1 Are Efficient Natural Language Parsers Robust ?

This paper discusses the robustness of four efficient syntactic error-correcting parsing algorithms that are based on chart parsing with a context-free grammar. In this context, by robust we mean able to correct detectable syntactic errors. We implemented four versions of a bottom-up error-correcting chart parser: a basic bottom-up chart parser, and chart parsers employing selectivity, top-down filtering, and a combination of selectivity and a top-down filtering. The combined selectivity and top-down filtering parser was the most efficient. However, this parser failed to correctly repair more sentences than the other parsers, failing on 18 out of 119 ill-formed sentences, compared to no failures for the basic bottom-up chart parser. This paper examines trade-offs between parsing efficiency and robustness at the syntactic level.