A unit analysis of prose memory

Abstract

Four stories were divided into function word units. These units were assigned dependent variable values determined by the scoring of subjects' recalls and independent variable values determined by measures of gist, imagery, repetition, frequency of occurrence, serial position, grammatical connectedness, centrality in a propositional net, and subjects' intuitions of which units would be remembered. The independent variables were all statistically significant predictors of recall. Subjects' intuitions and gist were the best predictors of the more structured stories, while repetition and serial position were the best predictors of the less structured stories. For each story, the underlying rank ordering of function word units from most to least likely to be remembered was the same for all subjects (i.e., scalable). While changes in the retention interval, subject population, and motivation level affected the amount recalled, these changes had little affect on the rank ordering of the units from most to least likely to be remembered. Changes in the retrieval task from free recall to prompted recall and recognition affected both the amount and rank ordering of units. © 1978 Academic Press, Inc.