Abstract
Encountering a homonym in a sentential context that biases interpretation towards its subordinate meaning makes that meaning easier to access later. This word-meaning priming effect is not restricted to homonyms and may be supported by general episodic memory processes. Such an account predicts that word-meaning priming may be affected by factors that affect episodic learning (e.g., predictability). Specifically, the contextual predictability of an incoming stimulus has been shown to affect episodic memory in that both highly expected as well as highly unexpected input leads to better memory for that input, via two different underlying neurocognitive mechanisms. We hypothesised that if word-meaning priming is supported by episodic memory processes, both highly expected as well as highly unexpected target words should lead to stronger word-meaning priming effects than target words of intermediate predictability. Four pre-registered online experiments tested whether contextual predictability affects word-meaning priming and incidental memory for language. We exposed participants to sentences that emphasised a particular aspect of a sentence-final target word’s meaning. Importantly, target words differed in how predictable they were based on the sentence context (e.g., “You can get in a good workout by riding / lifting a bicycle”). Associate production and semantic relatedness judgement assessed the strength of word-meaning priming. Contrary to our hypotheses, while there was evidence for priming across experiments, we found no significant effects of contextual predictability in any experiments. Our findings suggest that predictability-guided updating is rarely triggered during language comprehension, and we outline the theoretical implications of these results for the architecture of the language system.