[conclusion of Kaufer et al. article at top of this page]

THE WRITER'S WORKBENCH, in addition to serving as a tool for editing and instruction, appears to have promise as a research tool. The relationships of features of writing identified on the WORKBENCH with other approaches to evaluating the features of a writing sample (e.g., holistic scores, error analyses) provide somewhat detailed evidence about these features. The data analyzed on the WORKBENCH for this study suggest that certain characteristics of writing that are attended to by a human reader are related to, and therefore are likely to have influenced, the evaluation of a piece of writing. These data provide some interesting clues, which need to be investigated with further research. The results provide additional information about the features of writing of which readers may not be conscious, but that may contribute to a score. This observation is parallel to the experience of the readers who sensed that they were attending to somewhat different features of writing when applying the Discourse/Sentence-level (D/S) method of scoring, and would have expected the D/S and holistic scoring methods to yield different scores.