The representation of voice and contrasts in children's early phonological systems

Studies on the acquisition of voice contrasts by children have focused on when they appear and their acoustic properties. The present study examines the issue of how children's early voice contrasts should be represented. It explores this issue with three classes of consonants: voiced stops, voiceless stops, and nasals. Data are first presented from phonological analyses of selected children that show these three classes need to be represented by different features dependent on syllable position. The tendency is for children to merge the voiced and voiceless stops in syllable initial position, leading to a distinction in the feature [sonorant] to distinguish them from nasals. In syllable final position, however, some children associate final voiced stops with final nasals, in contrast to final voiceless stops, requiring a feature distinction of [voice] rather than [sonorant]. If children's phonologies are developed under the influence of language universals, it is proposed that a possibility should also exist in adult phonological systems. Data are presented from Japanese, Quichua, Warao and Mixtec, based on studies by Piggott (1992) and Rice (1993), to demonstrate that such may be the case. For example, syllable initial nasals in Japanese must be represented as [sonorant] so that they do not block Rendaku. Meanwhile, syllable final nasals must be represented as [voiced] in order to participate in regressive voicing assimilation. Also, some data exist indicating that nasal harmony from syllable initial position is blocked by obstruents, while nasal harmony from the end words may be blocked by voiceless segments. The article concludes with a suggested typology of four kinds of languages based on whether nasals as represented as [sonorant], [voiced] or some combination of both.