The Plausibility of Semantic Properties Generated by a Distributional Model: Evidence from a Visual World Experiment

Distributional models of semantics are a popular way of cap-turing the similarity between words or concepts. More re-cently, such models have also been used to generate prop-erties associated with a concept; model-generated properties are typically compared against collections of semantic feature norms. In the present paper, we propose a novel way of testing the plausibility of the properties generated by a distributional model using data from a visual world experiment. We show that model-generated properties, when embedded in a senten-tial context, bias participants ’ expectations towards a semanti-cally associated target word in real time. This effect is absent in a neutral context that contains no relevant properties