And the Real Reason It's Impossible

1998–2007 Justin B Rye


For some years now I've been amusing myself by planning exactly what I would try in the way of “spelling reform” if I woke up one morning and found that the Revolutionary Stalinist–Linguist Party had mounted a coup and appointed me as World Dictator.  Details of my proposal for a Revolting Orthography (modestly titled “Romanised English”) are unlikely ever to become available; for now I want to get it clearly established exactly how mad this scheme is.  The problems with our current system are sufficiently well known that I feel no need to rehearse them all here; and people have been protesting about the situation for centuries.  So just what is wrong with the idea of switching to something better?  Anti‐reformists come in thirteen basic flavours, with arguments summarisable as follows.

Throughout this essay, example spellings, pronunciation guides, and so forth are marked out as follows…

Key to Notation
English text: angle‐bracketed like this
Foreign text: ditto, italicised comme ceci
Revised spellings: double‐bracketed layk dhis
Pronunciation hints: capitalised in quotes “LYKE THISS”
Phonemic transcriptions: slant‐bracketed IPA /l ɑɪ k ð ɪ s/

For explanations of the IPA characters see also my Phonemic Transcription Key.  (If you're wondering where the angles are, sorry: it'll be because your browser is ignoring my CSS…)


The existing spelling system is traditional; if it was good enough for my grandparents then it's good enough for everybody!  I refuse to learn any new system, whatever its supposed merits!

The normal reply by your run‐of‐the‐mill wimpish gradualist reformer tends to be something along the lines of: Oh dear!  I'll have to try to persuade you it's a good thing.  Well, uh, look; the old style gives GH well over a dozen possible pronunciations: BridGHam, CallaGHan, doGHouse, drouGHt, EdinburGH, eiGHth, GHost, ginGHam, hiccouGH, houGH, HuGH, KeiGHley, lonGHand, louGH, ouGHt, siGHt, touGH!  The new version is quicker, easier, more elegantly logical, and less cruel to small children (or indeed the billions of adults apparently doomed to learn English as a world language).  Please try to be a bit more open‐minded!

I on the other hand prefer the kind of reply that goes: Eat leaden death, loathsome bourgeois counter‐revolutionary running‐dogs!  (Did I say giving me Absolute Power would necessarily be a good thing?)


Giving English a phonetic writing system, with one symbol for each sound, would produce a range of ridiculous ill‐effects, such as the following:

The correct response to this argument, overlooked surprisingly often by supposed experts, is: You [ʩǂ̼ʚ̃ʡ]wit!  Who said anything about a phonetic system?  All we need is one that's roughly graphemic (“one reading per grapheme”) and preferably phonemic (“one spelling per phoneme”) and/or morphemic (“one spelling per morpheme”).

In such a system,


If we spelled words as they're pronounced, confusion would reign (or rain) since homophones like fisher/fissure, minor/miner, session/cession, and two/to would become indistinguishable.

Reply: These words already are indistinguishable when spoken, but when did this fact last cause you any significant inconvenience in a conversation?  People naturally avoid ambiguities in speech unless they're trying to contrive a pun, so if you write as you would speak homophones are no problem.  Contrariwise, ambiguous spellings like axes, bass, bow, buffet, close, does, dove, lead, live, minute, moped, number, putting, ragged, read, resent, row, sow, supply, tarry, tear, use, wind, wound currently are a problem; and such misleading homographs (or do I mean heterophones?) could be sorted out by the most moderate of spelling reforms.

Besides, there will be plenty of slack in the system to distinguish between fisher and fisyur, maynor and mayner; and as for cession… what does it mean, anyway?  I'm not making these examples up, you know.

Other major world languages faced with the homophony problem have found solutions such as the following:


Any phonemic script would need to provide distinct graphemes for each of the forty or so phonemes of English, which means seriously expanded typewriters!  We'll need either ugly diacritics or entirely novel letters – for instance, show (two phonemes, /ʃ/ + /oʊ/) will have to become something like šō!

Answer: At present almost every letter of the alphabet is severely overstrained – it's “EIGH” as in beAuty, “BEE” as in numB, “SEE” as in musCle, “DEE” as in hanDkerchief, “EE” as in siEvEd, “EPH” as in oF, “JEE” as in Gnomonic, “AITCH” as in Hour, “EYE” as in busIness, “DGEIGH” as in mariJuana, “CAIGH” as in Knee, “ELL” as in couLd, “EM” as in Mnemonic, “EN” as in damN, “EAU” as in leOpard, “PEE” as in Pneumonic, “KEW” as in lacQuer, “AHR” as in dossieR, “ESS” as in iSle, “TEE” as in husTle, “YOO” as in bUild, “VEE” as in kalashnikoVs, “DOUBLEYOO” as in Wry, “ECKS” as in fauX, “WIGH” as in mYrrh, “ZED” as in capercailZie!  But in a reform, what's to stop us using two‐letter graphemes (as in sh ow)?  That way there are more than enough possibilities; we can even retire Q, X, and our existing ugly diacritic, the apostrophe!  One new vowel symbol would be handy; I'd go for Scandinavian‐style slashed O as in Bjørn.

