Context Can Constrain Lexical Access. Implications for Models of Language Comprehension

When people encounter polysemous words (i.e., words with two or more relatively common meanings), how is the contextually appropriate meaning selected? Contextual information could be used in either one of two general ways: (a) to limit initial meaning activation to a single contextually appropriate sense; (b) toguide choice of an appropriate sense only after all available meanings are activated. Interactive models of language comprehension favor the first, the context-constraining alternative; noninteractive, strictly bottom-up models favor the context-independent alternative. We demonstrate that context can constrain initial meaning activation, limiting lexical access to a single, contextually appropriate sense. This supports the general notion that language comprehension involves interactions among levels of speech and discourse analysis, at least at the level of word recognition. Specifically, lexical information is supplemented by and integrated with sentence-level contextual information toguide and constrain initial stages of word recognition. © 1986 American Psychological Association