Computer-aided corpus linguistics for referring chains analysis: methodological questions and technical solutions

Through the coreference annotation project presented in this volume, we show how the links between annotator, tool, and corpus are rich and complex. This is due to the fact that the annotation work is not linear. A first facet of this work is the constitution of the corpus, which leads to using and creating various file formats, in order to allow the manipulation and to ensure the continued existence of linguistic data. The corpus constitution involves the definition of the annotation structure. It has to be shared among a group of researchers, and evolve. A second facet of the annotation work groups annotating and querying data, a confrontation between corpus and analysis table. Problems and needs for correction may emerge at each step of the work. Then, to support efficiently the annotation work, the annotation tool has to be designed as an instrument that continuality needs to evolve.