[Journal of the Simplified Spelling Society 1987/1, pp5-10 Later designated Journal 4]

The Transcription of Pronunciation in Dictionaries
and its Implications for Spelling.

David Brazil.

Dr Brazil is a phonetician in the English Department of the University of Birmingham, where he has evolved a new system for describing the intonation system of English. Some of its insights are incorporated in the forthcoming Cobuild English Language Dictionary. We here publish, with his permission an edited version of his address to the Society on 21.6.1986.


1.1 Learners' needs.
Intonation is an important part of pronunciation, but what has it to do with spelling? In approaching this question, I should first explain I am involved with the lexicography and dictionary activities at Birmingham University, mostly sponsored and stimulated by Collins Publishers. We are working on an advanced learner's dictionary of English, for which I have been asked particularly to provide the phonetics entries. I am proceeding on the principle that any advanced learner who goes to a dictionary, whether to find out the meaning of a word or its spelling, is equally likely to want to find out how to pronounce it.

1.2 New findings on intonation.
Most dictionaries make an attempt, after the headword, to provide something that looks like, and sometimes is, international phonetic script (though it is sometimes something quite different). In approaching my task, I have had to re-examine the question of representing pronunciation graphically, particularly in the light of what I have discovered about the organization of intonation and the effect that this has on things like stress. So what I want to discuss is primarily how people sound, and I hope this will then shed light on how those sounds might be represented in an orthodox spelling system.

1.3 Phonographic complications.
Spelling reformers have long said the ideal orthography should have a one-to-one relationship between sound and symbol. To some extent I shall be suggesting that the sound part of this equation is even less straightforward than people have so far recognized. So in a sense, instead of offering easy solutions to spelling reform, I am as it were muddying the stream. But it is surely nevertheless true that any sensible reform will have to be based on a real scientific appreciation of what the sound-system is like. Otherwise it is doomed to failure from the start.


2.1 Pronunciation look-up procedure.
This extract from the Cobuild advanced learner's dictionary illustrates the sound symbols by using key words.



We here see that the symbol /ɑ/ stands for the vowel in heart, and the full representation of heart is /hɑ:t/; the symbol /æ/ stands for the vowel in act, which then appears as /ækt/, and so on. Clearly the dictionary has to give the whole symbol/sound-inventory, but these examples suffice for our purpose. The assumption on which the system is based is that learners who come across a word whose pronunciation they don't know first look the word up in the dictionary, see what the pronunciation symbols are, then trace those symbols back to their occurrence in the key words, and by a process of substitution finally derive the pronunciation of the word in question. So with the word <disrespectful>, the transcription /dιsrιspɛkfυl/ shows the first vowel as the /ι/ in build, the second vowel, <e>, similarly as the /ι/ in build, the third as the /ɛ/ in bet, and the final vowel as the /υ/ in good. If foreign learners use a dictionary, that is usually the only procedure open to them. The dictionary seems to assume that advanced learners have already mastered the sound-system of English, that they can already produce those sounds, and that all they are doing is making an appropriate selection from what they already know.

2.2 Does this meet learners' needs?
Now I am not sure that this is clear to those who write and talk about dictionaries. One can say those symbols obviously resemble the symbols in the International Phonetic Alphabet, so they could well be standing for these phonemes, if not for phonetic entities. Thus they show there is a difference between /i:, ι/, one standing for a vowel that is further forward and produced slightly higher in the mouth than the other, and those who don't happen to have that contrast in their own language must be taught the difference and how physically to produce those two sounds, so that they don't say ship for sheep or vice versa. It seems to me that making discriminations like that is no part of a dictionary's job, that a dictionary cannot in principle tell people how to pronounce sounds unless they already know how to do it. If people are confusing /i:, ι/, certainly they need to be corrected in class, perhaps with reference to phonetics, but that is quite different from what happens when the learner looks up the word in a dictionary. For one thing, I simply do not believe that most people who use dictionaries, even advanced learners and native speakers, are sufficiently sophisticated to be able to know all about such distinctions. It is a very much more rough and ready substitution that is involved. The dictionary does not have the task of teaching people to pronounce the language. It assumes they can already pronounce the language in the words they know, and simply need to do the same sort of thing in new words that they can't pronounce.

