You shall know a term by the company it keeps: collocations of the term evidence in general and legal corpora

Despite the progress made in the study of collocations, their use in specialised languages by and large continues to be underresearched. The present article attempts to go some way towards filling this gap by looking at variation in collocations of a single term (evidence) as extracted from a general corpus and a legal one, and by exploring the implications of such variation for the retrieval of legal collocations. In particular, the study looks at a) the overrepresentation of collocations in the legal corpus, b) the underrepresentation of collocations in the legal corpus, and c) the potential of both corpora for collocation retrieval. The findings suggest that there are striking differences between the use of collocations in each corpus and that such differences can radically affect the lists of collocations obtained from each corpus.