Cross-language and language-specific acoustic correlates of gemination in Berber and Japanese

摘要

It is usually posited that closure duration is cross-linguistically the most reliable and consistent acoustic feature of gemination. In this study, we conduct an acoustic analysis of singleton vs. geminate voiced and voiceless stops, produced in fast and normal speaking rates, in two varieties of Berber (Tarifit and Taqbaylit) and Japanese. Speech rate is varied in order to evaluate the robustness of the phonological contrast. The main thrust of the present investigation is to show how the singleton vs. geminate contrast functions in two related languages, Tarifit Berber and Taqbaylit Berber, which belong to the same Afro-Asiatic family, compared to Japanese which belongs to a different language family, the Altaic family. The corpus, of elicited materials, consists of three different corpora (one corpus for Tarifit, one for Taqbaylit and one for Japanese) of singleton vs. geminate minimal stop pairs, voiced and voiceless, inserted in carrier sentences. Acoustic analyses are based on measurements of absolute durations of intersegmental and intrasegmental parameters for the three languages, in the two speaking rate conditions. Relative timing is also observed by examining the proportion of the target singleton and geminate consonants within the CV syllable. Results show, on the intersegmental level, that closure duration is the main cue for this phonological contrast at the two speaking rates, cross-linguistically. Intrasegmental results reveal that the acoustic silent phase for voiceless stops and consonantal occlusion for voiced stops are the acoustic correlates that distinguish the two phonological categories across the three languages. As concerns the other parameters (VTT, VOT and the flanking vowels), they did not show significant differences, except for Japanese where vowels are longer before geminate consonants than before their singleton counterparts.