한·중·조 뉴스 코퍼스 기반 어휘 결합 선호도 분석 - '인민'이 포함된 어절을 중심으로 -

Abstract

This work aims to analyze and compare the usage and distributional differences of the word "people" across Korean texts from China, North Korea, and South Korea. The analysis is conducted by constructing a specialized corpus and applying correspondence analysis. A Korean news corpus comprising 3 million "eojeol" is built, incorporating articles from China’s Yanbian Daily, North Korea’s Rodong Sinmun, and South Korea’s JoongAng Ilbo. From this corpus, all "eojeol" units containing the word “people” were extracted for analysis. The frequency and mutual information (MI) values of collocating morpheme types were calculated, and correspondence analysis maps were generated based on these data. The results reveal significant cross-national variation in the usage patterns of the word “people.” This not only demonstrates the effectiveness of correspondence analysis in exploring intra-language usage tendencies in Korean, but also offers theoretical implications for language education and automated text classification.