The stances around machine translation are often quite unsubtle: translators are against the technology because it is seen as a threat or they are embracing it as the next great wonder. While it is often thought that translators are using machine translation for projects as a whole or that they are not using it, there is also a middle ground. In this article I share two other ways to benefit from machine translation – even without embracing it in full.
A middle ground for adopting machine translation
Professional translators quite often see machine translation as a threat. That might not be a surprise given the pace at which new technological advancements are introduced. Some translators view colleagues that are considering or using translation engines with suspicion because they may be a threat to the future of our profession. While these attitudes and points of view are understandable, they do not have to be as black and white as they are now.
Professional translators can benefit from machine translation or get used to the technology in quite simple ways, at their own pace, and sometimes with surprising results.
But whatever your approach to adopting or implementing machine translation, please be always aware of the fine print: using machine translation providers to translate your content often means that you are required to share the source text, possibly putting your client at a risk.
Way 1: Using machine translation to speed up the translation of phrases
Using a translation engine in the translation process, often leads to interesting, humorous, or sometimes embarrassing results. Put simply, machine translations are currently not good enough to trust. But while they can not succeed at chapter, paragraph, or even sentence level, they often do a good job of translating phrases or parts of sentences. While you cannot use a great deal of the translation engine’s output, you can still make use of those parts that are working. Enable the use of automatic suggestions or similar lookup settings in your CAT tool to insert those phrases that work well quickly without re-typing them.
Depending on the quality of the translation engine these phrases can form a big part of the sentence, sometimes speeding up the whole translation process by more than 50 per cent. On the other hand you should read those phrases thoroughly to check whether they are 100 per cent correct and fit in your overall translation.
This method also only works when you are using a connection to a translation engine, like Google Translator or SDL’s Language Cloud.
Way 2: Using machine translation output to learn from
Robots are viewed more with suspicion than with romantic feelings. When it comes to translation engines most of us believe that humans are better at translating than machines. So far they have proven to be right. However, that does not mean that translation engines are performing that badly. In the past few years I have been using Slate Desktop as my translation engine. This engine uses my own translation memories to create an engine – sometimes with really poor results and sometimes with a surprisingly high quality, depending on my input. It happens every now and than that a suggested translation is too poor to use but contains words I had never considered in my manual translation. Translation engines can therefore be a mirror or an educator, pointing at vantage points we had never considered ourselves and bringing our passive vocabulary to the foreground.
This way we can still learn from the robots, no matter how weakly related we are to them.