Paul Hagstrom
Johns Hopkins University
Q-movement
Syntax 1/10/99 10:20am

In Japanese, questions are formed with a question particle ("Q") ka or no.  This
talk argues that in wh-questions, Q moves syntactically from a clause-internal
position to the clause-final position it occupies on the surface.  Evidence for
this conclusion includes "intervention effects" where Q-movement is prevented
over another instance of ka (e.g., in dareka 'someone') and a crosslinguistic
comparison with another wh-in-situ language, Sinhala, where Q occupies its
clause-internal position on the surface.  Investigating questions with islands
reveals that Q moves from a position just outside the island (visible in Sinhala
and deducible in Japanese), which correctly predicts a striking and
counterintuitive result: if a question is ill-formed because Q must move over an
intervening instance of ka (1), embedding the intervenor and wh-word together
inside an island improves the sentence (2), since Q, moving from outside the
island, never need cross the intervenor.

(1) ?? dareka-ga nani-o katta-no?
       someone-nom what-acc bought-Q
       ('What did someone drink?')
(2)    Mary-wa [ dareka-ga nani-o atta atode ] dekaketa-no?
       Mary-top someone-nom what-acc bought after left-Q
       'Mary left after someone bought what?'