Syntagmatic Relations in Corpus and Natural Language Processing: Focusing on Korean Particles

This paper presents the syntagmatic relations of Korean particles in spoken corpus and develops the grammatical procedure of the feature computation which can generate those relations. The syntagmatic relations of Korean particles are summarized as follows : (A) Lexical case particle ⇒ Lexical case particle, (B) Lexical case particle ⇒ Delimiter, (C) Lexical case particle ⇒ Structural case particle, (D) Lexical case particle ⇔ Conjunctive particle, (E) Conjunctive particle ⇒ Structural case particle, (F) Conjunctive particle ⇒ Delimiter, and (G) Delimiter ⇒ Structural case particle. These seven syntagmatic relations are captured via the operation of the feature computation. Some syntagmatic relations among these are simulated in the field of the natural language processing using the feature computation. Furthermore in the theoretical point of view this paper propose non-template approach to Korean inflection.