In the last few years, Large Language Model (LLM)–based chatbots have increasingly been adopted in language learning contexts as conversational partners. The use of chatbots allows for innovative language learning environments, in which such tools can become personalized practice assistants for students, potentially tailoring conversational exchanges on learners’ preferences. While previous research focused on chatbots’ pedagogical potential, less attention has been paid to the interactional qualities that shape learner experience, particularly politeness and empathy. This study analyses 72 written interactions between Italian university students learning English as a Foreign Language (EFL) and two LLM-based chatbots, ChatGPT and Pi. The data belong within the PRIN UNITE inter-university project and were collected through experimental sessions based on two tasks, role-play and small talk, followed by post-hoc student questionnaires on students’ perceptions about the experience. Using a Critical Discourse Analysis framework, the present contribution delves into how the chatbots enact politeness and empathy, and how students relate to this interactional experience. The findings show that both chatbots employ politeness and empathetic strategies, contributing to the construction of a supportive and non-judgemental interactional environment. Questionnaire data indicate that students positively value the quality of interaction with chatbots, feeling included and motivated. By fostering a minimally stressful learning context and through their continuous availability, chatbot-mediated interactions expand opportunities for EFL learners to practice English. This study therefore offers insights into the strengths and limitations of chatbot-mediated language learning techniques that could complement traditional learning methods, while highlighting the central role of empathy and politeness in sustaining interactional engagement and conversational fluency.
Brasi et al. (Wed,) studied this question.