But by the way, while we're addressing hypothetical typewriter manufacturers, I'd better warn them that the old QWERTY keyboard will be declared ungoodthinkful too.  Its deliberately unergonomic layout, designed to slow down common sequences on early manual typewriters, is a thoroughly pointless legacy once we're typing different common sequences on unjammable palmtop keypads.


This revised spelling scheme looks completely alien to English orthographic traditions.  If schoolchildren are taught only the new version, we'll lose touch with our literature; our cultural heritage will be lost unless kids can read Shakespeare in the original!

Normal reformers' reply: Aren't you overreacting a bit?  We'll phase it in slowly, so there's plenty of time to reprint the classics – most of the editing required is simple search‐and‐replace work.  Compare the gradual process of metrication.  Other languages manage spelling reforms once a generation; and the Japanese seem to be perfectly happy using several very different writing systems in parallel!

My additional remarks: First – if, as is here conceded, the old orthography looks so very unlike a reasonable one… why stick with it?  People complained about the jarring novelty of electric lights, but I don't hear anyone these days campaigning for a change back.  Second – anyone caught using pecks and bushels after the tenth anniversary of my glorious rule will be branded on the forehead with the word idiot.  And third – trying to read Shakespeare “in the original” is futile.  As originally composed, it was…

In other words, the whole thing is unintelligible without either an annotated translation, which might as well be in a reformed spelling, or weeks of specialised training, which would be no more worthwhile than teaching every child how to pilot a biplane.


Adult readers recognise whole words by their overall silhouettes, not by decomposing them into the sounds.  What's the point of improving the correspondence of sounds and symbols?  It'll only mean we have to relearn the silhouettes!  (And then of course we'll have to go through the whole thing all over again the next time the language changes…)

Reply: Actually, there are three skills involved in fluent reading…

The upshot is that spelling reform might be briefly awkward for word‐recognisers, but would eventually be an advantage even for them – if only because it allows more hieroglyphs to fit on a page!  For children (and many, many adults), it would be an enormous, immediate, and permanent improvement.  Or at least, as good as permanent; if the orthodox system can outlive its best‐before date by half a millennium, we can leave the next reform for Buck Rogers to worry about.


What about a spelling reform's incidental effects on word‐games, abbreviations, and so on?  If the dictionary contains more Ks and Zs than Ds and Hs, the Scrabble‐players are going to riot!

Reply: Ah, yes, a much more intelligent point.  (Okay, I admit it, this one's a plant; I've never seen it considered anywhere else, but I thought it deserved an airing.)  Scrabble‐players will have to decide whether to play “historical” or “recalibrated” Scrabble; the rest of us will just have to get used to the idea that the E.U. is the Y(uropian) Y(union), K.O.s are N(ok)‐A(wt)z, the C.I.A. is the S(entral) I(ntelijens) E(yjensi), and a G.H.Q. is a J(eneral) H(ed)‐K(worterz)A.I.D.S. may still be A.I.D.S., but this is no longer the same as the word eydz; and since any serious reform would also change the names of the letters, even the unaltered initialisms may be hard to recognise in speech: A.I. for instance becomes “AH EE”.  If you think that's confusing, count yourself lucky I'm not reforming the Phoenician‐derived alphabetical order!

Come to think of it, I.D., O.K., and many others (especially tradenames) are already anomalies, not standing for any particular real series of English words; and acronyms such as laser, quango, or ufo are effectively independent of their original forms too.  Do we make it aydi, leyzer or I.D., L.A.S.I.R.?  And as for G.N.U. (“GNU's Not Unix”)… I don't particularly care what happens in these cases; but the marketing director of I.C.I. might.


The orthodox system, which spells changes, joints, and qualifications exactly as French does, is very useful for those who know French and want to learn English, or vice versa.  Changing the spellings to, say, ceynjiz, joyntz, kwolifikeysyonz will make polyglottism even rarer!

Reply: True, our Norman‐influenced orthography is a bridge between English and French.  But why force everyone to learn it as the only spelling system for English?  Most Asian (or even Scandinavian) learners of English care little for French; and Texans would be better off with a bridge towards Spanish.  Personally, I would have been happy to learn a bit about Anglo‐Norman during my years as a French student, but nobody wanted to tell me anything about it then!

