문법관계와 체언의 유표성 - 구문분석 말뭉치를 이용한 연구 -

Abstract

This paper describes quantitative patterns of Korean nominals(pronouns, proper nouns and nouns) based on grammatical relations, as found in the 21 Century Sejong Treebank. The distribution of nominals is estimated by correlation coefficients derived from statistical method χ2. This is used to systematically explore markedness of known or unknown linguistic features. Prominence scales of grammatical relations, semantic roles, and old/new information, known as universal markedness in previous works, are similarly applied to Korean nominals. An interesting observation is that markedness of demonstrative pronouns is concerned with the relative spatial locations, and that proper nouns and nouns presented as new information have an tendency to prefer different grammatical relations.