Research on Generative AI-Empowered Formative Assessment of Maritime English: Evaluation Indicators and Practical Paths

Abstract

The Manila Amendments to the STCW Convention set compliance requirements for seafarers’ maritime communicative competence; however, higher-vocational Maritime English assessment still relies mainly on summative final-term evaluation, with high oral-assessment costs and delayed feedback. Drawing on the Outcomes-Based Education (OBE) philosophy and formative assessment theory, and supported by generative AI and intelligent speech-judgment technology, this study constructs a three-dimensional evaluation indicator system and designs a human-AI collaborative assessment closed loop of “pre-class diagnosis, in-class dynamic monitoring, and post-class iterative feedback”. A 16-week quasi-experiment shows that this model significantly improves students’ maritime English communicative competence, with the largest gain in the oral dimension (18.7% improvement in the experimental class versus 5.3% in the control class), and enhances students’ learning initiative. The study also analyzes the validity defects and academic-integrity risks of AI-driven assessment and proposes human-AI collaborative verification mechanisms, providing practical references for curriculum assessment reform of higher-vocational Maritime English and for STCW-aligned talent training.