Google Translate (GT) was launched in April 2006 as a statistical machine translation service, which used United Nations and European Parliament documents to collect linguistic data. During translation, it looked for patterns in these documents to help decide what the best translation was. The application now includes more than 100 languages, and is able to translate 37 languages via photo, 32 via voice, and 27 via real-time video. Quite impressive if it weren’t for its downsides.
It is true that the tool can be helpful when you are travelling and need to understand the directions to the train station in a foreign language, but it’s a different situation when your clients need to translate important confidential or creative documents, which will involve subtle nuances and a good understanding of the source language. If the materials involve legal contracts, financial reports, compliance training materials, health related documents, pharmaceutical drug trials or global branding videos, GT is unlikely to be the right tool for your business.
It is important to observe that companies may think it’s more cost effective to avoid using a translation company and take advantage of the free service instead, oblivious to the fact that Google collects all of the translated data. This is especially dangerous when key confidential information is involved.
Machine translation saves time and reduces costs, but free translation platforms are not ideal when important/confidential/creative documents are involved. Clients could choose a translation agency and rest assured that their documents will be safe and protected.
Machine learning and machine translation tools are known to have bugs. GT now uses a combination of machine learning and the help of human volunteers to make sure translations are more accurate, but it’s still far from perfect. Translation agencies have quality assurance practices already established to ensure better outcomes.
As The Guardian points out, if the English version of The Girl from Ipanema had been automatically translated by Google, Frank Sinatra would be singing: “Girl in the golden body, sun from Ipanema, the it swung its more than a poem,” which clearly doesn’t make much sense. Literature, music, and poetry are not under immediate threat due to the nuances needed to convey the original ideas when translating, which GT still doesn’t grasp.
Back translations using GT are no guarantee of checking if the content has been translated accurately.
To use GT, you need to be connected to the internet at all times.
Interpreters can’t use GT because they would have to type everything people said in meetings and conferences into the tool to be able to provide a translation. It would be time consuming, inefficient, and inaccurate.
The tool will presumably improve the accuracy of its translations the more human volunteers are involved in providing corrections and with the growth of the database of documents it analyzes, but context and a language's unique subtleties will be a difficult task for it to tackle. A machine doesn’t have a sense of humor and isn’t able yet to find the perfect words for a particular text.
Although enhanced technology is changing our approach to translation, the traditional translation industry is safe for now simply because the tools don’t translate cultural values, which is appreciated by companies who are serious about reaching their audience in the best and most natural ways possible.