Rhetorical Specification in Essay Examination Topics

In developing essay examination topics, professional test makers have relied almost exclusively on critical judgments and pilot testing rather than on principles derived from experimental research. One popular approach favors topics that create naturalistic writing problems by giving writers information about the purpose of a composition, its audience, its speaker, and its subject-what is commonly called full rhetorical specification.' Its advocates say that such specification allows writers a better chance to do their best work under test conditions; it enables them to understand more fully the demands of particular writing tasks and consequently to produce more fully realized, coherent compositions. (Many contemporary composition teachers would agree, having used such an approach successfully in the classroom.) But despite its persuasive, common-sense appeal, this position is unfounded: there is almost no experimental research evidence to support the idea that full rhetorical specification assures essay examination topics that will elicit the writers' best writing. Of course, there has been research on topics, and some of it has attempted to determine the effects of certain topic variables-for example, the mode of discourse in which writers are asked to write and the cognitive and experiential demands placed on writers by different essay questions.2 But the preference for rhetorically specified writing problems in large-scale writing assessments, such