A Comparative Investigation of Study ROI: Multimodal Per-sonalized English Learning Environment Versus Traditional English Learning Environment

Time constraints remain a central bottleneck in university-level English as a Foreign Language (EFL) vocabulary learning. We propose a web-based, AI-driven environment that combines multimodal personalization with a time-based return-on-investment (ROI) framework to evaluate learning efficiency. The system uses a large language model to generate contextualized practice and tutoring, provides pronunciation support via text-to-speech, and visualizes progress through an interactive 3D mastery display to facilitate self-regulated learning. Learner knowledge is represented as a discrete mas-tery state (m ∈ {0, …, 5}) updated after each response, and an adaptive scheduler al-locates practice across mastery strata to prioritize fragile knowledge while maintaining review stability and preserving score validity when answers are revealed. Learning ROI is quantified as newly mastered words per unit study time, computed from logged be-havioral traces of practice time and mastery transitions. In an initial deployment, learners mastered more words than with conventional vocabulary practice under comparable time budgets, while the multimodal design supported engagement beyond isolated word recall, particularly for listening-oriented rehearsal. These findings offer an implementable blueprint for reliable generative-learning workflows and position time efficiency as a first-class target for evaluation.