The WMM Spelling System

(version 2.3, April 14, 2005)

WMM (Weirdly Marked Minglish) is an alternative spelling system for English.  It used to be called GASS, standing for Grave Accent = Short Stressed.  The system was designed around the idea that there are benefits to using the unaccented vowel letters (a, e, i, o and u) to stand for both the schwa and the corresponding unstressed short vowel, ambiguously.  (Unstressed vowels are often difficult to classify precisely, and often vary from speaker to speaker as well.  Because almost all words are recognizable even if the unstressed vowels are distorted, it may well make spelling easier if the orthography represents them imprecisely.)  The remainder of the system, other than this ambiguity, is quite precise and mostly phonemic.  This, the second version of WMM, has been further enhanced to make it applicable to both American English and RP (British Received Pronunciation) with minimal respelling.

WMM has a very unfamiliar appearance, caused mostly by these factors:

These factors make it very unlikely that WMM will achieve any significant following.

WMM was primarily an experiment in the representation of English vowels.  I believe it to be an unsuccessful experiment.  The failure is the result of the importance of stress to the WMM system.  For both long and short vowels, the spelling depends on the stress, and I often have difficulty determining stress accurately.  If I am representative of the general population in this regard, most people would find WMM difficult to use.

There is a WMM dictionary here.  This dictionary was developed using FEWL and FLEWSY, as described here.  The process of developing the dictionary was key to the evolution of WMM, which ending up looking nothing like my original idea.  My guess is that, without the ability to quickly generate a new dictionary, it would have taken months rather than weeks to have developed WMM to my satisfaction, and I might well have given up long before then.

Even though I consider WMM to be a failure, it is an interesting failure - I think it shows the perils of trying to completely differentiate stressed and unstressed vowels.  And who knows, it's possible that you might like it more than I do.

Here is that old chestnut "The Star" spelled in WMM.

It wuz on dhe fërst dá ov dhe neú yeer dhat dhe anàunsment wuz maid, ôlmoast siemultainýusly frum thré obzërvatôryz, dhat dhe moashon ov dhe planet Nèpteün, dhe àutermoast ov ôl dhe planetz dhat wheel abàut dhe Sun, xad bikùm very iràtik. A reetârdaishon in its velòsity xad bin suspèkted in Dêsèmber. Dhen, a faint, rêmoat spek ov liet wuz diskùverd in dhe reejon ov dhe pertërbd planet. At fërst dhis did not kôz eny very grait êksietment. Síentìfik peepel, xàuèver, fàund dhe intèlijens rêmârkabel ênùf, eeven bêfôr it bêkaim noan dhat dhe neú body wuz rapidly gróiñ lârjer and brieter, and dhat its moashon wuz kwiet diferent frum dhe ôrderly progres ov dhe planetz.

You can use the Wyrdplay converter to convert traditionally spelled English text into WMM (and four other reformed spelling systems).

The Rules of WMM

Compared to my other systems, WMM is pretty simple.

Short Vowels

Stressed short vowels are written with a grave accent:  pàt, pèt, pìt, pòt, pùp.  Unstressed short vowels are written without any markings: fastìdýus, presteejh, tiepist, monsuen, subjèktiv.  The schwa is also written without any markings, using the same letter as the traditional spelling: soafa, petìshon, vërmin, reejon, kampus.  In cases where the schwa is traditionally represented by a digraph, one of the letters is used:  màuntan, fôren, deevýus, shoafur, tôrtis.

In many words containing an unstressed i sound, the spelling ê is used in place of i, as in êkstreem and përfêkt.  See below for discussion of this convention.

In a one-syllable word with a short vowel, the diacritic is omitted: kat.  In a two-syllable word where the first vowel is short and stressed and the second vowel is a schwa, the diacritic is omitted: lemon, kombinaishon.

Long vowels

Long vowels are written in one of two ways, with an acute accent (á, é, í, ó, ú) or as a digraph (ai, ee, ie, oa, ue).  The accented character is used when the vowel is unstressed, at the end of a word (or a component of a compound word), and before another vowel.  The digraph is used in all other cases.

The long u, spelled ú or ue, is a pure long u, the vowel of crew, not of few.  The latter word is written as feú. (eú is spelled eü when stressed and followed by a consonant, as in feüd.)  For further information, see British complications below.

Both é and ý are used for the long e sound.  As a general rule ý is used when the sound is unstressed, and é when it is stressed.  However, é is used even when unstressed if the following letter is a consonant, except when it ends a component word with the same spelling.  The accent may be left off ý when no ambiguity is possible - at the end of a word, or before a consonant. 
If the first syllable of a word is a stressed short vowel, and the second syllable ends with y/ý, the accent is left off the first vowel.

