How expert teachers promote students’ deep learning: an analysis based on teacher-student dialogue

Research Question: This study aims to identify the types of teacher discourse associated with students’ deep learning and to explore how expert teachers adjust their discourse strategies to promote such learning. Methodology: The study employed three instruments for data analysis: the Teacher Discursive Moves (TDM) coding catalogue, the Structure of Observed Learning Outcomes (SOLO) framework, and Lag Sequential Analysis (LSA) to analyze mathematics classroom videos from four expert teachers. Through this mixed-methods approach, we systematically examined the interplay between teacher discourse and student cognitive development. Key Findings: The results revealed that specific discourse behaviors—such as probing, asking for alternative points of view, requesting clarification, direct affirmation, consulting for negotiation flow, and prompting evaluations—were closely linked to different levels of deep learning as classified by the SOLO framework. Furthermore, three discourse interaction structures were identified as particularly effective in facilitating deep learning: (1) PRO → M → DA → PRO → C, (2) CNF → M → DA → APV → C, and (3) APV → M → DA → APV → A, (4) APV → M → DA → PLE → CNF → A. These patterns illustrate how expert teachers adaptively scaffold students’ progression from surface-level understanding to deeper, more abstract cognitive engagement. Contributions: By innovatively integrating LSA and the SOLO framework, this study moves beyond traditional static analyses of classroom interaction, offering a dynamic account of how teacher discourse promotes deep learning. The findings provide new theoretical insights and practical implications for optimizing instructional discourse to foster students’ cognitive development and engagement.