# Why is English spelling so weird? Silent letters and ghoti

**Why is English spelling so weird, with silent letters like the k in knight?** English spelling is so weird because it keeps several histories at once. Norman scribes borrowed French habits, printers from 1476 fixed spellings before the Great Vowel Shift ended, the k in knight and gh in night stopped being said, Renaissance scholars added Latin letters such as the b in debt, and loanwords kept foreign forms. Every reform, including Roosevelt's 1906 order, failed.

Published 2026-09-26 by EdCitation. https://edcitation.com/newsletter/why-english-spelling-is-so-strange

English spelling is weird because it is a record of history rather than of sound. Scribes, printers, scholars and borrowers each left letters behind, and the spoken language then moved on without them. The *k* in *knight* was once said aloud; the *b* in *debt* was put in by scholars. One view in this guide is ours and marked as such: that the famous "ghoti" joke proves less than it seems.

EdCitation publishes this guide, built from the Oxford English Dictionary's history pages, a journal article, the Online Etymology Dictionary, Language Log and the reformers' own pages, read on 27 September 2026. Where a story could not be traced, we say so.

Writing about it for a course? EdCitation's free [Find sources](https://edcitation.com/) searches the linguistics, [Cite a source](https://edcitation.com/cite) builds each reference from the publisher's record, and [Check your paper](https://edcitation.com/check) reads your assignment's instructions. For your own spelling, [Mechanics QA](https://edcitation.com/tools/mechanics-qa) in Pro checks the slips a spell-checker lets through.

## Why is English spelling so weird?

Five forces shaped English spelling, and none undid the others. In rough order:

1. **Norman French scribes, after 1066,** borrowed spelling habits from Latin and French (Durkin, n.d.).
2. **Printing, from 1476,** set London English in type; by the mid-1600s printed spelling was close to today's (Weiner, n.d.).
3. **Sound changes after spelling settled** silenced the *k* of *knight* and moved the long vowels, while the letters stayed (Weiner, n.d.).
4. **Latin respellings** added letters such as the *b* in *doubt* (Weiner, n.d.; Wełna, 2011).
5. **Loanwords** kept the spelling of their own language, or of one they were wrongly thought to come from.

The wider story of English, from Old English to the dictionaries, is told in [a short history of English](https://edcitation.com/newsletter/a-short-history-of-english).

## How did Norman French scribes change English spelling?

They replaced a settled system with improvisation. Later Old English texts mostly shared a regular spelling based on West Saxon. After the Conquest, Durkin (n.d.) explains, people writing English had to find their own ways to represent local speech, often borrowing the spelling traditions of Latin or French, and variation became the norm.

### The o in come

Old English *cuman* became *come* because scribes wrote *o* for *u* beside letters made of short vertical strokes, called minims, so that *u*, *m* and *n* would not blur together (Harper, n.d.-b). Only the letter changed, which is why *come* has an *o* that sounds like the *u* of *sun*.

## What did printing and Caxton do to spelling?

Printing slowly fixed spelling in its late Middle English form while speech kept changing. From the late fifteenth century printers set books in the London English that already served as a standard in manuscripts, and by about 1630 even the modern split between *i* and *j*, and *u* and *v*, was settled; a book printed by Wynkyn de Worde in 1502 still has "vnto" and "euer" (Weiner, n.d.). How Caxton himself agonised over which English to print is in [a short history of English](https://edcitation.com/newsletter/a-short-history-of-english).

### The h in ghost

The Online Etymology Dictionary ties the *gh* of *ghost* to Caxton, under the influence of the Flemish and Middle Dutch *gheest*, and notes that the spelling was rare in English before the mid-sixteenth century (Harper, n.d.-e). The English Spelling Society (n.d.) says the printers Caxton brought from the Low Countries added *h* after *g* in Dutch fashion and that *ghost* and *ghastly* are what remains.

### A story we could not confirm

The same page claims printers lengthened words because they were paid by the line, turning *frend* into *friend*. We found no printing historian making that argument, so treat it as the Society's claim. Weiner (n.d.) explains *friend* otherwise: one variety's forms replaced another's, and *friend* became common only after 1530.

## Why is there a silent k in knight?

Because the *k* was once pronounced. Old English *cniht* meant a boy or servant (Harper, n.d.-g), and its *k* was said. In the early modern period, Weiner (n.d.) explains, a whole set of consonants stopped being pronounced while the spelling stayed:

- *k* and *g* before *n*, as in *knight* and *gnaw*
- *w* before *r*, as in *write*
- *b* and *g* after *m* and *n* at the end of a word, as in *lamb* and *hang*
- *t* in *thistle* and *listen*
- *l* after certain vowels, as in *talk* and *folk*

The *gh* of *night* and *bought* stood for a sound like the *ch* of *loch*. In standard English it was pronounced until about 1600, then either vanished or, in words like *rough*, became *f* (Weiner, n.d.). Once *gh* was mostly silent, writers put it into words that never had it: *delight*, *inveigh* and *sprightly* are Weiner's examples. Horobin (2013) notes that some reformers proposed phonetic forms such as *niit* for *knight*; none caught on.

