| |
Abstract:
In this poster we present two questionnaire experiments
designed to investigate the effects of discourse status on the
processing of NPs and the resulting implications for sentence
complexity. Both experiments manipulated NPs in the most embedded
positions of doubly center embedded sentences in order to determine
how different NPs affected the intuitive complexity of the
sentences. According to the dependency locality theory (DLT) of
linguistic complexity (Gibson, 1998; Gibson & Ko, 1998), the
intuitive complexity of a sentence is roughly approximated by the
maximal integration cost incurred at any parse state during its
processing, where integration cost is the cost associated with
connecting a newly input word into the structure for the sentence
thus far. Integration cost is hypothesized to be heavily
locality-based, such that the greater the distance (the more
processing that occurs) between an incoming word and the location
to which it attaches, the greater the cost.
The two experiments investigate the role that building and
identifying referents plays in the distance metric within the DLT.
Experiment 1 provided evidence that distance should be quantified
in terms of new discourse referents. Experiment 2 tested an
extension of this idea according to which distance is quantified in
terms of the difficulty of building/accessing a referent in a
discourse model (Haviland & Clark, 1974; Garrod & Sanford,
1982). It is hypothesized that the type of an NP is correlated with
the accessibility of its discourse referent, which is determined
according to a cognitive hierarchy extending from NPs in focus
(pronouns) to type identifiable NPs (indefinites) (Gundel et al.,
1993). Less accessible referents are predicted to require more
processing, so integrations that cross NPs with relatively
inaccessible referents are predicted to be harder than integrations
that cross NPs with easily accessible referents. Experiment 2
tested nested and right-branching versions of sentences with six
kinds of NPs from the hierarchy in the most embedded subject
position: first/second-person pronouns, third-person pronouns, name
from long-term memory, new name, definite description, indefinite
description.
(2) The salesman who the woman who (I, Bob, Microsoft, the
company, a company) hired dealt with was very polite.
The DLT predicted increasing complexity according to the
cognitive status hierarchy in the nested structures, but not in the
right-branching structures. This was the pattern that was observed.
Overall, the results of these experiments offer strong support for
a distance-based linguistic complexity theory, where distance is
quantified in terms of discourse referent accessibility.
References
Garrod, S. C., & Sanford, A. J. (1982). The mental
representation of discourse in a focused memory system:
Implications for the interpretation of anaphoric noun phrases.
Journal of Semantics,
1, 21-41.
Gibson, E. (1998). Linguistic complexity: Locality of syntactic
dependencies.
Cognition,
68, 1-76.
Gibson, E., & Ko, K. (1998). An integration-based theory of
computational resources in sentence comprehension. Paper presented
at the Fourth Architectures and Mechanisms in Language Processing
Conference, University of Freiburg, Germany.
Gundel, J., Hedberg, H., & Zacharski, R. (1993). Referring
expressions in discourse.
Language,
69, 274-307.
Haviland, S. E., & Clark, H. H. (1974). What's new? Acquiring
new information as a process in comprehension.
Journal of Verbal Learning and Verbal Behavior,
13, 512-521.
|
