Written Composition: Toward a Theory of Evaluation

PICK UP ANY RECENT PUBLICATION on composition and you will almost surely find some reference to the problem of evaluating writing. Teachers and researchers alike acknowledge that pronouncing judgment on a piece of writing is both important and difficult. Important because teaching students to write, sorting students for placement or admission, and research in composition all depend upon ability to discriminate levels of quality in writing. Difficult because the theoretical basis of evaluation remains unarticulated. In contrast, composition instruction has begun developing a coherent set of assumptions. For example, theorists may disagree on the relative merits of classical, tagmemic, dramatistic, and prewriting forms of invention, but they agree on the principle that invention is part of the writing process. Evaluation of writing proceeds without a similar set of principles. Yet evaluation does proceed. The need for deciding who shall attend which college, designating those competent to graduate from high school, identifying growth in writing, or determining our nation's educational progress have spawned various systems for evaluating writing. Holistic scoring, quantification of syntactic features, analytic scales, and primary trait scoring illustrate the range of existing methodologies for evaluating writing. Rather than evolving from commonly held assumptions about evaluation, each method rests upon its own set of assumptions. Statistical computations of reader responses provide the rationale for holistic scoring and analytical scales; developmental stages of language acquisition account for quantification of syntactic features; a triangular model of discourse underlies primary trait scoring. Each of these systems and the assumptions underlying it represent careful and intelligent thought, and my purpose here is not to denigrate any of them. I cite them simply as illustrations of my point. Driven by the necessity to evaluate writing, theorists have avoided examination of the nature of evaluation itself and have moved directly to devising means (and rationales for these means) for accomplishing this difficult task. In this article I wish to propose a more general theory of evaluation and to suggest how it might be worked out in practical terms. This theory grows out of a philosophical and linguistic debate on the question of meaning. The debate, best summarized by P. F. Strawson's distinction between