In 2006, Google came out with Google Translate and people went nuts arguing over whether or not that would mean job losses for translators. At the time, while sustaining that it takes a human to understand and convey language nuances and expressions of uniquely human abilities like sarcasm or humor, most translators agreed Google Translate was rubbish and there was no way a machine could ever really replace us. Back then, I argued that Google Translate would evolve, it would get better and some jobs would indeed be lost, and what’s more, they should be! Before you bite my head off for that last bit, please read on, I can explain.
Microsoft has announced the launch of its new Star Trek translation technology that renders real time translation of spoken language. Like its predecessor, Google Translate, it sparked the debate all over again. So again, here’s what I have to say: eventually some jobs will be lost. I’ve been following Google Translate since its inception and guess what? It was designed to learn and evolve –and it did. We’re talking about a program that in 2006 could not translate sentences with plural subjects or subjective clauses and now it can. In some languages, it even does so pretty well. The reason is that for the past 8 years users have been feeding Google Translate with the necessary corrections and information for it to evolve. But what can Google Translate get relatively right? Grammatically correct source sentences with basic language structures, particularly, simple instructions, like those found in manuals, mostly in Western languages being translated to and from English. If the source is grammatically incorrect, Google Translate will probably fail; but when the source is well-written and simple, then Google Translate can do the job about as well as a one or two-cent-per-word translator (yes, those exist!).
Recently, Microsoft embarked on a mission to create “even better” translation technology and we’d be lying to ourselves as translators if we failed to admit that a lot of what some translators do (again I mean one or two-cent-per-word translators) is really uncomplicated enough to easily be replaced by machine translation and then merely tweaked a bit by a human. Many argue that this doesn’t mean jobs will be lost; instead, the role of translators will simply change. Though the second premise is true, the first is mathematically unsustainable. Some translators will experience this change from linguist to editor, but it is not cost-effective for all translators to survive the cut. So while some cling to the editing raft, others will inevitably sink.
But is this really such a big deal, anyway? Even though technical manuals and similar texts constitute a large part of what actually gets translated, there is also a myriad of texts that are not that simple and still fall under the category of things containing uniquely human expressions and nuances that simply cannot be captured by a machine. In fact, sometimes, they can’t even be captured by another human with a lesser command of language, lower cultural level than the author of the source, or unsuitable background knowledge for the task at hand! Sometimes, even humans fail to convey the essence of a complex source text, and in my area of specialization, that includes legal translators, for lack of familiarity with either jurisprudence or nuances (not of legalese, but of legal systems altogether). So while I think good legal translators (like translators working in many other complex areas of translation) are nowhere near threatened by machine translation, I’m afraid I can’t say the same for those two-cent translators out there.