[Spelling Progress Bulletin Summer 1980 p7-13]
[John Beech: see Journal, Newsletter, Bulletins.]
Some Proposed Principles for Simplifying English Orthography,
by John R. Beech, Ph.D.**Dept. of Psychology, New Univ. of Ulster, Coleraine, N. Ireland.
Presented at the 1979 Conf. on Reading & Spelling, Nene College.
The English language is exceedingly difficult to learn to read because of a spelling structure which is very complicated and which is not altogether consistent. Consequently, when a child trys to learn to read, it would be a grave mistake for him to draw generalizations and rules about how a symbol might be pronounced, for before too long he is likely to encounter another word which contradicts the rules. To save space, I will not go any further into the reasons why we should reform our spelling system. Instead, I will begin immediately with three main criteria which should be considered in adopting a new spelling system.
Criteria for Reform.1. Obviously the main criteria is that it should enable children to learn to read much faster and more easily. This also implies that most children who would not have been able, to read the traditional system should now be able to read the new system. This criterion is probably fulfilled by most suggested spelling reform schemes and is the easiest to satisfy.
2. The next criterion is very important from the point of view of achieving reform: would the public be able to read the new system easily? The agonizing aspect of this point is that it runs counter to criterion 1. If I were to propose a law about the relation between these two criteria, it might run: the less rules of spelling one proposes (e.g. a system like WES), the more unlike traditional spelling text becomes; conversely, the more rules one proposes, the more one's text becomes like traditional spelling. Consequently a system that has the least rules would be the one in which each symbol or combination of symbols represents only one phoneme (i.e. the sound of a letter) but which also would be the least readable from the point of view of the public. A compromise has to be achieved in which these two opposing aims are delicately counterbalanced.
3. The third criterion would be that it should be reasonably easy to spell within the new system from the point of view of both the layman already familiar with traditional spelling and the child learning to read. One major reason for proposing this criterion is that a great deal of time is wasted by children having to learn literally thousands of spellings. Just take the long-e sound. This sound is represented by several different spellings: ea, ee, e, ie, e-e, eo, ey. ei, and so, although the learner knows that the long-e sound in a particular word could have a range of spellings, he has to decide which is the particular correct spelling for that word. Consequently he has to memorize a large proportion of the spellings of the words with the long-e sound. Furthermore, it is unlikely that the public will want to change to a spelling system which is overcomplicated or to one which is so compromised that one has to learn a large number of exceptions.
Proposed guidelines for regularizing English spelling.In devising a spelling scheme one should bear in mind the likely disruption to English text brought about by any spelling reform. The first guideline therefore is designed to minimize disruption:
1. One would examine the way words are presently spelt and where several symbols or combinations of symbols represent the same sound, adopt the rule most frequently used. However, in English spelling, it is sometimes the case that the type of spelling is contingent on the position of the sound in the word. For example, the spelling ou represents the ow sound (as in 'house') most frequently in the middle of a word, but ow represents the same sound at the end of the word. Consequently, the most frequent spelling for a particular position should be employed. This first principle ensures that disruption is reasonably minimal. It also means that symbol combinations which are not particularly phonetic should be retained, e.g. -tion at the end of a word and qu at the beginning or middle of a word. However, if the rule occurs for too few a number of words, even though it is the most frequent one, the rule is not necessarily applied. This is in order to cut down on the total number of rules that have to be learned. For example, in my own scheme to be described later, I have opted not to retain gu for the gw sound as the complication of learning that spelling rule out-weighs the number of words with that sound combination. To conclude this first principle: the aim here is that there is only one spelling for a particular sound and that that particular spelling, if so imposed, will prove to be the least disruptive when the new text is compared to the same text in traditional spelling.
The advantages of least disruption are two-fold. Firstly, the public would probably be more willing to change to a system which is similar to the present one, and secondly, children taught in the new system would still be able to read books in the old spelling if necessary, without it appearing to be completely different.
2. If it were at all possible, that is, in circumstances in which it would not be too disruptive, a spelling combination should be as simple as possible in order to aid the learner. To illustrate with the author's scheme to be described later, if u-e is used to represent the ue sound (e.g. 'tune'), it would be more straightforward to learn if the ue sound is also represented at the end of the word by u rather than by ew, even though ew is presently used the most frequently at the end of the word. Thus instead of 'new', the spelling would be nu.
3. This next principle has a two-fold purpose to make spelling both more efficient and easier. While marking exam scripts for my course in Cognitive Psychology this June, I made a note of a sample of the student's spelling errors. The most common error was with the double consonant. Here are some examples: chanel, sylable, interpretted, aggreed, etc. Therefore the double consonant should be abolished, except in a few cases (e.g. midday).
