Island-sensitive Sluicing in Turkish Although sluicing has been
Transkript
Island-sensitive Sluicing in Turkish Although sluicing has been
Island-sensitive Sluicing in Turkish Although sluicing has been analyzed as being island-insensitive (Ross (1969), Merchant (2001), among others), I demonstrate that sluicing in Turkish is island-sensitive. In Turkish, the wh-phrase in an adjunct clause does not raise to matrix Spec-CP. Turkish has an optional Q-like particle –ki which attaches to both yes-no and wh-interrogatives, (1-2). Because the clitic ki is optional, it cannot be a Q-particle. However, its position after the Q-particle (1b) shows that it is in the CP-domain. Furthermore, it occurs only in matrix clauses (3). The interaction between the wh-phrase in sluicing structures and the particle ki provides evidence that this particle is in the CP because only whphrase(s) and other elements in the C0 domain can be spelled-out in sluicing structures. First, it is not elided in matrix sluicing constructions (4). The elided element in B’s utterance is shown in (5). In (5) the wh-phrase raises to the CP domain. IP is elided, and the wh-phrase and the particle ki are sent to PF. That ki is pronounced with the wh-phrase is support that it is in the CP domain. Second, when a wh-phrase is in an embedded complement clause, it and the particle ki are both pronounced in sluicing constructions (6). However, a wh-phrase in an adjunct clause cannot co-occur with ki in sluicing structures. In nonsluicing cases ki can occur (merge) in matrix clauses (7a' -b' ), but in sluicing, a wh-phrase and ki cannot co-occur (7a-b), in contrast to sluicing in complement clauses. This indicates that a whelement in an adjunct clause does not raise to matrix Spec-CP in Turkish. The Complex NP Constraint is also observed in sluicing structures with ki (8). In non-sluicing cases, ki occurs in the matrix clause and the clause is grammatical with interrogative force (8b) whereas in sluicing structures, the wh-phrase and ki cannot co-occur (8c). Complement clauses are not islands. A wh-phrase can move out of the complement clause. In Turkish, a phrase in a complement clause can scramble to the right- or left-periphery of the matrix clause (9). This contrasts with adjunct clauses which are islands and which do not allow a phrase to scramble to the matrix clause (10). The ungrammaticality of (7) and (8) demonstrates that adjunct and complex NP islands retain their island-properties even in sluicing cases. Thus, sluicing in Turkish is island-sensitive. This finding in Turkish has theoretical implications: that adjunct islands are not simply PF islands, counter to Merchant’s (2001) assumption. If they were only PF islands, (7 a-b) would be possible (see also 7a' ' b' ' ); i.e., when they are elided, the ‘*’ (island) feature would be deleted. The facts indicate that the ‘*’ feature remains in the structure and therefore adjunct islands in Turkish are more than PF-islands. Examples: (1) a. Ahmet geldi mi ? A.-Nom come-Past-3.s. Q-particle ‘Did Ahmet come?’ b. Ahmet geldi mi ki? A.-Nom come-Past-3.s. Q-particle ki ‘Did Ahmet come, then?’ (2) a. Ahmet ne yedi? A.-Nom what eat-Past-3.s. ‘What did Ahmet eat?’ b. Ahmet ne yedi ki? A.-Nom what eat-Past-3.s. ki ‘What did Ahmet eat, then?’ (3) Hasan’ın ne yedi ini (*ki) duydun? H.-Gen. what eat-DIK-3.s.poss.-Acc. hear-Past-2.s. ‘What did you hear that Hasan ate (then)?’ (4) A: Dün seni biri aradı. Yesterday 2.s.pron.-Acc. one-Nom. call-Past-3.s. ‘Yesterday someone called you.’ B: Kim ki? Who-Nom. ki ‘Who, then?’ (as in “Oh? Who?”) (5) Kim dün seni aradı ki? (6) a. Hasan seni birinin aradı ını söyledi. Kim ki? H-Nom. 2.s.pron.-Acc. one-Gen. call-DIK-3.s.poss.-Acc. say-Past-3.s. who-Nom. ki ‘Hasan said that someone called you. Who then?’ a' . Hasan beni kimin aradı ını söyledi ki? a' ' . Kimi Hasan beni kimi aradı ını söyledi ki? b. Hasan Ahmet’in birisine para verdi ini söyledi. Kime ki? H.-Nom. A.-Gen. one-Dat. money give-DIK-3.s.poss.-Acc. say-Past-3.s. Who-Dat. ki ‘Hasan said that Ahmet gave money to someone. Who to then? (as in “Oh? Who to?”)’ b' . Hasan Ahmet’in kime para verdi ini söyledi ki? b' ' . Kimei Hasan Ahmet’in kimei para verdi ini söyledi ki? (7) a. Biri seni arayınca Hasan çok sinirlendi. Kim (*ki)? One-Nom 2.s.pron.-Acc. call-IncA H.-Nom.very get-angry-Past-3.s. Who-Nom. ki ‘When someone called you, Hasan got very angry. Who (then)?’ a' . Kim seni arayınca Hasan çok sinirlendi ki? a' ' . *Kimi kimi seni arayınca Hasan çok sinirlendi ki? b. Ali birinden para alınca Hasan çok sinirlendi. Kimden (*ki)? A.-Nom. one-Abl. money borrow-IncA H.-Nom.very get-angry-Past-3.s. Who-Dat. ki ‘When Ali borrowed money from someone, Hasan got very angry. Who from (then)?’ b' . Ali kimden para alınca Hasan çok sinirlendi ki? b' ' . *Kimdeni Ali kimdeni para verince Hasan çok sinirlendi ki? (8) a. A: Ahmet’in birini vurdu u söylentilerine inanıyorum. A.-Gen. one-Gen. shoot-DIK-3.s.poss. rumour-plu.-3.s.poss.-Dat. believe-Prog.-1.s. ‘I believe the rumour that Ahmet shot someone.’ B: *Kimi ki? Who-Acc. ki? ‘Who, then?’ b. Ahmet’in kimi vurdu u söylentilerine inanıyorsun ki? c. *Kimii Ahmet’in kimii vurdu u soylentilerine inaniyorsun ki? (9) a. Ahmet’ii Hasan [Ali’nin ti dövdü ünü] sanıyor. b. Hasan [Ali’nin ti dövdü ünü] sanıyor Ahmet’ii. ‘Hasan thinks that Ali beat Ahmet.’ (10) *Ali’yij Ahmet a ladı [adjunct Hasan tj dövünce]. ‘Ahmet cried when Hasan beat Ali.’ References Merchant, Jason (2001). The syntax of silence: Sluicing, islands, and the theory of ellipsis. Oxford University Press: Oxford. Ross, John R. (1969). Guess Who?, in R. Binnick, A. Davison, G. Green, and J. Morgan (eds.), Papers from the 5th Regional Meeting of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society, 252-286.