Contextual Specificity in the Activation of Predictive Inferences

Abstract

There is general agreement that predictive inferences are activated when there is strong contextual support in the discourse model; however, there has been debate concerning the specificity of these inferences. In a series of 3 experiments, the specificity of context was manipulated to test the effects of contextual support on inference activation. Experiment 1 demonstrated that with less specific context, activation of predictive inferences could be captured by more than 1 lexical item. Experiment 2 confirmed that this outcome was a result of inference activation and not semantic priming. Experiment 3 demonstrated that as contextual support increases, inference activation becomes increasingly constrained. Results of all 3 experiments support the assumption that inferences are not activated in an all-or-none process.