4. Subtle distinctions in sounds should be ignored. For instance, some spelling reform systems seek to distinguish the th sound in 'theory' from that in 'these.' These distinctions would place a burden on learners, especially those with difficulties such as poor auditory discrimination (Yule, personal communication). Sometimes traditional spelling makes a half hearted attempt at differentiating one sound from another. For instance, the s and z sounds are largely undifferentiated (e.g. the 's' in 'is' and in 'result' should be spelt with a 'z'). Also, it is difficult to differentiate the s from the x sound in plurals. My solution in this case would be that the letter 's' would conveniently represent both categories of sound, except at the beginning of the word.
5. We should aim to arrive at a situation in which, given the rules of spelling, anyone could correctly generate the spelling of a new word given that (a) he knows how it is pronounced formally (and this is a problem even with the most phonetic system); (b) he has mastered the set of spelling rules for that system. In situations in which there are ambiguities, the devisors of a new spelling scheme would fall back on asking ordinary people to generate new words, having previously instructed them in the new spelling scheme, and the form of spelling which would be adopted would be the most frequently used form for each word. Alternately, or in addition, if a word can be spelt in more than one way, then each version might be acceptable.
6. If spelling reform takes place, it should be done simultaneously by all English speaking nations. This next point is perhaps debatable, but bearing in mind the dominant role of English in the communication of science and in other spheres, the spelling and punctuation structure should be as standard as possible across nations. This will be a problem with the different pronunciation between English and American. There is also a problem of different regional accents. However, traditional spelling has tolerated these differences remarkably well, so a new spelling scheme should try not to aggravate the problem. One example of how well present spelling copes with English versus American pronunciations is that the '-ew' ending represents the ue and oo sounds, enabling Americans to pronounce 'new' as noo and the British to pronounce it as nue.
I have devised a spelling scheme based on the above principles and there now follows a very brief description of the scheme as it might be presented to the British layman (the layman would actually receive an expanded version). A list of rules for reading could be constructed along similar lines. Comments on and analyses of the scheme will follow afterwards.
A brief summary of the spelling rules of the new scheme.In this scheme, spelling is based on the sound of words as pronounced in formal speech. So here are some of the words which change in a straightforward manner in the new schemer bred, hart, cigaret, giv, hav, ar, gon, involv, twelv, carv, frend, bilding. Most words in the English language may be spelt unambiguously from the rules given below:
1. Most silent letters are abolished or substituted, e.g. thum instead of 'thumb,' parm instead of 'palm.' Most double consonants are abolished, e.g. bel ('bell'), comunity. But note that only a few cases in words like midday, cannot, withhold, etc. is the double consonant 'retained as these are really two words joined together.
2. Ten words which should be spelt rather differently in the new scheme are kept the same as in traditional spelling. These can be memorized by learning the following somewhat gruesome rhyme which incorporates all the ten words (which are italicized):
I was one of the ones who was there who tride to pul out all your hair.
The consonant sounds.The consonants are spelt exactly as before with the following qualifications:
1. The j sound: j represents the j sound in all cases, e.g. jam, chanj ('change'), jigantic, etc.
2. The k sound: the k sound is always represented by c, so k is abolished.
3. The qu sound: As in traditional spelling, qu represents the qu (or kw) sound, e.g. quality, liquid, equater, etc.
4. The s sound. s represents the s sound in all cases, e.g. les ('less'), chans ('chance'), stand, sit, etc.
5. The z sound: z represents the z sound only at the beginning of a word, e.g. zip, otherwise s represents the z sound in all other positions, e.g. visual, jas ('jazz'),etc.;
6. The ex sound: the ex sound (and the gz sound as in 'exact') at the beginning of a word continues to be spelt ex, e.g. exclame, exempt, (except for 'Xmas' and 'X-ray'), as is the case in traditional spelling. Similarly, x is employed in the same way as in traditional spelling for other positions in the word, e.g. mix, ax, conexion, (conection is an alternative spelling).
The vowel sounds.The simple single vowels (a, e, i, o, u) are spelt exactly the same as in traditional spelling (e.g. flag, bet, thin, spot, thug) but other vowel sounds are spelt as follows:
1.* The long a sound is represented by a-e (e.g. mate, vane) except at the end of a word where it is represented by ay, e.g. (day, say, thay ('they').
2. The long a plus r sound is represented by air, e.g. fairy, mair, ('mayor' and 'mare'), scairs ('scarce' and 'scares'). But note: layer, servayer ('surveyor'), player.
3. The ar sound is only differentiated from the intermediate a sound if necessary, for instance, these words are spelt with ar: card, farm, harm, carm but these words are not spelt with ar: casal ('castle'), bath, gras, cast, last. In other words, for Southern English speakers, the ar sound in front of the hissing ending is pronounced with the intermediate a (e.g. 'grass' is pronounced grahss), but is spelt with an 'a' in the new scheme. For other English speakers, this vowel sound is pronounced as a in cat and is spelt, accordingly (e.g. gras).
