The Effects of ICTE Robot on Students’ English-Speaking Skills: A Mixed-Methods Study on Non-English Majors at Nguyen Tat Thanh University

Amid Vietnam's growing emphasis on English as a second language, educators continue exploring AI-powered tools to help learners speak accurately, naturally, and fluently. While prior studies have examined well-known chatbots such as ELSA, Google Assistant, and ChatGPT, few have investigated a customized chatbot within a mixed-methods design. This study addressed that gap by examining the impact of the ICTE robot, a custom-built AI chatbot, on the speaking performance and perceptions of 24 non-English-major students at Nguyen Tat Thanh University. Pre- and post-tests adapted from the PET speaking rubric assessed vocabulary, grammar, pronunciation, and discourse management, while semi-structured interviews with nine participants explored their experiences. Paired-samples t-tests revealed significant gains in vocabulary, pronunciation, and discourse management, though grammar showed no significant improvement. Interviews indicated that students valued the ICTE robot's voice-based and multimodal features, reporting greater confidence, stronger vocabulary retention, and improved pronunciation, while also noting technical issues such as speech-recognition errors and connectivity problems. These results suggest customized AI chatbots can meaningfully support EFL speaking development, highlighting the need for improved feedback accuracy and expanded grammar practice.