2 4 )Japanese a. [Dare-ga kuru-nara] iidesu-ka? who-NOM com e-if good-QM ‘W ho is the person x such that if x comes it will be good?’ b. [Dono hito-ga kuru-nara] iidesu-ka? which person-NOM com e-if good-QM ‘W hich person x is such that if x comes it will be good?’ c. [John-ga M ary-ni doko-de ae-ba] iidesu-ka? J-NOM Mary-DAT where-at m eet-if good-QM ‘W hat is the place x such that if John meets M ary at x it will be good?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 208 d. [John-ga M ary-ni itsu ae-ba] iidesu-ka? J-NOM M ary-DAT when m eet-if good-QM ‘W hat is the time x such that if John meets M ary at x it will be good?’ e. [ Boku-ga taijuu-ga ikutsu are-ba] iidesu-ka? I-NOM weight-NOM how much w eigh-if good-QM ‘W hat is the amount x such that if I weigh x it will be good?’ f. [John-ga dooyatte kuru-nara ] iidesu-ka? J-NOM how com e-if good-QM ‘W hat is the means x such that if John comes by x it will be good?’ g. [Kono kanji-wa doo kaita-ra ] iidesu-ka? that Chinese character-TOP how w rite-if good-QM ‘W hat is the manner x such that if I write the Chinese character in x it will be good?’ (25) Japanese * [John-ga naze kuru-nara] iidesu-ka? J-NOM why com e-if good-QM ‘W hat is the reason x such that if John comes for x it will be good?’ (26) Chinese a. [shei lai] ni hui gao xiang? who come you will be happy ‘Who is the person x such that if x comes you will be happy?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 209 b. [na-ge-ren lai] ni hui gao xiang? which person come you will be happy ‘W hich person x is such that if x comes you will be happy?’ c. [John qu nali lai] ni hui gao xiang? John to where come you will be happy ‘W hat is the place x such that if John comes to x you will be happy?’ d. [John shenmeshihou lai] ni hui gao xiang? John when come you will be happy ‘W hat is the time x such that if John comes at x you will be happy?’ e. [ni zhong ji bang] ni hui gao xiang? you weigh how much you will be happy ‘W hat is the weight x such that if you weigh x you will be happy?’ f. [John zenmeyang lai] ni hui gao xiang? John how come you will be happy ‘W hat is the means x such that if John comes by x you will be happy?’ g. [John zenmeyang shuo] ni hui gao xiang? John how say you will be happy ‘W hat is the manner x such that if John talks about you in x you will be happy?’ (27) Chinese *[John weishenme lai] ni hui gao xiang? John why come you will be happy ‘W hat is the reason x such that if John comes for x you will be happy?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 210 Again, the contrast in grammaticality o f the sentences in (24-27) shows that naze (why) and weishenme (why) stand apart from the rest o f wh-words in Japanese and Chinese, respectively, posing a nontrivial problem to ECP based account in Huang (1982) under the uniform LF wh-movement hypothesis. Lastly let us turn to Sentential Subject Islands. 2.2.3. Sentential Subject Island Consider the examples in (28-31) involving Sentential Subject Islands. The examples in (28- 29) are from Japanese and the ones in (30-31) from Chinese. (28) Japanese a. [D are-ga kuru no-ga] iidesu-ka? who-NOM comes-NOM good-QM ‘Who is the person x such that that x comes is good?’ b. [Dono hito-ga kuru no-ga] iidesu-ka? which man-NOM come-NOM good-QM ‘W hich person x is such that that x comes is good?’ c. [John-ga doko-ni kuru no-ga] iidesu-ka? John-NOM where-to comes good-QM ‘W hat is the place x such that that John comes to x is good?’ d. [John-ga itsu kuru no-ga] iidesu-ka? John-NOM when comes good-QM ‘W hat is the time x such that that John comes at x is good?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 211 e. [Boku-ga taijuu-ga ikutsu aru no-ga] iidesu-ka? I-NOM weight-NOM how much weigh-NOM good-QM ‘W hat is the amount x such that that I weigh x is good?’ f. [John-ga dooyatte kuru no-ga] iidesu-ka? John-NOM how comes good-QM ‘W hat is the means x such that that John comes by x is good?’ g. [Kono kanji-wa doo kaku no-ga] iidesu-ka? that Chinese character how write -NOM good-QM ‘W hat is the manner x such that that I write the Chinese character in x is good?’ (29) Japanese * [John-ga naze kuru no-ga] iidusu-ka? J-NOM why come-NOM good-QM ‘W hat is the reason x such that that John comes for x is good?’ (30) Chinese a. [shei lai] hao? who come good ‘Who is the person x such that that x comes is good?’ b. [na-ge-ren lai] bijiao hao? which-CL-person come more good ‘W hich person x is such that that that x comes is better?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 212 c. [John qu nali lai] bijiao hao? John to where come more good ‘W hat is the place x such that that John comes to x is better?’ d. [John shenmeshihou lai] bijiao hao? John when come more good ‘W hat is the time x such that that John comes at x is better?’ e. [wo zhong ji bang] bijiao hao? I amount how much more good ‘W hat is the weight such that that I weigh x is better?’ f. [John zenmeyang lai] bijiao hao? John how come more good ‘W hat is the means x such that that John comes by x is better?’ g. [wo zenmeyang shuo ni] bijiao qiadang? I how say you more appropriate ‘W hat is the manner x such that that I talk about you in x is more appropriate?’ (31) Chinese *[John weishenme lai] bijiao hao? John why come more good ‘W hat is the reason x such that that John comes for x is better?’ Once again the contrast in grammaticality o f the sentences in (28-31) is along the lines of naze (why) and weishenme (why) and the rest o f wh-words in these languages. This provides Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 213 further evidence that Huang’s ECP account, based crucially on the dichotomy o f argument wh-word vs. adjunct wh-word, under uniform LF wh-movement hypothesis is not correct. To summarize, the data involving Complex N oun Phrase Islands, Adjunct islands and Sentential Subject Islands in Korean type languages suggest that adjunct wh-words, which correspond to English why in these languages indeed stand apart from the rest o f wh-words, in showing sensitivity to locality effects in these island constructions. This means that H uang’s ECP, which is based on the dichotomy o f argument wh-words on the one hand and adjunct wh-words on the other under the uniform LF wh-movement hypothesis, cannot be a correct locality constraint at LF, which motivates one to look for an alternative account. At this point before we further proceed to our own analysis of the data involving Complex Noun Phrase Island, Adjunct Island and Sentential Subject Island examined thus far in section 2, it is useful to review other important proposals in the literature to see how our own analysis later on compares with these proposals. In fact, there have been a growing body of research in the literature suggesting that wh-words in Korean type languages cannot be treated on a par with wh-words as in English (Nishigauchi 1990, Cheng 1991, Li 1992, Tsai 1994, Lin 1996, Shimoyama 2001, among others), departing from Huang (1982) who claims that the difference is only trivial with wh-words in the former undergoing movement at LF. This departure from classical uniform LF wh-movement as advocated in Huang (1982) is from the important observation that wh-words, if not all, in Korean type languages are indefinites in that they can have different interpretations depending on the context where they occur. Below I will review Tsai (1994) and Nishigauchi (1990), whose analyses for the locality effects o f wh-words in syntactic islands is based on this important observation, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 214 like the current proposal. Nishigauchi (1990) and Tsai (1994) argue that subjacency, but not ECP, is the valid locality condition at LF in Korean type languages: Subjacency is satisfied via the mechanism o f pied-piping in Nishigauchi, while it is satisfied via the m echanism of unselective binding in Tsai. Thus the ungrammaticality o f sentences with a certain adjunct wh-word inside an island is attributed to the unavailability of pied-piping in the former and the unavailability o f unselective binding in the latter. 3. Pied-piping Approach Nishigauchi (1990) proposes that wh-words in Japanese (Korean and Chinese) are indefinites and they should be raised into the question marker (or Spec o f CP in current terminology), where they can be unselectively bound in the sense o f Heim (1982) by the question marker under government for proper interpretation. Given that an indefinite can take scope out o f the clause, the question is why wh-words in Japanese should still undergo movement into the question marker to be properly interpreted. For this, he presents locality effects for wh-words in W h-island constructions, as illustrated in (32), and argues that this locality effect is indeed evidence for movement of wh-words in Japanese. (32) Tanaka-kun-wa [C p dare-ga nani-o tabeta-ka] oboteimasu-ka? T-TOP who-NOM what-ACC eat-QM remember-QM ‘Does Tanaka know who ate w hat?’ (Nishigauchi 1990: 28) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 215 According to Nishigauchi, none o f the wh-words in (32) can take matrix scope out o f the other, i.e., the question is only understood as a yes-no question, a standard instance o f a Wh- island effect. Based on the strict locality effect o f wh-words in (32), he thus claims that the examples in (5), repeated in (33) below, do not necessarily establish that wh-words undergo movement at LF in a way not conforming to subjacency, contrary to Huang’s (1982) claim. (33) [shei xie de shu] zui youqu? who write book most interesting ‘Who is the person x such that books x wrote are most interesting?’ (Huang 1982: 526) In fact, Nishigauchi proposes that subjacency but not ECP is the valid locality constraint for the interpretation o f wh-words in Japanese, and that this locality constraint is actually observed in (33), contrary to appearances. He claims that the apparent lack o f island effect in (33) is just an illusion, which is due to a special mechanism at LF, i.e., pied-piping o f the NP including the relative clause into Spec o f CP (or comp in his terminology). He notes that the pied-piping mechanism can find its motivation in the fact that in languages like Japanese the relative clause is a specifier o f the NP. Based on this marked structural property o f the relative clause in Japanese (and also Korean and Chinese), he crucially suggests a feature percolation mechanism between the specifier and the NP. To be specific, the wh-word inside the relative clause undergoes initial movement into Spec o f CP (or comp o f S’) from which point its wh-feature percolates into CP (or S’), which is the specifier o f the NP, and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 216 then further percolates into the NP. 5 Then the operation, which is responsible for wh- movement, moves the NP including the relative clause into Spec o f CP o f the matrix clause, which is headed by the QM as in (34). (34) [ cp [ c’ [ v p [ n p [ cp wh-wordj [ip t, ] ]] V ] QM ] ] The feature percolation mechanism which enables the eventual large scale pied-piping o f the NP including the relative clause is subject to several conditions, however. The head o f the NP including the relative clause should not have a referential feature, otherwise the feature percolation process is blocked from CP (S’) into NP, which forces movement of the wh-word alone, which will eventually be subject to subjacency, hence resulting in ungrammaticality as shown in (35). (35) a. [D are-ga kaita hon-ga] omosiroidesu-ka? who-NOM wrote book-NOM interesting-QM ‘Who is the person x such that books x wrote are interesting?’ b. * [D are-ga kaita sono hon-ga] omosiroidesu-ka? who-NOM wrote that book-NOM interesting-QM ‘W ho is the person x such that the books x wrote are interesting?’ Nishigauchi (1990:85) 5 Nishigauchi (1990) assumes without discussion that the QM in Japanese is base-generated in the head of CP. As shown in Chapter 3, this is not correct, however. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 217 According to Nishigauchi (1990), the contrast in grammaticality in (35) is due to the contrast in the referentiality o f the head nouns. The head noun in (35a) is nonreferential while the one in (35b) is referential, thus forcing the wh-word in (35b) to move alone out o f the island at LF into the matrix Spec o f CP, leading to ungrammaticality, due to subjacency violation. Another restriction on feature percolation is that the wh-word undergoing initial movement into Spec o f CP as in (34) either should match with the categorical feature [+N] o f the head norm or should be quantificational, otherwise the feature percolation from Spec o f CP o f the relative clause into NP will be blocked again. For this, consider the following examples in (36): (36) a. [Dare-ga kaita hon-ga] omosiroidesu-ka? who-NOM wrote book-NOM interesting-QM ‘Who is the person x such that books x wrote are interesting?’ Nishigauchi (1990:85) b. *[Kare-ga naze kaita hon-ga] omosiroidesu-ka? he-NOM why wrote book-NOM interesting-QM ‘W hat is the reason x such that books he wrote for x are interesting?’ Nishigauchi (1990: 89) c. [Kare-ga dooyatte kaita e-ga] itiban sukidesu-ka? he-NOM how drew picture-NOM most favorable-QM ‘W hat is the means x such that pictures he drew by x are most favorable?’ Nishigauchi (1990: 99) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 218 The contrast in grammaticality in (36ab) follows from the contrast in categorical feature mismatch between the head noun o f the relative clause with [+N] feature and the wh-word. The wh-word in (36a) has [+N] feature such that it can percolate the wh-feature up to NP, hence leading to large scale pied-piping at LF eventually, avoiding subjacency violation. Adjunct wh-word naze (why) in (36b), with [-N] feature, on the other hand, cannot percolate its wh-feature up to the NP, and hence is forced to move across the island into the matrix clause Spec o f CP, leading to ungrammaticality due to subjacency violation. The adjunct wh-word dooyatte (how) with a means construal in (36c), which has a [-N] feature, according to Nishigauchi (1990: 99), can still void subjecency since the wh-word is quantificational, in contrast with naze (why) in (36b), so that the wh-feature can percolate up to NP, which eventually undergoes pied-piping. 6 The status o f dooyatte (how) as a quantificational wh-word in contrast with naze (why) is shown by the examples in (37): (37) a. *[ Kare-ga naze kite mo] boku-wa uresii. he-NOM why come MO I-TOP happy ‘I will be happy for whatever reason he com es.’ b. [ Kare-ga doyatte settoku-si-yoo-to site mo] zettaini kyohi-suru zo. he-NOM how persuade try do M O absolutely reject will ‘No matter how he tries to persuade me, I will absolutely reject it.’ (Nishigauchi 1990: 98, 100) 6 Nishigauchi (1990) analyzes itsu (when) and doko (where) as NP wh-words, while dooyatte (how) as a non-NP wh-word, based on the fact that the latter cannot be case-marked in contrast to the former. Refer to Nishigauchi (1990:99) for the relevant data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 219 Mo in the embedded clause, which conveys a universal quantificational force, determines the quantificational force o f the adjunct wh-word dooyatte (how) in (37b), which, however, is not possible for naze (why) in (37a). Nishigauchi attributes the ungrammaticality o f the sentence in (37a) to the property o f naze (why) as a non-quantificational sentential operator. To recap, LF pied-piping o f the wh-word will be possible if the head noun o f the relative clause is nonreferential, and the wh-word has categorical feature [+N] or is quantificational. The contrast in grammaticality in section 2 involving Complex Noun Phrase Islands will thus follow, assuming that the adjunct wh-words way (why) in Korean, weishenme (why) in Chinese and naze (why) in Japanese cannot have [+N] feature nor can they be quantificational in his system. The contrast in grammaticality in examples involving Sentential Subject Islands and Adjunct Islands in section 2 will also follow in his system, given the crucial assumption that S’ has [+N] feature, following Stowell (1981: 38) such that S’ (CP in our terminology) including a wh-word which has a [+N] feature or is quantificational can be pied-piped eventually, while the adjunct wh-words way (why) in Korean and weishenme (why) in Chinese and naze (why) in Japanese cannot be pied-piped, assuming they cannot have [+N] feature nor can they be quantificational. Nishigauchi’s pied-piping, however, seems to face serious problems, technical, empirical and ontological in nature. Technically, his pied piping mechanism encounters a serious problem when it comes to strong CNPIs as discussed in section 2, which was originally noted by von Stechow (1996: 69). Consider the example in (16a) in Japanese involving a strong CNPI, repeated in (38). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 220 (38) Kimi-wa [dare-ga kaita hon-o] yomimasita-ka? you-TOP who-NOM wrote book-ACC read-QM ‘What is the person x such that you read books x w rote?’ (Nishigauchi 1990: 48) Suppose the null operator is sitting in the Spec o f the relative clause. In this case, the wh- word cannot move into Spec o f CP to induce categorical feature percolation into NP, which will eventually force movement o f the wh-word across the island alone, leading to ungrammaticality, contrary to fact. This important issue was not discussed at all in Nishigauchi (1990). 7 Let us turn to the empirical problem o f pied-piping, i.e., wrong semantics. Nishigauchi presents two important arguments for LF pied-piping, which actually argue against it. The first argument involves the short form answer ending with desu (be), and the other has to do with a ‘scope problem ’. Let us review his first argument. According to him, the hearer o f the question below in (39) can answer it by giving short answers ending with desu (be) with information for the wh-word, as in (40A), or by mentioning the NP in which the wh-word is embedded, as in (40B), apart from by giving a full answer repeating the entire sentence. (39) Kimi-wa [dare-ga kaita hon-o] yomimasita-ka? you-TOP who-NOM write book-ACC read-QM ‘You read books that who w rote?’ (Nishigauchi 1990: 51) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 221 (40) A. Austin-desu. ‘(It’s) Austin.’ B. Austin-ga kai-ta hon-desu. Austin-NOM wrote book-be ‘(It’s) the book that Austin w rote.’ (N ishigauchi 1990: 51) Nishigauchi claims that the answer in (40B) is an answer for the (derived) wh-expression in Spec o f CP, i.e., NP including the relative clause, based on his observation that the short answer with desu supplies no more information than what corresponds to the value for the wh-expression in the question (Nishigauchi 1990: 50). This pattern o f answer shows that the NP including the relative clause with a wh-word in it was pied-piped into Spec o f matrix CP, according to him. He claims that the answer in (40B) provides ‘the identity o f books making crucial use o f the identity o f the person who wrote them ’. He also claims that the answer in (40A) is derived from (40B) by some pragmatic deletion rule. His claim that the answer in (40B) is the primary one nevertheless seems to raise the problem o f wrong semantics, as Partee (cited in Nishigauchi (1990: 52)) and von Stechow (1996) note. W hat the question in (39) is after is not the identity o f books but rather the identity o f authors. The problem is that Nishigauchi’s pied-piping does not yield the right semantics for (39) which, under the LF pied-piping, will have the following LF in (41): (41) [ C p [darex-ga kaita hon-o]y [ e [ i p kimi-wa [ V p y yomimasita ] ] k a x y]] who-NOM wrote book-ACC you-TOP read QM ‘for which x, y, x a person and y books x wrote, did you read y? 7 This technical problem noted by von Stechow (1996) is also mentioned in Shimoyama (2001: 20). