Acoustic Properties of Japanese and English Vowels: Effects of Phonetic and Prosodic Context MIWAKO HISAGI, KANAE NISHI, AND WINIFRED STRANGE The City University of New York-Graduate Center 1. Introduction Adult learners of foreign languages have considerable difficulty learning to perceive and produce some, but not all, non-native vowels (e.g., Nishi et al., 1998). According to Flege’s Speech Learning Model (Flege, 1995) non- native vowels that are “phonetically similar” to native vowels will be per- ceptually assimilated to native categories, resulting in interlanguage (IL) representations which subsume both L1 and L2 vowels. This will lead to accented production of the L2 vowels, and to perceptual confusions if two L2 vowels are assimilated to the same L1 category. In contrast, non-native vowels which are highly dissimilar phonetically from L1 vowels will (with L2 experience) come to be represented separately from all L1 categories in IL, and may be produced as more native-like. In predicting these patterns of perceptual assimilation, many previous studies have inferred phonetic similarity/dissimilarity from cross-language comparisons of the acoustic structure of vowels produced in only one or a few phonetic and prosodic contexts (often in CV or CVC syllables produced in lists). Previous re-