Accounting for the accented perception of vowels: Universal preferences and language‐specific biases.

Abstract

Strange things happen in cross‐language and second‐language vowel perception: Nave non‐native listeners have been reported to rely on acoustic properties which are nonfunctional in their L1 and dysfunctional for the perception of non‐native vowels; naïve non‐native listeners’ perception is guided by a preference for vowels that are peripheral in the articulatory/acoustic vowel space; and, in general, naïve non‐native listeners’ perception is not well predicted by comparative analyses of vowels of the native and the non‐native language. This presentation reviews the accented perception of vowels by focusing on two forces which shape non‐native vowel perception: universal perceptual preferences which non‐native listeners (and infants) bring to the task of vowel perception, and perceptual biases which non‐native listeners transfer from their native to the non‐native language. Strange and her colleagues have shown that these biases cannot be predicted from acoustic comparisons; rather, they have to be examined directly through assessments of the perceived cross‐language similarity of vowels. This presentation addresses several of the still unresolved questions regarding the design and the interpretation of perceptual assimilation tasks used to account for the accented perception of vowels. [Work supported by Danish Research Council for the Humanities, Canadian Natural Sciences and Engineering Research Council.]