| |
Abstract:
Readers and listeners rapidly construct a semantic
interpretation of a sentence, and do so incrementally, as the
individual words come in. Kutas and Hillyard (1983) for instance
observed that a semantically anomalous word immediately elicits an
enlarged N400 component. Most follow-up research has studied this
N400 effect in isolated sentences only. We conducted two ERP
experiments to see whether we could elicit the same effect with
words that, although semantically coherent in a single sentence,
made no sense in the context of the wider discourse. If people
first relate word meaning to the local sentence context before
relating it to the wider discourse, then a discourse-dependent
semantic anomaly might elicit a delayed N400 effect or a
qualitatively different ERP effect. In the main experiment, we
contrasted sentences like (1a) "Jane told her brother that he was
exceptionally *quick* today" and (1b) "Jane told her brother that
he was exceptionally *slow* today" by presenting each in a wider
discourse context that selectively rendered one of the two critical
words anomalous (e.g., in the discourse, the brother had just won a
race). The discourse-anomalous words elicited a solid N400 effect.
Furthermore, the discourse effect was indistinguishable from a
standard sentence-dependent N400 effect. This supports theories
that, in terms of actual processing, do not distinguish between
semantic integration at the level of the sentence and of the
discourse.
|
