Wen Liang, Li Siyan, Zackary Rackauckas, Julia Hirschberg (2026) — Computation and Language (cs.CL). 📄 Full text (arXiv)
This survey provides the first systematic review of automated presentation coaching systems, organizing them along a five-dimensional task taxonomy: segmental pronunciation, lexical stress, suprasegmental prosody, pacing, and content faithfulness. The authors review systems spanning pronunciation tutors, fluency and prosody coaches, multimodal trainers, and conference Q&A practice tools, identifying coverage gaps in pacing and content feedback dimensions.
The survey connects to intelligent tutoring systems by examining how real-time feedback delivery mechanisms in coaching systems mirror educational feedback principles. The taxonomy serves as a framework for professional skills training, particularly for second language learners, and highlights the role of scaffolded feedback in skill development.
Key open challenges include the scarcity of annotated presentation corpora, achieving accent-fair feedback across diverse L1 backgrounds, and delivering real-time multimodal feedback without overwhelming the learner — all of which connect to broader personalized learning and AI literacy concerns in education technology design.
Related Pages
- feedback-loop — Feedback loops in education
- scaffolding — Scaffolding in AI tutoring
- intelligent-tutoring — Intelligent tutoring systems
- professional-training — Professional training applications
- language-learning — Language learning with AI
- personalized-learning — Personalized learning systems
- ai-literacy — AI literacy concepts and programs