DEVELOPING CROSS-CULTURAL COMMUNICATIVE COMPETENCE IN ADVANCED LANGUAGE LEARNERS THROUGH GENERATIVE AI-BASED SOCIO-PRAGMATIC SUPPORT

This paper examines the effectiveness of Generative AI (GenAI) as a socio-pragmatic aid that can improve Cross-Cultural Communicative Competence (CCCC) in advanced language learners (N = 80). Although such learners may be highly proficient in their language, A so-called pragmatic gap may also be faced in attempts to negotiate complex social interactions and in high-context cultural interactions. The study followed a quasi-experimental pre-test/ post-test control group study design where the influence of a four-week structured GenAI intervention was quantified. The experimental group went through several cycles of safe failure by using GenAI models that had to simulate different cultural personalities, whereas the control group used the traditional instructional materials. Analysis of the statistical results showed that the experimental group (n = 40, total sample N = 80) recorded a significant mean score of 20.85 points on the Written Discourse Completion Task (WDCT) as compared to a 2.65- point gain in the control group, which was not statistically significant relative to the intervention's impact. A one-way ANCOVA, controlling for initial proficiency, showed a significant main effect for the intervention (F(1, 77) = 295.70, p < .001), with a Partial Eta Squared (η² = .794) indicating that the GenAI framework accounted for nearly 80% of the variance in post-test performance. Also, the effect size (Cohen d = 3.89) was found to be substantial, which implies that GenAI can be regarded as an effective artificial interlocutor to learn the nuances of cross-culture. The results have strong empirical support to include AI-informed socio-pragmatic support within the modern curricula in order to bridge the gap between grammatical accuracy and cultural fluency. The research paper ends by summarizing the pedagogical considerations of the concept of AI-mediated safe failure and outlining the existing limitations in the form of the lack of non-verbal communication in text-based AI models.

Hali tarjima qilinmagan