(Journal of the Simplified Spelling Society 1987-1, pp11-17 Later designated Journal 4]

A New Orthography on Morphophonematic Principles.

Edward Smith.

Edward Smith first became interested in spelling reform when at the age of five he wondered why three was not spelt the same as free. He was secretary of the Simplified Spelling Society from 1971 to 1972, and chairman from 1972 to 1976, as well as a transport planner for Great London Council. He later lived in San Francisco, CA, USA.


The work in this paper is based on these premises:
1. The traditional English orthography is an unnecessary and unacceptable obstacle for children learning to read, write, and speak English.

2. A new orthography should treat the many accents of Standard English as variants of a single language, as far as possible accounting for phonetic variation by differences of phonetic realization rather than differences of underlying representation.

3. The new orthography should be linguistically as simple as possible to ensure ease of learning. This simplicity cannot be achieved by chopping and changing the traditional orthography. Truly simple spelling must build on a rigorous and comprehensive analysis of the sound patterns of the language.


1.1 Phoneme fundamentals.
1.11 A speaker of Hindi learning English might wonder at the use of <t> for 2 different sounds in <stop> and <top>. To an Englishman, however, the difference between these sounds is not significant because it is decided by the phonetic context and is never the only phonetic difference between words. A set of sounds with complementary distribution is a "phoneme", and is enclosed between /slasciz/. A notation that groups sounds into phonemes is "phonematic". The various sounds belonging to a phoneme are "allophones".

Note: The word "phonemic" is more common than phonematic", but is bad Greek.

1.12 The <i> in <medium> has two pronunciations. No native speaker of English attaches importance to the "free variation" between these pronunciations: good reason to suspect that the 2 sounds of <i> in <medium> are allophones of one phoneme. This theory will be sustained if no pair of words, or "minimal pair", can be found that are different words, in some accents consistently distinguished phonetically by only these two sounds.

1.13 The <a> of <par> is for many Americans phonetically similar to the <o> of <pod>, while many Scots and Irish would identify it with the <a> of <pad>. However, in the word <parity>, where the first element of meaning or "morpheme" is clearly the same as <par>, the vowel is in every accent phonetically more similar to the <a> of <pad>. Therefore it is preferable to analyse the <a> of <par> as an allophone of /a/. A phonematic analysis designed to minimize the disturbance to morphemes is "morphophonematic". The morphophonematic spelling of a word is its "underlying representation".

Note 1: The American will instinctively read his own pronunciation into <par> because any other allophone of /a/ would give a sound sequence not possible in his speech.
Note 2: Traditional orthography is sporadically morphophonematic, but compare <deign> and <disdain>, <line> and <align>, <affray> and <afraid>.
Note 3: See also Harris (1951), Appendix to 7.4.

1.14 The depth of a morphophonematic, analysis must be appropriate for the purpose. For example, it would be possible for <brother> and the stem of <fraternal> to have the same underlying representation, but the rules for the phonetic realization would be too complicated for a practical orthography.

1.2 Internal open juncture.
1.21 Allophones otherwise found only at the beginning or end of words will appear at certain morpheme boundaries within words. This feature is known as "internal open juncture" and can be shown by the location of stress marks /'/, /-/) or the sign /+/. The phonetic importance of internal open juncture depends on the difference between allophones, which varies among accents. A phonetic difference between /row+d/ <rowed> and /rowd/ <rode> is obvious to a Scot but unknown to a Londoner. A sequence of allophones with neither pause nor open juncture is "close juncture".

night rate
tarry (adj)
furrier (adj)
tarry (verb)
furrier (noun)

1.22 In the Midlands, North Wales, South Lancashire, and New York City [ŋ] is always followed by /k/ or /g/ and this sound is an allophone of /n/. The contrast other accents show in /sing+r/ <singer> and /fingr/<finger> is the result of internal open juncture.

Note 1: The rime <singer: finger> in the Midlands accent is explained not by an absence of internal open juncture in <singer> but rather by the absence of an allophonic differentiation of /ing/ that would be necessary for an underlying internal open juncture to be phonetically evident.

