Abstract
Young and elderly adults listened to spoken passages that were segmented for immediate recall either at natural syntactic boundaries (such as after sentences or major clauses), or at random, nonsyntactic intervals. In addition, speech rate was manipulated using time-compression of the speech materials. Results showed that random segmentation was especially detrimental to the elderly subjects' recall, as was the effect of increasing speech rate. An analysis of subjects' recall errors offered evidence for reconstruction in short-term segment recall in a manner similar to that usually associated with long-term memory.