While useful for communicating the basics, machine and online translation tools still can’t grasp the nuances of language
An article by Gülay Eskikaya, published online in the Guardian on Wednesday 25 September 2013, 15.00 BST (http://www.theguardian.com/education/2013/sep/25/lost-in-machine-translation-star-trek)
My childhood memories of Star Trek might be hazy at best, but one image did stick in my mind. On some occasions when Captain Kirk met with an alien species he would communicate with them through a universal translation device. I can’t remember whether he flicked a switch or prodded at a device but there was definitely some intraspecies communication going on. The pointy-eared or four-eyed alien would speak in his own language but Captain Kirk and the audience would hear him in English. I remember being fascinated by this as a concept.
Fast forward four or five decades since Star Trek was conceived and there are those who would argue that we are not far off that idea. No, we’re not communicating with little green men, but with the advent of online machine translation it is so much easier to at least attempt communication with people from our own planet.
I have come across long-distance relationships, lovers cruelly divided by language as well as oceans, conducted entirely through Google Translate; language students swearing that they wouldn’t have passed exams without it. The tool was even said to have been used last year in a UK court when the court-appointed interpreter failed to arrive for the hearing – albeit just to inform the defendant that the hearing was being adjourned because the interpreter was absent.
This is good news if you need to know or communicate something quite basic or to just get the gist of a text. I recently came upon a blog and could not identify the language and just copying a few words into Google Translate did the trick (it was Swahili). It’s less good news if you have some serious stuff to translate and want your language to appear – human.
Earlier this year, there was a kerfuffle in the Turkish press after an interview with Noam Chomsky went awry. The Turkish daily, Yeni Şafak, interviewed the political commentator about developments in Egypt over email in English and then translated his answers into Turkish. In a published transcript of Chomsky’s “original” replies, the following sentence appeared: “Contrary to what happens when everything that milk port, enters the work order, then begins to bustle in the West.”
When the Turkish original: “Aksine ne zaman ki her şey süt liman olur, düzene girer işte o zaman Batı’da telaş başlar” is fed into Google translate then you do indeed get that garbled English sentence above. The words “süt” and “liman” do mean milk and port on their own, but taken together they form an idiomatic expression to indicate calm.
A human translator might have put it thus: On the contrary, when everything has calmed down, then this will be when the West starts panicking.
At the moment, much of machine translation on its own is not sophisticated enough to replicate natural human language. Companies hoping to reduce sometimes considerable translation costs are attracted by a machine/human hybrid collaboration increasingly being offered by translation agencies.
This involves running texts through translation programmes and the resulting copy is post-edited by a human linguist. Of the many winners in this formula, unfortunately the translator is not one of them. Finding themselves at the bottom of a production chain, they find that they have increasingly less room to manoeuvre in a crowded market. Newbie translators, who are perhaps recent language graduates and keen to gain experience and start building their network of clients, may be more likely to be tempted by lower paying jobs such as these.
The modern world demands that messages be conveyed quickly and in a variety of languages. The British Academy’s 2011 Language Matters More and More report notes that “the proportion of internet usage conducted in English is already on the decline, falling from 51 to 29% between 2000 and 2009.” English may still be the big player on the internet but communicating globally means more languages, not less.
To that end, machine translation serves a particular purpose and serves it well. However, machines can only translate words and not meaning and will be unable to grasp concepts, abstractions or cultural references. Ultimately, machine translation fails to differentiate between the language of a literary masterpiece and a car manual, a United Nations convention and a text message. To the machine it’s one and the same.
The homogenisation of language may be the dream of science fiction writers and futurologists but even a machine programmed to have a brain the size of a planet and fed every article in the Encyclopaedia Britannica will still be missing a heart. Algorithms may be able to work well with data but can they deal with syntax, idiom or nuance? Translations should be elegant as well as accurate. Good translation is good writing which reflects a lifetime of experience, creativity and imagination.
I think Captain Kirk would agree that we are not yet in possession of that magical universal translator which will allow us to communicate effectively and accurately. I for one hope that time remains light years away.
Gülay Eskikaya is an English/Turkish interpreter and translator and runs Turkish Business Translations.