| |
Abstract:
We describe two on-line experiments that extend the
investigation of long-distance dependencies by examining the
processing of two different types of
wh-
questions: questions headed by
who
or
what
(e.g., The soldier is pushing the unruly student violently into
the street. Who [1] is the soldier [2] pushing ____ [3] violently
[4] into the street?), and questions headed by
which-
NP (e.g., The soldier is pushing the unruly student violently into
the street. Which student [1] is the soldier [2] pushing ____[3]
violently [4] into the street?). These two types of
wh-
questions can be distinguished on linguistic grounds. For example,
which
phrases are considered referential (i.e., they include
participants taking part in the event described, e.g., "student")
while
who
and
what
questions are considered non-referential. Related to their
referentiality,
which
phrases pick out an individual from a set of individuals
explicitly mentioned or inferred from the discourse and are
therefore considered discourse-linked (D-linked), whereas
who
and
what
phrases do not have to be. We summarize these differences and
examine their implications for sentence processing.
For each experiment, normal listeners (15 subjects each) were
randomly assigned to one of each of four probe positions
(between-subjects). A cross-modal priming task was used, whereby
lexical probes either related or unrelated (control) were presented
visually (within-subjects). Results are shown in the tables
below:
We found that for
who-what
questions an antecedent mentioned in a previous sentence (e.g.,
"student") was activated at the
wh-
word (probe position 1), then re-activated at the wh-gap (position
3). For
which-
NP questions, we also found activation for the antecedent at the
which-
NP, but found unambiguous evidence for gap-filling only several
hundred milliseconds downstream from the gap (position 4). We
interpret our data in terms of a sentence processing account that
integrates discourse representations into putatively automatic
operations of the parser. Specifically, we suggest that
which-
NP questions that require contact with the discourse subsume
greater processing load, thus delaying gap-filling. Our results are
buttressed by data from normal adults (e.g., Shapiro & Hestvik,
1995) and children (e.g., Avrutin, 1999), and from aphasic patients
(e.g., De Vincenzi, 1996; Hickok & Avrutin, 1995).
References
Avrutin, S. (1999). Comprehension of discourse-linked and
non-discourse-linked questions by children and Broca's aphasics.
Paper submitted for publication.
De Vincenzi, M. (1996). Syntactic analysis in sentence
comprehension: Effects of dependency types and grammatical
constraints.
Journal of Psycholinguistic Research,
25, 117-133.
Hickok, G., & Avrutin, S. (1995). Representation,
referentiality, and processing in agrammatic comprehension: Two
case studies.
Brain and Language 50,
10-26.
Shapiro, L. P., & Hestvik, A. (1995). On-line comprehension of
VP-ellipsis: Syntactic reconstruction and semantic influence.
Journal of Psycholinguistic Research, 24,
517-532.
|
