© 1999 by Oxford University Press
Building an adult ESL knowledge base: an exploratory study using an expert system
Southern Illinois University at Carbondale, Carbondale, IL 62901-4518, USA E-mail: kimw@siu.edu
The major focus at the first phase of this study was to examine 57 language learning (LL) background features in relationship to overall grade, rate of progress, and completion of an intensive English program (IEP). Data analyses for 319 completed cases allowed the researcher to identify variables showing divergent patterns for high versus low success groups for each of the success categories. Eleven features were targeted for omission. In the second stage of the study, the focus shifted to determining the extent to which success could be accurately predicted by the expert system when basing inferences on feature constellations including LL background features only, entry proficiency features only, and LL background and entry proficiency features combined. Results indicated that entry proficiency variables were most effective when predicting low grade and high IEP completion success. Constellations including only LL background features were most effective when predicting high grades and low rate of progress. Constellations including both LL background and entry proficiency features were most effective when predicting high rate of progress and low program completion success.