Features
Part of Misquotes and fact checks: the questions that matter
Correcting misquotes after a fact check: four repairs and what the note must say
How to correct a misattributed quotation: four ways to repair it, what an erratum must state, updating graphics and social posts, and publishing unresolved checks.
The check is finished, the attribution was wrong, and the line has been on the page for two years. What happens next is the part nobody plans for, and it is where most of the damage is either contained or multiplied.
Correcting a quotation is not the same as correcting a number. A wrong figure can be swapped out and the sentence still works. A wrong attribution has usually been load-bearing: the line was there because of whose line it was, and removing the name often removes the reason for the paragraph.
What to take away
- Say what the page used to say. A silent swap is worse than the original error.
- The correction has to reach the graphic, not only the text.
- Correct the claim you actually made, which is usually about a person rather than about words.
- An unresolved check is still worth publishing, with its limits stated.
Four repairs, and how to choose
Remove the name and keep the line. Works when the sentence was doing the work and the attribution was decoration. The line becomes an unattributed observation, which is what the evidence supports.
Choosing a repair
Was the attribution decoration?
Remove the name, keep the line
Was the evidence incomplete?
Keep the name and change the label. Works when there is real but incomplete evidence: sources credit the person, no primary source has been located, and saying so is accurate. This is the honest use of an attributed label, not a way to keep a line you failed to source.
Replace the line entirely. Usually the right answer when the attribution was the whole point. If the paragraph needed a named authority, and that authority never said it, the paragraph needs rebuilding rather than patching.
Cut the paragraph. Undervalued. A passage that existed to frame a borrowed line is often not missed at all, and the piece gets shorter and better.
What the correction has to say
An erratum that does not let a reader reconstruct what changed is not much of a correction. Four elements cover it.
- The previous wording, quoted exactly and marked as the old version.
- The new wording, or a plain statement that the line has been removed.
- The reasonwhat the check found, and what the evidence does not support.
- The claim the line was carrying, and whether it survives the correction.
What the correction must say
- The wording as published, quoted exactly
- What was wrongattribution, wording, or framing
- The evidenceearliest appearance, or none found
- What the page says now
Established newspaper correction practice has converged on roughly this shape over a long time, and there is no reason to reinvent it. The one addition worth making for quotations specifically is the fourth element, because these corrections often change the argument as well as the fact.
A worked example. The line "I disapprove of what you say, but I will defend to the death your right to say it" is widely printed under Voltaire's name.
Old wording: attributed to Voltaire. New wording: credited to Evelyn Beatrice Hall, who wrote it in her 1906 book The Friends of Voltaire to summarize his position. Reason: no source in Voltaire's own writing has been found, and the earliest appearance located is Hall's book.
The part that gets skipped
Text is the easy half.
- The image with the line set over a photograph, which was shared far more widely than the article.
- The slide, which is now in other people's decks.
- The social post, which may not be editable, and which is where most people saw it.
- The printed program, the wall, the mug, the newsletter that went out by mail.
Anything you cannot edit still needs a note somewhere it will be found. Anything you can edit needs editing on the same day as the page.
Keep the finding written down as well. The same line comes back, from a different writer, six months later. Record the exact wording, the earliest dated appearance, and the searches that found nothing older.
Our reference on the mechanisms that damage quotations is the vocabulary to record it in, and the labels that describe evidence rather than verdicts are what the record should use.
Correcting the claim you actually made
This is the step people miss. A quotation in a piece is rarely just a quotation. It was doing something: lending authority, establishing that a respected person shared your view, standing in for an argument you did not make.
When the attribution fails, that work fails with it. The correction should say so. If a paragraph argued that a position is sound because a famous figure held it, and the famous figure never said it, correcting the wording while leaving the argument standing is a half-measure. The honest note says the supporting claim has been withdrawn.
| What the line was doing | What the correction owes |
|---|---|
| Decoration | A note that the attribution was removed |
| Evidence for a claim | A withdrawal of the claim, not only the line |
| Characterizing a person | A clear statement that the person did not say it |
| Framing a whole piece | An editor's note at the top, not a footnote |
The third row matters most where the person is alive. Our material on quotations attributed to current leaders covers why an unsourced line about a living person is a different category of problem from a stray misattribution to a historical figure.
When the check did not resolve
Most checks end without a verdict, and there is a way to publish that honestly. State what you searched, state the earliest appearance you located and its date, and state that nothing older was found in those corpora. Then say what you have done: kept the line without a name, or removed it.
Readers handle this well. What they do not handle is a hedge that sounds like a finding. "Often attributed to" alone tells a reader nothing about whether anyone looked.
Most of these lines have a history rather than an author. Our account of what a saying with no single author looks like covers that pattern, and the earliest dated appearance you found is the most you can give a reader.
Common questions
Should the original wording stay visible?
Yes. A reader who saw the old version needs to recognize it, and removing it entirely makes the correction impossible to match against memory. Mark it clearly as the previous wording.
Is a quiet fix ever acceptable?
For a typo, yes. For an attribution, no. The attribution was a factual claim about a person, and withdrawing a factual claim silently is the thing that costs you trust when someone notices.
How long should a correction stay up?
Permanently, on the page. It costs nothing and it is the only evidence that your process works. Removing it later leaves the page looking as though it was never wrong.
What if the wrong version is now more famous than the corrected one?
Expect that. The circulating version usually won because it is shorter and better balanced, and a correction will not beat it. Publish the accurate version anyway and stop treating adoption as the measure of success.
