Formalist Mode of Inquiry
Nature of "Knowledge"
Formalist inquiry if characterized first by the use of formal language and second by the development of an axiomatic, deductive structure, and its object is the building of models and their interpretation—that is, determining the relationship between the model and some portion of the real world. (cf. Euclid's geometry) An Experimentalist isolates the features of some portion of the empirical world and works to correlate the relationships between them (cause-effect). A Formalist, by contrast, investigates an analogy. The basic idea is to construct, usually in formal language, a model whose internal logic resembles the phenomenon under study. A fully realized Formal theory will have a model with a set of interrelated propositions stated in a formal language, and second it will have one or more sets of rules that relate the model to the empirical world (rules of correspondence or rules of interpretation).
Formalist researchers in writing have a problem: Language (speech) is the production of some system. What is the nature of this system? (that's their research question). The problem is that the researcher is denied direct access to this system, so the Formalist examines the output data or whatever indirect methods of observation are available to devise a model system that might be made to produce the relevant data (in this case, language). The claims that might be made for the resulting model are not empirical; it is not a description of the way that humans make language, but a model whose internal logic is such that, given the right input, it can turn out a product, "language."
Formalist knowledge, like Experimental knowledge, has a paradigmatic structure: community activity is directed by reference to one or another models. The difference lies in the assumed relationship between the paradigm and the empirical world. Instead of following an inductive logic (or probably inference), the Formal paradigm or model has two other logics: a tautologic, in the sense that as a model it is internally complete; and an analogic, by which it can be tested against, interpreted in terms of, one or more empirical, real-world systems.
A key danger (especially for Compositionalist) in Formalist inquiry is to mistake the elements of the Formal model for the "real" things to which they correspond. You can't mistake the model for a description. If a Formalist model is not a description, how does it work? While an Experimentalist wants to quantify phenomena, the Formalist wants to qualify them. That is, both methods are positivist in their orientation, assuming that despite surface appearances of chaos, the world is fundamentally an orderly place and that humans have access to that order. Whereas an Experimentalist wants to record enough instances of x happening after y happening after z to be able to link this sequence causally, the Formalist will set out to discover its underlying logic—the how or why or by what mechanism. He will do this by choosing a plausible logic or logical sequence that might explain the system, construct a model, and then work at testing the correspondence between the model and the system in question, refining the model to account more and more fully for the empirical system. This last process, usually called successive approximation, is at the heart of Formalist work. Testing and refining a model refers to determining, via its bridge principles, its "conditions of relevance"—the extent to which it has explanatory power in any given domain.
Advantages of Formalist inquiry—avoid problem of variable explosion in Experimental inquiry. Also, we are free to build from the top down; rather than having to work toward a full explication by painstakingly accounting for innumerable component correlations, we get to make some sort of educated guess right away, and then go back to see if or how the details will or will not support our guess. The eventual potency of any model we construct would depend on the degree of correspondence between its output data and the output data of the writing process, texts.
Inquiry—Formalist Inquiry
- Identification of Problem/Constructing of a Model
- Formalization
- Testing and Refining the Model: Successive Approximation
- Dissemination to a Wider Audience: The Uses of Formalist Knowledge
The problem's of Flower and Hayes model:
It does not set out rules of interpretation. It puts its Formal language in everyday language, so the model looks as though its making empirical statements about observable phenomena. Their terms, actually, have very special, restricted meanings, and they do not correspond to observed psychological states or entities or processes. (BIG question—but their terms DO seem to describe processes and states of the writer???)
Methodological problem with the use of protocols? Not in the nature of protocols themselves, but in the use to which they are put. Instead of designing a model that treats texts as its primary output data, and whose "black box" operations can be somewhat illuminated by protocol analysis, Flower and Hayes have given the protocols primary status. The model they posit is designed so as to offer a far better account of how protocols, rather than texts, are produced.
Beware the fallacy of affirming the consequent. When discrepancies appear between deduced predictions and data, changing the assumptions to conform to the data. The error lies in forgetting that the same consequences can be deduced from any number of assumptions, not only from the original one.
Any use of Composition's models as other than first approximations would probably be ill-advised. Danger of propaganda use of model. The theory can readily perform the essential function of propaganda, which is to pass off an idealized picture as a true account of reality.
Central figures: Linda Flower and John Hayes. Robert de Beaugrande, Ann Matshuhashi, Mike Rose
Modes of Inquiry Home
This site contains direct excerpts from The Making of Knowledge in Composition by Stephen North. Portsmouth: Heineman, 1987.
Lirvin Researching | Site created by Lennie Irvin, San Antonio College (2007) | Last updated
August 20, 2007