Developing acoustic measures to evaluate the emergence of phonological contrast.

Abstract

Acoustic analysis has been used in studying phonological development since at least the 1960s when Preston and colleagues measured voice onset time (VOT) in English- and Arabic-learning toddlers. Studies applying measures such as VOT, for stop voicing, and spectral peak frequency, for fricative place, show that the emergence of robust phonemic contrast is more gradual than it seems when phonetic transcription is used alone. For example, Japanese-speaking children’s short-lag productions of target /d/ labeled as “[t]” might show somewhat shorter VOT than their productions of target /t/. Studies that supplement acoustic measures with continuous labeling tasks, such as a visual analog scale with “d” and “t” as target endpoints, support the idea that not only is this developmental phenomenon of “covert contrast” ubiquitous, the associated subphonemic variation can be perceived by adult listeners given the appropriate task. Moreover, listeners with different kinds of experience with young children have different sensitivities to this acoustic variation. This finding suggests that appropriate acoustic measures can be developed to provide psychometrically valid measures over the full course of development from initial stages of producing undifferentiated or inappropriate cues to producing the adult community pattern for the contrast. [Work supported by NIDCD grant 02932 and NSF grants BSC-0729306, BSC-0729140, and BCS-0729277.]