한국어교육에서의 병렬 말뭉치 활용 연구 동향 분석 = Analysis of Research Trends in the Use of Parallel Corpora Within Korean Language Education

국문 초록 (Abstract)

민유미·이정희, 2026, 한국어교육에서의 병렬 말뭉치 활용 연구 동향 분석, 어문연구, 209 : 273~304 본 연구는 1992~2025년 국내 한국어 병렬 말뭉치 활용 연구 895편과 한국어교육 연구 141편을 대상으로 키워드 네트워크 분석, LLR 검정을 통해 연구 동향을 분석하였다. 병렬 말뭉치 활용 연구는 〈21세기 세종계획〉과〈모두의말뭉치〉를 전환점으로 성장하였으며 2020년대 한국어교육 연구가 26%로 폭발적 증가하였다. 한국어교육은 대조 분석 중심의 독자적 군집을 형성하였고 LLR 분석에서 중국어·영어·일본어 외 베트남어 등 저자원 언어 연구가 주목받았고 ‘오류’가 학습자 연구와 교육 응용을 연결하는 핵심 키워드였다. 한국어교육 연구에서 저자원 언어는 대조 분석에만 머물러 교육 응용 확장이 미흡함을 확인하였다. 이러한 결과를 바탕으로 본 연구는 향후 병렬 말뭉치 연구가 형태·통사 중심에서 확장되고 저자원 언어의 교육적 응용 연구 및 관련 분석 인프라 구축이 강화되어야 함을 제언하였다.

다국어 초록 (Multilingual Abstract)

This study analyzes research trends in the utilization of Korean parallel corpora by examining 895 studies published in South Korea between 1992 and 2025, with a focus on identifying the characteristics of 141 studies within the field of Korean language education. Data were collected from the Research Information Sharing Service, and research patterns were examined using keyword network analysis and Log-Likelihood Ratio test. The findings indicate a continuous growth in domestic parallel corpus research, with the 21st Century Sejong Project (2007) and the Modu Corpus (2021) as primary catalysts. Notably, research in Korean language education surged in the 2020s, accounting for approximately 26% of overall research volume. Keyword network analysis revealed that Korean language education studies formed an independent cluster centered around contrastive analysis, distinct from translation studies and machine translation research. Furthermore, LLR analysis identified distinctive keywords specific to Korean language education, highlighting active research on low-resource languages—such as Vietnamese, Indonesian, and Thai—alongside traditionally studied languages such as Chinese, English, and Japanese. Thematic concentrations primarily involved the contrastive analysis of grammatical categories, notably causative and passive constructions. Betweenness centrality analysis confirmed that the keyword “error” acts as a crucial node connecting learner-focused research with educational applications. However, studies on low-resource languages have predominantly remained at the structural contrastive analysis stage and have not expanded into practical teaching and learning applications. Based on these findings, this study suggests that future parallel corpus research should expand its scope from morpho-syntactic features to discourse-level analysis and it highlights the need to strengthen applied educational research for low-resource languages.