The Large-Scale Production of Syntactically Analysed Corpora
R. Garside · Literary and Linguistic Computing · 1993
This paper describes the technique of skeleton parsing, developed at the University of Lancaster to produce large, high-quality syntactically analysed corpora. Texts to be annotated are word-tagged by the Claws system, the wordtags are manually checked and corrected, and then a special-purpose editor EPICS is used to mark and label the main syntactic constituents of each sentence. The paper discusses the advantages of the technique of skeleton parsing, and also describes the mechanisms for quality control of the annotation process.