Two studies examined the effects of different types of imaginal elaborations on recall. In Experiment 1, 6-year-olds, 9-year-olds, and college adults were given 20 word pairs embedded in one of four types of sentences: short plausible, bizarre, personalized, or self-generated. With no mention of a memory test, subjects were asked to use the sentences to generate images and rate their vividness. Findings indicated that recall varied with age and type of elaboration. Subjects who generated their own sentences had the greatest recall. Analysis of the generated sentences showed younger children were more likely than adults to generate personalized elaborations. However, the presence of personalized referents was not enough to enhance memory. Recall was actually lowest in the personalized condition. In Experiment 2, 6-year-olds, 9-year-olds, and college adults generated sentences connecting 20 unrelated word pairs for one of three purposes: to generate images of the words in the pairs by way of helping to create materials for children's stories, to generate images that would help them remember the word pairs, and to help them remember the words. Yoked controls were given sentences generated by age-mates. Findings indicated that memory was better in all conditions requiring sentence generation than among yoked controls. However, recall varied depending on the purpose of generating the sentences. It is concluded that findings have implications for discussions of children's elaborative abilities. (Author/RH)