A comparative Study of AI English Tutors and Tamil Nonverbal Communication Patterns for the Development of Culturally Adaptive AI tools in ESL Context

Abstract

Artificial Intelligence (AI) powered English tutoring is a paradigm shift in English as a Second Language (ESL). It is an AI-centred teaching and learning approach rather than a teacher- or student-centred approach for Tamil speaking ESL learners and teachers. The AI tutors facilitate training for the target groups to communicate verbally and nonverbally like native speakers. Though AI tutors are an effective tool for improving verbal communication among learners and teachers, the cultural nonverbal communication system would be affected as the AI-powered tutors are designed based on Western communicative norms and communication practices. This creates a significant cultural gap for Tamil ESL learners whose communicative behaviours are deeply influenced by indigenous cultural traditions and performance arts such as Bharatanatyam. This study aims to comparatively investigate the nonverbal communication patterns embedded in AI English tutors and Tamil nonverbal communication practices derived from Bharatanatyam in order to contribute to the development of culturally adaptive AI tools for ESL contexts. The study specifically examines the extent to which current AI English tutors incorporate culturally diverse facial expressions, gestures, eye movements, and emotional communication patterns relevant to Tamil learners. A qualitative comparative research design was employed for the study. Ten AI English tutoring systems were purposively selected and analysed using multimodal discourse analysis and nonverbal communication frameworks. The communicative behaviours of these AI tutors were compared with Tamil nonverbal communication patterns referenced through the Bharatanatyam concept of Abhinaya, including Mukhabhinaya (facial expressions), Netrabhinaya (eye expressions), gestures (Mudras), head movements, eyebrow movements, and body posture. Also, the analysis combined the Navarasa framework, which classifies nine cultural and emotional expressions such as Shringara (love), Hasya (Joy), Karuna (sandness), Raudra (anger), Veera (courage), Bhayanaka (fear), Bibhatsa (disgust), Adbhuta (wonder), and Shanta (peace). The findings of the research revealed that almost all the existing AI English Tutors consisted of limited Western-oriented facial expressions like broad smiling, sustained eye contact, frowning for anger, and raised eyebrows for surprise and visible facial tension for disagreement. While these expressions support basic emotional interaction, the AI tutors demonstrated limited incorporation of culturally nuanced Tamil nonverbal communication features, which are commonly represented in Bharatanatyam traditions, such as respectful downward gazes, side glances, subtle eyebrow movements, controlled emotional expressions, head tilting, gentle nodding, and culturally contextualised emotional transitions were largely absent or underrepresented within the analysed AI English tutors. The study emphasises that an AI English Tutoring system without Tamil cultural nonverbal communication patterns may contribute to cultural incongruence in an AI-centred ESL learning context. Therefore, the study newly propose and conceptualize a culturally adaptive framework for the creation of a ‘Tamil AI English Tutor’ as a paradigm shift towards facilitating Tamil ESL learners while preserving the indigenous communicative identities, acknowledging the diverse, and evolving concept of Englishes. In addition, this research contributes to the fields of AI-assisted language learning, applied linguistics, intercultural communication and human-computer interaction by introducing a culturally responsive framework for designing AI education technologies for ESL context.