[Journal of the Simplified Spelling Society, 1987/2, p30-32 later designated Journal 5.]

Escaping from a Dialect Straitjacket 2.

David Stark.

This is the fifth in a series of articles of which the first three appeared in the Simplified Spelling Society Newsletter Summer 1985, Spring 1986 [J2], Summer 1986 [J3], and the fourth in the 1987/1 issue [J4] of this Journal.

My last article set out principles for making a revised orthography more applicable to various accents by allowing certain pairs of similar phonemes, which can be easily confused in some accents, to be represented by single graphemes. This is possible because differing phonemes do not just appear in different accents, but are the result of phoneme splits or convergences in the historical development of language. These changes have taken place in some accents but not in others, and are consequently only recognized in some of them.

The next two articles explore the potential for such an approach by looking at some of the features which separate the major accents. The main source of reference for this study is Accents of English by J C Wells. This excellent book should be on the bookshelf of every serious orthography reformer.


The London newsreader's announcement that groups were campaigning against the impending extinction of Wales, largely brought about by excessive wailing by Japanese, filled me with indignation and disbelief at the fate awaiting my fellow Celts. The momentary misunderstanding was caused by the RP feature of reducing /hw/, or a slightly different phoneme in my accent, to /w/. This is known as Glide Cluster Reduction and makes pairs like Wales: whales and wailing: whaling into homophones.

This feature, although present in the south of England before the time of Shakespeare, did not become acceptable in educated speech until the late 18th century. While some RP speakers still use /hw/, probably because of spelling pronunciation, most accents of England, Wales, West Indies and the southern hemisphere have only /w/. Although there is evidence that Glide Cluster Reduction is growing in the United States, it is not common in North America, and it is not present in the accents of Scotland and Ireland.

As discussed in my previous article, the functional load of the opposing phonemes is probably small enough for them to be represented by the one grapheme.


The distinction between the vowels in lass and pass was discussed in the last article and it was concluded that it would be unwise to express them as two separate graphemes because of the variability of these vowels in various English accents. Someone who does not possess the distinction at all, or in a different way to RP, would have to learn lists of spellings non-phonetically, so that he knew that lass, gaff, hath, mascot, hassle, manse, rant, gander and trample had different initial vowels from pass, staff, path, basket, castle, dance, grant, slander and example. Minimal pairs like aunt: ant are few, and t.o., predating the lass: pass split, normally represents both phonemes with the same grapheme. It would seem sensible for a revised orthography to do the same.


Most accents of English distinguish between the vowels in put: putt or stood: stud and those in pool: pull. The fact that some do not is the result of several changes in pronunciation since the original, relative consistency in the Middle Ages. The results of these changes, which in themselves would require lengthy explanation, are shown the following table.

Middle English
Great Vowel Shift
Output (RP & Gen Am)
Output (N England)
Output (Scotland)



The result is that put:putt and could:cud are homophones in some local accents in the north of England and in Ireland. If one could represent these vowels with the same grapheme in a revised orthography, the alphabetic advantages of spelling affected words would be enjoyed by adherents of these local accents as well as the rest of the English-speaking world. There are not many minimal pairs, indicating a low functional load in accents which have this distinction. The common ones are book:buck, could:cud, put:putt, rook:ruck, stood:stud and took:tuck. The main reason that there are so few minimal pairs is that the lexical set of words containing the put vowel is relatively small, numbering about 60 common words plus the <-ful> ending. One could therefore conclude that it is possible for the two vowels to be given the same grapheme, or alternatively one would accept that a small section of the English-speaking world would require to learn these 60 words by non-alphabetic means.

The above table also shows that pool:pull are homophones in most of Scotland. Also, these two phonemes, while present in most accents, have some lexical variance in different areas. For example, words like hoof, roof, tooth, soon and spoon are pronounced with a short /ʋ/ in parts of the west of England, as opposed to the /u:/ of RP, and even within RP some words like broom, groom and room can be pronounced either way.

Again one is faced with a similar choice, whether or not to allow the one grapheme for both. There are even fewer minimal pairs than with the putt:put vowels, these being fool: full, pool: pull and too: two: to. However, if one does combine the pool:pull vowels one would exclude the opportunity to combine the put:putt vowels, as the number of minimal pairs for /u:, ʋ)/ v./ ʌ/ is large, numbering about 35 common pairs of words.

