Photo by Brooke Anderson Photography/Getty


Who decides what words mean

Bound by rules, yet constantly changing, language might be the ultimate self-regulating system, with nobody in charge

by Lane Greene + BIO

Photo by Brooke Anderson Photography/Getty

Decades before the rise of social media, polarisation plagued discussions about language. By and large, it still does. Everyone who cares about the topic is officially required to take one of two stances. Either you smugly preen about the mistakes you find abhorrent – this makes you a so-called prescriptivist – or you show off your knowledge of language change, and poke holes in the prescriptivists’ facts – this makes you a descriptivist. Group membership is mandatory, and the two are mutually exclusive.

But it doesn’t have to be this way. I have two roles at my workplace: I am an editor and a language columnist. These two jobs more or less require me to be both a prescriptivist and a descriptivist. When people file me copy that has mistakes of grammar or mechanics, I fix them (as well as applying The Economist’s house style). But when it comes time to write my column, I study the weird mess of real language; rather than being a scold about this or that mistake, I try to teach myself (and so the reader) something new. Is this a split personality, or can the two be reconciled into a coherent philosophy? I believe they can.

Language changes all the time. Some changes really are chaotic, and disruptive. Take decimate, a prescriptivist shibboleth. It comes from the old Roman practice of punishing a mutinous legion by killing every 10th soldier (hence that deci­- root). Now we don’t often need a word for destroying exactly a 10th of something – this is the ‘etymological fallacy’, the idea that a word must mean exactly what its component roots indicate. But it is useful to have a word that means to destroy a sizeable proportion of something. Yet many people have extended the meaning of decimate until now it means something approaching ‘to wipe out utterly’.

Descriptivists – that is, virtually all academic linguists – will point out that semantic creep is how languages work. It’s just something words do: look up virtually any nontechnical word in the great historical Oxford English Dictionary (OED), which lists a word’s senses in historical order. You’ll see things such as the extension of decimate happening again and again and again. Words won’t sit still. The prescriptivist position, offered one linguist, is like taking a snapshot of the surface of the ocean and insisting that’s how ocean surfaces must look.

Be that as it may, retort prescriptivists, but that doesn’t make it any less annoying. Decimate doesn’t have a good synonym in its traditional meaning (to destroy a portion of), and it has lots of company in its new meaning: destroy, annihilate, devastate and so on. If decimate eventually settles on this latter meaning, we lose a unique word and gain nothing. People who use it the old way and people who use it the new way can also confuse each other.

Or take literally, on which I am a traditionalist. It is a delight to be able to use a good literally: when my son fell off a horse on a recent holiday, I was able to reassure my mother that ‘He literally got right back in the saddle,’ and this pleased me no end. So when people use literally to say, for example, We literally walked a million miles, I sigh a little sigh. I know that James Joyce, Vladimir Nabokov and many others used a figurative literally, but as a mere intensifier it’s not particularly useful or lovely, and it is particularly useful and lovely in the traditional sense, where it has no good substitute.

So I do believe that when change happens in a language it can do harm. Not the end of the world, but harm.

There is another fact to bear in mind: no language has fallen apart from lack of care. It is just not something that happens – literally. Prescriptivists cannot point to a single language that became unusable or inexpressive as a result of people’s failure to uphold traditional vocabulary and grammar. Every language existing today is fantastically expressive. It would be a miracle, except that it is utterly commonplace, a fact shared not only by all languages but by all the humans who use them.

How can this be? Why does change of the decimate variety not add up to chaos? If one such ‘error’ is bad, and these kinds of things are happening all the time, how do things manage to hold together?

The answer is that language is a system. Sounds, words and grammar do not exist in isolation: each of these three levels of language constitutes a system in itself. And, extraordinarily, these systems change as systems. If one change threatens disruption, another change compensates, so that the new system, though different from the old, is still an efficient, expressive and useful whole.

Begin with sounds. Every language has a characteristic inventory of contrasting sounds, called phonemes. Beet and bit have different vowels; these are two phonemes in English. Italian has only one, which is why Italians tend to make homophones of sheet and shit.

There is something odd about the vowels of English. Have you ever noticed that every language in Europe seems to use the letter A the same way? From latte to lager to tapas, Italian, German and Spanish all seem to use it for the ah sound. And at some level, this seems natural; if you learn frango is ‘chicken’ in Portuguese, you will probably know to pronounce it with an ah, not an ay. How, then, did English get A to sound like it does in plate, name, face and so on?

Look around the other ‘long’ vowels in English, and they seem out of whack in similar ways. The letter I has an ee sound from Nice to Nizhni Novgorod; why does it have the sound it does in English write and ride? And why do two Os yield the sound they do in boot and food?

Nobody in a 15th-century tavern (men carried knives back then) wants to confuse meet, meat and mate

The answer is the Great Vowel Shift. From the middle English period and continuing into the early modern era, the entire set of English long vowels underwent a radical disruption. Meet used to be pronounced a bit like modern mate. Boot used to sound like boat. (But both vowels were monophthongs, not diphthongs; the modern long A is really pronounced like ay-ee said quickly, but the vowel in medieval meet was a pure single vowel.)

