SEMANTIC MEMORY AND STRUCTURE OF TEXT: EFFECT OF MANIPULATION OF ORGANIZATION OF CONNECTED DISCOURSE ON RECALL OF SEMANTIC UNITS BY ADEQUATE AND POOR READERS AT THE COLLEGE LEVEL.

摘要

The purpose of this study was to investigate the effects of the manipulation of three types (descriptive, causal, and control) and two positions (before vs. after target details) of organizational structure. Subjects were 60 adequately and 60 poorly comprehending undergraduates. Six versions of a three-paragraph passage were constructed. After reading the passage, subjects wrote a free recall.;The following significant results were obtained. Subjects reading passages with a topical structure recalled more related propositions (SS, RP, SP, and EP). Adequate comprehenders given the topical structure before the target recalled more SP and EP than controls. Adequate comprehenders with the topical structure after target paragraph recalled more SP than controls. When poor comprehenders were given topical structure at the end, their performance was better than control subjects for recall of RP and EP. When total recall was measured, subjects who read passages with topical structure before the target performed better than subjects with topical structure after the target. Adequate comprehenders recalled more of all types of propositions than poor comprehenders.;It was concluded that the presence of a high-level organizing topical structure improves recall for details. The facilitation effect is made possible by virtue of the structure of semantic memory, i.e., high-level propositions influence comprehension and memory of lower-level propositions through the use of referential operations. This was explained in terms of the theories of psycholinguistics, text coherence, and discourse analysis.;For adequate readers, the Proactive Subordinate Processing model was constructed to show that reading comprehension is an interactive process involving the semantic memory of the reader and the hierarchical structure of text. Poor comprehenders do not actively integrate related propositions; therefore, they are not able to use this strategy.;Recalls were scored for five types of details using Frederiksen's (1975) model: (1) total recall (SS), (2) unrelated (NP), (3) similar (RP), (4) subordinate (SP), and (5) schematically related (EP). The design was a 3(Organizational Structure) x 2(Position) x 2(Reader Type) factorial with ten subjects in each cell.;The practical significance of this study involves readability formulas, textbook construction, and suggestions for making poor readers more adequate comprehenders.