A Study Of The Effect Of Proficiency Level In Paired Role Play Tape-mediated Oral Tests

In recent years great importance has been attached to the performance assessment of oral communicative competence. It can be influenced by a wide range of features of the test taker, the task, the assessment criteria and the interlocutor/examiner (Milanovic and Saville 1996). Recent research has highlighted the importance of the effects of test taker characteristics on candidate performance. Despite a body of work investigating the issue of variables within the test taker, there are a number of aspects that remain little explored. The present study focuses its attention on the influence of test taker's language proficiency levels. The purpose of this study is to investigate whether different language proficiency levels may result in difference in a candidate performance in paired oral speaking tests involving role play activity. This study considers the issue in a tape-mediated test delivered in a language lab, where no interlocutor is present.Before the study a pre-study questionnaire and think-aloud verbal report protocol were carried out to conclude the research hypotheses and related research questions. Then a group of 66 Chinese sophomore students in college attended a simulated oral proficiency test (SOPT). They were divided into two groups (high-proficiency group and low-proficiency group) according to their scores of Oral Test, TEM-4, and performed the tasks with different partners. All performances were recorded, rated by five-trained raters (two using holistic rating scale and three using analytic rating scale) after the test, and transcribed for language analysis later. A One-Way ANOVA was adopted to examine the scores and the language.Results from test scores and language analysis both suggest evidence of a partner-proficiency-levelâ–¡effect for low-proficiency group, with candidates achieving higher scores and grammatical accuracy when working with a high-proficiency partner. They improved greatly in aspects of tense, prepositions, articles and agreement of subject and predict. However language analysis shows that there is no effect on grammatical complexity for them when they were paired at different proficiency levels. Furthermore for high-proficiency group there is no great change in terms of test scores, the analysis of grammatical complexity and grammatical accuracy whether their partner is a high-proficiency candidate or a low-proficiency one.This paper concludes with discussions of limitations, implications and suggestions for further research. Despite its limitations, the results of this study confirm that this performance variable is in need of urgent and extensive study.