Generation Effects With Nonwords: The Role of Test Appropriateness

The generation effect refers to the superior memory that is found when to-be-remembered items are subject produced rather than experimenter produced. Our theoretical interpretations of this phenomenon have been constrained by the fact that experimenter-created nonwords, when generated, fail to produce any retention advantage over a simple read or copy condition. We account for this failure by demonstrating that a robust generation effect can occur with meaningless items, as long as our retention tests show sensitivity to what subjects actually generate in the nonword case. In Experiment 1, which used the switching of two letters as the generate task, subjects showed better memory for generated nonwords when the retention test assessed memory for which letters had been switched; in Experiment 2, a generation effect for nonwords emerged when subjects were required to generate items again at test, prior to the recognition decision. We suggest that when subjects generate nonwords, the functional product of generation is often the fragment of the entire response unit that is actually manipulated. A generation effect with nonwords will emerge only when the subject is able to use this generation product to mediate performance on a retention test.