Realization of the English postvocalic [voice] contrast in F1 and F2

Imperceptible incomplete neutralization: Production, non-identifiability, and non-discriminability in American English flapping

“…The [voice] feature has a number of known acoustic correlates (Lisker, 1986), including preceding vowel duration (Chen, 1970), closure duration (Kluender et al, 1988), effects on the F0 and F1 of surrounding vowels (Hombert et al, 1979;Kingston and Diehl, 1994), and F1-F2 divergence in the offglide of closing (i.e., upgliding) diphthongs (Thomas, 2000;Moreton, 2004; before /t/-flaps specifically: Kwong and Stevens, 1999). First, the duration of vowels preceding the flapped segments was measured from the onset of voicing to the onset of the flap closure.…”

“…While preceding vowel duration tends to be the focus of studies of incomplete neutralization, it is far from the only known correlate of voicing (Hombert et al, 1979;Kluender et al, 1988;Kingston and Diehl, 1994;Kwong and Stevens, 1999;Thomas, 2000;Moreton, 2004). As such, the following measurements were also taken: closure duration, percent voicing during closure, and the slope of F0 and F1 at 10 ms preceding the onset of flap closure and 10 ms following the offset of flap closure.…”

Differences in coda voicing trigger changes in gestural timing: A test case from the American English diphthong /aɪ/

“…But data from coda voicing environments contradict it. As Moreton (2004: 2) summarizes, previous production studies have shown that low monophthongal vowels such as [æ] and [ɑ] (i.e., vowels associated with high F1 values) have even higher, not lower, F1 values before voiceless compared to voiced codas, even though duration is shorter. Likewise, high diphthongal offglides, such as that found in /aɪ/ (i.e., offglides associated with low F1 values), have even lower, not higher, F1 values in this environment.…”

“…Another change concerns spectral quality. Vowels exhibit more peripheral formant values before voiceless codas, compared to more centralized values before voiced codas (Moreton 2004 and references cited therein). The puzzle is that these two changes appear disconnected from one another, because when vowels shorten in most other conditioning environments, we typically see centralization, not peripheralization.…”

“…For high monopthongal vowels, there is a curious gap in the speech production literature (a few acoustic measurements are reported for the stimuli used in perceptual studies, but these are difficult to interpret systematically: Fischer & Ohde 1990, Hillenbrand et al 1984, Parker & Walsh 1981, Walsh & Parker 1984). We follow the spirit of Moreton (2004) in assuming that the relevant generalization most likely concerns low and high vowel qualities, rather than monophthongal versus diphthongal status per se. If this is the case, the generalization is that formants for both high and low vowel qualities peripheralize before voiceless codas, and centralize before voiced codas.…”