[Journal of the Simplified Spelling Society, 1989/1 pp13-19 later designated J10]
[Thomas Hofmann: see Journals, Newsletters, Bulletins.]

Showing Pronunciation in EFL teaching.

Thomas R Hofmann.

Thomas Hofmann [1] is Professor in the Faculty of Foreign Languages at Hokuriku University, Kanazawa, Japan, & previously contributed International Requirements for Spelling Reform to JSSS J7 1988/1. The following is a condensed version of an article which is to appear in the Journal of Hokuriku University & includes discussion of classroom teaching techniques. This article develops the concept of diaphones also discussed in several articles by David Stark & has important applications in defining spelling patterns in English, & how to devise a spelling system that is equally suitable for different accents.

Abstract.

Systems for the representation of pronunciation in teaching English language to foreign learners are based on the linguistic theories of nearly half a century ago. While linguistics introduced a number of invaluable techniques into language teaching & changed its emphasis from the written to the spoken language, its limitation to single accents & its exclusion of the written language do not serve the needs of practical language teaching well. Moreover linguistics has now abandoned the phonemic theory that serves as the basis on which pronunciation is represented.

Surveying the requirements of a system for representing pronunciation in language teaching, we identify a number of points over & above simply representing sound. The most important are that it should not conflict with orthography but preferably mesh with it as far as possible, & that it be valid at least for the 2 major standards of General American (GA) & British Received Pronunciation (RP). This second requirement led to defining a type of diaphonic representation, & showing how it applied in the severe test of English vowel sounds. The traditional system of long & short vowels is used, but expanded with 4 more unit vowels, giving rise to a system that applies to most accents with minor mergers. An English accent can thus be fairly accurately described by a characteristic set of diaphonic mergers, plus some distinctive preferences in word choice. A rough sketch of the linguistic analysis of this diaphonic system provides a surprisingly simple account of vowel features that works across accents, suggesting that these are the features that an English speaker actually learns to identify when meeting speakers of varying accents.

With minor additions such as a circumflex accent to mark 'continental' vowel-values, this system is found to be consistent with ordinary orthography. This allows a language learner to learn only one system of representation, 'spelling plus', that specifies both the spelling & the sound. Being equivalent to a phonetic representation with some extra marks, it can serve both the foreign learner & the native-speaking child learning to read, as well as supporting the teaching or learning of English spelling. It is found to be clearly superior to the traditional systems of phonemic representation used in English language teaching.

It is hoped that a future issue of the Journalwill describe this system applied in teaching English in Japan.

0. Introduction.

For all the effort devoted to teaching English as a second language, as a foreign language, & today as a world language, & for all the related research that is being carried out, rather little attention has been given to the best way of representing English pronunciation. Traditionally linguistic analyses have been used on the assumption that they are not only scientifically valid, but also appropriate to the teaching task.

However, the learner faces different pronunciations depending on accent, while these commonly used linguistic analyses each represent only a single accent, whether that of the US or of the UK. They start from the assumption that the language is the same everywhere. In fact communication takes place (sometimes imperfectly) despite different accents, rather as in Canada where French & English speakers may understand each other's language, although they do not speak it.

The result of this situation is that most course materials & most teachers only teach the standard accent of England or of the US. Students may then need to undertake considerable relearning if they travel to an area where a different accent is prevalent.

Of course a good program will introduce students to both standards - but at the risk of possible confusion & slower learning. In fact the effect of students being confronted with alternative accents is unknown, but it can hardly be beneficial. What is needed is a foundation which can apply to both dominant standards, as well as allowing for other sometimes quite serious variations the student will encounter in daily life.

There is furthermore fluidity in the accents used & understood by native speakers themselves: they vary their accent according to the linguistic situation in which they find themselves (Labov), speakers of the standard accents understand local variations, & speakers of local accents understand the standard accents.

In teaching foreign learners it would therefore be desirable to use a phonetic representation that can allow for these variations. Fortunately the standard orthography goes a considerable way towards providing such a phonetic representation, & must be learnt anyway, as the student must learn to write.

We shall here attempt first to determine what features such a script needs, & provide a theoretical basis for it. We shall conclude that the traditional distinction between 'long' & 'short' vowel values is the key to such a system, which is superior in many respects to the linguistically-based representations of pronunciation at present widely used.

