CHATGPT-ASSISTED SPEAKING ACTIVITIES TO IMPROVE MATHEMATICS EDUCATION STUDENTS' SPEAKING SKILLS: A CLASSROOM ACTION RESEARCH

Abstract

The rapid advancement of artificial intelligence (AI) has created new opportunities to enhance English language learning, particularly in developing students' speaking skills through interactive and technology-assisted environments. This classroom action research aimed to: (1) investigate the effect of ChatGPT-assisted speaking activities on the speaking performance of Mathematics Education students, and (2) explore students' perceptions of using ChatGPT in speaking practice. The study employed the Kemmis and McTaggart model of Classroom Action Research, conducted in two cycles with 35 second-semester students of Class A1 in the Mathematics Education Study Program during the 2025 academic year. Data were collected through speaking tests, classroom observations, questionnaires, and field notes. Findings revealed a continuous improvement in speaking performance from the preliminary study (M=58.40) to Cycle I (M=69.15) and Cycle II (M=81.37), with notable gains in pronunciation, grammar, vocabulary, fluency, and comprehension. Observation results showed increased classroom participation (from 65% to 88%) and confidence (from 60% to 85%). Questionnaire responses demonstrated positive perceptions of ChatGPT across ease of use (M=4.42), usefulness (M=4.58), motivation (M=4.47), and confidence (M=4.53). The study concludes that ChatGPT-assisted speaking activities effectively enhance students' speaking competence and create a supportive learning environment. Integrating generative AI into EFL speaking instruction offers a promising pedagogical approach, particularly for non-English-major students.