I have been thinking for some time now about creating a scheme for English spelling reform. There are many reasons for English spelling reform, the most important ones being that English is not only hard to spell for native speakers, but also hard to read. I’m not out to sell the idea of reform on this post, but just to lay out some ideas on how it could be done. So far, my ideas stem from the current spelling system, and the way English reading and spelling rules are currently taught, and some principles underlying Dutch spelling. The driving principle is to regularize the spelling system and do away with difficult to learn exceptions whenever possible, and to make the spelling correspond to pronunciation as much as possible. The goals of this reform are, in order of importance: first, to make learning to read it less difficult for native speaking children; second, to make it easier to spell for every native speaker; third, to make it easier to learn to read and spell for non-native speakers.
Some of these ideas are taken from John Wells’ English accents and their implications for spelling reform, an excellent article and required reading for anyone interested in English spelling reform.
Consonant sound spelling
The current system has a lot of letter sequences, usually 2 or 3 letters long, which are used to either represent English phonemes that the latin alphabet was ill prepared to represent with single letters, or to indicate something about the pronunciation of the previous vowel sound. There are good reasons to keep this principle going in a reformed system, and I don’t think we need to strive for a perfectly alphabetic, one-phoneme-per-letter system. However, I think some changes should be made about the spelling of consonant sounds. Here are the ideas I have so far:
1- Expanding VV and introducing SSH and possibly TTH and YY to indicate the previous vowel is “short”, to go along with other such already existing letter sequences like SS, TT, DD, CK, TCH, etcetera (ever -> evver).
2- SH and SSH would uniformly indicate both the voiceless sound in show and the voiced sound in pleasure (chef -> shef, pleasure -> plessher, plagiarism -> plasherizm, asian -> aashn). In the case of a word-final or word-initial pleasure sound, which are rare and found in loans, perhaps it is better to spell them as J (see 9 below). This would prevent spelling pronunciations of it as the sound in show and would also reflect the fact that many people pronounce it as J to begin with, because of its unnaturalness within the English sound system (genre -> janre, rouge -> rooj).
3- Uniformly spelling /s/ with S or SS, and not “soft C” (cent -> sent).
4- Uniformly spelling /z/ with Z or ZZ, and not S, SS or X (please -> pleez, peasant -> pezzant, is -> iz, has -> haz, as -> az).
5- Uniformly spelling /k/ with K or CK, and not “hard C”, CC or CH. As a result of this and the previous two changes, C would be relegated to the sequences CH, TCH and CK (court -> kort, accord -> akord, ache -> aak). Also uniformly spelling /ks/ and /gz/ as X except in cases where the /ks/ or /gz/ is created by adding a morpheme (fox, exit as is, accident -> axident, accent -> axent, except -> exept, backs -> baks, bags as is).
6- Eliminating silent consonants that give no clues as to how to pronounce a word. This includes, but is not limited to, the B in debt, the H in honor, the GH in high, the L in should, the K in known, the G in gnome, and so on (-> det, hi, onner, shud, noen, noem). This will also include spelling /r/ uniformly with R / RR (right – write -> riit, rhythm -> rithm). Other equally useless silent vowel letters will also be eliminated (guard -> gard, and see below).
7- Drawing on the two previous changes, spelling /kw/ uniformly as KW (quit -> kwit).
8- Spelling /v/ uniformly with V or VV. This would include spelling of as ov. Also spelling /f/ uniformly with F / FF (photography -> fotografy, laugh -> laf) and /h/ with H (who -> hoo).
9- Spelling the initial sound of jet uniformly with J or DJ (gin -> jin, badge -> baj, badger -> badjer). Because of this, G will only represent /g/ (its current “hard” value).
10- No longer allowing doubled consonants (SS, TT, DD, CK, TCH, VV, SSH, etc) to end a word (will -> wil, full -> ful).
11- In an idea borrowed from Cut Spelling, doubled consonants are not doubled when they do not indicate important information about pronunciation, i.e. when the vowel they follow is a reduced vowel or a “long” vowel (connected -> konekted, pole – poll -> poel, polling -> poling).
Vowel sound spelling
This is where these ideas draw on Dutch spelling. In Dutch, there are many vowel sounds and just five vowel letters, as in English. But in the Dutch system, eight of the vowel sounds are called “long” or “short” versions of A E O U, just like in English there are ten vowel sounds called the “long” and “short” versions of A E I O U. The difference is that Dutch takes a much more regular approach to spelling them. It goes like this:
Short: a e o u
Long: aa ee oo uu
Single syllable word examples: man, lek, bot, kus (short), maan, leek, boot, vuur (long)
Multiple syllable word examples: mannen, lekken, botten, kussen (first syllable is short), manen, leken, boten, vuren (first syllable is long).
