We describe a method for evaluating a grammar checking application with hand-bracketed parses. A randomly-selected set of sentences was submitted to a grammar checker in both bracketed and unbracketed formats. A comparison of the resulting error reports illuminates the relationship between the underlying performance of the parsergrammar system and the error critiques presented to the user. INTRODUCTION The recent development o f broad-coverage natural language processing systems has stimulated work on the evaluation of the syntactic component of such systems, for purposes of basic evaluation and improvement o f system performance. .
