Translator, Translate Thyself!
Regular readers of this blog know that I rarely stray from the functional description given in its tag: “Commentary on the European non-English-language press.” Naturally, I make this assertion as the prelude to one rare instance where I violate that mandate. In mitigation, though, it should also be fairly obvious how close to the functioning of this blog issues of translation are, which moves me to bring up for discussion the “I, Translator” article by Princeton Translation Program director David Bellos published in last Saturday’s New York Times.
As you might expect, the growing capability of machine translation (with the translation facilities provided for free by Google in the vanguard) presents me with a number of fairly challenging questions. Did I simply waste all that time of my past, of my precious youth, learning the various languages that I claim to be able to use at present? (And am I wasting it now as I continue to study others? I’m afraid I can’t stop myself.) Are the translation assignments I occasionally get to earn a bit of money fated to dry up? Is there indeed any point anymore to a weblog supplying “Commentary on the European non-English-language press” when anyone can now plug any given article into Google Translate and read it? (I still don’t believe that last part is actually true.) With these worries in the back of your mind, you expectantly click on an article like I, Translator, one that purports to defend human translation and foreign language capabilities, hoping for a encouraging ego-boost for the home team, for your side, for those who master foreign languages the old-fashioned way. I mean, hey, this is from a Princeton guy!
Did anyone else suffer as bitter a sense of disappointment at what the article actually turned out to say as I did? As for you, Mr. Bellos – Did it have to be so hard? All you needed to do was provide a convincing listing (and explanation) of machine translation’s disadvantages vis-à-vis human translation, maybe with a few disadvantages in the other direction thrown in at the end to preserve an even-handed, judicious aura. What we got instead was almost the opposite. Machine translation (although from Carnegie-Mellon, not Google) saved lives in the Haitian earthquake! Google should be OK “for maybe 95 percent of all utterances,” probably even for use in translating lower-quality literature that “employ[s] only repeated formulas” in its language.
Damn, Mr. Bellos, you’ve given away most of the store by this point; what’s left, if anything, that human translators would be able to do better? True “literary translation” is what’s left, “works that are truly original – and therefore worth translating,” although even then “human beings have a hard time of it, too,” i.e. will still be liable to get things wrong. Gee, thanks. Of course, three paragraphs previously we already learned that there’s no need to use machines for literary translations anyway, since there are more than enough humans ready to do that work. Bellos seems to lose sight of the fundamental consideration that, although there are more than enough human translators available, all or most of them will demand to be paid for the work, while machines will not.
Add in the various other sloppy elements here – “two important limitations” to statistical machine translation are announced, but it’s never clear what the second one is; there’s a brief history given of machine translation, but one of doubtful relevance especially when space is at a premium in a high-profile column like this – and one comes to the end desperately hoping that Bellos did actually deliver a convincing treatment of the whole translation question but that it fell victim to brutal disfigurement at the hands of a human editor prior to publication. (It does not seem to have been fact-checked, in any case; see the appended error correction about Warren/William Weaver.) As it stands, with public advocates like this, those of us who remain exponents (and practicers) of human translation certainly stand in no need of any more detractors.
