“The purpuse of the pruposal”.

Therese had typed those words and she was now staring at the highlights her botched spellings had caused. She retyped: “the perpose of the praposal”. Again, she could only stare with stunned unfamiliarity at the highlighted words. When she tried to select “perpose,” her finger glided too close to the line above. The software asked “Did you mean to select ‘perpose’?” When she tapped “Yes,” only one alternative was offered: the correct spelling. Even looking at this correct spelling, there was a moment where Therese still felt that “purpose” must be incorrect. (“Purr pose?” she thought.) There was then another moment where she had to remember what the word meant, and then a further moment where she had to remember what she was typing in the first place.

What she was typing was a slightly better phrasing in a software-translated document. She no longer translated entire documents on her own, as she did when she started with Multiversant. And she no longer looked for the software’s errors, as she did for several years. There were virtually no errors anymore—no errors of definition or errors of grammar, certainly. And, since the version before the last, Therese could hardly improve upon Multiversant’s choices with idioms and metaphors. There were always optional differences in translation, according to context or style or audience. And there were the ambiguities inherent in language, as well as the sloppiness inherent in authors, or typists. But Multiversant even recognized passages that it could not clearly translate. It recognized them, termed them as “disambiguation candidates,” and sent them to Therese’s department. In these senses, the software made no errors.

Therese made errors, lots of errors, constantly. But she instinctively felt the difference between “We did not complete the review” and “We have not completed the review”. She would have trouble explaining just why there was such a difference, but it wasn’t her ability to explain that mattered here. It was, ultimately, her ability to feel hurt that mattered. The ability to detect intention in “I didn’t read your e-mail” versus “I haven’t yet read your e-mail” is what mattered. And this seemed like the only reason that a human was still involved in translating sentences from one language to another: to mind the sensitivities.

Her ability to feel hurt was being well exercised today. Her head hurt. Not terribly, but it made thinking feel like strained exertion rather than weightless flight. She could go have a cigarette, but she’d have to ask for one when the office thought she had quit. She thought of an early lunch, but the only food that appealed to her was sugary or starchy. Her body craved exactly what made her feel unattractive. Was she all raw nerves and dumb instinct? Were all of her judgments as approximate and slow as her memory now was? She possessed a nearly native proficiency in four foreign languages and yet had to write her eight-character password down so she didn’t forget it.

She wanted to find an error in the document, an error that revealed the limited domain of the machine. She caught a possible error with “interested parties that are closed to the measure”. She repeated that phrase—“that are closed to the measure”. Then she realized it actually said “that are opposed to the measure”. The software appeared to process a thousand-word document all at once, whereas she had to go word by word, sometimes mouthing the words, and then doubling back to reread what she hadn’t paid attention to.

Her girls were using Multiversant now. In the last light of school-night evenings, they would run barefoot across the back lawn, giggling their thrillingly forbidden curse words into their phones. One would then hold her phone out to the other while Multiversant’s calmly certain voice translated these words into Senegalese, Tamil, Romanian. When the other’s phone would then recognize the language and translate it back into English, the decoded secret would send them into debilitating laughter.

Therese listened keenly to these translations. She listened for the usual tab-keeping on her daughters. And she listened out of hope that a language she knew would be mistranslated by the software. That hadn’t happened yet.

At breakfast, Therese asked her oldest daughter whether she wanted to take Spanish next year. Carey held her hand over her mouth and whispered into her phone. The phone’s smugly omniscient voice said “Ya yo hablo español”.

“You do not speak Spanish, Carey.”

Quivering in her seat, Carey sang “Yo ya! See? Yo ya-ya-ya! I already do.”

Even these teases of Spanish brought the linguist out of Therese. She stood over her daughter at the table and showed her just how little she already knew: “Cuando usted viaje, sera grosero sostener el teléfono frente a las caras de las personas.

But Carey’s eyes just widened while she held down a smile. She lifted her phone into her mother’s face while it smugly translated everything Therese had just said: “When you travel, it’ll be rude to hold your phone up to people’s faces.” On the phone, inches from her face, Therese could see her company’s logo: it was a Picasso-esque icon of several faces in one face—facing forward and in left and right profile—talking in different directions. The intended effect of a “multi-conversant” person ended up looking like a cartoon of someone shaking their head after a concussion.

Carey sassily mimicked along with her mother’s translated words, doing her eleven-year-old’s version of an impression of Therese. This involved scrunching her face up into a caricature of nastiness and hinging and unhinging her mouth, doll-like, to emphasize the complete dismissal of her mother’s words. The impression told Therese that she was nit-picking about things that didn’t matter.

There was a new Multiversant ad that seemed to agree with Carey: “You Don’t Do Long Division In Your Head. So Why Conjugate Another Language In Your Head?”

No, she didn’t do much math in her head. She hardly even dealt with double-digit figures in her head anymore. A 25% discount on $30, she thought, must be … less than $10 because 10 would be a third of 30. But 25% of $30 would be … what? Where the answer should be, or at least the method to the answer, was a terrifying absence.

This shouldn’t be comparable—numbers and language. Except that the software’s universal fluency made them comparable. (“Converse in the Multiverse,” another ad read.) Numbers added to other numbers equaled an already known number. A sentence in one language more or less equals an already known sentence in another language. (Her whole career, and the stature of her entire being, seemed to depend on the shrinking size of that “more or less”.) You could still do the math on paper and you could still translate with an slow, error-prone head (a slow, hurting head). But the answers would soon no longer be more correct than what was provided automatically.

Learning another language once gave Therese an outsized pride. It made her as boastful as her girls were when they thought they were juggling with just two tennis balls. Language wasn’t just a skill or an answer. It was an embodiment, a story, a relationship. Each language you learned was like entering into a new romance. You were lightly flirting as you alighted at the sound of those first introductory words. You were kissing as you flaunted around those first phrases; people were enviously watching and you felt instantly, prematurely fluent. You were dating as, night after night, you absorbed the basic grammar. Dating was sometimes fun and dating was sometimes arduous, but it ended once you started to look beyond the present tense. When you started to deal with the irregularities, with the charming and bewildering quirks of gender, with the new families that came in past and future tenses—when you were determined to make all of this work, this was the commitment. If you didn’t make this final commitment, you could end up learning and then forgetting an entire language, an entire relationship.

Therese considered herself extremely fortunate to have had translation jobs that kept her actively in love with her four lingual husbands: her Mandarin, her Spanish, her German, and her Bengali husbands. She used to be needed anytime someone wanted to talk to her husbands. It was petty that she was jealous that anyone could now talk to them without her, but the jealousy prickled her nevertheless.

For a short time, the ability to speak to any human being on earth through a reliable translator in your hand had seemed miraculous, even to Therese. She loved testing it out, trying to trace the limits of its nuances. But then instant translations had become just another utility, like free long-distance phone calls or mapping directions online. They weren’t miraculous anymore. And neither, to a worrisome extent, were other languages.

 

Next »