Consonant environment specifies vowel identity

Publisher Website
Google Scholar
Abstract
While vowels can be produced with static vocal tract configurations, the resulting steady-state tokens are misidentified frequently by naive [human] listeners. The perception of isolated vowels was compared with vowels spoken in a fixed consonantal frame by the same set of 15 talkers. Vowels in /p-p/ syllables were identified with far greater accuracy than were comparable isolated vowels in both single and multiple-talker conditions. Acoustical analyses of the test tokens showed that the poor intelligibility of isolated vowels could not be attributed to talkers'' failure to produce these vowels correctly. Vowels in syllables in which the initial and final stop consonant varied unpredictably from item to item were still identified with greater accuracy than were isolated vowels. Dynamic acoustic information distributed over the temporal course of the syllable is apparently utilized regularly by the listener to identify vowels.

This publication has 4 references indexed in Scilit: