An acoustic analysis of Cantonese vowels and tones produced by hearing-impaired speakers

Abstract

This study investigates the acoustic characteristics of Cantonese vowels and tones produced by speakers with hearing impairment, in order to determine the deficiency in their pronunciation. Speech samples of all the 11 vowels [i y ɛ oe a ɔ u ɪ ɵ ɐ ʊ] and 9 citation tones [55 33 22 21 25 23 5 3 2] in Cantonese were elicited from six hearing-impaired (HI) and two normal-hearing (NH) Cantonese speakers who were in their 20s. The six HI speakers included three males and three females with severe to profound hearing loss, who have received cochlear implants or are wearing a hearing aid. The formant frequencies (F1F2) of the vowels, fundamental frequency (F0) of the tones, and durations of the vowels and tones were measured and compared between the two groups of speakers. The main findings are presented as follows. (a) Duration. The HI speakers make sufficient duration contrast between the long and short types of vowels and tones in Cantonese as the NH speakers do, while vowel prolongation is observed in the speech of some HI speakers. (b) Formant frequencies. Compared to the NH speakers, some HI speakers tend to have a reduced size of vowel space, due to centralization of the peripheral vowels, in particular the backward shifting of the front vowels. However, no significant effect on the differentiation of vowels in the reduced vowel space is observed. (c) Fundamental frequency. The HI speakers show deficiency in tone production, with a major problem in the flattening of the F0 contours of the contour tones, which leads to the merging with the level tones. Overall, the HI speakers are more successful in producing the vowels than the tones, which may suggest a difference in articulatory mechanism between tone production and vowel production.