Pattern mining uncovers social prompts of conceptual learning with physical and virtual representations

To succeed in STEM, students need to connect visual representations to domain-relevant concepts, which is a difficult task for them. Prior research shows that physical representations (that students manipulate with their hands) and virtual representations (that they manipulate on a computer) have complementary advantages for conceptual learning. Further, physical and virtual representations are often embedded into different social classroom practices. Thus, to optimally combine these representation modes, we need to understand what social events prompt students to connect representations to concepts, and if different representation modes afford different social prompts. A multiple-case study with 12 high-school students addresses this question. Student pairs worked with physical and virtual representations of chemistry. Frequent patterns obtained from discourse data show that students incrementally co-construct concept-representation connections, and that instructor prompts are key triggers of these connections for both representation modes. Meta-cognitive statements serve as important prompts in the absence of an instructor when students work with virtual representations. I discuss implications for interventions that combine physical and virtual representations.