Why not Spelling Reform?

Back to Spelling Reform Page

Some answers to common objections to reformers.

A quick note on symbols

Normally when talking about how people speak we use the IPA alphabet but this involves special fonts that old browsers (like mine) can't handle. Here I use the ASKII equivalent: SAMPA.

English spelling is so eccentric in its use of letters to represent sounds that to label the sounds by a letter, or a group of letters, is confusing. Hence it is essential to hav some way of making it clear that one is talking about the letters written or the sounds spoken. Fortunately linguistics has a notation to cover this. I'll explain logic behind these later but for now:

<ight> means the actual letters written.

/aIt/ means the basic sound that we use in words like night. I say basic because linguists would use these brackets [] to specify the exact sound so [?] is the glottal stop that you get when in London a /t/ is "dropped". I will explain more fully why this distinction is important later on.

There is no point in reform because the way we speak is always changing.

Yes, the way we speak is always changing but the process is quite slow. Serbo-Croat adopted a reformed spelling in 1850 and there has been no need for significant changes since then. Spain is currently undertaking a reform but it involves simply taking into account that two letters that were pronounced differently in the past are now spoken the same. Once a language has a reformed alphabet it only requires quite minor changes to keep it regular.

The idea that pronunciation changes too quickly for spelling reform to keep up comes from the myth that English got into the mess it is because of neglect. But English did not get into the mess it is today as a result of the natural change in the way the people spoke. England in 1065 had a near perfect sound based spelling system known as the West Saxon Standard but what occurred next was not neglect but a series of linguistic disasters. The first disaster was the Battle of Hastings. This meant that for a couple of centuries the scribes spent most of their time writing in Norman French which unlike Saxon had a highly illogical spelling system. These scribes tended to write English in a French way but note - only tended. The old Saxon conventions still survived to an extent so that English spelling was now a rather unpredictable mixture of the two conventions.

Hastings was a defeat for a phonemic English alphabet as well as King Harold.

But things were to get worse. The fact that u and v looked the same meant that the convention came about that where the sound was /uv/ then <ov> should be written - hence words like love. It is sometimes suggested that this is how words like <come> came to be spelt that way but, according to Scragg. (A history of English Spelling, D G Scragg p49) this almost certainly originated in Latin spelling and crept into the English spelling system as a result of the Norman conquest. Then English spelling fell into the hands of the etymologists. In a fad that had its height in the 16th Century a whole series of words that were originally spelt as they were pronounced in English came to be spelt as they had been in Ancient Rome or Athens.

The fad for etymology began to ebb by the end of the sixteenth century. By then, however, printing had come and as the printers were a conservativ lot spelling began to be fixed in its most chaotic form. But printers did not simply provide a bulwark against any reform. They also added a few inconsistencies of their own. The early printers were often Dutch or had, like Caxton, learned their trade in the Netherlands and they often introduced Dutch spelling conventions into English words. We must personally blame Caxton for the <h> in ghost which had been spelt gost until then (Scragg p66).

Changes in pronunciation must take some of the blame for the mess we are in. The magic e convention has its origin in the way a whole series of two syllable words lost their final e. As they lost the final e, the vowel in the first syllable become long. Change has continued since the printers fixed English spelling. The <k> in knight was pronounced in the seventeenth century. However there has been a compensating trend whereby the power of spelling has led people to assume that the spelling must indicate the true pronunciation. Hence words like corpse and falcon were original spelt (and pronounced) cors and faucon. The modern spellings were originally etymological spellings at odds with the way people spoke. This kind of back to front reform has even worked on words where the etymology was quite false. The <a> of the Middle English word amyrel was assumed to be a truncated form of the Latin prefix ad- so producing admiral. The word in fact comes from the Arabic Amir (ie Emir) but now pronunciation has caught up with the spelling.

The big problem is that the combination of conquest, 16th century pedants and foreign printers has so wrecked English spelling that English people do not expect English to hav much relation to sound. To a Spaniard the idea that alphabets are supposed to represent sound is so obvious that any divergence from spelling and pronunciation and spelling seems an outrage. Hence minor divergences are spur to reform for a Spaniard. To the English the chaos seems so normal that change seems utopian.

We need to keep the spelling to preserv history.

