ERP representational similarity analysis reveals the prediction of semantic features in minimal phrasal contexts

Abstract

Existing studies demonstrate that comprehenders can predict semantic information during language comprehension. Most evidence comes from a highly constraining context and it is less investigated that whether individuals predict following semantic information in a less constraining context. In the present study, we investigated semantic prediction when reading minimal adjective-noun phrases using electroencephalography (EEG) combined with representational similarity analysis (RSA). Native Chinese Mandarin comprehenders were presented with animate-constraining or inanimate-constraining adjectives, followed by animate-congruent or animate-incongruent nouns. EEG amplitude analysis revealed an N400 for incongruent conditions. Critically, we quantified the similarity between patterns of neural activity, and animate-constraining adjectives revealed greater similarity than inanimate-constraining adjectives before the presentation of the nouns. This pre-noun similarity effect suggests pre-activation of animacy-related semantic information of nouns, and provides evidence for the prediction of semantic features of upcoming words, even in minimal phrase contexts.