日本語類別詞の意味構造と体系

原型意味論による分析

抄録

The semantic structures and system of Japanese numeral classifiers are examined, using the framework of prototype semantics (Coleman & Kay 1981, Fillmore 1982). The analysis is based in part on the results of experimental studies of the acceptability of classifier use for various objects. It is observed that many classifiers, including-wa, -hon, -dai and-ken, form a complex prototype structure, defined not only by necessary conditions for appropriate use, but also by “prototype conditions”, which do not have to be satisfied but, if satisfied, makes its referent a representative or prototypical member of the category.
It is also claimed that there is a priority order in the choice of classifiers when more than one classifier is potentially usable. It is pointed out that a more informative classifier in general restricts the use of a less informative classifier and that this is due to a pragmatic principle of informativeness (Grice 1975).
The findings also suggest that classifiers as a whole do not form a neatly structured taxonomy as formerly suggested (e. g. Sanches 1977). Rather, they constitute a loose system, each classifier having its own peculiarities that reflect the nature of its referents, suggesting the inadequacy of a strong form of structural semantics (Fillmore 1983).