If you want to know the origins of a word check in a dictionary because the spelling is very unreliable. Two of the main ways of spelling a long e come directly from the Saxon-Norman split. The <ie> spelling is french and the <ee> spelling is in origin West Saxon. Strictly speaking the scribes of Saxon Winchester did not use <ee> but a simple <e>. This was because a long e was in those days simply a short e (Nowdays it iz pronounced like a lengthened short i and the vowel we call a long i is in fact a diphthongs.) Only later did words with a long rather than a short e acquire the second <e>. When you see <ee> for long e it is likely that the word is Saxon but by no means always. Fee and feeble are in fact from the French.

Likewise the tendency that words with <ie> are French has exceptions. Thief and fiend are in fact Saxon. According to Scragg (p49) they were spelt with and <eo> as theof and feond. And just to muddy the waters further this convention survives in a word which is French to the core, that of people. If you're on the ball you'll be saying "but hang on didn't the Saxons use a simple <e> for long e?". The scribes of Winchester did indeed use <e> for long e and when Wessex was victorious their conventions, the West Saxon Standard, became universal throughout England. In the West Saxon Standard <eo> was a diphthong gliding from e to u. However before Wessex had conquered all England there were other conventions for other dialects and in Anglian texts <eo> was used for long e. (The English Language, A Historical Introduction: Charles Barber. p 108).

Originally many borrowed words had their spelling adapted to English spelling. The reason that English spelling often is preserves the spelling of the original language is down to the group of sixteenth century scholars obsessed with the etymology. They tried to track down the origin of words and then changed the spelling to conform to the original spelling. Trouble is, they often got it wrong. The <c>s in scissors and scythe are there because these sixteenth century etymologists wrongly believed these words came from scindere (to cut). In fact scissors comes from the Latin cisorium (cutting instrument) and scythe isn't even Latin - it is an Old English word. The word isle was a borrowing from the French with the spelling (and pronunciation) of ile. When the etymologists realized it originated from the Latin insula and so inserted an <s> to isle they did the same for island despite this being from old English. The Old English origins can be seen in names like Sheppey (Sheep island) and so the origin is the Old English for island plus land. (Scragg p 57 )

You can't hav an alphabet based on sounds because English has too many accents

Linguists who study dialects use an alphabet that is truely phonetic in that it shows the smallest differences in people's pronunciation. No language however attempts to use such phonetic system as it's writing system. What "say-write" languages attempt is to describe all the sounds that are needed to avoid confusion of meaning. They ignore differences in pronunciation that do not affect meaning. When /b/ begins a word it is different in sound from when ends a word. For people using a language this difference is of no importance and for that matter they probably don't realize it exists. To cover this linguists use the term phoneme. Hence we say the consonant that begins a word like bed is the same phoneme as the consonant that ends a word like cab. The phoneme is the smallest distinction in sound that conveys meaning.

For example, in London you will hear a /t/ between two vowels as a glottal stop. That is <butter> will be pronounced as bu'er or in SAMPA as [bV?@]. I use square brackets here to indicate that I'm referring to the exact pronunciation for good reason. When we hear a Londoner use a glottal stop in this way we know they mean a /t/. Even tho a glottal stop has a very different sound it plays the same role to convey meaning. Hence we can say the glottal stop of a Cockney is the same phoneme as /t/ and we write what a Cockney says when he means butter as /bVt@/. Crucially, a Cockney kid when learning to read is unlikely to hav much problem with spelling if butter with a <t>. All that he needs to understand is that when he says [?] he writes <t>.

Now I hope you see why the difference between using square brackets and slashes is crucial. When we write something between slashes we are only concerned with meaning but square brackets implies we are concerned with the exact sound. Hence /bVt@/ can equally represent the spoken sound of [bVt@] or [bV?@]. Of course phonemes are language specific. A speaker of Arabic will find the idea that [?] and [t] are the same quite strange because in Arabic the glottal stop is a phoneme in its own right with its own letter.

