The effects of pictures on first graders comprehension of written discourse

The nature and extent of the effects of pictures on first-graders' comprehension of written discourse was studied. Sixty subjects, randomly chosen from three suburban schools, were randomly assigned to one of four treatments: (a) text with text-related pictures (RFT); (b) text alone (T); (c) text-related pictures alone (RP); and (d) text with text-unrelated pictures (URPT). Each subject read aloud and immediately retold a basal reader story in one of the four treatments. A structured interview on the function and usage of pictures was administered two weeks later.

The retellings were scored for explicit and inferential information using modifications of Taylor's semantic analysis technique and Warren, Nicholas and Trabasso's categorization of inferences. Inter-rater reliability at initial agreement was 83% for explicit information and 72% for inferential information.

The following were studied: (a) the effect of pictures on first-graders' recall of explicit, inferential, and text-inconsistent story information; (b) the effect of pictures on the kind of inferences that first-graders make during recall; (c) the predispositions that first-graders have toward the function and usage of pictures that accompany a story; and (d) the effect that these predispositions have on the amount and kind of information recalled.

Statistical analysis of the data was carried out using the nonparametric Kruskal-Wallis one way analysis of variance test (ANOVA). This more conservative procedure was chosen over a parametric one way ANOVA because the data were shown to violate two of the assumptions of the parametric procedure (p < .05), i.e., homogeneity of variance and normality of population. The Kruskal-Wallis showed significant differences (p < .05) for explicit information (free and total recall) and for text-inconsistent information (free and total recall), but not for inferential information. Further analysis was done using the nonparametric multiple comparisons test.

For explicit information (free and total recall), there were significant differences (p < .05) favoring RP over T, T over URPT, and more importantly, RPT over URPT. Also for explicit free recall, RPT was significantly better than RP (p < .05). For text-inconsistent information (free and total recall), there were significant differences (p < .05) favoring each of the other treatments over RP.

Descriptive techniques were utilized to determine the predispositions that first-graders have toward the function and usage of pictures that accompany a story. The results of the picture interview, the observational notes, and the story retellings indicated that all of the subjects used pictures to some degree. The most frequent usage was for word recognition and the least frequent usage was for comprehension.

The following conclusions were drawn: (a) RPT do not assist nor hinder explicit, inferential, nor text-inconsistent comprehension; (b) URPT hinder explicit comprehension but not inferential comprehension; (c) neither RPT nor URPT influence the kind of inferences that are utilized by the subjects; (d) when pictures accompany text, they are utilized mainly to aid in word recognition; and (e) with RPT, those who use pictures to a relatively greater extent may have greater explicit comprehension than do those who use pictures to a relatively lesser extent.

These conclusions imply that when materials are designed for teaching beginning reading, publishers should avoid using pictures which are too unrelated to the written discourse. Educational practices in the teaching of reading may influence picture usage. Also, at present it is unknown whether or not picture reading ability is a developmental process.