| |
Abstract:
Four experiments examined the relative strength and
time-course of cross-modal associative priming when the primes were
presented in isolation versus in sentence context. In a pretest,
native speakers of British English produced association responses
to monosyllabic English words of high frequency; 32 such words, all
with strongly preferred associate responses, were chosen as prime
words. The prime-target pairs included antonym (give-take),
attribute (rock-hard), collocation (cup-saucer), associative
(bone-dog) and co-ordinate (verse-poem) relationships. In all four
priming experiments, the subjects' task was to decide as rapidly as
possible whether each target word was a real English word. The
targets were presented visually, each preceded by auditory
presentation of its prime. We compared correct "YES" responses to
the targets preceded by the related prime (give-take) versus by an
unrelated control prime (night-take). In each experiment the 32
experimental trials were embedded in a list with 68 other trials,
in 50 of which the visually presented target was a nonword. 128
native speakers of British English participated, 32 in each
experiment.
In the first two experiments, the primes were spoken in
isolation. For instance, listeners heard "give" and saw on the
screen TAKE, or they heard "night" and saw TAKE. In Experiment 1
targets were presented immediately at offset of the spoken prime;
in Experiment 2 targets were presented 500 ms after prime offset.
In both experiments significant priming effects were observed. At
offset, mean response time (RT) to targets was 565 ms following
related primes and 606 ms following unrelated primes; 500 ms after
offset, RT was 539 ms following related primes and 577 ms after
unrelated primes. In both cases the facilitatory effect of a
related prime (41 ms, 38 ms) was significant at < .001; the
priming effect did not interact with the factor prime-target
interval.
In the following two experiments the same primes were spoken in
sentences, e.g. "We wondered whether we should give them such low
scores". The sentences contained no other words strongly associated
to the prime, and the part of the sentence preceding the prime did
not allow the prime word to be predicted. Again the target occurred
at prime offset in one experiment and 500 ms later in the other. No
significant priming effects occurred in either experiment. At
offset, mean RT to targets was 542 ms following related primes and
547 ms following unrelated primes; 500 ms after offset, RT to
targets was 55 ms following related primes and 562 ms following
unrelated primes. The disappearance in the sentence context of the
robust effects found in isolation is unexpected given the many
successful demonstrations of cross-modal associative priming in
sentences. We suggest that facilitatory effects are more likely to
occur with longer prime words, or with more constraining contexts.
In any case associative priming with related pairs such as those of
the present study does not occur automatically in sentence
processing. In other words, cross-modal associative priming in
sentences is not a direct reflection of lexical access or
activation.
|
