Final voicing and devoicing in American English

Abstract

English is typically described as a language in which voicing contrast is not neutralized in word-final position. However, a tendency towards devoicing (at least partial) of final voiced obstruents in English has been reported by the previous studies (e.g., Docherty (1992) and references therein). In the present study, we examine a number of acoustic correlates of obstruent voicing and the robustness with which each one is able to differentiate between voiced and voiceless obstruents in the word-final position in the speech recorded by twenty native speakers of the Mid-Western dialect of American English. The examined acoustic properties include preceding vowel duration, closure or frication duration, duration of the release portion, and duration of voicing during the obstruent closure, frication, and release. Initial results indicate that final voiced obstruents are significantly different from the voiceless ones in terms of preceding vowel duration and closure/frication duration. However, release duration for stops does not appear to correlate with voicing in an equally reliable fashion. A well-pronounced difference in terms of closure voicing between voiced and voiceless final stops is significantly reduced in fricative consonants, which indicates a tendency towards neutralization of this particular correlate of voicing in the word-final fricatives of American English.