Justifying The Validity Of The Listening-based Retelling Task In The National Natriculation English Test

Listening-based oral retelling is a frequently used integrated speaking task. This study aimed to define its task construct, describe the task characteristics, and create a valid rating scale. Based on the Assessment Usage Argument (AUA)(Bachman&Palmer2010), the present study built a validity argument for the listening-based oral retelling task in order to investigate and justify the validity of the task. The argument consisted of two claims and their relevant warrants, i.e."The interpretations about the ability to be assessed are meaningful" and "Assessment records are consistent" Both theoretical and empirical evidence was obtained to justify the claims and warrants.From the perspective of cognitive psychology, this study defined the listening-based oral retelling task construct as the ability to recognize and recall the detailed information of the listening stimulus and reproduce the information orally by applying one’s linguistic and non-linguistic knowledge and strategies. This provided the theoretical evidence for the warrant supporting the interpretations claim, i.e."The definition of the construct is based on the theory of communicative language ability and the related researches". The empirical study involved5experienced raters and89high school graduating seniors. Through the rater judgement, test takers’verbal protocol reports, raters’verbal protocol reports, discourse and text feature analysis, Kruskal-Wallis test, and correlation analysis, the study collected sufficient evidence for the warrants supporting the interpretations claim, i.e."The procedures for producing listening-base oral retelling task records focus on those aspects of the performance that are relevant to the construct we intend to assess","The listening-base oral retelling task engages the ability defined in the construct definition","The listening-based retelling task specifications clearly specify the conditions under which we will observe or elicit performance from which we can make inferences about the construct we intend to assess","The listening-base retelling task records can be interpreted as indicators of the ability to be assessed" and the warrants supporting the assessment records claim, i. e."Ratings of different raters are consistent" and "Different ratings by the same rater are consistent". The study finally created the data-claim inferential links to support and justify the validity argument.The results manifested that the listening-base oral retelling task engaged the ability defined in the construct definition. Besides, compared with narrative retelling and prepapred retelling, expositive retelling without preparation time could better elicit students’performance from which inferences were to be made about the construct intended to assess. Lastly, the holistic rating scale including the qualities concerning content, language usage and language delivery was proved to be valid.The construction and validation of the listening-based oral retelling task validity argument shed light on applying practical argumentation theories and the latest validation theories into English speaking assessment practice in China. The study also provided guidance to the design and implementation of listening-based oral retelling tasks. Furthermore, these findings had several implications for China’s high school English testing practice.