Word Fragments as Retrieval Cues: Letter Generation or Search through Nonsemantic Memory?

Abstract

plored. Type of nonsemantic cue during study and test was factorially varied. Words were studied either (a) without modifying cues, (b) in the presence of context cues emphasizing initial letters, or (c) in the presence of contexts stressing endings. Regardless of study emphasis, test cues shared either beginnings or endings with their targets. Results showed that emphasizing either portion during study produced no benefit to recall. However, cue switches from study to test reduced recall. In addition to these manipulations, the category sizes defined by these cues were varied, as was cue embeddedness. In general, cues defining smaller categories were more effective. This result was generally linked to the use of specific cues, i.e., beginningcategory size effects were obtained with beginning but not with ending cues. Embedding the cue within a meaningful word substantially reduced recall. The findings were interpreted within a model of cued recall that assumes that nonsemantic cues operate by defining search sets of nonsemantic memory units. With one exception, the results were consistent with this model. In the exception, the results suggested that subjects used a lettergeneration strategy.