A word cannot end in a doubled consonant (see 10 above).
12- A lot of this is reminiscent of English, if only it were regular in how it spelled vowel sounds. The idea here is to draw on this principle to regularize English vowel sound spelling. It would spell “short” A E I O U and “long” A E I O U in the following way:
Short: a e i o u
Long: aa ee ii oe uu
(OO is not used to represent “long O” since it will be used to spell the /u/ sound in GOOSE, in the way it’s often done in the current system.)
So for example:
Single syllable words: back -> bak, bet -> as is, will -> wil, hop -> as is, buck -> buk; blame -> blaam, seem -> as is, right -> riit, vote -> voet, cute -> kuut
Multiple syllable words: backing -> as is, betting -> as is, willing -> as is, hopping -> as is, bucking -> as is; blaming -> as is, seeming -> seming, righting -> riting, voting -> as is, cuter -> kuter.
A consequence of the rules outlined above is that the “magic silent E“, that has countless exceptions in the current system, is done away with completely.
A vowel also is read as “short” if it is followed by two consonant sounds, and needs to be doubled if it is “long”. This applies in the case of X, since X represents a sequence of two sounds (the traditional occasional /z/ value of X is removed by 4 above). This also applies if the second consonant sound following the vowel is a syllabic consonant represented by L, M or N (see 24 below).
This, of course, only takes care of 10 vowel sounds. There are more vowel sounds in English, so we need to think of ways to spell the entire vowel system. From here on I draw on Wells’ lexical sets for reference.
13- TRAP, DRESS, KIT, LOT and STRUT would be spelled with short A E I O U, with doubled consonants following if necessary. FACE, FLEECE, PRICE, GOAT, and my addition CUTE would be spelled with long A E I O U (i.e. A E I O U / AA EE II OE UU), as in the examples above.
14- PALM would also be spelled with short A, so TRAP, PALM and BATH would be spelled the same way.
15- GOOSE would be spelled with OO (goose -> goos, loose -> loos, lose -> looz), CHOICE with OY (noise -> noyz) and MOUTH with OW (mowth). START would be spelled with AR and NORTH and FORCE with OR (heart -> hart, star as is, starring -> staring, bore -> bor, bored – board -> bord, boring as is). There is a collision here with the current spelling of staring, which would be spelled stearing (and stare – stair -> stear) The same would happen with sparring.
16- LETTER would be spelled with ER (separation -> seperaashn).
17- HAPPY (word-final /i/ in polysyllabic words) would be spelled uniformly with Y. This would not change when suffixes are added (jockey -> jocky, happiness -> happynes, happier -> happyer).
18- NEAR and CURE would be spelled ER/EER and UR/UUR, i.e. long E and U plus R (near -> neer, cure -> kuur, nearest -> nerest, curing -> kuring).
19- SQUARE would be spelled EAR, as in bear and swear (square -> skwear).
20- NURSE would be spelled UR / URR, i.e. short U plus R (earth -> urth).
21- I am also pondering whether FOOT should be spelled with OO or short U. Ultimately, this is a very small set of words, but both choices are justifiable. In my proof of concept drafts, I am using short U (should -> shud).
22- CLOTH and THOUGHT present some special considerations. To begin with, not all speakers distinguish them. Also, those who do might merge them with other sets, for example, CLOTH-LOT. Other speakers merge CLOTH-THOUGHT and PALM-LOT instead (this is to a large extent my case). If we use a distinct spelling, say, AW, speakers like me will spell long as lawng, while others might be inclined to make no change because for them, it rhymes with LOT. Many speakers in America, on the other hand, make no distinction between PALM-LOT-CLOTH-THOUGHT whatsoever. Something will have to give, but maybe the best solution here is to just spell PALM with TRAP and BATH as I suggested before (with short A), and then just spell LOT, CLOTH and THOUGHT with short O. I remain open minded here, and in proof of concept drafts, I am using AW and following my own pronunciation in regards to the use of that letter sequence (cause -> kawz).
23- Also, regarding FACE, I am wondering about something different, which is if a spelling variant should be used when the sound is morpheme-final. I have been going with this in my drafts, and using EY, though AY could also be a defensible choice (both letter sequences appear in very high frequency words in the current system). The result of this is that the word they is not changed at all while laying -> leying (instead of thaa and laaing if this variant was not used). Of course, if we were less concerned with the change being too radical, and with preserving familiar word shapes, this could be dispensed with.
