Processing category terms in context: Instantiations as inferences

Three experiments were conducted to isolate conditions in which readers make their representations of text more concrete by instantiating (inferring a specific instance of) category terms. In previous research using single-sentence contexts and a modified Stroop task as an on-line measure of comprehension, no evidence for instantiation was obtained. In the present experiments, the modified Stroop task was used to test for instantiation when category terms were referred to anaphorically in subsequent sentences and when the category terms served as the subjects of single-sentence contexts. The pattern of the Stroop effects was dependent upon whether typical or atypical exemplars were suggested by the context. Facilitation of color naming was obtained in the former case, and interference was obtained in the latter case when the instances followed the appropriate biasing contexts. The results highlight the role of foregrounding and context in guiding the use of elaborative inferences.