자동 발음열 생성을 위한 한국어 음운현상의 계량적 연구

Abstract

This paper sets out to present a systematic description of phonological phenomena and to analyze the frequency of their occurrence for purposes of automatic Korean pronunciation generation based on phonological knowledge and the analysis of dictionary entries. The dictionaries used for the study were a general Korean dictionary (Yonsei Korean Dictionary) containing about 50,000 entries and a dictionary of particles and endings (Eomi Josa Dictionary) containing 2,200 particles and endings. The research resulted in the identification of 18 phonological phenomena: 13 regular phenomena were observed under phonological and morphonolgical environments; and 5 irregular phenomena showing idiosyncratic characteristics. This will help to provide an understanding of the Korean language from a phonological point of view and to develop an automatic Korean pronunciation generator for TTS and speech recognition.