Navigating elementary EFL speaking skills with generative AI chatbots: Exploring individual and paired interactions

Tzu Yu Tai, Howard Hao Jan Chen

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)

Abstract

Generative artificial intelligence (GAI) and automatic speech recognition (ASR) have ushered in promising tools for foreign language learning, notably GAI chatbots. This study investigated the impact of GAI chatbots on elementary school English as a foreign language (EFL) learners' speaking skills, focusing on two interaction configurations—individual and paired. Eighty-five elementary school EFL learners participated in a three-week summer program, engaging in daily 45-min interactions with CoolE Bot. The participants were randomly assigned to three groups: (1) individual interaction with CoolE Bot (I-Bot group), (2) paired interaction with CoolE Bot (P-Bot group), and (3) interaction with teachers and peers in a conventional English classroom (No-Bot group). In each class, participants in the Bot group received worksheets with a topic, prompts, and vocabulary to guide their interactions with CoolE Bot, while those in the No-Bot group also received worksheets for comparable activities. Quantitative (English-speaking tests) and qualitative data (semi-structured interviews) were collected and analyzed. Results revealed that the I-Bot and P-Bot groups' post-test speaking skills were significantly higher than those of the No-Bot group. CoolE Bot significantly improved the speaking skills of EFL learners. Both individual and paired interactions with CoolE Bot demonstrated positive effects, with no significant differences between groups. Interviews highlight CoolE Bot's adeptness in coherent interaction, charismatic conversational style with a human-like voice, diverse topic discussions tailored to learners' interests, and supportive functions. The participants found GAI chatbot-assisted EFL speaking enjoyable, motivating, and engaging appreciating its cartoonish, human-like characters, conversational style, and voice. Additionally, CoolE Bot fostered rapport and a supportive environment enhancing learners' confidence and reducing anxiety regarding EFL speaking. Individual interactions encourage personalized engagement and self-directed learning, whereas paired interactions involve social dynamics, shared learning experiences, and mutual resolution of language challenges.

Original languageEnglish
Article number105112
JournalComputers and Education
Volume220
DOIs
Publication statusPublished - Oct 2024

Keywords

  • Chatbots
  • Cooperative/collaborative learning
  • EFL speaking
  • Elementary education
  • Generative artificial intelligence (GAI)

ASJC Scopus subject areas

  • General Computer Science
  • Education

Fingerprint

Dive into the research topics of 'Navigating elementary EFL speaking skills with generative AI chatbots: Exploring individual and paired interactions'. Together they form a unique fingerprint.

Cite this