[Journal of the Simplified Spelling Society 1987/1 pp27,28. Later designated Journal 4]
Escaping from a Dialect Straitjacket 1.
David Stark.This is the fourth in a series of articles of which the first three appeared in the Simplified Spelling Society Newsletter News8 Summer 1985, J2 Spring 1986, and J3 Summer 1986 issues.
The story so far: Phonics is useful when learning to read as it allows one to determine the pronunciation of the occasional unknown word in an effort to identify it and learn it with the minimum of help from others. The target however is quick visual recognition of all words met.
Phonics is useful in spelling as it affords an additional, easier way of remembering spellings than purely by memorizing the visual patterns of words. One has to remember the Standardized Spelling Pronunciation (SSP) for each word, which may or may not fully equate with the perceived pronunciation in one's own accent.
As it is impractical to fix a living accent for use as the pronunciation standard for spelling, the regularized spelling of each word must be used for this purpose. However, once one has determined an SSP from the spelling, this should approximate to a living pronunciation likely to be heard. The comparison of a word's SSP and its actual pronunciation helps reinforce the SSP in one's memory.
In the introduction to New Spelling, it is stated that "anyone who knows the pronunciation of a word should be able to spell it". It is not said how people throughout the English-speaking world would know this pronunciation, nor how practical it was for ordinary people without phonetic training to define every single phoneme in every word met. In the same way, many reformers seem to think it enough to describe their reform proposals on a single page, by matching a proposed set of graphemes with an assumed set of phonemes. The relationship of this to the lexicon is usually based on their own accent.
Just before World War 2, when New Spelling was being formulated, Received Pronunciation was the prestige accent in Britain, all others being regarded as inferior, at least in well-educated circles, with the possible exception of the cultured Scottish accent. So dominant was RP then that representatives of the Simpler Spelling Association in America concurred with slightly amended New Spelling proposals in 1956.
New Spelling did try to allow for accents other than RP, recognizing differences between rhotic and non-rhotic accents, and some different vowel-sounds in Scottish and other non-Southern English accents. However, the limited knowledge of dialects then did not allow a comprehensive study of accent problems and the development of an appropriate reform-policy. Since the status of RP has declined, and no dialect is, or looks likely to be, in common currency throughout the English speaking world, it would be wrong to base SSP too closely on one accent.
Some phoneme-grapheme relationships will be the same and uncontroversial for all major English accents. There will also be some phonemes which, although different in two separate accents, will translate directly to a single grapheme. For example, the vowel in the word dress is /e/ in RP and /ε/ in General American. It does not matter that the vowel quality differs in the two accents, one merely defines both as represented by the grapheme <e>.
However, there are other situations where there is no direct translation, as the following table illustrates.
In such cases, standard spelling will be unable to accommodate all accent variations. Assuming that alternative spellings of words are unacceptable, some people will have to learn SSPs which do not correspond with their own pronunciation standard. Alternatively, they will need to memorize spellings visually.
There are opportunities to reduce the number of phonemes represented by the orthography. We have only 5 vowel letters to represent some 13-19 vowel sounds, the number depending on the accent. If we could reduce the number of vowel phonemes represented by the orthography towards the bottom of this range, there would be less need to invent new vowel symbols to represent them, and those with fewer vowels in their accent would be better able to relate to the orthography.
Consonants too could be usefully simplified when represented in writing. For example, many English accents have both /w, hw/ phonemes, which relate fairly regularly to the graphemes <w, wh> in t.o. The latter phoneme has disappeared from RP, and RP-speakers consequently need to remember individual <wh> spellings independently of pronunciation. However, if <wh> merged with <w>, RP speakers could spell relevant words phonetically, accent adherents with both phonemes would write both as <w>, and existing literates caught in the transition period when revised spellings are being introduced could simply change <wh> to <w> without recourse to pronunciation.
This example in RP may be described as a phoneme convergence, where two phonemes become one. The single phoneme representation is more useful to more people in an alphabetic orthography. It also reduces the number of phonemes to be learned, or identified, by foreigners learning English, who may have fewer phonemes in their native orthography than there are in most English accents.
The opposite of phoneme convergence is a phoneme split, with one sound becoming two. If the single sound is still widespread among some common English accents, it would be better to combine the two possible phonemes for representation by one grapheme, for the same practical orthographic reasons as with phoneme convergence.
For example, the vowels in the words lass and pass are different in RP. The sound split from a previously single vowel did not occur in General American or in many other dialects, and even in RP the sound change was incomplete, with some words pronounced with short /æ/, as in lass, and others with lengthened /α:/ as in pass, and no phonemic rule to split them. Also in some accents where the split has occurred, it has taken place in different ways to RP. For example, many Australians use the shorter vowel whenever /n/ or /m/ follows. Other Australians, West Indians, New Zealanders and South Africans always use the longer vowel. Scottish and Northern Irish accents always use the shorter one. One may conclude it would be easier for all to use the one grapheme for both phonemes.