Note 2: There are phonologic reasons for preferring to treat [ŋ/ in the non-Midlands pronunciation of <ring> as the realization of a cluster analogous to /mb, nd/ rather than a simple sound analogous to /m, n/. Like /mb, nd/, and unlike /m, n/, [ŋ] never occurs initially and is never syllabic, /wiiknd/ <weakened> being shorthand for /wiikn+d/.

1.3 Syllables.
1.31 Speech sounds may be classified according to whether they are "syllabic" or "non-syllabic". Syllabic sounds are articulatorily longer, and auditorily more prominent than non-syllabic sounds. in the word /top/, /o/ is syllabic, while /t/ and /p/ are non-syllabic. In /pinian/ <pinion> /i/ is once syllabic and once non-syllabic. A "consonant" may be defined as a phoneme with only non-syllabic allophones, a "semiconsonant as a phoneme with both non-syllabic and syllabic allophones, and a "vowel" as a phoneme with only syllabic allophones. From our previous examples it happens that /t/ and /p/ are always non-syllabic and therefore consonants. The phoneme /o/ is always syllabic and a vowel. The phoneme /i/ has both syllabic and non-syllabic allophones, and is a semi-consonant. The phonetic context of a sound may be defined as "presyllabic" if the sound is followed with close juncture by a syllabic sound. An "open syllabic" may be defined as a syllabic followed by either open juncture or a presyllabic sequence of non-syllables that can also occur at the beginning of a word. Examples are /a/ in /spa, plastr/. A "closed syllabic" is any syllabic that is not open.

1.32 The phoneme /r/ followed with either close or open juncture by a non-syllabic or by a pause, as in /mustr/ <muster>, is for Scots phonetically similar to /r/ in other positions, as in /ring, merit, fur+i/ <ring, merit, furry>. The [e] or [i] that the Scot inserts before [r] in /mustr/ is not phonematically significant in a transcription that distinguishes the internal open juncture in /mustr+ing/ <mustering> and the close juncture in /string/. In some other accents the phonetic differences among the allophones of /r/ can be considerable, but there is no difficulty in assigning these allophones to the same phoneme.

1.33 The phonetic difference between the endings of /feilir/ <failure> and /howli+r/ <holier> reflects open juncture in /howli+r/. Chomsky and Halle (1968) give <ingenious: genial> as an example of "near contrasts", but any contrast of syllabic and non-syllabic /i/ in those words is peculiar to their accent and not a regular feature of the language. Moreover, any attempt to assign syllabic and non-syllabic /i/ to separate phonemes would be embarrassed by free variation in <medium> and countless other words. Similarly, <w> in <wit> and <u> in <put> do not contrast, vary freely in /kaziwal/ <casual>, and should be assigned to a single semiconsonant phoneme, /w/.

Note: Non-syllabic /i/ does contrast with the /ii/ <ee> of /biit/ <beet>. Compare /iowk/ <yoke> and /iioliθ/ <eolith>.

1.34 In any combination of semiconsonants it is regularly predictable which semiconsonant will be syllabic. For /ir/ the /i/ is syllabic unless the combination is weakly stressed and posttonic. Thus in the Scottish pronunciation of /gird/ or /sir'kumfarens/ <circumference> the /i/ is syllabic, while in /feilir/ <failure> the /r/ is syllabic. Similarly, weakly-stressed posttonic /r/ is syllabic after /w/, as in /forwrd/ <forward>, but compare /fowr-wurd/ <foreword>, in which secondary stress preserves the /u/.

1.4 "Long vowels" and cluster analysis.
1.41 Fundamental to the analysis of a language is the question whether diphthongs should be treated as unit phonemes or as clusters of other phonemes. In English the morphematic relationship in <vain: vanity, divine: divinity> suggests that the "long vowels" in <vain, divine> are unit phonemes differing in one feature from the corresponding "short vowels" in <vanity, divinity>. However, in all other respects, including distribution, the diphthongs are clusters with /i, w/, perfectly analogous to clusters with /m, n, l, r/. Examples follow of such a cluster analysis:-


/saw, kawm/
/wwlf, bwwt/
saw, calm

wolf, boot

1.42 For the analysis of <beet> to be valid one must assume that <Yiddish> is a foreign word. The contrast in /liist/ <least> and the ending of /howli+ist/ <holiest> is the result of internal open juncture. <Inappropriate> must be analysed as /ina'prowpriat/, with the rule for some accents that weakly-stressed /iat/ is phonetically [i+it], not [iətl. Note that /ww/ includes 2 allophones, the <oo> of <boot> and the <wo> of <wolf>. The first allophone always follows close juncture or precedes /z/ or open juncture. The second allophone appears otherwise.

