What are corpora for?

I get this queestion every time we talk Translation Memories, so here are a few links to good reading materials for translators:

http://www.jostrans.org/issue07/art_wilkinson.php

Exploring corpora with concordancers can help translators to improve the quality of their translations by, for example, providing them with information about collocates; by helping them to choose between terms; or by enabling them to confirm intuitive decisions. But corpora also allow unpredictable, incidental learning: the user may notice unfamiliar uses in a concordance and follow them up by exploratory browsing. The article discusses the potential of corpora to throw up previously unknown information that may be relevant to a translation assignment, and illustrates how advanced search strategies can increase the likelihood of “accidentally” finding relevant information.

http://www.jostrans.org/issue10/art_maher.php

Abstract: Translators and editors who work in a specialised field—a particular branch of medicine, technology or finance, for instance—may find it difficult to acquire (or enhance) their domain-specific knowledge other than by learning as they go or going back to college. Both strategies can be slow and costly. Our paper describes a faster, more economical way to climb the specialist learning ladder, namely a corpus-guided approach to translating, revising and editing. We describe two tools for analysing a corpus of model texts: on the one hand, a user-friendly concordancer with an intuitive interface; on the other, an equally easy-to-use desktop-based indexer. Finally, we propose an approach to the issue of corpus size (sampling adequacy) that provides a practical solution for the working translator: we recommend creating a carefully chosen, cleaned text collection that functions as a reliable substrate corpus for language pattern guidance and adding to it an ad-hoc ‘quick and dirty’ corpus to further narrow the topic focus as needed.