It is generally believed that all languages, no matter how tonal, have grammatical intonation of some kind. A fall in pitch at the end of ordinary declarative sentences is thought to be universal. Hong Kong Cantonese has six contrastive tones, two rising and four level, according to the impressionistic account given here. The experimental results presented here indicate that sentence-final tone lowering does occur in Cantonese while contrastive stress apparently does not have a significant effect on pitch. In addition, it appears that although intelligibility is somewhat worse sentence-finally than sentence-medially, the tonal distinctions are not neutralized in final position.

No abstract

No abstract

The experiment reported here is a perceptual study of the six contrastive tones of Cantonese. The monosyllable [jλu] was synthesized, and a large number of closely spaced Fo contours were applied to it. Listeners were asked to identify each synthetic stimulus as one of six Cantonese words which all have the segmental shape [jλu] and differ only in tone. The response data were used to compare two impressionistic accounts of the tonal distinctions and determine which reflects actual Fo more accurately. In general, the results accord fairly well with straightforward predictions made from the impressionistic accounts, but there is one major exception. It appears that something in addition to differences in Fo is involved in distinguishing the low-low tone from the other tones.

The Japanese consonant alternation known as rendaku is a nonautomatic process whereby morphemes with an initial voiceless obstruent in isolation sometimes occur with an initial voiced obstruent as the second element of compounds or stem-affix formations. A putative phonological constraint known as Lyman's Law states that the initial consonant of a morpheme never undergoes rendaku if that morpheme already contains a voiced obstruent. There are only a handful of exceptions to this constraint in the existing vocabulary, but the experimental results presented here indicate that it is psychologically real only for a small minority of speakers. Since Lyman's Law appears to have been a part of Japanese for centuries, these results raise the problem of accounting for its persistence.