4. The long e sound is represented by ea, e.g. meal, sleap. In the case in which the e in the word, as spelt in the new scheme, has more than one letter between itself and the end of the word, it is spelt just as e, e.g. experiens ('experience'), feld ('field'), equal, secret. However, when the word ends in ch, st or th, ea is still used: e.g. 'teach, east, teath. In the case of the word ending in the long e sound, this is spelt as e, e.g. me, be, ne ('knee'), fe, ple, ('plea'), we, he, she, tre ('tree').
5.* The long i sound is represented by i-e in the middle of the word, eg. tribe, nite, and is represented by y at end of a word, e.g. by, scy ('sky').
6.* The long o sound is represented by o-e, e.g. throte, gote, rote, throne, those, coxe ('coax'), and by o at the end of the word, e.g. solo, helo, belo, bo, tho, so, go.
7. The oi sound is represented by oi, e.g. coin, emploiment, groin, and by oy at the end of a word, e.g. boy, coy, toy, etc.
8. The oo digraph continues to represent the two different sounds in words such as brood, booc, ('book'). Here are some examples: boo, doo, zoo, groo, scroo, troo, bloo, rood, tooc, looc, hooc, etc.
9. The awe sound (or 'or' sound in Rec'd Standard) is represented by or, e.g. horl ('hall' and 'haul'), story, for ('for' and 'four'), por ('pore, 'poor', 'pour'), orltogether, orlso.
10. The diphthongal intermediate vowel plus long oo sound is represented by ou in the middle of a word and by ow at the end, e.g. hous, proud, cow, sow. Note the following: pouer, touer, ouer (power', 'tower', 'our', respectively).
11.* The long u sound is represented by u-e, e.g. fume, use, huge, except at the end of the word where it is spelt u, e.g. nu, lu ('new', 'few'), valu, continu, retinu.
*Note that in the silent e rule, when the vowel (a, i, o, or u) is separated from the end e by more than one consonant, the vowel is left unqualified, e.g. utensal ('utensil'), human, ulogy ('eulogy') criterion, blind, child, sical ('cycle'), stranj, broch ('brooch'), tost, most, loth, etc.
1. The -er, -or and -ar endings: many words are pronounced as a slurred er at the end even though they are presently spelt '-ar' or '-er.' These are all spelt er in the new scheme, e.g. tracter, raser ('razor').
2. Words ending in the l sound: these words are all spelt l at the end, e.g. pil, fil, lul, butiful, etc. But note that when there is a slurred vowel sound between the last consonant before l and l itself, this slurred vowel is always spelt with an a. Here are some examples to illustrate: pepal ('people'), reliabal, viabal, prinsipal, political, etc. Note that the silent e rule continues as before, e.g. pole, role, gole, pile, mule, etc.
3. The -sion, -tion, -zion and -ion endings: the -tion ending represents the shun, zhun and chun sounds, e.g. pention ration, divition, fution, question. The xion ending can represent the exshun sound, e.g. conexion, sexion, but these words may alternatively be spelt collection and section, respectively.
4. The -y ending: this continues to represent both the short i and long ie sounds at the end of words, e.g. sily, scy ('sky').
5. Other endings: these are spelt as in traditional spelling, e.g. ed, er, ing, ist, etc., e.g. sealed, oner ('owner' and 'honour'), sealing ('ceiling' and 'sealing'), sicling ('cycling'), tacing ('tacking' and 'taking'). As in traditional spelling, the final e of the root word should be dropped when adding an ending beginning with a vowel, but the e should be kept before a consonant, e.g. drive, driving, driven, live, lived, lively, liven, living, etc. Note the change in the -y ending: spy, spied, try, tried, etc. The rules for plurals are the same as in traditional spelling, e.g. booc-boocs, lady-ladies, hero-heroes; and for plurals of words with a hissing ending: gas-gases, wish-wishes, church-churches, fox-foxes, etc. Other endings have been described when necessary under the heading "Vowels" listed above.
An example of some text in the new scheme is given in Appendix 1. The scheme was devised keeping the proposed principles in mind and using Wijk (1959), the spelling counts of Dewey (1970) which were of some limited use, and my own spelling counts on a modest scale from ordinary text. The pronunciations are based on Hornby (1978), which has both English and American pronunciations and is prepared with foreigners in mind.
Comments on the new scheme.
