The dependency of schema formation on type of verbal material: Linear orderings and set inclusions

Two paragraph structures, linear orderings (Potts, 1972) and set inclusions (Frase, 1969), were equated on relevant independent variables to examine recognition memory for relationships between pairs of terms. Experiment 1 replicated Potts' findings: Proportion correct was greater on remote than on adjacent linear pairs. For set pairs, however, there was an interaction between remoteness and truth, giving better performance on remote false and adjacent true pairs. It was suggested that the memory representation (schema) for linear structure was predicted by Potts' end-term anchoring and rating-scale strategies, while set structure performance reflected logical errors in processing. Experiment 2, an "open book" test, confirmed this view: In the nonmemory task, there was no evidence of schema formation for linear structure, whereas set performance still merely reflected logical errors in processing.