Exploring the role of Punctuation in Parsing Natural Text

Abstract

Few, if any, current NLP systems make any significant use of punctuation. Intuitively, a treatment of punctuation seems necessary to the analysis and production of text. Whilst this has been suggested in the fields of discourse structure, it is still unclear whether punctuation can help in the syntactic field. This investigation attempts to answer this question by parsing some corpus-based material with two similar grammars --- one including rules for punctuation, the other ignoring it. The punctuated grammar significantly out-performs the unpunctuated one, and so the conclusion is that punctuation can play a useful role in syntactic processing.