From metrics to meaning: large language models and the computational turn in embodied educational research

Educational science has long grappled with a methodological tension: quantitative metrics offer scale but often lack depth, while qualitative inquiry offers depth but is difficult to scale. Mixed-methods approaches have sought to address this tension by combining both paradigms, yet practical constraints of time, labor, and analytical capacity have typically limited how fully researchers can integrate interpretive depth with large-scale analysis. At the same time, theories of embodied cognition emphasize that learning, language, and communication are grounded in sensorimotor, affective, and socio-cultural experience rather than in abstract, amodal symbols. The emergence of Large Language Models (LLMs) and multimodal Artificial Intelligence (AI) provides new opportunities to bridge these methodological and theoretical developments. In this article, we conceptualize “Computational Hermeneutics” as the interpretation of meaning at scale, informed by embodied perspectives on cognition, language, and communication. We outline three computational methodologies applicable to educational research: Semantic Similarity Rating (SSR), AI-based Qualitative Content Analysis (AI-QCA), and Computational Ethnography via multimodal video analysis. We show how these approaches can operationalize embodied, hermeneutic processes, such as interpreting student reflections, tracing metaphorical embodiment, or analyzing classroom habitus, at a scale previously reserved for standardized testing. By detailing the theoretical basis, opportunities, and limitations of these methods, we argue for a “computational turn” in educational science and research that remains faithful to embodied, multimodal meaning-making while overcoming traditional scalability constraints.