During the Great Vowel Shift, ee and oo started to move towards the sounds they have today. Nobody knows why. It’s likely that some people noticed at the time and groused about it. In any case, there was really a problem: now ee was too close to the vowel in time, which in that era was pronounced tee-muh. And oo was too close to the vowel in house, which was then pronounced hoose.

Speakers didn’t passively accept the confusion. What happened next shows the genius of what economists call spontaneous order. In response to their new pushy neighbours in the vowel space, the vowels in time and house started to change, too, becoming something like tuh-eem and huh-oos. Other changes prompted yet more changes, too: the vowel in mate – then pronounced mah-tuh – moved towards the sound of the modern vowel in cat. That made it a little too close to meat, which was pronounced like a drawn-out version of the modern met. So the vowel in meat changed too.

Throughout the system, vowels were on the move. Nobody in a 15th-century tavern (men carried knives back then) wants to confuse meet, meat and mate. So they responded to a potentially damaging change by changing something else. A few vowels ended up merging. So meet and meat became homophones. But mostly the system just settled down with each vowel in a new place. It was the Great Vowel Shift, not the Great Vowel Pile-Up.

Such shifts are common enough that they have earned a name: ‘chain shifts’. These are what happens when one change prompts another, which in turn prompts yet another, and so on, until the language arrives at a new equilibrium. There is a chain shift underway now: the Northern Cities Shift, noticed and described in the cities around the Great Lakes of North America by William Labov, the pioneer of sociolinguistics. There is also a California Shift. In other words, these things happen. The local, individual change is chaotic and random, but the system responds to keep things from coming to harm.

What about words? There are only so many vowels in a language, but many thousands of words. So changes in the meanings of words might not be as orderly as the chain shifts seen in the Great Vowel Shift and others. Nonetheless, despite potential harm done by an individual word’s change in meaning, cultures tend to have all the words they need for all the things they want to talk about.

In researching Samuel Johnson’s dictionary for my new book, Talk on the Wild Side (2018), I made a startling find. Johnson, in describing his plan for the dictionary to the Earl of Chesterfield in 1747, wrote that

[B]uxom, which means only obedient, is now made, in familiar phrases, to stand for wanton; because in an ancient form of marriage, before the Reformation, the bride promised complaisance and obedience, in these terms: ‘I will be bonair and buxom in bed and at board.’

When most people think of buxom today, neither ‘obedient’ nor ‘wanton’ is what comes to mind (To my wife: this is why a Google Images search for buxom is in my search history, I promise.)

Turning to the OED, I found that buxom had come from a medieval word buhsam, cognate to the modern German biegsam, or ‘bendable’. From physical to metaphorical (the natural extension), it came to mean ‘pliable’ of a person, or – as Johnson put it – obedient. Then buxom kept on moving: a short hop from ‘obedient’ to ‘amiable’, and then another one to ‘lively, gay’. (William Shakespeare describes a soldier of ‘buxom valour’ in Henry V.) From there, it is another short jump to ‘healthy, vigorous’, which seems to have been the current meaning around Johnson’s time. From ‘good health’ it was another logical extension to physical plumpness, then to plumpness specifically on a woman, to big-breasted.

The leap from ‘obedient’ to ‘busty’ seems extraordinary until we look at it step by step. Nice used to mean ‘foolish’. Silly used to mean ‘holy’. Assassin is from the plural of the Arabic word for ‘hashish(-eater)’, and magazine from the Arabic word for a storehouse. This is just what words do. Prestigious used to be pejorative, meaning glittery but not substantive. These kinds of changes are common.

I don’t know how we did without hangry so long in English, because I spent about a third of every day hangry

Two paragraphs ago, I used the words ‘leap’ and ‘jump’. But we see the ‘leaps’ only when lexicographers, looking back, chop up a word’s history into meanings for their dictionaries. Words change meaning gradually, as a small number of speakers use them in a new way, and they in turn cause others to do so. This is how words can change meaning so totally and utterly; mostly, they do so in steps too small to notice.

Again, no chaos results. Every time buxom changed meaning, it could have theoretically left a hole in the lexicon for the meaning it had left behind. But in each case, another word filled its place: in fact, the ones I have used above (pliable, obedient, amiable, lively, gay, healthy, plump and so on). For useful concepts, it seems, the lexicon abhors a vacuum. (I don’t know how we did without hangry so long in English, because I spent about a third of every day hangry. But sure enough, someone coined it.)

There are several predictable ways that words change meaning. Some people insist that nauseous means only ‘causing nausea’. But going from cause to experiencer is a common semantic shift, just as many words can be used in both active and agentless constructions (consider I broke the dishwasher and The dishwasher broke). Yet true confusion is rare. For nauseous’s old meaning we have nauseating.

