LX CAS 500 Topics in Linguistics: Questions Spring 2006 Paul Hagstrom
Februarys End, 2006 7a: Prosody and Japanese wh-questions
Japanese subjacency violations under stress So is it really ungrammatical to ask what John wants know whether Mary bought? What does it mean to say that it can be rendered acceptable through heavy stress on what? Recall this, from a couple of weeks ago: Watanabe, Akira (1992). Subjacency and S-structure movement of wh-in-situ. Journal of East Asian Linguistics 1:255291. Japanese allows wh-words in islands (1); however, one type of island seems to cause problems; wh-words in wh-islands seem to be about as bad as they are in English (2). (1) John-wa [nani-o katta hito]-o sagasiteiru no? John-TOP what-ACC bought person-ACC looking.for Q What is John looking for the person who bought?
(2) ?? John-wa [Mary-ga nani-o katta kadooka] siritagatteiru no? John-TOP Mary-NOM what-ACC bought whether know.want Q What does John want to know whether Mary bought? Whatever our explanation for (1) is, it somehow doesnt work for (2). Kitagawa, Yoshihisa (2005). Prosody, syntax, and pragmatics of wh-questions in Japanese. English Linguistics 22(2):302346. But, then again, the judgments on (2) are notoriously flaky. Why? It turns out that a lot turns on the prosody one usesthat certain ways of pronouncing (2) makes it sound better than others. (Along with that, we can also observe that the default prosody seems not to be the one that makes (2) workthus, the sentence is usually rated as being not very good.) Normal Japanese prosody is bouncy (rhythmic). (3) John-wa yoku ohiru-ni ramen-o toru. John-TOP often lunch-for ramen-ACC order John often has ramen noodles delivered for lunch. (YK3b)
But in wh-questions, theres a flattening out, after the emphasized wh-phrase (Deguchi & Kitagawa 2002, Ishihara 2002). (4) dare-ga yoku ohiru-ni ramen-o toru no? who-NOM often lunch-for ramen-ACC order Q Who often has ramen noodles delivered for lunch? (YK2b) DAre-ga yoku ohiru-ni ramen-o toru no !
(5)
(6)
# dare-ga yoku ohiru-ni ramen-o toru no ! ?
Kitagawa calls this flattened region the emphatic prosody domain (EPD).
This flattening lasts between the wh-word and its scope (its ka). (7) John-wa [ Mary-ga NAni-o eranda-ka ] imadamo siranai. John-TOP Mary-NOM what-ACC chose-Q still does.not.know John is yet to learn [what1 Mary chose t1 ]. (YK4b)
(8)
John-wa [ Mary-ga NAni-o eranda-to ] imademo omoikonderu no ! ? John-TOP Mary-NOM what-ACC chose-C still believe Q What1 does still believe [that Mary chose t1 ]. (YK5b)
Thus, a question that is ambiguous when written can be disambiguated by prosody: (9) a. keisatsu-wa [kanojo-ga ano-ban DAre-to atteita-ka] imadamo sirabeteiru no!? police-TOP she-NOM that-night who-with seeing-Q still searching Q Are the police still investigating [ who1 she was with t1 that night ]? keisatsu-wa [kanojo-ga ano-ban DAre-to atteita-ka] imademo sirabeteiru no!? police-TOP she-NOM that-night who-with seeing-Q still searching Q Who1 are the police still investigating [ whether she was with t1 that night ]?
b.
(But waitisnt that a wh-island violation? Hmm.) Theres kind of a similar thing that seems to happen between NPIs like daremo (between the NPI and the negation), and concessive clauses (between the indeterminate and mo), except that instead of a low flattening, its a sustained high pitch. _____________________ (10) sonna kantanna koto-o daREMO YAROU-TO SINAkatta. such easy matter-ACC anyone will.do did.not No one was willing to do an easy thing like that. __________________________ (11) ano-ko-wa [ [ naNI-O TABETAI-TO] OMOTTE]-mo kuchi-ni-wa-dasanai that-child-TOP what-ACC want.to.eat.C think -even does.not.mention Whatever that kid may want to eat, he will not mention it. Watanabes example, with local EPDof course its no good, kadooka is for y/nqs. (12) ?? John-wa [Mary-ga NAni-o katta kadooka] imadamo siritagatteiru no!? John-TOP Mary-NOM what-ACC bought whether still know.want Q Does John still want to know whether Mary bought what? With global EPD, its acceptable and interpretable for most speakers (even better with ka) (13) John-wa [Mary-ga NAni-o katta ka(?dooka)] imadamo siritagatteiru no!? John-TOP Mary-NOM what-ACC bought whether still know.want Q What1 does John still want to know whether Mary bought t1 ?