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 222 Now the problem in (41) is, as pointed out by von Stechow (1996), that only the variable ranging over books occurs in the nuclear scope o f the wh-quantifier and as such its value varies. In other words, what is asked for primarily is books but not authors. However, as just noted, this is not the correct reading for the sentence in (3 9 ).8 Let us review the other argument o f Nishigauchi for his pied-piping mechanism, which has to do with what he calls the ‘scope problem .’ Consider the example in (42) as an illustration: (42) ??[ Nan-ka-koku kara kita daihyoo-ga] atumarimasita-ka? how-many-countries from came representative-NOM gather-QM ‘Representatives who came from how many countries gathered at the summit m eeting?’ (Nishigauchi 1990: 58) According to Nishigauchi, the above question is absurd in that what it asks is ‘which number x is such that representatives from x countries gathered.’ Consider the LF representation in (43) for the sentence above in (42), which is generated under his LF pied-piping analysis: (43) [c p [n p [ cp [nanz [ z ka-koku]]x [ ip x kara kita]] daiyoo-ga]y [ C ’[ ip y atum arim asita]-kay /z]] how many countries from came representative-NOM gathered-QM ‘for which y, z, z a number, y representatives that came from z many countries, y gathered?’ 8 I will not discuss von Stechow (1996) since his proposal is essentially the same as the one in Nishigauchi (1990), except that the level of representation where pied-piping takes place is what he calls WH-structure, and the pied-piped phrase reconstructs at LF. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 223 It should be noted that the logical form in (43) actually yields a coherent reading, asking which representatives o f countries gathered, as observed by von Stechow (1996). As a result o f LF pied-piping o f the NP including relative clause into the matrix Spec o f CP, the number variable is removed from the nuclear scope, so that the question does not ask the number of countries the representatives are simultaneously representing. Thus the pied-piping mechanism gives a coherent reading for the question in (43), contrary to Nishigauchi’s claim and the facts, again ending up with wrong semantics. 9 Nishigauchi’s pied-piping mechanism, in addition to the technical problem and the empirical problem o f wrong semantics as discussed above, also invites a nontrivial ontological question: If wh-words are indefinites in general in Korean type languages as he claims, why then can they not take scope out o f syntactic islands without movement? Does that mean the grammar has another type o f indefinite that is subject to island constraints? Nishigauchi offers the locality effects for wh-words in W h-island constructions in Japanese as supporting evidence for the movement o f wh-words in Japanese, claiming that the movement o f wh- words out o f other syntactic islands, such as Complex Noun Phrase Islands, actually conforms to subjacency, with a large scale LF pied-piping mechanism voiding the would be 9 Choe (1987) presents the following sentences involving weak crossover as evidence for LF pied- piping. I will not discuss this since Nishigauchi himself notes that weak crossover cannot be evidence for pied-piping, given the independent mechanism of quantifier-raising (see Nishigauchi 1990 and also von Stechow 1996). (i) a. John-wa Mary-ga proi yomumae-ni [Judy-ga kaita honj-o] yonda-no? J-TOP M-NOM read-before J-NOM wrote book-ACC read ‘Did John read the book] that Judy wrote before Mary read it,?’ b. * John-wa Mary-ga pro; yomumae-ni [dare-ga kaita honj-o] yonda-no? J-TOP M-NOM read-before who-NOM wrote book-ACC read-QM ‘Did John read the bookj who wrote before Mary read if?’ (Choel987: 352) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 224 violation. However, as will be discussed in Chapter 5, W h-island effects do not show up in Korean type languages contrary to his claim, further undermining his pied-piping mechanism as responsible for subjacency at LF in these languages. Given the inadequacy o f pied-piping as the right mechanism responsible for subjacency in Korean type languages, it follows that a different mechanism is indeed responsible for subjacency in these languages, assuming that subjacency is the valid locality constraint at LF in these languages at least, given the inadequacy o f ECP as shown in section 2. W hat is the mechanism? Tsai (1994) claims that it is unselective binding, which is essentially the position we are taking in this thesis, although the details o f the current proposal and the one in Tsai and thus the empirical predictions the two proposals make differ, as will be shown below in section 4 and 5. 4. Tsai (1994) Tsai (1994) claims that argument wh-words in the sense o f Huang (1982) and the adjuncts zenmeyang (how) under means and resultative construal and weishenme (why) under purpose construal in Chinese are nominal wh-words and are indefinites, being interpreted in situ as a wh-interrogative by being unselectively bound by the null operator in Spec o f CP at LF, while zenmeyang (how) under manner construal and weishenme (why) under reason construal are nonnominal wh-words and are not indefinites, and should undergo movement into Spec o f CP at LF. Putting aside the legitimacy o f introducing the null operator for the interpretation o f wh-words as interrogatives, the grammaticality o f the examples with the adjunct wh-words ettehkey (how) in Korean, doo (how) in Japanese and zenmeyang (how) in Chinese in islands under manner construal as shown in section 2 poses a nontrivial problem R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 225 to his proposal, since given his dichotomy regarding indefmite-hood as summarized above, these adjuncts should be subject to locality effects in syntactic islands, since they have to undergo movement at LF for proper interpretation, hence leading the sentences with them to ungrammaticality, contrary to the facts. I will illustrate this point with examples involving Complex Noun Phrase Islands in Korean type languages in section 2. These examples are repeated below in (44-46). 1 0 The examples in (44a-46a) involve a strong CNPI and the ones in (44b-46b) involve a weak CNPI. (44) CNPI in Korean a. Ne-nun [os-ul ettehkey ipnun salam-ul] coaha-ni? you-TOP clothes-ACC how wear man-ACC like-QM ‘W hat is the manner x such that you like a man who gets dressed in x?’ b. Ne-nun [M ary-ka ne-eytayhay ettehkey malhakotanintanun somwun-ul] tuless-ni? you-TOP M -NOM you-about how saying rumor-ACC heard-QM ‘W hat is the manner x such that you heard a rumor that Mary is talking about you in x?’ (45) CNPI in Japanese a. Kimi-wa [kimi-nituite doo itteiru hito-o] sukidesu-ka? you-TOP you-about how say man-ACC like-QM ‘W hat is the manner x such that you like a person who talks about you in x?’ 1 0 In fact, Lin (1992) for example, argues that in Chinese only weishenme (why) is subject to locality effects involving various islands, which indeed is supported by the present data in section 2. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 226 b. Kimi-wa [M ary-ga John-nituite doo hanosita toyuu uwasa-o] kiita-no? you-TOP M -NOM J-about how saying rumor-ACC heard-Q ‘W hat is the manner x such that you heard a rumor that Mary is talking about John in x?’ (46) CNPI in Chinese a. ni bijiao xihuan [zenmeyang shuo ni de ren]? you more like how say you person ‘W hat is the manner x such that you like better a person who talks about you in x?’ b. ni tingshuo quo [John zenmeyang shuo ni de gushi] you hear John how say you rumor ‘W hat is the manner such that you heard a rum or that John talks about you in x?’ These examples with an adjunct wh-word under manner construal above in (44-46) are all grammatical, which strongly suggests that the dichotomy o f wh-words in Tsai (1994) regarding indefiniteness along nominal wh-words vs. nonniminal wh-words is inadequate. 5, Account At this point, please recall the asymmetric scope taking strategies o f the propositional adjunct wh-words and nonpropositional wh-words proposed in Chapters 2 and 3. According to our proposal, the propositional adjunct wh-words have to undergo movement into Spec o f CP at LF, driven by the need for proper interpretation by forming an operator variable chain, while R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 227 nonpropositional wh-words are unselectively bound by the QM, which behaves as a wh- operator and thus marks their scope. By the propositional adjunct wh-word, I m eant a wh- word corresponding to why, as it can typically quantify over propositions, while by nonpropositional wh-words, I meant the rest o f wh-words quantifying over individuals, means, measures or manners. Below it will be shown that the present proposal can give a straightforward account for the asymmetry o f locality effects with the two types o f wh-words as presented in section 2. Let us consider the examples in (8-9) and (16-19) involving Complex Noun Phrase Islands in Korean type languages in section 2, as repeated in (47-52). (47) Korean a. Ne-nun [nwu-ka ssun chayk-ul] ilkess-ni? you-TOP who-NOM wrote book-ACC read-QM ‘Who is the person x such that you read a book x w rote?’ b. Ne-nun [ enu cakka-ka ssun chayk-ul] ilkess-ni? you-TOP which writer-NOM wrote book-ACC read-QM ‘Which writer x is such that you read a book x w rote?’ c. Ne-nun [John-i eti-se ssun chayk-ul] ilkess-ni? you-TOP J-NOM where-at wrote book-ACC read-QM ‘W hat is the place x such that you read a book John wrote at x?’ d. Ne-nun [John-i encey ssun chayk-ul] ilkess-ni? you-TOP J-NOM when wrote book-ACC read-QM ‘'What is the time x such that you read a book John wrote at x?’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 228 e. Ne-nun [cheycwung-i elma nakanun salam-ul] coaha-ni? you-TOP weight-NOM how much weigh man-ACC like-QM ‘W hat is the weight x such that you like a man who weighs x?’ f. Ne-nun [John-i ettehkey kulin kulim-ul] coaha-ni? you-TOP J-NOM how painted painting-ACC like-QM ‘W hat is the means x such that you like pictures John drew by x?’ g. Ne-nun [os-ul ettehkey ipnun salam-ul] coaha-ni? you-TOP clothes-ACC how wear man-ACC like-QM ‘W hat is the manner x such that you like a man who gets dressed in x?’ (48) Korean *Ne-nun [John-i way ssun chayk-ul] ilkess-ni? you-TOP J-NOM why wrote book-ACC read-QM ‘W hat is the reason x such that you read a book John wrote for x?’ (49) Japanese a. Kimi-wa [dare-ga kaita hon-o] yomimasita-ka? you-TOP who-NOM wrote book-ACC read-QM ‘Who is the person x such that you read a book x w rote?’ (Nishigauchi (1990:48) b. Kimi-wa [dono hito-ga kaita hon-o] yomimasita-ka? you-TOP which man-NOM wrote book-ACC read-QM ‘W hich person x is such that you read a book x w rote?’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 229 c. Kimi-wa [kare-ga doko-de kaita hon-o] yomimasita-ka? you-TOP he-NOM where at wrote book-ACC read-QM ‘W hat is the place x such that you read a book he wrote at x?’ d. Kimi-wa [kare-ga itsu kaita hon-o] yomimasita-ka? you-TOP he-NOM when wrote book-ACC read-QM ‘What is the time x such that you read a book he wrote at x ?’ e. Kimi-wa [taijuu-ga ikutsu aru hito-o] sukidesu-ka? you-TOP weight-NOM how much have man-ACC like-QM ‘W hat is the weight x such that you like a person who weighs x?’ f. Kimi-wa [kare-ga dooyatte kaita e-o] kaimasita-ka? you-TOP he-NOM how painted painting-ACC bought-QM ‘W hat is the means x such that you bought a picture he drew by x?’ g. Kimi-wa [kimi-nituite doo itteiru hito-o] sukidesu-ka? you-TOP you-about how say man-ACC like-QM ‘W hat is the manner x such that you like a person who talks about you in x?’ (50) Japanese *Kimi-wa [kare-ga naze kaita hon-o] yomimasita-ka? you-TOP he-NOM why wrote book-ACC read-QM ‘W hat is the reason x such that you read a book he wrote for x?’ (Nishgauchi 1990: 41) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 230 (51) Chinese a. ni bijiao xihuan [shei xie de shu]? you more like who write book ‘Who is the person x such that you like better a book x w rote?’ b. ni bijiao xihuan [na-ge-ren xie de shu]? you more like which person write book ‘W hich person x is such that you like better a book x w rote?’ c. ni bijiao xihuan [John zai nali xie de shu]? you more like John at where write book ‘W hat is the place x such that you like better a book John wrote at x?’ d. ni bijiao xihuan [John shenmeshihou xie de shu]? you more like John when write book ‘W hat is the time x such that you like better a book John wrote at x?’ e. ni bijiao xihuan [zhong ji bang de ren]? you more like amount how much person ‘W hat is the amount x such that you like a person who weighs x? f. ni bijiao xihuan [John zenmeyang hua de hua]? you more like John how paint painting ‘W hat is the means x such that you like better a painting John painted by x?’ g. ni bijiao xihuan [zenm eyang shuo ni de ren]? you more like how say you person ‘W hat is the manner x such that you like better a person who talks about you in x?’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 231 (52) Chinese *ni bijiao xihuan [John weishenme xie de shu]? you more like John why write book ‘W hat is the reason x such that you like better a book John wrote for x?’ If not necessary, I will use head-initial language notation for the LF representations for Korean type languages for the reader’s convenience throughout this chapter. I will suppress a full DP structure for the representation o f nonpropositional wh-words throughout the chapter, since it is not relevant for our discussion here. Abstracting away from the parametric variation regarding the position o f the QM, the LF for the sentences above in (47, 49, 51) with a nonpropositional wh-word thus will look like the following, with irrelevant details suppressed:1 1 (53) [ QMi [V P V [ n p [ c p [ ip ... WH-wordi ...]]]]] The interpretation o f the nonpropositional wh-word by unselective binding in (53) vacuously satisfies subjacency since it has not undergone movement into the matrix Spec o f CP position for interpretation.1 2 1 1 One needs to postulate an abstract QM in the LF representation for Chinese, since it is deleted at the surface level freely. 1 2 A similar idea to the present proposal is suggested in Chomsky (1993: 26, 1995: 193), as stated below. ‘The LF rule that associates the in-situ wh-phrase with the wh-phrase in [Spec, CP] need not be construed as an instance of move-a. We might think of it as the syntactic basis for absorption in the sense of Higginbotham and May (1981), an operation that associates two wh-phrases to form a R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 232 W hen it comes to the propositional adjunct wh-words in (48, 50, 52), given their LF movement driven by the need for proper interpretation by forming an operator variable chain, the LF representation for these sentences will be (54) (with irrelevant details suppressed): (54) [ c p why\ [c’ QM [ n p [ c p . . . h . . . ] ] ] ] The movement o f the propositional adjunct wh-word in (54) crucially violates subjacency, crossing barriers, CP and NP. This violation o f subjacency is the nature o f the ungrammaticality o f the sentences in (48, 50, 52). Essentially the same account can extend to examples involving other islands in Korean type languages as presented in section 2, which I will leave to the reader. Before closing this section, it should be noted that our proposal does not have the problem o f wrong semantics in Nishigauchi (1990) regarding the CNPI constructions, as discussed in section 3, since the NP including the nonpropositional wh- word is interpreted in situ as a wh-interrogative bound by the QM in the matrix clause. For this, consider the example in (39), repeated in (55). (55) Kimi-wa [dare-ga kaita hon-o] yomimasita-ka? you-TOP who-NOM write book-ACC read-QM ‘You read books that who w rote?’ (Nishigauchi 1990: 51) generalized quantifier. If so, then the LF rule need satisfy none of the conditions on movement.’ [sic] The difference between the suggestion in Chomsky (1995) and ours in this thesis is that we argue for two independent mechanisms for the interpretation of in situ wh-words, at least in Korean type languages, i.e., the movement of the propositional wh-word at LF and the unselective binding of nonpropositional wh-words by the QM. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 233 The LF representation for the sentence above in (55) will be the one below in (56), under our system: (56) [Kimi-wa [ dare,-ga kaita hon-o] yomimasita-kax] you-TOP who-NOM wrote book-ACC read-QM ‘Who is the person x such that you read books that x w rote?’ As shown by the logical notation in (56), the variable ranging over authors occurs in the nuclear scope o f the wh quantification, correctly predicting that the speaker is asking information primarily about identity o f the authors in (55). The present proposal can also account for the absurd reading for the sentence in (42), repeated as (57). (57) ??[Nan-ka-koku kara kita daihyoo-ga] atumarimasita-ka? how-many-countries from came representative-NOM gather-QM ‘Representatives who came from how many countries gathered at the summit m eeting?’ (Nishigauchi 1990: 58) The following in (58) is the logical form representation for (57), with irrelevant details suppressed, under our system: (58) [[nanx ka-koku kara kita daiyoo-ga] atumarimasita-kax] ‘W hat is the number n such that representatives from n number o f countries gathered?’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 234 As one can notice, the variable standing for the number o f countries remains in the nuclear scope of the QM, and as a result the absurd reading obtains. 6. Conclusion The asymmetry o f locality between propositional adjunct wh-words and nonpropositional wh-words in Korean type languages in constructions involving Complex Noun Phrase Islands, Sentential Subject Islands and Adjunct Islands was observed and accounted for in conjunction with critiques o f the important proposals in the literature, i.e., the ECP approach under the uniform LF wh-movement hypothesis o f Huang (1982), pied-piping approach of Nishigauchi (1990) and the unselective binding approach o f Tsai (1994). I claimed that the ECP approach of Huang (1982) is rather problematic, given that the adjunct wh-words in Korean type languages, which correspond to how in English can still occur inside o f islands. The proposal o f Tsai (1994) is also problematic in that the adjunct wh-words how under manner construal, which are nonnominal wh-words and thus are not indefinites in his system can still occur inside islands in Korean type languages. N ishigauchi’s pied-piping mechanism runs into the problem o f wrong semantics, among other problems, which actually argues against pied-piping. As an alternative, I tried to show that the asymmetry of locality effects between propositional adjunct wh-words and nonpropositional wh-words in Korean type languages in constructions involving Complex Noun Phrase Islands, Sentential Subject Islands and Adjunct Islands receives a straightforward account, given the different means of scope taking o f the two types o f wh-words at LF: unselective binding o f the nonpropositional wh-words by the QM vs. the movement o f propositional adjunct wh-word. Movement o f the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 235 propositional adjunct wh-word out o f islands into its scope position, marked by the QM at LF, leads to ungrammaticality due to subjacency violation, while unselective binding by the QM o f the nonpropositional wh-words vacuously satisfies subjacency, since no movement is involved. