Morphological and universal devoicing in English and Swedish

Abstract

Voicing alternations in English and Swedish have been analyzed as assimilating a voiced consonant to a voiceless consonant bidirectionally by Anderson (1979) and Halle and Mohanan (1985), among others. According to these accounts, when one of the obstruents in the cluster is voiceless, the other also becomes voiceless, irrespective of their relative temporal order. Expanding on Cho (1990a, b), I argue against the traditional assimilation accounts, and propose several arguments for the position that earlier level voicing agreement in English and Swedish is due to morphologically conditioned devoicing whereas voicing alternation in inflection and in the postlexical domain is triggered by Universal Devoicing. The paper is organized as follows. First, several pieces of independent evidence will be summarized to argue for the position that voicing is a privative feature. Second, a brief typology of voicing assimilation will be presented in which only one value of the feature [voice] is motivated. Third, a universal constraint regarding a voicing contour within a syllable will be argued for. Finally, voicing agreement in English and Swedish will be accounted for by two independent mechanisms. The Linguistic Review 11 (1994), 221-239 0167-6318/94/0011-221 © Walter de Gruyter Brought to you by | Michigan State University Authenticated Download Date | 6/10/15 2:46 PM 1. Whether the feature [voice] should be considered binary in phonetics is an open question. The point of this paper is that no phonological rules need to refer to the minus value of [voice]. 2. Some languages are represented as having voiceless sonorants which may not be distinctive. 3. lai is the only language in which the feature [voice], but not aspiration, seems to play a role. Brought to you by | Michigan State University Authenticated Download Date | 6/10/15 2:46 PM4. There is only one language (T'ien Pao) in which aspirated stops and voiceless sonorants are assigned to different tone series. 5. A phonetic rule of vowel devoicing is often triggered by voiceless consonants and in the phrase-final environment, and governed by non-phonological, non-assimilartory factors.Brought to you by |