「全民英檢」學習者自我評估量表發展

Abstract

「全民英檢」自民國89 年施測以來，廣受採用，迄今已達400萬人次報考，成績獲申請入學、畢業、甄選等參考。有鑑於「全民英檢」的重要性，並為更進一步協助學習者選擇適當的級數報考或作為考前了解自己的強、弱點的參考，本研究以「全民英檢」的能力指標為本，經實證程序發展適合「全民英檢」初至中高級的聽力、閱讀自我評估量表。本研究首先經專家判斷程序製定聽力22 題、閱讀21 題的能作敘述 (can-do statement)，編製自我評估量表，並於2009 年間邀請共8,006 位「全民英檢」考生回答該量表。接著，以Rasch 模式與次序羅吉式迴歸分析方法 (ordinallogistic regression) 探討學習者自我評估的結果與「全民英檢」成績間的關係，並計算出三個級數的切分點與正確預測級數的機率模型。分析結果顯示：該模型能正確預估學習者能力的機率分別為聽力 (68%)、閱讀 (65%)，明顯高出隨機機率 (33%)，證明該量表具相當的預測效度。該量表可提供全民英檢指引與教學參考，教師亦可使用量表中的能作敘述訂定更明確的教學目標與設計課堂活動。

The General English Proficiency Test (GEPT), a five-level criterion-referenced test, was developed to assess the general English proficiency of TaiwaneseEFL learners. Since its first administration in 2000, the GEPT has won wide recognition, with four million Taiwanese having taken the test. This paper reports onan empirical study of the relationship between self-assessment and test performance. The self-assessment tool, developed in accordance with the test construct in the GEPT, consists of 22 listening and 21 reading can-do statements. Eight thousand and six Taiwanese EFL learners were invited to take a GEPT test and respond to the self-assessment statements. The data were analyzed by Rasch model and ordinal logistic regression. Results show that the self-assessment statements have achieved acceptable accuracy (0.68 for listening; 0.65 for reading) in estimating learners' language levels. The paper concludes that the GEPT self-assessment statements can be considered a useful tool allowing learners to pre-estimate their ability before registering for a GEPT test and therefore recommends that the tool be utilized to enhance learners' awareness of their learning and proficiency, and ultimately encourage autonomous learning.