A Comparative Study On Selected Response And Constructed Response In Assessing Communicative Listening Ability

This paper examines the differences of two response types, i.e. selected response (or multiple-choice question) and constructed response in the context of communicative language testing of listening comprehension, from the perspectives of test validity, reliability, authenticity, and impact on language teaching and learning.The first half of this paper lays out a theoretical background for this research:Chapter one gives a short introduction to this paper, including rationale, hypothesis, background and purpose of this study.Chapter two concerns the nature of language testing and dwells upon four considerations in evaluating the two response types: validity, reliability, authenticity and impact on language teaching and learning.Chapter three reviews the development of language testing and the evolution of response types used in different approaches. The communicative language testing calls for response types capable of assessing communicative language abilities, other than multiple-choice questions that features the psychometric-structuralist approach.Chapter four elaborates on listening comprehension ability, which has been granted more importance in the new College English Curriculum Requirements, as the construct in the following empirical research. The author has pointed out features of listening comprehension and the linguistic and non-linguistic components in this process, followed by a survey of listening comprehension testing in TOEFL, 1ELTS and CET, which is featured by communicative orientation and more constructed response types.With the theoretical foundation laid thus far, the second half of this paper probes into the differences of the two response types:Chapter five first elaborates on definition, taxonomy, and significance of response type. A review of theoretical researches proves constructed response to excel selected response in validity, authenticity and impact on language teaching andlearning. A review of empirical researches, though not as productive as theoretical ones, sheds light on how further exploration should be conducted.Chapter six describes the empirical research equipped with aforementioned insights. Language teachers' perception toward the validity, reliability, authenticity and backwash effect of the two response types was elicited by a questionnaire. Non-English major college students were asked to voice their views on affective factors, authenticity, validity, reliability, and the impact on language learning in a second questionnaire. Some of them performed on two listening comprehension tests using the two response types respectively. Six subjects were chosen among them to participate in an immediate retrospective verbal protocol examining their task-taking processes. The result shows that constructed response enjoys higher validity, authenticity and impact on language teaching and learning than selected response. Nevertheless, further researches should go into how to ensure reliability in constructed response questions, especially when used in large-scale testing.In the conclusion, the author proposes CET to take up a larger proportion of more constructed response, in accordance with the need for communicative testing. Some issues are raised for further investigation.