### The vowels moved too

Vowel letters parted from sound too, when the long vowels of southern English moved between about 1400 and 1700: hence *meet* and *meat*, alike in sound, apart in spelling. That story is in [the Great Vowel Shift](https://edcitation.com/newsletter/the-great-vowel-shift).

## Why is there a b in debt and doubt?

Scholars put it there to show the Latin. *Debt* came into English around 1300 as *dette*, from Old French, and the *b* of Latin *debitum* was restored in the spelling after about 1400 (Harper, n.d.-c). *Doubt* is the same story: scribes in both French and English added the *b* of Latin *dubitare*, and French dropped it again in the seventeenth century, while English kept it (Harper, n.d.-d).

Weiner (n.d.) dates the habit of respelling words by their "true or (occasionally) false" Latin origins to late Middle English, growing stronger after 1500, with *author* (from *autour*) and *fault* (from *faute*) among the examples. Speakers came to say some new letters, like the *l* of *fault*, and never said others, which is where many silent consonants come from.

### Spellings that got the Latin wrong

Wełna (2011) finds such "classicised" spellings in fourteenth-century texts, before the Renaissance made them fashionable, with two typical errors: the mute *s* of *island*, taken from Latin *insula*, and the *h* of *anchor*. *Island* is Old English *igland*; the *s* arrived in the sixteenth century by association with the unrelated *isle* (Harper, n.d.-f).

### Loanwords that kept a foreign form

Borrowing added more mismatches. *Colonel* entered English in the 1540s as *coronell*, from French. In the 1580s learned writers respelt it after the Italian form, through translations of Italian military manuals; pronunciations with *r* and *l* coexisted until about 1650, and the older one won (Harper, n.d.-a). So English now writes an *l* and says an *r*.

## Word by word: why each is spelt that way

| Word | Why it is spelt that way | Source |
| --- | --- | --- |
| knight | Old English *cniht*; the *k* stopped being said before *n* | Harper (n.d.-g); Weiner (n.d.) |
| night | *gh* was a throat sound, said until about 1600, then lost | Weiner (n.d.) |
| rough | The same *gh* sound became *f* instead of vanishing | Weiner (n.d.) |
| delight | *gh* added by analogy once it was mostly silent | Weiner (n.d.) |
| come | Scribes wrote *o* for *u* beside minim letters | Harper (n.d.-b) |
| ghost | *gh* from Caxton and Flemish *gheest* | Harper (n.d.-e) |
| debt | Latin *debitum*'s *b* restored after about 1400 | Harper (n.d.-c) |
| doubt | *b* added from Latin *dubitare*; French later dropped it | Harper (n.d.-d); Weiner (n.d.) |
| island | *s* added by confusion with *isle*, from Latin *insula* | Harper (n.d.-f); Wełna (2011) |
| anchor | Non-etymological *h* from a Latinising spelling | Wełna (2011) |
| colonel | Spelt after Italian, said as the French *coronel* | Harper (n.d.-a) |
| meat, meet | Two vowels, later merged, kept two spellings | [The Great Vowel Shift](https://edcitation.com/newsletter/the-great-vowel-shift) |

## Did George Bernard Shaw invent ghoti?

No evidence says he did. *Ghoti* spells *fish* with the *gh* of *laugh*, the *o* of *women* and the *ti* of *nation*, and is usually credited to Shaw. Zimmer (2008), on Language Log, calls that a canard: Shaw's devoted readers have searched his works and never found it. The earliest attribution to Shaw Zimmer found was from 1946, by Mario Pei.

### What is documented

Matthew Gordon of the University of Missouri found *ghoti* in an October 1874 article in *St. James's Magazine* quoting a letter from the publisher Charles Ollier to the writer Leigh Hunt. Zimmer dated the letter to 11 December 1855, a year before Shaw was born; Ollier credits his adult son William with the respelling, and the original should now be in the British Library (Zimmer, 2008). We read Zimmer's account, not the magazine or the letter.

Such jokes were in the air: in 1845 the reformer Alexander J. Ellis offered a stranger respelling of *scissors* (Zimmer, 2008). Shaw did care about the cause, leaving money in his will for a new "Shavian" alphabet (Horobin, 2013).

