The Development Of Chinese Listening And Speaking Skill Test For 3～4 Grades

Objective: The Chinese Listening and Speaking Skill Test for 3～4 Grades was preliminarily developed. the items were analyzed, and the reliabilities and validities were tested. The purpose of this study was to provide an effective measurement instrument for the Chinese listening and speaking skill, and for the clinic diagnosis and curriculum reform, anoter purposr is to study the use in the test development of the SEM.Methods: It was based on theoretical and experiential strategies, after drawing forefathers' research production, interviewing with expert and the practice of teaching, we confirmed the structure and content, developed the test. The test, made up of listening and speaking sub-scales, contains six sub-tests of memory,discrimination,thinking,evaluation,retelling and narration. Through testing and electing the items, we've formed A and B which are believed equal. Formal sample consists of 270 pupils (135 in A and 135 in B) from 3 primary schools in Changsha and Changde, including urban, town and countryside. In addition, 17subjects did the same tests after a week, 30 subjects did the parallel tests after one day, 30 subjects did the retest and parallel tests after a week. 88 samples were collected to test validity, and the test was valuated by 5 experts. The items, reliabilities, and validities were testing by the different measurement methods.Rusults: 1. General result: thd variant editions, grades, and genders were not significant difference. The higher grade was, the higher raw score of the test was. 2. Item analysis: Generally, the difficulties of items among 0.20～0.80 amounted to 94%, the difficulties of subtests and subscales were 0.49 to 0.63, the full scale was 0.50 or so; more than 94% items discriminations were above 0.30, the discriminations of the subtest and subscale were 0.30 to 0.63, every subtest and subscale correlate to the whole score was above 0.58. 3. Reliability analysis: The test-retest reliability of subtests was 0.78 to 0.96 and 0.74 to 0.91, the subscales and full scale ranged from 0.91 to 0.96, the full scale was above 0.90. The parallel reliability of full scale was 0.91,the subscales and subtests ranged from 0.75 to 0.99. The retest and parallel reliability ranged from 0.54 to 0.96. The Cronbach's a voefficient from 0.30 to 0.74, the subscales and full scale ranged from 0.71 to 0.81. The scorers reliabilities of speaking scales were above 0.90. The standard error of the subtests,subscales and full scales keep about 2 standard deviation. 4. validity: The content validity was good. There were significant differences in the variant schools and grades. CFA supported the model of the scale. The correlation between the score and abilities rank was 0.42 to 0.85. The correlation between the score and academic test was 0.49 to 0.66, and it proved good predictive validity.Conclusion: 1. The difficulties were appropriate and discriminations were good. 2. A series of reliabilities testing conformed to the measurement standards. 3. The content validities,construction validities and concurrent validities were good. 4. Test A and B were parallel on the whole.