When Language Stops Naming Things
As reading weakens and AI writing spreads, public language grows less precise about harm and responsibility.
In a January 2026 announcement, Amazon said it was making “additional organizational changes” to “reduce layers, increase ownership, and remove bureaucracy,” changes that would affect about 16,000 jobs. That is what a layoff sounds like once it has passed through corporate language.
The people losing their jobs are still there, just pushed farther away. The executives who made the decision are harder to see. The process moves into the foreground while human impact fades from view.
There is a shift in public language that works like this. You hear it in school emails, HR memos, app notifications, even the way a landlord explains a rent increase. Not by saying less. By saying more in ways that blur what happened. We are surrounded by words: statements, captions, summaries, talking points, auto-generated replies, corporate reassurances, political euphemisms. Public life is loud with language, but loud is not the same as clear.
You can hear the strain in the way different anxieties keep collapsing into one another. Reading scores fall. AI tools promise to write on our behalf. Public language grows softer, vaguer, and less willing to say plainly who did what. These are not the same problem. Still, people keep reacting to them as if they belong to one larger shift.
“The grammar changes first. Responsibility recedes with it.”
Part of what they are reacting to is easy enough to recognize once you hear it. More and more language now sounds as if it has already passed through legal review, brand review, damage control, and someone else’s estimate of what people can handle. By the time it reaches you, the rough edge is gone. So is the person responsible. What remains is a sentence that sounds calm, finished, and untouched by the thing it is describing.
This is one reason reading matters more than people like to admit. Reading is not ornamental. It trains people to hear when language gets slippery and when a sentence is built to soften blame or blur what happened.
And there is evidence that many people are getting less practice with that kind of reading. The Nation’s Report Card found that the average reading score for 12th graders in 2024 was three points lower than in 2019. Thirty-two percent scored below NAEP Basic, the highest share recorded in 12th grade reading.
Those numbers do not explain everything. They do suggest that many people are entering adulthood with weaker reading habits, weaker comprehension, or less confidence in sustained language. That matters in a culture where so much depends on being able to tell the difference between an explanation and a gloss.
Civilians are not killed. Instead, there is collateral damage. A company does not raise prices. It adjusts for market conditions. A bad outcome becomes a challenge. A decision becomes a transition. The grammar changes first. Responsibility recedes with it.
Not every shift in language is a crisis. People shorten things. They improvise. They borrow shorthand. They find new ways to say what they mean. Serious scholarship makes a more modest claim than internet panic usually does. They found language does not trap people inside a fixed mental box, but it does shape how people notice, sort, and describe the world.
That is enough to make the problem worth taking seriously. The danger is not that fewer words make thought impossible. It is that thinner language leaves people with fewer good tools for describing what is happening.
This is where artificial intelligence enters the picture.
The strongest case for AI writing tools is easy to understand. They save time. They lower barriers. They help people begin. Some of that is real. But they also make it easier to skip a part of thinking that matters.
A recent MIT Media Lab study on essay writing found that participants using an LLM showed weaker brain connectivity than those writing without one. They also reported less ownership over what they wrote and had more trouble accurately quoting their own essays. MIT’s paper is a preprint and has not yet been peer-reviewed, so it should be treated cautiously. Still, the concern it raises is plain enough: What happens when fluent language shows up before thought has had time to deepen?
Writing is not just a way of recording thought. Very often it is how thought takes shape. A person reaching for the right word is doing more than swapping synonyms. They are testing distinctions. They are trying to find out whether what they mean is anger or humiliation, confusion or evasion, inconvenience or harm.
A culture that loses patience for that work does not become speechless. It becomes easier to satisfy with language that almost fits. And once that happens, powerful institutions gain room to move. The approved phrase arrives first. The softened phrase sounds reasonable. The person trying to name the harm sounds shrill, imprecise, or late.
You can see the same truth clearly in a classroom. Language is not just a vehicle for delivering information. It shapes whether people understand what is being asked of them, whether they feel confident enough to answer, and whether new knowledge connects to anything they already know. UNESCO’s 2025 guidance on multilingual education found learners do better when they are taught in languages they understand. UNESCO says 40% of learners globally are not taught in their mother tongue, and in some low- and middle-income countries the figure rises to 90%.
In Mozambique, UNESCO found that children who enter school and hear the teacher and their classmates using a familiar language are less afraid to speak. And across Africa, children taught in a familiar language were 30% more likely to read with understanding by the end of primary school. That is the larger point. Language does not only describe reality after the fact. It shapes whether people can meet it clearly in the first place.
Seen together, these pressures begin to describe the same kind of loss. A society reads less deeply. Its institutions speak more euphemistically. Its technologies offer to write on people’s behalf. And reality starts to feel harder to name.
That sensation gets described in different ways. Sometimes it is called an information problem. Sometimes propaganda. Sometimes a literacy problem. Sometimes AI disruption. Those are not the same diagnosis. But they produce the same vulnerability. They make it easier for institutions, companies, and other people in power to get away with weaker explanations.
That, more than anything, seems to be what people are reacting to. Not silence. Not the end of thought. Something more ordinary and, in its own way, more dangerous.
The words keep coming, but mean a little less.
The sentence gets smoother as the truth gets harder to hold.
People still know something is wrong, but have fewer durable ways to say what, exactly, is being done to them.
And if you cannot name a thing with some precision, it becomes much harder to refuse it. ⁂
