Towards a ‘Science’ of Corpus Annotation: A New Methodological Challenge for Corpus Linguistics

Corpus annotation—adding interpretive information into a collection of texts—is valuable for a number of reasons, including the validation of theories of textual phenomena and the creation of corpora upon which automated learning algorithms can be trained. This paper outlines the main challenges posed by human-coded corpus annotation for current corpus linguistic practice, describing some of the methodological steps required for this indispensable part of the research agenda of Corpus Linguistics in this decade. The first part of the paper presents an overview of the methodologies and open questions in corpus annotation as seen from the perspective of the field of Natural Language Processing. This is followed by an analysis of the theoretical and practical impact of corpus annotation in the field of Corpus Linguistics. It is suggested that collaborative efforts are necessary to advance knowledge in both fields, thereby helping to develop the kind of methodological rigour that would bring about a ‘science’ of annotation.