Cross language study of the effects of voiced/voiceless consonants on the vowel voice source characteristics

Abstract

Source characteristics of a vowel may differ according to the voiced/ voiceless nature of adjacent consonants. The postvocalic consonant could be particularly crucial, as vocal fold abduction for a voiceless consonant may be initiated considerably before oral occulsion. Here, CV (:) C utterances (where C = voiced/voiceless labial stop or fricative) were analyzed for female and male speakers of Swedish, English, and French. Characteristics of the voice source were measured from inverse filtered data using the four‐parameter LF model developed at the KTH. Airflow recordings with a Rothenberg mask allowed inferences regarding incomplete vocal fold closure during the vowel and at onset and offsets. Results indicate major effects of a following consonant; the later part of the vowel preceding a voiceless consonant shows a marked drop in excitation strength and a steeper spectral slope, as might be expected when the vocal folds are opening but vibrating. A spectral consequence of this abducting gesture is a widening of the F1 bandwidth and an upward shift in formant frequencies. These effects were much less in evidence in French than in Swedish and English. The preceding consonant had comparatively little effect; full excitation strength is achieved almost immediately.