Variability in language processing: Processing non-native speech.

摘要

Research on language processing typically focuses on the way people process the speech of native speakers, implicitly assuming that non-native speech is processed in a similar manner. Yet according to the Expectations-Guided Processing model proposed here, listeners adjust their manner of language processing according to their expectations of the speaker. Therefore, examining the way people process non-native speech can add to our understanding of the mechanisms underlying language processing, including their flexibility and constraints. In addition, such research can contribute to our understanding of interactions between native and non-native speakers.;According to the Expectations-Guided Processing model, listeners adjust their manner of language processing when listening to non-native speakers, because they expect non-native speech to be less reliable in conveying the speaker's intentions. Specifically, listeners adjust to non-native speakers by increasing their reliance on top-down processes and decreasing the amount of information they take from the speech, sometimes sufficing with "good-enough" representations. The model also proposes that such adjustment to non-native speakers depends on available cognitive resources.;The first three studies support the model's proposals by showing differences in the type of information listeners rely on when they process native and non-native speech, differences in the final interpretation of native and non-native speech, differences in the ability to recall details from native and non-native speech, and different patterns of lexical competition and lexical access with native and non-native speakers. These studies also show that these adjustments depend on listeners' working memory, with higher working memory allowing more adjustment. The fourth study examines one potential social implication of adjustment to non-native speakers -- differences in the tendency to make the Correspondence Bias. These findings have implications for theories of language comprehension as well as for our understanding of interactions between native and non-native speakers.