Lelia M. Glass

This dissertation explores a linguistic inference pattern called distributivity. Told that Alice and Bob smiled, one infers that they each smiled (distributive); whereas, told that Alice and Bob met, one does not infer that they each met (nondistributive). Told that Alice and Bob lifted the table, one is not sure whether they each did so (distributive) or did so jointly (nondistributive). This dissertation defends an analysis attributing these inferences not to any difference in the logical representations of these sentences, but rather to extralinguistic knowledge about the events they describe. To make this analysis explanatory, it investigates which predicates are understood distributively and/or nondistributively, and why. The project then identifies large-scale patterns in the understandings available to different classes of predicates, and grounds these patterns in an analysis of the events they represent, uncovering the way knowledge about the world creates patterns across the lexicon people use to portray it.