I am one of those evil translators who love machine translation and try to exploit its potential as much as possible. Not only MT has become the preferential choice of many big clients who need billions of words translated in a very short time, but it is also a good ally when you’re translating for private clients who can’t locate a high budget on their translation projects.
How does it work? You need a MT engine and a source text. That could be all. Unfortunately, the quality of the output text is not very easily predictable (remember: garbage in / garbage out!) and needs to be checked by a human linguist who post-edits the output.
Light post-editing (a.k.a “make it not offensive for native speakers”) involves minor but necessary changes in order to make the text readable and understandable for native speakers, and includes fixing typos, grammar mistakes and rewriting sentences when they make no sense.
Full post-editing aims at creating texts that sound natural in the target language: typos, grammar mistakes and weird sentences must be corrected and the tone, style and terminology must be kept consistent throughout the text. Idioms and cultural references must be adapted and a strict compliace to the guidelines is usually required (localize this and not that, write the numbers this way and not that way etc.)
What’s the point of using MT? Believe it or not, the outcome is sometimes brilliant and once you become familiar with the rules of post-editing, you’ll be relieved: Google Translate is not stealing our job! Curious about MT in Audiovisual translation? Check this!