An autosegmental study of Japanese pitch-accent systems

Abstract

According to early autosegmental treatments of pitch-accent languages, such as Goldsmith (1976), Haraguchi (1977, 1978, 1979), each pitch-accent language has a finite number of Basic Tone Melodies, and tones are mapped to tone bearing units with regard to the universal convention known as the Well Formedness Condition. Rather than the concept of Basic Tone Melody, the present paper, basically continuing on from the idea given by Pulleyblank (1983), suggests that each tone is introduced to the tonal tier, one by one, by tonal rules. The correlative assumption of this approach is that tones are not automatically associated to tone bearing units by the Well Formedness Condition, but mapped to them only if specified by a particular association rule.
The ultimate purpose of this study is to develop a type of approach to Japanese pitch-accent systems that incorporates recent developments in the autosegmental theory of other tonal and pitch-accent languages. Based on the notions such as floating tone, extratonality, etc., this paper presents a theory of tonal rules, which aims to explain all the varieties of Japanese pitch-accent systems, whereby all predictable information is not included by underlying phonological entries, but rather is ultimately supplied by a finite number of parameters.