General comments.The application of the "most frequent spelling" rule means that 's' represents the z sound except at the beginning of a word. In Dewey's corpus of 364,381 sampled words, out of 11,089 occurences of the z sound, only 247 were spelt 'z' whereas 10,695 were spelt 's'. However, to clarify the beginning of a word - an important part for the learner - z is used for the z sound. In the case of the k sound, out of 10,010 occurences, 6403 were spelt 'c' compared to 1,854 which were spelt as 'k', so c represents the k sound in all cases in the new scheme and 'c' is no longer used for any other sound. In the case of the j sound, an exception to the "most frequent spelling" rule was made because there were only 1,582 instances of the sound and although 948 were spelt with a 'g', it was considered that in view of the infrequency of this sound, it would not be too disruptive to spell it as j. Furthermore, unlike the s and z sounds, j and g are not so similar to each other so it should be easier for the beginning reader to differentiate one from the other. On the other hand, with the s and z sounds, it can be difficult to distinguish one sound from the other, especially for those people with poor auditory discrimination, as mentioned previously. It might be noted that the sion and zion endings are not differentiated for much the same reasoning and are both spelt tion.
There was a problem with the ea and oa endings in that there seemed to be much variability in the way these words ended. The eventual decision to represent these sounds by just e and o, respectively, was based on several considerations. Firstly, some very high frequency words end this way. Secondly, this spelling is the most economical and is an abbreviation for the other common alternatives, and thirdly, the layman can easily guess how they are supposed to be pronounced. The ue sound ending as in 'few', continue', etc. is represented by ew 199 times and by ue 45 times according to Dewey. It was decided that these occasions were sufficiently few to warrant spelling this sound with u alone so as to make spelling easier for learners, rather than spelling it with ew which looks rather different from u elsewhere in the word. Furthermore, this means that words ending in the three long vowels e, o, and u all follow the same rule and end the word with their respective single vowel. This should be an aid to learning the system.
As has been noted previously, the spelling combination gu which usually represents the gw sound, as in 'language, languish', etc. has not been adopted because the rule applies to too small a set of words. The same applies to the ue sound at the beginning of a word which is sometimes spelt as 'eu' as in 'eulogy' etc. Again, the set of words here is far too small for this rule to be worthy of adoption. For the same kind of reasons, it was decided to change the 'le' ending, which is common after a consonant in traditional spelling, to al because the 'le' ending is not very frequent (684 occurences in Dewey's corpus), although it is more frequent than the 'al' ending. A further reason for this was that e on the end of a word is already serving the functions of (a) making a vowel long in sound (e.g. cote) and (b) making a long e sound (e.g. she). Adding the third function of a silent, non-functional e as in the 'le' ending would have produced an added complication for the beginning reader.
The slurred vowel sound (the schwa), which is the seventh most frequently occurring phoneme out of 41 (see Dewey 1970, Table 3), and the second most frequent vowel sound, has presented something of a problem. In traditional spelling, the schwa vowel presents quite a spelling problem as Dewey lists 23 graphemic representations for it! Dewey found that the schwa sound is represented, in order of frequency, by 'a' which accounts for 5602 occurrences, by 'e' which occurs 5027 times, by 'o' which occurs 2901 times, and by 'u' which occurs 369 times. There is a total of 15,024 occurences of the schwa sound, or 4.1% of Dewey's corpus of phonemes. My solution has been aimed at tampering as little as possible with existing spelling because I found in practice that changing the schwa vowel would be moderately disruptive. Therefore for the endings '-er', '-ed', and '-at' which are often slurred (and 'er' in the middle of a word), these are left exactly as they are at present except that words which end in '-le' (e.g. 'principle') are changed to an -al ending (e.g. prinsipal, pepal, etc.). The remaining slurred vowel sounds are spelt exactly as they are in traditional spelling, for instance, here are some words spelt in the new scheme: seven, student, hundred, dificult, etc. The main advantage of this scheme is that the layman familiar with traditional spelling does not have to keep deciding whether a sound is sufficiently slurred for it to be spelt with a uniform schwa vowel such as 'a' or 'e'. Unfortunately it means that the child learning to spell will have to learn the different spellings for these words. But in relation to the enormous reduction in the overall spelling problems, this should be a minor burden and it should present no problem in reading. Note that there are still many vowel sounds which will be spelt phonetically in the new scheme, e.g. imerged, bineath, devotion, marcit, etc., because they are clearly pronounced differently from the vowels presently used to represent them.
The homograph problem.
Inevitably, the new system creates more homographs (i.e. words with the same spelling but different meanings, such as air for 'air' and 'heir') than previously because traditional spelling occasionally tries to differentiate between words of the same sound but of different meaning. But this creation of more homographs is seen as a major strength by the author because one major spelling burden for children is learning how to spell the different homonyms. The particularly difficult homonyms to learn tend to be the more abstract ones, for instance, 'their' and 'there', 'to' and 'too', etc. So if people were able to spell homonyms the same (e.g. 'some' and 'sum' both become sum) then a major spelling difficulty would be eliminated. Furthermore, a reading problem would not be created due to increased ambiguity because the context of the running words should aid identification. The spoken homophone is not normally difficult to identify, and to put the problem into context, there are now a few hundred homophones which are differentiated by their spelling (e.g. 'grate' and 'great'), but there are thousands of words with different meanings but with the same sound and spelling. Here are a few examples from Dewey (1971) with the number of meanings of each word in brackets: 'bay' (5), 'fair' (3), 'right' (3), 'sound' (3), 'spring' (3), etc. There is another group of homographs in which the words are spelt the same in traditional spelling but sounded differently. Many of these now become differentiated in the new spelling scheme e.g. 'bow' (bo, bow), 'row' (ro, row), 'read' (read, red), 'live' (liv, live), 'tear' (tair, tear), 'wound' (woond, wound).