2.3 Variations in speech.
But already in disrespectful the position is less straightforward than it looks, since we see the above transcription lacks the /t/ that is present in the spelling <disrespectful>, and there might be disagreement as to whether the <t> is pronounced or not. There is also uncertainty about the second vowel, the first <e> in that word: the transcription gave it as /ι/, but if speaking fluently, I might well pronounce it as a schwa, /ə/. Then the last vowel in disrespectful is here transcribed as /ʋ/, but if I'm speaking rapidly, I may make this also into a schwa, with the same pronunciation as in raffle. In fact, in normal speech the native speaker will accept both values of the vowel equally.

2.4 Which sound in the dictionary?
What are the implications of this? In the above three cases the dictionary transcription is faced with wide variation in usage. Why should we say that a given vowel is /ι/ rather than /ə/? On what basis are we saying, such and such is the correct form? Foreign language users are well accustomed to the Daniel Jones-Gimson tradition, in which someone has told them what is the correct value for the vowels. But there is a fairly big discrepancy between what they are told and what actually occurs. That is the first problem for me, because I want a dictionary to be useful to learners not only when they are producing, but when they are listening. If they have been taught the pronunciation as transcribed, and they at once hear a different pronunciation, there is a worrying discrepancy if they are that sensitive. The chances are they are not that sensitive, but globally, given that this kind of problem is going to arise with perhaps 50% of words in the dictionary, it is a worrying phenomenon.

2.5 Which accent?
In the Dictionary you have got to define the kind of English you are representing. A dictionary of British English would not for instance represent the <r> in heart in the pronunciation. But we have to take account of a rather wider spectrum of usage than what is normally described as RP.


3.1 Spelling by sound or vice versa?
Furthermore, some people may say they pronounce the second vowel in disrespectful with something of the sound /i:/, on account of the letter <e> used in the spelling. In that case we are suggesting a full range between /i:, ι ə/ for that vowel-letter. But if you say that a good speller pronounces the word more closely according to the spelling, you are determining the pronunciation from the spelling. So where does that leave the proposition that we should base spelling on pronunciation? Our perception of a word influences our pronunciation. But you can't on the one hand say spelling should follow pronunciation, and on the other that pronunciation follows spelling - it's a circular argument.

3.2 Consistency the aim of spelling.
In terms of spelling reform, one must conclude from this that any reform should make the spelling system itself consistent without reference to phonetic representation. This turns out to be a complicated task, quite different from what most spelling-reformers in the past have attempted, since they considered that the great defect of English spelling was the breakdown of the sound-symbol link. In essence, the sound system has an existence and a consistency of its own which permits all sorts of variations within certain limits, and to start by assuming that it is predictable, regular, consistent, and that all reformers have to do is to relate symbols to it, is not going to work.


4.1 Dictionary indication of stress.
The key to the kind of variation I have been discussing is stress. In /dιsrιspɛktfʋl/ a dictionary will give primary stress to /spɛkt/ and secondary stress to /dis/, thus /,dιs rι 'spɛkt fυl/ and it is taken for granted that the foreign learner, and indeed the native speaker, can interpret those symbols. I'm not sure whether he can or not: I'm not sure whether you don't need to know how to pronounce the word before you know what the symbols mean. But that's another issue.

4.2 Variable stress.
But now let us consider what happens if you say disrespectful behaviour. The distribution of primary and secondary stress, as indicated in /'dιsrιspɛktfυl bι 'hɛιv jə/, may now be spread over the larger chunk of speech. In disrespectful behaviour the /spɛk/, far from having primary stress, has at most tertiary stress. Something rather interesting happens too if you say his behaviour was disrespectful, /hιz bι,hειv jə wəz dιsrι'spɛkfυl/: the /dιs/ loses its secondary stress. And it turns out that this is something whose significance has been largely overlooked in phonetic writing. Yet it is tremendously important. I believe that something very similar may happen in all languages.

