Rational Cloze

Abstract

國內各級英文考試及英語文課堂內的練習裡都常見克漏字測驗，而既有的克漏字題目究竟達成了何種評量目標？也就是說，它是測出了語彙能力、語法知識或是篇章 結構知識呢？我們當如格在命題後檢視此一問題？這就是本文所關注的議題。本文例舉若干實例分析，藉以闡述探究克漏字測驗題目之建構效度的過程。此外，本文 說明漏/刪字策略及選目設計足以影響選擇式搭漏字測驗的建構效度。本文建議命題者先對擬選用的文章做言談/篇章結構分析，再據以審慎決定漏/刪字(詞)對 象(可為單字或片語)。分析篇章結構後再行命題的克漏字測驗可免於測驗內容集中於較低層次的語彙及語法知識而忽略高層次的篇章結構及語用知識。

The cloze test has been widely used in English tests in Taiwan. However, what the developed cloze tests assess remains questionable. This article addresses the following issues. How can one determine what developed cloze items assess? How can one use the cloze to measure discourse skills? How do item-generation approaches affect the construct validity? The article presents the results of a qualitative study. It reports on post-hoe analyses of a cloze test that was included as part of the standardized test, demonstrating that in order to determine the construct validity one needs to examine the cognitive demands required to responds to the item. It also reports on discourse analysis of two text selections, demonstrating that discourse analysis could help generate discourse-oriented cloze items. In addition, the article reports on analysis of several items of standardized test, demonstrating that the deletion strategy and the alternative design decide the item-generation approach to the multiple-choice cloze and the approach in turn affects the type of test construct.