Taking on Testing: Teachers as Tester-Researchers

Good writing teachers typically mistrust most writing assessments, and for very reason. As writing assessment is now customarily conducted, conflicts exist between how writing is tested and how writing is taught in classrooms where students in their own and their teachers' eyes seem to be learning to write purposively, meaningfully, and effectively. In the opinion of many writing teachers, testers rarely test what writing teachers believe they should teach; in the opinion of student writers who are making progress and taking pleasure in doing so, testers rarely provide them opportunity to compose in written language for purposes meaningful either to them or to the audiences for whom they typically write. And in the joint opinion of writing teachers and students who are learning, testers for reasons that are understandable, given present conceptions of their task focus attention upon features of texts and factors in their production that, all too often, seem marginal rather than central to meaningful characterizations of something that might be called good writing. Those who look carefully and reflectively into the teaching of writing and the testing of writing will find that a gap exists between the perspectives and practices of teachers in classrooms where writing and reading are being learned and put to use, and the practices and perspectives of those who test students' language competence. In our time, the gap is widest between those teachers whose practices are guided by current theories of language learning and development and those testers whose practices are informed by standard test theory. The gap is most apparent in those settings whether at the district, state, or national level where large-scale assessments are being attempted. One cause of the present gap between the teaching and the testing of writing is that assessments are more often designed to serve administrators and policy makers than teachers and students, and all too typically, these constituencies see themselves as having different needs. Administrators, as they see their responsibilities, must demonstrate to policy makers, and policy makers to the public, that learning is or is not happening. Broad generalizations and numerical summaries of the sort that large-scale assessments are designed to produce are viewed as the generally accepted vehicles