Writing, working memory and dual-TASK
All the works cited above allowed characterizing
the dynamic of the writing processes as well as
their processing demands. Their findings allow two
general observations. On one hand, a certain
stability in the general pattern of activation of the
writing processes and in their processing demands
appears. However, these general functioning char-
acteristics are affected both by inter-individual
differences and by situation-specific factors. On
another hand, individual differences in the writing
strategies (i.e., the way the writing processes are
activated) are well described in terms of transitions
between the writing processes (such the 'writing
signatures' of Levy & Ransdell, 1995). So far, not
enough studies have been conducted to reach firm
conclusions about the effects of individual
differences on the dynamic of writing. Future
researches should more systematically explore the
effects of individual differences both on the
dynamic of writing and on writing performance.
Concurrent activation of the writing processes.
The experiments described above on the composing
strategies of writers only deal with sequences of
writing processes, as if they were activated only
sequentially. However, to the extent that sufficient
resources are available, writers can activate several
processes in parallel (Kellogg, 1996). A very few
studies addressed this issue. In this section, I will
report results of a dual-task experiment that Olive
and Kellogg, (2002) conducted in order to
investigate the corodination of transcription with
planning, translating and reviewing during writing.
In this study, as a primary task, we asked our
participants to compose a text and copy it using a
digitizer tablet. It was thus possible to identify
when they were actually handwriting or pausing.
The secondary task we used was a probe task.
However, unlike the triple-task, writers did not
perform any verbalization. The findings of this
experiment revealed that in a composition task,
when the probes occurred while adults writers
transcribed their text, RT interference (or latency)
was longer than when the probes occurred while
writers were pausing. Moreover, RT interference
was even more brief when writers were transcribing
their text but during a copying task.
These differences can be easily explained if one
considers that adult writers, at the same time that
they write, simultaneously activate the high level
writing processes (planning, translating and
reviewing). By contrast, during pauses only these
high level processes are activated and during
handwriting in the copying task only transcription
processes are activated. Furthermore, we have
shown that this concurrent activation of the low
and high level writing processes mainly resulted
from the automatization of the transcription
processes. For instance, in third graders children,
RT interference associated to transcription was
higher than in adults but no longer differed in the
copying and composing tasks in children. This
finding corroborates the idea that the increase of
resources required by transcription processes no
longer allows children activating the high level
writing processes at the same time they write down
their text.
We provided further evidence consistent with
this interpretation showing that adults composing
with an unfamiliar calligraphy (with upper-case
cursive handwriting) also activated the low and
high level writing processes step-by-step (see also
Olive & Piolat, 2002 for the same results with a
suppression of visual feedback). These results point
out the adaptive management of writing. Some
authors (Fayol, 1999; Levy & Ransdell, 1995)
have besides suggested that writers' level of
expertise may partially be linked to this ability in
adapting the dynamics of activation of the writing
processes to the changing demands of writing.
Conclusion
As shown in this short review, the dual-task
technique largely contributed in showing how the
limits of working memory force the composition of
a text. This technique allows highlighting the
relationship between each component of working
memory and the various writing processes. They
also allows researcher investigating how writers
strategically control the course of writing for
adapting to its changes in cognitive demands.
The dual-task technique has raised very
important discussion in cognitive psychology. It
is clear that psychologists must be attentive when
selecting a secondary task and proposals like those
of Fisk, Derrick and Schneider (1986-87)
contributed in such selection. It must be noted,
however, that some of the criteria of Fisk, Derrick
and Schneider are not always relevant. For
example, with concurrent memory load tasks,
performance of either the primary task or the
secondary or both can be affected. Moreover, their
criterion about the community of resources has no
more relevance with multi-components models of
working memory such as Baddeley's one (1986).
In spite of this, the present article has shown
that several variants of the dual-task technique can
be fruitfully implemented but that the choice of a
particular secondary task is crucial because it
depends on how it is expected to interact with the
primary task. For instance, when aimed at
studying relationships between a cognitive process
or activity and a working memory component, the
decrease of performance resulting from the
secondary task can be observed on either the
primary or the secondary tasks, or both. By
contrast, when one attempts to evaluate central
demands of the primary task in usual conditions of
realization, performance of only the secondary task
must decrease (the amplitude of its decrease
indicating the amount of resources that was
required by the primary task). Finally, the use of
the dual-task technique can aim at inducing
modifications of performance of the primary task.
Studies on subject-verb agreement processes
conducted by Fayol and its collaborators (see