156
A. PIOLAT & JX. ROUSSEY
1) the reviser's overall representation of what he/she is supposed to do and of
the linguistic aspects he/she must monitor (e.g. Bartlett, 1982; Collier, 1983;
Daiute, 1989; Flower, Hayes, Carey, Schriver, & Stratman, 1986; Kurth, 1987;
Matsuhashi & Gordon, 1985; McCormick Calkins, 1980);
2) the execution of the revising, viewed either as the punctual solving of a given
problem (e.g. Scardamalia & Bereiter, 1985; Flower, Hayes, Carey, Schriver,
& Stratman, 1986) or as the step-by-step transformation of an entire text
(Piolat, Roussey, & Guercin, 1989).
The strategies and cognitive processes of revision
Among the various proposals made to define and study revising strategies, the one made
by Hayes, Flower, Schriver, Stratman, and Carey (1987) is undeniably the most precise. These
authors give a more complex description of the «reviewing» process, previously thought to
be composed of only two sub-processes («evaluate» and «revise»; Flower & Hayes, 1981),
by breaking down the process into four components and by further specifying the kind of
knowledge the revising activity involves and generates. They grant a more important role
to the reviser's selection of what knowledge to apply and what strategic choices to make
as he/she
a) defines the task,
b) evaluates the text and defines the encountered problem,
c) selects a strategy involving either going back to the preceding processes or
going on to modify the text, and
d) modifies the text either by revising it or rewriting it.
From a functional standpoint, the above sub-processes of revision are organized
hierarchically. Each of the four steps required to make a correction is necessarily subordinate
to the preceding one. The reviser can nevertheless decide not to go on to the next step, and
restart the sequence at any one of the higher-order sub-processes. This process-sequencing
flexibility accounts for the functional variants so fully described by Hayes et al. (1987). We
have briefly summarized them below, while mentioning the choices writers can make as they
first define the task and then select a revising strategy.
In order to revise, writers must have, and if not build, a representation of what they
consider to be involved in evaluating and improving a text. They must plan what they are
going to do by specifying:
- The goals to be reached (for instance, revise to make the text clearer).
- The characteristics of the text to be examined (for instance, revise the local
or global aspects of the text).
- The means that can be used to reach the defined goals (for instance, correct
the text several times in succession).
Hayes et al. (1987) attribute a clearly metacognitive role to the notion of task definition.
Indeed, this definition serves as the control manager for the sequencing of complex sub-
processes by setting the goals, constraints, and criteria required to guide the entire revising
activity. These authors make an inventory of the various «definitions of the task» that revisers
of differing degrees of expertise can verbalize, and thus ascertain that experts have more
meta-knowledge and knowledge likely to promote the setting of objectives like ((check for
correct meaning, text length, and text type)), or «check the number, density, and complexity
of the problems and errors in the to-be-improved text», etc. The scope of experts' processing
unit is the entire text, whereas that of novices is the sentence. At this point, Hayes et al.
(ibid.) violate their own usage rule for the term «strategy>>, reserved for the «selection»
component (see below). They speak of a «single-sentence strategy>> (p. 217) and a « limited-