F0 perturbation in a "pitch-accent" language

Abstract

This study investigates the relation between consonant voicing and F0 in modern Tokyo Japanese, as produced by young female speakers. In a tone language, the F0 perturbation related to onset voicing has been reported to be inhibited, so that F0 can be maximally used in tonal contrasts. According to this explanation, the same pattern should be found in Tokyo Japanese, as F0 should be maximally used to signal its “pitch-accent”. Contrary to this prediction, our data show that in Tokyo Japanese, for the initial mora, F0 is remarkably lower after voiced than voiceless stops, and this effect lasts till the final part of the mora. However, the F0 level of the mora endpoint is maintained at H or L so that the pitch-accent pattern is well preserved. We thus argue that the competing role of F0 in a pitch-accent language, or even a tone language does not necessarily impose limitations on the F0 perturbation effect. We also found that voiced stops are very often phonetically voiceless in utterance-initial position, while being phonetically voiced in utterance-medial position. Therefore, we question whether Tokyo Japanese is undergoing an incipient tonogenesis, given that the VOT cue is giving place to an F0 cue.