1. Showing pronunciation in teaching English.

Since English orthography is not a reliable guide to pronunciation, it is useful to have an alternative system when teaching foreign learners. It can be used by students when referring to dictionaries & by the teacher in the classroom. It allows the students to take notes [2] on the sounds, to do exercises on the pronunciation without supervision, & to be tested on that knowledge; furthermore it enables the teacher to refer to the sounds simply & unambiguously. Above all it enables students to form a clear idea of which sounds are the same & which are different, as is necessary if hearing & pronunciation are not to be defective.

The linguistically-based systems commonly used today have advantages & disadvantages. Their greatest problem is that they disagree & are not convertible between accents. We shall discuss particularly the system derived from Daniel Jones' An English Pronouncing Dictionary (for British varieties of English) & the one derived from Kenyon & Knott A Pronouncing Dictionary of American English (for North American varieties). We shall not consider how these systems are applied to non-standard accents, but will compare them as broad systems (henceforth called the J-system for Jones, & the K-system for Kenyon & Knott) with the system of 'long' & 'short' vowels [3]. Other linguistic analyses are that of Trager & Smith (Outline of English Structure) & of Chomsky & Halle (The Sound Pattern of English), which we shall call the T - & C - systems, though we shall have little to say of them as they have failed to find significant application in English teaching.

We will first survey the aims of a phonetic [4] System for indicating pronunciation in teaching English as a second language, & then apply the 'long'/'short' vowel distinction consistently to at least 90% of the syllables in running text, or 95% of the words in English. We shall then give a theoretical basis for such a system & explain why it is the best for language teaching: it is based on modern linguistic insights such as were not available 50 years ago, & adopts a wider perspective than hitherto - the English language as a whole, spoken & written, throughout the world.

2. Requirements of phonetic representation.

1. The prime requirement is of course accurate representation of pronunciation & the contrasts made. A system can also show non-contrastice variation. Such a system is easy to design, & linguistics has refined the techniques for doing so in the first half of this century.

2. The system should not mislead. J-systems that show tense vowels with a colon (e.g. tense & lax high front vowels shown as /i:, i/) can be misleading for students whose languages contrast long & short vowels. Japanese students need a lot of remedial work to cure the habit of pronouncing it, bit the same as eat, beat, only shorter. In fact the vowel qualities are significantly different, while length in English primarily reflects a following voiced consonant, as when the vowels in both bid, bead are longer than in either bit or beat.

3. The system should be easy to learn. Here a J-system is better than a K-system, as it has fewer symbols, & they are mostly familiar from the roman alphabet & have nearly the same values as in the student's native language. However the J-symbols strictly represent only a rather poor approximation to English sounds.

4. The system should show the student the underlying system of English pronunciation. Although the J- & K- systems show the off-glides of low vowels as in cline, clown, neither system shows the off-glides in the mid tense vowels (as in rain, loan, or more clearly in final position in ray, low). Although not all speakers always pronounce the off-glide medially, especially not before <r> (Scots do not), & it is quite difficult for a native to perceive the glide except finally, students from languages without off-glides should not be misled into ignoring them by symbols which stand for pure, unglided vowels in their native language.

As students in Africa, Asia or South America are particularly likely to encounter non-standard speech, including British or American regional accents, they must be able to accommodate readily to other varieties. The traditional linguistically-based J- & K-systems offer no help here, since they presume a uniform accent. Today it is realized that accents vary, & that foreign learners must be able to cope with variation. Dictionaries & textbooks have to aim at a world market, & be usable regardless of the teacher's accent, or the accent of the student may be aiming to acquire.

To overcome this problem, Daniel Jones proposed the concept of 'diaphones' to represent all the contrasts in at least the major varieties of English. However, different accents contain different contrasts, & the task was logically insolvable within the then prevailing theory of phonemes. It is however not difficult to devise a set of diaphones valid for a pair of accents, although some diaphones may be differently pronounced in the two accents.

5. The phonetic system should indicate all the contrasts in British 'Received Pronunciation' & a US standard such as 'news broadcast pronunciation'. Most native speakers understand them & may be able to approximate to one of them, and indeed most other varieties relate to these in fairly simple ways.

6. Since the students must also learn English spelling, the closer the phonetic system is to the orthography, the less additional learning there will have to be. The old linguistically-based J- & K-systems do not begin to meet this requirement, since they aimed only to represent the sounds & were perhaps deterred from using the orthography by its very irregularity. More recently, Chomsky & Halle, [5] & perhaps Bolinger, have attempted to link English orthography with morphology & phonology, but this notion has yet to be applied to the practical needs of the classroom.

