Passive vocal tract enlargement during voiced stops

Voiced stops should become devoiced within 5 to 10 m of stop closure if there is no vocal tract enlargement to delay the inevitable reduction in transglottal pressure drop. Since stops may be voiced longer than this, some cavity expansion must take place, either passive through tissue compliance or active through larynx lowering and the like. To estimate the duration of voicing in stops when only passive enlargement occurs, subjects produced isolated nonsense words of the form VC:V, where V was one of a variety of English vowels and C: was an artificially prolonged /b/, /d/, or/g/. Oral air pressure was vented through a catheter leading from the pharynx to the atmosphere via the nasal cavity. At unpredictable times, a solenoid-activated valve closed the catheter and the consequent build-up of oral pressure extinguished the voicing. Voicing continued after this closure longer during /b/ than /d/ or /g/, and longer when coarticulated with high vowels than with low vowels. These results can most plausibly be explained by reference to differences in compliance of the surfaces on which oral pressure impinges during the various consonants and vowels. [Supported by NSF and NIH.]