한국어 말하기 평가에서의 채점 편향성 분석

Analysis of Scoring Bias in Korean Speaking Test

The purpose of this study is to examine the grading tendency of raters’ consistency, severity and bias in IBT-based Korean speaking performance test. For this purpose, 82 Korean language learners were assessed and 6 Korean teachers participated in the scoring. The Facets program which developed based on the multi -facets Rasch model was used to identify the problems of scoring. The results of the anlaysis are summarized as follows: ⑴ The reliability among the raters was high, and the reliability in the raters was found to be overly consistent in one rater. ⑵ In the rater bias analysis, statistically significant bias was found in all the raters, mainly in the grade 4 or higher candidates and the middle and high grade evaluation items. ⑶ As the test taker’s proficiency increased and the item difficulty increased, many situations requiring subjective judgment of the scorer occurred, and it was found that the rater’s bias appeared in such situation. Consequently, the implication of this study was proposed.