7. Ideally, the phonetic notation should be familiar to ordinary users of the language. Not only could teachers then be trained more easily, but students could then get information from untrained people. If a student who is unable to distinguish the vowels of boat & dog were to ask whether ogre is pronounced with an open or closed <o>, or with an open/low or mid/non-low vowel, few people, teachers even, would understand the question. On the other hand the distinction between long & short <o> would be far more readily understood. A system that does not require technical terms will therefore be more useful, in that students could have their queries more easily answered by available non-specialists.

8. The system should be simple enough for children to use. The J- & K-systems are indeed simple, but Chomsky's would require special training; the system used in the 2nd Webster's is also complex, in that it expands the traditional 'long' & 'short' vowel values to some 30 special diacritic values, with an explanatory key needed on each page. Thus we should seek a system with a small number of easy-to-interpret symbols: these two goals tend to conflict, but we hope to be able to show that such a system can be constructed.

9. Very importantly, the symbols should be available on ordinary typewriters, or else easily obtainable. The J-systems come much closer to satisfying this requirement than do the K-systems.

10. It should be possible to annotate existing texts easily, so that the teacher can prepare self-pronouncing material & the student can annotate text when required. The benefit of using such notation can be judged from the experience of using traditional phonemic transcriptions: the student is constantly presented with the correct enunciation. Although phonemic transcription may be necessary for learning unwritten languages, it is rare in English because the typefaces are hard to find & printing expensive. Furthermore, the student must learn the conventional spelling anyway. If however standard spelling can be annotated to show the pronunciation, ordinary typing & printing can be used, & the texts annotated.

These requirements are of varying importance, & in some countries, with students of some ages or having specific needs, not all of them will be relevant. But in general, they all contribute to efficient teaching.

3. Orthographies, phonemic & pronouncing.

The J- & K-systems are based on the principle of one symbol for each phoneme, that is for each sound or class of sound by which words can differ. This concept is largely abandoned today for two reasons, one linguistic (units of sound are not in general segments in time) & one pedagogic while 2 sounds may contrast in some positions a word, elsewhere they may not do so. For example, the vowels of eat, it are quite different, but they do not contrast before <r, ng>, nor, for some people (speakers of my native accent for instance), before <n, l> either. This means that such a sound can be equally represented by the symbol for either phoneme. If however only one symbol represents the correct spelling, the choice is arbitrary. No alphabetic system permits only the significant distinctions to be written. [6]

The advantage of 1-for-1 sound-symbol & symbol-sound correspondence (if it can be devised) is that learning to read & write it is easy. However, in most languages (including English) there are necessarily sounds which can be written in more than one way. The writing system of English at least cannot be reduced to such absolute simplicity.

As far as spelling reform is concerned (whether for general use or for teaching children), many researchers now accept that a single symbol for each distinctive sound is not of primary importance (though some still cling to the phonemic ideal); but they do insist there should only be one pronunciation for each symbol. Reading such a system is unambiguous, but writing is less straightforward. Thus English has two normal ways of writing /f/, as either <f> or <ph>. If <ph> is learnt as a 'fancy' <f>, reading is no problem, & writing faces only the minor problem of remembering which words require <ph>. This kind of pronouncing orthography' is useful both for native & foreign learners.

Alternative symbols for some sounds in some environments do not cause serious problems. This indeed is exactly what is needed to define diaphones, as suggested by Jones: symbols valid for more than one accent. While the vowels in hot, father are distinct in England, they are usually the same in North America. A phonemic system valid for both RP & GA needs different symbols for them: the single sound in North American English therefore requires two symbols. Let us consider this more closely.

4. Theory of diaphonic systems.

We are here concerned with phonological features only as aids to teaching & learning. We shall therefore talk almost entirely in terms of the classical segmental phonemes or 'broad phonetic' transcription. Diaphones can also be described in terms of phonemes. The terms vowel & consonant are used here ambiguously for letters & sounds (phonemes or diaphones) - the context should make clear which is intended, with long & short designating letters & high, low, back, tense describing sounds.

For any pair of accents, a diaphonic system can be established whereby one accent is represented by the merger of some contrasts, & the other accent by the merger of other contrasts. In general one symbol is used when both accents agree, & two symbols when one of the accents contrasts two sounds, but the other accent does not.