4.3 Stress communicatively significant.
What I am going to suggest is that the primary and secondary stresses are communicatively significant in themselves, and not merely a function of having chosen particular word. They affect what we are saying, and are a very important part of the communicative significance of spoken language. Consider the transcriptions
/d ι s r ι s p ɛ k f υ l/
/d ι s r ι s p ɛ k f υ l   b ι h ɛι v j ə/
/h ι z   b ι h ɛι v j ə   w a z   d ι s r ι s p ɛ k υ l/
/v ɛ r ι   d ι s r ι s p ɛ k f υ l   ι n d i: d/.

Here I've used a larger character for vowels which occur in syllables which receive primary or secondary stress.


5.1 Psychological pre-assembly.
I could even go so far as to give as a single chunk
/ιt wəz ,vɛri dιsrιspɛkfυl bi:hɛιvjə ιn'di:d/.

In fact the only limit on length here is the amount speakers can pre-programme, that is, how much they can set up in their neuro-linguistic circuitry before they speak. It is an interesting fact that I can't say very disrespectful indeed unless I have conceived the whole phrase in my mind when I start speaking. This sort of minimal unit of pre-assembly may be psychologically important. With people who don't know the language very well and who are assembling an utterance piecemeal, one word at a time, if they say it was very - oh, what's the word I need? - disrespectful indeed, you don't get this kind of stress-patterning. Even with native speakers, if they have some sort of cognitive block, or are being very careful, thinking on their feet, deciding what to say, the intonation will reflect their moment by moment formulation as they go along.

5.2 Words and tone units.
So the learner, using the dictionary, looks up first very, then disrespectful, then indeed, and pronounces each word with its individual sound value, which is probably not what was wanted at all. This is the significance of the triangle below:


/          \
CON tro VER sial
/               \
a CON tro versial deCISion
/                     \
his deCISion was contro VERsial
/                           \
it was VERy controversial inDEED
/                                \
a VERY controversial decision inDEED

There exists in spoken English a clearly recognizable unit which is not a word, but rather a tone unit, as represented by the utterances on the lines in the triangle. The significance of the pronunciation of a word as it is given in a dictionary is that that is how you will say the word if you happen to want to say it as a single tone unit. One of the problems, I am convinced, that spelling reform has got to come to grips with in some way or another, is that spelling is always focussed upon the word as a unit. Speech doesn't do this. I'm not sure that the word has much significance in the examination of speech at all, and it may well be that we ought to be talking about tone-units as a minimum unit, and say that the word is the bit that is left when you have reduced the tone-unit to its smallest proportion.

5.3 Prominences.
If you listen to a person speaking, and concentrate not on the meaning but on the intonation and stress-distribution, you find that what is said falls apart into chunks, characterized by either secondary followed by primary stress, or, in the minimal case, just primary stress. I call them neither primary nor secondary stress, but rather 'Prominences', to distinguish them from stress. The transcript of a recording of a radio discussion made some time ago will show what I mean. A group of lawyers were talking about the operation of the social security system, and one of them said:

Well, I think on the whole ... these officials ... do a remarkably good job. We have to remember ... that they are required ... by administrative practice ... to take these decisions ... on paper ... and very often ... when they get these decisions wrong ... it's because ... they haven't had an opportunity ... of talking ... face to face ... with the claimant ... and really finding ... the facts.

5.4 'Expression'.
It so happens that, perhaps because the speaker is a lawyer, perhaps because he is trying to make what he's saying sound important, he leaves a long gap between tone-units; but you can hear the tone-units even if there are no such gaps. If spoken without gaps, the statement doesn't differ significantly from the original rather slow and ponderous version. But without the prominences, we get a monotonous delivery that sounds like a speech synthesiser rather than real human speech. Likewise at school we used to have to learn poems by heart and recite them just to prove that we knew them, and the teacher would criticize us for not speaking them with 'expression'. By 'expression' was meant treating the poem as though it were English communication.


