Select-a-Kibitzer: A Computer Tool that Gives Meaningful Feedback on Student Compositions

Select-a-Kibitzer is a computerized tool that gives feedback to students on their compositions in a unique way. The project is based on composition research which describes the process of writing as one of simultaneously solving multiple, possibly conflicting constraints. In Select-a-Kibitzer, each constraint is personified by a different character. A student enters a composition into the tool and then asks for feedback. A variety of natural language processing techniques are used to analyze the text. Then, each of the characters gives feedback on the text from its particular point of view. The student can focus on particular aspects of the writing process by choosing to get feedback from only one or a small subset of the critics. Select-a-Kibitzer differs greatly from standard “style checker” mechanisms that focus on surface features of the text. By using Latent Semantic Analysis, Select-a-Kibitzer can address a wide-range of meaning-oriented composition issues, including coherence, purpose, topic, and overall quality. This paper describes the composition research that forms the basis of the project, and the interaction and implementation of Select-aKibitzer. It focuses on the uses of LSA to provide semantic feedback to the writer. ∗This work was supported by a grant to the first author from the James S. McDonnell Foundation’s Cognitive Studies for Educational Practice Program.