Flower, Swarts, and Hayes, "Designing Protocol Studies of the Writing Process: An Introduction"

Linda S., Heidi Swarts, and John R. Hayes. "Designing Protocol Studies of
the Writing Process: An Introduction." New Directions In Composition
. Richard Beach and Lillian S. Bridwell, eds. New York:
Guilford, 1984. 53-71.

The team of authors present this essay as a practical variation of protocol
analysis much different from the theoretical treatment they published in 1983.
Protocol studies, they explain, focus on "individual processes" and they operate
in the interest of what they call "close modeling" (53). "Close modeling"
designates that the model is an intermediary between the minutiae of acts
carried out by the individual writer and larger tendencies that can be traced
across most writers when they write (given a regular, stated task or

The protocol produces a sequential record of the individual processes, and,
because it is extraordinarily detailed, it produces too much information (56).
How do researchers respond when there is too much data? Computational
methods. Limited scope. More deliberate selections. Samples,
etc. From the collection regimen, protocol analysts proceed with the
classification of activity according to a regularized process for writing–the
cognitive process model Flower and Hayes introduced in 1981. Analysts,
called "judges," study the protocols for "episodes" or moments when there is a
shift in focus or attention. Also, protocol studies are compatible with
four methods: 1.) exploratory (what if); 2.) comparative (similarity and
differentiation); 3.) hypothesis and testing; and 4.) modeling of the writing
process. This fourth and final point gets the lightest treatment.
Why? Also, there is very little discussion of what models do, or how they
are presented, set into motion, circulated, and so on.

What discussion there is of what a model does is limited to this:

"Our own efforts have been to model the cognitive processes in writing, that
is, to describe the key mental processes and their organization. Modeling not
only creates a theoretical framework for studying writing, but tries to account
for how people actually carry out the complex process of composing" (68).

Models "describe"; they "create a theoretical framework," and they aid in
accounting for complex processes. A tall order! Can models do all of
this? Later, the group states that "[w]hether it is fully articulated or
barely conscious, we all bring a ‘model’ or set of assumptions to research,
which to a large degree guides our questioning" (69). Models are, then, "set[s]
of assumptions," too. I can’t hurry through each of these ways of pinning
"models" to a particular function. But it does begin to seem like the
notion of "modeling" itself becomes a free agent–a Katamari ball ricocheting
off of too many matters for it to make any sense at all. Are models
descriptive? Is this true of both discursive and presentational (i.e.,
visual) models? Do models create a "theoretical framework" or a
"conceptual framework," and what is the difference (see Pemberton)? Models
come to our rescue where complexity is concerned, but they might also move us
toward complexity rather than always away from it. I mean that while many
models simplify, they can also complicate–unsettle commonplaces, and so on.

My general sense is that "models" and "model-making" becomes a catch-all–a
safety net for switching scales. Move from minutia (specificity and
precision) to broader orders: turn to a model. Move from broader
magnitudes back to the atomistic: a model. Still, the visual model from
Flower and Hayes’ 1981 essay shows up here, unchanged (an ancestor in an aging


"While this profusion of unselective data may seem overwhelming, it is
actually this method’s hidden strength: the very completeness of the picture of
the writing process provides a check on the researcher’s hypotheses" (53).
The points about selectivity and hidden strength are interesting, but the
"completeness of the picture of the writing process" seems like an impossible
pursuit. Still, even in light of post-process departures from routine, we
have to wonder whether there is something lasting in the challenge of too much
data, the decisions of what to do when faced with too much data, and even the
use of less formal measures to get at what writers are thinking about when they

"Thinking-aloud protocols, which provide some of the content of the
writer’s thoughts, give us many more data from which to draw inferences" (55).

"A protocol is a versatile research tool: it captures information from a
writer while the writer is engaged in a whole range of composing subprocesses
and behaviors" (65). Reminds me of Twitter. And although much of
this work has slid into technical communications–studies of tasks and activity
among designers, for instance–it causes me to wonder about the capturing of
information–field-wide metadata–tied to other scales of production, like what
is called a "discipline."

Phrases: judges (68), float-dive in Atlantic Ocean (56) (method is not for
the un-prepared); planning, translating, reviewing (60); model (69), episodes