The ten words retained in old spelling.
In the new scheme, ten common words remain the same because if they were changed, the spellings would be changed too drastically and this would not smooth the transition from the traditional spelling to the new scheme. This is important from the point of view of the layman reading the scheme for the first time - there should be as much similarity as possible between the two schemes or he may give up straight away. This idea is not new - for instance, Zachrisson (1932) in his spelling scheme "Anglic" left 43 words unchanged. In the new scheme, the ten words are incorporated into a rhyme to facilitate memorization so learning these exceptions would present only a minor problem for the layman. For the beginning reader, these exceptions would be minute in relation to what the beginning reader today has to face. The ten words are derived from the word frequency count of Kucera and Francis (1967) based on a million words. Here are the words with their frequency rank included in brackets: the (1), of (2), to (4), was (9), I (20), one (32), you (33), all (36), there (38), who (46). Note that the following words spelt here in traditional spelling would now have the same core spelling as the previous ten words: 'into', 'two', too', 'towards'('together', 'today', etc.), 'whom', 'whose'. These words would be spelt: into, to, to, towards (together, today etc.), there, whom, whos, respectively. These words are all derivatives of the three words who, there and to, and as a further memory aid, the following sentence might help:
Whos plase is there car going into today?However, derivatives of some of the rest of the remaining ten words will be spelt phonetically, e.g. 'once' (wuns), 'aye' (ie), 'eye' (ie), 'altogether' (orltogether), 'although' (orltho), 'also' (orlso), 'ewe' (yoo).
Advantages.There are several advantages to the new scheme, as will be seen mainly in the next sections. However, at this point it might be noted that the almost universal application of the silent e rule will make the learning of the new scheme easy for the layman; he doesn't need to learn a vowel combination for each long vowel sound, except for the long e and air sounds. As for the beginning reader, the fact that long vowels in polysyllable words are not specified (e.g. the u in utensal) may not be a problem because in the early reading stages, mainly monosyllabic words are learned. By the time the longer words are being learned, the pronunciation problems will have been reduced. A similar kind of advantage of the scheme, mentioned earlier, is that three long vowel sounds all follow the same rule at the end of the word by being represented by the single vowel letter (e.g. she, blo, and nu). This is a rule which intuitively makes sense and is simple to apply. This rule cannot be applied to the other two vowels 'a' and 'i' because these two vowel sounds occur in both long and short forms at the end of a word whereas the other three vowels do not (e.g. data, play, pity, sly). Another advantage of the scheme is that like traditional orthography, it attempts to minimize the differences between British and American pronunciations. For instance, in the new scheme, the ar sound is spelt 'ar' only when strictly necessary, for instance, lard, bark, and hard; but in words such as casal and gras, the 'ar' spelling is not used.
Analyses on the new scheme.
Reading the new scheme.
For the child or foreigner learning to read this scheme, there would be the problem that several symbols represent more than one sound. For instance, 'i' in the middle or at the beginning of a word could represent the short or long i sound. In fact, there are nine symbols or symbol combinations (out of 56 including the vowels in certain positions) which represent more than one sound: 'a' (in certain positions can represent short a, long a, or ar), 's' (represents s or z), 'oo' (represents two different sounds as in look and aloof), 'th' (represents the two different sounds as in that and thesis), the four remaining vowels 'e', 'i', 'o' and 'u' which can represent their respective long and short sounds (e.g. equal, mention, situation, criterion, etc.), and finally, at the end of the word, 'y' represents both the long and short i.
However, an analysis of specimen texts containing 1,345 words (used by Wijk and others) was undergone to discover whether a reader, given a knowledge of the words in the English language, could mistake any words for words of a different sound. Out of this sample, only seven examples were found, which represents 0% of the sample. This is a negligible amount. It should be added that none of these alternatives would have been remotely appropriate in the context of the passage. The seven words were: fiting, ('fighting', 'fitting'), halo ('halo', 'hallow'), cors ('cause', 'course'), fase ('face', 'phase') - occurred twice, raped ('raped', 'wrapped'), grase ('graze', 'grace'). This lack of a one-to-one relationship between a symbol and a phoneme is far less than in traditional spelling. These ambiguities are retained because in some cases, a change would produce considerable disruptions to the text when comparing the new scheme with traditional spelling (e.g. changing all z sounds from 's' to 'z'). In some cases it would mean that extra symbols would have to be put in to clarify a sound (e.g. expeariens instead of experiens); this would be cumbersome and inefficient. Furthermore, our traditional spelling system has exactly the same ambiguities and many, many more besides. As for the foreign learner who would like the spelling structure to enable him to know completely how to pronounce a word, a good textbook giving him guides to the ambiguous pronunciations in his early stages of learning should serve this purpose.
