| |
Abstract:
Two cross-modal naming experiments contrasted constraint-based
and facilitated integration accounts of priming in syntactically
unambiguous sentences. Facilitated integration (e.g., Foss, 1982;
Hess, Foss, and Carroll, 1995) predicts that sentence contexts
will facilitate responses to target words as long as those words
make sense. Constraint-based accounts (e.g., Schwanenflugel and
LaCount, 1988; Schwanenflugel and Shoben, 1985) predict that
facilitation depends on the degree to which a sentence generates
expectancies and on the degree of competition between activated
lexical items (e.g.,Spivey-Knowlton and Sedivy, 1995). Thus, our
experiments manipulated the degree of sentence constraint (high
vs. low constraint sentence primes) and the proximity of the
target word and its nearest competitor. In Experiment 1, we found
facilitated responses to all expected target words, regardless of
the proximity of a competitor, and regardless of the degree of
sentence constraint. In Experiment 2, we presented the
second-most likely sentence-final word in high and low constraint
sentence frames. These second-most-likely target words were
either close or distant in multiple-cloze probability to the most
expected target word. The constraint-based account predicts
little priming for targets following high-constraint sentence
frames (due to the presence of a higher probability competitor;
see, e.g., Duffy, Morris, and Rayner, 1988), or an interaction of
constraint and distance between the expected and
second-most-expected target word on a modified account. As in the
first experiment, we find evidence for facilitated response to
the targets across all conditions, and no interation of
constraint and distance between expected and second-most-expected
target. We interpret these results in terms of the facilitated
integration account and in terms of a modified constraint
mechanism.
|
