Fricative voicing in the Buckeye corpus

Many claims about the prevalence of phonetic voicing in English obstruents have been made in the literature over the decades, and this study is focused on the fricatives [v, ð, z, ʒ]. Textbooks summarize the consensus that these consonants are voiced through no more than a small portion of their duration unless followed by another voiced sound. In this study, the prevalence of voicing in all 51,361 voiced fricatives of the Buckeye Corpus of American English (39 speakers) has been examined in the utterance-initial(i.e., following a pause), intervocalic, utterance-final (preceding a pause), and preconsonantal positions. The fricatives are indeed less fully voiced in general when not followed by a voiced sound, but are usually voiced for at least 50% of their duration, and often are fully voiced in these environments. Devoicing is most prevalent in utterance-final positions, more so than before a voiceless consonant.