The perceptual assimilation of non-native sounds (i.e., the manner in which non-native sounds are identified with native perceptual categories and their internal structure) may predict the ease with which these sounds can be discriminated and ultimately acquired in second language acquisition (SLA). The focus of the research is to examine the effects of two common sources of variability in the speech signal that may influence the perceptual assimilation of non-native consonants: vowel context and speaking rate. Methodological issues in the elicitation of identification and discrimination data will also be studied to determine the optimal pair of identification and discrimination tests to be used in evaluating speech perception and SLA models. Specifically, the role of Talker Variability (single talker vs. multiple talker within trials) and the Context in which identification judgments are elicited (same context as the discrimination test; isolation) will be investigated in Experiment 1 for target non-native sounds in isolated words and in words in a short frame sentence. The relationship between cross-language consonant perceptual assimilation and discrimination using stimuli produced and presented under different phonetic and prosodic contexts will be examined in the remaining two experiments using the optimal identification-discrimination test pair obtained from Experiment 1. The results and interpretation of the proposed study will: a) provide a more precise understanding of how different phonetic and prosodic contexts affect cross-language consonant perceptual assimilation; b) clarify important methodological issues that obscure our interpretation of perceptual assimilation experiments; and c) provide a theoretical and empirical background for formal modeling of cross-language speech perception and SLA. ? ?
