The Future of Language: Considering the Sapir-Whorf hypothesis for translation, emoji, and pop culture
Image taken from Flickr user The All-Nite Images
By Elisa Gabbert
There’s an idea in linguistics that until a culture creates a name for a color, they don’t really see it as a distinct category. It builds from the anthropological discovery that languages tend to develop color terms in the same order: first, for black and white (or roughly, light and dark), then for red, then for either green or yellow and then both, then blue (and so on). They don’t invent a word for blue, the thinking goes, much less for mauve or taupe, until they need it. Color terms proliferate in a world of dyes and spectrometry.
This has led some linguists and scholars to suggest that Homer’s “wine-dark sea” was not just a weird poeticism but evidence of the ancient Greeks’ entirely different perception of color: “As late as the fourth century BC, Plato named the four primary colors as white, black, red, and bright” (via
It’s a chicken-or-egg conundrum though: How can the name come after the concept if you need the name to understand the concept? This problem of circularity always made me resistant to the Sapir-Whorf hypothesis in its strong version, which states that our thoughts are bound by the restraints of our language. The weak version, that language merely affects our thoughts, seems trivially true. What doesn’t affect our thoughts?
Taxonomizing the pleasure provides a frisson of meta-pleasure, like the ability to identify butterflies or wildflowers.
In a strong Sapir-Whorf world, you’d expect a lot of untranslatable concepts. But the appeal of lists of “untranslatable words” is not that the concepts are truly untranslatable—if they were, we wouldn’t be able to read the lists. Rather it’s the novelty of the formations, which is only apparent when they are translated literally, usually into awkward strings and phrases that our own language doesn’t compress into a single “word.” They are not, as my husband put it, “elegantly translatable.”
Take the word schadenfreude: English speakers love it precisely because it gives a name to a feeling we are already familiar with. Learning the word doesn’t teach us how to feel it, but it does, perhaps, sharpen the feeling, by heightening its specificity—we become more apt to recognize it. (I’m a sucker for compound German words because one can’t help breaking them down into parts: in this case, harm-joy, or damage-delight.) Taxonomizing the pleasure provides a frisson of meta-pleasure, like the ability to identify butterflies or wildflowers.
Thought experiment (gedankenexperiment): try to imagine an untranslatable color, a color term from Swedish or Swahili that English has no equivalent concept for. It’s hard because there’s only one spectrum of visible light, and humans all have the same color receptors (red, green, and blue). Once you have a robust color vocabulary it’s easy to describe, or “translate,” any color you can imagine – like X but lighter; like Y but more blue. (Birds’ eyes have a fourth receptor, for ultraviolet; perhaps owls and hawks see untranslatable colors.)
There is such a thing as an “impossible color,” and ways to trick yourself into “seeing” a shade like “reddish green.” They involve optical illusions, not the kinds of mental gymnastics some people do to visualize multi-dimensional objects like hypercubes. I don’t follow the Twitter account
The following sentence came to me, several months ago, as a complete thought in language: The Sapir-Whorf hypothesis but for emoji.
What I meant, before I knew what I meant, is that we can only express via emoji what we have emoji for. Emoji have an inexact relationship to language—I don’t think of the eyes emoji as “side-eye” or “surprise” or “wow,” though it’s somewhere near that semantically; I think of it as “the eyes of emoji.” As such, it’s closer to a reaction gif than a word. Emoji, unlike words, have no parts of speech; there is no syntax of emoji. So you get these pileups that don’t coalesce into complex structures the way sentences do, with diagrammable clauses and referents; emoji are merely additive, like multiple exclamation points.
Emoji are useful when you have nothing much to say. They can be used tactically to end a text conversation in much the same way as favoriting a tweet. A cute, public “mark as read.” But they’re a cop-out, a cheat, an avoidance of the hard work of precise communication, due to the impoverished lexicon. Over 800 emoji are supported across most platforms, more on iOS 9, but many, like stock photos, are largely unusable. Have you ever used any of the various building emoji? Why would you bother with an interrobang emoji when you can type out an interrobang? English, on the other hand, gives us access to over one million words. Converting my thoughts into this limited set of pictographs entails a great deal of compression. Emoji are palpably “lossy,” like bad MP3s.
This has a whiff of Sapir-Whorf to it, suggesting thoughts that don’t convert readily into language aren’t thoughts at all, but something illusory that crumbles at the touch, like dreams you never retell or write down.
My reticence, my infelicity with emoji, could of course be a fluency issue. The Japanese, I don’t doubt, are capable of subtle emoji nuance. But I came to emoji late and don’t expect to achieve fluency. What I want when I write – even in email, even on Twitter – is clarity. Language maps less sloppily to ideas than emoji map to language, but complex ideas are still difficult to explicate, even in one’s “native tongue.” “If you can’t explain it simply, you don’t understand it well enough” is one of those apocryphal Internet quotes often attributed to Einstein. He probably didn’t say it, but the physicist Richard Feynman reportedly said, after being asked to prepare a freshman lecture on why spin-1/2 particles obey Fermi-Dirac statistics, “I couldn’t reduce it to the freshman level. That means we really don’t understand it.” This has a whiff of Sapir-Whorf to it, suggesting thoughts that don’t convert readily into language aren’t thoughts at all, but something illusory that crumbles at the touch, like dreams you never retell or write down.
