CAUSAL CONNECTIVES INCREASE INFERENCE GENERATION

The influence of interclause connectives on inference generation was examined in three experiments. In Experiment 1, participants supplied lexical decisions on inference words following the word-by-word presentation of sentences containing the connective because and when the clauses were presented as two independent sentences (i.e., no connective). The results indicated that the causal knowledge-based inferences were generated in the connective condition, but not in the no-connective condition. Experiment 2 examined whether this finding would generalize to an additive connective (i.e., and). This experiment replicated the results of Experiment 1 for because, but there was little evidence that and had elicited inferences. In Experiment 3, the temporal connective after was examined. The results indicated that after did not produce causal-based inferences, suggesting that the effect of because was not due to temporal cuing. The pattern of results across the experiments indicate that readers incorporate causal knowledge-based inferences in the discourse representations for sentences containing a causal connective and support the hypothesis that connectives elicit inferences which are based on the connective's meaning. The findings are discussed in the context of previous research on connectives and the connective integration model (Millis & Just, 1994).