Adult age differences in memory and linguistic integration revisited.

In recent studies questioning whether linguistic integration of semantic information declines with age, the performance measures were based on recognition of related sentences in the Bransford/Franks paradigm; and the conclusions about age differences in linguistic integration were based on equivalent recognition performance of young and old adults. In the current study, the measure of linguistic integration was based on inference in a linear ordering task. Subjects were students at the University of Maryland - 34 between 18 and 23 years of age and 33 between 60 and 72. Recognition performance was equivalent for old and young groups, but linguistic integration was age related. Young subjects inferred the truth status of more sentences than did old subjects. These findings indicate that linguistic integration based on inference is age related and that equivalent recognition performance is not adequate evidence for equivalent linguistic integration. Furthermore, recent findings indicate that age deficits in making inferences using new information presented auditorily are due to deficits in memory for facts and in working memory. This study supports those findings and indicates that they can be generalized to information presented visually.