DEVELOPMENT OF AN AUTOMATIC ENGLISH GRAMMAR DEBUGGER FOR CIBNESE STUDENTS: A PROGRESS REPORT

This paper reports the research about development of an automatic English grammar debugger (on a personal computer) for Chinese college students, based on an analysis of the errors they made in their compositions. The first stage of development is devoted to an error analysis of 125 writing samples and classification of the errors into 14 main types and 93 subtypes. To implement the grammar debugger we first built a small dictionary with 1402 word stems and necessary features, as well as a suffix processor which accommodates morpho-syntactic variants of each word stem. We then built up an ATN parser, which is equipped with correct phrase structure rules and error patterns. In addition, a set of disambiguating rules for multiple word categories was designed to eliminate the unlikely categories to increase the precision power of the parser. The current implementation enables detection of seven types of error for a text input and response of corresponding diagnostic feedback messages. Future research will be focused on refining the grammar debugger to detect more types of mistakes with more precision and on providing appropriate feedback messages to diagnose students' deficiency as well as operations for students to edit their errors.