クローズ法による日本語能力の測定
スポンサーリンク
概要
- 論文の詳細を見る
I. BACKGROUND A proficiency test which can measure the overall Japanese language ability has not yet been developed with regard to Japanese language education as a foreign language. The cloze test (or cloze procedure) , a highly regarded procedure developed by W. Taylor (1953), originally measured sentence readability. In this test, words are systematically (usually every n th word) deleted, and subjects are required to fill in the blanks with the words which they think are most appropriate. In contrast to the discrete-point test which measures elements of language individually, the cloze test as a foreign proficiency test can be said to be an integrative test which attempts to measure the subject's integrated language ability. Validity, reliability, efficiency and practicality can be regarded as conditions for a good test and from many supportive researches, the English cloze test is proven to be a test with high validity, reliability, and practicality. As a step in Japanese language education toward the development of a Japanese language proficiency test for foreigners, the auther would like to investigate the suitability of the Japanese language cloze test as a proficiency test for Japanese as a foreign language. The investigation will be achieved through the appliciation of the cloze test in the Japanese language. II. PURPOSE The primary purpose of this research is to see whether or not the Japanese language cloze test, with high validity and reliability as a testing device for measuring English language proficiency, is a suitable test which can measure the Japanese language proficiency of people whose native language is not Japanese. The secondary purpose is to see the relationship between the results obtained from the cloze test and the subject's aural comprehension, writing ability, speaking ability, and reading comprehension in the Japanese language. HYPOTHESIS 1: The cloze test is a test which has high validity as well as reliability as a Japanese language proficiency test for foreigners. HYPOTHESIS 2-1: The cloze test relates to the aural comprehension of foreigners in the Japanese language. HYPOTHESIS 2-2: The cloze test relates to the writing ability of foreigners in the Japanese language. HYPOTHESIS 2-3: The cloze test relates to the speaking ability of foreigners in the Japanese language. HYPOTHESIS 2-4: The cloze test relates to the reading comprehension of foreigners in the Japanese language. III. METHOD In November, 1980, 1) cloze test (also as the external criteria), 2) aural comprehension test, 3) sentence construction test, 4) vocabulary test, 5) reading comprehension test, 6) composition test, and 7) speaking ability test were carried out. The Exact Word Scoring Method was used in scoring the cloze test. In order to evaluate the reliability of the cloze test, item analysis of the cloze test was used, and 1) difficulty, 2) critical values of the correlation coefficient between the total score and each item were determined, and 3) x^2-test was employed. Also, in order to examine the internal consistency of the cloze test, Cronbach's α was determined. After eliminating the items with difficulty above 90% and below 10% as well as items with low-discriminating ability (critical values of the correlation coefficient of r<.306) from all of the other test that were carried out, the Cronbach'α was computed. In order to evaluate the criterion-related validity, the Pearson Product-Moment Coefficient Correlations was also computed. Furthermore, in order to evaluate the relationship between the cloze tests and 1) aural comprehension, 2) writing ability, 3) speaking ability, and 4) reading comprehension, the Pearson Product-Moment Coefficient Correlations were determined. Multiple Regression and Factor Analysis were carried out as well. IV. RESULTS AND DISCUSSION From the results of the cloze test's item analysis, the items in which the critical values of the correlation coefficient between the total score and individual items were significant were fifteen items out of thirty. The cloze test is made up of four clauses, but in the fourth clause concentration is on the significant items of the critical values of the correlation coefficient between the total score and individual items. Thus, the fourth clause is thought of as having higher reliability than the other clauses. The reliability of the three cloze tests was determined by referring to the results of the item analysis. CLOZE TEST 1: The score of all items had an α=.807. CLOZE TEST 2: The score of the fourth clause section had an α=.678. CLOZE TEST 3: From the results of item analysis of cloze test 1, the score of the statistically significant item had an α=.835. Moreover, in order to investigate the validity of cloze tests 1, 2, and 3, the Pearson Product-Moment Correlations were determined from the fifteen variables. As a result, cloze test 1, 2, and 3 had significantly high correlation with the other tests which were carried out, thus indicating that this is a test with high criterion-related validity. Therefore, it can be said that Hypothesis 1 has been supported. In order to verify Hypothesis 2-1, by determining the correlation coefficient between cloze test 1, 2, and 3 and aural comprehension, a definitely significant correlation from r=.53 to r=.73 (p<.001) was found. Therefore, it can be said that Hypothesis 2-1 has been supported. In order to verify Hypothesis 2-2, by determining the correlation coefficient between cloze test 1, 2, and 3 and writing ability, a definitely significant correlation from r=.71 to r=.76 (p<.001) was found. Therefore, it can be said that Hypothesis 2-2 has been supported. In order to verify Hypothesis 2-3, by determining the correlation coefficient between cloze test 1, 2, and 3 and speaking ability, a definitely significant correlation from r=.55 to r=.70 (p<.001) was found. Therefore, it can be said that Hypothesis 2-3 has been supported. In order to verify Hypothesis 2-4, by determining the correlation coefficient between cloze test 1, 2, and 3 and writing ability, a definitely significant correlation from r=.44 to r=.51 (p<.01) was found. Therefore, it can be said that Hypothesis 2-4 have been supported. Hypothesis 1, 2-1, 2-2, 2-3, and 2-4 have been supported. Thus, according to this research the Japanese language cloze test can be said to be a suitable integrative test which can measure the proficiency in Japanese language as a foreign language.
- 国際基督教大学の論文