There are three main problems with spelling English as Anglo‐Norman:

  1. Mediaeval French isn't Modern French.  The three examples above used to be pronounced roughly as spelt (“TSHAN‐DZHES”, “DZHO‐INTS”, “QUA‐LEAFY‐CATSY‐ONS”), but nowadays they're barely recognisable (“SHAHNGZH”, “ZHWENG”, “KALI‐FEEKASS‐YAWNG”).  French could do with a new broom of its own – I'd suggest xanjhz, jwentz, kalifikasionz!
  2. Mediaeval English isn't Modern English.  The biggest change is the Great Vowel Shift, which is responsible for our pronunciation of A, E, I, O, U as “EH EE EYE OWE EWE” (as in no other writing system on the planet), rather than approximately “AH EH EE OH OO” (as in Old English, Finnish, Latin, Indonesian, Swahili… etc.).  The first hurdle for language teachers is usually to persuade pupils that (e.g.) dei is “DAY‐EE” not “DEE‐EYE”; a spelling reform that made English less insular would be a great help here.
  3. Mediaeval French never was Mediaeval English.  Applying Romance orthographic prejudices to a Germanic language just caused trouble from the start – witness the Norman scribes' use of:
    • Cosmetic O in place of U in cOme, mOnk, wOnder, wOunds, and many others where a U would be awkward in clerical handwriting (too many consecutive vertical strokes).
    • “Soft C” in Cell.  Germanic “K”s didn't soften like this.  Result, confusions such as Celt, Coelacanth, maCintosh, sCeptic!
    • “Soft G” in Gin.  Again, English “G” sounds never obeyed this rule; hence the inconsistencies in Gaol, Give, judGment, marGarine.
    • Silent U to signal exceptions to the above (gUess, gUild, mosqUe, plagUe, qUoin) – especially unwelcome in that it interferes with the following.
    • QU for the “KW”‐sound in QUeen (the Anglo‐Saxons had preferred to write cwene).
    And then there's the confused way they handled the voiced fricative sounds:
    • “V” written V in VerVe (with a pointless final E, as usual).
    • “DH” written TH in THe (hopelessly mixed up with the “TH” in THirtieTH).
    • “Z” written S in uSerS (leaving idle the more appropriate Z).
    • “ZH” written S in viSion (never properly recognised as a distinct sound).
    • And the now‐silent sound written GH as in liGHt (simply liht in Anglo‐Saxon).

All in all, we're better off without our Anglo‐Norman heritage!


English is full of vocabulary items borrowed from other languages – some fully naturalised, some just temporary visitors.  This is largely because its anything‐goes attitude to spelling places no restrictions on words like cinquecento, connoisseur, or Fraulein.  If we reform these their sources will become unrecognisable!  Besides, what are we going to do with names like Caesar, Einstein, or Munich (and come to that, Rye)?

Reply: English is hospitable to immigrant words because it has simple morphology, rich phonology, and a cosmopolitan tradition.  Spelling is irrelevant – witness the words fatwa, futon, and glasnost, taken from languages that don't even spell them in the same writing system as we do!  My policy on imports would be:


Spelling wrestling as we do is a useful guide to the word's provenance.  In its Old English form the word was indeed pronounced with an audible “W”, “T”, and “G”.  If we change our spelling we'll lose all these clues!

Reply: If etymology is a sufficiently important subject that primary school children are forced to master a Mediaeval Reenactment writing system on this basis, why are those children never actually taught even the basics of linguistic history?  Surely any kid who has gone to the trouble of learning an etymological spelling for wrestling (etc.) should be entitled to go on and take the subject at GCSE level!  But somehow I suspect that most people find etymology supremely unimportant in their lives… If anyone ever needs to know the origin of the word reslinh, there will still be dictionaries about.  Come to that, they will be easier to use (you can find the word under R) and have more room for etymologies (as they need less room for pronunciation guides)!

Besides, why stop at Old English?  Why not write everything in Proto‐Indo‐European?  English spelling is much less help as a guide to lexical history than it would be if anyone cared, featuring as it does…

I'm not saying we should necessarily wipe out such etymological traces as the specific unstressed vowels in inter­administrative or even the Greek PHs in philosopher (which can all convey useful morphological information); just that etymology isn't one of an orthography's main concerns.


