A quantitative study of accent change and variation in Tokyo Japanese: Focusing on adjectives and nouns

This dissertation investigates accent variation and change in Tokyo Japanese within the quantitative framework using empirical data on adjectives and nouns. Over 4,000 tokens of various inflectional forms of adjectives with two distinctive accent types and 2,000 tokens of selected nouns in accent variation were elicited from 36 speakers from Tokyo and surrounding areas. The quantitative results indicate that paradigm (analogical) leveling has not yet proceeded extensively in adjectives and the traditional accent patterns are still maintained along with the new accent patterns, resulting in a highly variable state. Multivariate analyses show that adjective accent variation is strongly conditioned by grammatical factors (e.g., inflectional forms, the following grammatical elements, word frequency) and moderately affected by phonetic factors (e.g., the sonority hierarchy of the consonant of the syllable on which the accent falls, stem length, and the vowel devoicing constraint). In contrast, extralinguistic factors played almost no role. No significant social stratification by speaker\u27s age (10s–20s vs. 30s–45), sex, place of birth/residence within Tokyo, or style is evident. The validity of the hypotheses formulated to account for the loss of accent in nouns called heibanka (de-accentuation)—phonological simplification, dialect contact, influence from intonation, segmental effects, grammatical conditioning, semantic differentiation, and familiarity—was evaluated, but only a positive correlation with rendaku (compound voicing) was empirically confirmed. In both adjectives and nouns, undeniable lexical idiosyncrasies in the rate of variation were revealed with remarkable overall regularity across speakers, strongly indicating that the current accent changes are mediated through lexical diffusion. Despite the previous contention that accent membership switch in individual words is sporadic and secondary in importance to change that affects the accent melody itself, it is the major mechanism of accent change currently observed. The results of this study prompt a re-examination of the view which regards regular sound change as the primary agent of accent change