| Abstract | In such Southeast Asian tonal languages as Central Thai, the tone domain is ordinarily seen as the syllable, but some linguists have claimed that a segmental representation of tones best fits the grammar. In such a model, the five-way tonal contrast in Thai would be handled by various arrangements of three-level tones, underlying which are two binary features. Phonetic evidence, either in the form of fundamental f contours or perceptual data, supporting such a view is sought. Criteria are developed & applied to the production of Thai words isolated & embedded in sentences, but fail to provide any direct support for a segmental representation of these tones. Listening tests with controlled variants of fundamental f contours made with a speech synthesizer also fall short. It is concluded that phonological arguments for segmentation are weak, that the phonetic data make such a model implausible, & that the concept is psychologically unconvincing. |