English spelling is notoriously inconsistent and difficult, at least to its own native speakers. Ages ago I saw a map of Europe with percentages listed for how many words seven-year-olds spelt wrong in the main/dominant language of each country. English topped the list, with a full 67% of all words being misspelt – the only other language that even hit double-digits was French, but their error rate was still only half of English’s. And in my time teaching at primary schools, I’ve become very well aware of just how much time and effort we have to put into teaching kids spelling. There seem to be two main strategies:

  • Rote memorisation of “sight words”, those extremely high-frequency words that are usually the most irregular (like “was”, “would”, “done”, etc.)
  • Studying “families” of words that share the same spelling and pronunciation, like one week you might do all the “-air” words (chair, fair, pair…), then another week might be “-ine” words (line, fine, dine…). Basically it takes years to get through this approach.

Over the centuries, countless English speakers have thought to themselves: There must be a better way. And numerous people have made proposals for how we could make English spelling better, more logically consistent, and more phonetic. In the US they have the example of Noah Webster (external link), who promulgated a number of spelling reforms through the educational spelling book that he produced. The reforms he pushed forward were all very modest and didn’t really get to the heart of the problem, but nonetheless that’s just about the only example we have of successful spelling reform in modern English. A more radical proposal in the mid-20th century, the Shavian alphabet (external link), was invented as a way of addressing the problem that English has like 40 distinct phonemes and the Latin alphabet has only 26 letters. While clever, obviously, it never caught on.

At this point it seems highly unlikely that English will ever undergo a real, from-the-ground-up overhaul of its orthography. For a start, unlike many other languages, English has no centralised Academy with the kind of auth­ority required to push through reforms to its spelling. Many people would also point out that there are some advantages to our idiosyncratic spelling: while it’s not very consistent at all if you consider how words are pron­ou­nced, it is a pretty good guide to the history and etymology of individual words. If an English speaker goes on to learn French, for example, they’ll find a ton of words that are not really recognisable in spoken form at all are recognisable in writing, because both languages have been pretty conservative orthographically. There are also a number of ways in which English spelling is more consistent in terms of meaning than pronunciation: one example that sticks out in my brain is that the second C in “electric”, “electricity” and “electrician” are all pronounced differently, but it’s there in all three words as part of the root morpheme “electric”. Another would be that -ed is relatively consistently used as the “past tense” marker in English (if you exclude irregular verbs), even though its pronounciation can vary from /t/ (as in “marked”), /d/ (as in “cried”) or /əd/ (as in “pitted”).

Nonetheless… I think I speak on behalf of all Anglophone linguistics nerds when I say it’s just kind of fun to imagine what could be – as in, what a phonetic system for spelling English could look like. And for some reason, after my cat woke me up before 6am this morning for breakfast and I was struggling to get back to sleep, this is what my brain fixated on 😅 So, on that note, here are some thoughts of mine on what a phonetic spelling system could look like.

Firstly, my thinking was that any new system should still look kind of like English, and be basically comprehensible (at least) to existing English speakers. While yes, you can go for a Shavian-like system with a 1:1 correspondence between phonemes and letters, that means giving up on the Latin alphabet (or at least adding umpteen new letters to it), and that makes any proposal a lot harder for interested randos to quickly pick up. It also means that I didn’t really want to drastically change what any letters represented. Like, you could certainly repurpose letters like Q, which doesn’t do anything K doesn’t, or X, which could be replaced by “ks” or “gz” pretty much everywhere, or C, which only really does anything unique as part of the digraph “ch”. But if you give them a new use then you’ve gotta explain what their new use is, which makes the system harder to immediately pick up. And these kinds of letters are not really the big problem with English spelling, anyway.

The big problem with English spelling has, for many centuries, been the vowels. English has umpteen different vowel phonemes (the exact number varies depending on accent) and, thanks in part to our very conservative spelling, we have the twin problems of letter combinations that can refer to multiple different sounds (like wheat, bread, great) and sounds that can be written with multiple different letter combinations (like bye, tie, dine, fly…). So in my brain at six in the morning, this is really what I wanted to solve, and I thought consonants could really be worried about later.

