Ducks, Eagles and Machine Translation

I can't remember when I first heard the metaphor of ducks and eagles, but I do remember that I took to it like a duck to water (or like an eagle to…air?). A quick Internet search reveals that the concept was first put forward by Ken Blanchard, a celebrated leadership author, and if you haven't heard it before, the basic idea is that some people are like ducks, constantly complaining ("quacking") and waddling around in groups doing their job without thinking about it, while others are like eagles, and use their intelligence, think for themselves and soar above the rest.

I was reminded of this metaphor the other day while reviewing a translation by an anonymous translator. The original document had been written in a hurry and had a few typos here and there, and there were also a few business-specific terms that were a bit hard to understand out of context. Far from trying to fix these problems in the translation, or (perish the thought!) bring the typos to the attention of the client and provide some added value, this thoroughly duck-like translator was more than happy to go ahead and reproduce errors in the target text. That's how the Spanish version ended up pronouncing that users had the right to "reuse" any treatment at any time (the intended meaning had obviously been "refuse"). And, my very favourite, participants in the study were warned of the real risk that their disease might get better (clearly the original had omitted the word "not"). You might think it was an automatic translation, but it was not. I could almost imagine the translator, presumably satisfied with their work, shrugging their shoulders as if to say, “Well, that’s what it says, and they pay me to say what it says, what else am I supposed to do?”

In translation today, as in many industries, ducks are a dying breed. Whether we like it or not, machine translation is making great strides, and it keeps improving every single day. For some language combinations and subject matters, progress might seem slow, but it is happening, always chipping away at the language puzzle behind the scenes.

As translators we tend to focus on its weaknesses, of which there are many: machine translation is mindless and bound to make comical mistakes, and it is obviously inadequate when it comes to cultural adaptations, transcreation (reimagining content for a new language audience), and anything subtle that cannot be approached one line at a time. If the world is divided into ducks and eagles, you could say that machine translation is the ultimate duck: doing what it's told, sucking in words at one end and spitting them out the other, without ever processing anything at the level of meaning. The problem for translators who bring this same mindless attitude to their work is that machine translation is not just another duck—it is Superduck! Think about it: Google Translate can translate 20,000 words in an instant, before most of us have had time to blink, let alone read our to-do list in the morning. On top of that, it’s available all hours of the day and night, never complains about working conditions or tight deadlines, and it is often dirt cheap or completely free.

Just about the only competitive advantages that we humans can offer are our eagle-like qualities: our capacity to think outside the box, to put ourselves in the shoes of the writer and the target readers and to use our intelligence to understand, process, and communicate information. We are not just "saying what it says," as machine translation does at its very best, but rather, pursuing the deeper goal of communicating a specific message from one human mind to another. Sometimes, with very straightforward informational materials, we might feel that we are like little machines, part dictionary and part grammarian, systematically working through the text one sentence at a time. But in many cases we employ much more than our language skills: we use our uniquely human mental resources to infer what the author was trying to say, even if they said it confusingly, or with atrocious punctuation, or omitted a trivial little word like not. In these complex but very common cases, producing a good translation requires human creativity and ingenuity, and a deep understanding of the subject matter and the world in general. Depending on the context, adding an explanatory note, rearranging some text or discreetly slipping in an extra word or two to clean up a sentence might do the trick. Quite often, it is necessary to go back to the client and diplomatically ask, “did you really intend to say that?"

In the final analysis, whether it happens right away or takes many years, the reality is that machine translation will displace duck translators. We live in a time when computers are starting to take over many things we once thought of as uniquely human skills, so it should not come as a surprise that many expect machine translation to supplant the much slower and more expensive organic translators.

Once that brave new world dawns, the only translators still flying high will be the eagles: those who can read more than just the written words, see beyond them and make creative leaps based on true understanding.

For now, our job is not just to educate our clients about the value of professional human translators, but also (and perhaps more importantly) to live up to that value in our daily work, and let the difference speak for itself.