To sluice or not to sluice? On sluicing in wh-in-situ...
Transcription
To sluice or not to sluice? On sluicing in wh-in-situ...
1 To sluice or not to sluice? On sluicing in wh-in-situ languages Andrew Murphy Humboldt-Universität zu Berlin amurphy87@gmail.com 1 Overview 2 I will argue against a common claim that sluicing in wh-in-situ languages are not derived from cleft-like structures. Instead, it will be claimed that we need to assume clausal ellipsis (as in English). I will show how we can explain sluicing in wh-in-situ languages by appealing to an analysis using focus movement followed by deletion. What is sluicing? Sluicing is arguably one of the most widely-discussed elliptical constructions. Typical sluicing examples are given in (1): (1) a. Joe says he is investigating someone, but he won’t tell me who. b. She’s complaining, but we don’t know about what. c. My library card has just been cancelled; it’s unclear why. (Chung 2013) The standard analysis of sluicing (Ross 1969, Merchant 2001) assumes full clausal ellipsis: (2) a. Joe says he is investigating someone, but he won’t tell me who he is investigating. b. She’s complaining, but we don’t know about what she is complaining. c. My library card has just been cancelled; it’s unclear why my library card has just been cancelled. Case-matching (Ross 1969): (3) Hans will jemandem schmeicheln, aber Hans wants.to somebody.DAT flatter but wem/*wen who.DAT/who.ACC ‘Hans is trying to flatter someone, but we don’t know who.’ wir wissen we know nicht not We observe that the wh-phrase in the second clause of (3) also bears dative case. Interestingly, the verb in the second clause wissen (‘know’) can assign only accusative case to its direct object (4). (4) Wir wissen *der/die Antwort we know the.DAT/the.ACC answer ‘We don’t know the answer.’ nicht. not 2 We can account for this by assuming that dative case is assigned to the wh-phrase from the verb schmeicheln (‘to flatter’) in an elided clause: (5) Hans will jemandem schmeicheln, aber Hans wants.to somebody.DAT flatter but [CP wem [TP Hans <wem> schmeicheln will]]. who.DAT wir wissen we know nicht not The standard analysis of sluicing (following Merchant 2001) assumes two steps: wh-movement deletion of TP The C head bears an ‘E-feature’, which licenses ellipsis of its complement at PF: (6) John wants to see someone, but I don’t know who. know CP whoi C’ C[E] <TP> John wants to see ti Sluicing in wh-in-situ languages: Problem: what about languages which lack wh-movement. We would predict that these cannot derive sluicing. Mandarin Chinese lacks wh-movement to form questions (it is wh-in-situ) (8). Nevertheless, they still evidence sluicing structures (7): (7) Zhangsan xihuan shei? Zhangsan like who ‘Who does Zhangsan like?’ (8) Zhangsan xihuan mouren, keshi wo bu zhidao Zhangsan like someone but I NEG know ‘Zhangsan likes someone, but I don’t know who.’ shi shei. SHI who Therefore, it seems we cannot derive sluicing in wh-in-situ languages in the same way as English. 3 3 Deriving sluicing from clefts As a result, it has been widely claimed that sluicing in wh-in-situ languages is derived from a cleft-like structure. This analysis comes in two main forms: the reduced cleft analysis and the pseudosluicing analysis. 3.1 Reduced cleft analyses First option is a so-called ‘reduced cleft’. We take an ordinary cleft structure and elide the relative clause: (9) [wh-phrase [RelC <...> ]] Japanese (Takahashi 1994): (10) Mary-ga nanika -o katta rasii ga, boku-wa Mary-NOM something-ACC bought likely but I-TOP ka wakara-nai. Q know-not ‘It is likely Mary bought something, but I don’t know what.’ (11) (12) 3.1.1 [Taroo-ga tataita no]-wa Taroo-NOM hit REL-TOP ‘It was Hanako that Taroo hit.’ Hanako( -o) Hanako-ACC …boku-wa [CP [CP I-TOP wakara-nai. know-not katta no ]-ga bought REL -NOM kanozyo-ga she-NOM nani-o what-ACC (da) COP da. COP nani -o what-ACC (da) COP ka] Q Arguments against reduced clefts However, a number of scholars have shown that wh-phrases in sluicing have a number of properties, which differ from the wh-phrases found in clefts. 3.1.1.1 Case-matching effects Turkish (Ince 2009): (13) a. [Hasan-ın borç al-dı-ğ-ı ] Hasan-GEN debt take-PAST-COMP-POSS.3SG ‘Who is it that Hasan borrowed money from?’ b. [Hasan-ın borç al-dı-ğ-ı ] Hasan-GEN debt take-PAST-COMP-POSS.3SG ‘It is Ibrahim that Hasan has borrowed money from.’ KIM-Ø? who-NOM IBRAHIM-Ø Ibrahim-NOM 4 (14) Hasan biri-nden borç al-dı, ama Hasan-NOM someone-ABL debt take-PAST but bil-mi-yor-um. know-NEG-PRES-1SG ‘Hasan borrowed money from someone, but I don’t know who.’ kim(-den/*-Ø) who(-ABL/*NOM) Mongolian (Sakamoto 2011): (15) Oyuna-gin id -sen ni tort -Ø/ *-ig. Oyuna-GEN eat-PERF NI cake -NOM/-ACC ‘It is the cake that Oyuna ate.’ (16) Oyuna-Ø yamar_negen_zuil-ig id –sen gevch Oyuna-NOM something-ACC eat-PERF but med-eh-gui. know-INF-NEG ‘Oyuna ate something, but I don’t know what.’ 3.1.1.2 Adjuncts bi I yu(-g/*- Ø) ni what(-ACC/*NOM) NI Turkish: (17) *Ali-nin git-tiğ-i dün / ne zaman? Ali-GEN go-COMP-POSS.3SG yesterday what time ‘It’s yesterday that Ali went/When is that Ali went?’ (18) Ali-Ø Ankara-ya git-ti-Ø, ama Ali-NOM Ankara-DAT go-PAST-3SG but ‘Ali went to Ankara, but I don’t know when’ ne zaman bil-m-iyor-um. what time know-NEG-PRES-1SG Mongolian: (19) *Oyuna-Ø ene nom-ig hudal_dag_av_san Oyuna-NOM this book -ACC bought ‘It was yesterday that Oyuna bought this book.’ (20) Bat-Ø Mie yav-san, gevch bi hezee -g Bat-NOM Mie go-PERF but I when-ACC ‘Bat went to Mie, but I don’t know when.’ 3.1.1.3 Multiple sluicing ni NI uchigdur. yesterday ni] med-eh-gui. NI know-INF-NEG Turkish: (21) *[Ahmet-in ti tj al-dığı] Hasan-dani kitapj. Ahmet-GEN take-REL Hasan-ABL book-NOM ‘It’s a book from Hasan that Ahmet borrowed’ 5 (22) Ahmet-Ø biri-nden birşey al-mış-Ø; ama kim-den ne Ahmet-NOM one-ABL thing take-HRS-3SG but who-ABL what bil-mi-yor-um. know-NEG-PRES-1SG ‘Ahmet borrowed something from someone, but I don’t know what (or) from whom.‘ 3.2 Pseudosluicing Kizu (2000) in particular claimed that all wh-in-situ languages exhibit ‘pseudosluicing’ structures, which look like sluicing in English but have a distinctly different structure. (23) a. A celebrity is coming to dinner, but John won’t tell me who is coming to dinner. b. A celebrity is coming to dinner, but John won’t tell me who it is. (24) Pseudosluicing structure: [ …indefinitei …] , but I don’t know [proi copula wh-phrase] 3.2.1 Arguments for pseudosluicing 3.2.1.1 Copula Pseudosluicing can provide a natural explanation for the fact that the copula often occurs in sluicing: Mandarin Chinese: (25) Lisi mai le yi-jian liwu, danshi ta bu gaosu Lisi buy LE one-CL present but he NEG tell ‘Lisi bought a present, but he didn’t tell me what.’ Korean: (26) Mimi-nun nwukwunka-lul manass-nuntey, Mimi-TOP someone-ACC met-but ‘Mimi met someone, but I don’t know who.’ wo *(shi) me COP shenme. what nwukwu-i-nci molukessta. who-COP-Q not.know 3.2.1.2 Island repair (27) a. John married [DP a woman, [CP who speaks Bulgarian]]. b. Which language did John marry [DP a woman, [CP who speaks ____ ]]? (28) They want to hire someone who speaks a Balkan language, but I’m not sure [CP which Balkan language [TP they want to hire [DP someone [CP who speaks _____ ]]]]. (29) [DP [CP Mei-ge cong mou-gei guojia lai de] xuesheng] dou tong every-CL from some-CL country come de student all pass guo kaoshi, wo xiang zhidao proi shi na yi-ge guojia. ASP exam I want know SHI which one-CL country ‘Every student from some country passed the exam, and I wonder which country (it is).’ 