Writers' Cognitive and Decision Processes: Revising After Feedback
Author: Barbara Sitko
Degree: Ph.D. in Rhetoric, Carnegie Mellon University, 1989
This dissertation explores the thinking processes of writers as they revise their texts after receiving feedback from members of their intended audience. Three studies establish that in this situation writers follow a definable problem-solving sequence. Studies 1 and 2 used protocol analysis to trace the processes of experienced and less experienced adults and 11th grade students. Although experienced writers tended to accept more feedback, all writers who revised followed the same pattern. They first assessed the feedback; if they detected that readers did not understand the text as they intended, they set a new goal, located the problematic text, generated alternatives, and made changes. Writers who do not revise and who report their reasons for not doing so exit this sequence in predictable ways. Notably, they fail to search their text for a cause, attributing the problem instead to the reader or to their own inability.
Study 3 used an experimental design to test whether a prompt specifying this sequence might guide writers to accept more feedback than would writers given a general prompt. The study also examined whether such revision was judged to improve text quality. The experiment used concurrent protocols, retrospective self-reports, and textual evidence. Twenty-four college freshmen participated in two sessions of one hour each. First they composed a text intended for twelfth-grade readers. One week later, the subjects revised their texts as they wished and then received the transcripts of readers of their texts who had been directed to summarize main points and predict future text. Subjects received either a general prompt to do their best or a prompt that specified the successful revision sequence via a figure and suggested questions. The between-subjects variable was the prompt. Dependent measures were the proportion of feedback statements used for revision and a quality rating of the post-feedback drafts.
Results indicate that the experimental and control groups did not differ in the proportion of reader comments used for revision. Both groups received approximately the same amount of feedback, and both responded in nearly equal proportions. All post-feedback drafts were judged by three adult raters to be of higher quality than pre-feedback drafts, but there were no between-group quality differences. Two interpretations are possible. Either the prompt was insufficient to influence revision behavior, or the feedback itself was strong enough to influence the writers. Post-hoc analysis of protocols shows that the thinking processes of both groups were consistent with the model of revising after feedback as a problem-solving process.