Words also weaken with frequent use: The Lego Movie (2014) was on to something with its song ‘Everything Is Awesome’, because Americans really do use this word rather a lot. Once powerful, it can now be used for anything even slightly good, as in This burrito is awesome. It can even be near-meaningless, as in Steven Pinker’s lovely example: ‘If you could pass the guacamole, that would be awesome.’

But do we really lack ways of communicating that we’re impressed by something? No language does, and English-speakers are spoiled for choice from the likes of incredible, fantastic, stupendous and brilliant. (All of which have changed from their etymological meanings of ‘unbelievable’, ‘like a fantasy’, ‘inducing stupor’ and ‘shiny, reflective’, by the way.) When those get overused (and all are in danger of that), people coin new ones still: sick, amazeballs, kick-ass.

The thousands of words in the language are a swirling mass constantly on the move. Again, when one piece moves, threatening a gap or an overlap, something else moves too. The individual, short-term change is random; the overall, long-term change is systemic.

At the level of grammar, change might seem the most unsettling, threatening a deeper kind of harm than a simple mispronunciation or new use for an old word. Take the long-term decline of whom, which signals that something in a question or relative clause is an object (direct or indirect), as in That’s the man whom I saw. Most people today would either say That’s the man who I saw or just That’s the man I saw.

What word is the subject in a clause, and what is the object, is a deeply important fact. And yet, precisely because this is so, even radical grammatical change leaves this distinction intact. Readers of Beowulf are in no doubt that virtually every word in that epic poem is vastly different from its modern counterpart. What those who can’t read Old English might not realise is how different the grammar is. English was a language like Russian or Latin: it had case endings everywhere: on nouns, adjectives and determiners (words such as the and a). In other words, they all behaved like who/whom/whose does (there was even a fourth case).

Today, just six words (I, he, she, we, they and who) change form when they are direct or indirect objects (me, him, her, us, them and whom). In a longer view, modern Anglophones speak godawful, brokendown Anglo-Saxon, lacking all the communicative power that those endings provided. How, one can imagine Alfred the Great asking, do English-speakers know what is the subject of a sentence and what are the objects without those crucial case endings?

The answer is boring: word order. English is a subject-verb-object language. In I love her, case is evident by the form of I (a subject, in the nominative case) and her (a direct object, in the objective case). But the meaning of Steve loves Sally is just as clear, despite the lack of case endings. Subject-verb-object order can be violated in special circumstances (Her I love the most) but it is expected; and that expectation, shared by all native speakers, does the work that the case endings once did.

To my six-year-old, everything is epic, which strikes my ear as awesome must have done my parents’

Why did the case endings disappear? We don’t know, but it was probably sped up as a result of two waves of conquest: adult Vikings and Normans coming to Britain, and learning Anglo-Saxon imperfectly. Then as now, things such as fiddly inflections are hard for adults to learn in a foreign language. Many adult learners would have neglected all those endings and relied on word order, raising children who heard their parents’ slightly stripped-down version. The children would then have used the endings less than earlier generations, until they disappeared entirely.

Once again, the grammar responded as a system. No civilisation can afford to leave the distinction between subjects and objects to guesswork. Word order was relatively flexible in the Anglo-Saxon period. Then the loss of case endings fixed it in more rigid form. The gradual disappearance of case signalling resulted in a potential loss of information, but the solidification of word order made up for it.

We now have a framework in which both the prescriptivists and the descriptivists can have their say. Sound changes can be seen as wrong, understandably, by people who learned an older pronunciation: to my ear, nucular sounds uneducated and expresso is just wrong. But in the long run, sound systems make up for any confusion in a delicate dance of changes that makes sure the language’s necessary distinctions remain. Word meanings change, by both type (a change in meaning) and by force (a change in how powerful a word is). To my six-year-old, everything is epic, which strikes my ear the way awesome must have done to my parents. A lunch just cannot be epic. But when epic is exhausted, his kids will press something else into service – or coin something new.

Even the deepest-seeming change – to the grammar – never destroys the language system. Some distinctions can disappear: classical Arabic has singular, dual and plural number; the modern dialects mostly use just singular and plural, like English. Latin was full of cases; its daughter languages – French, Spanish and so on – lack them, but their speakers get on with life just the same. Sometimes languages get more complex: the Romance languages also pressed freestanding Latin words into service until they wore down and became mere endings on verbs. That turned out OK, too.

Spontaneous order doesn’t sit well with people. We are all tempted to think that complex systems need management, a benign but firm hand. But just as market economies turn out better than command economies, languages are too complex, and used by too many people, to submit to command management. Individual decisions can be bad ones, and merit correction, but we can be optimistic that, in the long run, change is inevitable and it will turn out all right. Broadly trusting the distributed intelligence of your fellow humans to keep things in order can be hard to do, but it’s the only way to go. Language is self-regulating. It’s a genius system – with no genius.