The Effect Of Language Proficiency Based Parings On Interactive Patterns And Conversational Features Of Paired Oral Tests

The design of the speaking test should reflect the nature of speaking, which is described as a part of the social activity of talking (Luoma,2004). Paired oral tests, as one of the formats that motivate test takers co-construction of interaction and communicative features, are considered as a practical application of the nature of speaking. Therefore, paired oral tests have gradually become a widely accepted oral test format and received close attention of the researchers. Studies on paired oral tests mainly focus on the relationship between test takers’variables and test result as well as performance. A variety of variables, including familiarity, gender, personality, language proficiency, have been found influencing test results, among which language proficiency has been regarded as the major contributing factor in paired oral interaction, but whether the effect of language proficiency beneficial or undermining is still unclear. The inconsistency of the results is due to the only use of quantitative methods and a lack of using qualitative methods, which can re-picture the testing process. As little qualitative data is available regarding to this issue, the present study aimed to foster conversation analysis (CA) to examine the effect of language proficiency on paired oral assessment discourse. Consequently, the research question and two sub-research questions were addressed in the current study:How does proficiency-based pairing affect paired oral tests discourse?1. How does proficiency-based pairing affect interactive patterns of paired oral tests?2. How does proficiency-based pairing affect conversational features of paired oral tests?The research employed a semi-direct mode of speaking test-a simulated TEM4speaking test and conversation analysis of the test transcripts.44pairs of students joined in the test. All of their testing process was recorded and18transcripts of18special arranged pairs were further analyzed.The analysis highlights global patterns of interaction for both same proficiency pairs and different proficiency pairs. Within the same proficiency pairs, including high-high and low-low proficiency pairs, test takers engaged collaboratively with one another’s ideas, demonstrating a range of conversational features, such as providing strong listener support, frequent reference and expansion on partners’ideas, summarization of complicated ideas, etc. However, no clear interactive pattern was observed in intermediate-intermediate pairs. Within different proficiency pairs, including low-intermediate, intermediate-high, and low-high pairs, the resulting interactions displayed a clear expert-novice collaborative pattern with higher proficiency test takers facilitating less able partners in co-building the interaction by using conversational strategies, such as inviting contribution, using open-ended sentences, offering explanation of complex or abstract ideas, giving up floor-taking opportunities, etc. The study discussed theoretical and methodological implications for further research on paired oral tests and provided teachers with pedagogical implication on classroom-based pair work so as to motivate students to co-construct interaction in a more collaborative way. Finally, some limitations were shown in the present study, such as the small sample size and the control of potential variables. Recommendations for further research were presented based on the findings and limitations of the study.