Manuscript received April 16, 2026; accepted May 17, 2026; published May 26, 2026.

Abstract—This research investigates the use of Artificial Intelligence (AI)-powered chatbots on English speaking proficiency and speaking anxiety among English as a Foreign Language medical and nursing students in a Gulf Cooperation Council higher education context. Prior studies report positive results from AI-mediated speaking practice but significant research gaps remain. Existing literature primarily focuses on non-medical student populations on ready-made AI language apps that are not easily adaptable by instructors. Existing literature rarely addresses speaking proficiency and speaking anxiety concurrently. The Gulf Cooperation Council is a promising area of research since given that few studies in this region explicitly articulate how large language models are pedagogically guided through structured prompting. This leaves an underexplored gap in prompt engineering for language learning in the region. To address these gaps, this research will employ a pretest–posttest control group design that involves foundation program medical and nursing students at the Arabian Gulf University in the Kingdom of Bahrain. The experimental group will engage in structured speaking practice using Grok Voice Mode, an AI-powered conversational chatbot, while the control group will follow the standard curriculum. Data will be collected using standardized English speaking proficiency assessments and an adapted Foreign Language Speaking Anxiety Scale administered before and after the intervention. In addition to measuring learning outcomes, this research aims to develop and validate a structured prompt engineering framework to guide learners’ interactions with Grok. This framework will deploy pedagogical objectives into customizable prompts aligned with speaking components, proficiency levels, and anxiety-reduction strategies. The findings aim to contribute theoretically to AI integration in language learning, English as a Foreign Language curriculum design, and evidence-based prompt engineering practices.