한국어 품사 체계에 관한 비판적 고찰: 의존 명사를 중심으로

Abstract

This paper critically reveiws the currently prevalent parts of speech system of the Korean language and suggests a partial modification. In order to do so, this paper explains that the parts of speech system of a given language play a crucial role in defining syntactic well-formedness of a given word sequence. For that reason, parts of speech system should be established with reference to distributional characteristics. It implies that parts of speech should be dealt with on the syntactic dimension. Traditionally morphological characteristic has been regarded as the most reliable criteria to determine the identification of parts of speech while others argue that functional characteristic should be primarily consulted. This paper shows that morphological criteria and functional criteria can not determine the identification of parts of speech by themselves, but rather that they can work properly only when the distributional information of a linguistic form in question is provided. It should be noted that the distributional information of a given word is not merely a position in a linear sequence of words, but rather a postion in relation to other words that appear in the same local structure. This paper applies the distributional criteria to the noun category of the Korean language and criticizes that the current parts of speech system fails to distinguish the dependent noun category and the classifier category from the noun category. This paper shows that the so-called dependent noun must be preceded by modifying phrases due to its semantic emptyness, and thus can not appear at the beginning of sentences. This paper also points out a strong concordance constraint about the dependent noun that it can not be specified by numerals. To the contrary, words belonging to the classifier category co-occur only with numerals, but not with nouns and relative clauses that can modify dependent nouns. To sum up, the dependent noun category and the classifier category has no insersection at all in distribution. Therefore the two categories in point must be separated from the noun category and from each other as well so that they must be described as independent syntactic categories.