台灣大學指定考試英語科測驗克漏字在測量甚麼能力?

What do the Cloze Items in the English Subtest of the Advanced Subjects Test in Taiwan Measure?

本研究主要目的乃是探討台灣大學指定考試英語科測驗中克漏字題目的構念效度。本研究向大學考試入學中心申請2015及2016年隨機抽樣各5,000位考生每題克漏字答題的得分資料，並邀請五位專家們根據Purpura (2004)理論架構將2015及2016年克漏字題目進行題目分類，然後藉由驗證性因素分析方法檢此分類架構與考生的答題資料間是否適配，並期能更進一步找出實徵資料的最適配模型。驗證性因素分析結果顯示，專家的題目分配模型與實徵資料的適配不是最佳的，最佳的適配度模型反而是單一整體特質模型。也就是說，本研究結果顯示，大學指定考試英語科測驗克漏字題目乃是在測量整體英語閱讀能力。最後，本研究也根據研究結果，針對高中英語教師編寫克漏字試題相關人員，提出具體建議。

 

In each July, high school seniors in Taiwan sit for the Advanced Subjects Test (AST), of which an English subtest (AST-ES) is an integral part. Their scores on the AST-ES affect significantly which universities they are qualified to be admitted to. A question of their great concern is: what English ability does the AST-ES truly measure? This study was aimed to address this question through studying the cloze items of the AST-ES. This study adopted a simpler version of Purpura’s (2004) model in which four components of grammatical knowledge were used to classify the AST-ES cloze items. Classification was performed by five experienced tertiary- level English teachers. Then a series of confirmatory factor analyses (CFAs) were applied to determine the degree of fit between their classifications and test-takers’ responses. The findings showed that their classifications did not fit the test-takers’ responses and that a one-component model best described the data. The finding of a one-component model was consistent with the unitary competence hypothesis of Oller (1976, 1979), who advocated using cloze tests for measuring overall language ability. This study concluded with some implications and limitations of the findings.