Another example the old spelling of ship was scip because the Saxons originally pronounced in the same way as we do the word skip. However quite early on all the /sk/ sound changed to an sh (/S/) sound. This created no problem because to the Saxons of that time /S/ was just a way of saying /sk/. Hence to a Saxon reader it was perfectly logical to write it as sc and indeed they were probably convinced that they said s + k when they said [S]. Unfortunately then the Vikings overran a large part of England. The Vikings did still say sk as s+k and so a host of words with a sk sound got introduced into the language that probably included the modern word skip. There now was a problem in that if you read scip you had to check the context to be sure if the writer meant a vessel that floated on the sea or a short jump. In the short term this was not so great a problem because educated scribes regarded the English of Winchester as the only dialect worth considering. They would never hav dreamt of using uncouth words like skip. Only when [sk] words became respectable did it become necessary to start writing ship in the way we do today. (The fact that /sk/ was originally two phonemes and was then reduced to one, muddies the water a little but in other respects the principle remains the same.)

In modern English the sh sound, that is /S/, is a phoneme. In pre Viking England /S/ was not a phoneme. If you said you a saw a [skIp] people might find your accent a little archaic but they wouldn't hav any doubt what you meant. In short to say ship as [skIp] rather than [SIp] was a phonetic difference but not a phonemic difference.

It is possible to come up with different groups of phonemes. The t in trap is a quite distinct from the t in tap and indeed is quite close to the /tS/ of church. Because this distinction is purely phonetic we hav no problem with the spelling trap. To spell these words with a ch would however be equally correct. When linguists first came up with the idea of the phoneme there was some debate as to whether the [h] of hat was the same phoneme as the [N], that is the of sang. This was because in English you never find [h] at the end of a word nor [N] at the beginning of a word. They balked at this even tho it would be quite logical to hav done so. The basic test of whether a distinction is sound is significant enough to create separate phonemes is whether you can find a 'minimal pair'. In Southern Anglo-English they say the word but as [bVt] but the word put as [pUt]. Is this distinction a phonemic difference? Yes because this distinction is what separates the two words put and putt. Put and putt are then a minimal pair that shows that /V/ and /U/ are separate phonemes.

In an ideal writing system every phoneme would hav a letter. In a way when linguists try to find the phonemes of a language they are in effect say "What is the smallest number of letters we can hav for the alphabet of this language and ensure that no two words are spelt the same if they hav a different sound."

Many of the differences of pronunciation that make up accents are phonemically neutral. People may differ in the way they say a phoneme but if they do so consistently then there is no problem. Unfortunately this is not always the case but the phonemic differences between accents are a great deal smaller than the phonetic differences.

I'v already mentioned the Southern Anglo-English distinction between /bVt/ and /pUt/. In Northern Anglo-English they are both pronounced with a [U]. Hence put from putt are pronounce the same way. Clearly this is a problem for a phonemic alphabet. The spelling system can reflect Northern pronunciation and use the same letter for /V/ and /U/ or it can use two letters and reflect Southern pronunciation but not both

There are a number of ways round this.

One is to pick on once accent and make that the standard. This is the option taken by lots of reform schemes because it is very easy to do. If you base your reform on either standard American or the English RP then the bulk of the work of devising a reformed alphabet has already been done for you. It is also the principle behind the West Saxon Standard which was based on the speech of Winchester of the time.

Another option is to allow differences in spelling. This was the option adopted in Serbo-Croat but the difference between Serbian pronunciation and Croatian was no more than one phoneme. It would be unworkable for English.

Another is to let the current spelling be the guide where there is a large minority whose accent is best described by the old spelling. On this basis southerners would continue to use u for both /U/ and /V/ but they would be no worse off than they are at the moment so this is unlikely to create great problems.

Finally there is option of taking everybody's accent into account and patching together a compromise that allows everyone to get their way some of the time. This would take a lot of work to devise, would produce an alphabet that didn't conform exactly to anyone's pronunciation but would nonetheless be a massiv improvement for everyone.

But really this all misses the point. English spelling is so hopeless at representing phonemes that a reform based on virtually any accent would be an improvement for everyone else. There are, further, a vast number of reforms that would be accent neutral.

For instance words like psychology are pronounced by no one as /psychology/. If we were to spell the <ps> as <s> and the <ch> as <k>, we'd get sykology which is closer to everyone's pronunciation. Then there are words with <ph> like photograph and, for that mater, phoneme where everyone says /f/.