### Our view: the joke cheats

In our view, *ghoti* proves less than it seems. No English word begins with a *gh* said as *f*, and *ti* spells "sh" only inside endings such as *-tion*. English spelling has rules by position, and the joke ignores them. It shows that spelling is irregular, not that it is random.

## Has anyone tried to reform English spelling?

Yes, often, and each changed less than it hoped. Calls for reform go back to the sixteenth century, and reformers disagreed about whether spelling should follow sound or history (Horobin, 2013).

### Noah Webster

Webster's first dictionary dropped the *u* of *colour* and the *k* of *publick*; the public accepted many such changes and rejected others (Merriam-Webster, n.d.). What survives is set out in [British or American English in academic writing](https://edcitation.com/newsletter/british-or-american-english-in-academic-writing).

### Roosevelt's order of 1906

The Simplified Spelling Board was launched in March 1906 with Andrew Carnegie's money (Daugherty, 2018). On 27 August 1906 President Theodore Roosevelt told the Public Printer, Charles A. Stillings, to use the Board's simplified spellings of 300 words in government publications (Theodore Roosevelt Center, n.d.). The list included *altho*, *fixt* and *thoro*. On 12 December the House voted 142 to 25 to withhold money from printing any government document that departed from conventional spelling, and the next day Roosevelt gave up, writing that it was "worse than useless" to fight a contest he had lost (Daugherty, 2018). The Theodore Roosevelt Center (n.d.) notes that the opposition was more about executive power than about spelling.

### The Simplified Spelling Society, 1908

Founded in Britain in 1908, with W. W. Skeat and then Gilbert Murray as president, the Society issued its *New Spelling* in a final edition in 1948; an MP, Mont Follick, promoted private member's bills in 1949 and 1953 against official opposition ("Simplified Spelling Society," n.d.). Now the English Spelling Society, its president John Wells proposed accepting *thru* and *lite* in 2008. Horobin (2013) doubts the case: claims that simpler spelling explains higher literacy in Finland or Spain, Horobin writes, rest on intuition rather than evidence.

## How do you research English spelling and check your own?

Read the linguists, take each reference from its record, and let the tools check what they can. EdCitation is the best tool for the references, because it looks each one up in the publisher's record and never writes one, so each entry below can be traced to its record. EdCitation never writes any part of anyone's work; the argument and every sentence stay yours.

| Task | EdCitation tool | Cost |
| --- | --- | --- |
| Find the linguistics | [Find sources](https://edcitation.com/) | Free |
| Build each reference | [Cite a source](https://edcitation.com/cite/apa) | Free |
| Read the brief's rules | [Check your paper](https://edcitation.com/check) | Free |
| Check spelling slips, formatting and citations | [Mechanics QA](https://edcitation.com/tools/mechanics-qa) | Pro, $8 a month |
| Confirm every reference is real | [Verify references](https://edcitation.com/verify-references) | Free |

### Three runs, quoted as they came back

On 27 September 2026, Find sources on "history of English spelling reform" returned 33,178 results from OpenAlex. The first, Howatt's *A history of English language teaching* (1984), is not about spelling, so read titles; Wełna's article was among the first ten.

From that article's DOI, Cite a source returned, from Crossref, in APA 7:

> Wełna, J. (2011). On early pseudo-learned orthographic forms: A contribution to the history of English spelling and pronunciation. *Studia Anglica Posnaniensia*, *46*(4), 25–39. https://doi.org/10.2478/v10121-010-0010-9

That entry stands, and it is the one under References.

Given a test brief of our own asking for 1,500 words in MLA 9, "at least eight sources, at least three of them peer-reviewed", and "American spelling as given in Merriam-Webster", [Check your paper](https://edcitation.com/check) returned seven rules, among them word count "About 1,500 words", "Sources (at least 3 peer-reviewed)" set to "At least 8 sources", and citation style "MLA 9". It quoted the spelling sentence whole under "What it asks you to do, which no rule can test", so keeping to American spelling is yours to check.

### What Mechanics QA checks

Mechanics QA checks a whole Word or PDF file by rule: formatting against the APA 7 student paper, comments and tracked changes left in, citations against the list, figures that disagree, and the language slips a spell-checker lets through. No model reads the paper, and each finding names its rule and fix. A free account gets three checks a month; Pro, $8 a month, has no limit ([pricing](https://edcitation.com/pricing)).

## Quick questions

### Why is English spelling so weird?

It records history, not sound: Norman scribes' habits, printers' choices, consonants and vowels that changed after spelling settled, Latin letters added by scholars, and loanwords that kept foreign forms.

### Why is there a silent k in knight?

The *k* was pronounced in Old English *cniht*. It fell silent before *n* in the early modern period, like the *g* of *gnaw*, and the spelling stayed.