Mentally I think of English as having a few different logical groups of vowels (including diphthongs), so noting my admittedly Australian perspective these are:

  • Regular short vowels: [æ] (cat), [e] (bed), [ɪ] (kit), [ɒ] (dog), [ʊ] (put), [ʌ] (cup)
  • “Long vowels”: [ɑː] (father), [iː] (machine, lean, keen), [ɔː] (law), [uː] (rune, tool)
  • “R-coloured vowels”, which are not actually r-coloured in my non-rhotic accent, but form a group in my mind all the same: ar as in cart, the “er” sound in words like herd, bird or hurt, the other “er” sound (“air” sound?) in words like pear, fair or there, the “ir” sound in words like fear or weird, the “or” sound in cord, and the “ur” sound in words like pure or tour in some people’s accents. (For me, “ar” has merged with the “father” vowel, and “or” and “ur” both with the “law” vowel, but I know other accents distinguish them!)
  • Other diphthongs: [aɪ] (bye, tie), [eɪ] (day, prey), [ɔɪ] (joy), [aʊ] (house), [oʊ] (goat) and [ju] (huge) (the /u/ → /ju/ shift in English is so ancient though that in many cases the /j/ has coalesced with a preceding consonant or dropped entirely in many contexts, though)
  • Finally, schwa. In Australian English we just have [ə] (which we actually pronounce the same as the vowel in “cup”, just unstressed); some other accents distinguish between that and unstressed [ɪ] or [ɨ] (lacking the weak vowel merger) and rhotic accents also have “r-coloured schwa”, or [ɚ]. This may not be a large group, but it’s a very important one due to the high frequency of schwas!

So today I did some brainstorming, and tested my own systems out writing stream-of-consciousness paragraphs, and this is how I ended up allocating the written signs to spoken vowels:

  • Short vowels: Note, as far as I’m aware none of these can go at the end of a word, so if you were to see one of these vowels at the end of a word, take it to mean the long vowel instead.
    • a for [æ] as in cat
    • e for [e] as in bed
    • i for [ɪ] as in kit
    • o for [ɒ] as in pot
    • u for [ʊ] as in put and [ʌ] as in cup. This was mainly because I was running out of letters, and merging FULL-FOOL (leaving this letter exclusively for the “cup” vowel) seemed like a worse alternative. I guess you could put the cup sound on y, but to me that seemed to fail the “intuitive to existing English-speakers” test, as “y” is usually thought of as more like “i”.
  • Long vowels: Represented here with a macron, although you could choose a different accent, or maybe double the vowel (which I didn’t do mainly because “oo” already has a different meaning in real English spelling). As mentioned, at the end of a word I don’t think you’d need a macron or any other accent, because short vowels don’t go at the ends of words.
    • ā for [ɑː] as in father
    • ī for [iː] as in machine
      • to look more “Englishy”, you could use y for this at the end of a word.
    • ō for [ɔː] as in law
    • ū for [uː] as in rune
  • R-coloured vowels:
    • ar for [ɑr]/[ɑː] as in cart
    • er for [ɛr]/[eə] as in fair, square, there, bear
    • ir for [ɜr]/[ɜː] as in bird, fur, perk, work
    • or for [ɔr]/[ɔː] as in cord, more, four
    • ur for [ʊr]/[ʊə] as in sure, tour
    • yr for [ɪr]/[ɪə] as in fear, weird
  • Other diphthongs:
    • iy for [aɪ] as in bye or tie (you could use ay as well, which would be closer to the IPA transcription, but it already means something different in real English spelling and this sound is often linked to the letter I in English so may as well use “iy”)
    • ey for [eɪ] as in hey, day
    • oy for [ɔɪ] as in toy, join
    • au for [aʊ] as in house, wow
    • ou for [oʊ] as in so, grow, though
    • yu for [juː] as in huge
  • Schwa:
    • Most disappointingly… ə for a normal schwa [ə], as in nation, cadence
    • ər for r-coloured schwa [ɚ], like in letter or winner

I would have liked not to have a whole separate letter for schwa. I guess I didn’t allocate any sounds to y (by itself) or ē – I could’ve either used “y” for schwa or moved the “bed” vowel to ē then used e for schwa, but the former doesn’t come across intuitively like a schwa, and I didn’t want the short vowel in “bed” treated as a long vowel.

