Each time I read an article claiming that the guild of human translators will soon be forced to bow down before the terrible swift sword of some new technology, I feel the need to check the claims out myself, partly out of a sense of terror that this nightmare just might be around the corner, more hopefully out of a desire to reassure myself that it’s not just around the corner, and finally, out of my longstanding belief that it’s important to combat exaggerated claims about artificial intelligence. And so, after reading about how the old idea of artificial neural networks, recently adopted by a branch of Google called Google Brain, and now enhanced by “deep learning,” has resulted in a new kind of software that has allegedly revolutionized machine translation, I decided I had to check out the latest incarnation of Google Translate. Was it a game changer, as Deep Blue and AlphaGo were for the venerable games of chess and Go?
Back when I was working as a freelance translator, the early automated translation systems were just starting to reach critical mass. I recall at the time that the European Commission was starting to use one, turning translators into editors and proofreaders. A couple of my clients wanted me to work with pre-translated texts, and pay me less; I refused. Not only because the machine translations weren’t good, but because you may end up spending more time fixing a bad translation, and you are influenced by the text you start editing, and end up with something that is a hybrid between computerese and real language. While this type of automatic translation can be useful for technical translations, where nuance isn’t needed, is it up to snuff for more subtle translation?
Douglas Hofstadter is not only the author of, perhaps, the most interesting book about the mind ever written – Gödel, Escher, Bach: An Eternal Golden Braid (Amazon.com, Amazon UK) – but also of the most interesting book on translation ever written: Le Ton Beau de Marot (Amazon.com, Amazon UK). Hofstadter has a deep sensitivity for language, and this book discusses language and translation.
In this Atlantic article, Hofstadter takes a look at Google Translate, wondering exactly how good it is. After all, beating go professionals was an unexpected event; could it work as well with language?
He shows why not, and points out some everyday subtleties that machine translation can’t handle. However, much of machine translation is based on corpus linguistics (huge databases of source and target texts), and, over time, it will improve. It will never match a human translator – translation really is an art – but for all but literature, it my become viable.
A minor quibble about Mr Hofstadter’s French translation of the first example. Where he suggests “ils ont tout en double,” I would instinctively say “tout vient par deux.” This is a more colloquial expression, and, while Google doesn’t seem to be very familiar with it – a Google search for the phrase only turns up seven hits – it’s definitely something I recall hearing when I lived in France. A more common expression – according to Google – and one that is closer to Mr Hofstadter’s choice would be “tout est en double.” Given the context – In their house / dans leur maison – one wouldn’t need to use a pronoun (ils) in the second clause of the sentence.
But this is what makes translation an art; different translators will use different expressions according to their idiolect.