類型論から見た文法理論

This paper discusses the possibility of making a set of descriptive models for languages according to morphological typology.<br>Two word properties L (lexcical or semantic) and G (grammatical ro syntactic) are postulated in order to characterize three traditional morphological types of languages. An inflectional language has words with both {L, G}, whereas an agglutinating one, either {L} or {G}, and isolating one, only {L}.<br>Further, each language type is re-defined respectively as 'definite category language', 'infinite category language' and 'non category language', according to the the number of variables G, that of the value M for each variable G.<br>Definite category languages like most of Indo-Europian languages including English have a constant number of variables G. The number of values of each Gi is also fixed.<br>The constancy of these numbers and the existence of G within a word domain of the languages give paradigmatic characteristics to their words. On the other hand, indefiniteness of these numbers or the absence of G within a word domain give the rest of the two types syntagmatic characteristics.<br>The author claims that linguistic theories are based mostly on the knowledge of definite category languages, which would cause difficulties when they are applied to the other types of languages.<br>For example, the existence of only one definite verb in a sentence is unique to the definite category languages, though most linguistic theories regard this as one of language universals. Applying this principle to an indefinite category language like Japanese would require multiple embedding sentences, thus, it would cause unnecessary complexity in syntactic description.