The emergent generation effect and hypermnesia: influences of semantic and nonsemantic generation tasks.

The generation effect is moderated by experimental design, affecting recall in within-subjects designs but typically not in between-subjects designs. However, N. W. Mulligan (2001) found that the generation effect emerged over repeated recall tests in a between-subjects design, calling into question the generality of this limiting condition. In addition, the generate condition but not the read condition produced hypermnesia (increased recall over tests). The present experiments demonstrate that semantic-based (semantic-associate and category-associate) generation tasks produce this pattern of results whereas nonsemantic (letter transposition, rhyme, word fragment) generation tasks do not. Thus, the emergent generation effect appears to be a byproduct of semantic elaboration rather than a direct product of generation. In addition, high- and low-imagery words produced equivalent hypermnesia and emergent generation effects, arguing against a mediating role for imagistic encoding. Finally, there is no evidence of an emergent generation effect for nonwords, another traditional limiting condition of the generation effect.