Аңдатпа
Problem-based Learning (PBL) is widely used to promote problem solving, yet outcome measures alone provide limited insight how students reason, negotiate ideas, and regulate inquiry while solving authentic problems. This study examined patterns of problem-solving discourse across repeated synchronous written online PBL discussions among pre-university physics in Malaysia. A comparative two-semester qualitative content analysis, supported by descriptive discourse coding, was conducted using 18 Microsoft Teams discussion sessions: three groups per semester and three discussions per group. Semester 1 represents 21 students and Semester 2 represent 22 students. The corpus comprised 890 substantive students turns and 1390 PIM eligible meaning units. Cognitive presence was classified using the Practical Inquiry Model (PIM), while supplementary turn-level indicators captured mathematical reasoning, evidence use, concept integration, uncertainty awareness and limitation awareness, peer interaction, reflection, and peer facilitation. Semester 1, showed a modest rise in the study-specific Higher-Order Inquiry Ratio (HOIR) (Integration plus Resolution), from 27.8% in Discussion 1 to 33.6% in Discussion 3. Semester 2 displayed different pattern, beginning at 41.8% and stabilizing at approximately 31% in later discussions. Group-level trajectories were heterogeneous, indicating that semester averages did not represent a uniform progression. Across the two semesters, concept integration and explicit attention to uncertainty and limitations were prominent in the second semester corpus, whereas mathematical and evidence-based reasoning varied with problem demands. Peer correction also became more visible in later Semester 2 discussions. The findings suggest that synchronous written PBL can make the structure of scientific problem-solving observable, while also showing that inquiry patterns are shaped by task and group context. The study highlights the value of analyzing how students’ reason together, rather than relying only on final answers or participation counts.