The study of adult age differences in comprehension of and memory for text is a now burgeoning enterprise in cognitive gerontology, in part because of the potential for direct application of the findings to everyday life. To date, the work on discourse processing suggests the existence of age deficits of varying magnitudes, deficits that are largely quantitative rather than qualitative in nature. The work thus suggests that older adults use the same processing mechanisms as younger adults but with poorer results (e.g., Mandel & Johnson, 1984; Zelinski, Light, & Gilewski, 1984).

Beyond this summary the literature yields few simple generalizations; indeed, the findings on any given variable (e.g., educational level) tend to be complex and inconsistent. Consider the literature on the recall of ideas that differ in their importance to the meaning structure of the text. The usual finding with young adults (called the “levels effect”) is that the probability of recalling information from text is directly related to the information's importance level in the text as defined by a model (e.g., Kintsch's, 1974) of the hierarchical structure of that text. When young and elderly adults have been compared, different experiments have produced contradictory results (for a recent review, see Zelinski et al., 1984; see also Cohen, this volume). The most frequent findings are (1) parallel levels effects for younger and older adults (e.g., Zelinski et al., 1984); or (2) an exaggerated levels effect for the older adults, with the greatest age deficit seen at low importance levels (e.g., Dixon, Hultsch, Simon, & von Eye, 1984, for high verbal ability subjects; Spilich, 1983).