This paper examines the distribution of floating numeral classifiers in Korean, including their well-known subject-object asymmetry. We argue that their distribution can be accounted for by combining Bošković’s (2004) generalization that adjunct quantifiers cannot float in θ-positions and Hong’s (1994) generalization that Case-markers are optional to arguments in the complement position but obligatory to arguments in other positions. Even though numeral classifiers are free to be an argument or an adjunct, the complement position is the only position where Caseless numeral classifiers can occur as an argument because it is the only position where Case-markers are optional (Hong’s generalization). In Spec,vP position, only Case-marked numeral classifiers can occur; Caseless ones cannot occur either as an argument or as an adjunct because arguments in this position must have a Case-marker (Hong’s generalization), and because adjuncts cannot occur in θ-positions (Bošković’s generalization). In the other positions, which are non θ -positions, Caseless numeral classifiers can occur only as an adjunct (Bošković’s generalization) and Case-marked ones only as an a (moved) argument (Hong’s generalization). It is also shown that the distinction between argument floating numeral classifiers and adjunct floating numeral classifiers is supported by their different prosodic properties.
1. 서론
2. 격 표지에 대한 일반화
3. 유동 양화사에 대한 일반화
4. 유동 분류사의 운율 구성
5. 결론