Even with words that are pronounced differently in different accents improvements for all are possible. Scottish Vernacular English often has a short vowel in words like bone and make. In most English accents the final e is needed to show the vowel as long. If we were to replace the magic e rule by two letter combinations so that bone was spelt boen this would be equally illogical for the Scots. However where the magic e rule causes problems is in spelling. Words where the vowel is short must sometimes be doubled when vowel endings are added (such as ing etc) if they are not to become long. If we reformed the magic e out of existence then we could do away with all double letters and this would benefit Scots along with everyone else.

What causes reading failure?

The earliest theory was that poor readers had a visual problem - hence the term for dyslexia "word blindness". This theory bit the dust when it was a number of experiments showed that poor readers were no worse at visual tasks. For instance one experiment with children of the same age and intelligence but different reading ability showed that the poor readers were just as good at copying Hebrew letters and even English letters. They were also just as good at remembering the visual patterns of Hebrew letters. (Children's Reading Problems: Bryant and Bradley p27 The original work was by Vellutino). There is also no evidence that dyslexia is caused by a problem with reversing letters. Poor readers hav no greater a proportion of reversal errors than good readers (Bryant and Bradley p24). That letter reversal is an indicator of dyslexia is probably simply that letter reversal is more noticeable amongst poor readers because the total number of errors is greater.

There are however a number of experiments that seem to show that poor readers hav various deficits, especially in verbal ability. To explain why these do not show what they appear I hav to explain something about experiment design. Most experiments take two groups of children of the same average and intelligence but differing in reading ability. These are the so called "age matches". The problem with this kind of experiment is that if you show that the poor readers hav poor verbal ability you hav to ask which caused which. It could be that the poor verbal ability caused the poor reading. But reading opens up a whole world of verbal experience which could hav expanded the verbal ability of the good readers. Hence showing a connection with an age match proves nothing. All age matches are good for is excluding possibilities. If an age match shows that that poor readers hav no weakness in visual ability then we can be sure that is not the problem. It must be something else that slows their learning to read. An age match is like the first hurdle that a theory on the causes of poor reading must cross.

If we hav time, we can try longitudinal studies. Here we measure an ability and see if it predicts another ability some time in the future. For instance Bryant and Bradley took a group of four and five year olds and tested them for their ability to remember words. If this ability helped the children to read then this should hav helped predict which children would be the good readers when the children were tested later. That is to say, when they tested the children a year and a half later those who had had good memories at the first stage should hav been the best readers. They were not. That Bryant and Bradley found no such link suggests that the age match experiments that found a link between reading and memory for words had got the cause and effect the wrong way round. They then tested the children two and a half years later still. They found that the children's reading ability at the earlier stage did predict their memory for words at the final stage. So it was not the children's ability to remember words that caused their success in reading as had been assumed but their experience of reading that improved their memory. (Children's Reading Problems: Bryant and Bradley p36).

If we don't hav the time to do a longitudinal study there is an alternativ to the age match. It is the reading level match. The reading level match is the mirror image of the age match. Hence it is good for ruling in possible causes of reading ability rather than ruling them out as age match test do. We take a group of poor readers and compare them with a group of younger children who are matched in reading ability. That is to say if the poor readers are ten years old but hav the reading ability of normal eight year olds we compare them with a group of normal eight year olds. They hav the same reading ability so any weakness in ability can't be caused by a difference in reading ability. However the younger children will be less intelligent because of their younger age even if they hav the same IQ level. Hence the poor readers may cover up a real deficit in ability by using their superior intelligence to compensate. One might speculate they are more likely to understand the instructions and so know better what was expected of them for instance. Hence the odds are in this case stacked in favor of the poor readers and if we find no difference in ability this proves nothing. On the other hand if despite their advantages the poor readers are shown to be worse at a task we hav found something important.

Reading match experiments hav shown that poor readers are poor in phonological skills. They are bad at detecting rhymes and the onset (ie the initial consonant cluster). They are also poor at reading nonsense words like "slosbon" that can only be read by interpreting the sound from the letters. (Phonological Skills and Learning to Read, Usha Goswami and Peter Bryant pp84-87). There are strong grounds then poor readers are let down by their weakness in perceiving the sounds of English. It is probably the main cause of poor reading.