6 3.2.1.3 Sprouting Chung et al. (1995) discuss cases in which there is no overt antecedent for sluicing. This can be explained by the pseudosluicing analysis if we say that pro needs an overt antecedent. (30) a. John is eating, but I don’t know what. b. *Zhangsan zhengzai chi, dan wo bu zhidao Zhangsan PROG eat but I NEG know Int. ‘Zhangsan is eating but I don’t know what.’ 3.2.2 Arguments against pseudosluicing in Mandarin Chinese 3.2.2.1 Distrubtion of shi pro? shi shenme. SHI what In sluicing in Mandarin Chinese, the copula shi is obligatory with simplex wh-arguments (who, what) and optional with complex wh-arguments (which person) and adjuncts. (31) Zhangsan kandao mouren, danshi wo bu zhidao Zhangsan saw someone but I NEG know (shi) shenme ren. SHI what person ‘Zhangsan saw somebody, but I don’t know who/what person.’ *(shi) SHI shei / who (Wei 2004: 7) Why does the distribution depend on the wh-phrase? Adams (2004) and Wei (2004) claim that simplex wh-phrases are not predicates (<e>) (and therefore require shi as a linking verb. Complex wh-phrases are predicates (<e, t>) and therefore only optionally occur with shi (for emphasis?). Pseudoclefts: (32) a. Zhangsan zai yi ge canguan chifan, danshi wo bu zhidao Zhangsan at one CL restaurant eat but I NEG know [DP pro] (shi) shenme canguan. SHI which restaurant ‘Zhangsan is eating at a restaurant, but I’m not sure which (restaurant).’ b. [DP [CP Zhangsan zui xihuan chifan de] defang] *(shi) shenme canguan? Zhangsan most like eat DE place SHI which restaurant ‘Which restaurant is Zhangsan’s favourite place to eat?’ The na-construction: (33) a. Zhangsan Zhangsan pro (shi) xihuan yi ge ren, danshi like one CL person but shenme ren SHI which person ‘Zhangsan likes someone, but I don’t know who.’ wo I bu not zhidao know 7 b. Zhangsan Zhangsan na *(shi) xihuan yi ge ren, danshi like one CL person but shenme ren DEM SHI which person ‘Zhangsan likes someone, but I don’t know who.’ 3.2.2.2 wo I bu not zhidao know Sloppy readings Sloppy readings with VP ellipsis: (34) John knows how to make his mother smile and Peter does too. a. Strict reading: ‘John knows how to make his (own) mother smile and Peter also knows how to make John’s mother smile.’ b. Sloppy reading: ‘John knows how to make his (own) mother smile and Peter also knows how to his own mother smile.’ (35) a. Johni loves hisi cat and Billj does [love hisi cat] too. b. Johni loves hisi cat and Billj does [love hisj cat] too. (strict) (sloppy) (Hardt 2003) Sloppy readings in sluicing in MC: (36) Laowu zhidao ta weisheme yao jiehun, Lisi ye zhidao weisheme. Laowu know he why get marry Lisi also know why ‘Laowui knows why he is getting married, and Lisij also knows why hei/j is getting married.’ (Wang & Wu 2006) Identity Condition on sloppy readings: Pseudosluicing cannot explain why sloppy readings are only available if both wh-phrases are identical (Wei 2009): (37) a. Zhangsani zhidao [shei zai piping tai], dan Lisij bu zhidao shi shei. Zhangsan know who at criticize him but Lisi not know SHI who ‘Zhangsani knows who is criticizing himi, but Lisij doesn’t know who (is criticizing himi/j).’ (strict/sloppy) b. Zhangsani zhidao [shei zai piping tai], dan Lisij bu zhidao (shi) weishenme. Zhangsan know who at criticize him but Lisi NEG know SHI why ‘Zhangsani knows who is criticizing him, but Lisij doesn’t know why (that person is criticising himi/*j).’ (strict/*sloppy) (38) Lexical Identity Condition on Sloppy Identity in MC A sloppy reading is only available if every item in the antecedent TP (TPA) is lexically identical to an item in the elided TP (TPE). 8 (39) Zhangsani bu zhidao [CP [TP tai weishenme bei ma]], dan Lisij Zhangsan not know he why PASS scold but Lisi (shi) weishenme [TP tai weishenme bei ma]. SHI why he why PASS scold ‘Zhangsani didn’t know why hei was scolded, but Lisij knows why (hei/j was scolded).’ (strict/sloppy) zhidao know TPA = [ta weishenme bei ma] TPE = [ta weishenme bei ma] For (37b), the wh-phrases in the TPs do not match (shei vs. weishenme) and therefore no sloppy reading is licensed. The ‘na-effect’ (Wei 2009): (40) Zhangsani bu zhidao [tai weishenme bei ma], dan Lisij Zhangsan not know he why PASS scold but Lisi na (shi) weishenme. that SHI why ‘Zhangsani didn’t know why hei was scolded, but Lisij knows why (it is) hei/j was scolded.’ (strict/*sloppy) 3.2.2.3 zhidao know Binding with picture-NPs In English, reflexives such as himself can only be co-referent with an antecedent in the same clause: (41) Johni said [that Peterj likes himself*i/j]. Only himself is possible in sluicing, despite the antecedent being in a different clause: (42) a. Harryi likes that black-and-white photo of Sally you took, but I don’t know which photo of himselfi/*himi. b. I heard Johni drew a picture of himself and I’d be interested to know what kind of picture (of himselfi/*himi). (43) I heard Johni drew a picture of himself and I’d be interested to know [CP what kind of picture of himselfi [TP Johni drew what kind of picture of himselfi]]. (44) Zhangsanj zhidao [ Lisik xihuan ta-zijik/*j] Zhangsan know Lisi like he-self ‘Zhangsanj knows that Lisik likes himselfk/*j’ (45) Zhangsani renchu le yi zhang ta-zijii de zhaopian, keshi wo bu Zhangsan recognize LE one CL he-self DE picture but I not jide [TP pro shi (ta-zijii de) na-zhang zhaopian.] remember SHI he-self DE which-CL picture ‘Zhangsan recognized a picture of himself, but I can’t remember which picture (of himself).’ 9 4 A Focus Movement Analysis We have seen evidence for an elided clause even in wh-in-situ languages. We cannot appeal to wh-movement in these languages (although see Takahashi 1994). For a number of languages, it has been proposed that there is focus movement followed by deletion of TP: (46) FocP wh-phrase Foc’ Foc0[E] <TP> … <wh-phrase> (47) A: Wen hat Peter gestern Abend getroffen? B: Peter hat [MARIA]F gestern Abend getroffen. Focus markers and copulas: Heine & Reh (1984) have shown that there is a clear grammaticalization path from copula to FM: French (48) & Haitian Creole (49): (48) C’est Jean [RelC qui a tué it.is John who has killed ‘It was John who killed the deer.’ le cerf] the deer (49) Se [sou pis. flea chen mèg] yo wè FOC LOC dog thin 3PL see ‘It’s on the thin dog that fleas can be seen.’ (Heine & Kuteva 2002) Japanese sluicing as a focus construction (Hiraiwa & Ishihara 2002): ‘da’ as head of a focus phrase: (50) a. [FocP [CP TAROF-ga kono-ringo-o Taro-NOM this-apple-ACC ‘It is Taro who ate this apple.’ b. [FocP [CP Taro-ga KONO-RINGOF-o Taro-NOM this-apple-ACC ‘It is THIS APPLE that Taro ate.’ tabeta ate no] da] COMP FOC tabeta ate no] da] COMP FOC 10 (51) Mary-ga nanika -o katta rasii ga, boku-wa [CP [FocP nani-o Mary-NOM something-ACC bought likely but I-TOP what-ACC [TP Mary-ga nani-o katta] da] ka] wakara-nai. FOC Q know-not ‘It is likely Mary bought something, but I don’t know what.’ (52) CP C0 ka FocP wh-phrase Foc’ Foc0 da[E] <TP> … <wh-phrase>… Mandarin Chinese (Wang 2012, Murphy 2014): Recap: the distribution of shi = obligatory with simplex wh-arguments (who, what), optional with all others. (53) Zhangsan xihuan mouren, keshi wo bu zhidao *(shi) Zhangsan like someone but I NEG know SHI ‘Zhangsan likes someone, but I don’t know who.’ shei/ (shi) shenme ren. who/ SHI which person Following Wang (2012), I assume that shi is the head of a focus phrase: (54) FocP wh-phrase Foc’ Foc0 shi <TP> … <wh-phrase> (55) a. Shi [Zhangsan]F zai Beijing xue yuyanxue SHI Zhangsan at Beijing study linguistics ‘Zhangsan studies linguistics in Beijing (and not my brother).’ b. Zhangsan shi [zai Beijing]F xue yuyanxue Zhangsan SHI at Beijing study linguistics ‘Zhangsan studies linguistics in Beijing (not in Shanghai).’ (Hole 2012: 48) 11 Assumption: simplex wh-phrases are minimal/maximal phrases such as clitics, which incorporate into a head1: (56) Marco lo+ vedrà [VP <vedrà> (*lo)] Marco will.see CLT ‘Marco will see him tomorrow.’ domain. tomorrow These bear a feature (I will simply call it EPPmin), which triggers head movement of the closest head: (57) FocP shi+shei[EPPmin] Foc’ Foc0[E] <shi> <TP> Zhangsan T‘ T νP <Zhangsan> xihuan <shei> This derives the optionality of shi, since shi+wh-phrase form a complex head. Complex wh-phrases do not have this property (cf. Footnote 1) and therefore shi is optional. (58) CP C[EPPmin]+shi FocP shenme ren Foc‘ <Foc0>[E] shi <TP> Zhangsan T‘ T νP <Zhangsan> xihuan <shenme ren> 1 In Murphy (2014), I present empirical evidence for a number languages that the Foc0 position is never realized in sluicing (Sluicing-FOC Generalization) and therefore shi has to move higher in order to ‘escape ellipsis’. Evidence for this comes from swiping. Merchant (2002) analyses the inversion of preposition and sluice as (i) Mary got a letter today, but she wouldn’t say [who+from]. (ii) *Mary got a letter today, but she wouldn’t say [which person+from] 12 Therefore: if shi does not move, shi is deleted. If it does move, it is realized (preceding the wh-phrase). Summary Problem: the standard analysis of sluicing has wh-movement a pre-requisite. This posed a derivational problem for wh-in-situ languages. We have seen that sluicing structures tend to behave differently from clefts in wh-in-situ languages (case-matching effects, adjuncts, multiple sluicing). Pseudosluicing (…I don’t know who it is.) has been widely adopted for Mandarin Chinese in recent years (Adams & Tomioka 2012, Li & Wei to appear). I presented a number of arguments against this analysis (distribution of shi, sloppy readings, binding). I have shown that despite the presence of the copula (which can be analysed as a focus marker), we can appeal to a focus movement analysis for wh-in-situ languages. Such analyses have been proposed for Korean (Kim 1997), Japanese (Hiraiwa & Ishihara 2002), Bahasa Indonesia (Fortin 2007) Farsi (Toosarvandani 2008), Mongolian (Sakamoto 2011) and Turkish (İnce 2009). What next? Look at lesser-studied languages such as Vietnamese! (Duffield 2013): (59) Amy co mua mot cai gi do, nhung Amy ASR buy one CLS what DEM but cai gi. CL what ‘Amy bought something, but I don’t know what.’ toi khong I neg biet know ?(la) COMP References Duffield, Nigel. 2013. Minimalism and Semantic Syntax: Interpreting Multifuctionalism in Vietnamese. Ms. Konan University, Japan. Hiraiwa, Ken & Shinichiro Ishihara. 2002. Missing Links: Cleft, Sluicing, and “No da” Construction in Japanese. MIT Working Papers in Linguistics 43. 35–54. Hole, Daniel. 2012. The information structure of Chinese. In Manfred Krifka & Renate Musan (eds.). The Expression of Information Structure. Berlin/New York: de Gruyter Mouton. 45-70. İnce, A. 2009. Dimensions of ellipsis: Investigations in Turkish. PhD thesis, University of Maryland. Merchant, Jason. 2001. The Syntax of Silence: Sluicing, Islands and the Theory of Ellipsis. Oxford: OUP. Murphy, Andrew. 2014. The Syntax of shì: A Focus Movement Account of Sluicing in Mandarin Chinese. MA Thesis. Humboldt-Universität zu Berlin. Sakamoto, Yuta. 2011. Toward the Reconsideration of the Cleft Analysis of Sluicing in Wh-in-situ Languages. Proceedings of GLOW in Asia Workshop for Young Scholars. 277-291. Takahashi, Daiko. 1994. Sluicing in Japanese. Journal of East Asian Linguistics 3. 265-300. 13 Wei, T.-C. 2004. Predication and Sluicing in Mandarin Chinese, Ph.D. dissertation, National Kaohsiung Normal University, Taiwan. Wei, T.-C. 2009. Some Notes on Sloppy Identity in Mandarin Chinese. Concentric: Studies in Linguistics 35.2. 269-306