Translation Myth: The Rise of Machine Translation

Why do most post-apocalyptic films seem to center on the “rise of the machines?” You have to wonder why we keep imagining that giving a machine enough intelligence means we’re doomed to mechanical overlords. It’s funny because this instinct carries over to our field – that somehow, the rise of machine translation is making human translators obsolete.

Nothing could be more false though. Machine translation is nowhere near as capable at translating, well, anything when compared to human translation. Smartling CEO Jack Welde recently spoke about this in an interview for GLG Research’s Innovation in Action series:


Machine Translation > Human Translation?

Not a chance. As we’ve noted before, machine translation isn’t some throwaway technology. It’s quite advanced. But it can’t compare to professional human translation, and it likely never will.

Computers are great at pure logic but humans understand context. Human translators can deduce meaning and contextual clues; computers can’t. Yes, we can teach a computer to play Jeopardy! But that’s not nearly as complex as understanding language and culture.

For example, machine translation can pick up on regional preferences. Cold air in Brazilian Portuguese is ar frio but a different word for cold is used when describing water: água gelada. It’s not a huge error to use água fria for cold water but it does sound odd in Brazil. It’s similar to using the wrong term for public transportation. Here in New York, it’s the subway; in Washington, DC, it’s the Metro; in Boston, it’s the T; etc. A professional translator understands this difference; computers do not.

Professional translators can also decide on what words to use based on the tone of the original source text. Room temperature water in Brazilian Portuguese is água em temperatura ambienteor água em temperatura natural. A professional translator would know the first option is better suited to instruction manuals and the second option is better suited for casual, conversational use. Again, machine translation can’t pick up on these subtleties.

So, it seems the future of translation is safe from the machines. As for the rest of the world…