Google Translate offers a free, instant translation service between 64 different languages. It’s rumoured to be the best, most advanced online translator, used by over 200 million people every month, boasting the ability to translate a variety of words, sentences and web pages in any combination of its supported languages.
Having spent the last four years studying French and German to degree level, I find it quite insulting that people assume they can achieve just as accurate a translation by using this online translator. It sends out a message that those with a working knowledge of foreign languages are simply wasting their time; their skill is no longer necessary thanks to the wonders of new technology.
However, Google Translate is not as perfect as we’re led to believe. It searches hundreds of millions of documents, produced by native language speakers and human translators, to look for established patterns, before deciding which translation is best. Known as “statistical machine translation”, it searches documents accurately translated by human translators, in order to make an intelligent guess as to the most appropriate rendering of the text into the target language.
The creators of Google Translate admit that as all of the translations are generated by machines, they are not all perfect. They argue that the more human-translated documents that are available for Google to analyse, the higher the quality of the produced translation. Is that not suggesting therefore, that there is a greater need for human translated documents if Google Translate ever hopes to improve its accuracy?
In the UK, if you’re looking to translate into a foreign language, it’s more common for people to be able to speak French, German and Spanish; these are the foreign languages widely taught in UK schools. Therefore, there will be fewer documents available for Google Translate to use that switch between English and Icelandic, for example. According to Google Translate, this explains the reason for variance in translation accuracy across different languages.
Coming from a language background myself, I personally wouldn’t recommend Google Translate for anything other than a simple word-to-word dictionary. Having had a bit of a play with the options, it’s clear that machine translators have come a long way from the previous Babelfish attempts; however, they’re not to be trusted blindly. If you have some kind of knowledge already of the translation language, you’ll have a better sense of judgement and realise when the computer churns out complete gobbledegook.
The overriding problem with machine translators is that they have no context to go off; without context, you’ll never achieve a perfect translation. Another problem is that a lot of machine translators produce very literal translations; clearly, they cannot be expected to abide by the many different syntactical rules of a foreign language, yet this is a crucial aspect that needs to be right when translating, otherwise the target audience won’t have the faintest idea what you’re talking about.
In order to really understand language, you need to have cultural awareness; living abroad is the best way to achieve a true understanding of how a language really works on a day-to-day basis. I’ll leave you with a somewhat literal translation of the French idiom “to get your story straight”: “accorder ses violons” becomes “give his violins”. Not very likely to persuade someone to tell the truth now, is it? Think I might stick to a dictionary for future translations, though for a free service that works instantly, it’s probably better than relying on those 2 years of high school French classes that you didn’t really pay attention in.
Also, I discovered this neat little trick to make Google Translate beatbox. Click here and let the beats begin. If you’re really skilled you might even be able to get Google Translate to provide the rap in a new window… though good luck with that!