Recognition Memory for the Deep-Structure Propositions of Sentences

Abstract

This study tested a model concerning the structure underlying the retention of complex sentences; in the model, the deep structure constituent propositions of the multi-proposition sentences are hierarchically related. Latencies and error rated to recognize these constituent propositions as true or false were measured and were the same for all propositions. These results did not confirm the model but showed that for recognition (but not necessarily for other memory tasks), each deep proposition of a sentence is separately or equally accessible. Stated differently, the recognition process does not make use of hierarchical organizations. The findings were obtained while the imagery-arousing capacity of the deep propositions was controlled. In previous research imagery was not controlled and it was found that some propositions and idea-units were better remembered than others; consequently, the suggestion was made that imagery might be related to accessibility, in that high-imagery items might be more accessible than low-imagery items. Closer analysis of other aspects of the findings indicated that another interesting factor might also be operating.