Effects of Stylistics on Essay Graders.

Eight public examiners in English from the Western Australian Tertiary Admissions Examination marking panel and 103 final-year student teachers graded essays that were experimentally controlled for voice and diction. Control of these independent variables was achieved by providing four rewritten versions of the secondary student essays, each version identical to others except for one of four styles: active voice with Saxon diction, active voice with Latinate diction, passive voice with Saxon diction, and passive voice with Latinate diction. Grades were transformed to an interval scale using the Rasch Polychotomous Rating Model, and the transformed scores treated by repeated measures multivariate analysis of variance to test shifts in grader judgments from one version of an essay to the others. Results indicated a significant effect for diction, in favor of Latinate diction, when rated for overall quality, mechanics, logic, and organization. There was no effect for voice and no interaction effect. The students' original, unmodified essays received grades that placed them significantly higher than their Saxon versions but significantly lower than their Latinate versions. The experimental essay, with its intense uniformity of Latinate diction all in sentences of one voice, was preferred to the naturally written, variable, and flexible essay from the pen of a student. Among several implications arising from this study is the possibility that if English teachers instruct secondary students to write simply and clearly, students are being deceived, for such writing is not the most highly valued by teachers and public examiners alike. (Extensive appendixes include notes on the rewriting of essays and samples of stylistic versions, instructions to essay graders and examiners, and complete statistical results of data from three years.) (Author/HTH)