1.5 Syllabics followed by <r>.
We noted in §1.13 that the value of /a/ in /par/ is in many accents materially affected by a following /r/. This potential phonetic influence by a following /r/ applies to all syllabics. Economy, practicality, and elegance require that a phonematic analysis group a sound that occurs only before /r/ in the same phoneme or phoneme cluster as a sound that never occurs before /r/. The simplest procedure is reference to accents which morphophonematic evidence shows to be unaffected by the combinative development, in this case Scottish and some Irish accents. Examples follow of each phoneme and phoneme cluster that appears syllabically before /r/, with an example of the same phoneme or phoneme cluster not followed by /r/.






1.6 Syllables weakly stressed.

1.61 The <a> in <England> is not the same sound as the <a> in <filmland>. However, the first <a> is only found with weak stress, while the second <a> always has strong stress. Hence these 2 sounds have complementary distribution and can be assigned to the same phoneme:- /*ingland, film-land/. Indeed in a morphophonematic analysis it is desirable that the <a> in <principal> should be a member of the same phoneme as the <a> in <principality>:- /prinsipal: prinsi'paliti/. Similarly weakly stressed /e, o, u/ underlie what is phonetically a central syllabic in /aksident, kiwrios/ <curious>, /industri/. Compare /aksi'dental, /kiwri'oziti, in'dustrial/. For the purposes of an orthography it is not proposed that an underlying syllabic should be written if its phonetic realization is zero: - /kurij, fa'milir/ rather than /kureij, fa'miliar/ that would be suggested by /ku'reijos, famili'ariti/. All occurrences of the weakly-stressed central syllabic that have no strongly-stressed parallel can by convention be assigned to the /a/ phoneme: /a'jenda, ta'riin/ <agenda, tureen>.

1.62 Under weak stress open syllabic allophones of /i/ overlap with /e/. This overlapping or "archiphoneme" can be represented by /i/ except where morphophonematic patterns show underlying /e/.- /i'nuf, i'levn/, but /repe'tisian/ (not /repi'tisian/), /re'petitiv/ (not /ri'petitiv).

1.63 Similarly /o/ and /ow/ do not contrast when weakly-stressed open syllabics. This archiphoneme can be represented by /o/.- /sparo, o'bei, boro+d/.


2.1 Orthographic variety impractical.
English is commonly read 3 or 4 times as fast as it is spoken. Variety that gives colour to the spoken language would be impractical in an orthography suited for today's needs. It would not do for the Londoner to write [drawrimbawd] for <drawingboard> or the American to write [dwwdii] for <duty>. An orthography that follows the historic distribution of phonemes allows each speaker to read his own pronunciation into /drawing-bowrd/ or /diwti/. These words have each a single underlying representation. The phonetic realization in any accent can be predicted by the rules for that accent.

2.2 Traditional pronunciation preferable.
A few words have variants that can not be predicted by phonetic rules and require more than one underlying representation. The traditional pronunciation - that which preserves the rimes, metre, alliteration, and puns of English literature - may be the most acceptable basis for an orthography.


But she was up to sleek her clothes,
And would be sweet as any rose.
-Thomas Churchyard "Old Time Service", 1575

Wave Munich! all thy banners wave,
And charge with all thy chivalry!
-Thomas Campbell "Hohenlinden"


3.1 Consonant miscellany.
3.11 Some speakers pronounce [p] following the /m/ in the consonant clusters of /emti, pumkin, camfr, wormθ, semstris, re'demsian/ <empty, pumpkin, chamfer, warmth, seamstress, redemption>. This [p] is redundant in a phonematic transcription that distinguishes the close juncture in /*hamsted/ <Hampstead> and the open juncture in /howm-sted/ <homestead>.

3.12 Except following a pause, /t/ and /d/ followed weakly stressed and without a pause by any syllabic except /n/, regardless of juncture, are not contrasted in the speech of many Americans. Tho they pronounce /pating/ <patting> and /pading / <padding> alike, these Americans do contrast /pats/ and /padz/ <pads>. Therefore an orthography that does not distinguish /pating/ and /pading/ would be no simpler for Americans, while it would be much less satisfactory for other speakers of English.

