Further Issues on the Duration Differences in Vowels due to the Voicing of the Following Stops in English

It is a well-known phenomenon that vowel duration in English is generally longer before a voiced stop than a voiceless one. Past research has postulated that the closure duration of the voiceless stop is generally longer than that of the voiced stop and that the duration of a preceding vowel is determined complementarily by the closure duration of the stop. To shed further light on the phenomenon, this study examined fourteen native speakers of American English who read the monosyllabic words [bVC] (V = [i, ɪ, eɪ, ɛ, æ, ʌ, ɑ], C = [t, d]). First, we found that mean vowel duration was 38 ms longer before the voiced stop than the voiceless (mean duration ratio = 1.24). Second, mean closure duration of the voiced stop was only shorter by 5 ms compared to the voiceless stop (mean duration ratio = 0.97). Therefore, for our subjects, vowel duration was not determined complementarily by the closure duration of the following stop. Third, vowels with longer inherent durations (viz., tense, diphthong, and low vowels) tended to show larger duration ratios in the voiced and voiceless contexts than the vowels with shorter durations (viz., lax vowels). This indicates that the lengthening of inherently shorter vowels before a voiced stop is limited in order to avoid overlapping with longer vowels in the duration range. Fourth, there was no significant gender difference in vowel duration ratios in the contexts of voiced and voiceless stops. Finally, considerable individual differences were found in the vowel and consonant duration ratios.