Vowel and nasal duration as cues to voicing in word-final stop consonants: spectrographic and perceptual studies

To determine durational differences between vowel and nasal segments preceding word-final /t/ and /d/, spectrograms were made of adult speakers' productions of minimal pairs of the type /pent/-/pend/. Vowel, nasal, and vowel plus nasal (vocalic nucleus) durations were greater before /d/ than before /t/. Assuming the voiceless context as a base, the increase in nasal duration in the voiced case was proportionately greater than the increase in vowel duration. This outcome suggests that nasal duration is a more powerful cue to the voicing characteristic of the following consonant than is vowel duration. To test this, adult listeners were presented synthetic CVNC utterances in which the nasal and vowel segments were independently varied in duration over a range of 40 msec to 200 msec and were instructed to label the final stop consonant as either voiced /d/ or voiceless /t/. Although changes in both vowel and nasal duration were sufficient to cue both voiced and voiceless judgements, listeners' categorization of final consonants shifted more rapidly as a result of varying nasal rather than vowel duration. Nasal duration, therefore, appears to be a stronger cue than vowel duration for the word-final voiced-voiceless consonant distinction in CVNC utterances.