Proceedings of the Annual Meeting of the Cognitive Science Society

Universal constraints on word meaning apply to both lexical and logical words. Across languages, a well-known gap in the logical vocabulary is that 'not all' is never lexicalized. This gap extends beyond determiners to the modal and temporal domains; e.g. 'not must' and 'not always' are typically not lexicalized (Horn 1973). The challenge is to explain this gap. The non-lexicalization of 'not all' has been explained as resulting from a cognitive bias against intrinsically marked meanings (e.g., Katzir and Singh 2013). Recent alternative accounts, however, have explained this same gap relying on considerations of communicative efficiency rather than cognitive markedness (e.g., Enguehard and Spector 2021). In a series of word learning experiments, we disentangle these views by testing whether learners are more likely to infer that a novel word means 'some' rather than 'not all' and whether this varies depending on the communicative needs in the context.