Toward a New Theory of Writing Assessment

Abstract

Many composition teachers and scholars feel frustrated by, cut off from, or otherwise uninterested in the subject of writing assessment-especially assessment takes place outside of the classroom for purposes of placement, exit, or program evaluation. This distrust and estrangement are understandable, given the highly technical aspects of much discourse about writing assessment. For the most part, writing assessment has been developed, constructed, and privatized by the measurement community as a technological apparatus whose inner workings are known only to those with specialized knowledge. Consequently, English professionals have been made to feel inadequate and naive by considerations of technical concepts like validity and reliability. At the same time, teachers have remained skeptical (and rightly so) of assessment practices do not reflect the values important to an understanding of how people learn to read and write. It does not take a measurement specialist to realize many writing assessment procedures have missed the mark in examining students' writing ability. At the core of this inability to communicate are basic theoretical differences between the measurement and composition communities (White, Language). Writing assessment procedures, as they have been traditionally constructed, are designed to produce reliable (that is, consistent) numerical scores of individual student papers from independent judges. Traditional writing assessment practices are based upon classical test theory, with roots in a positivist epistemology assumes that there exists a reality out there, driven by immutable natural laws (Guba 19). The assumption is student ability in writing, as in anything else, is a fixed,