Production and perception of the Cantonese tones
Student thesis: Master's Thesis
This study investigates the production and perception of the six Cantonese long tones, namely /55 33 22 21 25 23/, by native Cantonese speakers in Hong Kong. It explores the relationship and interaction between production and perception of the tones in Cantonese. The obtained data are also used to determine the tonal representations and confusion patterns of the Cantonese tones at the present time. Three minimal sets of meaningful monosyllabic words consisting of the syllables /si/, /sɛ/, and /fu/ associated with the six Cantonese long tones were used for investigation. A group of college students, ten male and ten female, who are native Cantonese speakers in Hong Kong, produced the speech samples. The speakers also took part in a perception test (Perception Test 1) to identify the six Cantonese long tones which are clearly distinct in natural speech of a native female Cantonese speaker. Another perception test (Perception Test 2), also with ten male and ten female subjects, was carried out to identify the different variant forms of the six Cantonese long tones produced by a male and a female speaker in the present study. The production results show that the F0 contour of each of the six Cantonese long tones has many phonetic realizations. In some cases, the tones may merge to have the same phonetic realization. The merge is especially pronounced between the /33/ and /22/ tones. The merge between the /25/ and /23/ tones as reported in many previous studies is not apparent in the present study. In general, the /55/ tone is frequently realized as , /33/ as  or , /22/ as  or , /21/ as , /25/ as  or , and /23/ as , , or . The data demonstrate that the six tones, except for the /33/ and /22/ tones, are still distinct from each other in production, although their phonetic realizations are quite different from the tonal representations used in the previous studies. As for perception, in Perception Test 1, all the six tones have a high correct identification rate over 80%. Perceptual confusion is mainly observed between the /33/ and /22/ tones and between the /25/ and /23/ tones. The former corresponds to the merge of the two tones in production, but the latter does not. In Perception Test 2, the correct identification rate remains high for the /55/ and /21/ tones, but it is reduced for the /33 22 25 23/ tones, compared with that in Perception Test 1. This is due to the fact that the /33 22 25 23/ tones used for Perception Test 2 may have the same phonetic realization as that of the other tones. In some cases, while the tones, such as the /33/ and /22/ tones, have the same phonetic realization, they are still correctly identified in over 50% of the cases. The result suggests that some acoustic features other than F0 may also be involved in perception of the Cantonese tones. The production and perception data in the present study show that there is a close relationship between production and perception of the Cantonese tones, but the performance in production and perception is not symmetric. Perception is based on the phonetic realizations of the tones. The tones which are clearly distinct in speech have a high correct identification rate in perception. However, similarity of the tones in production may not necessarily give rise to perceptual confusion. In addition, subjects who tend to merge the tones in speech may not have difficulties in identifying those tones in perception. This indicates that the subjects’ performance in tone perception is better than in tone production.
- Cantonese dialects, Tone, Pronunciation, Speech perception