3.13 Most accents contrast /nc/ in /benc/ <bench> and /nsi/ before a weakly-stressed syllabic in /rezi'densial/ <residential>.

3.14 Variation between /θ/ and /ð/ is heard in <brothel, beneath, booth, forthwith, therewith, though, smithy, herewith, logarithm(ic), rhythm(ic), wreath, with, withe, withdraw, withstand, withhold, within, withal, without>. The normal developments, recommended for the orthography, are:

/broðil, bi'niið, bwwð, fowrθ'wiθ, ðeir'wiθ, ðow, smiði, hiir'wiθ, loga(')riθ)m(ik), riθm(ik), riið, wiθ, wiθ, wiθ'draw, wiθ'stand, wiθ'howld, wið'in, wið'awl, wið'uwt/.

3.15 In a few words pronunciation varies between historic /z/ and irregular [s). For an orthography /z/ is recommended in /pro'fiwz, blwwz, benizn, baizan, di'fiwz, di'zern, keizmant, kon'saiz, kum'parizn, griizi, garizn, spuwz, sakrifaiz, sa'faiz, venzn, veiz, huzi, huzif, liiz, liizing, izoleit, o'biiz, o'beziti, orizn/ <profuse, blouse, benison, bison, diffuse (adj), discern, casement, concise, greasy, garrison, spouse, sacrifice, suffice, venison, vase, hussy, housewife, lease "glean", leasing "lie", isolate, obese, obesity, orison>, and always in the endings of /ver'bowz, ver'boziti/ <verbose, verbosity> and - after a vowel, /w/, or /i/ - in the endings of /i'liwziv, i'liwzori/ <illusive, illusory>.

3.2 [ʒ] and [∫]: origins
3.21 The distribution of the sound [ʒ] is very different from that of /v. ð, z/, suggesting that [ʒ] is not an independent phoneme. The parallel between /feil: feilir/ and [klowz: klowʒr] points to /zi/ as the phoneme sequence underlying phonetic [ʒ] in [klowʒr]. This analysis is supported by the absence in many accents of phonetic [zi] before weakly-stressed syllabics, and when [zi] does appear in this context it is always in free variation with [ʒ].

3.22 The sound [ʒ] in [dʒudʒ] <judge> cannot, however, be analysed as /zi/ because phonetic [zi] does appear before stressed syllabics and finally:- /pre'ziwm, deizi/ <presume, daisy>. But these instances of [ʒ] are part of the phonetic cluster [], which has a distribution similar to /b, d, g/ and will be defined as a unit phoneme /j/.

3.23 Most Scots and some northern English contrast /verjr/ <verger> and /verdir/ <verdure>, with non-syllabic [di] in the latter. Those accents with phonetic [] in <verdure> have phonetic [di] before a weakly-stressed syllabic only in free variation with [dz]. it will be assumed, therefore, that the underlying representation is always /di/, not /j/, when the Scottish pronunciation is [di]. In practice the traditional orthography is a trustworthy guide, underlying /j/ always corresponding to <j, g, dg,> except in the word <soldier>, which is always /sowjr, sojr, sowljr/.

3.24 The distribution and analysis of [∫] is closely parallel to [ʒ]. The word [pre∫r] is analysed as /presir/, [t∫] in [t∫ip] <chip> is defined as a unit phoneme /c/. Those accents which contrast /verjr verdir/ do not rime /tiicr: fiitir/ <teacher: feature>, pronouncing non-syllabic [tij in <feature>, while in other accents an underlying /ti/ is a second source of phonetic [t∫]. Some Irish have [t∫] from /ti/ before syllabics with primary stress, as in <Tuesday, tulip>.

3.25 There remain many instances of [∫] that do not follow [t] and do not precede a weakly-stressed syllabic. Consider however that while there is no case of initial /zb, zd, zg/, initial /sp, st, sk/ are quite normal, and the common appearance of [∫] in positions where native words have no corresponding [ʒ] suggests the realization of /s/ plus a voiceless non-fricative consonant (fricatives such as /f, θ/ only sporadically following /s/). The only voiceless non-fricative consonant not otherwise attested after /s/ is /c/. Therefore <fish, ship, English> will be analysed as /fisc, scip, inglisc/.

