Empowering Vocabulary Learning Through Teaching AI: Using LLMs as a Student to Perform Learning by Teaching in Vocabulary Acquisition

Synthesis: Uchida et al. (2026) develop an LLM-based system that lets English vocabulary learners learn by teaching an AI "student." The system generates dynamic, contextually relevant questions for learning-by-teaching, replacing rigid template-based question generators. In a study with ten participants, learning with the system produced high learning effectiveness and improved memory retention at 3 and 7 days after learning, and the researchers identified learner traits linked to better outcomes.

The Approach

Learning-by-teaching helps learners deepen understanding by explaining concepts to others, with questions playing a vital role in identifying knowledge gaps and reinforcing comprehension. However, existing systems for generating such questions rely on rigid templates and are expensive to build. This work uses LLMs to create dynamic, contextually relevant questions for learning-by-teaching in English vocabulary acquisition — effectively using the LLM as a student the learner teaches.

Key Findings

• High learning effectiveness. Learners using the proposed system showed improved performance versus a baseline vocabulary-learning system, with test-score distributions shifting positively.
• Improved retention. Results showed improved memory retention at 3 and 7 days after learning.
• Learner traits matter. The study identified traits linked to better learning outcomes, highlighting potential for tailored/adaptive approaches.
• Scalable and cost-effective. The LLM-based approach supports scalable, cost-effective learning-by-teaching across fields, addressing the expense and rigidity of template-based systems.

Implications

1. LLMs as teachable students can operationalize learning-by-teaching in language learning without expensive bespoke systems.
2. Dynamic question generation from an LLM supports contextual relevance that rigid templates cannot match.
3. Personalization potential — identifying learner traits that benefit most points toward adaptive learning-by-teaching.

Connected Concepts

• Learning By Teaching
• Generative AI
• Language Learning
• Student Experience
• Scaffolding

Connected Articles

• Chatgpt Teachable Agent Programming Lbt 2024 — ChatGPT as a teachable agent in programming
• Prompting Teachability Novice Personas Lbt 2026 — Designing novice personas for teachability

Citation

Uchida, T., Watanabe, K., Vargo, A., Ishimaru, S., Rose, R. L., Sugawara, A., Dengel, A., & Kise, K. (2026). Empowering Vocabulary Learning Through Teaching AI: Using LLMs as a Student to Perform Learning by Teaching in Vocabulary Acquisition. In Augmented Humans International Conference 2026 (AHs '26).