The trouble with a more phonologically representative spelling system is that it would reveal the nonstandard ways dialect speakers interpret the graphemes of written English.  Tutor for instance is “TOODUR” to a Nebraskan, “TEWTRR” to an Aberdonian, and “CHOO'AH” to a Cockney; woe betide any speaker of BBC English who tries to impose some lah‐di‐dah “standard spelling dialect” on the inhabitants of the East End!

Reply: At last we're getting to the non‐trivial arguments!  Yes, there's an important problem here that the system has to deal with carefully.  But its nature is still obscured by several layers of misunderstanding, which I'll try to handle quickly:

There are four basic ways in which accents can vary:

  1. Phonetic (or “realisational”) variation.  Trifling but obvious features like the way Cockneys pronounce bay almost as “BUY” (while buy becomes more like “BOY” and boy like “BOOY”).  Cockneys have no trouble distinguishing them and lining them up correctly with the written forms, so this is irrelevant to the orthography.
  2. Phonemic (or “systemic”) variation.  Added or lost distinctions, such as between “TH” and “F” (Cockneys pronounce thin the same as fin).  If the spelling system makes more distinctions than you do, you can ignore them while reading, and your difficulties in learning to write will be nothing new or serious (“Hmm, is it spelled theft or feft?”).  On the other hand if it makes fewer distinctions you'll have serious trouble reading (“Hmm, does this say THREE CLOVES or FREE CLOTHES?”).  The lesson I draw from this is that the spelling system should make all the available phonemic distinctions – and not just the ones the Queen makes.
  3. Phonotactic (or “distributional”) variation.  This is variation dependent on the phonetic context, like the way Cockneys – and in fact the English generally – drop any “R” sound that isn't followed by a vowel (so that “LARDER” = “LADA”).  Again, the orthography should side with those who keep the distinctions clear, which in this case means spelling a lot of words with an R omitted by BBC newsreaders.
  4. Lexical (or “selectional”) variation.  Disputed idiomatic cases such as “GRASS/GRAASS” or “DOSSLE/DOHCYLE”.  Where these are real regional standards rather than merely outbreaks of “spelling‐pronunciation” (like saying “CUP‐BOARD” for “KUBBERD”), they have as much right to be tolerated as alternative spellings as they have to be tolerated as alternative pronunciations.  Obviously, you ought to be consistent, but if your recipes refer to tomeyto they will communicate at least as effectively as if you “standardised” it to tomahto.

In summary, then… as long as people understand the ways accents vary (a body of knowledge which will clearly be one of the main influences on the system's rules, but which any Cockney already needs for communication with non‐Cockneys), there is no reason to imagine that there are any insurmountable problems here – how many of the people who claim that creating a pandialectal scheme is impossible have ever even tried?


A purely phonemic system (obeying the principle of One Spelling Per Phoneme) would often mean giving divergent spellings to different forms of a single morpheme, concealing relationships between words in contexts such as… One of the few merits of the old style is that it makes obvious the connection between nation and national, which will be disguised if they're respelt neyshn and nashønal.

Reply: Absolutely – the morphemic principle (One Spelling Per Morpheme) conflicts with the phonemic system and is worth making concessions over.  Affixes that still work as productive processes, like plural ‑s or past tense ‑ed, should be given consistent single spellings wherever possible (including words such as raging/agEing, publicly/cyclicALly, tangos/mangoEs, wiry/fiEry where the conventional spellings are flagrant breaches of this principle).  Likewise, compromises can be found for the stress‐shift and consonant‐softening cases, though there is room for debate about how far it should be allowed to complicate things…


All this talk is pointless.  The anglophone nations are too lazy, ignorant, and superstitious; even if you were world dictator, you'd never get them to cooperate on a project that involved this much work and was this insulting to all their ludicrous national traditions.  Americans think any attack on their honor is un‐American, Brits are still stuck in the Middle Ages, and Australians of course think literacy's for poofs…  Besides, none of them can think straight about phonological issues, largely because their brains are hopelessly clogged with Anglo‐Norman delusions.

Reply: Well, I'm certainly glad I didn't say that…

Imagine the heartaches / Of diplomatic attaches / When the wind detaches / Their false moustaches


In case you're wondering, no, I don't believe that this sort of wholesale spelling reform would be a workable proposition, but I'm so sick of watching Aunt Sally reform proposals being pelted with ridiculously inadequate arguments that I thought it would make a nice change if I wrote something equally biassed and unfair in the other direction…  So don't expect me to provide a Mailbox like the one on my anti‐Esperanto page!  The flaws of the standard orthography are indefensible – but it has an extensive Installed User Base, and can thus afford to ignore criticism in exactly the same manner as QWERTY keyboards, Fahrenheit thermometers, and certain software packages, which can all rely on conformism, short‐termism, and sheer laziness for their continued survival.