Implicit Memory and Word Ambiguity

Abstract Previous research has indicated that in the cued recall task, words with few preexisting associates are more likely to be recalled than words with many preexisting associates. Since this research has avoided using ambiguous words as stimuli, it was unclear if this set size effect would generalize to ambiguous words. In a series of three experiments, the set size effect was demonstrated using ambiguous words. The magnitude of the effect was reduced for homographs relative to nonhomographs and more apparent when dominant, rather than subordinate, meanings were accessed by the retrieval cues. These results indicate that meaning can be selected at retrieval by the retrieval cue, and that the associates of the dominant meaning play a more critical role in retrieval than do the associates of the subordinate meaning. Additionally, the effect of encoding context was consistent across experiments and demonstrated that, when the encoding context biased the to-be-tested meaning of the homograph, performance was facilitated relative to when the encoding context biased a different meaning.