Tones in whispered Cantonese

Abstract

Acoustic analysis and perceptual experiments were carried out to investigate the acoustical characteristics of tones in whispered Cantonese and to identify possible perceptual cues for tone identification. The isolated vowel /a/ embedded in a framing sentence produced by 20 (10 male and 10 female) native Cantonese speakers using modal and whispered phonation was recorded. Formant frequencies, duration and intensity of the vowels were measured from the samples using signal analysis software. During tone identification tasks, the speech samples were presented to 20 listeners who were native Cantonese speakers. The listeners were instructed to identify the tone of the target vowels in the presented sentences, based on which percent correct identification of tones was calculated. Results of the study reveal the role of second formant, duration, average intensity and intensity contours in perception of Cantonese whispered tones. Speaker’s maneuvers in production of whispered tones were also discussed.