You have to assign your ideas to language (or equations, or musical notation) if you want to convey them to others, but it does more than make them communicable—it seems to reify the ideas, as though we’re translating them back to ourselves, turning raw data into something we can read. This happens automatically with other systems; we have no conscious awareness of our brains converting the eyes’ input into an approximate world model. In comparison, thought-into-language processing can feel exquisitely slow.
In his essay “E Unibus Pluram: Television and U.S. Fiction,” David Foster Wallace quotes an unnamed former instructor from his graduate school days who said “serious fiction must be timeless,” avoiding references that date a story to the “frivolous Now.” I got into a discussion with a few writer friends about this—if we want to be read five or twenty or two hundred years in the future, how are we to handle pop culture (or culture at all)? Can we put emoji in our short stories? Should we mention brand names that, while currently ubiquitous, may not exist in a few years? The example one friend used to illustrate such a decision was “phone” versus “iPhone” versus “Samsung Galaxy S7.”
To me the real issue with writing “Samsung Galaxy S7” into a story is not the prominent timestamp but the over-specificity. If a character in a movie opens a can of soup, it needs to be a specific kind of soup (just one of many reasons why advertisers are a lot more interested in movies than books). But a character in a book can just make a can of soup; we as readers can visualize whatever soup we like. There’s no blank space in our mental picture where the label should be. Or, if there is, we don’t notice it, because our attentions are elsewhere—like those experiments where, asked to focus on a moving ball, you don’t see a guy in a bear suit walk through the room.
Yes, language changes, but often culture changes while language doesn’t: same words, totally different connotations. In a matter of decades, subtleties could be lost. Will our children know about Apple-product snobbery?
A writer may decide to provide a much more granular level of detail (e.g. Amy’s Organic Lentil Vegetable), but that choice should be governed by something, some consistency in the resolution of the world being built. It’s like significant digits—it doesn’t make sense to forecast tomorrow’s temperature to three digits after the decimal point when there’s variance from block to block and our prediction models aren’t that accurate anyway. Why zoom in on the brand of phone or soup if the rest of the world—the characters’ faces, the layout of the floorplan, the numbers on the clock, be it digital or analog—are out of focus? It would be like having only part of the screen in HD. You can do it, but there should be a compelling reason, or at least a compelling effect.
Is “timeless” fiction even theoretically possible? Whether or not you mention brand names, the cultural milieu is going to creep in. Sometimes I wonder why we don’t have more diachronic (across-time) English-to-English translations. Yes, language changes, but often culture changes while language doesn’t: same words, totally different connotations. In a matter of decades, subtleties could be lost. Will our children know about Apple-product snobbery? Must we footnote everything? In the introduction to her new translation of Madame Bovary published in 2010, Lydia Davis notes:
These markers are the “frivolous Now” of mid-19th century France; their Pompadour clock is our iPhone.
For every translator, there must come a moment of reckoning, of wondering, What precisely am I translating? Flaubert was famously a stylist, who believed “a good sentence in prose should be like a good line in poetry, unchangeable.” (If unchangeable, then untranslatable.) Odd, then, Davis writes, that “many of the translations do not try to reproduce that style, but simply tell this engrossing story in their own preferred manner.” She cites the “lush, loquacious” 1948 translation by Gerard Hopkins (nephew of the poet), which “added material in almost every sentence.” A text is only made of language, and yet the language seems to generate some other, ineffable, epiphenomenal essence—a spectral text that survives when all the language changes. Infinite translations, infinite texts.
I read that Kafka’s Metamorphosis is difficult to translate into Japanese because of “insect appreciation”—that is to say, the Japanese do not experience revulsion at the prospect of a man-sized beetle. What to do, then—convert Gregor Samsa into something that the Japanese do find disgusting? Or let it become a new story in a new context? Which is more accurate, more faithful to the original? Imagine reading Metamorphosis without understanding why Gregor’s family is repulsed by him—after all, he has transmutated into something wondrous, something perhaps better!
Everything is translatable, but nothing is perfectly translatable: tidy words become gangly phrases, the “Kafkaesque” becomes fantastical, innuendos appear or disappear, polysemy and rhyme seem to teleport to a new location in the poem. Meaning dissipates in the processing, decays over time, but it’s remarkable how much is retained, the way it’s remarkable how good the Lascaux cave paintings are (always I am struck: such good horses!). Problems in translation are not an argument against translation. The work can remain what it is while also being transposed, twisted, given new significance, like a band of paper taped into a Mobius strip, a glove turned inside out.
Elisa Gabbert is the author of L’Heure Bleue, or The Judy Poems, forthcoming from Black Ocean in September 2016; The Self Unstable; and The French Exit. Recent poems and essays can be found in Jubilat, Harvard Review, and The Smart Set. Follow her on Twitter at