Tracing the development of multimodal representational competence in science reading: a longitudinal eye-tracking study across grades 4–6

Reading science texts requires students to actively engage in meaning-making processes by coordinating multiple semiotic resources, specifically texts and diagrams. However, the development of multimodal representational competence remains largely unexplored. To address this gap, this three-year longitudinal eye-tracking study examined the trajectories of science reading comprehension and eye-movement indicators of multimodal representational competence among 175 students from Grade 4 to Grade 6. Utilizing latent basis growth curve modeling, the analysis revealed a distinct non-linear developmental mismatch between students’ multimodal navigation strategies and their comprehension scores. From Grade 4 to Grade 5, text-to-diagram saccades per page decreased slightly, and comprehension temporarily declined. Conversely, by Grade 6, these integration strategies plateaued while comprehension improved substantially. Furthermore, text mean fixation durations grew progressively longer across the grades, consistent with a strategic slowing to accommodate the escalating conceptual difficulty of science materials. Additionally, baseline reading ability positively predicted initial comprehension but did not significantly predict the growth slope of integration strategies. These findings suggest that text-diagram integration has yet to become habitual in these grades. Consequently, effective science education necessitates explicit pedagogical interventions targeting multimodal representational competence to bridge general literacy with domain-specific scientific meaning-making.