Sensorimotor maps and vowel development in English, Greek, and Korean: A cross-linguistic perceptual categorization study.

Learning to speak involves learning the association between different articulatory maneuvers and their associated auditory-perceptual characteristics. These associations may change over the course of development as a consequence of the nonlinearities inherent in vocal-tract growth. A set of recent studies by Ménard and colleagues has examined these developmental changes by studying adults’ categorizations of synthesized vowels based on an articulatory model of the growing vocal tract. These have shown that vowels modeled on younger vocal tracts tend to be perceived as more front than those modeled on older vocal tracts. They have also shown language-specificity in the vowels that are identified in growing vocal tracts, with French listeners labeling fewer sounds as /u/ than English ones, presumably because the high-vowel space of French includes three vowels /i y u/, while English has only two. The current experiment expands on this by examining the perception of synthesized vowels based on seven vocal-tract growth stages by speakers of three languages with very different vowel systems, American English, Korean (seven vowels), and modern Greek (five vowels). Preliminary analyses show language-specific patterns in the perception of vowels in developing vocal tracts. [work supported by NIDCD 02932 and NSF grants BCS0729140 and BCS0729277]