Early adolescent composing within a generative-evaluative computerized prompting framework

摘要

A review of cognitive process research on writing suggests that a generative-evaluative model of composing could significantly alter instruction on higher level writing skills and strategies. To test the model, a total of 164 children in the sixth, seventh, and eighth grades were blocked according to low and high writing ability and then randomly assigned to one of two groups: (a) a control group or (b) an experimental group which received computer prompts, think sheet scaffolds, and expert modeling of prompts. Results from dimensional and holistic assessments of written products indicated that students did not improve their writing performance as a result of exposure to prompts over a 6-week period. Internalization of writing-control strategies as measured by the Index of Writing Awareness (IWA; Bonk, Middleton, Reynolds, & Stead, 1991), open-ended guidance questions in writing, and a prompt sort task all failed to favor the treatment group. In contrast, keystroke analyses revealed that students in the treatment group lowered the number of surface level changes they were making per 100 words of text when the prompts were available, while at the same time increasing their textual repositioning within text. No differences, however, were found in meaning changes as a result of these altered revisionary practices. As expected, though, students' holistic scores in writing were significantly correlated with their awareness of writing skills and strategies. Surface changes were negatively correlated with students' holistic scores, while repositioning within text correlated positively with holistic scores. A computer attitudes questionnaire indicated that the treatment group considered the computer helpful in creating a sense of audience and affording text evaluation opportunities. Various grade and ability differences provided evidence that procedural assistance was most appropriate at the sixth-grade level.