| |
Abstract:
Many researchers have theoretically proposed that in
processing Japanese, an SOV language, Case-markers attached to
arguments play an active role in hypothesizing the sentence
structure even before the verb information becomes available
(e.g., Inoue, 1991; Mazuka and Lust, 1990). Indeed, Yamashita
(1994, 1997) experimentally demonstrated that a parser
hypothesizes how the sentence will unfold as they read a sequence
of Case-marked arguments.
Exactly how the information from words marked by Case markers
or by other particles play a role in processing Japanese,
however, has never been extensively investigated. Most
critically, it is not yet known how an ambiguity in Case marking
such as the particle 'ni,' which is a Dative Case marker
(John-ni, 'to John') but which is also an adverbial morpheme
(kantan-ni, 'easily'), affect the processing. Because both the
Dative marked argument and the manner/temporal adverbs (adjuncts)
are composed of a noun phrase marked by 'ni,' and they both
appear at the same position, the only way to discriminate such
adjuncts from Dative marked arguments is by accessing the content
of the NP and combine it with 'ni' marking. At the very initial
stage of processing, does the parser simply scan the information
from the marker 'ni' for structural decision, or is
argument/adjunct distinction made immediately?
Experimental results using a lexical decision task demonstrate
that argument/adjunct distinction is immediately made, at least
within 400 ms. after the presentation of the ambiguously marked
word. The condition (A), which contains two arguments, each
marked by Nominative and Accusative, followed by a 'ni'-marked
manner/temporal adjuncts and a transitive verb target, was
compared to two other conditions: condition (B) containing three
Case-marked arguments (Nominative, Accusative, and Dative marked
arguments, respectively) followed by a ditransitive verb target,
and condition (C) containing the same Case-marked arguments
followed by a transitive verb target. It was predicted that if
the parser did not recognize the 'ni'-marked adjunct in condition
(A), the parser would treat the adjunct as 'ni'-marked argument,
and the condition (A) will pattern with (C). On the contrary, (A)
would pattern with (B) if argument/adjunct distinction is made
immediately. The results showed that (A) patterned with (B).
These results strongly indicate that Case-marked arguments are
immediately distinguished from adjuncts. Information on the
number of arguments is available immediately to the parser, and
it is utilized in the expectation of what type of verb
(transitive or ditransitive) would follow.
Inoue, A. (1991). A Comparative Study of Parsing in English
and Japanese. Doctoral Dissertation, University of Connecticut,
Storrs, CT.
Mazuka, R., and Lust, B. (1990). "On parameter setting and
parsing: Predictions for cross-linguistic differences in adult
and child processing." In L. Frazier and J. De Villiers (eds.),
Language Processing and Language Acquisition.
Kluwer Academic Publishers, Dordrecht.
Yamashita, H. (1994). Processing of Japanese and Korean.
Doctoral dissertation, The Ohio State University, Columbus,
OH.
Yamashita, H. (1997). "Processing of scrambled sentences in
Japanese and types of pre-verbal information."
Journal of Psycholinguistic Research,
26, 163--188.
|