In order to gauge the degree of disturbance in the text from the point of view of the layman who is used to traditional orthography and is trying to read the new spelling, the same text of 1,345 words was examined. It was found that 69% of the words in the text remained unchanged, which is quite good. Then the first 1,000 more frequent words were examined from the count of over one million words by Kucera and Francis (1967). This sample of 1,000 words represents 68% of the sampled million words, correcting for the omission of non-words and proper names, and so it represents a good proportion of vocabulary. In this sample, 48% of the words remained unchanged. This reduction is mainly because in the new spelling scheme, ten very frequent words remained unchanged and in ordinary text these occur with sufficient frequency to inflate the proportion of words unaffected by a change in spelling scheme. The words that were changed were scored according to how many letters of the word in traditional spelling were deleted as part of the transition to the new spelling. The purpose of this scoring was to ascertain the amount of context that would remain unchanged under the new scheme. A preliminary analysis revealed that 84% of the words in the sample of 1,000 remain unchanged or had only one letter deleted as a result of the change. Table 1 shows the % of words in the sample as a function of the % of deletion. This table reveals that it is comparatively rare to have over 40% of the word deleted.
Table 1.The % of the 1,000 most frequent words from Kucera and Francis (1967) as a function of the % of letters deleted by changing from traditional to new spelling.
|% of letters deleted||1%-20%||21%-40%||41%-60%||61%-80%||81%-100%|
|% of word||27%||19%||4%||1%||0%|
The average % of a word deleted, if a part was deleted, was 25%, the standard deviation was 12.9%. The average word length was 6.1 letters for words that had to be changed, and 5.1 letters for words that remained unchanged. An examination of frequency distributions according to length showed that the most frequent word length (i.e. the mode) was four letters for the unchanged words and five letters for the words that were changed. Furthermore, for all word lengths under five letters, more words were unchanged than changed; conversely, for all word lengths from 5 to 11 letters, there were more words changed than unchanged.
Spelling in the new scheme.
The question to pose here is: how many rules would the beginning reader (i.e. child or foreigner) have to learn before he could spell most words in the English language without difficulty. In order to obtain an objective measure, the new scheme was considered in terms of the number of simple propositional statements that would have to be learned. Also, all words which are exceptions to the scheme (e.g. I, was, one, etc.) count as one statement each. To illustrate, the rules concerning the z sound may be expressed by the following statements:
1. The z sound at the beginning of a word is spelt z.So this would count as two statements covering the z sound. Altogether, it was estimated that the beginner would need to learn about at least 97 propositions. This included 22 propositions for consonants and 32 for vowels. The task for the layman familiar with traditional orthography was a total of 72 propositions which included 8 propositions for consonants and 32 for vowels. However in this case, these propositions were in the main highly familiar to the reader as they constituted the most frequent spelling rules in traditional spelling. The words retained in their traditional spelling were estimated to be 20, which included the ten common words and most of their derivatives. In the case of the beginner, all the individual spellings of the schwa sounds would have to be learned; by contrast, the layman would already know these spellings. The author and his wife found the system easy to learn, but clearly an experiment is needed to find out how quickly the system can be mastered by others.
2. The z sound is spelt as s except at the beginning of the word.
An analysis was undertaken to work out the economy of spelling in the new scheme. It was found that in the 1,000 most frequent words, when the spelling was changed, the word length was reduced by 14%. The mean length of the traditionally spelt word to be changed was 6.0 letters, and this was reduced to 5.2 letters when the word was spelt in the new form. To put this another way, on average, words that had to be changed were six letters in length and they lost one letter when changed. Further analyses on the spelling economy of the scheme are described in the last section of this paper.
Comparisons with other spelling schemes.Here are very brief descriptions of some major schemes:
World English Spelling (WES). This is from the Simplified Spelling Assoc. and is an almost completely phonetic spelling system which is very similar to i.t.a. except that it uses the Roman alphabet. Here is an example: "... or eni naeshon soe konseevd and soe dedikaeted, kan long enduer." Spelling in WES is quite straightforward to learn but it is very different in appearance from traditional spelling. For instance, translation into Lincoln's address only leaves 41% of words unchanged. But the system would probably be just as useful as i.t.a. as a spelling medium to start children reading.
Anglic. Proposed by Prof. Zachrisson (1932) is again a phonetic system like WES and in fact, the above sample text for WES would be identical in Anglic. But the main difference is that 43 common words are allowed to remain unchanged in Anglic.
