Ça marche comme ça: The Humanity of Google Translate

Today, one of my French colleagues teaching French in Montpellier, France, sends me this quick reply to an email I sent her yesterday:

Merci, Kurk.
Ça marche comme ça.

Running that through Google Translate to get an output into English, we get this:

Thank you, Kurk.
It works like this.

You need more context, you say.

Of course you do, I respond.

We are trying to decide together whether one of the English learners in the university program I run might serve her French learners well in the French House of our institution. I’ve been in touch with the student a bilingual native speaker of French from the former Francophone nation of Rwanda, and she’s with the native English speaking students there in France.

You need more context, you say about my colleague’s reply.

And I agree with you again. Google Translate has given only so much to say, “It works like this.” To “know” French, and to communicate it into English, the machine translator would need to know a bit more than just the usual translations of this phrase done by humans. We could find a collection of them at a website like this one:

http://www.linguee.fr/francais-anglais/traduction/

This is somewhat how the human programmers of the machine translator “Google Translate” have created the algorithms for such translating. (Here at BLT, we’ve discussed Google Translate and its ways here, here, here, and here – if you’re interested.) The lack of context — even with all the different ways humans might move from French to English with “Ça marche comme ça” as shown in the link above — points to the lack.

Without my showing you the email thread, or without my telling you about it a bit, there’s no way you could really ascertain why my colleague would reply to me this way. And Google Translate has a hard time knowing and then showing you much either, in this case.

In this case, my colleague presumes I know what is deeply profound for her. Or she just says what she says without presuming my familiarity with her mother tongue. As it happens, she grew up speaking French in a French speaking home and going to schools in which French is the language of instruction in France. And as a little girl, she heard and sang and perhaps even moved her body to a song with this phrase in it. With some help for the monolingual English speakers, here are a couple of examples:

and

(from here)

For me the context of my thinking about this today is what Gideon Lewis-Kraus wrote in an article in today’s New York Times. The title implies a binary: “Is Translation an Art or a Math Problem?” The essay is a brilliant and wonderfully insightful look at translation running from the Enlightenment through Star Trek through “the principles of Chomsky’s ‘universal grammar’” through “’traduttore, traditore,’ a common Italian saying that’s really an argument masked as a proverb” and through its meanings in English (i.e., “literally, ‘translator, traitor,’ … semantically on target [if without an English phonological] match [to] the syllabic harmoniousness of the original [Italian], and thus proves the impossibility [of full translation to English] it asserts”) and through the “wonderful survey of the history and practice of translation, ‘Is That a Fish in Your Ear?’ [by] the translator David Bellos [who] explains that the very idea of [translator betrayal or] ‘infidelity’ has roots in the Ottoman Empire…. [in] a hereditary caste of translators, the Phanariots.”

Lewis-Kraus then brings us English readers to “a new Phanariot class, … native speakers of C++, … not particularly loyal to any language at all.” He’s describing the programmers for Google Translate. He says they only know C++ and, well, only English. In other words, some of these “translators” or C++ Google Translate coders assert that it is better to be English monolinguals. He laments that translation can be so divorced from, well, from languages, from bilinguals, from linguists, from translation artists and translation scientists. Translation, he laments, seems now married to, or at least is having illicit affair with, mathematicians. He quotes Susan Bernofsky:

As the translator Susan Bernofsky put it to me, “They create the impression that translation is not an art.”

He opposes that with his quotation of a computational linguist:

One computational linguist said, with a knowing leer, that there is a reason we have more than 20 translations in English of “Don Quixote.” It must be because nobody ever gets it right. If the translators can’t even make up their own minds about what it means to be “faithful” or “accurate,” what’s the point of worrying too much about it? Let’s just get rid of the whole antiquated fidelity concept. All the Sancho Panzas, all the human translators and all the computational linguists are in the same leaky boat, but the machinists are bailing out the water while the humans embroider monograms on the sails.

I like the metaphor of the boat, of the common place of the math machinists who get at translation and the humans who translate. I reject the binary, nonetheless. What would happen if a bilingual linguist translator human knew C++ and French and English?

And I love the concluding sentences Lewis-Kraus crafts in English to end one of his well-constructed paragraphs:

In a sense, their machines aren’t actually translating; they’re just speeding along tracks set down by others. This is the original sin of machine translation: The field would be nowhere without the human translators they seek, however modestly, to supersede.

He is confessing, conceding, that the would-be mathematicians-only never ever get away from the also-artistic humans translating.

Roll as we walk,
as we walk.

It works like this.

Ça marche comme ça.