Why Is English Spelling So Irregular? Six Centuries of Sounds That Moved

Old English, the language of Beowulf, was spelled more or less as it was spoken, using an alphabet with a few extra letters for sounds Latin lacked. The Norman conquest of 1066 replaced English with French as the language of government and writing for three centuries, and when English came back into writing in the fourteenth century the scribes trained in French applied French habits to it. They wrote qu for cw, so that cwen became queen; they wrote ou for the long u sound, giving house; and they replaced the old u in many words with o, because u looked too much like the neighbouring letters in the handwriting of the time, which is why love, come and son have an o they are not pronounced with.

The largest cause is a change in sound that ran through the language between about 1400 and 1700. Every long vowel in English moved: the a in name, which had sounded like the a in father, rose to its present sound; the e in see, once like the a in name, rose to ee; the i in time, once ee, became the diphthong it is now; and the same happened to the back vowels, so that the oo in moon, once like the o in stone, became oo, and the ou in house, once oo, became ow. Spelling had been settled by the printers before the shift finished, so English writes its long vowels as Chaucer said them and pronounces them as Shakespeare did. It is the reason no other European language uses the letters a, e and i for the sounds English gives them.

Many silent letters were once pronounced. The k of knight and knee and the g of gnaw were said until the seventeenth century; the gh of night was a throat sound like the ch of Scottish loch, still heard in the German equivalent; the l of walk and half and the w of write were sounded, and the b of climb and comb. Those spellings are fossils of pronunciation. Others were never pronounced at all and were added on purpose, by Renaissance scholars who wanted English words to show their Latin ancestry:

• •The b in debt and doubt, inserted to match Latin debitum and dubitare; the words had come from French without it
• •The c in scissors and the s in island, both wrong: scissors is from Latin cisorium, and island is Old English and never had an s
• •The h in ghost, added by a Flemish typesetter in Caxton's print shop who spelled it as in his own language
• •The p in receipt, the l in salmon and the ch in schedule, all Latin or Greek costume on words that had lost those sounds long before

Printing arrived in England in 1476, in the middle of the vowel shift, and fixed spelling before the sounds had settled. Caxton's compositors came from the Low Countries and brought their own habits; early printers also lengthened or shortened words to justify lines of type, adding a final e or dropping one as the line required, and the variants then circulated. Samuel Johnson's dictionary of 1755 settled most of the remaining choices, generally by picking the most common spelling in print, and after Johnson the spelling was fixed while the pronunciation went on drifting. The American divergence came later, when Noah Webster's dictionaries of the early nineteenth century deliberately dropped the u from colour and the k from musick and turned the ending of centre and theatre round, partly on principle and partly to give the new republic a spelling of its own.

Reform has been proposed since the sixteenth century and has never taken. Benjamin Franklin designed a new alphabet; Shaw left money in his will for one; Theodore Roosevelt ordered the American government to adopt three hundred simplified spellings in 1906 and was mocked into withdrawing the order within months. The obstacles are practical. English is spoken in dozens of accents that do not agree on the sounds, so a phonetic spelling would have to pick one; every existing book would become archaic; and the current spelling, for all its faults, keeps related words looking related, so that sign and signature, nation and national share a spelling they do not share a sound. Reforms succeed where a government controls the language and the population is small: Dutch, German, Portuguese and Norwegian have all revised their spelling in the last century. English, with no academy and a billion users, has kept its fossils.

English spelling is less random than it feels, because many of its rules are about vowels rather than individual letters. Compare hoping and hopping, or later and latter. A single consonant after a vowel usually signals a long vowel, as in the silent e of hope, while a doubled consonant signals a short one, as in hop. Once you notice this, you can predict the spelling of thousands of words, and you can see why writers double the p in shopping but not in sloping. Most everyday English words follow patterns of this kind; it is a noisy minority that gives the system its reputation.

A second pattern explains many of the oddities that remain: English borrowed words from everywhere and often kept the spelling of the source language. The ph in photograph and the ch pronounced k in chorus are Greek conventions passed through Latin, the silent t in ballet and the zh sound in beige are French, the double z in pizza is Italian, and the ch in yacht records a Dutch word. When you meet a strange spelling, asking which language a word came from is often more useful than asking how it sounds, because each source brought its own rules with it.

The claim that fish could be spelled ghoti works by borrowing the gh from enough, the o from women and the ti from nation. It is a good joke and a poor argument, because each borrowed piece only makes its sound in a particular position. The gh spelling for an f sound only ever appears at the end of a word or syllable, as in laugh and tough, and never at the start. No English reader would pronounce a word beginning gh as anything but a hard g, as in ghost.

The other parts fail the same way. The o in women is a one-off, a leftover from an older pronunciation, and appears in no other common word with that sound. The ti in nation only sounds like sh when it comes before a vowel in endings such as tion and tial. Put the letters where ghoti puts them and a reader following the real patterns would say something like goatee. The exercise is useful mainly as a reminder that English spelling depends on position and context, which is exactly why readers can usually cope with it.

Some apparent irregularities are actually useful, because English often spells a piece of grammar the same way however it is pronounced. Say walked, played and wanted aloud and the past-tense ending is a t sound, then a d sound, then a separate syllable, yet all three are written ed. The plural ending works the same way: cats ends in an s sound, dogs in a z sound and horses in an extra syllable, all spelled with s. A purely phonetic spelling would write walkt, playd and dogz, and readers would lose the instant visual signal that tells them a word is past or plural.

A quieter pattern separates content words from grammatical ones. Nouns, verbs and adjectives in English almost never have fewer than three letters, so short content words are padded out, as in inn, bee, egg, odd and add, while grammatical words are allowed to be shorter, as in in, be, of and I. Homophones are often kept apart in the same spirit, so that there, their and they're, or right, write and rite, look different on the page although they sound the same. These conventions cost learners effort when spelling but help practised readers recognise words quickly, and that trade-off runs through the whole system.

English spelling is irregular because it was fixed by scribes and printers between about 1400 and 1750, while the Great Vowel Shift changed every long vowel and consonants such as the k in knight and the gh in night fell silent, and because Renaissance scholars added Latin letters that were never pronounced. Johnson's dictionary froze the result, Webster adjusted it for America, and every reform since has failed on the diversity of accents and the weight of what is already in print.