The contribution of CV transition duration to the perception of final‐consonant voicing in natural speech

The experiments reported here were attempts to replicate the results of previous research which had employed synthetic stimuli [Raphael et al., J. Acoust. Soc. Am. Suppl. 1 57, S49(A) (1975), and 58, S57(A) (1975)]. A speaker recorded a single token of the syllable [sed]. Under computer control the syllable was edited to produce three continua: [sed] to [set], created by deleting pitch pulses from the vowel; [ded] to [det], created by deleting the [s] friction from the [sed‐set] continuum; and [ed] to [et], created by deleting the initial transitions from the [ded‐det] continuum and reiterating pitch pulses from the vowel to compensate for the duration of the deleted transitions. For each continuum listeners were asked to identify the final consonants as voiced or voiceless. The [d]‐[t] phoneme boundaries, plotted as a function of vocalic duration, fell within 3 ms of each other, indicating (as did the earlier experiments) that initial CV transitions contribute equally with “steady‐state” formants to the ...