, Volume 10, Issue 3, pp 225-245

Japanese Vowel Devoicing: Cases of Consecutive Devoicing Environments

Rent the article at a discount

Rent now

* Final gross prices may vary according to local VAT.

Get Access


In Japanese, the high vowels [i, m] become devoiced when they occur between voiceless segments, e.g., [kitai] 'expectation'. When a word contains a sequence of devoiceable syllables, however, not all high vowels are devoiced, and some are necessarily voiced: [ki∫itsu] 'temperament', [sekit∫itsu] 'room made of rock'. These words contain two consecutive devoiceable vowels, but only one of them, either the first or the second vowel in the sequence, is devoiced. Previous studies have claimed that it is impossible to predict which vowels are to be devoiced in consecutive devoiceable syllables. In this paper, I argue that devoicing sites are predictable when we understand the conflicting factors at work. Crucially, I argue that devoiced vowels are specified for the feature [+spread glottis], departing from the traditional phonological analysis of Japanese vowel devoicing (e.g., McCawley (1968)), which considers devoicing as an assimilation of the feature [−voice]. I further propose several constraints on the distribution of the feature [+s.g.]. I show that the locus of devoicing in consecutive devoicing environments is determined by the interaction of these constraints. The analysis is couched in Optimality Theory (Prince and Smolensky (1993)).