The Weekly Newsletter of MIT Linguistics

Syntax Square 4/15 - Ayaka Sugawara

Speaker: Ayaka Sugawara
Title: A-scrambling, Reconstruction and the Computation of Alternatives under Prosody in Japanese: Evidence from Acquisition
Date/Time: Tuesday, Apr 15, 1-2p
Location: 32-D461

(Joint work with Ken Wexler.)

A major open question in the theory of language acquisition is why children speaking English seem to have difficulty interpreting inverse scope of negation and a universal subject quantifier. Our results contribute both to the solution to this puzzle and provide evidence for particular approaches to the A-movement of Japanese and the theory of contrastive topic. We will argue that children have difficulty with at least some forms of reconstruction, but do not have a problem with interpreting a particular logical form. In Japanese, a scope-rigid language, (1) is unambiguous, while its English counterpart is not.

(1) Minna-ga siken-o uke-nak-atta.
     Everyone-NOM exam-ACC take-NEG-PAST
     ‘Everyone didn’t take the exam’ (ok“all>not”, *“not>all”)

One way to get wide scope of negation is to scramble an object over the subject (Miyagawa ‘01, ‘10, a.o.). The scrambled sentence in (2) is ambiguous between the all>not reading (preferred) and the not>all reading (less preferred), while the non-scrambled sentence in (1) does not have the not>all reading.

(2) [siken-o] minna-ga uke-nak-atta.
     Exam-ACC everyone-NOM take-NEG-PAST
     ‘Everyone didn’t take the exam.’ (ok“all>not” ok“not>all”)

In Miyagawa’s analysis, (2) receives the not>all reading because it does not violate rigid scope; the scrambled object optionally moves to [Spec, T], leaving the subject in [Spec, v], thus c-commanding negation. If children accept the not>all reading in (2), then they understand the not>all LF and can access it when reconstruction is not necessary. Our first experiment shows that indeed children accept the not>all reading of (2).

Another way, and the only unambiguous way to obtain the not>all reading is to have a high pitch contour on the universal quantifier followed by a topic marker –wa (Contrastive Topic), as in (3) (Hara ‘06, Nakanishi ‘07, a.o.). The not>all reading is derived by adopting Büring’s (1997) Alternative Semantics approach to German Topic-Focus sentences.

(3) [Minna-wa]_F siken-o uke-nak-atta.
     Everyone-TOP exam-ACC take-NEG-PAST
     ‘Everyone didn’t take the exam’ (*“all>not”, ok“not>all”)

Our second experiment shows that children completely fail to get the unambiguous not>all reading in (3). The difficulty seems to be related to the same type of “alternatives comparison” difficulty that is the major explanation of children’s difficulties with scalar implicatures.