Conclusion: Watanabes example doesnt show a Subjacency effect in Japaneserather, it just shows that its hard to assign the correct prosody. Additional wh effects can be explained in a similar way. First, a detour into the prosodic interpretation of multiple wh-questions.
Two wh-words can be assigned the same scope by giving them overlapping EPDs: (14) a. keisatsu-wa [ano-ban DAre-ga DAre-to atteita-ka] minna-ni tazuneta no!? police-TOP that-night who-NOM who-with seeing-C everyone-DAT asked Q Did the police ask everyone [who was with whom that night]? (YK18a)
b.
keisatsu-wa [ano-ban Mary-ga DAre-to atteita-ka] DAre-ni tazuneta no!? police-TOP that-night Mary-NOM who-with seeing-C who-DAT asked Q Who1 did the police ask whom [whether Mary was with him1 that night]? (YK18b)
c.
keisatsu-wa [ano-ban DAre-ga DAre-to atteita-ka] kimi-ni tazuneta no!? police-TOP that-night who-NOM who-with seeing-C you-DAT asked Q Who1 did the police ask you [whether he1 was with whom that night]? (YK18c)
d.
keisatsu-wa [ano-ban DAre-ga dare-to atteita-ka] kimi-ni tazuneta no!? police-TOP that-night who-NOM who-with seeing-C you-DAT asked Q Who1 did the police ask you [who2 he1 was with t2 that night]? (YKfn11i)
(15)
[ano-ban DAre-ga dare-to atteita-ka ] keISATSU-WA DAre-ni tazuneta no!? that-night who-NOM who-with seeing-C police-TOP who-DAT asked Q Whom1 did the police ask t1 [whom2 Mary was with t2 that night]?
And now, we can understand the additional wh effect: (16) a. John-wa [Mary-ga NAni-o katta kadooka] DAre-ni tazuneta no!? John-TOP Mary-NOM what-ACC bought whether who-DAT asked Q Who did John ask whether Mary bought what?
b. ?? John-wa [Mary-ga NAni-o katta kadooka] Tom-ni tazuneta no!? John-TOP Mary-NOM what-ACC bought whether Tom-DAT asked Q Did John ask Tom whether Mary bought what?
b".
John-wa [Mary-ga NAni-o katta ka(dooka)] Tom-ni tazuneta no!? John-TOP Mary-NOM what-ACC bought whether Tom-DAT asked Q What did John ask Tom whether Mary bought?
(17)
a. ?? John-wa [DAre-ga NAni-o katta kadooka] Tom-ni tazuneta no!? John-TOP who-NOM what-ACC bought whether Tom-DAT asked Q Did John ask Tom whether who bought what? b. John-wa [DAre-ga NAni-o katta ka(dooka)] Tom-ni tazuneta no!? John-TOP who-NOM what-ACC bought whether Tom-DAT asked Q Who1 did John ask Tom [ whether he1 bought what ]?
Given that such sentences are often not saved when judged by assigning a global EPD, there must be some kind of bias toward local EPD. (Also, this is evidence that we do assign prosody, even when judging a written sentence silently to ourselves. Things that influence prosodic structure like the length/placement of relative clauses will also bias judgments for readers.) The bias may be due to something like Avoid monotony. That monotony degrades things can also explain the antisuperiority effect on the additional wh-word. (18) John-wa [Mary-ga foAGURA-O shikago-no fuRENCHI-resutoran-de John-TOP Mary-NOM foie.gras-ACC Chicago-GEN French-restaurant-at maWARI-NO-HIto-ni toMERARERU-made oNAKA-IPPAI tabeta koto]-o surrounding.people-by stopped-until full ate fact-ACC imademo ]. shiranai still do.not.know John is yet to know the fact that Mary ate foie gras at the French restaurant in Chicago until she became completely full and was stopped by people on the scene.
(19) # John-wa [Mary-ga NAni-o shikago-no furenchi-resutoran-de John-TOP Mary-NOM what-ACC Chicago-GEN French-restaurant-at mawari-no-hito-ni tomerareru-made onaka-ippai tabeta ka] surrounding.people-by stopped-until full ate Q imademo shiritagetteiru ]. still wants.to.know John still wants to know what1 Mary ate t1 at the French restaurant in Chicago until she was stopped by people on the scene.
(20) ?# John-wa [Mary-ga NAni-o shikago-no furenchi-resutoran-de John-TOP Mary-NOM what-ACC Chicago-GEN French-restaurant-at mawari-no-hito-ni tomerareru-made tabeta ka] surrounding.people-by stopped-until ate Q imademo shiritagetteiru ]. still wants.to.know John still wants to know what1 Mary ate t1 at the French restaurant in Chicago until she was stopped by people on the scene. (21) John-wa [Mary-ga NAni-o shikago-no furenchi-resutoran-de tabeta ka] John-TOP Mary-NOM what-ACC Chicago-GEN French-restaurant-at ate Q imademo shiritagetteiru ]. still wants.to.know John still wants to know what1 Mary ate t1 at the French restaurant in Chicago. John-wa [Mary-ga NAni-o tabeta ka] imademo shiritagetteiru ]. John-TOP Mary-NOM what-ACC ate Q still wants.to.know John still wants to know what1 Mary ate t1.
(22)
If monotony can degrade a sentence, putting the additional wh-word at the front seems like its not going to help as much as putting it at the end. (23) ? John-wa DAre-ni [Mary-ga NAni-o katta kadooka] tazuneta no!? John-TOP who-DAT Mary-NOM what-ACC bought whether asked Q Who did John ask whether Mary bought what? And of course, one has to pay attention to the relative salience of the situation being discussed on the prosodically good interpretation. (24) ?* Sato-kun-wa [ Suzuki-kun-ga NAni-o tabeta-kadooka] oboete-imasu ka!? Mr. Sato-TOP Mr. Suzuki-NOM what-ACC ate-whether remember Q What1 does Mr. Sato remember [whether Mr. Suzuki ate t1 ]? Mr. Suzuki is suffering from food poisoning and the identity of some specific food item as its cause is being sought. Mr. Sato is believed to remember whether or not Mr. Suzuki ate some specific food item, which may be the crucial piece of information. In the quest for the identity of this food item, the question in (24) was asked of the person who is believed to know the answer.
(25)
hokenjo-wa [ sokuchudoku-kanja-zenin-ga NAni-o tabeta ka] health.dept-TOP food.pois-victim-all-NOM what-ACC ate Q kakunin-shiyou-toshiteiru no!? trying-to-confirm Q What1 is the Dept. of Health trying to confirm [whether all of those who suffered from food poisoning ate t1 ]?
One wonders Tanaka, Hidekazu (200?). Remarks on Becks effects: Linearity in syntax. Linguistic Inquiry 34(2):314 (26) a. ?* dare-mo nani-o kawa-nakatta no? anybody what-ACC buy-NEG.PAST Q (What did nobody buy?) b. nani-o dare-mo kawa-nakatta no? what-ACC anybody buy-NEG.PAST Q What did nobody buy? dare-mo [John-ga nani-o katta ka] sira-nakatta. anybody John-NOM what-ACC bought Q know-NEG.PAST. Nobody knew what John bought.
(27)
a.
b. ?* nani-o1 dare-mo [John-ga t1 katta ka] sira-nakatta. what-ACC anybody John-NOM bought Q know-NEG.PAST. (Nobody knew what John bought.) There is disagreement on the status of (27b). But what would make it bad? Nani needs to be associated with the Q, which is in the embedded clauseso it has to be put back for interpretation. At that point, it should be like (27a). It seems that actually theres something thats going wrong with daremo. You cant (longdistance) scramble over daremo? Tanaka notices that what (26a) and (27b) have in common is that the linear dependencies cross: (26) (27) (28) a." b." a. NPI wh Neg Q wh NPI Q Neg John-ga [ Mary-ga dare-ni nani-mo watasa-nai ka] siritagatteiru. John-NOM Mary-NOM who-DAT anything pass-NEG Q want.to.know John wants to know who Mary doesnt pass anything to.
a. ? dare-ni1 nani-mo2 John-ga [ Mary-ga t1 t2 watasa-nai ka] siritagatteiru. who-DAT anything John-NOM Mary-NOM pass-NEG Q want.to.know John wants to know who Mary doesnt pass anything to. b. ?* nani-mo2 dare-ni1 John-ga [ Mary-ga t1 t2 watasa-nai ka] siritagatteiru. anything who-DAT John-NOM Mary-NOM pass-NEG Q want.to.know John wants to know who Mary doesnt pass anything to. Scramble both, both have to go back, order still matters. Tanaka proposes that t...