Abstract
This article reports on three studies about the reading reliability of lay and expert readers in rating three kinds of writing assignments. Readers had to rate the Con tent and Language Usage of students' writing performances. The studies show that expert readers are more reliable in rating Usage, whereas both lay readers and expert readers prove to be reliable raters of Content. When the writing assignment becomes more restrictive, as is the case with interlinear revision tasks, lay readers are also reliable raters of Usage. The conclusion is that the effect of expertise is dependent on the rating task the readers have to perform. Expertise pertains most to language usage and to the cases in which the writing assignments as well as the scoring instructions are relatively free.
