NEGOTIATING LANGUAGE THROUGH ALGORITHMS: A QUANTITATIVE INVESTIGATION OF TRANSLANGUAGING IN AI-ASSISTED EFL WRITING

Résumé

The rapid integration of artificial intelligence (AI) writing tools into English as a Foreign Language (EFL) education is transforming multilingual writing practices. Yet, limited empirical research has examined how AI-mediated environments interact with learners’ deployment of their full linguistic repertoires through translanguaging. This quantitative study addresses this gap by investigating patterns of language alternation, the functions of translanguaging in AI-assisted writing, learners’ satisfaction, and the relationship between translanguaging functions and satisfaction across institutional contexts. Data were collected from 100 Moroccan university students engaged in AI-supported writing tasks. Analyses included descriptive statistics, the Kruskal–Wallis H test, and Spearman’s rank-order correlation. Results showed relatively high satisfaction with translanguaging practices (M = 4.04, SD = 0.94). Learners employed diverse language alternation patterns, primarily to support idea generation, lexical retrieval, and comprehension. No significant differences in satisfaction emerged across language patterns or institutions, and the relationship between translanguaging functions and satisfaction was weak and non-significant (ρ = −.167, p = .097). These findings suggest that translanguaging in AI-mediated writing operates as a flexible, agentive practice rather than a determinant of user satisfaction. Pedagogically, the results support integrating AI tools in ways that legitimize multilingual meaning-making.