6.1 Prominences carry meaning.
This division into tone-units and decisions about which syllables to make prominent within the tone-units cannot be predicted simply from the grammar or the vocabulary or any other formal aspect of the language. It is a quite separate decision made by the speaker, carrying quite separate language information. It is something that written language does not convey, but it is assumed we can deduce it. The punctuation can give a hint of it, but doesn't do so at all consistently. Take this example: yesterday I was asked whether I was free today, and I said, No, I'm going to a meeting in London. This statement is divided into two tone-units, one being I'm going to a meeting (like disrespectful behaviour), and the other, with only one prominence, in London. I could alternatively have said, I'm going to a meeting in London, but that would have implied a different meaning, appearing to answer the question, where are you going?, and it presumes one knows that the person is going to a meeting. But my actual reply, broken into two tone-units, gives both items of information, both the purpose and the destination of the journey.

6.2 Breath-groups not primary.
Traditionally this kind of structure of utterances has often been described in terms of breath-groups, but I reject that entirely. If it was just a matter of breath-groups, it would simply be a question of physiological need not under the control of the speaker: you would just speak until you ran out of breath, and the utterance would not be structured in a voluntarily meaningful way. But this structure by tone-units is voluntarily meaningful.

6.3 Interactive context.
This division of speech into tone-units is just as important as its organization into grammatical units, but separate from it. The way I organize my speech into tone-units depends on my apprehension of the here-and-now state of communication between myself and the hearer. I am constantly changing what I do to mesh with the set of expectations my hearer is likely to have. This is why it is nonsense to suggest there is a correct intonation for a particular sentence. That is never so, because the intonation depends not on anything in the sentence, but on the speaker's awareness of the interactive context. Everything that is said is uttered against a unique special background of understanding, and it is that which is being reflected in the intonation.

6.4 Little problem for foreigners.
That is a real problem for dictionary makers. Interestingly however the provisional findings of research suggest that foreign learners are not likely to choose a wrong pattern of intonation for a sentence, even though people often have the impression that what constitutes a foreign accent has something to do with the intonation. But in fact there are all kinds of things that the man in the street calls intonation that are not part of this system of meaningful oppositions - there are perhaps characteristically French tunes superimposed on intonation, just as in the British Isles there are characteristically Birmingham tunes, Welsh tunes, Cockney tunes, and so on, which have to be considered separately from the manipulation of the system.

6.5 Incorrect/correct intonation.
It requires some concentration to produce random intonation, and if I do it successfully, you find it difficult to follow because there is an obvious mismatch between what I am trying to say and the assumptions I ought to make about what the audience wants to hear. But that doesn't mean there is a single correct intonation: everything depends on what the speaker presumes the listener already knows.

6.6 Non-communicative intonation.
This has implications for language-teaching. One of the difficulties with teaching a foreign language is that initially you have to use artificial samples of that language without any real context. That is why recently there have been efforts to introduce communicative language teaching, which has a real context and seeks to convey real information. It may be that some speakers of English do tend not to interact with their listeners, but just produce sentences without these communicative stress-patterns. Thus one often hears such a lack of communicative intonation in public announcements, as on the railways, or from a tourist-guide, because the speaker is not interacting but merely reciting what has been learnt. It happens not infrequently, I'm sorry to say, in academic lectures, especially where lecturers are not native speakers of English. Then the lecturer can internalize a piece of his lecture as ritual, from his notes, not relating to the audience but just to the lecture, with consequent long runs of non-communicative intonation.


7.1 Nature of the problem.
This has serious implications for dictionary-makers, because the presentation of any word is dependent upon the here-and-now situation in which that word is used. So how can we arrive at a definitive pronunciation for any word? What do we mean when we say disrespectful has first secondary stress and then primary stress? All we mean is, that is how we say the word if we say it in isolation. But as most people want to use the word in a context, the dictionary must admit a range of possibilities.

