weasel words

How to spot it

Weasel words are phrases that leave the impression something specific has been said when nothing a reader could check has been said at all. Some of them shrink a claim until it cannot be wrong, like "may help" or "up to half." The kind this lesson teaches props a claim up instead, by crediting it to an authority nobody names.

The test is one question, and you ask it of the sentence rather than of yourself. Who, exactly, and where would a reader go to read them? A sentence that can answer names somebody. A sentence that cannot has handed its authority to a plural noun and walked off. "Experts say" is the form everyone recognizes, but the family is much larger than that one phrase and every member works the same way, so watch for critics who have noted, analysts who suggest, observers who point out, scholars who contend, and industry reports that indicate. Each borrows the grammar of sourcing. None of them leaves a name behind.

What the question catches is narrower than it first looks, and keeping it narrow is most of the skill. It does not catch a writer summarizing a real disagreement whose participants turn up a line later. It catches a claim that needs a witness and calls for one it never produces. So read what the sentence is actually asking you to accept. If that is a verdict, a number, or a judgment about who is right, the plural noun is standing exactly where a name belongs. If it is a shared premise nobody in the room would contest, the vagueness is costing the reader nothing, and you can leave it alone.

One more reading helps, and it works on the page rather than on the sentence. A draft that opens section after section on a different unnamed crowd is not reporting back from several different rooms. It has one move for "this ought to sound sourced," and it reaches for that move every time it needs to sound sure of something.

Why models overdo it

Weasel words of this kind come from habit more than deceit: a model finishes a sentence the way sentences like it usually finish, and an opening that sounds sourced is one of the cheapest continuations in English. The grammar of attribution costs nothing to produce and earns a great deal on a fast read, because a reader who sees the shape of a citation tends to grant it the weight of one long before checking whether anything is attached.

The difference is what each writer has at stake. Naming a person puts a checkable thing on the page and accepts that somebody may go and check it. The plural noun is that same sentence with the risk taken back out, and taking risk out is what a likelihood-driven process does all day. So the sentence arrives finished, sounds researched, and commits to nothing. It makes a very good opener and a very poor conclusion.

For the same reason, these accumulate instead of appearing once. A single borrowed authority can be an honest summary of a field. A page built on a dozen of them is repeating one gesture, and a reader who notices the gesture stops granting it anything. That is the cost that matters, and it is easy to miss, because the sentences do not read as false. They read as unsourced, which is worse, since the writer has now spent credibility that was never his to spend.

When it's fine

Vagueness passes when the naming is coming, and good writers often put it in the very next sentence. Maja Baretić and her co-authors open a passage in Technology beyond Biology on an unnamed camp, the critics for whom the World Health Organization's definition of health does not reflect what digital technology has done to well-being, and then, immediately, they name them: Sherry Turkle with a year attached, Luciano Floridi with another. The plural noun is doing topic-sentence work there. It says a camp exists before it says who is in it, and the debt is settled on the next line.

It is also fine when the emptiness is the joke. Jonathan Swift's narrator credits "a grave author, an eminent French physician" for a claim about diet and birth rates that the reader is meant to doubt, and the missing name is the signal to doubt it. Mark Twain tallies the rules of romantic fiction and adds that "some say twenty-two," a scholarly quarrel he plainly has no intention of documenting. Both writers know the crowd behind them is empty. Both are counting on you to hear that it is.

What all three have in common is that the writer knows the sentence is vague and can say why. If you can say why too, leave yours alone. Otherwise, pick one of two repairs: give the claim a name a reader can follow, or take the claim back and make it your own, with whatever evidence you actually hold. A claim in your own name can be wrong in public. That was always the point of naming a source.

Where it comes from

The device is far older than its name, and the name arrived in a short story rather than in a handbook. In 1900 Stewart Chaplin published "Stained Glass Political Platform" in The Century Magazine and called this kind of phrasing a weasel word, because such words "suck the life out of the words next to them, just as a weasel sucks the egg and leaves the shell." Theodore Roosevelt liked the image enough to carry it into a speech at St. Louis on May 31, 1916, warning that a weasel word set after another leaves nothing of the word it followed. Behind both sits a much older complaint out of logic, the one John Locke labeled argumentum ad verecundiam, the appeal to modesty: an authority argues something only when it is specific enough for a reader to weigh the credential against the claim. Take the name out and the appeal has nothing left to appeal to. Editors have been rewriting that rule for the modern page ever since. Wikipedia's own style guide sets "experts declare," "many scholars state" and "it is widely thought" side by side and says plainly that they create the impression of something specific having been said when only a vague claim was made.