If poor readers hav problems with phonemes won't a phonemic alphabet make things worse?

Perception of phonemes can be learnt. There has been a number of experiments comparing illiterates with literates and readers of a logographic script with readers of an alphabetic script. These suggest the experience of learning to read an alphabetic system improves phonological abilities. The snag is that with English spelling the phonemic code is hidden by fog of silent letters and the host of different spellings for each phoneme. The able children manage. Those children whose phonological ability is not too hot are however completely defeated and giv up. Had English spelling provided less high a hurdle many would hav managed.

Experiments on teaching methods tell the same story. Phonics methods hav been shown to be superior to whole word methods of teaching reading (pp118-20, Goswani and Bryant) but that in itself not conclusiv. A phonic method does not simply teach phonemes in isolation and it is possible that other parts of the phonic method are what helping children. Quite naturally, the phonic method teaches the phonemes in relation to the alphabet and it could be the resulting awareness of the alphabet that really helps. To be sure that teaching the sounds helps children we must teach children phonemes in isolation. Experiments show that when teach children about phonemes in isolation from any training with the alphabet their ability to learn to read is indeed helped (pp121-24, Goswani and Bryant). The first words that children learn often highly irregular. Is, come, one, have, said, you and was would in a phonemic system be spelt iz, cum, wun, hav, sed, yoo and woz. Because these are frequent words they are usually learnt without great problems, but they are poor training for perceiving phonemes. Not surprisingly people whose language is English hav as a result a poor sense of phonemes. Many manage to read fluently despite this handicap but many others do not.

You may notice that part of my argument is based on research that focused on rhymes rather than phonemes. It could be argued that all that is needed is to solv the problem is to ensure that English consistently spells rhymes. I don't see that as a very strong objection because the easiest way to ensure that all rhymes are consistent is to make English spelling phonemic.

But enough of theory. The real clincher for the advantages of a reformed spelling is i.t.a..

Hang on, i.t.a. failed didn't it?

It is a myth that i.t.a. was a failure. In many ways it was a dramatic success. Children taught using i.t.a. remained ahead of children taught from the beginning on TO even after the i.t.a. children had made the transition to TO. (G Sampson, Writing Systems p 196 who quotes Warburton and Southgate 69: i.t.a. and independent evaluation). i.t.a. avoided the problem of TO in that books for beginners must be graded to avoid confusing spellings. i.t.a. readers by contrast had far greater freedom in what they could read at a very early stage. (i.t.a. and the teaching of literacy: John Sceats 1967 p121)

And then most striking of all was the effect on the children. I include here some quotes from John Sceats book from heads of schools that experienced i.t.a..

"You do not get the frustrated infants you used to get."

"children are more reliant"

"The main effect of i.t.a. has been to restore children's confidence..."

In this school i.t.a. was started as a means of dealing with a reading problem where it was estimated that 42 per cent of the children were retarded in reading. Children from non-cultured homes hav been liberated by a simple method of learning to read..."

"If only they could keep to i.t.a. it would be marvelous; you see t.o. inhibits them."

"We found that with i.t.a., children were inclined to get a general interest in books; they read more and when they were reading it was very hard to distract them, they became engrossed in the book..."

Because it proved effectiv the use of i.t.a. spread till by 1975 10% of schools were using it yet seven years later it was used in only 280 schools (Sampson p196). What went wrong?

First i.t.a. did not solv all the problems of literacy. Some children still failed to get the hang of TO. But now they would be more visible as the children who still hung onto i.t.a. so i.t.a. got the blame. And the transition was difficult to manage. It is tricky when half the class has switched to TO and half are still using i.t.a. to know which to use when writing on the blackboard. Further i.t.a. only really comes into it's own if based on a phonic method of teaching. Children taught using whole word methods which tend to encourage children to memorize the sequence of letters in individual words got less benefit from i.t.a. but still had the problem of switching to TO. And heads and teachers move. When those teachers who were committed to i.t.a. were replaced, new teachers could easily switch back to TO by brushing the dust off the old TO books. The big problem for teachers committed to i.t.a. was money. The basic text books cost and the expense was even greater if the school library was to carry a fair number of more general books.