While New Spelling advocated <oo> for the pull vowel and invented <uu> for the pool vowel, it did not prohibit a Scotsman writing <oo> in all the relevant words. The necessity to invent a new grapheme <uu>, and then place it with the more common of the two sounds, would cause a large amount of alteration to traditional orthography, which itself does not positively distinguish the two phonemes. For these reasons, and the number of people who would be adversely affected by a written differentiation between the two, I would favour the combination of the pool:pull vowels in the one grapheme.


During the 16th century, the phoneme /ju:/ lost the /j/ element (the yod) in words like threw, which then became homophone of through. This occurred in certain environments, namely after palatals (as in chute, chew, juice), after /r/ (as in rude, crew, shrew, grew), and after consonants plus /l/ (as in blue, flue, flew, glue). While there are some accents in Wales, the north of England and America which remained conservative in this respect, the vast majority of accents now have a /u:/ phoneme in such words. Only a small number of words, like enthusiasm, suit, resume, can be pronounced both /ju:, u:/ in RP.

This process continued to develop in General American until the /ju:/ is now only present after a few consonants, for example beauty, cute, mutant. The /j/ element has disappeared for most Americans in the following environments:

/t-/ tune, student, attitude
/d-/ duke, reduce, during
/n-/ new, numerous, avenue
/θ-/ enthusiasm
/s-/ suit, assume, pseudonym
/z-/ presume, resume
/l-/ lewd, allude, solution.

We are then faced with the problem of coming to terms with the majority of people having /ju:/ in words where a large minority of the English-speaking world has /u:/. One might consider combining the two by representing them with a single grapheme. The number of minimal pairs is fairly small, about a dozen common words in RP, with a few less in General American where tooter:tutor, do:due:dew, overdo:overdue are homophones.

The phoneme /ju:/ is almost always represented within words in traditional orthography by <uCe> (C = consonant) with the <e> dropped if another vowel immediately follows, and simply by <u> if a vowel follows the /ju:/. Examples of this are cute, future, fuel. The same spelling rule exists for /u:/ in traditional orthography, as illustrated in words like rude, rudiment, gruel, this method of representing the phoneme being only slightly less common than <oo> as in boot.

There would be a disinclination to allow /u:/ and /ju:/ to be represented by the one grapheme by reformers who have always separated them, and by those who reckon that minimal pairs like coot:cute, booty:beauty, food:feud, ooze:use, moo:mew, moot:mute, poor:pure, who:hue/hew need to be differentiated in writing to avoid widespread semantic confusion. All that can be concluded is that, since these minimal pairs are few in number and the functional load of the difference between the phonemes is consequently low, it is possible to use one grapheme for both, and maximize the alphabetic spelling potential on both sides of the Atlantic for the relevant phonemes.


In association with yod dropping, a preceding /d, t, s, z/ sound will change in some accents to /ʤ, tʃ, ʃ, ʒ/ in words like during, tune, issue, visual. The only solution to this variation seems to be to retain the existing spellings, or at least to simplify them, so that the Standard Spelling Pronunciations indicate /dju:, tju:, tju:, zju:/ rather than /dʒu:, tʃu:, ʃu:, ʒu:/ respectively. These pronunciations and spellings would be learned irrespective of which way one would naturally say them. At the moment, many people who would normally display yod coalescence in their everyday speech would claim that the proper pronunciation contains the yod, probably being influenced by the spelling pronunciation.


The following table illustrates a problem for spelling reformers trying to ameliorate the difficulties of dialect in a revised orthography.

Gen Am

The pronunciation of the cloth, lot vowels has been subject to a lot of change throughout their development, explaining why they have become fixed differently in different dialects as these dialects developed at different speeds.

The RP vowel in lot can vary in Britain from /ɒ/ to /ɔ/, but in Southern Ireland, Canada and most of the USA it will have the /ɑ/ vowel of General American, making the vowel in pot and part homophonic.

Many non-RP speakers will perceive the vowel in thought as identical to that in cloth despite its different historical development. The thought vowel has several origins, although it was usually Middle English /a ʋ/ and /ɔʋ/, with spellings <augh, ough, au, aw, al> resulting. These vowels merged before /xt/ as in words like daughter and thought where the <gh> was traditionally pronounced /x/, the sound spoken in present day Scottish loch.

Middle English /a ʋ/ has not always developed to /ɔ/. It has also become the vowel in words like laugh, or in words like palm. Also, some RP speakers will use /ɒ/, not /ɑ/, for words like halt, salt, alter, fault.

