In the last few years there has been a growing interest in the use of Large Language Models (LLMs) systems, especially chatbots, in language learning contexts. Their adoption allows for innovative language learning environments, where AI-powered tools serve as conversational partners and personalized practice assistants for students, offering input and feedback to learners of a foreign language.

In our experimental project, we have collected 80 interactions between LLM-based chatbots (ChatGPT and PiAI) and L1 Italian university students learning English. The sessions comprised two English as a Foreign Language (EFL) learning scenarios: a task-oriented roleplay and small talk activities. The completion of a post-hoc questionnaire on students’ perceptions about the experience was also mandatory to conclude the sessions.

The present contribution delves into the analysis of two key aspects in the approach of conversational AI chatbots with learners of English: the use of empathy and the implementation of politeness strategies by the chatbots. By analyzing the conversational choices made by ChatGPT and PiAI, we aim to understand how these behaviors foster positive communication dynamics with the students and encourage them during their EFL learning journey.

Our case study, thus, exclusively addresses the chatbots entries in the chatbots-learners' interactions to unfold the chatbots behavioral choices in the conversations – ranging from hedging suggestions to providing direct and/or indirect constructive feedback – and to observe the atmosphere established by ChatGPT and PiAI in the fostering of a more positive learning experience.

Through a minimally stressful learning setting and in virtue of chatbots’ continuous availability, this experience constitutes an innovative language learning environment where AI chatbots can enhance opportunities for EFL learners to practice English and promote new techniques for language acquisition.

Therefore, the findings of this study offer insights into the potential of chatbot-mediated language learning to complement traditional learning methods; they also reveal how empathy and politeness in chatbot interactions contribute to improved learner engagement and conversational fluency, through a personalized, empathetic learning experience.

Keywords: Chatbot-mediated learning, English as a Foreign Language (EFL), Large Language Models (LLMs), conversational AI, AI-assisted language acquisition, AI and education, technology.