Predictive language processing revealing usage-based variation

While theories on predictive processing posit that predictions are based on one’s prior experiences, experimental work has effectively ignored the fact that people differ from each other in their linguistic experiences and, consequently, in the predictions they generate. We examine usage-based variation by means of three groups of participants (recruiters, job-seekers, and people not (yet) looking for a job), two stimuli sets (word sequences characteristic of either job ads or news reports), and two experiments (a Completion task and a Voice Onset Time task). We show that differences in experiences with a particular register result in different expectations regarding word sequences characteristic of that register, thus pointing to differences in mental representations of language. Subsequently, we investigate to what extent different operationalizations of word predictability are accurate predictors of voice onset times. A measure of a participant’s own expectations proves to be a significant predictor of processing speed over and above word predictability measures based on amalgamated data. These findings point to actual individual differences and highlight the merits of going beyond amalgamated data. We thus demonstrate that is it feasible to empirically assess the variation implied in usage-based theories, and we advocate exploiting this opportunity.