In order to find a solution to the orthographic problem posed, one might propose that all the vowels in palm, lot, cloth, thought be assigned a single grapheme. If General American can do without the distinction which RP speakers make between palm and lot words, it probably means that the number of minimal pairs is within the practical range for allowing a merger between the two. (I have not been able to check the minimal pairs involved as I do not possess an American pronouncing dictionary.) However, I suspect that there are many minimal pairs between palm and cloth or thought vowels, leaving this difference between RP and General American seemingly irreconcilable in a revised orthography.

There are only a few minimal pairs illustrating the difference between the cloth:thought vowels, god:gaud, cot: caught, tot:taut/taught being all I could find, this indicating that these vowels could combine easily under the one grapheme. This phonic simplification would solve part of the problem while leaving most American and some other dialects requiring to memorize the spellings of the lot type words, without alphabetic help from their own pronunciations.


Usually, weak unstressed vowels occur within words but the vowels at the end of letter, comma and happy are weak. The first two of these will be discussed in a future article, but the happy vowel is worth looking at on its own. The approximate pronunciations in common use can be tabulated as follows:

/ɩ/ most RP, conservative varieties of Gen Am, centre of the north of England (Manchester, Leeds), Jamaica and the American south.
/ɛ/ Nottingham and certain varieties of RP (particularly associated with army officers).
/i,/ some Americans (/i/ before a vowel, // elsewhere).
/i/ much of the south of England, the peripheral north of England (Liverpool, Newcastle, Hull, Birmingham), Gen Am and Canada.
/e/ some Scottish.

The happy vowel retains its identity despite various possible pronunciations, not only among different dialects but in literature and song, where different poets at different times will rhyme happy with tree or troy. Perhaps, as it is the only non-stressed vowel found at the end of words, apart from schwa, it has no phonetic opposition. That is, there are no minimal pairs illustrating its opposition to another vowel, and this allows it more leeway.

It is identified not only by its lack of stress in a final position, but by the usual <y> spelling, and the meaning of many words in which it is found. In the lexical set of words with the happy vowel, there are various subsets which seem to belong together - pet words, nicknames, familiar or informal terms, and words expressing common human emotions and states. For example: baby, lassie, dolly, budgie, piggy, doggy, pussy; Tommy, Johnny, Willie, Maggie, Sally; darky, Paki, lefty, commie, goody, baddy; happy, lucky, angry, silly, funny, fussy, bossy, edgy, tetchy, sexy, dopey.

At any rate, the happy vowel seems to be identifiable despite varying pronunciations, and it would seem unnecessary and complicating to impose one of these pronunciations on everyone in a revised orthography. It would surely be better for the happy vowel to be assigned the presently predominating <y> and for other spellings to be altered to this. This would not preclude the letter <y> being used for other phonemes where its environment will indicate its pronunciation, for example, yellow (<y> is a consonant before a vowel) and sty (the <y> is a stressed vowel).


J C Wells addressed the Society in January 1986 on the problems of accommodating different accents in spelling reform proposals. This article was in draft form at the time and, perhaps not surprisingly given the heavy reliance on his book for my material, my analysis and conclusions were very similar to his. In the summary list below, only my analysis of the yod vowel and the happy vowel are additional to the content of his address. The rest of the recommendations for the combination of pairs of phonemes being represented by a single grapheme are his as well has mine.

To most spelling reformers, the content of this and my previous article will be somewhat radical. However, most of it is in agreement with the analysis of one of the world's foremost authorities on English accents, and reformers would be wise not to reject my conclusions without a great deal of thought.

My next article continues the same theme, looking at a great divide between different accents of English, namely rhotic versus non-rhotic accents, that is, those which pronounce /r/ before consonants and those which do not, and the effect on vowels in the /r/ or /r/-less environment.

/hw, w/
/æ, ɑ:/
/ʋ, u: ju:/
/ɒ, ɔ, ɔ:/
whaling, wailing
ant, aunt
full, fool, feud
cloth, thought, lot (RP)
Suggested graphemes
<u >
<uCe, uCV, uV> or <oo>

Spelling Innovation in the Service of Advertising

"Are you going to join the flock at the baaah tonight? Kronenbourg 1664. A slightly non-konformist lager.

"0 for U" "Neau colouring. Neau sweetening. Neau artificial flavouring."

Back to the top.