Semantic Implicit Learning in Language

Previous studies of semantic implicit learning in language have only examined learning grammatical form-meaning connections where learning could have been supported by prior linguistic knowledge. Also, these studies assessed awareness by verbal report, which is arguably not the most reliable measure. Here we target the domain of verb meaning, specifically semantic preferences of novel verbs (e.g. a novel verb takes abstract objects). Using a reaction time methodology we show that after exposure to correct verb-noun combinations, reaction times to incorrect combinations are slowed down even for participants who are unaware of the semantic regularity. This effect was also obtained even when the semantic regularity was irrelevant to the tasks being performed, suggesting that the semantic generalisation is learned and exerts its influence automatically, hence satisfying one criterion for implicitness. Combined with a lack of verbalisable knowledge in any participant these experiments provide strong evidence for semantic implicit learning in language.