Machine translation softwares like online translation tools or translation apps are becoming more and more common, and while they have undoubtedly gotten better over the past decade, we believe that they cannot substitute human translations – yet, at the very least. The first and most important reason is the low quality of the result and the doubtful accuracy of the translated text. Have a look at our last two posts to learn more about the importance of quality in translation projects and how to spot a low quality translation yourself. Machine translation’s major flaws are what make any machine translation inherently low-quality.
The ability of the powerful algorithms and the immense datasets that power machine translations nowadays makes it tempting to place your end-of-year report into Google Translate and trust that the resulting translation is quite good. However, we must understand that because of the very pattern-oriented nature of machine learning algorithms, the translations provided by Google Translate, Microsoft Bing or any other online translation service, are impressive, considering that the machine does not understand any of the subtleties of language, apart from those that are explicitly coded into it or those that it “deduced” by looking at millions of phrases. But it is this very lack of understanding of the subtleties of language that make the resulting translations unacceptable for professional, technical, academic or legal translations. These subtleties of human languages and the capacity to understand its nuances is something that only humans can provide. To the best of our efforts, the best minds working on AI have not been able to design an algorithm that understands nuance or the intricate relationship between words in a phrase.
Without getting overly technical and linguistic, I would like to demonstrate what was shown by Terry Winograd back in the 70s, as brought back to my memory by this cool YouTube video (https://www.youtube.com/watch?v=m3vIEKWrP9Q): a human reader would easily understand who the “they” refers to in this famous example, depending on the schemas that arise for each verb choice:
- The city councilmen refused the demonstrators a permit because they [feared/advocated] violence.
We inherently understand, thanks to our years of built-up knowledge and our innate language skills, whether the “they” refers to the councilmen or the demonstrators, while a machine translation will usually fail with such schemas. This is an evident flaw, but it is only one of many that would be completely avoided by hiring an excellent translation agency. Translators at respectable companies, like Ready-to-Publish Translations, LLC, understand nuance, the intricate relationship between words in a phrase, the sort of language and style that best suits the target audience, and all of the other qualities that define good writing. They treat all of their translations like copywriting, instead of it being a mechanical process that replaces a word in one language with its translation in another. If you want your clients or readers to feel a connection when reading your translation or if you want your attorneys or the immigration office to fully understand everything that you want to portray to them, make sure you hire an excellent human translator or translation agency so that the translated document does not end up being an obstacle in your path towards growth or success.
