Fabio Said has a good summary of the BDÜ conference last week – which I did not attend. – with an emphasis on the presentations dealing with legal translation.

Patrik Mustu, he reports, presented on English-to-German machine translation of legal texts, giving examples of places where DeepL was not precise enough. On the screen behind Mustu one can see the parties hereto rendered in various ways, for example die Parteien hierzu, and the best solution, die Vertragsparteien, which DeepL presumably did not deliver.

Translators are increasingly using DeepL – it’s possible to use the Pro version, which means your private texts are not imported by DeepL – but they need post-editing. This is all a bit outside my current experience but needs keeping an eye on.

In the old days, when MT was rule-based and not statistical, it was recommended to use an MT program in-house and adapt it to one’s texts. Can this be done now with problem terms mentioned (damages, equity, quantum meruit)? Just wondering.