A preliminary analysis of collocational differences in monolingual comparable corpora

1. Introduction The notion of collocation has enjoyed mixed fortunes in the 50 odd years of its existence. Claimed to be obscure (Lyons, 1977: 612), counterproductive (Langendoen 1968: 63ff) and generally useless (Lehrer 1974) 1 by its detractors, the idea that part of the meaning of a word is somehow related to its " word accompaniment, the other word material in which [it is] most commonly or most characteristically embedded " has also had its supporters. These have suggested that words have no meaning out of context, and meaning itself is not contained anywhere, but rather dispersed as the " light of mixed wavelengths into a spectrum " (Firth 1957(1951): 192). The intuitive appeal of this view is evident if one thinks of the difficulty a compositional approach to meaning has (even allowing for subcategorisation frames) in explaining the patterned quality of language performance, as found in a corpus, and ultimately the speaker's or writer's effortless routine handling of co(n)textual restrictions. The hypothesis that " everything we say may be in some degree idiomatic – that […] there are affinities among words that continue to reflect the attachments the words had when we learned them, within larger groups " (Bolinger, 1976:102) provides a powerful argument in favour of the empirical study of collocations, with implications for theoretical, descriptive and applied branches of linguistics. In recent years, notwithstanding the vagueness of the notion and consequent methodological problems in investigating it empirically, the study of collocations has indeed defied difficulties and criticism and sparked renewed interest in a number of areas ranging from computational and corpus linguistics to lexicography, language pedagogy, and crucially for our purposes, translation studies. The hypotheses that " everything we say may be in some degree idiomatic " (Bolinger, above), and that " actual usage plays a very minor role in one's consciousness of language " (Sinclair 1991:39) raise a number of interesting questions for translation research. Is there any evidence that translators be aware of collocational restrictions in the source and target languages? Do they show sensitivity to phraseological (a)typicality and restrictedness? These are very complex issues, that can hardly be resolved in one fell swoop. For a start, theoretical as well as methodological problems remain as to what collocations are in the first place, 2 and how best they can be retrieved from corpora and compared (see e.g. Krenn 2000a). Secondly, different types of corpora for …