Abstract
Our brains house representations of word meaning that are constrained by both the form and the sentence context of the word. For example, when hearing the sentence “They deposited cash at the bank”, we do not think about the bank of a river. These context-dependent meaning representations enable cognitive flexibility. Existing studies aimed at isolating such meaning representations have confounded them with operational processing requirements, due to reliance on inferential and surprisal effects during natural language comprehension. We introduce a novel symbol-based procedure for disentangling context-sensitive word meaning representations from the processing demands of word meaning inference. Participants were instructed to learn associations between sentence-constrained meanings of homonyms (“They deposited cash at the [bank]”) and symbols. Explicit word meaning representations were successfully captured by a task that required the spatial arrangement of these symbols based on their associated word meaning, with smaller distances between symbols representing the same meaning than between symbols representing different meanings. Contrary to the spatial arrangement task, in a priming task assessing implicit word meaning representations related to the same symbols, there was no evidence for semantic meaning priming of reaction times. This work provides a novel experimental procedure for measuring explicit word meaning representations in a manner that can be disentangled from changes in word meaning inference, but raises questions about the suitability of the priming task used for uncovering implicit meaning representations.