| |
Abstract:
We explored the effects of reducing working memory resources
on the detection of syntactic and semantic anomalies. This was
accomplished by comparing event related brain potentials (ERPs) for
sentences with a preferred canonical word order (Subject-Relative
[SR]) with sentences having a non-canonical word order
(Object-Relative [OR]). The syntactic condition contained sentences
with a subject verb agreement error. The semantic condition
contained sentences in which the subject and object were reversed
thereby yielding implausible scenarios based on world knowledge. We
assumed that detection of syntactic versus semantic anomalies would
elicit P600-effects versus N400-effects. Anomalies occurred in the
sentence at the end of the clause. In Experiment 1 thirty-nine
participants judged the acceptability of the sentences. In
Experiment 2 forty participants read for comprehension. ERPs were
recorded from thirteen electrodes. Sentences were presented
visually (ISI 645 ms). In both experiments syntactic and semantic
anomalies in SR sentences elicited P600-effects (650-850 ms)
maximal at centroparietal sites. In Experiment 1 only, syntactic
anomalies in OR sentences yielded a delayed P600-effect. In
contrast in Experiment 2, syntactic anomalies in OR sentences
elicited an increase in an early posterior negativity (300-500 ms)
absent in SR sentences and Experiment 1. The finding of similar ERP
patterns for syntactic and semantic anomalies in SR sentences in
both tasks supports a single verbal working memory resource for
different components of language processing.
|
