Toward Convergence in Student-LLM Interactions: A Rapid Scoping Review and Taxonomy for Learning-Oriented Use

Synthesis: Borchers, Jansen, and Weidlich (2026) conduct a rapid scoping review of how student interactions with large language models are defined and categorized, finding that research remains conceptually fragmented. Across 46 categorizations drawn from 33 studies, they identify substantial variation in data sources, category-construction approaches, and units of analysis, which prevents comparison across studies and understanding of when LLM use supports learning. The review argues for a convergent taxonomy of learning-oriented LLM interactions. It provides a methodological foundation for the Student Experience literature and for interpreting Learning Analytics and self-regulatory evidence across heterogeneous studies.

A Fragmented Field

Generative LLMs are widely used by students, yet research on their role in learning remains conceptually fragmented. Existing studies describe student interactions using heterogeneous and often incompatible categories, ranging from speculative use cases to fine-grained dialogue acts, making it difficult to compare findings across studies and to understand when and how LLM use supports learning.

Method and Variation

This rapid scoping review examines prior work on student-LLM interactions with a focus on how interaction types are defined and constructed. Across 46 categorizations drawn from 33 studies, the authors identify substantial variation in data sources, category-construction approaches, and units of analysis, reflecting divergent theoretical commitments and research purposes.

Toward a Shared Taxonomy

The lack of shared terminology motivates a convergent taxonomy of learning-oriented use. The review situates these interaction categories within Self Regulated Learning and Learning Analytics, supporting Research Methods AIED for synthesizing evidence on AI Feedback Quality and the conditions under which student-LLM engagement produces learning rather than mere completion.

Connected Concepts

• LLM
• Generative AI
• Student Experience
• Learning Analytics
• Meta Analysis Systematic Review
• Higher Ed
• Self Regulated Learning
• Assessment
• AI Feedback Quality

Connected Articles

• Tracing GenAI Literacy Interaction Patterns
• LLM Sentiment Analysis Education Research
• Student Rationalization AI Writing
• Students LLM Usage Critical Thinking
• Learnmate2 LLM Adaptive Learning
• AI Generated Feedback Higher Ed

Citation

Borchers, C., Jansen, S., & Weidlich, J. (2026). Toward convergence in student-LLM interactions: A rapid scoping review and taxonomy for learning-oriented use. EdArXiv preprint.