The noncategorical perception of tone categories in Thai.
| Number | 279 |
|---|---|
| Year | 1979 |
| Drawer | 5 |
| Entry Date | 06/03/1999 |
| Authors | Abramson, A. S. |
| Contact | |
| Publication | In B. Lindblom & S. Ohman (Eds.), Frontiers of speech communication research. London: Academic Press. |
| url | http://www.haskins.yale.edu/Reprints/HL0279.pdf |
| Abstract | [Introduction] Theories of the categorical perception of speech seek to explain how levels of acuity of discrimination vary with the type of phonetic segment involved. The now classical paradigm requires the preparation of a continuum of variants along some physical dimension susceptible to auditory division into phoneme categories. Zones of ambiguity between the groups of variants labeled as phonemes are the so-called phoneme boundaries. In the ideal case of categorical perception, subjects will not do much better than chance in discriminating variants within the labeled categories, but will show rather high peaks of discrimination in the regions of the phoneme boundaries. This is best shown in experiments with stop consonants (Liberman et al. , 1957). A rather different set of results prevails with steady-state isolated vowels (Fry et al., 1962). Variants are indeed grouped into phoneme categories but with somewhat more overlap than for stops; however, discrimination is very good along the whole continuum without any special effects at the category boundaries. Two ways of explaining these differences have dominated the literature. The motor or articulatory-reference theory (Liberman etal., 1967) points out that stop consonants and presumably certain other types of phonetic segments are essentially discontinuous in their mode of production, while vowels are continuously graded in production. Thus, it is claimed that the extent of categoricalness in perception is shaped at some psychological level by the intervention of these radical differences in production. The other theory invokes short-term memory. Consonants, particularly stop sounds, have rapidly changing spectra, while vowels, certainly steady-state vowels, tend to be rather long in duration providing for more time to process the stimuli with less strain on memory (Fujisaki and Kawashima, 1969). Arguments continue over the categorical perception of phonetic segments versus continuous perception. There is still a need for experiments on types of phonetic segments not previously examined and types of segments that have been insufficiently examined. For example, experiments with voice onset time (Abramson and Lisker, 1970) yield data very similar in categoricalness to those of data with stop consonants differing in place of articulation, while experiments with distinctive vowel length in Thai (Bastian and Abramson, 1962) yield data very similar to those for steady-state vowels. In an earlier study on the identification and discrimination of phonemic tones in Thai (Abramson, 1961), a set of five fundamental-frequency contours with final points moving upward incrementally from a level base all with the same short drop at the end, was divided perceptually by native speakers into the mid and high tones. In ABX tests, the overall level of discrimination was very high with no convincing evidence of a discrimination peak at the category boundary, thus confirming a prediction of continuous perception. On the other hand, in a recent paper, Chan et al. (1975) presented contradictory results showing boundary effects for a fundamental-frequency continuum that yielded two of the four tones of Mandarin Chinese, namely the rising and level tones. The present study is an attempt to explore the matter further. |
| Notes |
Search Publications