Wijk's Regularized English (Regularized Inglish). This is a good system from the point of view of minimal disruption from old spelling to new. The advantage of this scheme is that it retains most of the rules of traditional spelling and also creates some new rules so that, given that one knows these rules, one has a very good idea about how a word should be pronounced. This is obviously a big advantage for the foreigner learning English. The major criticism of the scheme is that it tolerates to a large extent, the wide range of spellings for each sound. Consequently, learning how to spell in the scheme is complicated and similarly the foreign reader would have to learn a large number of rules before he would have mastered how to pronounce all words. A minor criticism of the scheme is that subtle sound distinctions are differentiated by different spellings which probably would be difficult to learn for people used to traditional spelling and for those with poor auditory discrimination. Here is a sample of Wijk's spelling: "... or eny nation so conceevd and so dedicated, can long endure."
Yule's spelling scheme. This is again a good system from the view point that a scheme should not be too disruptive compared to the old spelling system. Unlike the previous schemes, this is not a fixed scheme but suggests a series of minor reforms of spelling over time which should take place until a more nearly phonetic system is reached. The ideas for the early stages have a lot to offer and at a certain stage of development, come close in appearance to the scheme presented in this paper, as can be seen in the following example: "... eny nation so conseevd and so dedicated can long enduer." The differences between it and my scheme are that this version of Yule's scheme involves the eventual abolition of the silent e rule, omission of unstressed schwa letters, the use of 'k' instead of 'c' under certain conditions, 'ee' instead of 'ea' and the possibility of a limited number of distinctions for homonyms with other minor differences. There is also a tolerance of roughly the same common words spelt in traditional spellings. Valerie Yule kindly translated Lincoln's address for me into her scheme and 71% of the words remained unchanged.
The author's scheme.To summarize this scheme; it attempts to disturb traditional spelling as little as possible by adopting the most frequent spelling rules and by using as few spelling rules as possible. Thus, each sound can be spelt by only one type of spelling (unlike Wijk's scheme). Ten common words, incorporated into a rhyme, are left unchanged. The advantages of the scheme are that it is one of the best in terms of minimal disturbance from traditional to new spelling and it is relatively simple to learn to spell. Here is a sample of the scheme: "... or eny nation so conseaved and so dedicated can long endure."
Some comparisons across the scheme.
Disruption.Table 5 illustrates the % of words that remain unchanged in the sampled texts of 1,345 words. The % from Yule and WES were not available. WES would probably be slightly worse than Anglic in terms of the amount of disruption.
Table 5: The percentages of words which remain unchanged in sampled text for three schemes.
Titles, proper names and non-words were not included in this analysis. So the author's scheme is almost as good as Wijk's and Yule's in terms of minimal disruption produced by changing to a new scheme. This disruption criterion is by far the most important criterion in assessing a spelling scheme because the layman is going to be reluctant to give up his well-established reading habits to transfer to a system which is too different from what he is used to seeing.
Number of spelling rules.Figure 1 is a rough schematic representation of the number of rules of spelling that would have to be learned by a child learning each new scheme. It can be seen that Wijk's scheme would produce the greatest amount of difficulties.
Figure 1: Schematic diagram of the number of rules that would have to be learned by the layman to master each scheme.
|Minimum number of rules||Present number of rules|
It should be noted that the line should be ten (or more) times its length between Beech and Wijk to be truly representational.
Ambiguities in reading.Figure 2 is a rough schematic representation of the number of ambiguities which might be encountered in reading each scheme. For instance, in the author's scheme, 's' in a word might be pronounced s or z. It can be seen that most of the schemes are almost perfect in this regard but Yule's and Beech's do not, for instance, disambiguate the two sounds represented by 'th.' On the other hand, it could be argued that the other schemes overspecify sounds and that these schemes may slightly confuse the child with hearing difficulties.
Figure 2: Schematic diagram to the extent to which a letter or combination of letters represent one phoneme in the different spelling schemes.
|Complete phoneme to
i.e. no ambiguities.
|Ambiguities to the same|
extent as in traditional
Ambiguities in writing.Figure 3 shows the amount of ambiguities which might be encountered in writing each scheme. Because Wijk has several ways of spelling each long vowel sound, it poses problems on the learner's memory just like traditional spelling.
Figure 3: Schematic diagram of the extent to which a sound is represented by one grapheme (letter or combination of letters.)
|Complete grapheme to
|Ambiguities to the same|
extent as in trad. spell.
Economy.The economy of a spelling system refers to the % of extra or fewer letters that have to be used in the system. A system using more letters than previously probably has less ambiguity when the words are read, but it can be more cumbersome to write and consequently may be more unpopular with the layman. A system employing fewer letters may have more ambiguity, but it is more efficient to write, and this could be a factor greatly favoured by the public. Figure 4 is a schematic diagram of how the various systems would compare in terms of their respective efficiencies. Yule's scheme is similar to the author's in terms of efficiency.
