Machine translation and the apocalypse
As a translator, I am used to hear/read that my trade is doomed and dying. There are a lot of people saying and believing that Machine Translation is going to eliminate the need of someone reading and translating a text. Everything will be put in a translation software and will magically be transformed in a seamless text translated. We, the poor translators, will not be relevant anymore.
Personally, I do not agree with it, and I am not afraid of becoming obsolete. It is sufficient to think how dynamic and organic is a language, the many nuances of our thinking process to realize no algorithm can cover that accurately.
Of course I understand and acknowledge the value of machine translation for emergency situations, specific needs of day-today understanding, but the final result is never acceptable (or even understandable) for a professional text, a company presentation, for instance, or an add that will look right in the context and market it is being presented and bring the results a company expects. And never will be the correct option for a book. Literature will not benefit from machine translation.
I am sure that no hotel will attract clients with the following room description: “Fit with footage and differentiated decorating formed per room”. This a phrase originally in Portuguese, taken from a Brazilian site and processed in Google Translate. Can you understand it? Not really, huh?
There are several translation machines. I am mentioning Google Translate because they actually admitted they have a catch 22 situation in regard to translation quality. The very logic of the system is bringing an issue that is hard to tackle.
The matter was discussed by Google’s Research Director, Peter Norvig, in Nasa Innovative Advanced Concepts Conference in Stanford, in February, 2014. Norvig admitted that there are sites that use Google Translate to render a new language version of their entire content, which is perpetuating the errors it has created, because Google servers recognize that site as a multilingual platform and indexes it using as source for other translations, with the same low quality generated in the first place, thus not becoming a better engine, but stagnating.
Of course, that their allegation when admitting it is that they are working in ways to filter these situations.
Bottom line is, no company that wants to be taken seriously should rely only in machine translation. The final result will never be adequate, and someone that is not proficient in the new language will never know. But the client they are targeting will.
Some people have the opinion that machine translation should be used with a post-revision from a real translator (a human one). And that would lead to a whole new post. The client’s expectation that this would represent a lower cost is questionable. For us, translators, the work in an edit/review job of a machine translated text can be the same, or even harder, than to start a translation from scratch. The research, review and editing needs are similar, and more than that, the result will never be one that is satisfying to us, personally. It will never bring the pleasure of delivering your work, your text, it will always sound “foreign”, even to you, as it did not start with your style.
The translation market is indeed going through deep transformation and a huge rise in demand. Everybody is going Global. Machine translations, as well as CAT (computer aided translation) software have their role in this new market, but this role is not to eliminate the translator. I am sure that translator and machine will work together for years to come, and the translation apocalypse announcers will excuse me, but I do not believe in them. My world is not ending right now.