초록 열기/닫기 버튼
At present, international Chinese teaching has entered a new stage of connotative development. Improving the reliability and validity of Chinese proficiency test in listening, speaking, reading, and writing is one of the issues worthy of further discussion. This study investigates the impact of task types on writing score errors, and explores the most suitable task types and number of scorers for evaluating foreign students' Chinese writing ability. By using the analysis method of Generalizability Theory (GT) model and SPSS 22.0 UNINOVA program, the research group completed the G study. The results show that the writing ability of the subjects is the main factor affecting the writing score, and the scoring error caused by the scorer factor is greater than the task type. The D study shows that argumentative writing test is easier to achieve higher reliability, while narrative test has the lowest reliability. In addition, when the number of raters was 3, the generalizability coefficients of these 4 task types reached the ideal level of 0.7. In the end, this paper points out that argumentation should be the first choice for writing tasks of Chinese proficiency test above the intermediate level, and raters should be at least 3 in a group. Meanwhile, effective rater training should be carried out.
