Voice Onset Time Characteristics of Esophageal, Tracheoesophageal, and Laryngeal Speech of Cantonese

Abstract

Purpose To investigate the ability of standard esophageal (SE) and tracheoesophageal (TE) speakers of Cantonese to differentiate between aspirated and unaspirated stops produced at 3 places of articulation were investigated. Method Six Cantonese stops, /p, t, k, p h , t h , k h /, followed by the vowel /a/ produced by 10 SE, TE, and normal laryngeal (NL) speakers were examined through perceptual tasks and voice onset time (VOT) analysis. Results Perceptual experiments showed lower accuracy of identification of stops produced by SE and TE than by NL speakers, with mostly misidentification of aspirated stops as their unaspirated counterparts. Acoustic analysis revealed that aspirated stops produced by NL, SE, and TE speakers were associated with significantly longer VOT values than their unaspirated counterparts. Unaspirated velar stops showed significantly longer VOT than bilabial and alveolar stops in NL and SE speech. SE and TE speakers were still able to use VOT to signal aspiration contrast, but TE was unable to differentiate among different places of articulation.