Even though it has long been stipulated that Korean has subject as a grammatical category and i(/ga) is the subject marker in Korean sentences, there has been difficulties in identifying the subject in several types of sentences. Some of the types have eun(/neun) or the like instead of i, others doesn't have any NPs which can be evaluated as subject at any rate in their surface, and the others have more than one i-attached NPs. Hence, these have brought an argument that Korean doesn't have subject as a grammatical category. In this paper, I reconsider the notion of subject and those arguments that Korean doesn't have subject, and argue that there does exist subject as a grammatical category in Korean. In any language, the notion of subject is applied to the basic sentences of it according to Keenan(1976) and I regard that 'flow dimension'(Kibrik 2001) should be skimmed off from the discussion of the existence of subject as a grammatical category, because understanding the meaning of sentences including flow characteristics requires understanding the meaning of sentences not including them. Furthermore, in my opinion, the existence of non-lexically-selected i as in multiplenominative constructions is the proof that i as a nominative case marker is developed to subject marker when the condition of referentiality is met. In addition, I show some heuristics of identifying the subjects in Korean.

• 박철우(안양대학교)