Amy KONG
Asian Journal of English Language Teaching. 2024, 33(1): 53-86.
Second language (L2) testing research over the last two decades has witnessed a noticeable shift from measuring multiple validities to presenting evidence to the argument-based validation framework, but seldom did they examine the argu- ment-based validity of the Hong Kong Diploma of Secondary Education Exami- nation (HKDSE) English Language speaking test (hereafter referred to as HKDSE speaking test), which is taken in the form of group interaction that considers not only the validation components based on Levelt's (1989) speaking model but also international competence. The current study aims to validate the HKDSE speaking test by presenting evidence to back the warrants drawn in the domain description, explanation, and utilization inferences within the argument- based framework. The study adopted the qualitative approach by analyzing different sources of artifacts, including the constructs, task test samples, marking criteria, and authentic group discussion samples, as well as transcripts of the interview with four prospective task-takers. Results indicate that despite high consistency between most constructs, marking criteria, and theoretical expecta- tions, the non-specifications of the interaction context, participants' roles, and task purposes in the majority of the test prompts make it hard to fully justify both the target domain and explanation inferences. The interviewees' disregard of the representativeness of the test scores in manifesting the language use in real-life contexts has also undermined its validity in terms of utilization inference.
