Abstract
This study investigated the effects of mobile-supervised question-driven collaborative dialogues (QDCDs) on reducing lower-intermediate-level English as a foreign language (EFL) participants' tendency of their first language (L1) use in academic collaborative dialogues and on improving their academic foreign language (L2) oral performance. Throughout a whole semester, one group (n = 20) was involved in a mobile-supervised QDCDs intervention and a control group (n = 26) was involved in QDCDs with no supervision. Three semi-open-ended and three closed-ended academic questions were used to elicit pre-and post-study oral performance data from the participants. Independent-samples t-tests showed that after the intervention, the mobile-supervised group outperformed its control counterpart in a statistically significant manner in terms of Non-repeated L2 word production (NRW), T-unit count (TC), and Mean Length of Run after pruning (MLRP). The intervention group also significantly reduced their dependence on their L1-based speaker compensatory communication strategies (SC-CSs) in QDCDs. These results suggest that the intervention group outperformed the control group in their L2 academic oral performance and their language use tendency moves toward the L2 during QDCD. Based on the findings, we conclude that, even though L1 oral output may temporally enhance the quality of lower-intermediate-level EFL learners' tasks, it may inhibit their academic oral proficiency development in the long run. Methods for fragmental bilingual oral output analysis are introduced. Pedagogical implications of the findings for MALL are also discussed.