Language testing researchers and test developers have long preferred to use integrated tasks due to their authentic nature. However, such presumed authenticity is taken for granted without empirical justification. Therefore, in this study, I examined and compared the authenticity of two integrated task types, a speaking task following an audio lecture on an academic topic and a multimedia-mediated speaking task (MMST)—a similar task to the integrated listening-speaking task (ILST), but with a video lecture as its stimulus.

First, I used a questionnaire to analyze student perception of 1) situational authenticity regarding the task characteristics of the two assessment tasks compared with those of a target language use (TLU) task in the academic context and 2) interactional authenticity regarding the involvement of language ability in accomplishing the two assessment tasks. Second, as another way to analyze the interactional authenticity, I investigated if language ability, consisting of language knowledge and strategic competence, was used as defined in the construct of the ILST and the MMST. For the language knowledge analysis, I collected student speech samples and analyzed them from a systemic functional linguistic (SFL) perspective. For the strategic competence analysis, I conducted stimulated recall interviews to obtain students’ strategic behavior reports. Ninety-three international undergraduate and graduate students participated in the study.

Students perceived a similar degree of situational and interactional authenticity of both the ILST and the MMST overall. In addition, both the ILST and the MMST functioned well in eliciting some SFL features of the target language knowledge. However, some other SFL features were elicited more successfully by the MMST than by the ILST. For strategic competence, students in both the ILST and the MMST groups reported a frequent use of communication, cognitive, and metacognitive strategies. However, students in the MMST group reported a more frequent use of cognitive strategies than those in the ILST group. These findings lead to an enhanced understanding of how the two types of integrated assessment tasks work on student performance and can provide empirical evidence for the domain definition inference of a validity argument for an academic English proficiency test.