Language Policing Turns Ordinary Words Into Offences, and It Never Ends

What would you say is the most offensive word in the English language? Take your time. Now notice that your answer changed during the last forty years, and that it will change again, and that somewhere in an office with good lighting and a generous budget there is a person whose entire job is to find the next one. That is the whole story of language policing in one paragraph. The rest of this post is the footnotes.

I want to talk about a particular genre of document: the list. A list of words and phrases that a university, a company or an agency has decided people should stop using, with a friendly suggestion for what to say instead. These lists arrive with kind intentions, soft fonts and the vocabulary of care. And they share a structural flaw that no amount of kindness can fix, which is that they have no natural stopping point. Let me show you why, with a case that was funny until you thought about it.

The list that flagged "American"

In 2022, a web site called the Elimination of Harmful Language Initiative appeared on Stanford University's information technology pages, co-sponsored by the university's CIO Council and an affinity group called People of Color in Technology. It was a guide intended to help technology staff identify and remove language that its authors considered harmful from the university's websites and code. Its entries were sorted into categories, among them imprecise language.

It drew public attention in December 2022, after a critical newspaper editorial, and it was widely reported. Among the items it flagged was the word "American" when used to mean someone from the United States. The suggested alternative was "U.S. citizen". The reasoning, as the document explained, was that the word erases the other countries of the Americas, and that it implies that the United States is the most important country on the continents. The guide also advised against words that most people use without a second thought, and it did so in a tone that ranged from the earnest to the baffling.

The reaction was swift, and it came from every direction, including from people who are not conservatives and who thought that a university had gone off the rails. The site was hidden from public view on December 19, 2022 and taken down in early January 2023, and Stanford's chief information officer said it did not represent university policy and had been meant for discussion within the IT community. I take the university at its word on that. What interests me is not the retraction. It is how a group of intelligent people, with good motives, arrived at the conclusion that the name of the country in which they work and the people who live in it was a problem that needed solving.

Think about what that means. The name of the country appears in the title of the Declaration of Independence: "The unanimous Declaration of the thirteen united States of America". The word "American" is what we call the people of that republic, and the suggestion that it was a problem to be solved is the reductio ad absurdum, and it happened not in a satire but on a university website.

Why the list never ends

You might think the Stanford case was an overreach and that sensible institutions will avoid it. I would argue the opposite: the Stanford list was not a departure from the logic of language policing. It was the logic, followed all the way. Let me make the argument in numbered form.

1. The person who writes the list is rewarded for finding things. A committee that is set up to eliminate harmful language is judged by whether it eliminates something. A committee that reports, after a year, "we looked, and ordinary speech is fine," has justified neither its budget nor its existence. Every incentive points one way: toward more entries.
2. The test for inclusion is subjective and moves in one direction. The standard is whether someone, somewhere, might be offended or might find a word to carry a harmful history. Almost any word passes that test, because words have histories. A word's etymology, however remote from its current use, can always be cited. And because nobody earns credit for removing an entry, the list ratchets.
3. The cost of error is asymmetric. If the list wrongly includes a harmless word, the cost is a mild inconvenience spread across thousands of people. If the list is accused of missing a word, the cost is a public rebuke to the committee. The sensible response for the committee is to include.
4. There is no external check. The authors are not elected, they do not answer to the speakers, and the speakers are not consulted. A language is a vast collective decision made by hundreds of millions of people over centuries. A list is a memo from a working group.
5. Therefore the list can only grow, and the logic has no stopping point. If "American" can be flagged, then so can almost anything, and the interesting question is not which words will be added but which will survive.

Let me follow the logic one step further than its defenders do, as a labelled bit of hypothetical absurdity. Picture a future committee, working with the same method, reviewing the words "the", "a" and "and". The word "the" assumes there is only one of something, which excludes other candidates. The word "a" implies an arbitrary selection. The word "and" suggests that two things are equal when they may not be. The committee recommends that staff use "this one, among others, which is not meant to imply a ranking." This is a joke. I would be interested to know how many readers believe it could not happen.

Other lists, described only as documented

The Stanford page was not unique. Institutions of many kinds have published guides to inclusive or preferred language, and some of them have attracted similar criticism. I will not itemise them here, because I want to avoid repeating claims I cannot verify to my own satisfaction. My advice to readers is to look at the documents themselves, which are public. Find a list from your own university, local government, school district or employer, read it, and ask three questions. Who wrote it? On what authority? And what would happen to someone who ignored it?

That last question is the one that matters. A guide that says "you may wish to consider" is advice. A guide that appears in a performance review, a hiring decision or a disciplinary process is a rule, and a rule that governs speech in a public institution raises the constitutional questions I address in Speech, Pronouns and Compelled Words, Where the First Amendment Draws the Line. A guide that is a mere suggestion today has a way of becoming a requirement tomorrow, and the people who wrote it will explain that nothing has changed.

