Constructing macrostructure for expository text.

In Exp I, 108 (36 Ss at each level) 3rd, 5th, and 7th graders and 36 graduate students read short expository paragraphs and performed tasks that required the generation of macrostructure. Ss chose the best title, wrote a summary sentence, or wrote 1 additional sentence for each paragraph. Some paragraphs were not well structured; others contained an anomalous sentence. Results show that performance improved with age. The title task was easier than the summary task, which in turn was easier than the next-sentence task. Only adult Ss reflected the presence of anomalous information, and the effects were different on each of the 3 tasks. In Exp II, the title task with 4 response options was administered to 24 undergraduates. Results show that Ss broadened their representations to encompass the deviant sentence in both related and unrelated paragraphs. In the summary-sentence task, proficient adults—who monitored their own comprehension—responded like children. It is suggested that children need instruction variations in both task and in text, introduced gradually and systematically, in order to deal with potential sources of difficulty. (25 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)