This procedure can be applied to any number of accents, but with larger numbers the diaphones required may be too many to be easily grasped. It is therefore convenient for language teaching purposes to restrict ourselves to the two most standard & explicity described, Received Pronunciation (RP) as defined by D Jones & General American (GA) as defined in Kenyon & Knott.

As well as listing the diaphones, we need rules to specify which of them merge in which accent. We shall here describe the mergers informally.

'Exceptions' arise if a word is pronounced with a different phoneme in the two accents (mathematically, forming an equivalence class of one). Thus apricot has the long <a> of able, apron in England, but the short <a> of apple, apprehend in North America. As well as single words, we also find groups of words [7] forming exceptions.

A diaphonic system will be nearly valid for most other accents that derive from the same origin, so the choice of these two accents leads to a diaphonic system that is valid for most varieties of English.

5. A diaphonic system for English vowels.

The main phonemic differences between English accents concern vowels rather than consonants, so let us concentrate on this most difficult area.

It is convenient to begin by establishing the inventory of phonemes, & then seeing how the accents in question differ in corresponding to it. The traditional orthography is a useful starting point, since it tends to reflect the state of the language before the major accents diverged.

The vowel sounds in beet, bit, bait, bet, but, boat, bite, bout have nearly the same distribution, contrasts & pronunciation in both RP & GA. We can mark the vowels as 'long' & 'short' in the traditional way with macrons & hooks to create our phonemic system as follows:

vowel
A
E
I
0
U
OO
SHORT
băt, căt
bĕt, sĕt
bĭt, sĭt
bŏttle, rŏt*
bŭt, rŭt
look, wood
LONG
bāit, rāte
bēet, sēed, cēde
bīte, sīte, sīght
bōat, cōat
beaūty, cūte*
loop, mood*

Items marked * to be discussed below.
[the oo of 'look, wood' shoud hav a breev over both letters; the oo of 'loop, mood' shoud hav a macron over both letters.]

There are other vowels in English. To 3 of them we assign digraphs functioning exactly like unitary vowels.

OItoy, boilAUlaw, autumn OUcow, house
['oi, oy, ow, ou' should each have a macron over both letters; 'aw, au' should have a curve over both letters.]

Are these perhaps really simple vowels followed by off-glides? It is characteristic of many English vowels to be glided rather than pure, as with the off-glide in long-<i> & (barely perceptible except at the ends of words) long-<o> & long-<a>, while long-<u> has a preceding on-glide. Many linguists have tried to analyse the glided vowels systematically as a combined vowel plus glide (e.g. the C- & T-systems), but long <u> with its on-glide does not fit the pattern, & shows that all these vowels are elementary units of pronunciation.

More problematic for a diaphonic system are vowels before <r>, the low vowels & the high back vowels.

Whether silent (as usual in England) or pronounced (as usual in North America), postvocalic <r> does not cause a serious problem to the diaphonic system: we simply note that RP pronounces it as schwa, thereby lengthening a preceding low vowel, so that <or> merges with <au> & <ar> merges with continental <a>.

The low vowels are more difficult, as RP & GA distribute words differently between the three varieties: front, 'central' (back but unrounded) & back (rounded). RP has central <a> in bath, pass, can't where GA has front <a>, & hot, top, lock have central <o> in GA, but back <o> in RP.

Orthography resolves the difficulty however. Short <o> is back but becoming central (i.e. -Round) in most American accents (except before <g, ng> & the voiceless fricatives <f, s, sh, th>). Short <a> is front (mostly [æ] in GA, [a] in RP), but becoming central (+Back) in RP before <th, n, s> & elsewhere. Exceptions are words with central <a> in both accents (father, shah & some foreign words), which require an extra diaphone, the central low vowel of continental <a> in all accents. We thus have 3 low diaphones, but just the two sharply contrasting phonemes of short-<a> & short-<o> in either accent.


Low vowels
Front
short-<a>
Central
cont.-<a>
Back
short-<o>
 RP → ← GA 
 /æ, a ɑɔ/

The chart assumes RP short-<a>→continental-<a> is an allophonic (non-contrastive) variant before <f, n, s, th>. RP (but not Australian varieties) does have some contrasts:

+Back:
-Back:
pass
lass
castle
hassle
dance
manse
slander
gander
staff
gaff
path
hath.

