Q: Will machine translation replace human translators?
A: No—at least not in this lifetime.
The question Will machine translation replace human translators? may strike fear into the hearts of those in the profession. But translators shouldn’t worry. Machine translation will not replace skilled human translators anytime soon, if ever.
Why? Because languages don’t lend themselves to computer algorithms as readily as other human disciplines (e.g., arithmetic or design). To teach a computer to do arithmetic, you just need to teach it a set of rules. In math, those rules don’t change. The answer to 2 + 2 is always 4. No exceptions.
Design programs (from Adobe Illustrator to CAD software) work because the rules of geometry are constant. Tell a design program your desired radius, and you’ll get a perfectly drawn circle. Always.
Q: Why won’t machine translation replace human translators?
A: Languages are not closed systems, and their rules aren’t set in stone.
Human languages aren’t like math. They don’t have nice, neat one-to-one correspondences. There are several reasons for this:
- The rules that govern terminology and grammar change over time. Olds words die out, new terms are coined. In addition, syntactic conventions evolve. For example, using a split infinitive in English used to be grammatical heresy. Now it’s commonplace.
- One word can have multiple meanings. Linguists call this polysemy. Take the verb run. Its meaning in Usain Bolt runs fast is different than its meaning in My truck runs on diesel.
- One meaning can have multiple words. Consider contract, agreement, pact, treaty, concord, convention, and covenant. They are often used interchangeably.
Of course, technology can get better at dealing with the above. You can load synonyms into a database and program grammatical rules for a language into software.
You can even scrape the content from bilingual websites all over cyberspace and assemble a phrase-level machine translator. This is how Google Translate and Linguee, for example, work.
But these programs don’t translate per se. They can’t make congruity judgments. (A congruity judgment is the decision that a translator makes to best express the meaning of a source-text word or phrase in the target language.)
As our cartoon suggests, the advent of technology in a given industry doesn’t mean an end to the need for humans in that sector.
Take aviation, for example. Autopilot technology has been around for over 100 years. But we still use human pilots to fly planes.
Q: Are human translators anti-technology?
A: Not at all. In fact, translators use technology to make their work better.
We use find-and-replace tools to work smarter and faster. In addition, we use terminology databases to research terms of art and computer-assisted translation (CAT) tools to ensure terminology consistency in large documents. In short, we use technology to help us, not to do our job for us.
My prediction that machine translation won’t replace human translators doesn’t come from a place of fear. I freely admit that machine translation is getting better all the time. But when Google Translate produces the English translation It quail for the simple French phrase ça caille (It’s freezing), then the technology still has a ways to go.
So will machine translation replace human translators? No. In fact, if machine translation tools ever rival skilled human translators in terms of word choice, correct register, and stylistic nuance, I’ll be the first to admit it. I’ll even blog about it. But we’re not there yet.
Rest easy, human translators. If you’re good, you’ll have job security for a long time.