3.26 By analogy <beige, zho> can be analysed as /beizj, zjow/. A voiced fricative consonant followed with close juncture by a consonant is not normal, but then <beige, zho> are phonologically exceptional words in English.

3.27 Americans, New Zealanders, and a few English have phonologic /zi/ (phonetic [z/) rather than the normal /si/ (phonetic [∫] in <version, excursion, Persia>, and most other Latin words with presyllabic <rsi>, but not in <controversial>, nor in <inertia> and other words with <rti>.

Note: Formerly these pronunciations with [ʒ] were a feature of Scottish English (Walker, 1791).

3.3 /h/.
3.31 Palatal and velar fricatives, the <ch> in Scottish <bricht, loch>, can be assigned to the /h/ phoneme: /briht, loh/.

3.32 Presyllabic /hw/ and /w/, as in /hwet/ <whet> and /wet/, are contrasted by Scots, Irish, most Americans, and some northern English and New Zealanders.

Note: <Whelk> is /welk/ for "snail" but /hwelk/ for "pustule".

3.33 Every non-Germanic <h> was at first silent in words borrowed from French. Several words from other sources also have spurious <h> in the traditional spelling. The historic pronunciation of many such words is still common, particularly in Ireland and the southeastern United States.

(but /humbl-bii/)

"H"(the letter)

3.4 /r/.
Underlying /r/ is in all contexts phonetically significant in the accents of Scotland, Ireland, most of America, part of Wales, and southwest England and parts of the north. A few words have <r> in the traditional spelling but not the traditional pronunciation:

/welsc rabit/
/awnt ai/

Worstead, worsted
Welsh rarebit
aren't I


4.1 /ir, er, ur/
The contrast of strongly-stressed or pretonic /i, e, u/ before non-presyllabic /r/, as in /gird, serf, surf/ is preserved by most Irish and Scots and a few northern English. But as with the rest of the language the traditional spelling is a poor guide to die traditional pronunciation. A Dubliner pronounces <first birth> as /furst berθ/ and indeed Shakespeare rimed <first> with <accurst> and <birth> with <earth>.

Note 1: In some Irish speech, like 18th-century English speech (Walker, 1791), every non-presyllabic /ir/ coalesces with /er/.
Note 2: Some Scots substitute spelling pronunciations.

/ber / <birth>, /burd, burc, *burn/ <Byrne, Byron>, /twurl, durt, durk, durj, kirk, kun'ferm, kurb/ <curb, kerb>, /gird, girdl, girdr, gerθ, gerl, cirp, flurt, ferk, ferkin, ferm/ (adjective), /fur/ <fir, fur>, /furθ, *furθ, furst, furm (noun), /furmament, vertiw, verj, verjin, θurti, θurtiin, θurd, θurst(i), spurt, stur, skirt, skermise, skwurt, scurt, smerk, smurc, sirkit, sur, surneim, surloin, swurl, hwurl, hur/ <her>, /mir/<myrrh>, /merθ, murtl, murk/ (but in Scotland /mirk/ when a dog's name), /irk/.

4.2 /aw/
4.21 In most accents /aw/ before /f, v, m/ and, in some accents, before /n/ followed by a consonant has different allophones from that in other positions. But there is no difficulty in grouping as one phoneme cluster the syllabics of /saw, hawnt, hawf, sawv, kawm/ <saw, haunt, half, salve, calm>. Because the allophones of /aw/ are most commonly described phonetically as "pure vowels" it may be wondered why they are here analysed as the phonetic realization of a heterogeneous cluster. A phonologic cluster /aw/ is the underlying representation from which the various allophones can be most economically derived. Speakers of General American derive [kaf, haf] from /kawf, hawf/ with complete regularity. Moreover, in the southeastern United States the value of <aw> in <saw> is phonetically a diphthong ending in [w].

Note: Some New Englanders rime /grawnt: hawnt/ <grant: haunt>, with the syllabic of /kawm/ <calm> in both words. Londoners, however, contrast the syllabics of these words, the result of either a spelling pronunciation or accent mixing. In most other accents the syllabics of these two words contrast for a different reason, discussed in §4.23.

