From the concreteness advantage to semantic diversity: semantic precision as a predictor of word recognition

Concrete words are easier to process than abstract words, which has been taken as a sign that word meaning involves perceptual information. However, Bottini et al. (2021) showed that blind people also process concrete visual words faster, making an embodied interpretation of the effect entirely unviable. Here we investigate the possibility that abstract words are more difficult to process because their meaning is generally wider—while a “chair” is always an object to sit on, a “brave” individual might be a fireman saving lives or a child giving a presentation to the class for the first time. We quantified this idea of semantic precision through semantic diversity —the dissimilarity of contexts in which a word appears (Hoffman et al., 2013). We refined the computation of semD using more advanced computational models (Mikolov et al., 2013) and separated it from several established predictors (e.g., contextual diversity, valence) using Principal Component Analysis. We then reanalysed Bottini et al.’s (2021) data and assessed the role of semD in a large database (Brysbaert et al., 2019). The results revealed an ambiguous connection between concreteness and semD, but showed that the latter has an independent, statistically solid and inhibitory effect on word processing times.