As there is variation between accents, & as pronunciation is changing & the words with front <a> are the rare ones, the distinction can be ignored in the diaphonic system for teachings.

Traditional orthography again offers a solution with the high back vowels. The long-<u> loses its on-glide always after <r>, mostly after <l>, often after <s>, & for many Americans after <n, t, d> (though its total loss gives the impression of stupidity, as in this noo toon is sooper, & is not needed in the diaphonic system, though it is useful for spelling-consistency. The back vowel <oo> is centralized in selected words before <k, t, d>.

This completes a diaphonic vowel system best described in 3 classes, 'short', 'pure' & 'glided':

  FrontCentralBack
'Short' (not tense, not glided) vowels
 High
Mid
Low
short-<i>
short-<e>
short-<o>
short-<oo>
short-<u>


short-<o>
'Pure' (tense, but not glided) vowels
 High
Low
long-<e>
cont.-<a>
long-<oo>
<au>
Glided towards high front:
  -Low
+Low
long-<a>
long-<i>
<oi>
Glided towards high back:
 -Low
+Low
long-<u>

<ou>
long-<o>

We here have an adequate diaphonic system for vowels & <r> in RP & GA that is nearly valid for most other accents. The consonants only differ in a few words, except that some British & American accents merge <wh, w>.

The symbols we have assigned to the diaphones (conventional letters with a diacritic) have little phonetic substance, their realization varying significantly between accents. Nevertheless, we see in the table that they retain phonetic features to a surprising degree regardless of accent: the 'short' values are all -Tense, -Glided & form an ordinary 6-vowel system, while the traditional long values are all +Tense, some pure (+Glided) forming a standard 4-vowel system (some glided), 2 more or less standard 3-vowel systems, one with glides to the front, the other to the back. English-trained ears must note these features when listening to other accents; & foreign students must learn to do so.

6. Matching with traditional orthography.

The traditional letter-names & -symbols of the English alphabet have survived because they are appropriate to the phonology of English, known at least subconsciously to adult native speakers. As Chomsky & Halle showed, they are reflected in morphological alternations such as sane: sanity, tone: tonic, reduce: reduction. Furthermore, many diaphonic exceptions (e.g. apricot, dahlia, economic, italics, leisure, patriot, privacy, vitamin, zebra) vary between short & long pronunciation of the vowel in different accents. Apart from a hundred common irregularly spelt words, these diaphones have unambiguous, predictable spelling patterns. The major patterns are as follows:

A single vowel is short before 2 consonants, or word-finally before 1 consonant (counting <ph, sh, th, st> & consonants followed by <l> or <r> as single consonants). A vowel is usually long before 1 consonant followed by a vowel, & always long immediately before a vowel or word-finally (except <a, e, y>). We see these patterns in fade: fad, waste: vast, table: rabble, maple: apple, apron: appropriate. Special patterns occur word-finally & in function words (by, me etc), especially final silent <e>, which may indicate a long vowel or distinguish homophones.

Most long-vowel digraphs are unambiguous & do not need marking except for beginners & in dictionaries; thus <ai>=Iong-<a>, <ee>=Iong-<e>, <igh>=Iong-<i>, <oa>= long-<o>, <ue>=Iong-<u>. Less common patterns such as <eigh, eign> for long-<a> can be classed as exceptions.

Nevertheless, English spelling can hardly be called basically systematic. Systemically ambiguous are the endings <ow> (contrast low: cow) & <ea> (contrast bead: bread). Even here some patterns appear (<ea> is normally short-<e> before <d, th>), though they may not help the learner, & the diaphones need to be shown by the usual diacritics.

7. Diacritics as an aid to pronunciation.

In recent centuries English has not adapted the spelling of new loan-words to conform to English patterns. Most source languages (e.g. Italian, Spanish) have standard 'continental' values for the vowels, which gives a third value in English beside long & short, and has to be shown by a special diacritic. The circumflex accent, being associated with foreign words, seems most suitable. We thus have:

cont-<a> â (low central vowel as in shah)
cont-<e> [9]
cont-<i>
cont-<o>
cont-<u>
ê =
î =
ô =
û =
a (long-<a>)
e (long-<e>)
o (long-<o>)
oo (long-<oo>)

We can now mark nearly all vowels in ordinary spelling to show the pronunciation by means of these 'pronouncing diacritics'. Two more marks are needed: <x> over a vowel to show that it is irregular with neither long, short, nor continental pronunciation; & a slash through a vowel to show it is silent, as with the second vowel in every.

