Exploring science teachers’ questioning patterns for learning support through a computational grounded theory approach

Teacher questioning is pivotal for students’ conceptual construction and cognitive growth, yet prior studies rely on predefined taxonomies and static analyses. Using a computational grounded theory (CGT) approach that integrates lag sequential analysis (LSA) and pretrained language model–based semantic representations, we systematically examine questioning types and patterns in 32 upper-secondary physics lessons. We identify 19 types organised across three levels – behavioural regulation, cognitive development, and reflective questioning – and two recurring dynamic patterns: cyclical reinforcement and progressive scaffolding. These patterns guide key cognitive processes, including conceptual elaboration, causal reasoning, experimental design, and reflective integration. The study illustrates the value of CGT for semantic analysis of classroom discourse and, when combined with LSA, offers a data-driven yet theoretically grounded pathway to model questioning dynamics. Findings refine theoretical accounts of teacher questioning as a multilayered cognitive scaffold and provide empirical evidence on how exemplary science teachers orchestrate diverse questions to support learning.