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 236 CHAPTER 5 Symmetry of Locality in Wh-island Constructions 1.Introduction The asymmetry of locality between propositional adjunct wh-words and nonpropositional wh-words in Korean type languages in constructions involving Complex N oun Phrase Islands (Ross 1967), Sentential Subject Islands (Ross 1967), and Adjunct Islands (Huang 1982) was observed and discussed in Chapter 4, and it was shown to follow essentially from our proposed asymmetric scope taking strategies for the two types o f wh-words in Chapter 2 and Chapter 3, together with the assumption that subjacency violation in the sense of Chomsky (1986) at LF directly leads to ungrammaticality: The propositional adjunct wh- word should undergo movement at LF into Spec o f CP to mark its scope via spec-head agreement with the QM , driven by the need for proper interpretation, while nonpropositional wh-words are unselectively bound by the QM in the sense o f Lewis (1975) and Heim (1982). Interestingly, when it comes to W h-island constructions in Korean type languages, the two types of wh-words do not exhibit W h-island effects. This is rather surprising, given the asymmetry o f locality between the two types o f wh-words with other types o f islands as shown in Chapter 4. In this chapter, it will be shown that the distinct scope taking strategies o f the propositional adjunct wh-w ord and the nonpropositional wh-word each enable the respective type to scope out o f W h-islands, thus accounting for the surprising symmetry between them in W h-island constructions. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 237 2. Wh-island It was originally observed by Chomsky (1962) that a clause introduced by a wh-word is an island. The sentences below in (1) are typical instances of involving W h-island violation, resulting from the wh-word in the matrix clause having moved across the W h-island introduced by the wh-w ord in the embedded clause, crossing a barrier at some point during its derivation (Chomsky 1986).1 (1) a. ?*W hat do you wonder who saw? b. *How do you wonder who fixed the car? (Chomsky 1986: 48-49). W hen it comes to W h-island effects in Korean type languages, Huang (1982) originally claims that wh-words in Chinese undergoes movement at LF uniformly and the wh-word raised at LF into the embedded comp with [+WH] feature specification as in (2) forms a Wh- island, blocking the matrix scope o f the other wh-word out of it, depending on which type o f wh-word it is, an argument wh-word or an adjunct w h-w ord.2 (2) [ S’ comp [+WH] [s V [ S’ WH-word, comp [+WH] [s t, V W H-word ]]]] 1 It should be noted that judgments on the degree of grammaticality of the sentence in (la) have oscillated (see Chomsky 1977,1986,1995 among others). 2 1 should note that adjunct wh-words should come between the subject and verb in Chinese. I have abstracted away from this in the representation in (2). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 238 The following complex wh-questions with a wonder type matrix predicate, which takes a wh- clause w ith two wh-words as a complement are thus typical instances o f W h-island constructions in C hinese:3 (3) a. ni xiang-zhidao [shei mai-le shenme]? you wonder who buy-ASP what ‘Who is the person x such that you wonder what x bought?’ ‘W hat is the thing x such that you wonder who bought x?’ b. ni xiang-zhidao [Lisi zai nali mai-le shenme]? you want to know Lisi at where buy-ASP what ‘W hat is the thing x such that you wonder where Lisi bought x?’ ‘W hat is the place x such that you wonder what Lisi bought at x?’ c. ni xiang-zhidao [Lisi (zai) shenmeshihou mai-le shenme]? you want to know Lisi at when buy-ASP what ‘W hat is the thing x such that you wonder when Lisi bought x?’ ‘W hat is the time x such that you wonder what Lisi bought at x?’ d. ni xiang-zhidao [shei zeme mai-le shu]? you wonder who how buy-ASP book ‘Who is the person x such that you wonder how x bought a book?’ # ‘What is the manner x such that you wonder who bought a book in x?’ 3 Huang (1982: 520) notes that sentences as in (3a), for example, can be construed as a yes no question or as a statement with an indirect multiple question. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 239 e. ni xiang-zhidao [shei weishenme m ai-le shu] ? you wonder who why buy-ASP book ‘Who is the person x such that you wonder why x bought a book?’ # ‘W hat is the reason x such that you wonder who bought a book for x?’ (Huang 1982: 525-529) Please recall that Huang (1982: 528-530) means by argument wh-words, wh-words which can be arguments o f a lexical category such as V, P, N, or A, while by adjunct wh-words he means wh-words which cannot be arguments o f these lexical categories. Thus shei (who) and shenme (what) are bona fide argument wh-words. When it comes to wh-words such as nali (where) and shenmeshihou (when), they can be viewed as argument wh-words, since they can come together with a preposition, showing that they are serving as arguments o f a lexical category. Zeme (how) and weishenme (why) are genuine adjunct wh-words, since they cannot be arguments o f a lexical category. The argument wh-words in (3) can take matrix scope out of the W h-island, while adjunct wh-words cannot do so, according to Huang’s observation. Huang accounts for the observed asymmetry o f locality effects between the argument wh-words and the adjunct wh-words in (3) in terms o f ECP under the uniform LF wh-movement hypothesis. To be specific, LF raising of the argument wh-word scoping out of the W h-island to take matrix scope does not pose a problem even though subjacency (W h-island) is violated by crossing more than one bounding node, i.e., S at once during its derivation from the embedded clause into the matrix comp, since it is ECP but not subjacency, which says that R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 240 nonpronominal empty categories should be properly governed either by lexical government or by antecedent government that is the valid locality constraint at LF, according to Huang (1982): Huang argues that subjacency cannot be the relevant locality constraint at LF given that the argument wh-words in (3) can take matrix scope, without showing any marginality, quite contrary to the argument wh-word in (la ) in English, which shows marginality in grammaticality. The cross-linguistc contrast in grammaticality between the two languages can be accounted for once one assumes that ECP which is based on argument/adjunct dichotomy but not subjacency is the valid locality constraint for LF wh-w ord movement in Chinese, too. Thus depending on the base position o f the argument wh-word, its trace left behind will be eventually properly governed by either a lexical category or INFL, which is also a proper governor in Chinese. Meanwhile, the impossibility o f the adjunct wh-word scoping out o f the W h-island in (3) is due to the ECP violation, since the maximal projection S ’ intervenes between the antecedent in the matrix clause and its trace in the embedded clause, thus leading to failure o f proper government. Huang’s ECP account, which is based on argument and adjunct dichotomy, thus seems to account for the apparent asymmetry between argument wh-words and adjunct wh-words in scoping out o f the W h-island so formed under the uniform LF wh-movement hypothesis as in (2). However, I note that there is an independent factor, which may prevent the adjunct wh-words from taking matrix scope in (3d-e). Zeme (how) cannot quantify over manners in (3d). W hat is a manner that a buyer can pose when he buys a book? Can zeme (how) readily quantify over manner in this context? Reckoning with this factor, one cannot ignore the possibility R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 241 that pragmatic infelicity o f quantification over manner in this particular context may lead to the virtual impossibility o f matrix scope interpretation for zeme (how) over the other wh- word, which takes the embedded scope. If zeme (how) in (3d) is construed as quantifying over means, it can take scope out o f the island. One can think o f several relevant means easily, and may wonder who bought a book by a credit card, for exam ple.4 The particular example in (3e) also forbids weishenme (why) from taking matrix scope, apparently for a different reason. The adjunct weishenme (why) can quantify over myriad o f reasons, contrary to typical argument wh-words as in (3a-c), whose domain o f quantification is rather easily limited by the context. How come the hearer o f the question can be sure that at least one individual in the discourse domain may have bought a book for the reason he possibly comes up with so that the statement he [the hearer] is making turns out to be a felicitous one in soliciting information for the other wh-word, namely, shei (who)? For wh-questions with a wonder type matrix predicate as in (3) to be felicitous questions, not only should individuals as replaced by the first wh-word be known to both the speaker and the hearer but also these individuals should be the actual participants in the event denoted by its predicate, or at least there should be good reason to be assumed so. Thus the hearer o f the question in (3e) will rather choose the easy option o f saying ‘I want to know why John bought a book’, knowing that the individual he so picked is one o f the 4 I will use zenmeyang (how) hereafter, since zeme (how) can also be interpreted as why in Chinese (see Charles Li and Sandra Thompson 1991: 523). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 242 individuals who actually bought a book, without having to worry about making an absurd statement asking for information for shei (who). Thus it seems that the domain of quantification, either too restricted or unrestricted, o f the adjunct wh-words in wh-questions as in (3de) with xiang-zhidao (wonder) as the matrix predicate blocks the matrix scope reading o f the adjunct wh-words out o f the W h-island. W hen the matrix predicate is replaced with zhidao (know), with presupposition of knowledge, the reading where weishenme (why) takes scope out o f the W h-island is impeccable even for (3e), repeated with the replacement as (4). (4) ni zhidao [shei weishenme mai-le shu]? you know who why buy-ASP book ‘Who is the person x such that you know why x bought a book?’ ‘W hat is the reason x such that you know who bought a book for x?’ The point I am trying to make here is that the observed cases o f lack o f matrix scope interpretation for the adjunct wh-words embedded in the W h-island as in (3de) may be ascribed to the pragmatic factor o f domain o f quantification, when it comes to a wonder type predicate. Thus, when the factor can be controlled, even an adjunct wh-word can scope out o f the W h-island as shown by the example in (4) with a know type matrix predicate, whose semantics, i.e., presupposition o f knowledge, can suppress the role o f the pragmatic factor o f domain o f quantification. W ith the proposed intervening pragmatic factor for adjunct wh- words scoping out o f W h-islands in mind, let us turn to Korean first. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 243 2.1. Korean I will consider wh-questions in Korean corresponding to the ones in (3) with two wh-words in the embedded clause. Throughout I mean W h-island effect as referring to the phenomenon where one wh-word taking the embedded scope forms a W h-island and blocks the other from taking scope out o f it. I will consider wh-questions with a matrix predicate such as alko sipta (wonder) as well as alta (know) and kiekhata (remember), all o f which can take a wh-clause as their complement. The point o f including the latter predicates is to highlight the pragmatic factor o f domain o f quantification associated with a wonder type predicate with respect to Wh-islands. Consider the questions in (5) with alko sipta (want to know), first. (5) a. Ne-nun [nwu-ka nwukwu-lul chotayhayssnun-ci] alko sip-ni? you-TOP who-NOM whom-ACC invited-QM want to know-QM ‘Who is the person x such that you want to know whom x invited?’ ‘Who is the person x such that you want to know who invited x?’ b. Ne-nun [nwu-ka eti-se tampay-lul phiwessnun-ci] alko sip-ni? you-TOP who-NOM where-at cigarette-ACC smoked-QM want to know-QM ‘Who is the person x such that you want to know where x smoked?’ ‘W hat is the place x such that you know who smoked at x ?’ c. Ne-nun [nwu-ka encey wassnun-ci] alko sip-ni? you-TOP who-NOM when came-QM want to know-QM ‘Who is the person x such that you want to know when x cam e?’ ‘What is the time x such that you want to know who came at x?’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 244 d. Ne-nun [nwu-ka cheycwung-i elma nakanun-ci] alko sip-ni? you-TOP who-NOM weight-NOM how much weigh-QM want to know-QM ‘Who is the person x such that you want to know how much x w eighs?’ ‘W hat is the weight x such that you want to know who weighs x?’ e. Ne-nun [nwu-ka ettehkey wassnun-ci] alko sip-ni? you-TOP who-NOM how came-QM want to know-QM ‘Who is the person x such that you want to know how x cam e?’ ‘W hat is the means x such that you want to know who came by x?’ f. Ne-nun [nwu-ka M ary-eytayhay ettehkey malhayssnun-ci] alko sip-ni? you-TOP who-NOM M -about how talked-QM want to know-QM ‘Who is the person x such that you want to know how x talked about M ary?’ ‘W hat is the manner x such that you want to know who talked about M ary in x?’ g. Ne-nun [nwu-ka way chayk-ul sassnun-ci] alko sip-ni? you-TOP who-NOM why book-ACC bought-QM want to know-QM ‘Who is the person x such that you want to know why x bought a book?’ # ‘W hat is the reason x such that you want to know who bought a book for x?’ In Korean, according to the argument/adjunct criterion in Huang (1982), wh-words such as nwukwu (who), mwues (what), and elma (how much) are argument wh-words, since they can be arguments o f lexical categories such as V, P, N and A. When it comes to eti (where) and encey (when), they can be viewed as argument wh-words too, since they can be arguments o f a postposition. Ettehkey (how) and way (why) are bona fide adjunct wh-words. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 245 As for the examples in (5a-f), let us suppose the situation where the speaker and the hearer are chatting over John, Mary, Jack, Bill, and Tom, who were at the spring break party, to which each o f them was supposed to invite whoever they wished. Let also suppose a book store situation for the example in (5g). W ith the above contexts as a background, given the questions in (5), the hearer o f these questions may typically answer the questions simply by saying ‘yes’, as in (6), interpreting the questions as yes-no questions, the most economical way from his point o f view. (6) ung yes ‘yes’ In addition to the reading in (6), the questions in (5a-f), in contrast to the one in (5g), which does not allow the matrix scope o f the adjunct wh-word way (why) out o f the other wh-word also admit readings where one o f the wh-words takes scope out o f the other wh-word, which is o f our interest here. 5 This is indicated by the pattern of answers in (7-13), given the standard assumption that the scope o f a wh-word is reflected by the possible answers to the question (Baker 1970, among others). 5 The questions in (5a-f) also admit the reading where both wh-words scope out of the embedded clause. This reading and the yes-no reading as in (6) are not our main interest here, since they do not involve scoping out of the Wh-island. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 246 (7) a. Na-nun [John-i nw ukwu-lul chotayhayssnun-ci] alko sipta. I-TOP J-NOM whom-ACC invited-QM want to know ‘I want to know whom John invited.’ b. Na-nun [nwu-ka caki yecachinkwu-lul chotayhayssnun-ci] alko sipta. I-TOP who-NOM his girl friend-ACC invited-QM want to know ‘I want to know who invited his girl friend.’ (8) a. Na-nun [John-i eti-se tampay-lul phiwessnun-ci] alko sipta. I-TOP J-NOM where cigarette-ACC smoked-QM want to know ‘I w ant to know where John sm oked.’ b. Na-nun [nwu-ka silnay-eyse tampay-lul phiwessnun-ci] alko sipta. I-TOP who-NOM room-in cigarette-ACC smoked-QM want to know ‘I want to know who smoked in the room .’ (9) a. Na-nun [John-i encey wassnun-ci] alko sipta. I-TOP J-NOM when came-QM want to know ‘I want to know when John cam e.’ b. Na-nun [nwu-ka ilkopsi-ey wassnun-ci] alko sipta. I-TOP who-NOM 7-at came-QM want to know ‘I w ant to know who came at 7 o'clock.’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 247 (10) a. Na-nun [John-i cheycwung-i elma nakanun-ci] alko sipta. I-TOP J-NOM weight-NOM how much weigh-QM want to know ‘I want to know how much John w eighs.’ b. Na-nun [nwu-ka cheycwung-i 150 pound cengto nakanun-ci] alko sipta. I-TOP who-NOM weight-NOM 150 pound about weigh-QM want to know ‘I want to know who weighs about 1501bs.’ (11) a. Na-nun [John-i ettehkey wassnun-ci] alko sipta. I-TOP J-NOM how came-QM want to know ‘I want to know how John came to the party.’ b. Na-nun [nwu-ka bus-lo wassnun-ci] alko sipta. I-TOP who-NOM bus-by came-QM want to know ‘I want to know who came by bus.’ (12) a. Na-nun [John-i M ary-eytayhay ettehkey malhayssnun-ci] alko sipta. I-TOP J-NOM M -about how talked-QM want to know ‘I want to know how John talked about M ary.’ b. Na-nun [nwu-ka M ary-eytayhay nappukey malhayssnun-ci] alko sipta. I-TOP who-NOM M -about badly talked-QM want to know ‘I want to know who talked viciously about M ary.’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 248 (13) Na-nun [John-i way chayk-ul sassnun-ci] alko sipta. I-TOP J-NOM why book-ACC bought-QM want to know ‘I want to know why John bought a book.’ The pattern o f answers as given in (7-13) shows that argument wh-words do not show Wh- island effects at all, since they can take matrix scope out of the other wh-word taking the embedded scope and thus forming a Wh-island. The pattern of answers in (7-13) also shows that the adjunct wh-word ettehkey (how) under a means construal in (5e) or a manner construal in (5f) can take matrix scope too, on a par with argument wh-words, hence showing no W h-island effects. Meanwhile, the matrix scope reading o f adjunct wh-word way (why) is infelicitious, showing W h-island effect. Given the pattern o f answers in (7-13), the dichotomy o f argument wh-words on the one hand and adjunct wh-words on the other in Huang (1982) thus seems to be problematic, since the adjunct wh-word ettehkey (how) can also take matrix scope out o f the W h-island anyhow. W hy is it the case that an adjunct wh-word ettehkey (how) in (5el) can take matrix scope out o f the W h-island, while another adjunct wh-word way (why) in (5g) cannot? I suggest that the infelicitousness o f the matrix scope reading o f way (why) out o f the W h-island in (5g) has to do with our proposed pragmatic factor o f domain o f quantification, having nothing to do with the W h-island effects. