Advancing Verbal Communication Skills Through Transcript-Based Feedback : Designing an AI-Assisted Language Learning Application, “Stepping Stones, ” to Support ESL Students in Academic and Professional Contexts

Abstract

This thesis investigates communication challenges faced by English as a Second Language (ESL) learners from high-context language backgrounds, such as Korean, Japanese, and Chinese, in online academic and professional settings. Cultural norms around indirectness, shared context, and cautious self-expression often influence how learners communicate in English, leading to subtle but persistent barriers to clarity, participation, and confidence. Adopting a research-through-design approach, the study combines interviews, diary studies, and online meeting transcript analysis to surface recurring pain points in learner speech. These insights informed the development of Stepping Stones, an AI-powered learning application that provides context-aware feedback based on learners’ actual spoken interactions. While many conventional language learning tools often emphasize grammar instruction and scripted practice, this system draws on learners’ real conversational data to provide personalized feedback and lesson flows. By analyzing naturally occurring speech rather than pre-written responses, it addresses individual communication patterns shaped by cultural context and supports growth in expressive clarity, pragmatic fluency, and habit awareness. This thesis presents a design-driven exploration of how culturally rooted communication barriers might be addressed through personalized feedback grounded in real learner speech, aiming to raise new possibilities for how interaction design can respond to the needs of high-context ESL learners.