ヨルバ語の母音削除とトーン変化の規則について

Abstract

Verbs in Yoruba, or other elements such as prepositions, conjunctions or prefixes, have a CV endings, while nouns usually have the structure VCV. So when a noun follows a verb or the other element, there will occur a sequence of vowels. Especially in a ordinal conversation, it is common that one of the vowels in such a sequence is deleted. (It is obligatory in affixation.) Although it is not simple to tell which vowel in such a sequence will be deleted, the tonal configuration that results after the vowel deletion is quite predictable. In Pulleyblank(1986), he proposes some rules which derive the tonal results of the vowel deletion from the underlying tonal pattern. His main suggestion is that M tone is underlyingly unspecified, that means surface M-toned vowels are toneless at the underlying representaion and receive their tone by a default rule. This analysis, based on the Underspecification Theory, accounts completely for the behaviour of M tone under vowel deletion when one of the abutting vowels is M-toned. But when it comes to a case where those abutting vowels are H-toned and L-toned, his explanation for the tonal derivation is not satisfactory. In this paper, I reconsider 'H/L-spread' rules presented by Pulleyblank(1986:110-112) and show a new order of rules for the tonal derivation under vowel deletion. 'H/L-spread' is a rule to derive a falling/rising tone. 'H-spread' is also applied to link a H tone, which is set afloat after vowel deletion, to a L-toned vowel. Since these spreading rules are applied at the surface level, they should be ordered after assigning of M tones. This is what I demonstrate by giving evidence, and I also show a merit of the new order of the rules.