Figure 4: Schematic diagram of how many extra or fewer letters are used in the various spelling schemes. The % are based on Lincoln's Gettysburg address.
The efficiency of Beech's system for the whole sampled text of 1,345 words was a reduction of 4.1% letters and on the 1,000 most frequent words, there was a reduction of 8.1% of letters overall. Another advantage of efficiency in any reading scheme is that it implies a financial saving and, using the kind of calculation employed by Dewey (1971), an efficiency of 5% would mean a saving of 50 million dollars out of one billion dollars of writing and printing costs.
Conclusion.An author of a spelling system is perhaps not in the position to give an unbiased appraisal of his and other systems. However it does seem to me that the new system presented in this paper, or one that is similar in approach, has sufficient advantages to be put forward as a candidate for a spelling reform that is both likely to be accepted by the public and is likely to put an end to much of the misery which children are subjected to when trying to learn to read.
Acknowledgements.I am very grateful to Valerie Yule for many useful comments on earlier versions of the spelling scheme and also to my wife, Jenny, who carried out most of the counts.
Dewey, G. Relative frequency of English spellings. Teachers College Press, New York, 1970.
Dewey, G. English spelling: Roadblock to reading. Teachers College Press, New York, 1971.
Hornby, A.S. Oxford student's dictionary of current English. Oxford Univ. Press, 1978.
Kucera, H. and Francis, W.N. Computational analysis of present-day American English. Brown. Univ. Press, Providence, R.I., 1967.
Wijk, A. Regularized English-Regularized Inglish. Almqvist and Wiksell, Stockholm, Sweden, 1959.
Yule, V. 'Let us be practical about spelling reform.' Spelling Progress Bulletin, 1979, v.19, no.1, pp7-9.
Zachrisson, R.E. Anglic, an international language, with a survey of English spelling reform. W. Heffner & Sons, 1932.
Appendix 1.Lincoln's Gettysburg Address in the author's new scheme.
Forscor and seven years ago ouer fathers brort forth on this continent a nu nation, conseaved in liberty, and dedicated to the proposition that all men ar created equal.
Now we ar engajed in a grate sivil wor, testing wether that nation, or eny nation so conseaved and so dedicated can long endure. We ar met on a grate batle-feld of that wor. We hav cum to dedicate a portion of that feld as a final resting plase for those who hear gave there lives that that nation mite liv. It is orltogether fiting and proper that we shood doo this.
But in a larjer sens, we cannot dedicate - we cannot consicrate - we cannot halo - this ground. The brave men, living and ded, who strugled hear, hav consecrated it far abuv ouer poor power to ad or ditract. The world wil lital note, nor long rimember wot we say hear, but it can never forget wot thay did hear. It is for us, the living, rather to be dedicated hear to the unfinished werc wich thay who fort hear hav thus far so nobly advansed. It is rather for us to be dedicated to the grate tasc rimaning before us - that from theas onored ded we tace increased divotion to that cors for wich thay gave the last ful mesure of divotion; that we hear hyly risolv that theas ded shal not hav died in vane; that this nation, under God, shal hav a nu berth of fredom; and that guvement of the pepal, by the pepal, for the pepal, shal not perish from the erth.
Appendix 2.Reciting the alphabet.
The alphabet as it is presently recited would be misleading to the child in the case of the letters 'c' and 'g'; these are pronounced see and jee, respectively and so do not represent their actual sounds in the new scheme. Therefore these would be changed to ce (sounding like 'key') and ge (hard g). The letter k (pronounced cay) would be retained so that children could read the old spelling if necessary. Here are the pronunciations of the alphabet spelt in the new scheme: ae, be, ce, de, ea, ef, ge, ach, ic, jay, kay, el, em, en, oe, pe, pe, ar, es, te, yoo, dubal-yoo, ex, wy, zed.
Note that kay would be the only word spelt with a k in the whole of the new scheme.
However, if the alphabet is going to be changed in its pronunciation, it might be a good idea to make further changes so that all names of the letters incorporated the sound of the letter in their pronunciation. In the traditional alphabet, 'h', 'q', and 'w' do not contain the pronunciation of the letters they represent. I would suggest that the following pronunciations of these letters would not destroy the rhythm of reciting the alphabet: hay, que (pronounced 'kwee') and wed. So a phonetic alphabet, in the sense that the name of each letter is contained in the pronunciation of the letter, would sound as follows spelt in the new scheme: ae, be, ce, de, ea, ef, ge, hay, ie, jay, kay, el, em, en, oe, pe, que, ar, es, te, yoo, ve, wed, ex, wy, zed.
Or spelt in traditional spelling: ay, bee, kee, dee, ee, ef, gee, hay, ie, jay, kay, el, em, en, oe, pe, kwee, ar, es, tee, yoo, vee, wed, ex, why, zed.
As a footnote this new alphabet is called 'abece' (pronounced 'aibeekee').
Back to the top.