An Exploratory Study of the Relationships between Reported Imagery and the Comprehension and Recall of a Story
Mark Sadoski · Reading Research Quarterly · 1983
ion generalized from stories, and a recognition of theme calls for an advanced degree of semantic representation in reading. Deeper levels of semantic processing and text integration then seem to characterize the measures significantly related to reporting an original image of the climax of the story, a behavior which itself requires the reader to go beyond the specific illustrations of the story for that imagery. However, other variables in the study also seem to require deeper processing but are not significantly related to reporting climax imagery or other original imagery. A possible explanation for this difference could be that deeper processing as measured in this study may have different component factors. In order to investigate this possibility, an exploratory factor analysis was undertaken. Common factor analysis with communality estimates was employed, with iterations. Using an eigenvalue criterion of 1.00, four significant factors were extracted and rotated to an oblique solution approaching simple structure. The resulting factor structure matrix is given in Table 4. Retelling total scores also significantly favored Group II, probably as a result of the theme subscore; other retelling subscores favored this group but did not reach significance. Since scoring well on retellings would require a degree of language production and composing competence, this capability might at first be thought to account for the difference. However, this same competence would be useful for scoring highly on most of the subscores, particularly the ones calling for inferences beyond the language of the text, such as theme, plot, and character analysisd velopment, but the findings on these subscores are mixed, suggesting that language composing competence is not the distinguishing factor. However, recognition of both plot and character analysis-development would require much the same deeper semantic processing that theme recognition would. This might suggest that deep semantic processing is not responsible for the difference either, or that deep processing as measured in this study may have different aspects. The final significant group difference involved the comprehension process score. This score is derived from miscues, and reflects the degree to which readers are monitoring the semantic aspects of the text in a manner more conceptually abstracted than simply producing a literal rendition of the text. Higher comprehension process scores indicate that readers are producing story appropriate meaning, but not necessarily the specific wordings of the text. Therefore, high 116 READING RESEARCH QUARTERLY * Fall 1983 XIX /I This content downloaded from 157.55.39.181 on Thu, 29 Sep 2016 05:34:01 UTC All use subject to http://about.jstor.org/terms scoring on this variable also suggests a deeper level of text processing. Deeper levels of sematic processing and text integration then seem to characterize the measures significantly related to reporting an original image of the climax of the story, a behavior which itself requires the reader to go beyond the specific illustrations of the story for the imagery. However, other variables in the study also seem to require deeper processing but are not significantly related to reporting climax imagery or other original imagery. A possible explanation for this difference could be that deeper processing as measured in this study may have different component factors. In order to investigate this possibility, an exploratory factor analysis was undertaken. Common factor analysis with communality estimates was employed, with iterations. Using an eigenvalue criterion of 1.00, four significant factors were extracted and rotated to an oblique solution approaching simple structure. The resulting factor structure matrix is given in Table 4. A loading significance criterion of .35 (p <.01) was selected, a relatively conservative criterion as suggested by Child (1969) and Nunnally (1978) for factorial work. For purposes of this anaylsis, climax imagery was again scored dichotomously, number of images reported was included as a variable, and retelling total scores were eliminated because they were linear combinations of the retelling subscores, and could therefore account for no unique variance. A theoretical interpretation of these factors can be proposed that might explain all the significant relationships observed here and would be consistent with contemporary imagery theory, particularly the theories of Paivio (1971), regarding the relationships between imagery and verbal processes. Paivio contends that visual imagery and verbal processes are alternative mental coding systems for the symbolic representation of meaning. Imagery is theorized to be relatively better at coping with concreteness and the parallel processing of spatial information, largely free of logical restraints, and more rapid than verbal processes. Verbal processes are theorized as being superior in abstract and sequential processing, and contributing logic, order; direction, and organization to thought. A central element in this theory is the conceptual peg hypothesis which contends that key images serve as mental pegs to which associated information is hooked for storage and retrieval. In this regard, Factor I might be interpreted as a deeper level of comprehension that is associated with verbal processes. Factor I receives significant loadings from Table 4 Factor analysis of study variables