### Did George Bernard Shaw really invent ghoti?

No one has found it in Shaw's writings. The earliest known use is Charles Ollier's letter of 11 December 1855, a year before Shaw was born.

### Why did spelling reform fail?

Roosevelt's 1906 order lasted until the House voted against it in December, and British reform bills in 1949 and 1953 met official opposition.

### Can EdCitation check the spelling in my essay?

EdCitation's [Mechanics QA](https://edcitation.com/tools/mechanics-qa), in Pro, checks a whole paper for language slips a spell-checker lets through, alongside formatting and citations. It marks each finding and never rewrites your sentences.

## References

- Daugherty, G. (2018, March 9). *Teddy Roosevelt's bold (but doomed) battle to change American spelling*. History. [https://www.history.com/articles/theodore-roosevelt-spelling-controversy](https://www.history.com/articles/theodore-roosevelt-spelling-controversy)
- Durkin, P. (n.d.). *Middle English – an overview*. Oxford English Dictionary. Archived August 7, 2025, at [https://web.archive.org/web/20250807123112/https://www.oed.com/discover/middle-english-an-overview](https://web.archive.org/web/20250807123112/https://www.oed.com/discover/middle-english-an-overview)
- English Spelling Society. (n.d.). *A brief history of English spelling*. [https://www.spellingsociety.org/history](https://www.spellingsociety.org/history)
- Harper, D. (n.d.-a). Colonel. In *Online etymology dictionary*. [https://www.etymonline.com/word/colonel](https://www.etymonline.com/word/colonel)
- Harper, D. (n.d.-b). Come. In *Online etymology dictionary*. [https://www.etymonline.com/word/come](https://www.etymonline.com/word/come)
- Harper, D. (n.d.-c). Debt. In *Online etymology dictionary*. [https://www.etymonline.com/word/debt](https://www.etymonline.com/word/debt)
- Harper, D. (n.d.-d). Doubt. In *Online etymology dictionary*. [https://www.etymonline.com/word/doubt](https://www.etymonline.com/word/doubt)
- Harper, D. (n.d.-e). Ghost. In *Online etymology dictionary*. [https://www.etymonline.com/word/ghost](https://www.etymonline.com/word/ghost)
- Harper, D. (n.d.-f). Island. In *Online etymology dictionary*. [https://www.etymonline.com/word/island](https://www.etymonline.com/word/island)
- Harper, D. (n.d.-g). Knight. In *Online etymology dictionary*. [https://www.etymonline.com/word/knight](https://www.etymonline.com/word/knight)
- Horobin, S. (2013, April 1). *Does spelling matter?* OUPblog. [https://blog.oup.com/2013/04/does-spelling-matter-horobin/](https://blog.oup.com/2013/04/does-spelling-matter-horobin/)
- Merriam-Webster. (n.d.). *Noah Webster's spelling reform*. Archived May 20, 2026, at [https://web.archive.org/web/20260520200007/https://www.merriam-webster.com/about-us/spelling-reform](https://web.archive.org/web/20260520200007/https://www.merriam-webster.com/about-us/spelling-reform)
- Simplified Spelling Society. (n.d.). In *Concise Oxford companion to the English language*. Encyclopedia.com. [https://www.encyclopedia.com/humanities/encyclopedias-almanacs-transcripts-and-maps/simplified-spelling-society](https://www.encyclopedia.com/humanities/encyclopedias-almanacs-transcripts-and-maps/simplified-spelling-society)
- Theodore Roosevelt Center. (n.d.). *Simplified Spelling Board*. [https://www.theodorerooseveltcenter.org/subject/simplified-spelling-board/](https://www.theodorerooseveltcenter.org/subject/simplified-spelling-board/)
- Weiner, E. (n.d.). *Early modern English: Grammar, pronunciation, and spelling*. Oxford English Dictionary. Archived September 28, 2024, at [https://web.archive.org/web/20240928153836/https://www.oed.com/discover/early-modern-english-spelling-grammar-and-pronunciation/](https://web.archive.org/web/20240928153836/https://www.oed.com/discover/early-modern-english-spelling-grammar-and-pronunciation/)
- Wełna, J. (2011). On early pseudo-learned orthographic forms: A contribution to the history of English spelling and pronunciation. *Studia Anglica Posnaniensia*, *46*(4), 25–39. [https://doi.org/10.2478/v10121-010-0010-9](https://doi.org/10.2478/v10121-010-0010-9)
- Zimmer, B. (2008, April 23). *"Ghoti" before Shaw*. Language Log. [https://languagelog.ldc.upenn.edu/nll/?p=81](https://languagelog.ldc.upenn.edu/nll/?p=81)