Negative concord in focus sensitive yes/no questions
Transkript
Negative concord in focus sensitive yes/no questions
Negative concord in focus sensitive yes/no questions Turkish yes/no questions are formed with a clitic that attaches to a focused element, but in the presence of a negative concord item, the only possible attachment is verbal, even though this is not the broad focus placement. I argue that this ‘implicit negation’ is not due to a morphosyntactic configuration like c-command, and propose a characterization in terms of the focus alternatives that the interrogative has. From the SOV declarative in (1a), various YNQs can be made just with differing clitic placement. (1) a. b. c. d. Ali dün yemek yaptı. Ali yesterday dinner made ‘Ali made dinner yesterday.’ Ali mi dün yemek yaptı? Ali MI yesterday dinner made ‘Was it Ali who made dinner yesterday?’ Ali dün mü yemek yaptı? ‘Was it yesterday that Ali made dinner? Ali dün yemek yaptı mı? ‘Did Ali make dinner yesterday?’ Declarative Subject focus Adverb focus Polarity focus Broad focus is not expressed by (1d), but instead with attachment on the object (let’s call it the VP2 position). This is shown by the appropriateness of the two options in the ‘guess what’ scenario in (2). (2) A: Guess why the kitchen is in a mess. B1: Ali dün yemek mı yaptı ? Ali yesterday dinner MI made ‘Did Ali make dinner yesterday?’ B2: #Ali dün yemek yaptı mı? ‘Did Ali indeed make dinner yesterday?’ VP2=Broad focus Verbal=not Alternative questions reveal clearly the possible alternatives in the two cases. (2B1) can compare TP alternatives, while (2B2) can only compare polarity alternatives. The examples are now the first conjunct in the AltQs in (3). (3) a. b. Ali dün yemek mi yaptı, mutfakta kıyamet mi koptu? Ali yesterday dinner MI made kitchen-loc hell MI broke.loose ‘Did Ali make dinner yesterday, or did hell break loose in the kitchen?’ Ali dün yemek yaptı mi, yapmadı mı ? Ali yesterday dinner made MI not.made MI ‘Did Ali make dinner yesterday, or not?’ VP2=TP alts Verbal=Polarity alts The two attachment options cannot be switched around. For instance, verbal attachment cannot be used when comparing TP alternatives (4). Verbal attachment is in no way a default. (4) *Ali dün yemek yap-tı mı, mutfakta kıyamet koptu mu ? Ali yesterday dinner made MI kitchen-loc hell broke.loose MI ‘Did Ali make dinner yesterday, or did hell break loose in the kitchen?’ (cf. (3a) 1 As much as it indicated broad focus, NCI licensing is not possible in the VP2 attachment, or any other attachment than verbal. This hold even when the NCI is a caseless internal argument, which is presumably the lowest constituent in the structure, readily c-commandable. (5) a. Ali hiç yemek yaptı mı? Ali never dinner made MI? ‘Did Ali ever make dinner?’ Verbal=Negative concord b. *Ali <mi> hiç <mi> yemek <mi> yaptı? Ali MI never MI dinner MI made Intended: ‘Did Ali ever make dinner?’, ‘Did ALİ ever make dinner?’ etc. *Neg concord c. *Ali <mi> dün <mü> hiçbirşey <mi> yaptı? Ali MI ysterday MI nothing MI made Intended: ‘Did Ali do anything yesterday?’, ‘Did ALİ do anything yesterday?’ etc. *Neg concord One could think that the clitic is c-commanding the NCI in its verbal attachment, like the negative suffix that also appears on the verb. But the YNQ particle is not a Q particle. It does not occur in wh questions (6). It is more a focus particle, with the whole utterance in the scope of an interrogative speech act operator. We would expect this operator to be equally able to c-command the NCI in (5b-c), and not just when the (focus-sensitive) overt clitic appears there. (6) Kim yemek yaptı (*mı)? Who dinner made MI? ‘Who made dinner?’ Whq=No Q particle In fact, it cannot even be said that an interrogative speech act licenses NCIs in Turkish: wh questions (7). (7) *Kim hiç yemek yaptı? Who never dinner made? ‘Who’s ever made dinner?’ Whq=*Neg concord It looks like verbal attachment YNQs have an ‘implicit negation’ that warrants negative concord. Where is it? It looks like not in the morphosyntactic makeup, but in the meaning. First, not just polarity, but tense and verb focus can be expressed with verb attachment. But these cannot license NPIs. It is not enough that the clitic is in the right morphosyntactic position, it must have the right meaning. (8) a. Ali hiç yemek yaptı mı, Ali never dinner made MI ‘Did Ali ever make dinner, b. *Ali hiç yemek yaptı mı, Ali never dinner made MI ‘Did Ali ever make dinner, yapmadı mı? not.made MI? or not?’ yapacak mı? will.make MI? or will he?’ Polarity focus=Neg concord Tense focus=not In rhetorical questions with VP2 attachment, NCI licensing suddenly becomes possible. Here we lack whatever morphosyntactic makeup verbal attachment may have, but apparently it is not needed after all, when the implied meaning is negative. (9) Hiçkimse-ye yalan mı söyledim? nobody-dat lie MI said.1sg ‘Did I lie to anybody?’ = ‘I didn’t lie to anybody!’ I formalize this intuition in terms of focus alternatives and argue that a negative focus alternative licenses negative concord in Turkish. In polarity focus, the alternatives are {p, ¬p}. Broad focus alternatives, on the other hand, are {p, q, r . . . }, crucially excluding ¬p, which can be shown independently. This offers a new angle to the clitic placement in Turkish YNQs and NPI licensing crosslinguistically. 2