Oh yes, 1975. The year that i.t.a. peaked was the year before Britain got into the clutches of the IMF. In 1976 Britain faced a catastrophic run on the pound. A loan from the International Monetary Fund seemed the only solution but there was a price. The IMF insisted on brutal cuts in government spending and local government spending was hit especially hard. One can imagine that in those days education committees took a very dim view of teachers who wished to fork out a small fortune on books for some new fangled teaching scheme.

OK so a reformed spellings will help learners but won't it make it harder for adult readers?

This is an argument that you most often meet from linguists and the most eloquent statement of it that I know of is that of Geoffrey Sampson in Writing Systems. Sampson is certainly on strong ground when he argues that the fluent reader does not decode the letters of a word to get the sound and then use that sound to retriev the meaning. What probably happens is that the reader accesses the meaning of the word directly from the letter sequence. This is why reformed alphabets are quite jarring because you hav to think about the sound of what you are reading and this is briefly slower. Sampson further argues that a less phonemic system allows words to hav a greater variety in letter sequences so ensuring that words are less easily confused.

Sampson is prepared to concede that for beginners a more phonemic script would probably help. He also concedes that writers would benefit from a phonemic script. What he argues is that the interests of adult fluent readers are more important. After all in the age of printing most people read more than they write. Further beginners will eventually become fluent learners so if we help beginners but at the cost of making it harder for fluent readers we harming those beginners long term interests. Given that people now liv longer the part of peoples lives when they are fluent readers is relatively much longer. Hence it is worth the extra effort of learning a writing system like English with loads of silent letters because later in life those same silent letters will help them read more quickly.

The argument rests on the assumption that all, or nearly all, beginners progress to be fluent readers. This is not the case. In English speaking countries those whose literacy is so poor that they often fail to read the most basic signs is around 10%. Further there is an even larger group who read so laboriously that they only read if they hav no choice. The complexities of English spelling are such a high hurdle that many don't pass into the promised land of fluent reading.

But it is disputable that silent letters really provide any benefit for fluent readers. Sampson givs examples where children found learning individual Chinese characters easier that leaning the English words (p163). This, they argued was down to the distinctiv look of Chinese characters. However this kind of learning strategy does not work well with English. We can see this in that phonic methods are more effectiv than whole word methods. English words, it seems, lacks the distinctiv look that allows an effectiv learning strategy based on the idea that English words can be treated almost like Chinese characters. But this of course only applies to learners. Sampson provides no evidence that, to take an example that Sampson uses, the <b> in debt helps fluent readers distinguish that word from bet. It is rare that a fluent reader will hav any noticeable hesitation when reading the word live in which the single spelling disguises different meanings and pronunciations. It is extremely unlikely that there is any significant advantage to fluent readers in that the letter sequence of debt is more sharply distinguished from the word bet by the extra <b>.

Of course the transition to reformed alphabet will cause fluent readers short term problems. They will hav to go thru a period in which they get used to the new spelling during which they will hav to pay attention to the sound-letter code. This is likely to be fairly short for most people but nonetheless they would gain little long term benefit. Spelling reform benefits learners and unfortunately these tend to be children with less political clout. The different needs of learners as opposed to fluent readers may not be a serious objection to the value of spelling reform. This difference is, however, a serious political problem.

What will spelling reform mean?

English spelling as it is leads to reading failure for two reasons. The ability to distinguish phonemes is an important part of learning to read. It is this ability that is undermined by the many words, including some of the most frequent in which the letters do not represent the expected phoneme. Further the very complexity of the code means that many children get so browned off by the experience that, even when mastered, reading is something they do only when they hav to. A simplified spelling system would undoubtedly reduce these problems. However it would be misleading to claim that there would be no problems. Even if everyone learns quicker there will always be a minority who need special help because they take to reading less quickly than the other pupils. The reason why spelling reform is important is to ensure that fewer children fail to become fluent adult readers. That will not happen if spelling reform is used as an excuse for reducing the time devoted to teaching children to read. If reading failure is due to a weakness in phonic ability then making English more phonemic will help those whose phonic abilities is merely below average. There will still be a smaller group whose phonic abilities are so poor that even a reformed alphabet will be a real struggle.