本研究旨在探討高職生於作答理性選擇型克漏字測驗(mu1tiple-choice rational cloze)時，所運用閱讀策略程度與其作答表現之關係，以及藉由改善閱讀策略知識與運用而提昇其克漏字測驗作答表現之可行性。 克漏字測驗發展半世紀以來，對於是否能做為閱讀理解的適當測驗工具以及改善閱讀理解的教學技巧，國內外文獻迭有正反兩方論辯，贊成者稱克漏字可測出句間文意整體理解(intersentential sensitivity)，反對者謂只能測出句內的局部句法理解(syntactic informaton)。 近年來認知策略(cognitive strategies)與後設認知策略(metacognitive strategieS)成為閱讀相關研究重要課題，閱讀策略改善訓練(Strategic reading intevention)被認為可提升學生之閱讀理解，本研究著眼於觀察高職生接受閱讀策略改善訓練後，可否藉由克漏字表現提昇彰顯其閱讀理解之提昇，亦進而驗證理性選擇型克漏字測驗可以做為閱讀理解的適當工具。 本研究採用學者巴克曼(L.F.Bachman)論點，將克漏字題型由四類分為局部理解試題(local items)及整體理解試題(global items)兩大類型，認為克漏字測驗可做為閱讀理解的適當測驗工具，並假設可以經由改善受測者後設認知閱讀策略的知識與技巧，提昇其克漏字測驗作答表現。並且運用推論策略(Drawing Inferences)以及戴維(B.Davey)所提出之「五步驟教師出聲思考模仿法」(Five-step Teacher's Think-aloud Modeling)訓練高職生綜合運用認知策略及後設認知閱讀策略的能力，並且加強其利用文章凝結結構的推論策略(Inferencing)，以加強其採用由上而下(Top-down Model)的閱讀模式，建立較佳之整體文意理解。 受試者為中壢高級商業職業學校高三學生，測驗工具取自大學聯考（民國八十二至九十年）以及大學推甄學科能力測驗（民國八十三至九十年）英文科試題之克漏字測驗及閱讀測驗，混合編製而成，並根據局部理解(local items)及整體理解(global items)將試題分類，實驗組於前測後施以兩個月之閱讀策略訓練，提昇其綜合運用認知與後設認知閱讀策略的知識及能力，期滿實施後測，觀察其在克漏字及閱讀測驗作答之表現是否提昇。並且於後測隨機抽樣學生實施口頭訪談，以了解其閱讀行為是否產生改變。 經由閱讀策略問卷調查及前後測實施結果分析，發現受試者運用在克漏字測驗作答的閱讀策略偏向局部理解策略，而改善閱讀策略知識與運用技巧可提升其克漏字測驗之作答表現。實驗組與控制組前測作答表現之平均數差異檢定未達顯著差異(t=.048,p=.962)，後測作答表現平均數則達到顯著差異(t=5.791,p=.000)，且實驗組低分群、中分群、高分群學生之前後測成績皆有顯著進步。 其次，本研究發現高職生之克漏字測驗與閱讀測驗作答表現呈現正相關。 本研究結論發現：若經過適當設計，克漏字測驗可以做為適當的閱讀理解測驗工具，也可以做為提昇閱讀理解的教學技巧，改善高職學生後設認知閱讀策略的知識及運用技巧。 / The purpose of this study is to probe the correlation between the vocational high school students' strategy use and their performance of Multiple-choice rational doze test, and the feasibility of improving their doze performance through a strategic reading training program. The Cloze test, taking on various forms, has caused intense debates in the past decades on whether it can serve as a proper measurement tool of reading comprehension and a teaching skill of improving reading comprehension. Some scholars agree that doze tests can tap intersentential sensitivity------global comprehension of the text, while others contend that it can only measure intrasentential sensitivity------local comprehension such as vocabulary and syntactic information. This study adopts Bachman's argument that doze tests can tap readingcomprehension and brings out the hypothesis that students can improve their cloze test performance through the strategic reading training. A training program is run on vocational high school subjects using the strategy drawing inferences and Davey's Teacher's Think-alouds Modeling approach. The Experiment group receives the strategic reading training that lasts two months after the pre-test. After the program ends, the post-test is taken to examine the performance differences. An oral interview is done to nine subjects randomly selected from the Experiment group to observe to what extent the subjects have changed their reading behavior. The results show that the subjects depend heavily on lower-level reading strategies in taking cloze tests, and the strategic training program can help enhance the Experiment group's reading comprehension and in turn improve their doze performance. In addition, the results confirm that the doze performance correlates positively with the reading comprehension performance. The major findings are that doze tests can serve as a proper measurement tool of reading comprehension and as a teaching skill of improving reading comprehension and that strategic reading training helps improve vocational high school students' cloze performance.