A lot of technical translators use Google in all kinds of interesting ways - they've become quite sophisticated at it over the years, with all sorts of tricks. Any corpus is going to be biased. For example, if you're a medical translator, you'll have a shelf full (or various shevles full) of bilingual and monolingual medical dictionaries that will not include a huge number of words used in the current medical research literature, owing to the pace at which research moves the subject on (and also owing to the fact that dictionaries don't tend to focus on the frontiers of a subject). A five-year-old medical dictionary, however good, is next to useless if you're translating research papers for current publication in places like the Lancet, the BMJ, JAMA etc. It's actually far better sometimes to use Google and other search engines to see who is writing what where, and whether it's only French websites - for example - or Romance-language sites, using a particular expression (in other words, a highly unreliable expression), or whether it's also on the main English-speaking research websites (and particularly in journal article titles). If you're translating peer reviews or peer-reviewed papers, it's often the only way to track down certain novel terms. I think technical translators would be among the first to vote for open publishing of academic journal articles, which would aid enormously in getting rid of some of the biases they face with current online information. Louise Ferguson