4.22 The contrast of /a/ and /aw/ before /f, v, m/, as in /laf: hawf, hav: hawv, kam: kawm/ <laugh: half, have: halve, cam: calm>, is familiar to most northern English and some New Englanders. This contrast is preserved before /v, m/ by Australians, New Zealanders, and most English and New Englanders.

salve "ointment"

4.23 <Dance, command, grant, sample>, and many similar words have each two pronunciations arising from variation in Old French. From Parisian Old French come /dans, ka'mand, grant, sampl/, the forms traditionally favoured by poets (who often rime <command> with <hand> and <land>) and today by far the most wide-spread pronunciation. From Norman Old French come /dawns, ka'mawnd, grawnt, sawmpl/, which are largely limited to southern England, New England, and New Zealand.

4.3 /α/
There is no "Italian" <a> phoneme common to most varieties of English. This sound is very common in London English, but everywhere it is either lengthening of /a/ before open juncture, or before /f, θ, s, r/ - except /sc/ - regularly if a closed syllabic and exceptionally if an open syllabic, or the special development of /aw/ before /f, v, m/, or before /n/ followed by a consonant, or a recent borrowing or new formation. In a description of London English it is better to list exceptional forms than to postulate a phonematic contrast of such limited distribution. When occurring in exceptional phonetic contexts in recent borrowings and new formations [α] is identified with /ar/ or /a+/ - despite no morpheme boundary - by some English, with /a/ by some Scots, and with /o/ by most Americans. These words that may have [α] always have an alternative pronunciation with /a/, /ei/, or /aw/. Thus <khaki, tomato, Chicago> may be /kaki, ta'meito, *sci'kawgo/, as indeed they commonly are in American speech. <Prague, drama, vase, Marham, armada> are traditionally /*preig, drama, veiz, *maram, ar'meida/.

4.4 Syllabic miscellany
4.41 In the 17th century Standard English /wa/ became /wo/ except before /k, g, nk, ng, w, i/ and - sporadically - labial consonants.

Note: Non-presyllabic <al>, as in <wall, walk>, became /awl/ in the 14th or 15th century and was not affected by the above development. Then /wawlk/ became /wawk/ in the 16th century. (Dobson, 1968)





4.42 The contrast of /o/ and /ow/ before non-presyllabic /r/, as in /hors: howrs/ <horse: hoarse>, is preserved in Scotland and Ireland and parts of England and America.

4.43 Both /o/ and /u/ are heard, /o/ being older, in /mongr, mongril, a'mong, won/ <one>, /wons, non/ <none>, /nothing/.

4.44 The traditional orthography often represents /u/ by <o>, particularly next to <m, n, v, w>. In Middle English this practice was partly French scribal tradition and partly to improve the legibility of medieval script. The ambiguous spelling is responsible for variation between /u/ or /w/ and unhistoric /o/ in many words.

Note: In words Old French inherited from Vulgar Latin <o> before /m, n/ always represents /u/. (Pope, 1952) Thus the Scotch and American pronunciation of <constable> with /o/ cannot be historic and must be a spelling pronunciation or a case of prefix substitution.


4.45 Non-initial /ww/ is often realized as [w] by southern English and a few Americans before /m/, by Americans before /f/, and by both before /k/. But <ww> is recommended for the spelling of /rwwm, hwwf, bwwk/ <room, hoof, book>, and all other words in which pronunciation varies between /ww/ and /w/.

Note: Most Scots and northern Irish do not contrast non-initial /ww/ and /w/, pronouncing /fwwl/ <fool> and /fwl/ <full> alike.

4.46 Pronunciation varies between /wwr/ and /owr/ (or /iwr/ and /iowr/) in many words spelt <our, oor, (ur, eur)>, and in <More, sure, whore, (your)>. The pronunciation /wwr (iwr)/ is still common in all these words except /dowr, kowrt, flowr, fowr(θ), fowrtiin(θ), sowrs/.

Note: <Tourney, tournament, courteous, courtesy, courtesan, journey> do not belong here and are traditionally /turni, turnamant, kurtios, kurtsi, kurti'zan, jurni/.

poor, pour

4.47 Except in the combinative development of /tiw, diw, siw, ziw/ (§3.2), /iw/ is in all positions preserved as a diphthong by the Welsh, some New Englanders, and a few English. They contrast, for example, /iw/ and /ww/l in /ciw: scww, θriw: θrww/ <chew: shoe, threw: through>.

