IDEAS: Journal on English Language Teaching and Learning, Linguistics and Literature
Vol. 14 No. 2 (2026): IDEAS: Journal on English Language Teaching and Learning, Linguistics and Lite

Improving SMK Students' Speaking Skills with AI-Based Learning in Vocational Contexts: A Quasi-Experimental Study

Sitti Maesura Yahya (Postgraduate Program of English Language Education, State University of Makassar (UNM), Makassar, Indonesia)
Munir (Postgraduate Program of English Language Education, State University of Makassar (UNM), Makassar, Indonesia)
Amra Ariyani (Postgraduate Program of English Language Education, State University of Makassar (UNM), Makassar, Indonesia)



Article Info

Publish Date
02 Sep 2026

Abstract

English speaking proficiency is essential for vocational high school (SMK) graduates entering the workforce, yet speaking instruction often lacks authentic workplace contexts and scalable feedback mechanisms. Although AI-based language learning and vocational contextualized approaches are each documented in the literature, their synergistic integration remains empirically under-researched in Indonesian SMK pharmacy settings. This study examined whether AI-based learning embedded in vocational pharmacy tasks significantly improves the English speaking skills of Grade XI students at SMKN 1 Polewali. A quasi-experimental nonequivalent pretest–posttest control group design was employed with 31 Grade XI Pharmacy students (experimental n = 15; control n = 16). The experimental group received Gemini-assisted patient-counseling role-play instruction; the control group received conventional vocational-contextualized instruction without AI, across eight sessions over four weeks. Speaking performance was assessed using Heaton's (1988) rubric covering accuracy, fluency, and comprehensibility, and analyzed using paired-samples t-tests, an independent-samples t-test, and Cohen's d. The experimental group demonstrated significant pre–post improvement across all speaking aspects (p < .001) with large effect sizes (average d = 2.97), and significantly outperformed the control group at posttest (p = .006, d = 1.06). Comprehensibility showed the greatest gain. Theoretically, the study extends Swain's (1995) Output Hypothesis into AI-assisted ESP contexts by showing how chatbot-mediated role-play supports pushed output and immediate corrective feedback in resource-constrained SMK settings. The novelty lies in the synergistic integration of AI-based learning and vocational contextualized instruction—an approach not previously validated in Indonesian SMK pharmacy classrooms. AI-based learning in vocational contexts effectively improved speaking skills. Integrating accessible AI tools with workplace-relevant tasks offers a scalable model for English instruction in SMK programmes.

Copyrights © 2026






Journal Info

Abbrev

ideas

Publisher

Subject

Languange, Linguistic, Communication & Media

Description

IDEAS Journal is published twice a year in the months of June and December (P-ISSN 2338-4778 and E-ISSN 2548-4192); it presents articles on English language teaching and learning, linguistics, and literature. The contents include analyses, studies and application of theories, research report, ...