Translator vs Machine Translation

Picture credits: topbots.com

Machine translation (MT) is getting more and more effective everyday. Some online translation tools, such as Google Translate, are developed on a daily basis thanks to their users, who can correct wrong translations or add new ones to the system. The more data these translation tools get, the better they become at translating texts from one language to another. Other mobile apps and softwares, seemingly not related to translation at first glance, use translation as a part of their marketing strategy. Skype, for example, promises its users to translate 10 languages in real time during calls, and more than 60 languages for text messages.

With the expansion of these machine translation softwares and services, any person who is unfamiliar with the world of translation could wonder if translators are really necessary, when a machine translation tool can translate long texts in just a second, free of charge. Companies would then be right to simply use machine translation software, and get all their translations done for free, in the blink of an eye, right?

Picture credits: American Translators Association

Well, no. The thing is, a machine translator will never be able to fully grasp nuances in human speech like another human can. Moreover, it cannot always understand the cultural aspect or the context of the original text. This can lead to a text full of  nonsense and mistakes (like this example shown on the American Translators Association website of instructions on a ceramic plate) and to awkward situations (like this example taken just yesterday from Yandex Translate: my boyfriend told me this sentence, but depending on whether I translated from Russian to English or to French, the outcome is completely different for the second sentence).

Here, Yandex Translate gives the correct translation...

... but definitely not there

Moreover, MT softwares are good at translating single words or short expressions, because in that case they act as an online bilingual dictionary. But when we put a longer and more complex text in a Machine Translation software, the software sometimes struggles to get a good sentence in the target language.

The only way for a text to be correctly translated including all these different parameters is to resort to a human translator.

Human translators, however, use CAT (Computer-Assisted Translation) tools, translation memories, online dictionaries, etc. How is that any different from machine translation ?

CAT tools and translation memories are here to help improve the translator’s productivity: the translator goes faster and the quality of the translation is better. They are in no way intended to replace the human translator. Only a human translator can determine which style and tone of speech are better, and which expressions are culturally acceptable or not. CAT tools and translation memories are just instruments to reach the goal of the translator, which is to transpose a text in one language, into another language, without denaturing it.

We could also mention the work of editors in translation, who sometimes have to correct work entirely done by MT software. But, once again, although the work here is mostly done by MT, a human correction is still needed.

Machine Translation softwares are getting better, faster and are evolving rapidly, but will probably never « beat » a human translator.  Humans write with so many nuances, in so many different contexts, with homonyms and not always with good grammar. Machine Translation is good, but not that good. Maybe in a hundred years the profession of translator will no longer exist, but until then we can rest assured that we, as translators, will still have plenty of work to do.

Article written by Marine B.