Latent profiles of engagement in GenAI-mediated language learning

The role of teacher interpersonal behaviors

Bibliographic Data

ID11988993
AuthorsWeili Xing (0009-0004-2929-4257, Macao Polytechnic University), Mark Feng Teng (0000-0002-5134-8504, Macao Polytechnic University, corresponding author)
Year2026
Volume139
Pages104030-104030
Publication date2026-03-16
Open AccessYes
TypeARTICLE
VenueSystem (JOURNAL)
Journal identifiersISSN: 0346-251X • E-ISSN: 1879-3282
PublisherElsevier BV (PUBLISHER)
DOI10.1016/j.system.2026.104030
OpenAlexW7136760282
LanguageEN
Citations received1
References cited51

Despite growing interest in GenAI-mediated language learning, little is known about whether and how learners differ in qualitatively distinct engagement profiles within such contexts, nor how teacher interpersonal behaviors (TIB) differentially relate to these profiles. Addressing this gap, the present study adopts a person-centered approach to profile English as a foreign language (EFL) learners’ emotional, behavioral, and cognitive engagement in GenAI-mediated classrooms and to examine the role of TIB in shaping these patterns. Grounded in Self-Determination Theory (SDT), TIB were conceptualized as autonomy support, structure, and involvement and measured using a multidimensional survey scale. Survey data from 385 Chinese university students were analyzed using latent profile analysis (LPA), followed by profile-based group comparisons and multinomial logistic regression. LPA identified four distinct engagement profiles: Highly Engaged All-Rounders (44.4%), Moderately Engaged Emotion-Behavior Actors (16.4%), Relatively Engaged Cognitive Strategists (20.0%), and Minimally Engaged Learners (19.2%). Perceived TIB also varied by profile, with Highly Engaged All-Rounders reporting the strongest autonomy support, structure, and involvement, and Minimally Engaged Learners the weakest; intermediate profiles did not differ significantly from each other on TIB. Regression results showed that all three TIB dimensions increased the odds of belonging to more engaged profiles relative to the low-engagement group, with structure and involvement exerting particularly strong effects. Findings extend SDT to GenAI-mediated EFL contexts, reveal latent heterogeneity in engagement, and underscore the centrality of teacher-provided structure and involvement alongside autonomy support. Implications are offered for designing GenAI-supported instruction in which teachers scaffold and sustain engagement among diverse learners

Autonomy · Centrality · Cognition · Interpersonal communication · Learner autonomy · Multinomial logistic regression · Odds · Survey data collection · EFL/ESL Teaching and Learning · Motivation and Self-Concept in Sports · Second Language Learning and Teaching

Tools

Open DOIOpen Access