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 249 To the extent that the hearer cannot be sure that at least one individual in the discourse domain may have bought a book for the reason he can possibly think of, he will be discouraged to say a statement soliciting information for the other wh-word, namely, nwukwu (who). Thus the hearer o f the question in (5g) will rather choose the easy option o f saying ‘I want to know why John bought a book’, knowing that the individual he so picked is one of the individuals who actually bought a book, without having to worry about making an absurd statement asking for rather implausible information for nwukwu (who), which accounts for the infelicitousness o f the matrix scope reading o f the adjunct wh-word way (why). Meanwhile, please note that the matrix scope o f the adjunct wh-word ettehkey (how) in (5ef) is natural, given that it can relatively easily quantify over means and manner, respectively, o f course their domain being easily limited by the context, in contrast to the example in (3d) in Chinese, whose domain o f quantification is too restricted in the context. I suggest that this rather limited domain o f quantification thus enables the adjunct wh-word ettehkey (how) to take matrix scope out o f the W h-islands in (5ef) in Korean. At this point, to be fair, I should note that some linguists in Korean literature report a different judgm ent regarding the locality effects o f wh-words in W h-islands with a wonder type matrix predicate. Kim (1990) and Joo (1989) observe that a complex wh-question as in (14) can be construed only as yes-no question. (14) Ne-nun [nwu-ka mwues-ul sassnun-ci] kwungkwumha-ni? you-TOP who-NOM what-ACC bought-QM wonder-QM ‘Do you wonder who bought w hat?’ (Joo 1989: 110) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 250 Givin the present observation for the lack of W h-island effects in similar constructions with a wonder type matrix predicate as shown in (5), to the extent that a yes-no answer to the above question in (14) is viewed as the only possible answer, it may have to do with the dominant economy strategy activated by the hearer, thus putting on the speaker the burden o f providing maximal information for the hearer. Up till this point, I considered complex wh-questions with a wonder type matrix predicate in Korean. N ow let us consider complex wh-questions with factive predicates such as alta (know) and kiekhata (remember) to further examine whether wh-words exhibit W h-island effects in Korean. Consider the questions in (15) with alta (know) as matrix predicate, first. (15) a. Ne-nun [nwu-ka nwukwu-lul chotayhayssnun-ci] a-ni? you-TOP who-NOM whom-ACC invited-QM know-QM ‘Who is the person x such that you know whom x invited?’ ‘Who is the person x such that you know who invited x?’ b. Ne-nun [nwu-ka eti-se tampay-lul phiwessnun-ci] a-ni? you-TOP who-NOM where-at book-ACC smoked-QM know-QM ‘Who is the person x such that you know where x smoked?’ ‘W hat is the place x such that you know who smoked at x?’ c. Ne-nun [nwu-ka encey ttenassnun-ci] a-ni? you-TOP who-NOM when left-QM know-QM ‘W ho is the person x such that you know when x left?’ ‘W hat is the time x such that you know who left at x?’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 251 d. Ne-nun [nwu-ka cheycwung-i elma nakanun-ci] a-ni? you-TOP who-NOM weight-NOM how much weigh-QM know-QM ‘Who is the person x such that you know how much x w eighs?’ ‘W hat is the weight x such that you know who weighs x?’ e. Ne-nun [nwu-ka ettehkey wassnun-ci] a-ni? you-TOP who-NOM how came-QM know-QM ‘Who is the person x such that you know how x cam e?’ ‘What is the means x such that you know who came by x?’ f. Ne-nun [nwu-ka M ary-eytayhay ettehkey malhayssnunci] a-ni? you-TOP who-NOM M -about how talked-QM know-QM ‘Who is the person x such that you know how x talked about M ary?’ ‘W hat is the manner x such that you know who talked about M ary in x?’ g. Ne-nun [nwu-ka way chayk-ul sassnun-ci] a-ni? you-TOP who-NOM why book-ACC bought-QM know-QM ‘Who is the person x such that you know why x bought a book?’ ‘What is the reason x such that you know who bought a book for x?’ As for the examples in (15a-f), let us suppose the situation where the speaker and the hearer are chatting over John, Mary, Jack, Bill, and Tom, who were at the spring break party, to which each o f them was supposed to invite whoever they wished. Let also suppose a book store situation for the example as in (15g). The hearer o f the questions in (15) o f course can answer these questions by either saying ‘yes’ or ‘no’ in the minimally informative way. Or he can answer these questions by providing information for both wh-words. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 252 W hat is o f our interest here is w hether the reading where the two wh-words can take scope over each other is possible. In fact, the hearer can answer the questions in (15) by giving answers to these questions as in (16-22). (16) a. Na-nun [John-i nwukwu-lul chotayhayssnun-ci] anta. I-TOP J-NOM whom-ACC invited-QM know ‘I know whom John invited.’ b. Na-nun [nwu-ka caki yecachinkwu-lul chotayhayssnun-ci] anta. I-TOP who-NOM his girl friend-ACC invited-QM know ‘I know who invited his girl friend.’ (17) a. Na-nun [John-i eti-se tampay-lul phiwessnun-ci] anta. I-TOP J-NOM where-at cigarette-ACC smoked-QM know ‘I know where John smoked.’ b. Na-nun [nwu-ka silnay-eyse tampay-lul phiwessnun-ci] anta. I-TOP who-NOM room-in cigarette-ACC smoked-QM know ‘I know who smoked in the room .’ (18) a. Na-nun [John-i encey wassnun-ci] anta. I-TOP J-NOM when came-QM know ‘I know when John cam e.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 253 b. Na-nun [nwu-ka ilkopsi-ey wassnun-ci] anta. I-TOP who-NOM 7o’clock came-QM know ‘I know who came at 7 o ’clock.’ (19) a. Na-nun [John-i cheycwung-i I-TOP J-NOM weight-NOM ‘I know how much John w eighs.’ b. Na-nun [nwu-ka cheycwung-i I-TOP who-NOM weight-NOM ‘I know who weighs about 1501bs.’ (20) a. Na-nun [John-i ettehkey wassnun-ci] anta. I-TOP J-NOM how came-QM know ‘I know how John cam e.’ b. Na-nun [nwu-ka bus-lo wassnun-ci] anta. I-TOP who-NOM bus-by came-QM know ‘I know who came by bus.’ (21) a. Na-nun [John-i M ary-eytayhay ettehkey malhayssnun-ci] anta. I-TOP J-NOM M -about how talked-QM know ‘I know how John talked about M ary.’ elm a nakanun-ci] anta. how much weighs-QM know 150 pound cengto nakanun-ci] anta. 150 pound about weighs-QM know Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 254 b. Na-nun [nwu-ka M ary-eytayhay nappukey malhayssnun-ci] anta. I-TOP who-NOM M -about badly talked-QM know ‘I know who talked viciously about M ary.’ (22) a. Na-nun [John-i w ay chayk-ul sassnun-ci] anta. I-TOP J-NOM why book-ACC bought-QM know ‘I know why John bought a book.’ b. Na-nun [nwu-ka sale-cwung ie-se chayk-ul sassnun-ci] anta. I-TOP who-NOM sale-on was-because book-ACC bought-QM know ‘I know who bought a book because it was on sale.’ The answers in (16-22) strongly indicate that there does not exist any asymmetry between adjunct wh-words and argument wh-words in the sense o f Huang (1982), when it comes to W h-island effects. Even the infelicitousness o f the matrix scope construal o f the adjunct way (why) out o f a W h-island found with a wonder type matrix predicate as in (5g) simply disappears in (15g) as shown by the answers in (22) for (5g). The general lack o f W h-island effects when it comes to questions with a know type matrix predicate as illustrated by the answers in (16-22) thus directly contradicts the prediction of Huang (1982), according to which the adjunct wh-words in (15e-g) should exhibit W h-island effects due to ECP violation. Next, consider complex wh-questions with kiekhata (remember) as a matrix predicate as in (23) below. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 255 (23) a. Ne-nun [nwu-ka nwukwu-lul chotayhassnun-ci] kiekha-ni? you-TOP who-NOM whom-ACC invited-QM remember-QM ‘Who is the person x such that you remember whom x invited?’ ‘Who is the person x such that you remember who invited x?’ b. Ne-nun [nwu-ka eti-se tampay-lul phiwessnun-ci] kiekha-ni? you-TOP who-NOM where-at book-ACC smoked-QM remember-QM ‘Who is the person x such that you remember where x smoked?’ ‘W hat is the place x such that you remember who smoked at x?’ c. Ne-nun [nwu-ka encey ttenassnun-ci] kiekha-ni? you-TOP who-NOM when left-QM remember-QM ‘Who is the person x such that you remember when x left?’ ‘W hat is the time x such that you remember who left at x?’ d. Ne-nun [nwu-ka cheycwung-i elma nakanun-ci] kiekha-ni? you-TOP who-NOM weight-NOM how much weighs-QM remember-QM ‘Who is the person x such that you remember how much x w eighs?’ ‘W hat is the weight x such that you remember who weighs x?’ e. Ne-nun [nwu-ka ettehkey wassnun-ci] kiekha-ni? you-TOP who-NOM how came-QM remember-QM ‘Who is the person x such that you remember how x cam e?’ ‘W hat is the means x such that you remember who came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 256 f. Ne-nun [nwu-ka M ary-eytayhay ettehkey malhayssnunci] kiekha-ni? you-TOP who-NOM M -about how talked-QM remember-QM ‘Who is the person x such that you remember how x talked about M ary?’ ‘W hat is the manner x such that you remember who talked about M ary in x?’ g. Ne-nun [nwu-ka way chayk-ul sassnun-ci] kiekha-ni? you-TOP who-NOM why book-ACC bought-QM remember-QM ‘Who is the person x such that you remember why x bought a book?’ ‘W hat is the reason x such that you remember who bought a book for x?’ For the questions in (23a-f), let us again suppose the situation where the speaker and the hearer are chatting over John, Mary, Jack, Bill, and Tom, who were at the spring break party, to which each o f them was supposed to invite whoever they wished. Let also suppose a book store situation for the example in (23 g). The hearer o f the questions in (23) o f course can answer either saying ‘yes’ or ‘no’ in a minimally informative way. O r he can answer by providing information for both wh-words. W hat is o f our interest here is whether the two wh-words can take scope over each other. After all, the hearer can answer the questions in (23) by giving answers to them as in (24- 30). (24) a. Na-nun [John-i nwukwu-lul chotayhayssnun-ci] kiekhanta. I-TOP J-NOM whom-ACC invited-QM remember ‘I remember whom John invited.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 257 b. Na-nun [nwu-ka caki yecachinkwu-lul chotayhayssnun-ci] kiekhanta. I-TOP who-NOM his girl friend-ACC invited-QM remember ‘I remember who invited his girl friend.’ (25) a. Na-nun [John-i eti-se tampay-lul phiwessnun-ci] kiekhanta. I-TOP J-NOM where-at cigarette-ACC smoked-QM remember ‘I remember where John smoked.’ b. Na-nun [nwu-ka silnay-eyse tampay-lul phiwessnun-ci] kiekhanta. I-TOP who-NOM room-in cigarette-ACC smoked-QM remember ‘I remember who smoked in the room .’ (26) a. Na-nun [John-i encey wassnun-ci] kiekhanta. I-TOP J-NOM when came-QM remember ‘I remember when John cam e.’ b. Na-nun [nwu-ka ilkopsi-ey wassnun-ci] kiekhanta. I-TOP who-NOM 7-at came-QM remember ‘I remember who came at 7 o ’clock.’ (27) a. Na-nun [John-i cheycwung-i elm a nakanun-ci] kiekhanta. I-TOP J-NOM weight-NOM how much weigh-QM remember ‘I remember how much John weighs.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 258 b. Na-nun [nwu-ka cheycwung-i 150 pound cengto nakanun-ci] kiekhanta. I-TOP who-NOM weight-NOM 150 pound about weighs-QM remember ‘I remember who weighs about 1501bs.’ (28) a. Na-nun [John-i ettehkey wassnun-ci] kiekhanta. I-TOP J-NOM how came-QM remember ‘I remember how John cam e.’ b. Na-nun [nwu-ka bus-lo wassnun-ci] kiekhanta. I-TOP who-NOM bus-by came-QM remember ‘I remember who came by bus.’ (29) a. Na-nun [John-i M ary-eytayhay ettehkey malhayssnun-ci] kiekhanta. I-TOP J-NOM M -about how talked-QM remember ‘I remember how John talked about M ary.’ b. Na-nun [nwu-ka M ary-eytayhay nappukey malhayssnun-ci] kiekhanta. I-TOP who-NOM M -about badly talked-QM remember ‘I remember who talked viciously about M ary.’ (30) a. Na-nun [John-i way chayk-ul sassnun-ci] kiekhanta. I-TOP J-NOM why book-ACC bought-QM remember ‘I remember why John bought a book.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 259 b. Na-nun [nwu-ka sale-cwung ie-se chayk-ul sassnun-ci] kiekhanta. I-TOP who-NOM sale-on was-because book-ACC bought-QM remember ‘I remember who bought a book because it was on sale.’ The pattern o f answers in (24-30) for the questions with kiekhata (remember) as matrix predicate in (23) again shows that both argument wh-words and adjunct wh-words can take matrix scope out o f a W h-island, like questions with alta (know) as matrix predicate, again posing a nontrivial problem to Haung (1982) according to which adjunct wh-words should show W h-island effects. To summarize, complex questions with alkosipta (want to know) as a matrix predicate and complex questions with alta (know) or kiekhata (remember) as a matrix predicate seem to behave differently in one important aspect. W hen it comes to questions with predicates of the former type, matrix scope o f adjunct wh-word way (why) out o f a W h-island is infelicitous, which I suggested essentially has to do with the pragmatic factor o f domain of quantification. M eanwhile, when it comes to questions with alta (know) or kiekhata (remember) type matrix predicates, where the aforementioned pragmatic factor can be controlled, even the infelictousness o f matrix scope for the adjunct wh-word way (why) out o f a W h-island simply disappears. This state o f affairs is actually expected given the presupposition o f knowledge carried by know type o f predicates. The felicitousness o f the matrix scope o f the adjunct wh-word way (why) in wh-questions with alta (know) or kiekhata (remember) type matrix predicates thus lends support to our suggestion that the infelictousness o f the matrix scope reading o f way (why) in questions with a wonder type Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 260 predicate is essentially due to the pragmatic factor o f domain o f quantification but not Wh- island effects. Alternatively, one may suggest that the matrix scope o f the adjunct wh-word way (why) out o f Wh-islands with alta (know) and kiekhata (remember) may be possible due to the semantics o f these predicates. This position does not seem to be correct, however, given the ungrammaticality o f the sentences in English below in (31) where the adjunct wh-word why takes matrix scope out o f the Wh-islands. (31) a. *Why do you remember who met John? b. *Why do you know who met John? The ungrammaticality o f the examples in (31) with the adjunct wh-word why in (31) taking matrix scope out o f the W h-island, suggests that W h-island effects cannot be canceled by the semantic factor of presuppostion o f knowledge. This state o f affairs in English further indicates that wh-words in Korean do not show W h-island effects and that the infelictousness o f the matrix scope reading o f the adjunct wh-word way (why) with a wonder type matrix predicate as in (5g) as repeated in (32) is indeed due to the pragmatic factor o f domain o f quantification. (32) Ne-nun [nwu-ka w ay nucessnun-ci] alko sip-ni? you-TOP who-NOM why came late-QM want to know-QM ‘Who is the person x such that you want to know why x came late?’ # ‘W hat is the reason x such that you want to know who came late for x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 261 To the extent that the observation that I make in this chapter is correct, i.e., the general lack o f W h-island effects, Huang’s (1982) observation for the existence o f the W h-island effects and his account for them, which is based on argument and adjunct dichotomy under the uniform LF wh-movement hypothesis are seriously undermined. In his system, the ability to scope out o f the W h-island is predicted to vary, depending on which type o f wh-word it is, an argument wh-word or an adjunct wh-word. I will thus present an alternative analysis for the surprising symmetry o f locality effects o f wh-words in Korean type languages in section 3, i.e., the lack o f W h-island effects o f wh- words. Before doing so, I need to examine W h-island effects in Chinese and Japanese, which are typologically akin to Korean in that these two languages also belong to wh-in-situ languages. 2.2. Chinese and Japanese Let us turn to Chinese first. Consider the following questions in (33) with xiang-zhidao (want to know) as a matrix predicate. The example in (3e) from Huang (1982) is repeated as (33g): (33) a. ni xiang-zhidao [shei yaoqing-le shei]? you want to know who invited-ASP whom ‘Who is the person x such that you want to know whom x invited?’ ‘Who is the person x such that you want to know who invited x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 262 b. ni xiang-zhidao [shei zai nali xi yan-le]? you want to know who at where smoke-ASP ‘Who is the person x such that you want to know where x smoked?’ ‘W hat is the place x such that you want to know who smoked at x?’ c. ni xiang-zhidao [shei shenmeshihou zou-le]? you want to know who when leave-ASP ‘Who is the person x such that you want to know when x left?’ ‘W hat is the time x such that you want to know who left at x?’ d. ni xiang-zhidao [shei zhong ji bang]? you want to know who amount how much ‘Who is the person x such that you want to know how much x w eighs?’ ‘W hat is the weight x such that you want to know who weighs x?’ e. ni xiang-zhidao [shei zenmeyang lai-de]? you want to know who how come-ASP ‘Who is the person x such that you want to know how x cam e?’ ‘W hat is the means x such that you want know who came by x?’ f. ni xiang-zhidao [shei zenmeyang tanlun Mary]? you want to know who how talked Mary ‘Who is the person x such that you want to know how x talked about M ary?’ ‘W hat is the manner x such that you want to know who talked about Mary in x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 263 g. ni xiang-zhidao [ shei weishenme mai-le shu] ? you w onder who why buy-ASP book ‘Who is the person x such that you wonder why x bought a book?’ # ‘W hat is the reason x such that you wonder who bought a book for x?’ For the questions in (33a-f), let us continue to suppose the situation where the speaker and the hearer are chatting over John, Mary, Jack, Bill, and Tom, who were at the spring break party, to which each o f them was supposed to invite whoever they wished. Let also suppose a book store situation for the example as in (33g). The hearer o f the questions above in (33) can give answers to these questions by saying answers as in (34-40): (34) a. wo xiang-zhidao [John yaoqing-le shei] I want to know John invite-ASP whom ‘I want to know whom John invited.’ b. wo xiang-zhidao [shei yaoqing-le ziji de nu peng you] I want to know who invite-ASP his girl friend ‘I want to know who invited his girl friend.’ (35) a. wo xiang-zhidao [John zai nali xi yan-le] I want to know John at where smoke-ASP ‘I want to know where John smoked.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 264 b. wo xiang-zhidao [shei zai shine xi yan-le] I want to know who at room smoke-ASP ‘I want to know who smoked in the room .’ (36) a. wo xiang-zhidao [John shenmeshihou zou-le] I want to know John when leave-ASP ‘I want to know when John left.’ b. wo xiang-zhidao [shei jiu dian zou-le] I want to know who at 9 leave-ASP ‘I want to know who left at 9 o ’ clock.’ (37) a. wo xiang-zhidao [John zhong ji bang] I want to know John amount how much ‘I want to know how much John w eighs.’ b. wo xiang-zhidao [shei zhong yue 150 bang] I want to know who amount about 1501bs ‘I want to know who weighs about 150 pounds.’ (38) a. wo xiang-zhidao [John zenmeyang tanlun Mary] I want to know John how talked Mary ‘I want to know how John talked about M ary.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 265 b. wo xiang-zhidao [shei edudi tanlun Mary] I want to know who viciously talked Mary ‘I want to know who talked viciously about M ary.’ (39) a. wo xiang-zhidao [John zenmeyang lai-de] I want to know John how come-ASP ‘I want to know how John cam e.’ b. wo xiang-zhidao [shei da bus lai-de] I want to know who by bus come-ASP ‘I want to know who came by bus.’ (40) wo xiang-zhidao [John weishenme I want to know John why ‘I want to know why John bought a book.’ The pattern o f answers in (34-40) reveals that the adjunct wh-word zenmeyang (how) in (33ef) either under means or manner construal can take scope out o f the W h-island while matrix scope for the adjunct weishenme (why) out o f the W h-island is infelicitous. Thus the contrast in W h-island effects is essentially between the adjunct weishenme (why), and the rest o f wh-words, which again poses a nontrivial problem to the ECP analysis o f Huang (1982), based on the dichotomy o f argument wh-words and adjunct wh-words accroding to which adjunct wh-words including zenmeyang (how) should exhibit W h-island effects. mai-le shu] buy-ASP book Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 266 At this point, I should note that the intuitions vary when it comes to questions with a wonder type matrix predicate in Chinese. Shi (1994) originally observes that even the adjunct wh- word weishenme (why) can take matrix scope out o f the Wh-island. He reports that the following sentence in (41) does not show any asymmetry between the argument wh-word shei (who) and the adjunct wh-word weishenme (why) with respect to W h-island effects: (41) ni xiang-zhidao [shei weishenme cizhi] you want to know who why resigned ‘Who is the person x such that you want to know why x resigned?’ ‘W hat is the reason x such that you want to know who resigned for x?’ (Shi 1994:315) To the extent that the matrix scope o f the adjunct wh-word weishenme (why) is a possibility even with a xiang-zhidao (wonder) type matrix predicate in (41) as observed by Shi, it may be interpreted as supporting our present proposal for the pragmatic factor o f domain o f quantification. Although normally the domain o f quantification for the adjunct wh-word weishenme (why) is unlimited, if there is indeed good reason determined by the context for the hearer to believe that at least one o f the individuals the speaker and the hearer are talking about is affected by that reason, he [the hearer] can answer the question with a matrix scope of the adjunct wh-word, solociting information for shei (who). This particular context delimiting the domain o f quantification o f the adjunct wh-word thus may facilitate the matrix scope o f the adjunct wh-word weishenme (why) in (41). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 267 Next let us consider complex wh-questions with zhidao (know) as the matrix predicate, as in (42) below. (42) a. ni zhidao [shei yaoqing-le shei]? you know who invite-ASP whom ‘Who is the person x such that you know whom x invited?’ ‘Who is the person x such that you know who invited x?’ b. ni zhidao [shei zai nali xi yan-le]? you know who at where smoke-ASP ‘Who is the person x such that you know where x smoked?’ ‘W hat is the place x such that you know who smoked at x ?’ c. ni zhidao [shei shenmeshihou zou-le]? you know who when leave-ASP ‘Who is the person x such that you know when x left?’ ‘What is the time x such that you know who left at x?’ d. ni zhidao [shei zhong ji bang]? you know who amount how much ‘Who is the person x such that you know how much x w eighs?’ ‘What is the weight x such that you know who weighs x?’ e. ni zhidao [shei zenmeyang lai-de]? you know who how come-ASP ‘Who is the person x such that you know how x cam e?’ ‘W hat is the means x such that you know who came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 268 f. ni zhidao [shei zenmeyang tanlun M ary]? you know who how talked Mary ‘Who is the person x such that you know how x talked about M ary?’ ‘What is the manner x such that you know who talked about M ary in x?’ g. ni zhidao [shei weishenme mai-le shu]? you know who why buy-ASP book ‘Who is the person x such that you know why x bought a book?’ ‘What is the reason x such that you know who bought a book for x?’ For the questions in (42a-f), again suppose the situation where the speaker and the hearer are chatting over John, Mary, Jack, Bill, and Tom, who were at the spring break party, to which each of them was supposed to invite whoever they wished. As for the question in (42g), suppose a book store situation. Given the questions in (42), the hearer can answer these questions by giving answers to them as in (43-49), which again shows that adjunct wh-words as well as argument wh-words can take scope out o f Wh-islands. (43) a. wo zhidao [John yaoqing-le shei] I know John invite-ASP whom ‘I know whom John invited.’ b. wo zhidao [shei yaoqing-le ziji de nu peng you] I know who invite-ASP his girl friend ‘I know who invited his girl friend.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 269 (44) a. wo zhidao [John zai nali xi yan-le] I know John at where smoke-ASP ‘I know where John smoked.’ b. wo zhidao [shei zai shine xi yan-le] I know who at room smoke-ASP ‘I know who smoked in the room .’ (45) a. wo zhidao [John shenmeshihou zou-le] I know John when leave-ASP ‘I know when John left.’ b. wo zhidao [shei jiu dian zou-le]? I know who at 9 leave-ASP ‘I know who left at 9 o ’clock.’ (46) a. wo zhidao [John zhong ji bang] I know John amount how much ‘I know how much John w eighs.’ b. wo zhidao [shei zhong yue 150 bang] I know who amount about 1501bs ‘I know who weighs about 150 pounds.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 270 (47) a. wo zhidao [John zenmeyang lai-de] I know John how come-ASP ‘I know how John cam e.’ b. wo zhidao [shei da bus lai-de] I know who by bus come-ASP ‘I know who came by bus.’ (48) a. wo zhidao [John zenmeyang tanlun Mary] I know John how talked Mary ‘I know how John talked about M ary.’ b. wo zhidao [shei edudi tanlun Mary] I know who viciously talked Mary ‘I know who talked badly about M ary.’ (49) a. wo zhidao [John weishenme m ai-le shu] I know John why buy-ASP book ‘I know why John bought a book.’ b. wo zhidao [shei yinwei zai jiang-jia mai-le shu] I know who because were on sale buy-ASP book ‘I know who bought a book because it was on sale.’ The answering pattern as shown in (43-49) indicates that when the matrix predicate is zhidao (know), even the infelicitousness o f the reading where weishenme (why) takes scope out of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 271 the W h-island disappears, thus exhibiting no W h-island effects at all, exactly like K o rean .6 The general lack o f asymmetry for wh-words scoping out of W h-islands illustrated by the answers in (43-49) thus directly contradicts the prediction of Huang (1982), according to which the adjunct wh-words in (42e-g) should exhibit W h-island effects due to ECP violation. This state o f affairs in Chinese when it comes to wh-questions with a zhidao (know) type matrix predicate can be understood as suggesting that the infelicitousness o f the matrix scope reading o f weishenme (why) out o f the W h-island when it comes to wh-questions with a wonder type matrix predicate is essentially due to the pragmatic factor o f domain o f quantification but not W h-island effects, as suggested in section 2.1 for Korean. Let us turn to Japanese now. The judgments reported in the literature vary. Nishigauchi (1990), for example, reports that the sentence in (50) below has only a yes-no question reading, exhibiting W h-island effects (see Shimoyama 2001 for similar observation). (50) Tanaka-kun-wa [dare-ga nani-o tabeta-ka] oboteimasu-ka? T-TOP who-NOM what-ACC eat remember-QM ‘Does Tanaka know who ate w hat?’ # ‘W ho is the person x such that Tanaka remembers what x ate?’ # ‘W hat is the thing x such that Tanaka remembers who ate x?’ Nishigauchi (1990: 28) 6 Native speakers whom I consulted mostly agree that the adjunct wh-word weishenme (why) in questions with zhidao (know) as a matrix predicate as in (42) can take scope out of a Wh-island. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 272 Nishigauchi (1990: 29), assuming the uniform LF wh-movement hypothesis following Huang (1982), claims that the second wh-word in (50) undergoes movement to the embedded QM and thus forms the W h-island for the matrix scope o f the first wh-word, suggesting that wh- word movement is rightward, given that Japanese is a strictly head-final language. Based on this observation about the scope o f the wh-words in the above example in (50), Nishigauchi (1990: 10, 28) argues that Japanese has ‘covert’ version o f W h-islands on a par with those created overtly in English as in (1), repeated in (51), such that wh-words in Japanese are subject to W h-island effects. 7 (51) a. ?* W hat do you wonder who saw? b. *How do you wonder who fixed the car? (Chomsky 1986: 48-49) However, the intuitions reported in Nishigauchi (1990) do not seem to be readily shared by other linguists. 8 Richards (2000), for example, notes that argument wh-words as in (52) in Japanese can take scope over each other, although not that natural. 7 Nishigauchi (1990: 11), however, notes that some people can have a matrix scope reading of the subject wh-word, which he attributes to a discourse-linking effect. 8 Dayal (1996: 93) reports that Japanese speakers have no problem accepting matrix scope for the object wh-word in the following sentence: Dare-ga [ Mary-ga doko-de nani-o kat-ta ka ] sitte-imasu-ka? who-NOM M-NOM where-at what-ACC bought-QM know-QM ‘Who knows where Mary bought what?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 273 (52) Keesatu-wa [dare-ga dare-o korosita-ka] sirabeteiru-no? police-TOP who-NOM whom-ACC killed-QM are invesitigating-QM (Richards 2000: 188) The present research further suggests that even adjunct wh-words can take scope out o f Wh- islands in Japanese. For this, consider wh-questions in (53) below with siritagateru (want to know) as a matrix predicate, first. (53) a. Kimi-wa [dare-ga dare-o sootaisita-ka] siritagateru-no? you-TOP who-NOM whom-ACC invited-QM want to know-QM ‘Who is the person x such that you want to know whom x invited?’ ‘Who is the person x such that you want to know who invited x?’ b. Kimi-wa [dare-ga doko-de tabako-o sutta-ka] siritagateru-no? you-TOP who-NOM where-at tobacco-ACC smoked-QM want to know-QM ‘Who is the person x such that you want to know where x sm oked?’ ‘W hat is the place x such that you know who smoked at x?’ c. Kimi-wa [dare-ga itsu kaetta-ka] siritagateru no? you-TOP who-NOM when leave-QM want to know-QM ‘Who is the person x such that you want to know when x left?’ ‘W hat is the time x such that you want to know who left at x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 274 d. Kimi-wa [dare-ga taijuu-ga ikutsu aru-ka] siritagateru-no? you-TOP who-NOM weight-NOM how much weighs-QM want to know-QM ‘Who is the person x such that you want to know how much x w eighs?’ ‘W hat is the weight x such that you want to know who weighs x?’ e. Kimi-wa [ dare-ga dooyatte kita-ka ] siritagateru no? you-TOP who-NOM how came-QM want to know-QM ‘W ho is the person x such that you want to know how x cam e?’ ?‘W hat is the means x such that you want to know who came by x?’ f. Kimi-wa [dare-ga Tom-nituite doo itteita-ka] siritagateru-no? you-TOP who-NOM Tom-about how was saying-QM want to know-QM ‘Who is the person x such that you want to know how x talked about Tom ?’ ?‘W hat is the manner x such that you want to know who talked about Tom in x?’ g. Kimi-wa [dare-ga naze hon-o katta-ka] siritagateru no? you-TOP who-NOM why book-ACC bought-QM want to know-QM ‘Who is the person x such that you want to know why x bought a book?’ # ‘W hat is the reason x such that you want to know who bought a book for x?’ Please recall that in Japanese, dare (who), dono-N (which N), and ikutsu (how much) are bona fide argument wh-words, according to the argument/adjunct criterion in Huang (1982). W hen it comes to wh-words such as doko (where) and itsu (when), they can be viewed as argument wh-words since they can come together with a postposition, showing that they can be arguments o f a lexical category. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 275 Dooyatte (how), doo (how) and naze (why) are genuine adjunct wh-words, since they cannot be arguments o f a lexical category. For the questions in (53a-f), let us keep supposing the situation where the speaker and the hearer are chatting over John, Mary, Jack, Bill, and Tom, who were at the spring break party, to which each o f them was supposed to invite whoever they wished. Let also suppose a book store situation for the example in (53g). Now given the above questions in (53), the hearer may most easily answer these questions by simply saying ‘yes’ or ‘no’. W hat we are interested in is whether these questions in (53) can have the reading where the two wh-words in the embedded clause can take scope over each other, showing no W h-island effects. In fact, one can answer the above questions by giving answers as in (54-60): (54) a. Boku-wa [John-ga dare-o sootaisita-ka] siritagateru. I-TOP J-NOM whom-ACC invited-QM want to know ‘I want to know whom John invited.’ b. Boku-wa [dare-ga zibun-no koibito-o sootaisita-ka] siritagateru. I-TOP who-NOM his girl friend-ACC invited-QM want to know ‘I want to know who invited his girl friend.’ (55) a. Boku-wa [John-ga doko-de tabako-o sutta-ka] siritagateru. I-TOP J-NOM where-at tobacco-ACC smoked-QM want to know ‘I want to know where John smoked.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 276 b. Boku-wa [dare-ga heya-de tabako-o sutta-ka] siritagateru. I-TOP who-NOM room-in tabacco-ACC smoked-QM want to know ‘I want to know who smoked in the room .’ (56) a. Boku-wa [John-ga itsu kaetta-ka] siritagateru. I-TOP J-NOM when leave-QM want to know ‘I want to know when John left.’ b. Boku-wa [dare-ga kuzi-ni kaetta-ka] siritagateru. I-TOP who-NOM 9-at leave-QM want to know ‘I want to know who left at 9 o ’clock.’ (57) a. Boku-wa [John-ga taijuu-ga ikutsu aru-ka] siritagateru. I-TOP J-NOM weight-NOM how much weighs-QM want to know ‘I want to know how much John w eighs.’ b. Boku-wa [dare-ga taijuu-ga daitai 70kg am-ka] siritagateru. I-TOP who-NOM weight-NOM about 70 weighs-QM want to know ‘I want to know who weighs about 70kg.’ (58) a. Boku-wa [John-ga dooyatte kita-ka] siritagateru. I-TOP J-NOM how came-QM want to know ‘I want to know how John came to the party.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 277 b. Boku-wa [dare-ga basu-de kita-ka] siritagateru. I-TOP who-NOM bus-by came-QM want to know ‘I want to know who came by bus.’ (59) a. Boku-wa [John-ga Tom-nituite doo itteita-ka] siritagateru. I-TOP J-NOM Tom-about how said-QM want to know ‘I want to know how John talked about Tom .’ b. Boku-wa [dare-ga Tom-nituite waruku itteita-ka] siritagateru. I-TOP who-NOM Tom-about badly said-QM want to know ‘I want to know who talked badly about Tom .’ (60) Boku-wa [John-ga naze hon-o katta-ka] siritagateru. I-TOP J-NOM why book-ACC bought-QM want to know ‘I want to know why John bought a book.’ The pattern o f answers in (54-60) for the questions in (53) shows that not only argument wh- words but also adjunct wh-words dooyatte (how) under means construal and doo (how) under manner construal can scope out o f W h-islands. The matrix scope o f the adjunct naze (why) is infelicitous. 9 The scope facts as shown in the answering pattern in (54-60) thus suggests that the contrast in W h-island effects is essentially between the adjunct naze (why), and the rest o f wh-words, which poses a nontrivial problem to the ECP analysis o f Huang (1982), based on the dichotomy o f argument wh-words and adjunct wh-words accroding to 9 Yasuhiko Miura (personal communication), our primary informant, notes that the matrix scope of the adjunct wh-words dooyatte (how) and doo (how) in (53) is slightly deviant. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 278 which adjunct wh-words should not take matrix scope out of the W h-islands. Next consider questions with the matrix predicate sitteimasu (know) as in (61). (61) a. Kimi-wa [dare-ga dare-o sootaisita-ka] sitteimasu-ka? you-TOP who-NOM whom-ACC invited-QM know-QM ‘Who is the person x such that you know whom x invited?’ ‘W ho is the person x such that you know who invited x?’ b. Kimi-wa [dare-ga doko-de tabako-o sutta-ka] sitteimasu-ka? you-TOP who-NOM where-at book-ACC smoked-QM know-QM ‘Who is the person x such that you know where x smoked?’ ‘W here is the place x such that you know who smoked at x?’ c. Kimi-wa [dare-ga itsu kaetta-ka] sitteimasu-ka? you-TOP who-NOM when leave-QM know-QM ‘Who is the person x such that you know when x left?’ ‘W hen is the time x such that you know who left at x ?’ d. Kim i-wa [dare-ga taijuu-ga ikutsu aru-ka] sitteimasu-ka? you-TOP who-NOM weight-NOM how much weigh-QM know QM ‘Who is the person x such that you know how much x w eighs?’ ‘W hat is the weight x such that you know who weighs x?’ e. Kimi-wa [dare-ga dooyatte kita-ka] sitteimasu-ka? you-TOP who-NOM how came-QM know-QM ‘W ho is the person x such that you know how x cam e?’ ‘W hat is the means x such that you know who came by x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 279 f. Kimi-wa [dare-ga Tom-nituite doo itteita-ka] sitteimasu-ka? you-TOP who-NOM Tom-about how was saying-QM know-QM ‘Who is the person x such that you know how x talked about Tom ?’ ‘W hat is the manner x such that you know who talked about Tom in x ?’ g. Kimi-wa [dare-ga naze hon-o katta-ka] sitteimasu-ka? you-TOP who-NOM why book-ACC bought-QM know-QM ‘Who is the person x such that you know why x bought a book?’ ‘W hat is the reason x such that you know who bought a book for x?’ Let us suppose for the questions in (61a-f) the situation where the speaker and the hearer are chatting over John, Mary, Jack, Bill, and Tom, who were at the spring break party, to which each o f them was supposed to invite whoever they wished. Let us also suppose a book store situation for the example as in (61g). The hearer o f the above questions in (61) indeed can answer these questions by giving answers as in (62-68). (62) a. Boku-wa [John-ga dare-o sootaisita-ka] sitteimasu. I-TOP J-NOM whom-ACC invited-QM know ‘I know whom John invited.’ b. Boku-wa [dare-ga zibun-no koibito-o sootaisita-ka] sitteimasu. I-TOP who-NOM his girl friend-ACC invited-QM know ‘I know who invited his girl friend.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 280 (63) a. Boku-wa [John-ga doko-de tabako-o sutta-ka] sitteimasu. I-TOP J-NOM where-at tobacco-ACC smoked-QM know ‘I know where John sm oked.’ b. Boku-wa [dare-ga heya-de tabako-o sutta-ka] sitteimasu. I-TOP who-NOM room-in tobacco-ACC smoked-QM know I know who smoked in the room .’ (64) a. Boku-wa [John-ga itsu kaetta-ka] sitteimasu. I-TOP J-NOM when leave-QM know ‘I know when John left.’ b. Boku-wa [dare-ga kuzi-ni kaetta-ka] sitteimasu. I-TOP who-NOM 9-at leave-QM know ‘I know who left at 9 o ’clock.’ (65) a. Boku-wa [John-ga taijuu-ga ikutsu aru-ka] sitteimasu. I-TOP J-NOM weight-NOM how much weighs-QM know ‘I know how much John w eighs.’ b. Boku-wa [dare-ga taijuu-ga daitai 70kg aru-ka] sitteimasu. I-TOP who-NOM weight-NOM about 70kg weighs-QM know ‘I know who weighs about 70kg.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 281 (66) a. Boku-wa [John-ga dooyatte kita-ka] sitteimasu. I-TOP J-NOM how came-QM know ‘I know how John cam e.’ b. Boku-wa [dare-ga basu-de kita-ka] sitteimasu. I-TOP who-NOM bus-by came-QM know ‘I know who came by bus.’ (67) a. Boku-wa [John-ga Tom-nituite doo itteita-ka] sitteimasu. I-TOP J-NOM Tom-about how said-QM know ‘I know how John sad about Tom .’ b. Boku-wa [dare-ga Tom-nituite waruku itteita-ka] sitteimasu. I-TOP who-NOM Tom-about badly said-QM know ‘I know who said badly about Tom .’ (68) a. Boku-wa [John-ga naze okurete kita-ka] sitteimasu. I-TOP J-NOM why late came-QM know ‘I know why John came late.’ b. Boku-wa [dare-ga seeru dakara hon-o katta-ka] sitteimasu. I-TOP who-NOM sale because book-ACC bought-QM know ‘I know who bought a book because it was on sale.’ As shown by the pattern o f answers in (62-68), there does not exist any asymmetry between adjunct wh-words on the one hand and argument wh-words on the other in wh-questions Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 282 with sitteimasu (know) as matrix predicate when it comes to scoping out o f a W h-island. This state o f affairs in wh-questions with a sitteimasu (know) type matrix predicate can be understood as indicating that the infelicitousness o f the matrix scope reading o f naze (why) out o f the W h-island in wh-questions with a siritagateru (wonder) type matrix predicate is essentially due to the proposed pragmatic factor o f domain o f quantification but not Wh- island effects. To summarize, adjunct wh-words as well as argument wh-words in Korean type languages can take scope out o f W h-islands without exhibiting any asymmetry, to the extent that the infelicitous status o f the reading where the adjunct wh-word corresponding to English why scopes out o f a W h-island in wh-questions with a wonder type matrix predicate is due to the pragmatic factor o f domain o f quantification. If the present observation and suggestion are correct, the lack o f W h-islands in Korean type languages poses a serious problem to Huang (1982), according to which adjunct wh-words cannot scope out o f W h-islands due to the ECP. The present observation also undermines the proposal for LF pied piping in Nishigauchi (1990) for other syntactic islands as discussed in Chapter 4, which is crucially based on the W h-island effects in Japanese. 3. Analysis Please recall the scope taking strategies o f wh-words in Korean type languages: A nonpropositional wh-word is unselectively bound by the question morpheme that serves as a wh-operator and marks its scope, while the propositional adjunct wh-word, not being an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 283 indefinite, should undergo movement at LF into Spec o f CP to mark its scope via spec-head agreement with the QM, driven by the need for proper interpretation. W ith this asymmetric scope taking mechanism in mind, let us turn to the analysis o f scope taking in W h-island constructions in Korean type languages. For this, let us consider the examples in (15), (42) and (61) with a know type matrix predicate, as repeated in (69-71). (69) Korean a. Ne-nun [nwu-ka nwukwu-lul chotayhayssnun-ci] a-ni? you-TOP who-NOM whom-ACC invited-QM know-QM ‘W ho is the person x such that you know whom x invited?’ ‘Who is the person x such that you know who invited x?’ b. Ne-nun [nwu-ka eti-se tampay-lul phiwessnun-ci] a-ni? you-TOP who-NOM where-at book-ACC smoked-QM know-QM ‘W ho is the person x such that you know where x smoked?’ ‘W hat is the place x such that you know who smoked at x ?’ c. Ne-nun [nwu-ka encey ttenassnun-ci] a-ni? you-TOP who-NOM when left-QM know-QM ‘W ho is the person x such that you know when x left?’ ‘W hat is the time x such that you know who left at x?’ d. Ne-nun [nwu-ka cheycwung-i elma nakanun-ci] a-ni? you-TOP who-NOM weight-NOM how much weigh-QM know-QM ‘Who is the person x such that you know how much x weighs?’ ‘What is the weight x such that you know who weighs x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 284 e. Ne-nun [nwu-ka ettehkey wassnun-ci] a-ni? you-TOP who-NOM how came-QM know-QM ‘W ho is the person x such that you know how x came?’ ‘W hat is the means x such that you know who came by x?’ f. Ne-nun [nwu-ka M ary-eytayhay ettehkey malhayssnunci] you-TOP who-NOM M -about how talked-QM ‘Who is the person x such that you know how x talked about M ary?’ ‘W hat is the manner x such that you know who talked about M ary in x' (70) Chinese a. ni zhidao [shei yaoqing-le shei]? you know who invite-ASP whom ‘Who is the person x such that you know whom x invited?’ ‘Who is the person x such that you know who invited x?’ b. ni zhidao [shei zai nali xi yan-le]? you know who at where smoke-ASP ‘Who is the person x such that you know where x smoked?’ ‘W hat is the place x such that you know who smoked at x?’ c. ni zhidao [shei shenmeshihou zou-le]? you know who when leave-ASP ‘Who is the person x such that you know when x left?’ ‘W hat is the time x such that you know who left at x?’ a-ni? know-QM Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 285 d. ni zhidao [shei zhong ji bang]? you know who amount how much ‘Who is the person x such that you know how much x w eighs?’ ‘W hat is the weight x such that you know who weighs x?’ e. ni zhidao [ shei zenmeyang tanlun Jack]? you know who how talked Jack ‘Who is the person x such that you know how x talked about Jack?’ ‘W hat is the manner x such that you know who talked about Jack in x ?’ f. ni zhidao [shei zenmeyang lai-de]? you know who how come-ASP ‘Who is the person x such that you know how x cam e?’ ‘W hat is the means x such that you know who came by x?’ (71) Japanese a. Kimi-wa [dare-ga dare-o sootaisita-ka] sitteimasu-ka? you-TOP who-NOM whom-ACC invited-QM know-QM ‘Who is the person x such that you know whom x invited?’ ‘Who is the person x such that you know who invited x?’ b. Kimi-wa [dare-ga doko-de tabako-o sutta-ka] sitteimasu-ka? you-TOP who-NOM where-at book-ACC smoked-QM know-QM ‘Who is the person x such that you know where x smoked?’ ‘W hat is the place x such that you know who smoked at x?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 286 c. Kimi-wa [dare-ga itsu kaetta-ka] sitteimasu-ka? you-TOP who-NOM when leave-QM know-QM ‘Who is the person x such that you know when x left?’ ‘W hat is the time x such that you know who left at x ?’ d. Kimi-wa [dare-ga taijuu-ga ikutsu aru-ka] sitteimasu-ka? you-TOP who-NOM weight-NOM how much weigh-QM know QM ‘Who is the person x such that you know how much x weighs?’ ‘W hat is the weight x such that you know who weighs x?’ e. Kimi-wa [dare-ga dooyatte kita-ka] sitteimasu-ka? you-TOP who-NOM how came-QM know-QM ‘Who is the person x such that you know how x cam e?’ ‘W hat is the means x such that you know who came by x?’ f. Kimi-wa [dare-ga Tom-nituite doo itteita-ka] sitteimasu-ka? you-TOP who-NOM Tom-about how was saying-QM know QM ‘Who is the person x such that you know how x talked about Tom ?’ ‘W hat is the manner x such that you know who talked about Tom in x?’ The wh-questions in (69-71) with nonpropositional wh-words in the embedded clause can have both the reading where the subject wh-word takes matrix scope and the other wh- word takes embedded scope, and the reading where these wh-words take the inverse scope, i.e., the embedded scope for the subject wh-word and the matrix scope for the other wh- word. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 287 At this point, please recall that in Chapter 2 and Chapter 3, I took the position that QM is base-generated in the head o f IP in Korean and Japanese, following Kim (1990), while suggesting that it is in the head o f Topic Phrase in Chinese. Abstracting away from the parametric variation regarding position o f the QM in these languages, the relevant configuration where nonpropositional wh-words as in (69-71) can take scope over each other will be the one where the QM in the matrix clause and the one in the embedded clause each bind a w h-w ord.1 0 Thus when it comes to the reading where the subject wh-word and the other wh-word in (69-71) take matrix and embedded scope, respectively, the reading is derived by the unselective binding o f the subject wh-word by the matrix QM and the other wh-word by the embedded QM, respectively, as illustrated by the LF representation below in (72) with the irrelevant details suppressed. (72) [C P [QMj [V P V [C P [ subject WH-word, [ QMj [ V P V WH-word, ]]]]]]] Although Korean and Japanese are strictly head-final languages while Chinese is not, I will notate the LF representations as if they were all head-initial languages throughout this chapter, purely for the reader’s convenience. I mean by the QM the one with the feature specification o f [+Q, +WH] hereafter without otherwise specificed. W hen it comes to the in versed reading where the subject wh-word takes embedded scope and the other wh-word in (69-71) takes the matrix scope, the reading is derived by the unselective binding o f the subject wh-word by the embedded QM and the other wh-word by 1 0 An abstract QM should be posited for Chinese questions as in (72), since it is not realized phonologically in the embedded clause while optionally realized in the matrix clause. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 288 the matrix QM, respectively, with the embedded QM raised into a position, i.e., the head o f the embedded CP such that it can bind the subject wh-word, as illustrated by the LF representation in (73) with the irrelevant details suppressed. (73) [C P [ QMj [ vr V [C P [c QM, [ subject WH-wordi [ t, [ V p V WH-word, ]]]]]]]] Since the QMs in Korean type languages are base-generated in a position lower than the subject as suggested in Chapter 2 and Chapter 3, the embedded QM in (73) should be raised into the head o f the embedded CP from which it can bind the subject wh-word, which is a nonpropositional wh-word and thus is an indefinite. Otherwise the indefinite wh-word will not be properly interpreted. 1 1 1 2 As shown thus far, the unselective binding o f nonpropositional wh-words by the QM essentially accounts for the reading where wh-words in (69-71) take scope over each other without exhibiting W h-island effects. W hy do the questions in (69-71) not exhibit W h-island effects? Assuming W h-island constraint as subsumed under the explanatory um brella of 1 1 Suppose a situation where the embedded QM unselectively binds both wh-words in (73). This possibility should be excluded, since the matrix QM with [+Q,+WH] will end up with a vacuous quantification without binding any wh-word. 12 The unselective binding by the matrix QM of a wh-word within the scope of another unselective binder as in (73) is legitimate as shown by the example below. Nwu-jka wa-to proj taycheylo wuli-ul pangmwunhanta. who-NOM come-TO usually us-ACC visit ‘for most x, x an individual, if x comes, x visits us.’ ‘for every x, x an individual, if x comes, x usually visits us.’ The first interpretation clearly indicates that the unselective binder, i.e., the adverbial quantifier in the main clause can bind the indefinite within the scope of another binder, i.e., universal quantifier, to (see Reinhart 1997 for essentially the same view). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 289 subjacency in Chomsky (1986), W h-island constraint violation is not an issue from the very beginning, since none o f the two wh-words undergo movement at LF in (69-71). Next let us turn to examples with a nonpropositional wh-word and a propositional adjunct wh-word in the embedded clause as in (15g), (42g) and (61 g), repeated in (74-76). (74) Korean Ne-nun [nwu-ka way chayk-ul sassnun-ci] a-ni? you-TOP who-NOM why book-ACC bought-QM know-QM ‘Who is the person x such that you know why x bought a book?’ ‘W hat is the reason x such that you know who bought a book for x?’ (75) Chinese ni zhidao [shei weishenme mai-le shu]? you know who why buy-ASP book ‘Who is the person x such that you know why x bought a book?’ ‘W hat is the reason x such that you know who bought a book for x?’ (76) Japanese Kimi-wa [dare-ga naze hon-o katta-ka] sitteimasu-ka? you-TOP who-NOM why book-ACC bought-QM know-QM ‘Who is the person x such that you know why x bought a book?’ ‘W hat is the reason x such that you know who bought a book for x ?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 290 Consider the reading where the subject wh-word takes matrix scope and the propositional adjunct wh-word takes embedded scope in (74-76), first. The reading derives by the spec head agreement o f the propositional adjunct wh-word with the embedded QM at Spec o f the embedded CP and the unselective binding o f the subject wh-word by the matrix QM , as illustrated by the following LF in (77) with the irrelevant details suppressed. (77) [C P [ QMi [V P V [ c p why, [ c. QM k [ subject WH-wordi [ tk [ t, ]]]]]]]] Please recall that I argued that the propositional adjunct wh-word should undergo movement at LF to mark its scope via spec-head agreement with the QM, driven by the need for proper interpretation. The matrix scope reading o f the subject wh-word again does not show any W h-island effects, since no movement in involved. As shown above, the asymmetric scope taking strategies o f the two types o f wh-words enable wh-words to scope over each other without exhibiting the W h-island effects.1 3 Next consider the reading where the propositional adjunct wh-word takes matrix scope and the subject wh-word takes the embedded scope in (74-76). How will this be accounted for in our system? The only way the propositional adjunct wh-word in the embedded clause in (74- 76) can be interpreted is by undergoing movement, since unselective binding is not an option. The LF movement of the propositional adjunct wh-word should be first into the 1 3 A similar account to the present one was proposed in Shi (1994) for wh-words scoping out of Wh- islands in Chinese. The difference between his account and the present one has to do with the treatment of the propositional adjunct wh-word corresponding to why in English. Shi argues that all wh-words in Chinese are subject to the same binding mechanism when it comes to scoping out of Wh- islands, which, however, seems to be problematic, given that the propositional adjunct wh-word in Korean type languages is subject to other island constraints as discussed in Chapter 4. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 291 embedded Spec o f CP, assuming that movement is guided by the economy priniciple o f shortest movement (Chomsky 1995). The movement is indicated in the LF below in (78) with the irrelevant details omitted. (78) [cp whyj [ c QM j [subject NP [ t; [Vp V [ Cp tj [C ; QMk [ subject WH-wordk [ tk [ vp V ^ ]]]]]]]]]] The matrix QM is raised into the head o f the matrix CP, driven by the need to enter into spec-head agreement with the propositional adjunct wh-word in the matrix Spec o f CP, while the embedded QM is raised into the head o f the embedded CP to unselectively bind the subject wh-word, which is a nonpropsotinoal wh-word and thus is an indefinite. One may wonder why the propositional adjunct wh-word can move via the embedded Spec o f CP into the Spec o f matrix CP without checking off its wh-feature and thus ending up with having embedded scope. In fact, I suggest that the propositional adjunct wh-word cannot check off its wh-feature at the Spec o f the embedded CP in (78), based on the contrast in grammaticality in the following examples in Korean type languages as shown in (79-81): 1 4 1 4 The grammaticality of the sentences in (79a-81a) contrasts with the ungrammaticality of similar sentences in English where the in situ adjunct wh-word why somehow leads to ungrammaticality. *Who arrived why? (Reinhart 1998: 30) The ungrammaticality of the sentence as above has received a different account, based on different frameworks. For this, refer to Aoun, Homstein and Sportiche (1981), Huang (1982), and Reinhart (1998), among others. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 292 (79) Korean a. Nwu-ka way ttenass-ni? who-NOM why left-QM *‘Who came w hy?’ b.*Way nwu-ka ttenass-ni? why who-NOM left-QM *‘W hy did who com e?’ (80) Japanese a. Dare-ga naze soko-ni itta-no? who-NOM why there-to went-QM *‘Who went there w hy?’ b.*Naze dare-ga soko-ni itta-no? why who-NOM there-to went-QM *‘Why did who go there?’ (Saito 1994: 195) (81) Chinese a. Shei weishenme lai-de? who why come-ASP *‘Who came w hy?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 293 b. *Weishenme shei lai-de? why who come-ASP *‘W hy did who com e?’ The sentences in (79-81) exhibit an interesting contrast in grammaticality. To be specific, when the propositional wh-word precedes the nonpropositional wh-word, the sentences are ungrammatical as shown in (79b-81b), while with the opposite word order, the sentences are grammatical as shown in (79a-81a), an illustration o f the so called anti-superiority effect originally observed by Saito (1982, 1987, 1994) for Japanese (also see Nishigauchi 1990, A. Watanabe 1992, and S. W atanabe 1995 for Japanese and Chung 1995 for K orean).1 5 1 6 How can the contrast in grammaticality in (79-81) be accounted for? I argue that the contrast essentially follows, once one hypothesizes that the QM can do only one thing, i.e., either unselectively bind the nonpropositional wh-word or enter into spec-head agreement with the propositional adjunct wh-word, together with the proposal in the literature (Nishigauchi 1985, 1990, Saito 1994, Chung 1995) that in our terms the propositional adjunct wh-word can be licensed parasitically due to the existence o f the nonpropositional wh-word to its left. The argument for the parasitic licensing o f the propositional adjunct wh-word is based on the contrast o f grammaticalirty as shown in (82-84) in Korean type languages. 1 5 Chinese speakers whom I consulted note that a contrast in grammaticality exists between the two sentences in (81). 1 6 Note that I am assuming that the adjunct wh-word is base-generated in sentence initial position. For the argument for this position, please see Chapter 2. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 294 (82) Korean a. *[Ku-ka way ssun chayk-i] cemiiss-ni? he-NOM why wrote book-NOM interesting-QM ‘W hat is the reason x such that a book he wrote for x is interesting?’ b. (?)?[Nwu-ka w ay ssun chayk-i] ceymiiss-ni? who-NOM why wrote book-NOM interesting-QM ‘Who is the person x and the reason y such that a book x wrote for y is interesting?’ (Chung 1995:149) (83) Japanese a. *[Kare-ga naze kaita hon-ga] itiban omosiroidesu-ka? he-NOM why wrote book-NOM most interesting-QM ‘W hat is the reason x such that books he wrote for x are interesting?’ b. (?)?[Dare-ga naze kaita hon-ga] omosiroidesu-ka? who-NOM why wrote book-NOM interesting-QM ‘Who is the person x and the reason y such that books x wrote for y is interesting?’ (Nishigauchi 1985: 93, 96) (84) Chinese a. *ni kan [John weishenme xie de shu]? you read John why wrote book ‘W hat is the reason x such that you read books John wrote for y?’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 295 b. (?)?ni kan [shei weishenme xie de shu]? you read who why wrote book ‘W hat is the person x and the reason y such that you read books x wrote for y?’ Given the ungrammaticality o f the examples with the propositional adjunct wh-word inside the island as shown in (82a-84a) as discussed in Chapter 4, the improved grammaticality in the examples in (82b-84b) with the same wh-word inside the island thus indicates that the existence o f the nonpropositional wh-word to its left inside the island is responsible for this improvement o f grammaticality, thus lending support to the claim for parasitic licensing. 