한국어의 정형화된 표현 연구

Abstract

This corpus-driven study focuses on a role of formulaic expressions. By using computational technologies, we have provided empirical evidence for the most common recurrent sequences and their discourse functions in a given register. A few studies argue that compared to English, Korean has few recurrent sequences of words. However, those studies are based on orthographic words and do not represent morphological characteristics of Korean as a agglutinative language, where a string of affixes are attached to a root. Therefore using a morpheme-based n-gram analysis, the present study examines the distribution and the use of formulaic sequences in three registers. We compare the formulaic expressions in written academic text, spoken academic text, and conversation from the 21st Sejong Corpus. Structural patterns are described first, and then we present a functional taxonomy. The results here show that formulaic expressions are frequent in Korean written academic text. And, structurally, VP-based formulaic expressions are more common than NP-based expressions in all 3 registers and by far the most common expressions have ‘su(수), ges/ge(것/거)’. Functionally, stance expressions are most frequent than discourse organizers and referential expressions. Stance expressions are much more prevalent in academic registers than in conversation.