Budiyanto, Issy Yuliasri, W. Widhiyanto, Sri Wahyuni
This study investigated the effectiveness of integrating Artificial Intelligence (AI) tools, namely ChatGPT and ELSA Speak, into Project-Based Learning (PjBL) to improve students’ speaking performance in a Buddhist College in Central Java. A mixed-methods design with an exploratory sequential approach was employed, involving 32 students in the experimental class and 32 students in the control class. Data were collected through speaking tests, observation sheets, questionnaires, interviews, and document analysis, and were analyzed using SPSS and thematic analysis. The quantitative findings revealed that the experimental class achieved a significant improvement in speaking performance, with the mean score increasing from 63.47 in the pre-test to 81.59 in the post-test, compared to the control class which improved from 65.78 to 76.97. The Mann-Whitney U Test confirmed a statistically significant difference between the two groups (p < 0.05), indicating the effectiveness of AI-integrated PjBL. Qualitative findings supported these results, showing that AI tools enhanced students’ pronunciation, fluency, vocabulary, grammar, and confidence through interactive practice and immediate feedback. ChatGPT assisted in content development and language structuring, while ELSA Speak supported pronunciation accuracy. The study concludes that AI-integrated PjBL is an effective and innovative approach for improving speaking performance, as it combines technology, collaboration, and meaningful learning experiences in higher education contexts. © 2026 Authors.
Universitas Negeri Semarang, Indonesia