Whether associated with phoneme convergence or split, there are good practical reasons for reducing the number of phonemes represented in the standard orthography. These two factors are the result of accent development, but there is a third, only partly related to dialect, which I would like to consider under the same criteria. This is "context related phoneme convergence", where pairs of phonemes become indistinguishable when associated with certain other phonemes, or where pairs of phonemes become interchangeable depending on whether they are stressed or not. Schwa may be judged to be in this category, but to do it justice would require a separate article. Instead, I shall use as my example the phoneme pairs /f, v/, and /s, z/.
Formerly /f, v/ were not distinguished in English orthography and sometimes now it is difficult to separate the two in everyday speech. Luckily this seldom causes confusion in the context of speech, but when we try to formalize them in spelling, we often find ourselves in difficulty. When we add an <s> to a word ending in <f>, does the sound change to /v/? Traditional orthography is not sure, as it spells the plural of thief as thieves but the plural of chief is chiefs.
Where <f> is retained, the plural marker is pronounced /s/, but if one puts sufficient stress on the end of the word thieves to fully realize the <v> sound, the plural marker is pronounced /z/. If the difference between such pairs of sounds depends on their environment or the amount of stress they receive, are they really opposing phonemes, or merely versions of single phonemes? If the distinction between such pairs is often difficult and leads to spelling confusion, perhaps we should consider fusing them into single graphemes in an orthography.
Traditional orthography once had a separate grapheme for the /s/ phoneme, but dropped it, probably because its function was less important than the visual confusion caused between the upright <s> grapheme and <f>. Today the <s> grapheme is the most common way of representing both the /s, z/ phonemes, so accommodating the two phonemes possible with the <-s> inflection, and allowing alternative stress options to coexist without spelling alternatives, as in adv'ertisment: advert'izement.
In any language, a phonetician will be able to show the existence of more phonemes in most accents than are represented exclusively in the standard orthography. This will partly depend on the definition of a phoneme, but such a person will usually be interested in more accuracy than we would need, or be able to practically accommodate, in everyday orthographic use.
However, if we pursue the approach of simplifying and reducing phonemes represented in a revised orthography to its ultimate conclusion, we may eventually halve (half?) the number of phonemes that many people recognize in normal speech. Apart from the major change which this would cause in traditional orthography, it would also throw up a large number of pairs of words which are spelled the same but which have different meanings, and which sound different to many people. The same problem would be encountered as we have with the translation of homophones from traditional orthography to a revised orthography. The one spelling would now embrace two or more words which were previously spelled differently, thus making reading less efficient for experienced readers.
In considering whether we can allow two phonemes to be represented by one grapheme, we need to identify pairs of words which rely solely on the difference afforded by these opposing phonemes. These are called "minimal pairs". If there is a large number of minimal pairs relying on such a distinction, and sufficient of these would cause semantic confusion if revised, we have probably passed the bounds of practicality, and may have to reject the merger.
For example, in considering whether to combine the /w, hw/ phonemes into one grapheme <w>, we can cite the minimal pairs while: wile, Wales: whales, wailing: whaling, etc. These are homophones to RP speakers, but the distinction between the pairs is useful in many other accents and to readers of traditional orthography, and we must try to quantify the consequent loss there would be in a revised orthography. The value of such a phoneme distinction in the lexicon is called its "functional load".
Since a major accent like RP can do without the /w, hw/ distinction, it seems that the functional load of these phonemes is small enough to allow the merger to be translated into a revised orthography. A design decision can be made, in effect lubricating the sluggish evolutionary process of t.o., which in modem times has practically halted. In the past, when the orthography was more subject to change, the evolutionary process decided that, despite having two graphemes available for /ð/ and /ð/ one grapheme should be used for the two. There are few enough minimal pairs like thigh: thy to demonstrate a low enough functional load for this situation to have arisen.
Let me summarize. In order that a revised orthography is not trapped into a dialect straightjacket, we must look at ways of accommodating various features of different accents in the English speaking world. This would allow the alphabetic advantages of the orthography to be enjoyed by the greatest number of people. In searching for common denominators, one of these strategies is to allow pairs of potentially confusing phonemes to be allocated single graphemes, so simplifying the phonemes represented in a revised orthography. We will be guided and limited in this by consideration of the functional load of the phonemes involved, as illustrated by the relevant minimal pairs.
I am now able to investigate in more depth common differences between accents, and how to rationalize them in a revised orthography. This is my next task.
Back to the top.