Production of American English word-initial stops in the vicinity of a voiceless stop

Abstract

This study examined how native-American English speakers produced the stop voicing distinction when word-initial stops were positioned before a voiceless stop (/t/) in the sentence repeat CVC ten times. Sentence durations, durations of vowel (/i/) before word-initial stops, number of produced stop closures (/t/, /c/), durations of voicing in closures and VOT were measured. Preliminary analyses on three speakers show that they produced only one closure in most cases. Speakers occasionally produced two closures when slowing down their speaking rate. Vowels preceding /b, d, g/ were barely longer than vowels before /p, t, k/. When speakers produced one closure, closures were occupied by voicing for both /b, d, g/ and /p, t, k/. However, the average durations of voicing in closure were consistently longer for /b, d, g/ than for /p, t, k/. In addition, /b, d, g/ were produced in most cases with short-lag VOT, except when there were two closures, in which case, /b, d, g/ were produced with lead VOT. All /p, t, k/ were produced with VOT greater than 30 ms. Articulatory studies are needed to assess how the sequence of stops is produced and whether the gestures for /t/ are completely or partially eliminated.