1 7 Our exclusive licensing mechanism of wh-words by the QM accounts for the ungrammaticality o f the sentences in (79b-8 lb) in a rather straightforward way. Consider the LF representation in (85) for these sentences with the relevant details suppressed, assuming that the wh-word in the ‘scrambled’ wh-question constructions is base-generated as it appears at the surface. (85) [C p why( [c QM, [ q [ subject W H-word [ t, ]]]]] Let us suppose the QM, which is raised into the head o f CP in (85) enters into a spec-head agreement with the propositional adjunct wh-word in Spec o f CP. This will lead to the 1 7 When the word order between the nonpropositional wh-word and the propositional adjunct wh- word in (82b-84b) is reversed, the grammaticality does not improve. See the original observation by Saito (1994: 205) for Japanese. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 296 situation where the subject wh-word, which is a nonpropositional wh-word and is an indefinite cannot be interpreted, since it is not bound by the external quantificational element, namely, the QM. Suppose now the QM unselectively binds the subject wh-word. The problem in this case is that the propositional adjunct wh-word in Spec CP cannot be licensed via spec-head agreement, due to the unselective binding o f the subject wh-word by the QM. The grammaticality o f (79a-81a) also follows, given the parasitic licensing o f the propositional adjunct wh-word. W ith the above discussion on the exclusive licensing mechanism o f the wh-words by the QM in mind, let us come back to the LF representation in (78) again, repeated as (86). (86) [C P whyj [c QM, [subject NP [ t, [V P V [ C P t, [c QM k [ subject W H-wordk [ tk [ t j ]]]]]]]]]] The propositional adjunct wh-word in the embedded Spec o f CP cannot be interpreted there since the QM already unselectively binds the subject wh-word. As a last resort, it should be raised into the matrix Spec o f CP where it can be interpreted by entering into a spec-head agreement relation with the QM in the head o f the matrix CP at LF, leading to the matrix scope reading o f the propositional adjunct wh-word. Let us see whether the movement o f the propositional adjunct wh-word into Spec o f the matrix CP violates W h-island constraint, as subsumed under subjacency (Chomsky 1986). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 297 The movement o f the propositional adjunct wh-word via the embedded Spec o f CP into the matrix Spec o f CP does not violate W h-island constraint, since it does not cross a barrier during its derivation. 1 8 4. Conclusion As shown thus far, the propositional adjunct wh-word and nonpropositional wh-words in Korean type languages show a surprising symmetry in scoping out o f W h-islands, which is quite unexpected, at least on the surface, given the asymmetry o f locality between them in constructions such as Complex Noun Phrase Islands, Sentential Subject Islands, and Adjunct Islands as observed and discussed in Chapter 4. It was shown, however, that the asymmetric scope taking strategies o f the two types o f wh- words both enable wh-words o f the respective type to take scope out o f W h-island, without violating the constraint: The matrix scope of the propositional adjunct wh-word out o f a Wh- island does not violate the constraint, since it does not cross a barrier, and the matrix scope of the nonpropositional wh-word out o f a W h-island does not exhibit the effects either, since no movement is involved. To the extent that the current observation for the lack o f W h-island 1 8 Alternatively one may adopt Wh-island constraint subsumed under the minimal link condition (Chomsky 1995). The movement of the propositional adjunct wh-word into the matrix Spec of CP is via the embedded Spec of CP and thus does not violate minimal link condition. However one wishs to understand Wh-island constraint, i.e., either in terms of subjacency in Chomsky (1986) or minimal link condition in Chomsky (1995), it does notr affect the present discussion of the phenomena of Wh- island effects in Korean type languages. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 298 effects in Korean type languages is correct, H uang’s (1982) observation for the existence of W h-island effects in Chinese and his account o f the phenomena in terms o f ECP, which is crucially based on argument and adjunct dichotomy under the uniform LF wh-movement hypothesis is seriously undermined. In Huang’s system, W h-islands are invariably created by one wh-word being raised into the embedded comp at LF (Spec o f CP in current terminology), such that the ability to scope out o f the W h-island is predicted to vary for the other wh-word, depending on which type o f wh-word it is, an argument wh-word or an adjunct wh-word. The present observation for the lack o f W h-islands in Korean type languages provides additional argument against the proposal in Nishigauchi (1990) for LF pied piping. Nishigauchi’s proposal for large scale LF pied piping in islands such as Complex Noun Phrase Island as discussed in Chapter 4 is crucially based on his observation that wh-words in Japanese cannot take scope out o f Wh-islands. However, to the extent that one cannot find any positive evidence that wh-words cannot take scope out o f Wh-islands in Korean type languages, his claim for the large scale pied-piping does not carry any theoretical weight. If the current observations are correct that I make with respect to the W h-island construction in Korean type languages in contrast with other island environments, as discussed in the previous chapter, the present proposal that there exists unique scope taking strategy for each type o f wh-word along the lines suggested in this thesis seems to be defendable: A nonpropositional wh-word is unselectively bound by the question morpheme that serves Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 299 as a wh-operator and marks its scope, while the propositional adjunct wh-word, not being an indefinite, undergoes movement at LF into Spec o f CP to mark its scope via spec- head agreement with the QM , driven by the need for proper interpretation by forming an operator variable chain. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 300 References Aoun, Joseph, Nobert Hom stein, and Dominique Sportiche. 1981. On some aspects o f wide scope quantification. Journal o f Linguistic Research 1:67-95. Aoun, Joseph, and Dominique Sportiche. 1981. On The Formal Theory o f Government, paper read at the 1981 GLOW conference at Gotingen. Aoun, Joseph, and Audrey Li. 1989. Scope and Constituency. Linguistic Inquiry 20:141-172. Aoun, Joseph, and Audrey Li. 1993a. The Syntax o f Scope. Cambridge, Massachusetts, MIT Press. Aoun Joseph, and Audrey Li. 1993b. W h-Elements in Situ: Syntax or LF? Linguistic Inquiry 24:199-238. Aoun Joseph, and Audrey Li. 1993c. On Some Differences between Chinese and Japanese Wh-elements. Linguistic Inquiry 24: 365-372. Baker, Lee. 1970. Note on the description o f English questions: The role o f an abstract question morpheme. Foundations o f Language 6:197-219. Barwise, Jon, and Robin Cooper. 1979. On Branching Quantifiers in English. Journal o f philosophical Logic 8: 47-80. Barwise, Jon, and Robin Cooper. 1981. Generalized quantifiers and natural language. Linguistics and Philosophy 4:159-219. Beck, Sigrid, and Shin-Sook Kim. 1997. On wh-and operator scope in Korean. Journal o f East Asian Linguistics 6:339-384. Beghelli, Filippo. 1997. The Syntax o f Distributivity and Pair-List Readings. In Ways o f Scope Taking, ed. Anna Szabolcsi, Dordrecht, Kluwer Academic Publishers. Benmamoun, Elabbas. 1992. Functional and Inflectional Morphology: Problems o f Projection, Representation and Derivation. Doctoral dissertation, University o f Southern California. Bromberger, Sylvain. 1992. W hy-Questions. In On What We Know We D on’ t Know: Explanation, Theory, Linguistics, and How Questions Shape Them. The University of Chicago Press. Charles, Li, and Sandra Thompson. 1991. Mandarin Chinese: A Functional Reference Grammar, University o f California Press, Berkeley. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 301 Cheng, Lisa. 1997. On the typology o f wh-questions. New York, Garland Publishing Inc. Chierchia, Gennaro. 1991. Functional W H and W eak Crossover. In Proceedings o f the Tenth West Coast Conference on Formal Linguistics, Stanford, CSLI. Chierchia, Gennaro. 1993. Questions with Quantifiers. Natural Language Semantics 1: 181- 234. Cho, Eun. 1998. Why, Contrastive Topic and LF movement. In Proceedings o f the Japanese Korean Conference 8, Stanford, CSLI. Cho, Jae-Hyung. 1994. On Scrambling: Reconstruction, Crossover and Anaphor Binding. In Theoretical Issues in Korean Linguistics, Stanford, CSLI. Choe, Jae-Woong. 1987. LF-movement and Pied-Piping. Linguistic Inquiry 18: 348-353. Choe, Hyun-Sook. 1994. Syntactic wh-movement and Licensing. In Theoretical Issues in Korean Linguistics, Stanford, CSLI. Choi, Hyun-Pae. 1935. Wuli Malpon (Our Grammar). Seoul, Cengummwunhwasa. Chomsky, Noam. 1956. Syntactic Structures. Mouton, Hague. Chomsky, Noam. 1962. The Logical Basis o f Linguistic Theory. In Proceedings o f the Ninth International Congress o f Linguists, The Hague, Mouton. Chomsky, Noam. 1964. Current Issues in Linguistic Theory. The Hague, Mouton. Chomsky, Noam. 1976. Conditions on Rules o f Grammar. Linguistic Analysis 2:303-351. Chomsky, Noam. 1977. On wh-movement. In Formal Syntax, eds. Peter W. Culicover, Thomas W asow, and Adrian Akmajian, New York, Academic Press. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht, Foris. Chomsky, Noam. 1986. Barriers. Cambridge, M assachusetts, MIT Press Chomsky, Noam. 1993. A M inimalist Program. In The View from Building 20: Essays in Linguistics in Honor o f Sylvain Bromberger, eds. Ken Hale and Samuel Keyser, Cambridge, Massachusetts, MIT Press. Chomsky, Noam. 1994. Bare Phrase Structure, MIT Occasional Papers in Linguistics, No.5. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, M assachusetts, MIT Press. Chung, Daeho. 1995. On the representation and licensing o f Q and Q-dependents. Doctoral dissertation, University o f Southern California. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 302 Dayal, Veneeta. 1996. Locality in WH Quantification, Dordrecht, Boston and London, Kluwer Academic Publishers. Engdahl, Elisabet. 1986. Constituent Questions. Dordrecht, Boston, Lancaster and Tokyo, D. Reidel Publishing Company. Farkas, Donka. 1981. Quantifier scope and Syntactic Islands. In Proceedings o f the Seventeenth Chicago Linguistics Society. Fukui, Naomi, and M argaret Speas. 1986. Specifiers and Projection. In MIT Working Papers in Linguistics 8, MIT. Geach, Peter. 1962. Reference and generality. Ithaca, New York, Cornell University Press. Hagstrom, Paul. 1998. Decomposing Questions. Doctoral dissertation, MIT. Hamblin. C.L. 1973. Questions in M ontague English. Foundations o f Language 10: 41-53. Groenendijik, Jeroen, and M artin Stokhof. 1984. Studies on the Semantics o f Questions and the Pragmatics o f Answers. Doctoral dissertation, University of Amsterdam. Heim, Irene. 1982. The Semantics o f definite and Indefinite Noun Phrases. Doctoral dissertation, University o f Massachusetts, Amherst. Heim, Irene. 1987. W here does the Definiteness Restriction Apply? Evidence from the Definiteness o f Variables. In Representation o f (In)defmiteness, eds. Eric Reuland and Alice G.B. ter Meulen, Cambridge, Massachusetts, MIT Press. Heim, Irene, and Angelika Kratzer. 1998. Semantics in Generative Grammar. Malden M assachusetts, Blackwell Publishers. Higginbotham, James, and Robert May. 1981. Questions, quantifiers and crossing. Linguistic Review 1: 41-80. Higginbotham, James. 1991. Interrogatives I. In M ore Papers on W H-M ovement, eds. Lisa Cheng and Hamida Demirdash, MIT. Hoji, Hajime. 1985. Logical Form Constraints and Configurational Structure in Japanese. Doctoral dissertation, University o f Washington. Hong, Sung-Shim. 1985. A and A ’ Binding in Korean and English: Government and Binding Parameters. Doctoral dissertation, University o f Connecticut. Homstein, Nobert. 1995. Logical Form, from GB to Minimalism. M alden, M assachusetts, Blackwell Publishers. Huang, C.T. James. 1982. Logical Relations in Chinese and the theory o f Grammar. Doctoral dissertation, MIT. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 303 Joo, Shim Yanghee. 1989. A Cross-Linguistic Approach to Quantification in Syntax. Doctoral Dissertation, University o f W isconsin-M adison. Kang, M yung-Yoon. 1988. Topics in Korean Syntax: Phrase Structure, Variable and Movement. Doctoral dissertation, MIT. Katz, Jerrold, and Paul Postal. 1964. An Integrated Theory o f Linguistic Description. Cambridge, MIT Press. Karttunen, Lauri. 1977. Syntax and Semantics o f Questions. Linguistics and Philosophy 1: 3- 44. Kim, Soo-Won. 1991. Chain Scope and Quantification Structure. Doctoral dissertation, Brandeis University. Klima, Edward. 1964. Negation in English. In The Structure o f Language, eds. Janet Fodor and Jerrold Katz, N ew York, Prentice-Hall. Koopman, Hilda, and Dominique Sportiche. 1983. Variables and the Bijection Principle. Linguistic Review 2:139-160. Koopman, Hilda, and Dominique Sportiche. 1985. Theta Theory and Extraction. GLOW Newsletter 14, 57-58. Koopman, Hilda, and Dominique Sportiche. 1988. Subjects, ms., University of California at Los Angeles. Kuno, Susumu, and J. Robinson. 1972. Multiple W h-Questions. Linguistic Inquiry 3: 463- 487. Kuroda, Shige-Yuki. 1965. Generative Grammatical Studies in Japanese Language. Doctoral dissertation, MIT. Kuroda, Shige-Yuki. 1969. English Relativization and Certain Related Problems. In Modern Studies in English, eds. David Reibel and Sanford Schane, Englewood Cliffs, Prentice-Hall. Kuroda, Shige-Yuki. 1988. W hether W e Agree or Not: A Comparative Syntax o f English and Japanese. Lingvisticae Investigationes 12: 1-47. Ladusaw, William. 1980. Polarity Sensitivity as Inherent Scope Relation. New York, Garland Publishing INC. Larson, Richard, and Gabriel Segal. 1995. Knowledge o f Meaning. Cambridge, M assachusetts, MIT Press. Lasnik, Howard, and M amoru Saito. 1992. Move-a. Conditions on its Applications and Output. Cambridge, M assachusetts, MIT Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 304 Lasnik, Howard. 1999. Chains o f Arguments. In Working Minimalism, eds. Samuel David Epstein and Norbert Hom stein, Cambridge, M assachusetts, MIT Press. Lasnik Howard, and M amoru Saito. 1984. On the nature o f proper government. Linguistic Inquiry 15: 235-289. Lee, Hyosang. 1982. Asymmetry in Island Constraints in Korean, ms., University o f California at Los Angeles. Lee, Kisuk and Satoshi Tomoika. 2002. LF Blocking Effects are Topic Effects: Wh- Questions in Japanese and Korean. To appear in Japanese and Korean Linguistics 10, Stanford: CSLI. Lewis, David. 1975. Adverbs o f Quantification. In Formal Semantics o f Natural language, ed., Edward Keenan, Cambridge University Press. Li, Audrey. 1992. Indefinite W h in M andarin Chinese. Journal o f East Asian Linguistics 1:125-155. Lin, Jo-W ang. 1992. The Syntax o f Zenmeyang ‘how ’ and W eishenme ‘w hy’ in Chinese. Journal o f East Asian Linguistics 1: 293-331. Lin, Jo-Wang. 1996. Polarity Licensing and Wh-phrase Quantification in Chinese. Doctoral dissertation, University o f Massachusetts, Amherst. Liu, Feng-hsi. 1997. Scope and Specificity. Amsterdam and Philadelphia, John Benjamins Publishing Company. May, Robert. 1985. Logical Form. Cambridge, Massachusetts, MIT Press. Milsark, Gary. 1975. Existential Sentences in English. Doctoral dissertation, MIT. Nishigauchi, Taisuke. 1985. Japanese LF: Subjacency vs. ECP. In Seoul Papers in Formal Grammar Theory, ed. Ik-Hwan Lee, Seoul, Hanshin. Nishigauchi, Taisuke. 1990. Quantification in the Theory o f Grammar. Dordrecht, Kluwer Academic Publishers. O ’grady, William. 1991. Categories and Case: the Sentence Structure o f Korean. Amsterdam and Philadelphia, John Benjamins Publishing Company. Pesetsky, David. 1982. Paths and Categories. Doctoral dissertation, MIT. Pesetsky, David. 1987. W h in situ: Movement and Unselective Binding. In Representation o f (In)definiteness, eds. Eric Reuland and Alice G.B. ter Meulen, Cambridge, M assachusetts, MIT Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 305 Pesetsky, David. 2000. Phrasal movement and its kin. Cambridge, M assachusetts, MIT Press. Pollock, Jean-Yves. 1989. Verb Movement, Universal Grammar, and the Structure o f IP. Linguistic Inquiry 20:365-424. Reinhart,Tanya. 1976. The Syntactic Domain o f Anaphora. Doctoral dissertation, MIT. Reinhart, Tanya. 1997. Quantifier Scope: How labor is divided between QR and Choice functions. Linguistics and Philosophy 20: 335-397. Reinhart, Tanya. 1998. W h in situ in the M inimalist Program. Natural Language Semantics 6: 29-56. Van Riemsdijk, and Edwin Williams. 1981. NP-structure. The Linguistic Review 1:171-217. Richards, Norvin. 2000. An Island Effect in Japanese. Journal o f East Asian Linguistics 9:187-205. Rizzi, Luigi. 2000. The Fine Structure o f the Left Periphery. In Comparative Syntax and Language Acquisition, London and New York, Routledge. Ross, John. 1967. Constraints on variables in Syntax. Doctoral dissertation, MIT. Saito, Mamoru. 1982. Scrambling, Topicalization and Strong Crossover, ms., MIT. Saito, Mamoru. 1989. An Extension o f K J.H arata’s W H-Q Binding Analysis. Presented at the Fifth Annual M eeting o f the English Linguistics Society o f Japan. Saito, Mamoru. 1994. Additional W h effects and the Adjunction. Journal o f East Asian Linguistics 3: 195-240. Saito, Mamoru, and Hajime Hoji. 1983. W eak crossover and move a in Japanese. Natural Language and Linguistic Theory 1: 245-260. Shi, Dingxu. 1994. The Nature o f Chinese W h-Questions. Natural language and Linguistic Theory 12: 301-333. Shimoyama, Junko. 2001. Wh-constructions in Japanese. Doctoral dissertation, University of M assachusetts, Amherst. von Stechow, A. 1996. Against LF pied-piping. Natural Language Semantics 4:57-110. Sloan, Kelly. 1991. Quantifier-wh Interaction. In More Papers on WH-Movement, eds. Lisa Cheng and Hamida Demirdache, MIT W orking Papers in Linguistics, MIT. Suh, Cheongsoo. 1994. Korean Grammar. Seoul, Deep-rooted Tree Publishing. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 306 Suh Jin-Hee. 1990. Scope Phenomena and Aspects o f Korean Syntax. Doctoral dissertation, University o f Southern California. Szabolsci, Anna. 1997a. Background Notions in Lattice Theory and Generalized Quantifiers. In Ways o f Scope Taking, ed. Anna Szabolcsi, Dordrecht, Kluwer Academic Publishers. Szabolsci, Anna. 1997b. Quantifiers in Pair-List readings. In Ways o f Scope Taking, ed. Anna Szabolcsi, Dordrecht, Kluwer Academic Publishers. Tang, Ting-Chi. 1988a. Hanyu Cifa Jiufa Runji [Essays on Chinese Morphology and Syntax], Student Books CO, Taipei, Taiwan. Tsai W ei-tien Dylan. 1994. On economizing the theory o f A-bar dependencies. Doctoral dissertation, MIT. Ueyama, Ayumi. 1998. Two Types o f Dependency. Doctoral dissertation, University of Southern California. W atanabe, Akira. 1992. W h-in situ, Subjacency and Chain Formation. MIT Occasional Papers in Linguistics 2. Watanabe, Shin. 1995. Aspects o f Questions in Japanese and their Theoretical Implications, Doctoral dissertation, University o f Southern California. Williams. Edwin. 1986. A Reassignment of the functions o f LF. Linguistic Inquiry 17: 265- 300. Williams. Edwin. 1988. Is LF Distinct from S-structure? A Reply to May, Linguistic Inquiry 19:135-146. Vergnaud, Jean-Roger, and M aria Luisa Zubizarreta. 2002. Some Preliminaries to M inimalist Questions, ms., University o f Southern California. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Beyond words and phrases: A unified theory of predicate composition

PDF

Grammaticalization and the development of functional categories in Chinese

PDF

An anaphoric approach to clitic position in Spanish

PDF

Form and meaning: Negation and question in Chinese

PDF

A variationist study of relative clauses in Spanish

PDF

Acquisition of Spanish verb morphology by bilingual children. A longitudinal study between the ages of 2;9 and 3;3

PDF

Connecting phrasal and rhythmic events: evidence from second language speech

PDF

Children's scope of indefinite objects

PDF

Dynamic voltage and frequency scaling for energy-efficient system design

PDF

Ellipsis constructions in Chinese

PDF

Discourse functional units: A re-examination of discourse markers with particular reference to Spanish

PDF

Cognitive efficiency of animated pedagogical agents for learning English as a second language

PDF

"Master of many tongues": The Russian Academy Dictionary (1789--1794) as a socio -historical document

PDF

Issues in the syntax of resumption: Restrictive relatives in Lebanese Arabic

PDF

Gossip, letters, phones: The scandal of female networks in film and literature

PDF

Children in transition: Popular children's magazines in late imperial and early Soviet Russia

PDF

Correspondence and faithfulness constraints in optimality theory: A study of Korean consonantal phonology

PDF

"The unaccusative trap": L2 acquisition of English intransitive verbs.

PDF

Energy and time efficient designs for digital signal processing kernels on FPGAs

PDF

Coactivator synergy for nuclear receptors: Regulatory roles of arginine methyltransferase activity in transcription

Asset Metadata

Creator Choi, Young-Sik (author)

Core Title Asymmetry of scope taking in wh -questions

School Graduate School

Degree Doctor of Philosophy

Degree Program Linguistics

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag language, linguistics,Language, Modern,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Vergnaud, Jean-Roger (committee chair), Saltarelli, Mario (committee member), Zubizarreta, Maria Luisa (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-275811

Unique identifier UC11339861

Identifier 3094316.pdf (filename),usctheses-c16-275811 (legacy record id)

Legacy Identifier 3094316.pdf

Dmrecord 275811

Document Type Dissertation

Rights Choi, Young-Sik

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA