| |
Abstract:
Most symbolic theories of sentence processing explain
individual differences in comprehension as variations in a
biologically determined working memory capacity (Gibson, 1998; Just
& Carpenter, 1992); experience is largely limited to vocabulary
acquisition and assumed to have little effect on initial parsing.
We present an alternative account based on connectionist and
constraint-based approaches to language comprehension. This account
emphasizes variations in language experience as an important cause
of comprehension differences.
We explore the role of experience in interpreting unambiguous
subject relative (SR) and object relative (OR) clauses in English
such as:
(SR) The reporter that attacked the senator admitted the error.
(OR) The reporter that the senator attacked admitted the
error.
SRs are easier to process than ORs (Holmes & O'Regan, 1981;
King & Just, 1991). We suggest that the processing of SRs is
facilitated by their structural overlap with frequently occurring
simple transitive sentences (e.g., 'The reporter attacked the
senator'). In contrast, ORs have an infrequent irregular OV word
order. Direct exposure is therefore crucial for processing ORs, but
not SRs. In other words, the Frequency ( Regularity interaction
observed in word recognition and syntactic ambiguity resolution may
play an equally important role in unambiguous sentence processing:
The more irregular the sentence construction, the more processing
is dependent on prior experience with such irregular input.
This prediction was tested by manipulating language exposure in
human subjects. Subjects had two training sessions: An experimental
group read sentences with an equal number of SRs and ORs, and a
control group read sentence of similar syntactic complexity without
relative clauses. The Daneman and Carpenter (1980) reading span
task and a standard self-paced reading task involving SRs and ORs
were administered to the subjects before and after training.
The Frequency ( Regularity interaction predicts differential
training effects, with ORs benefiting more from training than SRs.
The results supported this prediction. Training induced differences
in reading time profiles mimicking differences previously
attributed to differences in working memory capacity (King &
Just, 1991): The profiles for the experimental and control subjects
resembled the profiles for high and low span subjects, respectively
- yet there was no increase in reading span as a result of
training.
These results, together with corroborating connectionist
simulations, point to the importance of experience in parsing -
consistent with connectionist and constraint-based accounts - and
offer a reassessment of the importance of a fixed "working memory"
in accounting for individual differences.
References
Daneman, M., & Carpenter, P. (1980). Individual differences in
working memory and reading.
Journal of Verbal Learning and Verbal Behavior,
19, 450-466.
Gibson, E. (1998). Linguistic complexity: Locality of syntactic
dependencies.
Cognition,
68, 1-76.
Holmes, V. M., & O'Regan, J. K. (1981). Eye fixation patterns
during the reading of relative clause sentences.
Journal of Verbal Learning and Verbal Behavior,
20, 417-430.
Just, M., & Carpenter, P. (1992). A capacity theory of
comprehension: Individual differences in working memory.
Psychological Review,
98, 122-149.
King, J., & Just, M. A. (1991). Individual differences in
syntactic processing: The role of working memory.
Journal of Memory and Language,
30, 580-602.
|
