An Article based on My Thesis Study
There is a moment in almost every speaking class that feels familiar: you ask a question, you wait, and the room fills with silence. Not because students don’t know the answer, but because they are not ready to speak. This moment stayed with me and eventually became the starting point of my research. I began to wonder whether the issue was not just about language ability, but about confidence, pressure, and the lack of safe spaces to practice. My study explored the use of AI—specifically ChatGPT—in supporting speaking instruction, focusing on its impact on students’ speaking performance and their willingness to communicate.
The idea grew directly from my classroom experience. I often saw capable students hold back during speaking activities. This made me question whether traditional classroom settings were always the best space for practice, and whether alternative tools could help students feel more comfortable using the language.
I conducted this study at the preparatory program of Fatih Sultan Mehmet Vakıf University, working with upper-intermediate students in two intact classes—one experimental group and one control group. Over six weeks, the experimental group engaged in AI-supported speaking practice using ChatGPT, completing structured speaking tasks outside the classroom twice a week. These tasks were designed to mirror real-life communication and were aligned with IELTS-style prompts.
To understand the impact, I collected both quantitative and qualitative data. Students completed pre- and post-speaking tests assessed through IELTS speaking criteria, and their willingness to communicate was measured using a standardized scale. I also analyzed changes in their speaking rate (words per minute) as an indicator of fluency.
What I did not expect was how naturally students would interact with the AI. Many described it as a space where they could try, make mistakes, and try again—without feeling judged.
The findings showed that students who used AI for speaking practice improved more in their speaking performance compared to those who followed traditional methods. They also reported higher levels of willingness to communicate, which, for me, was one of the most meaningful outcomes.
However, what stayed with me the most was not just the improvement in scores, but the change in students’ behavior. Students who were previously hesitant began to take more risks and participate more actively. There was a noticeable shift in their confidence.
It became clear that speaking development is not only about grammar or vocabulary, but also about reducing anxiety and creating opportunities for low-pressure practice. In this sense, AI functioned less as a “tool” and more as a supportive partner in the learning process.
This experience has reshaped how I think about speaking instruction. I now see AI as a way to extend the classroom—giving students more space to practice, reflect, and build confidence at their own pace.
If I were to say one thing to colleagues considering classroom research, it would be this: the most meaningful questions often come from those small, quiet moments in our classrooms. Paying attention to them can open the door to powerful insights—not only about our students, but also about our own teaching.
Elnaz Asadi
English and Spanish Instructor at Fatih Sultan Mehmet University/ Modern Languages Unit
This article is based on my MA thesis titled “Investigating the Impact of using AI-Tools for Speaking Instruction on Adult EFL Learners’ Speaking Performance and Willingness to Communicate,” conducted under the supervision of Asst. Prof. Dr. Mustafa Polat.