Another option would have been keeping schwa with the STRUT vowel (which is, in my accent at least, the same vowel but for stress anyway) at u. Worth noting that in my experience, when five and six-year-olds are learning to spell this is what they do intuitively – which can take you aback when you’re not expecting to see “paper” spelled like “papu”. I’d still have to work out what to do about r-coloured schwa though, since I already assigned ur to mean something else. So, for the purpose of this thought experiment, the least-bad option seemed to be to add the letter ə to the English alphabet. And hey, it’s not like we’d be the only language to do it… the letter is also used in Azeri…

To the best of my ability, this was designed to be a cross-dialectical system of spelling English. I’d like to caveat that by saying I’m mainly familiar with General Australian, General American, Received Pronunciation and to some extent other accents of the UK… I’m aware that my system already lumps the FOOT/STRUT vowels on the same letter but there may well be other splits I’m not aware of that it also doesn’t cater to 😅 What I am aware of, though, is that my system also draws distinctions between various vowels that are merged in different dialects. For example:

  • as mentioned, the r-coloured vowel “ar” sounds the same as “ā”, and “or” and “ur” the same as “ō”, in my own, non-rhotic accent
  • the cot-caught and father-bother mergers are both pretty widespread in North America, so for many North Americans they’ll hear no difference between “o” and “ō” or “o” and “ā” (or for some of them, maybe even all three?)

I guess there are two ways you can go with that. My own preference would be that English speakers have a little bit of awareness of what other accents sound like, and can apply that knowledge in their spelling. Option B would be that people write whatever vowel they use in their own accent, and people would just adjust, the way we do when we see the one out of -or/-our or -ise/-ize we don’t use ourselves 😛 At least those kinds of spelling differences would actually be meaningful. I have my preference, but either one is fine really, and for purposes like song lyrics or poetry option B might even be superior.

Anyway, so that’s vowels. As far as consonants go, I feel like it’s generally a much easier story, although there are still some tactical decisions to make. Some of these would include:

  • In real English spelling, the letter S can represent the sound /s/ (like both the ones in secrets) but also /z/ (like both the ones in roses). Perhaps where it sounds like a Z, it should be spelt with that letter, too.
  • The letter D is often pronounced /t/ where it appears as a past tense marker, like in jumped or scoffed. Do we keep the -ed ending for faster recognition as a past-tense verb form, or do we go to spelling it as it’s pronounced (e.g. jumt, scoft)?
  • C basically only does anything unique as part of the digraph “ch”; everywhere else it could be replaced by either S or K. The thing is that it’s mostly not harming anything by existing elsewhere, because 99% of the time it’s pronounced /s/ before E or I and /k/ in all other positions (those rare exceptions should get spelling-reformed out of existence, though 😛). I guess it does make life a little harder if you’re only just learning to spell, but not that much harder. It’s also a pretty common letter in real English spelling, so the “reformed” version looks less alien (and potentially off-putting) if you keep it.
  • Q, similarly, doesn’t do anything that K doesn’t do. But by the same token, having the digraph “qu” doesn’t harm anything because it’s not ambiguous and there aren’t all that many words with the /kw/ sound spelt differently.
  • G, like a more problematic version of C, is soft before E and I in words of Latin origin but hard in all other positions and before those letters in words of Germanic origin (like girl or get). Because of its inconsistency I think soft G should be changed to J, so that when you see G you’ll know it’ll always be in its hard form.
  • It would make sense to introduce the digraph “zh” for that sound you hear in words like mirage or casual. Just to have a real, explicit way of spelling it.
  • The letter combination “th” can mean either devoiced /θ/ (as in thing, fourth) or voiced /ð/ (as in this, with). To distinguish them, you could use a different letter combination, like “dh”, for the latter. Or you know, even re-introduce the letter Ð to the English language, if we’re adding schwa anyway.
  • X can refer either to the combos /ks/ or /gz/, and you kind of just have to know which. You could argue that it’s not a necessary letter anyway, so you might as well just replace it with whichever one of those combos applies to a given word.
  • With most arguments like “hey, we don’t even need such-and-such letter, let’s get rid of it!” I feel like it’s kind of pointless because unless you’re going to repurpose that letter for something else, why bother getting rid of it? It’s not harming anything. But in the case of X, there are some other languages where it means /ʃ/ (a sound we otherwise spell “sh”) and perhaps there is some value in having a single letter for that single sound. We could use X. In Catalan, they go a step further and replace “ch” with the digraph “tx” – we could also do that. The thing is, both of these would also look pretty alien to today’s real-world English speakers, so they might not be a great move if you really wanted to encourage adoption. As a thought experiment though, you can do what you like 😊
  • Real-world, modern-day English makes a lot of use of double consonants (like tapping, hopping) to make clear the preceding vowel is short (i.e. don’t get it mixed up with taping or hoping). With a different system for writing vowels where it’s clear from the vowel itself whether it’s long or short, then you don’t need double consonants for this purpose.
  • Should you distinguish between /ŋ/ (e.g. singer) and /ŋg/ (e.g. finger)? Many Austronesian languages do. You could put an extra G on the end where it’s pronounced (e.g. “finggər”) and eliminate any ambiguity that way.