7.2 Indicating varying stress pattern.
The dictionary entry
d ι s r ι s p ɛ k t f ə l
indicates that if disrespectful is used in a context where it needs a prominence at all, the prominence will occur on either the first or the third vowel, which are printed large. But which of the two takes the prominence depends on whether it is at the beginning or the end of the tone-unit. In fact the varying stress pattern we find in the context disrespectful behaviour and in the context his behaviour is disrespectful turns out to be a fully generalizable fact about all two-prominence words in the language.

7.3 Protected vowels.
The transcription tells us not only that the two vowels printed large are available as prominences if real speech demands it, but also that the first vowel is going to be /ι/, and the problem of variation that we had with the vowel in <resp> or <ful> does not arise: the first vowel going to remain /ι/ whether or not it is prominent. Likewise the vowel in <spect> is going to remain /ɛ/, regardless. The advantage of this notation is that it enables me not only to indicate where the prominences will be if there are prominences, but also to make a distinction between what I call protected vowels and unprotected ones. It turns out to be a very general rule that those vowels that are used in the prominent syllables of the citation form retain their value within fairly strict limits. (There are a few exceptions - no statements like this could probably be made absolutely.)

7.4 Unprotected vowels.
But the other vowels are unprotected, and there is the possibility of variation, occurring within the usage of an individual person. It depends not on dialect, but on such things as whether you speak carefully or quickly, where the vowel comes in the tone-unit, what its phonetic environment is, and so on. So my notation for the dictionary gives the information
d ι r ι s p ɛ k t  f ə l  
  / \ / \

which tells us that the second vowel in disrespectful can be realized as /ι/, or as /ə/, or as anything nondescript in between. When uttered in fluent speech, it is hard to tell what the precise sound-value is - it can be anywhere on the continuum between the two. It may even be a longer continuum than that: it has for instance been suggested that the vowel might be an /ɛ/, so there is a very wide range of possibilities. There is a similar range of possibilities in the final vowel: it may be /ɛ/, or /ə/, or even nothing at all, in which case we put a little superscript zero in, υ<->0

7.5 Which variant first?
The first of the two symbols in these notations is the Jones-Gimson type of vowel, and can be relied on to be acceptable - it is my sop to the conservative tradition, and says: this is what's normally taught. The second is quite likely to be the more common. In regulate, for instance, the /υ/ realization of the second vowel as in
/r ɛ g j υ l ɛ ι t/,
is the one given in most dictionaries, and that is the more formal. But I believe the dictionary that gives only that realization is giving a very up-market version, and that these traditional dictionary descriptions describe a language that few of us ever use - we are few of us that consistent. The unprotected vowel can be conditioned by what comes after, or by intonation. There may well be rules to be discovered, but nobody wants them in a dictionary. What I'm doing is presenting something that is useable by ordinary people. The principles we articulate may be sophisticated, but their application has got to be as transparent as possible.

7.6 Practicalities of publishing.
What I needed in my dictionary was some way of indicating all that information fairly economically, and I discussed various methods with the publishers. It is impractical to put the notation underneath each word, as I did in §7.4 to illustrate the system, because it creates problems for the printers, so I suggested putting it in brackets afterwards. But that was too complicated and would put off the user, who needs to have it very clear. One must remember that language teachers are not always receptive to change in such matters. There is a long-standing and widespread belief that Daniel Jones said nearly the last word, or if he didn't, then Gimson did, and anything that looks different will deter people from buying the book. In the end I put a superscript numeral against vowels that vary in this way, as
d ι s r ι1 s p ɛ k t0 f ə4 l

The superscript 0 after the /t/ indicates that it may or may not be present when the word is spoken. In the key to the transcriptions I explained what range of vowels the numeral stands for, thus
ι1 = ι<->ə and ə4 = υ<->ə

I have found that by using 10 superscripts, with 6 vowels (mainly /ə/ and /ι/, in 20 combinations, I have been able cover most of the common patterns. In more cases than not, the variation is so small it is not worth putting in. I found it necessary to allow for nine variants of schwa. I stopped there partly for the reason that I didn't want more than one digit as a superscript, and partly because you come to a point of diminishing returns: there may be only half a dozen cases where number 10 is needed, and it's better not to clutter up a dictionary too much.

