Robert A. Avery
Catherine T. Best
Dept. of Psych., Wesleyan Univ., Middletown, CT 06459 and Haskins Labs., 270 Crown St., New Haven, CT 06511
Listeners' native phonology constrains their perception of non-native phonetic distinctions. Although most non-native contrasts are discriminated poorly, recent findings that certain contrasts are discriminated well led to development of a ``perceptual assimilation model'' (PAM), which proposes that the perceptual assimilation of non-native phones to native categories predicts discrimination performance [e.g., Best et al., JEP:HPP 14, 345--360 (1988)]. The current study investigated native phonotactic influences on three assimilation patterns. Non-native contrasts that: (1) assimilate to two categories (TC) and show excellent discrimination; (2) assimilate equally to a single category (SC) and show poor discrimination; (3) assimilate as a category goodness difference (CG) and show good discrimination. American adults labeled and discriminated lip-rounding contrasts among Norwegian high front vowels [/i/--/y/; /(barred you)/--/y/] in CV's and CVC's. In both contexts, Norwegian /i/ of course was assimilated to English /i/; /(barred you)/ was assimilated to /u/. Although /y/ was assimilated equally to /(small capital eye)/ or /i/ in CVCs, it was virtually always called /i/ in CVs, compatible with the English phonotactic constraint against final lax vowels. Thus, /(barred you)/--/y/ showed TC assimilation and good discrimination in both syllabic contexts. In CVs, /i/--/y/ showed SC assimilation and poor discrimination; in CVCs, it showed TC or CG assimilation and good discrimination.