Reader-based and teacher-centered instructional tasks: Writing and learning about a short story in middle-track classrooms

Journal of Literacy Research, Mar 1996 by George E Newell

Posttests of story understanding. Posttests were scored for quality using a six-point scale as explained in Marshall's (1987) study of literary understanding. In general, students' responses were scored low when they simply summarized elements of the story using low-level inferences. They were scored higher when they reflected not only a recall of text elements, but also an ability to write analytically about the features of the text using the text to support and elaborate on their interpretation. This measure judges the quality, or "level," of response using criteria such as (a) "brief, shallow answer"; (b) "summarizes the text"; (c) "one or more inferences; little specific response"; (d) "one or more,but incomplete inferences with some specific support"; (e) "reports and associates details with inferences"; and (f) "one or more elaborated inferences with specific support from text." More specifically, each criterion is tailored to the nature of the posttest question. For example, for a description task, the level 6 criterion includes "Moves beyond description to make inferences about the story, using specific details to support the argument"; for an interpretation task, the criterion for the same level is "Makes one or more inferences, elaborating on them using specific evidence from the text"; and for a generalization task, the criterion requires "Uses elaborated world knowledge and text evidence to support a generalization" (Marshall, 1987, p. 47). After reaching a reasonably high agreement on training samples, two raters independently scored 35 posttests. Exact agreement was attained on 72% of the tests across the three question levels. When all scoring was completed, analysis-of-variance procedures were employed to examine posttest results for the students' story understanding in the two conditions. For this analysis, a factorial design with multiple measures was employed with instructional task (reader-based vs. teacher-centered) and mode of literary response (description, interpretation, and generalization) as betweensubject factors. The dependent measures included scores for the quality of written response to the three questions at each level of story understanding: description, interpretation, and generalization (Marshall, 1987; Newell, 1994).

Complete data sets for the posttests of story understanding and the essay responses to the story were available for 35 students - 20 from the reader-based condition and 15 from the teacher-centered condition. Because of student absences, complete data sets were not available for 6 students across the two classes. These were dropped from the analyses. In addition, data from the four casestudy students were not included in the analyses of group results. Accordingly, the quantitative data collected during the study consisted of posttests of story understanding (io5 paragraph-length responses) and the essays written in response to the writing task (35 essays).

Analyses of the retrospective interviews. The analyses of the interviews were qualitative in nature, involving careful reading, first separately and then in comparison with other interviews (Merriam, 1991). This process enabled me to discern patterns in the students' responses to the questions that were posed before they began writing, then again as they completed the first paragraph of their essays, and then finally as they completed the essay. Once identified, I sought specific evidence for and against these patterns by returning to the interviews for examples of the students' own language. Throughout the process of analysis, both the categories and the comments within categories were compared and contrasted (Bogdan eS Biklen, 1982). The analyses were concerned primarily with unpacking students' concerns at different points in the composing process. Thus, the strategy allowed for a focus on students as informants to explore how they conceptualized the writing task and reasoned about the story in an analytic mode. The results of the analysis provided a set of categories with students' language samples identified as to the time it occurred during the composing. Analyses of the essays and literary response statements. Based on Purves and Rippere (1968) and modified by Marshall (1987) and Newell et al. (1989), the measure codes individual sentences or T-units as falling within one of the alternative categories of response. Appendix B provides the complete instrument. Descriptive statements: Statements in which some part of the story is retold or some aspect of the story is described, for example,"Amy then decided that she would take a stand with her children." Personal statements: Statements in which the writer reacts in a personal way to the form or content of the story, for example,"I felt upset too, the way the husband treated her."