7.7 The example regulation.
We see from the phonetic representation
r ɛ  g j υ l ɛ ι  ʃ ə n
    / \ / \
that the first and third vowels, the /ɛ ɛι/, of regulation are protected, while the second vowel varies between /υ/ and schwa. For the final syllable I have said that the vowel can be anything between schwa and nothing, when the <n> becomes syllabic.

7.8 Unphonetic spelling criteria.
But consider a word like relentless, represented here as
r ι l ɛ n t l ι s 
 / \ / \

if you're trying to make the spelling more phonetic, You could change the <e> to <i> in the last syllable. However another consideration is the meaning of the word less: you may want to preserve the appearance of the word less at the end of that word, in which case you are applying another criterion besides the phonetic one. When one is considering how to make the spelling more consistent, there are other kinds of consistency beside pure phonetic consistency that one may want to take into account. And in the case of relentless we may say that to maintain the link with the word less, we want to spell the final syllable as <-less> rather than with an <i>, thereby also maintaining consistency with lent of the second syllable.

7.9 Conflicting criteria.
But before introducing non-phonetic criteria, we should reflect that English spelling is today so much in need of reform just because in the past a variety of criteria - fidelity to the source language from which a word has been borrowed is another such criterion - have been applied, and they often conflict. Historically, change of pronunciation has been another factor.


8.1 Children learn tone-units first.
There is a lot of research yet to be done here, but I am myself convinced that young children are operating the tone-unit aspect of language long before they pick up any actual segments of the language. This really is not surprising because I'm claiming that what happens here arises from the interactive situation, the awareness that people have of each other, and surely one of the strongest motivations of a young child that is beginning to speak must be the awareness of the growing relationship with the mother or another person. These things arise not from language as such, but from the normal interaction between people. Most research so far has examined what the child means in terms of grammatical units, but I think it would be much better if researchers asked, what does the child mean in terms of tone-units. But that is very speculative.

8.2 Chomsky-Halle thesis.
The Chomsky-Halle thesis says that the distribution of stress can be determined from syntax: you start off with the deep structures and transformations and so forth, and eventually arrive at a given stress-treatment for an utterance. I reject that because I don't think stress is predictable in this way. Stress relates not to the formal organization of an utterance, but from the function it is performing. I would claim that my analysis here provides an alternative explanation for all those cases where the Chomskyan nuclear stress rule works and it explains all those cases where it doesn't work.


9.1 Zero vowel-representation in CS.
The notation of a zero variant, as in ə<->0 for the final syllable of regulation, has affinities with the Cut Spelling representation of post-accentual schwa before <l, m, n, r>, which are spelt as syllabic in such forms as chapl, atm, fashn, propr, giving zero orthographic value to whatever vowel may occur before the final consonant

9.2 Spelling unprotected vowels.
A consideration that the Society has attached some importance to is limiting the amount of change. One practice adopted by the original New Spelling was to keep the letter <i> where it now occurs in an unprotected situation, as in vestige, but to regularize vowels in this position with <e> where other letters are now used, so that hostage becomes hostej. It has even been suggested that <e> should be used for all schwas, but there are two objections to this, one being the sheer number of instances, and the other being that all the schwas occur in unprotected syllables and at least 50% of them are subject to variation.

