Integration of acoustic cues in Spanish voiced stops

Abstract

In Spanish, voiced stops in initial position usually show the presence of a voice bar before the release burst. In this study, the perception of word-initial voiced stops when the voice bar is removed is examined. Nine men and nine women pronounced 15 Spanish words in which the initial consonant was /b/, /d/, or /g/ in combination with the vowels /a/, /e/, /i/, /o/, /u/. Seventeen subjects with normal hearing heard the stops in two conditions: (1) whole voice bar + 102.4 ms of signal from burst onset; (2) 102.4 ms from burst onset. In condition (2), tokens were perceived as the correct stop, as its voiceless counterpart, or as an ambiguous signal. Removing the voice bar caused 37% of /b/ tokens to be classified as /p/; 41% of /d/s to be classified as /t/s ; and 19% of /g/s to be classified as /k/s. The change in voicing characteristics was significantly more marked for women. The results indicate that the integration of the voice bar with the burst and the vowel plays a relevant role in the perception of voicing in voiced stops, especially in women’s speech, smoothing the dynamic frequency characteristics of burst and vowel.