Collaborative Multilingual Discourse Analysis in Science Classrooms: A Systemic Functional Approach

ACL ARR 2026 January Submission96 Authors

Keywords: Systemic Functional Linguistics, Discourse Analysis, Multilingualism, Syntactic Relations, Parts-of-Speech, Dialogue Acts

Abstract: This paper examines how a Systemic Functional Linguistic (SFL) framework can inform the design of computational features for analyzing dialogic interactions in multilingual classrooms. Classroom discourse with multilingual language learners (MLLs) often involves fluid use of multiple linguistic resources, yet standard natural language processing (NLP) methods rely on surface linguistic features that are weakly aligned with the functional meanings through which ideas, stance, and cohesion are enacted in multilingual classroom interaction. To address this gap, we map core dimensions of SFL onto interpretable linguistic indicators, including part-of-speech patterns (POS), universal dependency (UD) relations, cohesion cues, and dialogue act (DA) categories. Correlation analyses show that POS and UD patterns align most strongly with ideational meaning, DA signals provide the clearest evidence of interpersonal meaning, and textual meaning is only weakly represented across all indicator types. These findings clarify which features most effectively capture functional aspects of language use in multilingual discourse and provide a foundation for developing more inclusive and scalable NLP approaches to analyzing classroom interaction.

Paper Type: Long

Research Area: Discourse, Pragmatics, and Reasoning

Research Area Keywords: Discourse and Pragmatics, Multilingualism and Cross-Lingual NLP, Linguistic Theories, Cognitive Modeling, and Psycholinguistics, Human-Centered NLP

Contribution Types: NLP engineering experiment, Data resources, Theory

Languages Studied: English, Spanish

Submission Number: 96