9.3 Spelling by citation form.
Another approach to schwa the Society has tried is to base spelling on the citation form of words, their pronunciation when spoken in isolation, and using whatever vowel-letter is suggested by the citation form if the vowel-sound is then other than schwa. One is to some extent driven to make an exception of schwa if one is not prepared to introduce new letters into the alphabet, for the simple reason that no unambiguous letter is available to represent schwa. It is very often possible to substitute a sort of non-neutralized, non-reduced form for an unprotected vowel without making it sound un-English. What you cannot do is alter the protected vowels. So if I say photography, I can represent it as /fəυtogrəfi:/, and it is still recognizable English. This may be an adequate basis for spelling, but it is not the pronunciation that a dictionary could recommend any foreigner to learn. There are nevertheless speakers who tend to avoid relaxed neutralized forms - I once met a lady who said: /aι kʌm from hærəυgɛιt maιsɛlf bʌt ov cɔ:rs aι hæv nəυ æksɛnt/; what she said was truer than she realized, since she was speaking nobody's English. But this raises the question, should the reformed spelling reflect such pronunciation? If it did, the reason would be not so much to reflect that particular accent as not to depart too far from the visually familiar forms of present spelling. Above all what the Society wants is a single consistent, easily learnable system which must cover a multitude of pronunciations. The 1948 6th edition of New Spelling by Daniel Jones gave up the attempt to alter the unstressed short vowels, giving the derivative pedantic for instance as a reason for not altering the <a> in pedant, another example of mixing criteria.

9.4 Has pronunciation changed?
When New Spelling suggested forms like kotej, hostej for cottage, hostage, one wonders whether these words were perhaps pronounced like that 40 years ago. One factor was that in those days there was a widespread feeling that if people pronounced a vowel as schwa, they were speaking badly, and spelling reformers then had to allow for that attitude. Indeed even today many first year undergraduates, when taught the phonetic alphabet, will stoutly deny they use schwa in that way, and that if you say evening there is no realization of the second vowel. They will say that is careless speech - but none of us really knows how we speak.

9.5 The <s> inflection.
In the case of the <s> inflection, as in the plural morpheme and present tense of verbs, I would argue that the voicing or devoicing was purely conditioned by the voicing of the previous consonant, and therefore need not be reflected in the spelling. The Society has actually opted for <z>, partly because in the vast majority of cases the sound is in fact voiced, and partly because it often needs to be distinguished from the unvoiced /s/, as in the pair hens:hence.

9.6 Morphophonemic <-bl>.
One of the more disconcerting features of Cut Spelling that has been discussed on the Society's Working Party is the <- able, -ible> morpheme, as in eatable, edible, whose suffixes have the same pronunciation, somewhere in the region of schwa. The unpredictable spelling of this suffix causes users great difficulty, and the Working Party has been considering whether they could both be reduced to morphophonemic <-bl>, with no preceding vowel-letter at all.

9.7 Omit all unprotected vowels?
While morphophonemic <-bl> appears to be satisfactory, it might be interesting to pursue the possibility further. Taking this procedure to its logical extreme, one might find that vast numbers of vowel-sounds which are fully predictable from their context need not be spelt at all. Mightn't the logical outcome of this be to spell only the protected vowels, and omit all the unprotected ones?

9.8 Visual prominence.
In fact the Society was at pains to preserve those in the most visually prominent positions, especially those occurring before the primary stress in words. On the other hand unprotected vowels that may be pronounced schwa are systematically dropped in Cut Spelling before final liquids and nasals when these can be represented by syllabic <l, m, n, r>. Examples with <r> cover a wide range of t.o. spellings, as in burglr, teachr, amatr, Cheshr, doctr, vigr, centr, murmr, injr, martr.

9.9 Final <-nt>.
Similarly in Cut Spelling the <-ant, -ent> endings are, merged as <-nt>, so that combatant becomes combatnt, on the grounds that the precise pronunciation of the vowel is unimportant, indeed it is variable, and so it doesn't need to be indicated in the spelling. The question is asked whether the grouping of consonants within such strings would be ambiguous, but the regularity of the <-nt> ending should ensure that that reader faces no uncertainty. For the writer the, regularization of this error-prone morpheme would be of great benefit.


Brazil, D C, The Communicative Value of Intonation in English, E. L. R. University of Birmingham, 1985