TEXT REVISING STRATEGIES AND PROCEDURES
157
scope strategy» (p. 217). Finally, in addition to being much more elaborate, the task definition
of experts remains more flexible than that of novices throughout the correction making process;
this flexibility allows for modifying the goals that guide the revising whenever necessary.
One of the four components of this model of revision is exclusively devoted to the selection
of one of the five adequate strategies whenever a problem is detected or diagnosed. This
component is situated in the overall structure at the intersection between <comprehension-evaluation» and «production of a correction>>. The choice to use one of these strategies
is triggered by different factors, since the problems to be solved in text revising are very
diverse (Faigley & Witte, 1981).
Three of the five strategies («revising process implementation strategies>>) involve returning
to one of the higher-level processes to reprocess the problem encountered. They are:
- Ignore the problem, assessed to be of little importance or too difficult to solve.
- Delay solving the problem, since, for example, on the first pass through the
text, it was not compatible with the goal set at task definition time.
- Search for more information in memory or in the text in order to better
understand or define the problem.
The other two strategies («text modification strategies>>) pertain to the production aspects
of the model:
- Rewrite in order to conserve the idea, but not the text. Depending on the
nature and degree of difficulty of the to-be-solved problem, the reviser can
either make a new rough draft or paraphrase the text.
- Revise the text in order to conserve whatever can be saved. The various revisions
are grouped into a problem-solution matrix also called the «means-ends»
table (the ends concern both communicative goals and text organization levels;
the means include various highly diverse operations such as insertion, deletion,
moving, etc.).
The studies conducted by Hayes et al. (1987; correction of errors introduced in relatively
long texts, and analysis of the verbal protocols produced by the revisers during the revision
process) confirmed the fact that revisers do indeed implement the defined functional variants.
Thus the authors were able to conclude that the functioning of expert and nonexpert revisers
differs for each component of the model. To give an example, experts diagnose most problems
and revise more often than they rewrite, while nonexperts do not.
This model succeeds in reducing the wide variety of functional variants to the activation
of four processes and five strategies, while outlining the hierarchical nature of the reviser's
functioning and the conditions under which strategical options are triggered. The validity
of this model cannot be experimentally proven, since it remains both very broad in scope
and highly complex in content. Revising strategies that are more local, but still compatible
with this model, must therefore be developed.
Strategies and correction-sequencing procedures
In their model of the cognitive processes of revision, Hayes et al. (1987) specify the
various steps and options taken by a reviser to make a correction, but they speak little about
the order in which those corrections are made. They only hint at potential changes in the
definition of the task as the revising is in progress, and simply mention that successive passes
through the text are necessary for transforming the text to meet the preset goals.
But the revising activity most often consists of making a series of corrections of different
types (each one enabling the reviser to attain only one of the task sub-goals). The reviser
must therefore check the compatibility of each correction against a global text-improvement
plan. Indeed, insofar as the different linguistic levels of a text (micro- and macrostructural)