4.5 More miscellany.
4.51 Many Americans identify /i/ with /ii/, /e/ with /ei/, and /o/ with /aw/ when /g, nk, ng/ or presyllabic /r/ follows. Thus they rime /big: liig, eg: veig, moral: awral/.

4.52 Some Americans also raise /a/ before presyllabic /r/ to coalesce with /e/ and /ei/, pronouncing /meri: meiri: mari/ <merry: Mary: marry> alike.

4.53 The identification of /o/ with /aw/ before /f, θ, s/ regularly if a closed syllabic and sporadically if an open syllabic is now largely limited to Cockney and most Irish and American speech. Whatever the pronunciation, the vowel in /soft, moθ), lost/ is served well by the traditional spelling.

4.54 Many Scots, Northern Irish, and certain Americans (western Pennsylvania and increasingly elsewhere) never contrast /o: aw/, pronouncing /kot/ <cot> and /kawt/ <caught> alike.

4.55 Many northern English do not contrast /w: u/, pronouncing /kwd/ <could> and /kud/ <cud> alike.

4.6 Syllables in endings.
4.61 The suffix <ari, ary> is /eiri in /lai'breirian, sekre'teirial, sente'neirian/ <librarian, secretarial, centenarian>. Generally when final, as in <library, secretary, centenary>, the pronunciation varies with accent but in a morphophonematic analysis the underlying representation will be /eiri/- /laibreiri, sekreteiri, senteneiri/.

4.62 Secondary stress, preserved in some accents, shows the underlying representation of the endings <ery, bury> to be /eri, beri/, as in /semeteri, *scrowzberi/ <cemetery, Shrewsbury>.

4.63 Shakespeare rimed <oratory: story> (Rape of Lucrece), and the endings <ory, mony, ative> are still /owri, mowni, eitiv/ in American speech when the preceding syllabic is weakly stressed.



5.1 /m, n, l, r/
Some accents contrast syllabic and non-syllabic /n, l/, but not /m, r/. Note that the traditional orthography is no guide to the distribution of syllabic /n, l/.


5.2 [ə] from [i].

For many Irish, Americans, and Australians the central syllabic is often the phonetic realization of weakly-stressed /i/ followed with close juncture by a non-syllabic.

Examples: /teribl, sertin, sudin, markit, kurij, kurtin, kwscin, minit, letis, biznis, busiz, bwsciz, linin/

5.3 The traditional pronunciation of posttonic <ile> in Latin and Romance words is still preferred by most Americans. This ending under secondary or weak stress is always /il/ except in /krokodail, jentail, siinail, iidail/ <crocodile, gentile, senile, aedile>, assuming that /ail/ has primary stress in /rekun'sail, eg'zail/ <reconcile, exile>.


6.1 Stress levels.
It is necessary to distinguish 2 levels of strong stress primary and secondary - as in the end syllables of /e'vent, sam-pan/. It is proposed that /'/ be the sign of primary stress, and /-/ the sign of secondary. The sign precedes the stress. In practice it is not necessary to show primary stress if it occurs on the first syllable and not again in the word. Secondary stress need not be shown when on the first syllable. The sign /+/ signifies internal open juncture and is generally only necessary when weak stress follows.

Examples: /silw'et, film-land, ko'ordineit, ko-ordi'neisian, 'hawf'hartid, sing+r/

6.2 The traditional orthography is quite arbitrary in the selection of words written as compounds. It is proposed that all combinations with a reduced time interval between elements be written as compounds. Thus <long island> will be /long ailand/, but <Long Island> will be /*long'ailand/.


7.1 Uppercase mediaeval.
7.11 The distinction of uppercase and lowercase arose in the Middle Ages when scribes reserved the capitals for the beginning of important words. This practice is continued more because of tradition than utility. It is a needless extra burden for children and for typists and should be dropped.

7.12 In experiments testing reading speed, M A Tinker (1965) found that text set in all-capitals is read about 14 per cent slower than the same text in Roman lowercase. Italics were read about 5 per cent slower than Roman lowercase, while no difference was found between the reading speed of boldface and ordinary lowercase. The conclusion for orthography reform is that lowercase - ordinary and boldface - should alone be used. As a substitute for italics, the typographer has a choice of boldface, letterspacing, ordinary lowercase in a bigger size, angle brackets, and underlining.

