Idiomatic expressions have been called the last frontier between machine translation solutions and human translators, perhaps the one aspect of the translation process that machine translation solutions as currently designed are incapable of cracking. I will outline the factors in data-driven machine translation that make this so, and explain how we can use them to our advantage. Human translators can and must leverage our real-world knowledge in ways that no machine-based system can.
Using current examples from academic writing, corporate content and the media, I will suggest that our currently accepted standards of formality and register in the field may be outdated. I will show how we as translators, in a world where the level of discourse in the highest circles seems to be spiraling in a downward cycle of vulgarity, can take a more bold, creative and colorful approach to adding idiomatic expressions to our translation output, to help keep our edge over the machines.
Formerly affiliated with the Stanford University linguistics department, Kyle Wohlmut is now a linguist at large and has been translating full-time-and-then-some since 1999, first as an in-house legal translator and full-time freelance since 2004. In that time he has developed specializations in fields close to his heart, including culinary translation, literary translation, and translation for the performing arts. In recent years he has spoken at several major translation conferences in Europe. He works from Dutch and German into his native English.