| |
Abstract:
Understanding a sentence requires both parsing its syntactic
structure and retrieving the meaning of its words from the mental
lexicon, but what is the relationship between these two processes?
One possibility is that lexical retrieval and syntactic parsing
take place independently -- either in parallel or with syntactic
parsing occurring first and determining what is accessed from the
mental lexicon. Another possibility is that syntactic and lexical
processing interact on-line.
This study explored these issues by examining the influence of
clause boundaries on priming between associated word pairs as
reflected in event-related brain potentials (ERPs). A reduction in
the amplitude of the N400, an ERP component linked to
lexical/semantic processing, has already been correlated with
associative priming within a clause (Van Petten, 1993). Here we
compared the response to associated versus unassociated word pairs
within a clause with the response to the same word pairs, separated
by the same number of words, across a clause boundary, as
illustrated in the following example:
ASSOCIATED-WITHIN:
The book described the KINGS and the QUEENS but lacked details she
needed.
UNASSOCIATED-WITHIN:
The book described the KINGS and the HISTORY but was not very
accurate.
ASSOCIATED-ACROSS:
The book described the KINGS, but the QUEENS were not mentioned at
all.
UNASSOCIATED-ACROSS:
The book described the KINGS, but the HISTORY was more fiction than
fact.
Individuals read sentences like these word by word for
comprehension as their ERPs were recorded; the responses to the
targets (e.g., QUEENS/HISTORY) and the primes (e.g., KINGS) were
assessed. If lexical and syntactic processing take place
independently and in parallel, ERP priming effects should not
differ between the within and across clause conditions. If, in
contrast, syntactic parsing precedes and shapes lexical retrieval,
there would be no priming across a clause boundary (Carroll &
Slowiaczek, 1986). Alternatively, on an interactionist account,
priming would be expected in both clause conditions, but the nature
of the effects would be expected to differ qualitatively and/or
quantitatively.
All targets showed an N400 reduction relative to primes,
reflecting the gradual build-up of sentential constraints (Van
Petten Kutas, 1990). In addition, when prime and target were within
the same clause, an N400 priming effect was observed: associated
targets elicited smaller N400s than their unassociated
counterparts. When a clause boundary separated these same words,
however, the N400 effect was smaller than that observed within a
clause, even though the number of intervening words in the two
cases was identical. In addition, the ERP to across clause
associates contained a right frontal positivity from 300 ms onwards
that was not seen for their unassociated counterparts. In sum,
there is evidence for lexical associative priming across a clause
boundary, albeit different in nature from that occurring within a
clause. The results therefore imply that lexical processing is
affected by syntactic parsing during on-line sentence
comprehension.
WORKS CITED
Carroll, P. and Slowiaczek, M. L. (1986). Constraints of semantic
priming in reading: A fixation time analysis. Memory and Cognition,
14(6), 509-522.
Van Petten, C. (1993). A comparison of lexical and sentence-level
context effects in event-related potentials. Language and Cognitive
Processes, 8, 485-531.
Van Petten, C. and Kutas, M. (1990). Interactions between sentence
context and word frequency in event-related brain potentials.
Memory and Cognition, 18, 380-393.
|
