Task Difficulty in Semi-direct Speaking Tests-Code Complexity

Abstract

本項究究係「全民英檢」口語能力測驗效度研究計劃之一部份。本文主要探討「全民英檢」中級口說能力測驗朗讀短文之文段所使用的語彙對測驗難易度產生的影 響。每位考生朗讀八段短文，依語彙分成兩種具不同難易度之版本，以比較考生在兩者之滿日於成績、發音語調之錯誤及不當停頓之頻率、考生對難易度之差異。本 文也就如何有效控制不同試卷之難易度平行與確保測驗之公平性提出具體建議。

The paper reports preliminary findings form an intial stage of a study on task difficulty in the GEPT Speaking Test-Intermediate Level (GEPTS-1, 全民英檢中級口說能力測驗), a semi-direct speaking test. The aim of the study is to examine the effects on task difficulty of variables associated with various performance conditions, such as complexity of language in written input, test-takers' familiarity with non-verbal prepositional content, and planning time. This paper focuses exclusively on the effects on task difficulty of performance conditions associated with the code complexity of written input in the read-aloud tasks of GEPTS-1 candidates, participated in this study. Each subject was asked to read aloud a total of eight passages, four of this are standard passages as sued in the past administrations of GEPT-1and the other four passages were specially modified for this study. In the modified passages, lexical range was varied and its impact on the difficulty of the tasks was measured by comparison of test scores of accuracy and fluency, number of errors in pronunciation/intonation and unnatural pauses, and test-takers' reports on estimates of difficulty. Implications for the design of speaking assessments as well as directions for future research are discussed.