Remaining problems include the common <o> & occasional <oo> pronounced as short-<u> (come, son, blood). These can be shown by an umlaut or diaeresis (<ö> = short <u>): wörd, wörk, wön. Similarly, <a> with the value of short-<o> could be given an umlaut: wär, wäsh, swän, though I have not so far tried this in the classroom.

This essentially simple system of diacritics for long, short, continentl, & modified vowels (despite a few irregular forms to which it cannot apply) could equally be used for native-speaking children & foreign learners.

Like the i.t.a., it has only a single pronunciation for each symbol (= letter + diacritic) for all major accents, but there are alternative spellings for each sound. Writing therefore remains difficult, but learning to read is much simpler, with easy transfer to conventional spelling. Parents, teachers & children have no difficulty in reading either annotated or conventional script.

Though much simpler than reading ordinary text, the rules are still complex, & do of course have to be learnt, the system being further complicated by the rather large number of exceptions. The teacher marks each fully pronounced vowel with the traditional macron or breve for long or short values & a circumflex accent on continental values. Unmarked vowels are ignored (unpronounced) at the ends of words or beside another vowel, or they are pronounced with reduced value if surrounded by consonants.

This diaphonic system for indicating pronunciation can be combined with marking for rhythm & stress. Each complex symbol of vowel + diacritic has a single pronunciation, while unmarked letters are effectively decorative additions distinguishing homophones. The writer, of course, still has to learn all the letters required in TO. Transfer to TO for the reader is even easier than in i.t.a., since there are no changes in the shapes of letters or words, but only the disappearance of the diacritics.

In language teaching the system serves as a phonemic transcription (though with some extra letters), it is valid for most accents, it lends itself to EFL publishing (especially dictionaries), & it is less misleading than an international set of symbols.

Regular spellings are taught by removing the diacritics, which are only left in irregular cases; in this way the student confronts the irregular forms independently.

8. Comparison with J- & K-systems.

We will now compare the D-system with the J- & K- systems according to the EFL/ESL requirements listed in section 2 of this paper.

1. As explained, this D-system (diaphonic) was devised to represent both the GA & RP standards of pronunciation, with different mergers for the low vowels, vowel + <r> & long <u> with <oo>, but it can apply more or less to most dialects. For example, it is valid for Scots with a merger of long- and short-<oo>, and again for Northern England, where short-<u> is merged with short-<oo>.

2. Unlike the J-system, the D-system does not mislead by overtly symbolizing some vowels as longer than others. It is true the traditional terms 'long' & 'short' can mislead in the same way, but we can avoid confusion by describing the marked vowels as macron: breve, or bar: hook.

3. The D-system is much easier to learn than the K- system, needing only 4 new symbols, the diacritics. The long values are known from the letter-names, & the short values are learnt from the earliest words, while students often know the continental values of vowels from their native language. Learning the diacritics may appear harder than the 3 new symbols of the J-system (:, ɔ, ʌ), but the latter also entails learning different qualities for <e, e:>, <i, i:>, <u, u:>.

4. The D-system is far superior to the earlier systems in enabling the student to perceive & use the essential phonological structure of English in terms of its vowel-alternations. It makes clear that in English short-<i> relates to long-<i> rather than to long-<e>, although the student's hearing & the old phonetic systems of representation (long- <e> written as /i:/) both suggest otherwise. The student will then accept such long: short variations as are heard in actual speech (apricot, italic with the initial vowel pronounced either long or short).

5. By being applicable to the main accents, a D-system enables the student to cope with the different realizations of words likely to be heard. This the J- & K-systems cannot do, as they only represent a single accent.

6. A D-system is consistent with TO: the same representation of words teaches both their spelling & their pronunciation. The inconsistency of the J-, K- & T- systems, as well as of Webster's 3rd New International Dictionary in this respect is particularly striking in their representation of the back rising glide spelt <ou> [10] ( in TO (out around about the house) as <au>, which in TO represents the low back unglided vowel (audition, bauble, bawdy, awful, saw). While </au/> is phonetically justified (the sound s= with [a]) & may be used in the student's native language, it causes unnecessary confusion between spoken & written English, which leads to distrust of both the spelling & pronunciation systems of English.

