New Ways to Teach and Learn Pragmatics for Advancing Academic Communication Competence: An AI-Mediated Multimodal Platform

Abstract

Effective academic communication in multilingual higher education requires pragmatic competence, yet this dimension of language ability is often overlooked in instruction and insufficiently supported through feedback or authentic practice. Grounded in instructed pragmatics and technology-mediated learning frameworks, this study reports on the design, implementation, and empirical evaluation of an artificial intelligence (AI)-mediated multimodal platform developed to scaffold pragmatic competence in Indonesian within digital academic contexts. The platform integrates five interdependent features: automated text analysis to identify speech acts, politeness strategies, and violations of conversational principles; video analysis with multilingual transcription and theory-based annotation; conversation simulations providing context-specific academic interactions and real-time feedback; an AI consultation assistant offering explanations, examples, and targeted suggestions; and a digital politeness forum for peer and instructor exchange. A quasi-experimental mixed-methods design was employed with 240 undergraduates (120 in the experimental group, 120 in the control group) over a 5-week intervention. Pre- and postintervention assessments combined a validated pragmatic competence test, discourse completion tasks, and a self-report questionnaire on pragmatic awareness and confidence, complemented by student reflections and discussions. The experimental group demonstrated significantly greater gains in pragmatic test scores ( M = 15.0, p < .001) compared with the control ( M = 5.2, p < .001), alongside improvements on the discourse completion task (experimental = +14.1 vs. control = +4.8, both p < .001) and medium-to-large effects on awareness and confidence in the questionnaire. Regression analysis identified engagement with conversation simulations ( β = 0.58, p < .001) plus AI consultation ( β = 0.29, p = .012) as predictors of learning. Reflections and discussions underscored the platform’s role in enabling authentic practice, personalized feedback, and a safe environment for pragmatic experimentation. This study provides the first empirically validated model linking integrated AI-mediated scaffolding features to measurable and transferable gains in pragmatic competence, offering a scalable framework for multilingual and intercultural higher education.