Abstract
It is unfortunate that so many college teachers of writing and composition textbooks describe revision as the process by which a writer merely cleans up the mechanical and stylistic infelicities of an otherwise completed text. This simplistic view presupposes something akin to the three-stage linear model of composing set forth by Rohman and Wlecke in the 1960's.2 Research during the past decade, particularly that of Emig and Sommers, challenges the assumption underlying such a view of revision by demonstrating that revision is not the end of a linear process, but is rather itself a recursive process,3 one which can occur at any point during composing. Recent research also shows that different groups of writers revise in different ways, a finding reflected in, for example, the work of Beach, Bridwell,5 Faigley and Witte,6 Flower,7 and Murray,8 as well as Sommers. Finally, recent research has developed classification systems to explain those revisions. Such efforts appear, for example, in the work of Sommers,9 Bridwell,'o and Faigley and Witte. However much this body of research helps us to understand the results or effects of revision, it does considerably less to help us understand what causes writers to revise. The most promising research on the causes of revision, of course, is that of Flower and Hayes. Reporting on their use of composing-aloud protocols in a case study format,'2 they conclude that when expert writers redefine or clarify the audience and the goals of their texts, they frequently revise.13 This research offers the best hypotheses about the situational or contextual causes of revision. But while Flower and Hayes suggest that the produced so far becomes part of the situational context, they do not adequately explore specific textual cues that may prompt revisions. Indeed, apart from what little can be gleaned from studies which look to errors14 in the text for causes of revision, we know very little about