7. A D-system is also known to users of the language, whom the student can consult in the teacher's absence. Teachers are also more easily trained in it. Educated native speakers recognize the circumflex as representing continental vowels, & they are generally aware of the concept of 'long' & 'short' vowels, however problematic (e.g. some US systems represent long vowels with upper case, short vowels with lower case letters). Moreover, a basic rule for line-end hyphenation distinguishes long & short values, a hyphen being permissible immediately after a long vowel but requiring an intervening consonant after a short vowel.

8. While the J- & K-system rules indicate non- contrastive variation, a D-system has some unordered merger rules for different accents. The D-system is however simple enough to use with young children, indeed perhaps as efficient as the i.t.a. when its compatibility with TO is taken into account.

9. The diacritics are not available on ordinary typewriters, but they are available in many printing fonts. The non-roman special characters required by the J- & K- systems on the other hand are hard to obtain.

10. In any case, a D-system has the advantage that the teacher can annotate textual material as required for teaching, & specially printed material is unnecessary. For beginners the teacher can annotate all vowels, or perhaps just new words. Students can be tested by being asked to enter the diacritical marks themselves - a much simpler, less error-prone task than writing a phonemic transcription.

A further advantage of a D-system is that the vowels all have distinct & well-known names, which is not the case for all the non-roman symbols in the IPA.

9. Conclusion.

For teaching English as a foreign or especially a 2nd language, a D-system has many advantages over the usual J- and K-systems, & we may hope to see dictionaries & materials use it. It also has clear advantages in teaching native children to read, not least in allowing publishers to indicate pronunciation - as they dare not do at present.

This D-system also sets up the facts that any reformed spelling system must face: who pronounces which sets of words alike & who pronounces them differently, as well as identifying which present spellings are total exceptions, which are examples of limited patterns, and so on.

10. Samples of text annotated with diacritics.

In the first text mainly the stressed vowels are marked for pronunciation as described in the above article.

Rĕpresĕnting pronŭnciātion in tēaching Ěnglish

Critēria of an ădequate rĕpresentātion of pronŭnciātion for lănguage lĕarning are māde explĭcit. The linguĭstically bāsed rĕpresentātions ūsed in tēaching Ěnglish are thĕreby sēen to be clēarly inădequate & a pârtial solūtion is found in the tradĭtional nōtions of 'lŏng' & 'shŏrt' vălues for vowels. Howĕver, mīnor extĕnsions are nĕcessāry, as wĕll as a theorĕtical foundātion, & mĕthods for its ĭntrodŭction in the clăssroom.

[N.B. ou in found and ow in vowels should have joined macrons above both letters.

In the following text all vowels should be marked. There should be a single dot to indicate an obligatory schwa-syllabel above the italicized vowels in 'a, the, dialect, another, incompatible, even, applied, assumption'. Strikethru marks should be vertical. There should be an x over the italicized vowels in 'language, to'. The y of 'underlying, by' should have macrons. These digrafs should have macrons: pointed out.]


Ŏf cōurse, thĭs ĭs ŏbvîoŭslŷ nŏt trūe önce ĭt ĭs pointed out: whīle a lănguage stūdĕnt mŭst fāce dĭfferĕnt prōnŭncîātiöns & dĭfferĕnt dīalĕcts, the lĭnguĭstĭc mĕthŏds thăt prĕsĕnt lănguage tēachĭng ĭs bāsed ŏn wörk fōr ōnlŷ öne sĭngle dīalĕct ōr anöther. Ĭn făct thēse mĕthŏds lēad to quīte ĭncŏmpătible rēsŭlts ēven whĕn applīed to dīalĕcts ăs sĭmĭlâr ăs the ăccĕpted stăndârds öf prōnŭncĭātiön ĭn the ŪS & the ŪK. Sŭch rēsŭlts fŏllōw nătūrallŷ fröm the assŭmptiön ŭndĕrlyĭng nēarlŷ âll öf mŏdĕrn lĭnguĭstĭcs, thăt the lănguage ĭs shāred by a cŏmmūnĭtŷ öf spēakĕrs.

Notes.

[1] My thanks to Chr. Upward not only for discussion & encouragement but for the difficult task of cutting the length while preserving the essence, & to K Okana for comments leading to the improvement of this exposition. This work was begun in 1961 at the Institute for Exceptional Children, University of Illinois, under L Stalurow, aiming to find easier ways to teach English spelling to native children.

