A three-year reform of an EMI listening course involving 533 low-proficiency STEM students found that graded accountability alone was insufficient. Adding collaborative, voluntary, and reflective ungraded supports coincided with lower failure risk, sustained voluntary engagement, and stronger strategic awareness. Cohort-based reform does not prove that the supports caused the changes.
Key findings
- Assessment-only design was insufficient for progression. With ungraded low-risk supports, failure risk decreased, voluntary engagement persisted, and reflections showed greater strategic and goal awareness. Listening was increasingly viewed as a transferable academic skill rather than a course requirement.
Why this matters globally
EMI is expanding globally, and low-proficiency learners must process discipline content and language in real time. Balancing accountability with low-stakes support may inform more inclusive course design, subject to contextual adaptation.
Thai researcher contribution
A language-studies researcher at KMUTT's School of Liberal Arts used longitudinal evidence from Thai STEM learners to connect language pedagogy with outcome-based curriculum design.
Limitations to consider
This was not randomised; cohorts may differ in students, teachers, content, and context. Effect sizes and failure counts are not reported in the abstract, reflections may have response bias, and one course at one institution limits transferability.