Research studies have pointed to a similarity between reading and listening that may imply that listening to writing is likely to be a useful means of evaluating it. Therefore a study was designed to determine the extent to which holistic scores assigned to student essays by readers correlate with holistic scores assigned to the same essays by listeners. Essays were selected from those written by freshmen at the University of Southern California. At the listening session, each listener was presented with a copy of the essay question and with a rubric used during the original grading system. Once the socialization process associated with holistic scoring was established, each participant was asked to listen to a few essays that had been tape recorded and to evaluate the essays according to the identical criteria used when evaluating by reading. The correlation between listening and reading was discovered to be .764, although the listening scores were at least a point higher in all but five cases. Analysis of the data suggests that some facets of written discourse--in particular, content, structure, and task fulfillment--can be evaluated reliably by listening. Surface areas of discourse such as spelling, mechanics, usage, and diction are more difficult to detect by listening than by reading. (Author/HOD)