24- Finally, we come to a big whammy: how to spell COMMA, i.e. how to spell schwa. In the current system, it can be spelled with any single vowel letter and just about every sequence thereof. This has to be solved in one way or another, since it’s the reason behind many spelling mistakes in the current system, like effect for affect and vice versa. This problem is conflated by another, which is that some speakers contrast unstressed KIT with COMMA, while others don’t. I still don’t have a solution to this, but I think a first step in the right direction is to spell morpheme-final syllabic /l m n/ as just L, M and N. This is an idea I have borrowed from Cut Spelling and does away with the problem of how to spell unstressed vowel letters before morpheme-final syllabic L, M and N (people -> peepl, equal -> eekwl, human -> huumn, accustomed -> akustmd). There is some precedent for this in traditional spelling, like in the suffix -ISM (despotism) and the word rhythm.
Not all vowel sounds in English can occur in the same places. This affects the choice of how to spell them in a reformed system. Specifically, certain vowel sounds do not occur word-finally. These are /ɛ æ ɒ ʌ ʊ/. /ɪ/ occurs word finally in some accents instead of word-final /i/; however, in the current system these variants are often spelled Y, and my reform proposal spells them with Y invariably. In other words, TRAP, DRESS, KIT, LOT, STRUT and FOOT do not occur word-finally. However, PALM does.
25- This makes it possible to use E I U to spell long E I U word-finally, because it is impossible for the short versions to occur in those positions (see -> se, by – bye -> bi, cue -> ku). If AW is used, instead of a generalized short O spelling for LOT-CLOTH-THOUGHT, O could also be used for word-final long O instead of OE, preserving the current spelling of no and go and spelling toe as to and show as sho.
Which vowel sounds can occur in which places is also the reason why it’s not necessary to introduce NNG to indicate the previous vowel is “short”: no “long” vowels can occur before NG, only KIT, TRAP, STRUT, and LOT (or CLOTH-THOUGHT in some accents, like my own). The fact that in some accents the KIT-FLEECE contrast is neutralized before NG is a wholly predictable pronunciation rule and as such does not need to be reflected in the spelling.
Morphological spelling: possible exceptions
You might have noticed all of this leads in the direction of a “spell as you pronounce” system. The first thing in the way of that is accent differences, at least if we are going for a pan-regional, pan-dialectal system to be used by everyone and without too much in the way of equally acceptable spelling variants. Pretty much everyone will have to get used to the fact that while the new system is more regular and easier to learn, spellings of some words will still have to be learned by rote, because they are spelled according to a pronunciation distinction they do not make when speaking (while other speakers do).
26- The case can also be made for some exceptions in “spell as you pronounce” in the case of plurals, third person singular verb inflection and possessive marking on the one hand, and weak/regular verb past tense marking on the other.
In the first case, these word endings can be pronounced /z/, /s/ or /Vz/ (the vowel in question corresponding to either unstressed KIT or COMMA depending on the accent). However, they are spelled -S or -ES in the case of plurals and third person singular inflection and -‘S or -‘ in the case of possessive marking. In my drafts, I have been doing away with the possessive apostrophe and spelling the possessive invariably -S (nature’s -> nachers). What to do in cases of words ending in S to begin with begs the question – maybe, this could be an exception to the “no word final doubled consonant” rule and be spelled -SS (Wells’/Wells’s -> Wellss). As for plurals and third person singular inflection, I have been using -S except in cases of /Vz/, where I use -ES (becomes -> bekums, causes -> kawzes).
As for past tense marking, a case could be made for using -ED throughout. However, this is an exception to the “no silent letters that don’t say anything about pronunciation” principle. This is the reason I’ve chosen to use -ED when the pronunciation is /Vd/ and -D when it’s /d/ or /t/ (connected -> konekted, endowed -> endowd, established -> establishd). On the other hand, this still does little to help speakers avoid spelling mistakes like past for passed or vice versa, whether we spell passed as is or as pasd. Maybe a radical solution is the only way to avoid this problem, where passed -> past, begged -> begd and created -> kreeated. This would certainly make spelling easier. However, the question might be asked of whether it would make reading any harder. If it did, it might not be such a good idea.
