It's all s̶e̶m̶a̶n̶t̶i̶c̶s̶ pragmatics! The new faces, hands, and phallic vegetables of grammar

I feel pretty confident in saying that the people, on average, are more literate in today’s world than at any other point in history. I would even wager that we are quickly approaching an era, if we aren’t already in it, where the majority of our conversations with others are mediated via written rather than spoken language: I don’t know about you, but I receive far more Facebook messages on my birthday than phone calls or hand-delivered, mouth-congratulating, vegan ice cream cake. Though my birthday is on June 23 if you feel like proving me wrong.

Linguists like to repeat the mantra that spoken language is the primary empirical domain of linguistic inquiry — and it is true that while there are spoken languages that have never had writing systems, there are no written human languages that have never been spoken (barring that notebook of fresh rhymes you’ve been secretly working on). However, it is also naive to think that a language’s orthography doesn’t sneak back into the language itself to some degree: consider acronyms such as NPR, DVD, and USB, which are words formed from the initial orthographic letters of a larger phrase. Or, for the etymologically snobby, consider the history of the spelling and pronunciation of the article ye, where the initial “y” sound in modern English is the result of conflating two visually similar symbols.

Recently, Facebook unveiled a new slew of reaction emoji (I realize they may officially be called ‘graphicons’, but I’m happy calling a spade a shovel):

original image source: http://static6.techinsider.io/image/56cdc08a2e526556008b9359-1490-762/reaction.png

While they may have impressed your snooping mother as she ❤s your new Candy Crush score, they allegedly failed to exactly wow linguists:

In the article, Geoff Pullum and Susan Herring, linguists far smarter than me, weigh in on the syntax of the new reactions as being inconsistent and nonsensical. And they are right: interjections and verbs and nouns and adjectives are all distinct lexical categories. On the other hand, I think they are also wrong in thinking about emoji from a strictly syntactic perspective, as if we have simply swapped out certain words for cute pictures of disembodied appendages. Rather, I think it more helpful to think of what linguistic information a written orthography is attempting to codify; in general usage, emoji are primarily representing not phonological or lexical, but pragmatic information.


While semantics within linguistic theory is primarily concerned with the literal meaning of words, phrases and sentences, pragmatics is concerned with how these meanings are used by speakers to convey meanings beyond what is simply said — it is the study of how your spouse tells you they are ready to leave your holiday work party now. If semantics asks you, “But, dude, like, what does Every farmer who owns a donkey beats it REALLY mean, man?”, pragmatics tells you to stop eating those brownies and takes your car keys away.

One type of pragmatic meaning is conventional implicature, which is a lot of syllables dedicated to the idea that we sometimes agree that words or sentences have connotations that are not part of the at-issue meaning. When your roommate — let’s call him Keanu — asks you to buy some goddamn toilet paper, there are at least two meanings you get from this: the first is the literal semantic meaning that you need to buy toilet paper; the other is contributed by the goddamn, and tells you that your roommate is stranded on the potty with roughly 10 minutes before the pizza arrives, and that he likely has some strong feelings about your forgetting to buy toilet paper yesterday.

Linguists have many sly tricks to tease these types of meaning apart. One of them has to do with what can be negated by a follow up sentence:

Keanu: You didn’t buy any goddamn toilet paper!
You: Yes I did! It’s in the pantry!

In your response, you have corrected Keanu’s literal, at-issue assertion that you didn’t buy any toilet paper by telling him that you in fact did buy some, and inexplicable stashed it into the pantry with the new red velvet cake Oreos. However, what you cannot correct are the strong feelings he has conveyed about his belief that you dropped the ball.

The next day, Keanu sheepishly shows you the text he sent his first, and only, match on Tinder:

according to ur profile u like Chicago style hot dogs 😉

“Where did this all go spectacularly wrong?” he asks. What you have to tell him is although he made an innocent enough assertion semantically, he has conventionally implicated something far more lewd with the punctuating winky face. And, in fact, if he had not been already unmatched, a response of “i dont” would contradict his assertion, but not his creepiness.


Punctuation is an apt comparison, as I think it is the closest orthographic relative of modern emoji usage we have in English — and I am not simply referring to the evolution of emoticons into emoji (e.g. :P →😛) . Whether a written sentence is punctuated by a . or a ? or a ! tells you something about how it is pronounced, and, consequentially, how to interpret it; often there is even no overt change to the word order. Typing a word in ALL CAPS conveys the message that one is not only being louder, but also more enthusiastic about saying something. Typing all those “xXx”s around your screen name tells all your friends you’re waiting until you’re 21 to break edge. These are all orthographic tricks to convey the richness of spoken language that are otherwise invisible in writing — volume, intonation, stress — richness that can convey meaning beyond what is simply said.

Which raises the question: what linguistic information is a written system trying to capture? Writing systems that use an alphabet, such as English, seem to be primarily representing phonemic information; other systems, such as those based on Devangari scripts (e.g. Hindi, Khmer, Thai), Canadian Aboriginal syllabics (e.g. Cree, Inuktitut, Blackfoot), or Japanese katakana and hiragana, are called abugidas, and primarily represent syllabic information. Chinese Hanzi and Mayan glyphs are to some degree logographic, with characters representing words. Korean Hangul to some extent even represents phonetic information regarding manner and place of articulation.

Intuitively, all of the linguistic information primarily captured in the above writing systems has a spoken correlate. In the Wired article above, Professor Herring claims that emoji represent non-speech sounds. I agree, and would go further to say that emoji are innovative orthographic conventional implicatures — whew! that was a lot of polysyllabic terms; go ahead and take a second to digest that.

Writing systems, for all of our collective literary masterpieces, are crude sketches of the complexities that comprise language. And like any other technology, the longer writing systems loiter around, the more creative literate users get in hacking it and upgrading it. If you are a musician, there’s a good chance you might learned to read standard notation, to the point that it is easy to conflate the written sheet music for actual music. But notation is just that: an organized system of representation, which could capture rhythmic, melodic and harmonic structures reasonably well, but with the advent of modern classic, composers like Krysztof Penderecki found the need to notationally transcribe aspects that had been ignored up until that point. I am barely literate in standard notation, but can see that the sheet music for Penderecki’s “Threnody for the Victims of Hiroshima” captures far more nuance than what I have typically seen:

Still not the most abrasive thing you’ve heard on Youtube today

Oh yeah, sorry to make you listen to the sound of millions of souls being forcibly evicted from their bodies — you should have known with a title like that, though.

I believe what we are currently seeing with emoji is a revolution in being able to orthographically capture an entire domain of linguistic information that has previously only hovered intangibly around our sentences. They doesn’t always serve this purpose — sometimes an 🍆 is just an 🍆 — but they do allow the astute texter to capture how we have always talked in novel new ways. And those of us loathe to learn it are doomed to be poor, sad Keanu, trying to decode “lol byeee👋”.


Update [March 11]: Thanks to some sharp-eyed readers for pointing out that it is the Early Modern English ye and not the pronoun ye that is incorrectly pronounced in Modern English, as was mentioned in an early version of this article.

The earlier version of this article also referred to Chinese characters as kanji, which is the Japanese term; the update now refers to them as Hanzi, the Chinese term (though the characters have been adapted for use in other languages).