From Role-Play to AI Interaction: Speaking Confidence Development among EFL Nursing Students at AKPER YPTK Solok

Conversational artificial intelligence is now a common sight in classrooms for foreign languages, but evidence of it improving speaking confidence compared to established techniques like role-play is still not available. With a sample of 60 EFL nursing students who were part of the program AKPER YPTK Solok, in West Sumatra, Indonesia, the explanatory sequential mixed-methods study looked into how combining classroom role-play with structured ChatGPT-assisted rehearsal performed against a control group role-playing only. Speaking confidence, willingness to communicate, speaking anxiety, and oral performance were the four factors assessed using modified scales and an analytic speaking rubric, together with interviews of eight students to elaborate upon the quantitative results. A significant increase in spoken confidence was detected in the group with AI assistance by a very small margin. The gains in willingness to communicate and oral performance were relatively small, and as for speaking anxiety, no significant difference was found. The interview data suggest that the learners experienced greater confidence and readiness in class after repeatedly practicing dialogue that was free of any judgment, and they found that a conversation with a person and with AI resulted in significantly different reactions. Based on these findings, it seems that a good use of conversational AI is as a rehearsal tool that develops learners' confidence in speaking and prepares them for authentic human interaction it has not replaced.