Abstract
We contrasted four possible views of how causal connectives function during reading short texts: and end-of-clause reactivation view, a connective-elicited inference view, a resource-saving view, and a no-connective stimulating view. In Experiment 1, participants read a series of popular scientific texts joined by because or separated by a full stop. The on-line word-completion task used to measure inference activation revealed that although causal inferences are activated during reading, they are not enhanced by the presence of a connective. In Experiment 2, second-clause reading times data and memory data (prompted delayed recall test) revealed specific patterns of results according to the nature of the relation conveyed by the connective (additive vs. causal). We found evidence that only because gave rise to an immediate presence effect associated with faster end-of clause reading times. This supported the view that because enhanced the on-line integration process rather than the inferential process.