토픽 모델링과 언어 네트워크 분석을 활용한

Research Trend Analysis focused on the assessment of Korean Language through Topic Modeling and Network Analysis

This study analyzed research trends in the assessment of Korean language over the past 20 years using 354 pieces of research through LDA-based topic modeling technique and network analysis. A total of 754 keywords were extracted from 354 academic journals published between 2002 and 2022. As the result of the analysis, the keywords were categorized into six topics: assessment items, assessment for academic purpose, scoring methods, classroom based assessment, pragmatic competence, and assessments for other purpose. First, the analysis of the vocabulary frequency revealed that top-frequency words included “items”, “abilities”, “writing”, “speaking”. It indicated that research had been primarily focused on item analysis and descriptions of communicative competence, with an emphasis on writing, and speaking. Secondly, analysis of co-occurring words found that “academic purpose” was the most frequent, followed by “domain-items”, “communication-competence” and “reading-items”. The research before 2014 was mainly conducted on the categories of “communicative competence”, “items”, and “culture”, while the research after 2014 has gradually expanded into the categories of “academic purpose”, “scoring”, and “validity”. Lastly, a paradigm shift towards a learner-centered approach led to a focus on alternative assessment methods within the classroom in the research conducted after 2020. Furthermore, centrality analysis showed that the top five words with high degree centrality and betweenness centrality, indicating their importance as key concepts in the research.