The euphemism treadmill

There is a second reason the lists never end, and it comes from the nature of language itself. The psychologist Steven Pinker popularised a phrase for it: the euphemism treadmill. The idea is that when a word for a sensitive subject picks up negative associations, people invent a new word, and within a generation the new word picks up the same associations, because the stigma attaches to the thing and not to the word. So the new word has to be replaced too.

The history of the vocabulary for intellectual disability is the clearest case. Terms that were once clinical and neutral, and were adopted by the medical and educational professions of the early twentieth century in the hope of being respectful, became playground insults. The replacement terms followed the same course. In 2010 Congress passed Rosa's Law, signed by President Obama in October of that year, which replaced the term "mental retardation" in federal statutes with "intellectual disability". That was a decision made with good motives, and nothing in this essay objects to it. But it illustrates the treadmill: each term was introduced to be kind, and each acquired the taint of the thing it described.

What does this tell us? That changing the word does not change the attitude. If the contempt is in the heart of the speaker, he will carry it to the new word. If the respect is in the heart, the old word will not hurt. The reform that matters is the one in the speaker. The list-writer, who believes the vocabulary is the lever, will spin the wheel forever and the hamster will never get off.

George Orwell understood this a long time ago. In his 1946 essay "Politics and the English Language" he argued that careless and evasive language corrupts thought, and in Nineteen Eighty-Four (1949) he imagined a language, Newspeak, whose purpose was to make certain thoughts impossible by removing the words for them. Nobody on a university committee is trying to build Newspeak. But the mechanism by which vocabulary is narrowed in the name of some good end is the same, and Orwell's warning is a useful one for anyone who believes that a smaller vocabulary makes a better society.

Why ordinary people resist

Now let me say something in defence of the people on the receiving end. When a friend, a neighbour or a coworker is told that a word he has used all his life is a sin, he does not respond with gratitude. He responds with a sort of weary skepticism, and he is right to.

Ordinary people have good reasons. They know that most of the time they have used the word without malice. They know that the list was not written by them, or by anyone they know. They know that the rules seem to change faster than they can learn them, and that the penalty for a mistake can be social or professional. They also know that a society in which every sentence must be checked against a list is a society in which people say less, and where the people who benefit are those who are most skilled at navigating the list. It is not the kindest person who thrives in such a world. It is the best lawyer of language.

A free people talks the way it talks. It argues about words constantly, and over time the language changes, as it always has: through use, through humour, through the example of good writers and speakers. A good deal of what we now consider polite was once considered radical, and it was adopted through persuasion. The way to change a word is to make a better case for the new one, not to put the old one on probation.

The constitutional anchor

Here is the founding principle. The First Amendment says that Congress shall make no law abridging the freedom of speech. That limits the government, not private bodies, and a private company is free to adopt an internal style guide. But a public university is a part of the state, and a list backed by discipline from a public institution engages the First Amendment. The Court said in West Virginia State Board of Education v. Barnette (1943) that no official, high or petty, can prescribe what shall be orthodox in politics or other matters of opinion. A list of required and forbidden words, when it is enforced, is an attempt to prescribe an orthodoxy by vocabulary. I read the whole passage in Barnette's Fixed Star, Reading the 1943 Opinion on the Pledge Passage by Passage.

For a related example of the same impulse in the hiring process, see DEI Statements Were Loyalty Oaths, and the Old Oath Cases Show Why. For the social cousin of the word list, the habit of punishing people for words without judge or jury, see Cancel Culture Has No Judge, No Jury and No Statute of Limitations. And for the comic treatment of a world in which every product carries a warning, see A Modest Proposal to Put a Warning Label on Every Novel Before It Hurts Someone.

The strongest objection

The strongest defence of such lists is that language does change, and that words can hurt. Both are true. Words have been used to demean people for as long as there have been words, and nothing in this post defends slurs or cruelty. A university is entitled to expect that its staff will treat one another with respect, and a guide that helps a staff member avoid wounding a colleague by accident is doing something useful.

I would answer in three points. First, the distinction I draw is between advice and enforcement. A voluntary guide, offered with humility, is fine, and people may take it or leave it. Second, the distinction between a slur and an ordinary word matters. A slur is a word whose function is to demean. An ordinary word that someone has decided to find offensive is different, and the Stanford example is a case of the second. Third, the presence of a few legitimately hurtful terms on a list does not justify the thousand harmless ones that come with them. A rule that cannot tell the two apart is a bad rule.

The case in brief

1. Lists of forbidden words grow because the people who write them are rewarded for finding offences, and no one is rewarded for leaving ordinary words alone.
2. Stanford's Elimination of Harmful Language Initiative flagged the word "American" in December 2022 and was withdrawn from public view in January 2023, which shows the logic followed to its end.
3. The euphemism treadmill means that replacing a word does not remove the stigma attached to the thing, so the list can only keep running.
4. Ordinary people resist because they know the words were used without malice, and the constitutional line falls where a public institution turns advice into enforced orthodoxy.

What you can do: find your employer's, school's or university's language guide, read it, and if it is enforced, ask in writing who adopted it and under what authority.