Embodied Semantics and Prior Knowledge Jointly Support Learning of Abstract Concepts

Figurative language and prior knowledge both link unfamiliar concepts to familiar experiences, but their combined role in real-time learning remains unclear. We tested a prior-knowledge scaffolding account, predicting reliance on domain-specific semantic activation, against embodied-cognition accounts, which predict that figurative language supports learning through sensorimotor simulations. In a visual-world eye-tracking study, 104 Mandarin speakers learned new science concepts from spoken literal or figurative explanations while viewing a prior-knowledge associate, a sensorimotor-rich synonym, and a distractor. Gaze data showed that literal explanations enhanced fixation on the prior-knowledge competitor. Figurative explanations showed a dynamic shift in knowledge activation: early attention to the sensorimotor synonym, followed by simultaneous engagement of sensorimotor and prior knowledge. Early sensorimotor activation predicted better outcomes, whereas late reliance on sensorimotor activation predicted poorer performance. Overall, learning abstract concepts requires coordinating knowledge sources, so figurative teaching is most effective when combining cues that aid knowledge integration.