LONDON: We quite often talk, colloquially, about the written language having “evolved”. As the linguist David Crystal's new book demonstrates with some panache, almost exactly the opposite took place. In evolution, a series of random variations gives rise to something that appears orderly and designed. In the history of English spelling, though, we can trace a whole series of purposive, thought-through and often ingenious practical decisions — made over the years by scribes, compositors and lexicographers — whose net result is a complete flaming boggins.

Crystal sets out to explain how these decisions were made and why, and why the results after hundreds of years are so messy and confusing. He does so with great brevity and clarity, starting with the first Christian missionaries jerry-rigging the Roman alphabet to make sense of Anglo-Saxon vocables (that buzzing noise made with the tongue and the top teeth: how are we going to spell “that”?) right up to the likelihood that the internet's unending babble of “rhubarb rhubarb rhubarb” will, within a decade, more likely be “rubarb rubarb rubarb”.

He shows a brisk impatience with the tradition that likes to pretend that English spelling is senseless. The famous suggestion that you could spell “fish” “ghoti” (gh as in “rough”, o as in “women” and ti as in “motion”) is a witticism often ascribed to George Bernard Shaw but, Crystal says witheringly, has been doing the rounds since the middle of the 19th century. It is, he argues “complete naughtiness. The spelling ti is NEVER used with this sound at the end of a word in English, and the spelling gh is NEVER used with this sound at the beginning of a word.”

It doesn't do, then, to simply throw your hands up and say: “Isn't our language mad?” The real story is much more interesting than that. That “aberrant” has one B and “abbreviate” has two, for instance, isn't capricious or illogical at all — if you consider the sources of those words. “Ab + breviate” gives us the double B; whereas “ab + errant” doesn't ask for it.

The use of the silent E, for instance, makes very good sense: scribes were trying to find a way of indicating a long preceding vowel sound: that gives us the difference between “run” and “rune”. Approaching from the other direction was a monk named Orrm at the turn of the 13th-century. He suggested doubling consonants up to indicated a preceding short vowel: “hopping” as opposed to “hoping”. When scribes were being paid by the inch, maximalism could be profitable. Why write “run” when you could write “runne”? Kerching!

Spelling has been influenced by a whole range of complicated but perfectly decipherable factors. It has been influenced by the source languages of loanwords or by the mother-tongues of the scribes writing things down (French scribes after the conquest didn't like Anglo-Saxon words that ended in -s, for example, so they substituted -ce, giving us “mice”, “lice” and “ice” among others). It has been influenced by analogy (sometimes sensible, sometimes mistaken) with other words already in the language — “could”, for instance, acquired its “l” because of its association with “would” and “should”. It has been influenced by ease of writing (which explains why Z, despite being an incredibly common sound, is a rare letter), by ease of reading, or by the need to mark words out from already-established homophones.

“Stake”, for a pointy bit of wood, was already in the lexicon when the slab of meat came along in the 15th century. So after experiments with “steike”, “steyke”, “styke”, “steke” and “steake” it took its present form, and is now the only “-eak” word to rhyme with “break”. The thing on the end of your foot used to be called either “toe” or “too” in Middle English — but since one heard the word “too” too often (it having already amicably separated from its homophone “to”) it made sense to make that too “toe”.

What of the H in “ghost”? The word in Anglo-Saxon didn't have it: the Holy Ghost was a “Hali Gast”, and Chaucer's ghosts remained, likewise, H-free. But when William Caxton set up shop in London he needed compositors, and, of course, there weren't any English ones available. So he imported them — among their number the splendidly named Wynkyn de Worde — from the continent. They spoke Flemish, so “if a word reminded them of its Flemish counterpart, why not spell it the Flemish way?” A Flemish spook was called a “gheest” — and so “ghost” came in as a variation. By the end of the 16th century, “ghost” had seen off “gost” — and “aghast” and “ghastly” had seen off “agast” and “gastly”. Crystal salts his work liberally with such good examples, and is able to argue that English spelling makes sense without feeling obliged to pretend that's the same thing as being simple. More than one chapter ends with words to the effect: “If you think that's the last of the exceptions, brace yourself ...”

Crystal's practical burden, set out in two short appendix chapters, is that spelling can be, and would be better, taught in linguistic context: if you can explain why “accommodate” came to be spelt like that, it stands a better chance of sticking in the mind. And yes, he says: spelling can be fun. I agree. James Joyce transcribed the noise a cat makes as: “Mrkgnao”. That is surely something that everyone from dame-school prescriptivists to the most freewheeling of linguistic fieldworkers can agree is obviously ... well, correct.

By arrangement with the Guardian