A classic of Argentine literature, Antonio Di Benedetto's Zama is available for the first time in English. The novel, about a provincial magistrate of the Spanish crown named Zama, is a riveting portrait of a mind deteriorating as the 18th century draws to a close. Esther Allen brilliantly translates Di Benedetto's novel, and talks about the six-year process of bringing the book to U.S. readers.

No, Google Translate was in no way useful to my translation of the 1956 Argentine novel Zama: let's get that out of the way first thing.

However, during the six years I worked on putting Antonio Di Benedetto's masterpiece into English, I performed an experiment. Every so often, I'd run the first line through Google Translate, to see what it came up with.

The idea came from a 2010 New York Times piece that put Google Translate through its paces by having it "translate" the first lines of famous novels: Gabriel García Márquez's One Hundred Years of Solitude, Antoine de Saint-Exupéry's The Little Prince. The problem with this, as David Bellos pointed out in an Op Ed follow-up, is that Google Translate doesn't translate. It's a search engine, scanning a massive corpus of already-translated material for correlations with the phrase it's currently processing. And of course both novels are widely available in English. Connecting the original line to its existing translation is "an impressive trick for a computer," Bellos noted, "but for a human? All you need to do is get the old paperback from your basement."

In the Spanish-speaking world, Zama is an awe-inspiring, foundational work that has powerfully influenced generations of writers from Julio Cortázar, Roberto Bolaño, and Juan José Saer, to Sergio Chejfec, Daniel Saldaña Paris and Alvaro Enrigue. It's been translated into German, Italian and Polish, but has never before been published in English. Which made it the perfect Google Translate test case.

The first line is Salí de la ciudad, ribera abajo, al encuentro solitario del barco que aguardaba, sin saber cuándo vendría.

Here's Google Translate:

  • "I left the city, river bottom, to meet the boat alone waiting, not knowing when to come." (2010)

  • "I left the city, riverside down to meet the boat alone waiting, not knowing when it would come." (2012)

  • "I left the city, riverside below, solitaire game boat waiting, not knowing when to come." (2014)

  • "I left the city, banks down, the lone meeting the boat waiting, not knowing when he would come." (2016)

Google Translate's stock-in-trade is stock phrases, that vast majority of all speech and writing which simply repeats standard formulations. When you plug "Salí de la ciudad" into the Google search engine (which has rapidly become one of the literary translator's most indispensable and versatile tools for any number of purposes—such as, for example, determining whether something is a stock phrase) more than 3.5 million instances of that exact character string turn up. No surprise, then, that the translation of that first clause is a cinch—comes out perfectly every time. In my own agonized versions of the line, which also varied a good deal over the years, those three words—"I left the city"—never changed, either.

The Google Translate results feel less and less lucky as the sentence progresses, and with each new roll of the search engine dice. The 2012 is probably the best of the lot, but, like the 2010, does not hold together grammatically and gives the mistaken impression that a boat is actually there, waiting for the narrator. The 2014 conjures up a nonsensical game of solitaire, and the most recent version is the worst of all, adding a "he" that introduces a nonexistent second character.

Mistakes can be made by human translators, too—and sometimes forgivably. The biggest issue with the machine-produced language is its failure to convey the personality, character or time frame of the narrator: a man of the late 18th century (the year is 1790), marooned in Asunción, Paraguay, and in his own, echoing internal voice, existing in expectation of... something. (Zama is dedicated "To the victims of expectation.")

What has compelled Zama's readers most, over the decades, is its structure: both the overall structure of the whole and the calibrated, rhythmic intensity of each line, written, said Roberto Bolaño, with the "steady pulse of a neurosurgeon." Recreating that structure in a film, as Lucrecia Martel, one of Argentina's most renowned directors, has been doing for the past several years, involves translating the internal perspective of the novel's first-person narrator into the camera's inevitably external view. A translation of Zama into another literary language must recreate the character as the original novel does: through his voice, so that each successive word deepens our sense of the man speaking and his dilemmas, both internal and external.

Another unvarying element of the Google Translate versions is the word "boat" for barco, a perfectly correct rendering of a simple Spanish noun. But in English, we have the expression, "When my ship comes in," which ferries with it a long history of mercantile navigation and trade, Shakepeare's Merchant of Venice, investment, hope, expectation, salvation. (The closest corresponding expression in Spanish has its origins in the Bible: Cuando lleguen las vacas gordas, "When the fat cattle come.") Translating barco as "ship" brings an echo of that deep history of longing into our first, fleeting image of the narrator, Don Diego de Zama, who lives in the hope of a ship that will come to save him.

Google Translate has been great for literary translators like me, but not by making our work obsolete, as many people assumed it would. Instead, by inviting everyone to participate in the act of translation, it has created much greater awareness of how very challenging translation is and how elusive and evasive linguistic meaning—not to mention linguistic beauty—can be. Over the last decade or so, literary translation has experienced something of a Renaissance in the English-speaking world, becoming much more widely appreciated and studied. It may well be that Google Translate has played a part in that.

Nevertheless, many people still run phrases through Google Translate and assume that the results are some sort of unvarying, literal, mathematical, algorithmically precise translation. Once my translation of Zama is published, the search engine may locate my painstakingly-crafted interpretation of that first sentence, and present my version to its users as its own (probably with one preposition changed just a bit, since the search engine appears to be programmed to evade charges of plagiarism when matches are too exact). I wonder how long it will take before that happens, or if I'm increasing the chances of it happening by including my translation of the first line in this article.

But I'm not too worried about that, actually. In Zama, the taut balance of every sentence creates a rare literary phenomenon: silence. The words that make up the novel thread among vast, resonant silences. And silence is a human thing that no computer can ever be programmed to translate.

Here's the line:

"I left the city and made my way downriver alone, to meet the ship I awaited without knowing when it would come."