한국어 숙달도 평가 도구로서 유럽공통참조기준 자가진단표의 타당도 검증 및 활용 방안

Validation of CEFR Can-Do Statements as a Self-assessment Instrument of Korean Proficiency

This study demonstrates empirical evidence of the self-assessment battery designed to distinguish between intermediate and advanced learners of Korean with a focus on structural, content, and construct aspect of validity. All 24 statements were selected from the CEFR self-assessment grid and converted into 6-point Likert scale items and administered to 217 university-level Chinese students (Council of Europe, 2001). The item responses were analyzed by the AMOS version 20.0 and the R package version 3.4.3, using a structural equation modeling (SEM) and a generalized partial credit model (GPCM). The confirmatory factor analysis showed that the selected 24 statements constructed four factors (i.e., RC, LC, SP, and WR) and grouped into two meta-criteria (i.e., understanding vs. production skills) which are structurally aligned with the CEFR skill description. The Logit scale of item difficulty indexes showed that the items are sensitive and systematically distributed in relation to the expected difficulty range. The discrimination indexes revealed that only one item showed a low discrimination index below threshold, suggesting that the quality of items is generally acceptable.