Automatic access to verb continuations on the lexical and categorical levels: evidence from MEG

Little is known about how readers represent probabilistic information about verb continuations (VCs), or to what degree automatic tracking of these probability distributions constitutes anticipatory processing. Research shows that subcategorization frame entropy, a measure indexing uncertainty about the syntactic constituent following a verb, affects neural responses during verb recognition. This suggests readers have mental representations of VCs. However, the granularity of these representations remains unclear. Using MEG, we investigated how precisely, and to what purpose, the brain represents VCs in single-word processing. We found that, compared to SCF entropy, entropy over a granularized set of continuations (created by subdividing the prepositional phrase frame) better predicted behavioural and MEG responses, suggesting readers access continuations more fine-grained than grammatical category. Results also revealed a spatiotemporally separate sensitivity to the distribution of lexemes following verbs. Overall, our findings suggest that readers represent VCs on multiple levels, and likely engage in anticipatory processing automatically.