Most word-building runs forward: you take a word and add a piece, so edit plus -or gives editor. Back-formation runs the chain in reverse. Speakers meet the longer word first, hear an ending they can peel off — a real suffix or merely the shape of one — and subtract it to make a new word. The result is a real English word built backwards — and a trap for anyone who assumes the long word must have come first in every case.
What back-formation is
A back-formation is a word created by removing a real or supposed affix from an already existing longer word, as Merriam-Webster’s definition puts it — both cases count. Sometimes the ending really was a suffix in the source word: editor’s Latin -or genuinely marks “one who.” Sometimes it was only shaped like one: the -s of pease was never a plural ending at all. English speakers can subtract either. Editor entered English as one borrowed word in the 1640s, but it looks like “one who edits,” so the verb edit was probably carved out of it by 1791 — the leading account, though not the only one. Burglar looks like burgle plus the agent ending -ar, so burgle appeared in 1869. The subtraction works because the removed piece matches a pattern English really uses: -er, -or and -ar do mark “one who does,” and -s does mark a plural. The new word is wrong about the old word’s structure and still earns its place in the language.
The name itself is documented: the lexicographer James Murray, the Oxford English Dictionary’s primary editor, coined back-formation in 1888, as Wiktionary’s entry records. A century and a half of use has made the process ordinary — surveil (probably from surveillance, by 1903, per Etymonline) works the same way.
Pea and cherry: plurals that were never plural
The cleanest back-formations pretend to fix a grammar problem that did not exist. Middle English pease was the whole word for the vegetable — a single form that could name one pea or the collective crop, the way wheat works today. By the mid-seventeenth century, speakers were hearing the final -s as a plural marker and cutting it off: pea is, in the Online Etymology Dictionary’s phrase, a false singular. The subtraction invented a new singular that had never existed — and it worked so well that Merriam-Webster now enters pease chiefly as the plural of pea, inverting the real history.
Cherry walked the same path earlier. The word came over from Anglo-French cherise — modern French cerise — around 1300, and Middle English speakers again mistook the final -s for a plural and stripped it, exactly the error Etymonline notes in its cherry entry. One fruit-name losing its phantom plural would be a curiosity; two makes it a pattern. When a word ends in a sound that could be an ending, check whether the shorter form was ever the real original.
Edit from editor, burgle from burglar
The other classic shape turns a person-noun into a verb. Editor was borrowed from Latin in the 1640s as the person who puts a work out; the verb edit does not appear until 1791 — Etymonline calls it “perhaps a back-formation from editor,” while allowing that French éditer or Latin editus could have supplied it directly. Keep that hedge: the subtraction is the leading account, not the only possible one.
Burgle is firmer and funnier. Burglar was centuries old when burgle appeared in 1869 — the Etymonline entry calls it a humorous or erroneous back-formation. Speakers saw the -ar, assumed an agent verb, and got there anyway. The same mechanism probably produced surveil from surveillance by 1903. In each case the longer word is the earlier attested one and the shorter appears later — for burgle the subtraction is documented outright, for surveil it remains the leading account — the reverse of the order the spelling suggests.
How dictionaries mark a back-formation
You rarely have to spot a back-formation unaided, because dictionaries usually say so. The Online Etymology Dictionary writes it inline: “perhaps a back-formation from editor,” “probably a back-formation from surveillance.” Wiktionary’s etymology conventions give back-formations their own notation section, and Merriam-Webster marks them in the same bracketed etymology line the dictionary-entry guide decodes.
Two quieter signals help when the label is absent. First, dates: a verb that appears a century and a half after a matching person-noun is a back-formation candidate, because the ordinary direction would put the verb first. Second, hedges: a dictionary that says “perhaps a back-formation” is telling you the records allow more than one route — that caution is part of the answer, not a footnote to skip.
Back-formation versus folk etymology
Both processes are reanalysis — speakers remaking a word around what they think its pieces mean — but they move in different directions. Back-formation subtracts: pease loses an imagined -s and stays itself. Folk etymology substitutes: unfamiliar pieces get swapped for familiar words, as when the Spanish cucaracha became English cockroach — cock and roach were never inside the original. The folk-etymology guide covers that reshaping and the unsupported stories that imitate it; this guide covers the neater, quieter mechanism that cuts instead of rewrites. A third pattern packs pieces of two words into one — smog, motel, brunch — and the portmanteau guide follows that blending mechanism.
Back-formation can also hide inside another pattern. Currant began as “raisins of Corinth” — raysyn of Curans — before the final -s was mistaken for a plural: a false singular inside a place name, a subtraction rather than a folk-etymology rewrite; the place-names guide follows that thread. When you separate the mechanisms, you can see which part of a word’s story each source is actually explaining.
Spotting a suspected back-formation
Four checks turn a hunch into a claim worth keeping:
- Pattern. Does the removed piece match a real English ending — an agent -er/-or/-ar, a plural -s, an action -ion? The subtraction only tempts speakers when the leftover shape is plausible.
- Chronology. Is the longer word attested earlier? Editor (1640s) before edit (1791) is the whole argument; without the dates, it is just a guess.
- The label. Do dated references actually call it a back-formation, or hedge with “perhaps” and “probably”? Follow their confidence, not yours.
- The alternative. Could the shorter word have arrived another way — as edit might have come through French éditer? A second live route weakens the subtraction account without killing it.
And one caution that survives every checklist: not every word ending in -er or -ar conceals a verb. Paper, pillar and sugar have no pape, pille or suge behind them. Resemblance is the prompt, not the proof.
To try it live, run a suspected pair through the word lookup — edit, burgle, surveil are good candidates — and see whether the generated reading proposes the agent noun as the earlier form. Then check that proposal against the dated entries above: a Word Says Hi reading is a research starting point, not a verified history, so the dictionaries decide. For a contrast, our robot page shows the other way new words arrive — coined outright for a 1920 play — which is why robot has nothing to subtract.
Sources and further reading
- Online Etymology Dictionary: pea — the false singular out of Middle English pease.
- Online Etymology Dictionary: cherry — the same -s subtraction from Anglo-French cherise.
- Online Etymology Dictionary: editor — the 1640s noun, and the hedged back-formation note for edit.
- Online Etymology Dictionary: burgle — the 1869 humorous back-formation from burglar.
- Online Etymology Dictionary: surveil — the probable 1903 back-formation from surveillance.
- Merriam-Webster: pea — pease entered today as the plural of pea.
- Merriam-Webster: back-formation — the “real or supposed affix” definition.
- Wiktionary: back-formation — the term’s definition and Murray’s 1888 coinage.
- Wiktionary: Etymology — the notation conventions, including back-formations.
The checking steps and guide structure are original editorial guidance. Source summaries are intentionally brief. Sources consulted 1 October 2026.