A Parsing Methodology for Error Detection.
Davide Turcato, Devlan Nicholson, Trude Heift, Janine Toole, Stavroula Tsiplakou · 2000
Introduction We discuss the design of grammars for syntactic error detection. The topic is equally relevant for grammar checkers and Intelligent Language Tutoring Systems (henceforth ILTS). The proposed methodology addresses three interrelated issues recurrent in the error detection literature: Efficiency. In error detection the search space is larger than in ordinary parsing. Therefore there is a need to keep the search space manageable and to introduce some control mechanism over parsing. Modularity. Different knowledge sources need to be used, some of which are domain-dependent. In order to combine efficiency with modularity, different knowledge sources should be amenable of being stored separately but accessed in parallel at parsing time. A related issue is also the ability to reuse the same grammar for both error detection and ordinary parsing. Expressive power. There is a tendency to use unification-based frameworks of th