[2] A mature learner denied an adequate medium will use an ad-hoc creation, perhaps based on the sounds of his own language, & surely inadequate when he does not already know English.

[3] A moderate extension of this system is found in Fowler et al.'s Concise Oxford Dictionary & further extensions in the 2nd edition of Webster's New International Dictionary. See also The Random House Dictionary.

[4] We shall call them all 'phonetic systems', without suggesting thereby that they should or do indicate non-distinctive variations, as the only other terms available are either awkward ('phonological system') or unpopular if not discredited ('phonemic').

[5] It is perhaps best seen as a philosophical 'position', the antithesis of the phonemic J-, K- & T-theories that assumed English spelling was irremediably unsystematic.

[6] It is possible in principle for languages with a CV structure, or if written with symbols for syllables, but even in such languages, it may not be possible, as e.g. Japanese ji can be written as voiced varieties of either shi or chi.

[7] E.g. the <-alm> words balm, calm, palm, psalm, with continental-<a> in RP & GA but <au> or <au + l> in some accents.

[8] That is, if centralizing one of the lower (-Back) group is not felt to be a serious mispronunciation.

[9] An <e> in a foreign word may also be anglicized as short-<e>, but we already have a symbol for that, so another symbol is unnecessary. The 1st <e> in fete, for example, is given a short-<e> value in North America, but a continental-<e> pronunciation (i.e. long-<a>) in England.

[10] We here take <w> as replacing <u> word-finally.


References.

Abercrombie, David. Elements of General Phonetics (1967) Edinburgh University Press.

Betts, E A. 'Learning word-perception skills' (1983) in Spelling Progress Bulletin 23 No.2: 6-9 & No.3: 10-12.

Betts, K P. 'Language, orthography & the schwa' (1979) in Spelling Progress Bulletin 19 No.3: 11-16.

Chomsky, Noam. 'Phonology & Reading' (1970) in Levin H & Williams J P (Ed) Basic Studies on Reading, New York Basic.

Chomsky, Noam & Halle, M. The Sound Pattern of English (1968), New York: Harper & Row.

Downing, John. 'Linguistic awareness, English orthography & reading' (197?) in Journal of Reading Behaviour 10 No.l: 103-114. (1979) Spelling Progress Bulletin 19 No.3: 5-10.

Fowler H W & Fowler F G & McIntosh (Rev). The Concise Oxford Dictionary of Current English (1911, 1951: 4th) Oxford University Press.

Hofmann Th R. 'Music, verse & supra-segmental aspects of language' (1982), Memoirs 11: 49-68, Senzoku-Gakuen, Kawasaki, Japan.

Jones, Daniel. An English Pronouncing Dictionary (1917, 1956: 11th), Cambridge University Press.

Random House Dictionary of the English Language, Flexner, Stuart B (ed) (1987: 2nd), N.Y.: Random House.

Scragg D G. A History of English Spelling (1974), Manchester University Press & New York: Barnes & Noble.

Shorter Oxford English Dictionary on Historical Principles, Onions C T (Ed) (1973: 3rd), Oxford University Press.

Stubbs, Michael (ed). The Synchronic Organization of English Spelling (1986), Committee for Linguistics in Education, BAAL & LAGB, England.

Takebayashi, Shigeru. 'Eigo-no tsuzuriji-niyoru hatsuon hyouki hoshiki (Pronunciation system based on English spelling) (1) - (3)' (1973-76) Lexicon 2: 13-27, 3: 28-44, 5: 53-72.

Takebayashi, Shigeru. Eigo-no Fonikkusu: Tsuzuriji & hatsuon-no ruuru (English Phonics: Spelling and Pronunciation Rules) (1981) Tokyo: Japan Times.

Trager, George L & Smith, H L. An Outline of English Structure (1951) Norman, Okla: Battenburg (Studies in Linguistics: Occasional Papers No.37).

Trudgill, Peter. International English (1985, 2nd) London: Edward Arnold.

Upward, Christopher. English Spelling & Educational Progress (1988), Working Paper No. 11, Committee for Linguistics in Education BAAL & LAGB, England.

Webster's New International Dictionary (19?? : 2nd) Springfield: Mass: Merriam. (1961: 3rd)

Wells, John C. Accents of English (1982) Cambridge University Press.

Wells, John C. 'English accents & their implications for spelling reform' (1986), in Simplified Spelling Society Newsletter, J3 Summer p.5-13.


Back to the top.