| |
Abstract:
A series of experiments investigated whether spreading
activation, inclusion in a schema, or integrative processing
affected the speed with which words in sentences were processed.
According to some accounts, readers activated precompiled schemas
when processing sentences (Morris, 1994; cf. Kintsch, 1998; Schank
& Abelson, 1977, Schank, 1982). Words should be processed
faster when they refer to concepts activated when the schema is
accessed. Alternatively, incoming words that are associated with
previously mentioned concepts should be processed more rapidly
because of spreading activation, irrespective of any schema-based
influences (Meyer & Schvaneveldt, 1976; Seidenberg et al.,
1982; Schustack et al., 1987). The facilitated integration
hypothesis, by contrast, proposes that the source of priming (i.e.,
fast processing time relative to some baseline) observed in prior
studies is based upon ease of integration with an evolving
discourse representation (Hess et al., 1995; Foss, 1982; Foss &
Ross, 1983; Foss & Speer, 1991; Traxler & Foss, in press).
We investigated these competing hypotheses in a set of
eye-tracking, self-paced reading, and naming experiments.
In the first set of experiments, we manipulated degree of
association and plausibility independently, in sentences like (1):
(Eye tracking Means on "spices")
1a. The cook combined the spices as she had been trained to do.
358 425
1b. The cook seasoned the spices as she had been trained to do.
350 452
1c. The woman combined the spices as she had been trained to do.
355 414
1d. The woman seasoned the spices as she had been trained to do.
365 493
If association or inclusion in a schema is the mechanism that
supports priming, then we should have observed faster processing on
the word "spices" in 1a relative to 1c or in 1b relative to 1d. If
fast processing is based on integrative processes, then we should
have observed faster processing on "spices" in 1a and 1c relative
to 1b and 1d. Eye-tracking, cross-modal naming, and self-paced
reading experiments all failed to show any benefit of association
or inclusion in a schema, but all of the experiments showed some
effect of plausibility (our operational definition for ease of
integration). In the second set of experiments, we used repeated
words as the targets because such words represent the highest
possible degree of association. Self-paced reading data for
sentences like (2) showed no benefit of association on either the
target word or the following word, but did show a benefit of
semantic plausibility on the following word. (Eye tracking means on
2nd "maid")
2a. The maid watched the maid in the grand mansion on the hill.
370 462
2b. The maid cleaned the maid in the grand mansion on the hill.
363 531
2c. The girl watched the maid in the grand mansion on the hill.
404 493
2d. The girl cleaned the maid in the grand mansion on the hill.
418 545
Eye-tracking data, however showed an early benefit of repetition
that disappeared by the first fixation following the target
word.
We conclude that lexical access is not affected by spreading
activation from previously encountered associates or by inclusion
in a putative schema. Further experiments are in progress that will
allow us to distinguish between visual-form and repeated lexical
access explanations for the early benefit of repetition in the
second set of experiments.
|
