Memorability functions in verbal memory: a longitudinal approach.

Middle-aged (N = 252, M = 39.91), young-old (N = 486, M = 60.77), and old-old (N = 137, M = 74.42) participants in the Seattle Longitudinal Study were tested on two occasions on their recall of a 20-item word list. Proportions of participants in each group correctly recalling each word-unit served as the dependent measure. Word-unit scores obtained in 1991 were regressed on those from 1984, yielding linear functions that varied by age group. Each set of word-unit scores (three groups on two occasions) was then regressed on word familiarity, imageability, primacy, and recency. The relative influence of each of these variables on memorability was then compared, both between and within cohorts. Primacy and familiarity were consistently strong predictors. Imageability and recency were predictive of memorability in the middle-aged group, but less so in the young-old, and not at all in the old-old group. Results and possible implications are discussed.