Do I expect that I will ever be in a position to foist my feelings about English spelling on the rest of the world? Absolutely not. I’m not an expert and I haven’t exactly done exhaustive research before composing this post; I’m more just thinking aloud, working stuff out for my own gratification.

Anyway, to put this rambly post together, here is the beginning of the story Rumpelstiltskin rewritten to use my proposed respellings (with at least some system for dealing with the consonants).

Biy ðə siyd ov ə wud, in ə cuntry ə long wey of, ran ə fiyn strīm ov wōtər; and upon ðə strīm ðer stud ə mil. Ðə milər’z haus woz clous biy, and ðə milər, yu must nou, had ə very byūtiful dōtər. Shi woz, morouvər, very shrūd and clevər; and ðe milər woz sou praud ov hir, ðat hi wun dey told ðə king ov ðə land, hu yūst tu cum and hunt in ðə wud, ðat hiz dōtər cud spin gold aut ov strō. Nau ðis king woz very fond ov muny; and wen hi hird ðə milər’z boust hiz grīdynes woz reyzd, and hi sent for ðə girl tu bi brōt befor him. Ðen hi led hir tu ə cheymbər in hiz paləs wer ther woz ə greyt hīp ov strō, and geyv hir ə spining-wīl, and sed, “Ōl ðis must bi spun intu gold befor morning, az yu luv yor liyf.” It woz in veyn ðat ðə por meyden sed ðat it woz ounly ə sily boust ov hir fāthər, for ðat shi cud du nou such thing az spin strō intu gold: ðə cheymbər dor woz lokt, and shi woz left aloun.

Shi sat daun in wun cornər ov ðə rūm, and began tu beweyl hir hard feyt; wen on a sudən ðə dor oupend, and ə droul-luking litl man hobld in, and sed, “Gud morou tu yu, miy gud las; wot ar yu wīping for?” “Alas!” sed shi; “Iy must spin ðis strō intu gold, and Iy nou not hau.” “Wot wil yu giv mi,” sed ðə hobgoblin, “tu du it for yu?” “Miy neckleys,” repliyd ðə meyden. Hi tuk hir at hir wird, and sat himself daun to ðə wīl, and wisld and sang:

Raund abaut, raund abaut,
  Lou and behold!
Rīl awey, rīl awey,
  Strō intu gold!

And raund abaut ðə wīl went merily; ðə wirk woz quickly dun, and ðə strō woz ōl spun intu gold.

I’ll admit that despite my best intentions it does still look very different from real written English, but at least I’m sure it’s easier to wrap your head around than the Shavian alphabet 😆 At any rate, this is really all more of a thought experiment, done in good fun, than a real serious proposal. A lot more consideration would have to be put into it before that, and considering the implausibility of English spelling ever being reformed, why bother! Still, it is fun to play around.