7.2 <th>.
The traditional orthography uses <th> for 8 different values in <this, thing, thyme, hothouse, eighth, clothes>, and the traditional pronunciation of <Rotherhithe>, which are better spelt: /ðis (ddis), θing (thing), taim, hot-huws, eitθ, (eitth), klowz, *redrif/.

Note 1: The letters <θ, ð>, from Greek and Old English, are used with these values in the International Phonetic Alphabet. The digraph <dd> is familiar in Welsh names such as <Rhondda>.
Note 2: Formerly <Rotherhithe> was often spelt <Redriff(e)>. These spellings are now standard in the names of 2 roads in London.)

7.3 <wh, quh>.
The Old English spelling <hw> accurately describes the pronunciation of many today. <Wh> arose in Middle English by graphic analogy with <th, sh, ch>. <Wh> and Scottish <quh> were never intelligent spellings and should be dropped.



7.4 Punctuation.
English punctuation is a much later development than the traditional spelling and is in most respects entirely satisfactory. However, the following new conventions are proposed.

a) The sign <*> will be optional before proper names: /*lundan, *jon/.
b) The sign <=> will be used at the end of a line to *continue a word on the next.
c) The punctuation mark that ends a sentence will also begin it.
d) The sign <?> should begin and end only sentences ending with a rising tone. For example, the question /?wil iw hav tii or kofi?/ (ending with a rising tone) does not assume the listener will have either, and is different in meaning from the question /.wil iw have tii or kofi./ (ending with a falling tone), which assumes the listener will have one or the other.
e) Contractions need no special symbol, though they often show internal open juncture: /ai+l/ <I'll>, /ðei+r/ <they're>, /kwdnt/ <couldn't>.
f) The sign <+> can also indicate letters omitted in standard abbreviations: /m+/ <Mr.>, /k+/ <Co.>.
g) Latin abbreviations and initials might be replaced by English words and phrases:
/neimli/ <viz.>, /ðat iz/ <i.e.>, /for eg'zampl/ <e.g.>

7.5 Word signs.
It is not desirable that the orthographic representation of <the> and <to> should vary with pronunciation, and - for brevity as well - the word signs /ð, t/ are proposed.




Chomsky, Noam, and Morris Halle, 1968, "The Sound Pattern of English", Harper & Row, New York.

Dobson, Eric John, 1968, "English Pronunciation 1500- 1700", 2nd ed., Oxford University Press.

Harris, Zellig S., 1951, "Structural Linguistics", University of Chicago Press.

Pope, Mildred Katharine, 1952, "From Latin to Modern French with Especial Consideration of Anglo-Norman", 2nd ed., Manchester University Press.

Tinker, Miles Albert 1965, "Bases for Effective Reading", University of Minnesota Press, Minneapolis, and Oxford University Press.

Walker, John, 1791, "A Critical Pronouncing Dictionary", London.


1 .3 waiz men ov *gowtm
went t sii in a bowl:
and if ð bowl had bin strongr,
mai song had bin longr.

2 .orinjiz and lemanz,
sei ð belz ov *sint'klemants.

.iw ow mii 5 farðingz,
sei ð belz ov *sint'martinz.

.hwen wil iw pei mii,
sei ð belz ov *owld'beili.

.hwen ai grow ric,
sci ð belz ov *scowrdic.

.hwen wil ðat bii,
sci ð belz ov *stepni.

.aim scwwr ai downt now,
sez ð greit bel at *bow.

.hiir kumz a kandl t lait iw t bed,
hiir kumz a copr t cop of iwr hed.
3 ."ð taim haz kum", ð wawlras sed,
"t tawk ov meni θingz:
ov scwwz - and scips - and siiling-waks -
ov kabijiz - and kingz -
and hwai ð sii iz boiling hot -
and hweðr pigz hav wingz".
- *liwis karal

4 ."a fo'netik speling wwd dww muc t stedi ð inglisc langwij in ðis tem'pestiwos steij in its historic spowkn az it nuw iz bai hundridz ov milianz awl owvr ð wurld."
-*arθr loyd jeimz

Back to the top.