Perceiving non-native speech.

In a series of experiments using monosyllabic words produced by a native and a non-native speaker of English, native English speakers monitored the word-initial consonants of the words to decide which of two consonants was present on each trial. In some of the experiments, a secondary task of a linguistic nature, deciding whether the target-bearing word was a noun or verb, was also required. When the words were presented in silence, the native and non-native stimuli were processed in a like manner. Specifically, when the secondary task was not required, phonemic decisions tended to be made on the basis of prelexical information, whereas when the secondary task was required, they tended to be made on the basis of postlexical information (see Eimas, Marcovitz Hornstein, & Payton, 1990). However, when the listening conditions were degraded by presenting the words at a lower level and in noise, the two types of stimuli yielded different patterns. Native speech was processed as before, whereas for non-native speech phonemic decisions now tended to be made on the basis of postlexical information both when a secondary task was required and when it was not. The contrasting results for native and non-native speech are discussed in terms of models of phoneme processing.

Base-language effects on word identification in bilingual speech: evidence from categorical perception experiments.

The categorical perception paradigm was used to investigate whether French-English bilinguals categorize a code-switched word as French or English on the basis of its acoustic-phonetic information alone or whether they are influenced by the base-language context in which the word occurs, that is, by the language in which the majority of words are spoken. Subjects identified stimuli from computer-edited series that ranged from an English to a French word as either the English or the French endpoint. The stimuli were preceded by either an English or a French context sentence. In accord with previous studies (Grosjean, 1988), it was found that the base language had a contrastive effect on the perception of a code-switched word when the endpoints of the between-language series were phonetically marked as English and French, respectively. When the endpoints of the series were phonetically unmarked and thus compatible with either language, however, no effect of the base language was found; in particular, we failed to find the assimilative effect that has been observed with other paradigms (Grosjean, 1988; Soares and Grosjean, 1984; Macnamara and Kushnir, 1971). The current results provide confirming evidence that the perception of a code-switched word is influenced by the base-language context in which it occurs and, moreover, that the nature of the effect depends on the acoustic-phonetic characteristics of the code-switched word. In addition, the finding that a contrastive effect occurs across all paradigms used to date, but that an assimilative effect occurs in only some paradigms, suggests that these two context effects may arise at different stages of processing.