Four versions of several paragraphs had the same second sentence and were referentially coherent, but they differed in causal relatedness of the two sentences. Results showed that despite referential coherence, recognition and recall memory for the causes was poorest for the most and least related causes and best for causes of intermediate…

Twenty-three subjects drew deductions from ten different pairs of conversationally conveyed requests, of which the first conveyed a positive request, the second, a negative. Some evidence shows the listener constructs the literal meaning before the conveyed meaning. (CK)

Memory representations of statements from a story were predicted to cluster into separate episode chunks in memory. It was shown that the recall of episode depends on the length of that episode, but not on the lengths of other episodes. The chunking idea was confirmed. (SW)

Three experiments were conducted to ascertain the effect of contextual restraints on pronoun assignment. Pronoun selection is based on integration of the context even where it is already syntactically constrained. Integration occurs during and not following the assignment of the pronoun. (PMJ)

Five experiments were conducted that illustrate microprocesses of discourse comprehension related to foregrounding. An intermediate case in which prior memories exist but must be reinstated to active memory was studied in detail. (SW)

A schema-based framework for representing knowledge and prose organization was studied. Testing of a script pointer and tag hypothesis confirmed that memory discrimination is better for atypical actions in a passage than for typical script actions and that there is no memory discrimination for very typical actions. (SW)