Proceedings of the Annual Meeting of the Cognitive Science Society

People vary in what they know, yet models of languageprocessing do not take this variability into account. Weharnessed the temporal sensitivity of event-related brainpotentials alongside individual differences in Harry Potter (HP)knowledge to investigate the extent to which the availabilityand timing of information relevant for real-time wordcomprehension are influenced by variation in degree of domainknowledge. We manipulated meaningful (category, event)relationships between sentence contexts about HP stories andcritical words (endings), assessed via behavioral ratings and bymeasuring similarity of word embeddings derived from a high-dimensional semantic model trained on HP texts. Individuals’ratings were sensitive to these relationships according to thedegree of their domain knowledge. During reading, N400amplitudes (neural measures of semantic retrieval) alsoreflected this variability, suggesting the degree to whichinformation relevant for word understanding is availableduring real-time sentence processing varies as a function ofindividuals’ domain knowledge.