27- Removing the apostrophe from the possessive S begs the question of whether to leave it alone in other cases, or not. In all other cases, the apostrophe basically stands for a missing sound in an abbreviated version of a word or sequence of words. Given how common the mistake of spelling it’s for its and vice versa is, I am of the opinion that using an apostrophe in common contractions causes more problems than anything, and so this reform proposal removes it. So in this reformed spelling we would spell its, doent, iznt, arnt, iim, duznt, didnt, havnt, haznt, aant, and so on. Spelling theyr still does not solve the problem of they’re – there – their, though, not does its theyr spelling clearly indicate its pronunciation either, whether stressed or unstressed. And we have a two-part problem (theyr – thear) instead of a three-part one. This is still something that calls for a solution. Maybe the only possible one is to spell them all thear.
The apostrophe could still optionally be used with less common abbreviations if the writer believes it is necessary (‘kawz, o’er, etc).
28- In drafts, spelling words like opinion, unalienable, transient, experience presented an interesting problem. In these words, the /iV/ sequence (the Vowel in question is unstressed) can be pronounced either with a syllable boundary, /i.V/, or as a diphthong, /jV/. Because of this, and because of the clunkiness of spelling them like opinneeon, I chose to spell them opinnyan, unalyanabl, tranzyant, experyans. Note how I am treating this Y as a vowel for the purposes of doubling the previous consonant. In this respect, this Y is like -Y (see 17 above), not like the consonantal syllable-initial Y in yet.
In English, stress deeply affects the pronunciation of vowel sounds, in what is commonly called vowel reduction. This vowel reduction affects syllables that are unstressed because of lexical stress as well as ones unstressed because of sentence stress. Traditional spelling does not reflect how sentence stress affects vowel reduction. Not does this reformed spelling. To do so would create variant forms for a large amount of words, and would remove a degree of ambiguity in writing that is sometimes desirable (sometimes we want for there to be more than one interpretation about the sentence stress and prosody of a sentence). This underlying principle of traditional spelling is solid, and in this reform proposal, as in traditional spelling, words are spelled as they would be if they were stressed.
29- A problem with English spelling that this proposal does not solve is the fact that there is no clear way in English spelling to indicate where a polysyllabic word carries stress. In my system, object is spelled objekt, but this does nothing in the way of letting the reader know whether it should be read “OBjekt” (as in the noun) or “obJEKT” (as in the verb). Context should be enough to solve this in most situations, because verbs do not appear in the same contexts as nouns; and this is the case under the current system. The problem is that, if the reader is reading a word with a reformed spelling that is still unfamiliar to them, and the word could be read two ways, if both possibilities belong to the same class of words (two nouns, for example), context becomes more and more relevant, and it might not be enough to resolve the situation.
An example: because of the elimination of “soft C”, in my reformed spelling decent is spelled desent. However, because of the lack of stress marking, it could be read like descent – dissent. This is a potential problem because the reader could go in either direction when reading it aloud. I see no way to solve this without introducing an exception into the system and spelling decent as deesent, unless a satisfactory way to spell the unstressed vowel differently is found (desant?). But I don’t want there to be a “pure exception” in the system that is hard to predict because it only affects one or a handful of words. Perhaps the syllabic N rule (see 24 above) could be expanded to words where the syllabic N in the final syllable is followed by T, so that decent -> deesnt (a rule such as this would also make the spelling of some contractions in 27 above more regular in regards to the system as a whole).
An interesting idea regarding teaching children to read and word stress, borrowed from John Cowan, is using diacritics to indicate word stress in children’s books and teaching materials. As a matter of fact, it wouldn’t be a bad idea to do this in ESL teaching materials either.
Unreduced unstressed vowel sounds
Should a reformed spelling clearly indicate unreduced unstressed vowel sounds? In other words: if a word has an unreduced unstressed vowel sound and that sound is usually indicated by a “long” or “short” A E I O U, should the reformed spelling indicate this with doubling the vowel letter or the following consonant letter if necessary? Some examples: the words education and separation have an unreduced unstressed DRESS vowel in the first syllable. In this case, however, we may consider them predictable cases of words ending in -TION and not clearly indicate the DRESS vowel in the spelling, just like in the traditional spelling (ejukaat, seperaat, ejukaashn, seperaashn). A different case is words like hallelujah; should they be spelled hallelooya or halelooya? I have still to figure this out, and in the meantime I am following the traditional spelling in this regard, thus going with hallelooya in this case. But if we want to create a spelling system that is self-sustaining and will be able to spell anything, even new words which did not previously exist, and assimilate foreign loanwords, we need a system that has a clear cut rule of whether we should indicate the quality of these unreduced unstressed vowels or not.
Should trisyllabic laxing be taken into consideration as far as the reform is concerned or should words be spelled without taking it into account? This is something that I still have to work out.
As always, feedback is appreciated.