Abstract. Adult subjects were given concrete and abstract textbook passages to study by using either an imaginai or verbal strategy. Two days later, they were given a multiple-choice test and a production test of comprehension. The verbal strategy produced better comprehension than the imaginai strategy; concrete passages were comprehended better than abstract passages, but only according to the production test; and strategy and concreteness did not interact. Differences between these results and results obtained in imagery research on word lists are discussed, and caution is advised before generalizing the word research to meaningful prose learning by adults.The effectiveness of imagery for facilitating recall of word lists has been amply documented (e.g., Paivio, 1969Paivio, , 1971. In general, concrete words are remembered .better than abstract, and when subjects are instructed to use imagery as a mnemonic strategy, they often remember more words than when they use other strategies, especially rote repetition. Moreover, these effects of concreteness and strategy have been shown to interact. Paivio and Foth (1970) showed that for concrete words, an imaginai strategy produces better recall than a verbal mediational strategy, whereas for abstract words, the verbal strategy is better than the imaginai. This interaction is important theoretically because it provides support for distinguishing between an imaginai system of representation specialized for concrete materials and a verbal system specialized for abstract materials-a distinction compatible with Paivio's dual-coding view of memory (e.g., Kieras, 1978; Paivio, 1971). The dual-coding view has been questioned, however, and it has been contrasted with the view that all of long-term memory can be represented in terms of abstract propositions. The logic of these views has been explored in a number of recent discussions (Anderson, 1978;Kieras, 1978;Kosslyn & Pomerantz, 1977;Paivio, 1974;Pylyshyn, 1973).The dual-coding view is not restricted to the representation of words in experiments on list-learning, and so it is meaningful to ask whether the results of Paivio and Foth generalize to prose learning as well. An interaction between concreteness loo Journal of Reading Behaviorand strategy when prose materials are used would support the generality of the dualcoding view; the absence of an interaction would raise questions about dual-coding as it applies to prose and would increase the need for an alternate approach to memory representation for prose learning.Other, research on imagery effects in prose learning has concentrated mainly on children (e.g., Pressley, 1977), whereas the main concern in the present study was learning by adults. Of the few studies investigating imagery in adult learning of extended prose, two measured recall of words rather than comprehension (Montague & Carter, 1973;Yuille & Paivio, 1969). Anderson and Kulhavy (1972), on the other hand, used test items intended to assess learning of specific and general information. Overall, they found ...