Comparison of Vowel Devoicing for Speakers of Tokyo- and Kinki Dialects

Japanese vowel devoicing is claimed to be less frequent in the Kinki dialect than in the Tokyo dialect (standard Japanese). In the present paper, frequency of vowel devoicing of /i/ in the non-sense /CiCe/ words is compared among Kinki- and Tokyo dialect speakers. The results indicated that, within pitch accent and speech rate groupings, some of the Kinki dialect speakers showed devoicing rates similar to that of the Tokyo dialect speakers. It was also confirmed that the devoicing rate for some Kinki dialect speakers was significantly less than that for Tokyo speakers. For Tokyo and Kinki speakers who had a high rate of devoicing, variation in rate of devoicing due to consonantal environment was 1) highly frequent (almost without exception) when a stop or an affricate was at least on one side, 2) less frequent (seemingly random) when fricatives were on both sides, and 3) infrequent (seemingly random) when there was a following /h/. A similar tendency was found for the Kinki speakers with a smaller rate of devoicing. The present results suggest that devoicing phenomena may be complex, occurring not only at higher (phonological) levels but also at the lower (phonetic) level.