Cross-language speech perception: Initial capabilities and developmental change.

Abstract

This article reports three studies designed to increase our understanding of developmental changes in cross-language speech perception. In the first study, we compared adult speakers of English and Hindi on their ability to discriminate pairings from a synthetic voiced, unaspirated place-of-articulation continuum. Results indicated that English listeners discriminate two categories (ba vs. (Ja), whereas Hindi listeners discriminate three (ba vs. da, and da vs. Da). We then used stimuli from within this continuum in the next two experiments to determine (a) if our previously reported finding (Werker & Tees, 1984a) of a reorganization between 6 and 12 months of life from universal to language-specific phonetic perception would be evident using synthetic (rather than natural) stimuli in which the physical variability within and between categories could be controlled, and (b) whether the younger infants' sensitivity to nonnative speech contrasts is best explained by reference to the phonetic relevance or the physical similarity of the stimuli. In addition to replicating the developmental reorganization, the results indicate that infant speech perception is phonetically relevant. We discuss the implications of these results. Since the early 1970s researchers have been studying infant speech perception, partly in an attempt to determine if there is evidence for a specialized phonetic mode of processing among infants. In a now-classic study, Eimas, Siqueland, Jusczyk, and Vigorito (1971) demonstrated that English-learning infants aged 1 to 4 months show far better discrimination along a synthetic voice onset time (VOT) continuum for two stimuli that straddle the adult /ba/-/pa/ phonetic category boundary than they do for two equally acoustically distinct stimuli from within the same phonetic category. This differential discrimination at some, rather than other, points along a single synthetic contin