Comparing Corpora Comparing Genres: Evaluating Collocation in Specific Discourse

Corpus linguistics approaches are more and more often exploited in the study of specialised languages (Bowker and Pearson 2002) and corpora have proven to be invaluable tools in the identification and analysis of specific genres, domains and discourses. Keywords, collocates and clusters have become key concepts in both language research and teaching, particularly so in the teaching of LSP (Gavioli 2005). The present study aims at contributing to the debate on the criteria used in corpus planning, gathering and assembling by focusing on the comparison between a set of existing large specialised corpora, assembled for various research projects: the CorDis corpus (conceived as representative of the 2003 conflict in Iraq, Morley and Bayley 2009), the IntUne corpus (Williams and Bayley forhcoming) and the WHoB corpus (Venuti et al 2009). Previous studies have focused on the different meanings and connotations that the lexical item ‘war’ acquired in the various sub-corpora of the CorDis corpus, i.e. newspapers, television news, Parliamentary debates, and White House Press Briefings (Zanca 2011, Riccio Venuti 2008). We now want to verify to what extent the collocational profile of this and other lexical items can be influenced and possibly restrained by the selection criteria of the whole corpus by comparing the use of the same terms in the IntUne and WHoB corpora, which contain the same genres and domains (newspapers and TV news programmes the former, White House Press Briefings the latter) and are representative of the genres in specific time spans rather than according to a specific topic. The findings should contribute to a more coherent and considered determination of the criteria underlying the collection of large corpora and the interpretation of results both in linguistic research and language or LSP teaching.