A two-accent model of Japanese word prosody ∗

要旨

In generative phonology, it has been commonly assumed since McCawley (1968) that a Japanese word has either one accent or no accent (the ‘one-accent model’). Some problems for the model are discussed, both cross-linguistically and within Japanese itself. To solve the problems, I propose an alternative model, according to which a Japanese word has either one or two accents (the ‘two-accent model’). The new proposal yields a more consistent treatment of metrical prominence and of the relation between stress and tone in different languages, such as English, Chinese, and Japanese.

主題