|
Abstract:
An event-related brain potential (ERP) study of Japanese
questions was conducted. Stimulus sentences contained interrogative
or demonstrative pronoun direct objects either "in-situ" (following
the subject and preceding the verb) or "scrambled" (preceding the
subject), as shown in English gloss below (ACC=accusative,
NOM=nominative).
The local newspaper-to according
[what-ACC/that-ACC] the reckless adventurer-NOM finally
[what-ACC/that-ACC]
discovered-Q(UESTION)
'According to the local newspaper, did the reckless adventurer
finally discover what/that?'
Two questions motivated the study. The first was whether ERPs
elicited by scrambled objects would resemble those elicited by
object wh-questions and relative clauses in English and German:
most notably, slow anterior negative potentials, which are often
left-lateralized and are purportedly an index of holding a
displaced element in working memory pending reassignment to its
in-situ position (Kluender and Kutas 1993; King and Kutas 1995,
among others). The second was whether there would be any evidence
of processing specific to wh-words in-situ, which are common in
Japanese. ERP responses to scrambled sentences replicated effects
seen in other languages: (1) slow anterior negative potentials
between scrambled element (filler) and in-situ position (gap), (2)
phasic LAN effects at gap positions, and (3) P600 effects at
pre-gap positions (demonstrative pronoun sentences only).
Furthermore, (4) phasic right anterior negativity at the sentence
final verb+question particle (Q) position suggested increased
processing load for both types of wh-sentences, perhaps due to a
dependency between wh-words and the question particle.
|