Examples: main, meen, mien, moan, muen, meül, máonaiz, théater, bíòlojy, dìslócait, vërchú, raidýó, légàlity, enerjy, absenté, fantasyz, xapynes, laidybug, glôrýus, antysoashal, résiekel.

Other vowels

The vowels â, ô and û are used as in drâma, ôful and pûsh.  The sequence ër is used for the stressed er sound of jërm.  And the sequences àu and ôi are used for the diphthongs of klàun and bôil.  Note that the accent cannot be omitted from these sequences, even in a one-syllable word.

The nasal vowels ã and õ are used in borrowed French words like kontretã (contretemps) and kõsyèrjh (concierge).

British complications

The sounds of â and ò are the same in American English.  A vowel is written as ò in WMM if it is pronounced with a short o sound (Sampa /Q/) in British English (RP), or as â otherwise.  Thus, in WMM one writes "balm" as bâm, but "bomb" as bom.

RP has three diphthongs not present in American English, occurring before the letter r: Sampa /e@(r)/, /I@(r)/ and /U@r/.  The corresponding sounds in American pronunciation are /Er/, /Ir/ and /Ur/.  In WMM, these American sounds are normally written èr, ìr and ûr respectively; this is changed to air, eer and uer in words where the diphthong is used in the British pronunciation, as in WMM fairy, steer and tuerist.  It turns out that the American /Ur/ sound is almost always spelled uer.  The spelling of the other two sounds can generally predicted from the traditional spelling: the WMM spelling èr is used mostly when TS uses er or err (very, terrify), and the spelling ìr is used mostly when TS has ir, irr or yr (miracle, irrational, pyramid).

RP often has the sequence /Vr/, which does not occur in American English, in words spelled with -urr- or -our- like "hurry" and "encourage".  These words are pronounced by Americans with /3`/.  WMM uses ùr rather than êr to spell these words: hury, inkùrij.

English has two long u sounds, /ju:/, as in "cute", and /u:/, as in "rude".  For a large number of words, such as "tune", British English uses the /ju:/ pronunciation, and American English the /u:/ pronunciation.  WMM represents such words using the British pronunciation, in part because a minority of Americans also use this pronunciation, using the spellings eü (when the sound is stressed and precedes a consonant) or eú (in other cases).  Thus, WMM spells "few", "new" and "crew" as feú, neú, and krú, and "immune", "tune" and "June" as imeün, teün and Juen.  Often, the long u sound is shortened into a /U/ or a schwa, which may be /jU/ or /j@/ in either or both varieties of English.  WMM uses a preceding e for these vowels too, as in regeular
and kòmpeûtaishon.  Sometimes, the preceding /j/ may be dropped in American English, in which case WMM still uses the e, as in inseular and teûbërkeuloasis.

RP speakers generally pronounce the ending -y of <happy>, and the i of words like <radio>, with a short i rather than a clipped long e sound.  The letter ý is used to indicate either pronunciation.  This allows WMM to represent the distinction between <booty> and <bootee>: buety versus bueté.

Two other ways in which RP differs systematically from American English are illustrated by the words "bath" and "cross".  "bath" is /b{T/ in American English, but /bA:T/ in RP; "cross" is /krO:s/ in American English, but /krQs/ in RP.  WMM does nothing special about these differences.  I consider it sufficient that the WMM spellings differ only in the diacritics: bath and bâth, krôs and kros.

Note that when WMM is used to write RP, the r's remain in the spelling, even when they are not pronounced.  The traditional spelling rules work very well in this regard, and there is little gain to be had by opening a chasm between British and American spelling by omitting them.


The normal consonant symbols of English are used, except for the following.  Most sounds which are usually written with a digraph in traditional spelling are also written with a digraph in WMM: chip, ship, vijhon, thum, tugèdher, whisper.  The English ng sound is written with ñ: yuñ, siñer, fiñger, thiñker.  The sound traditionally written as h is written x in WMM: xeven, mìsxàp.

The German or Scottish ch sound is written in WLM as kh, as in lokh (loch).

Other rules

Regular plurals are always written with an ending z, whether it is pronounced as z or s.  Similarly, a regular past tense is always written with an ending d, whether the pronunciation is d or t.  Examples: katz, dôgz, dansez, rekd, dragd, aded.

In a compound word, accents are removed only if allowed by the rules for the entire word, not for the component words.  Examples:  sùnbërn, kràbàpel.

Word relationships in WMM

Due to several of its features, notably the way it handles the schwa, WMM is considerably better than most reformed spelling systems at revealing the linkages between related words, as in the spellings feüry/feürýus, medisin/medìsinal, foatogràf/fotògrafy.  I consider this a very desirable property for a spelling system to have.

There is a complication, however, in the form of the frequent use, in current spelling, of the letter e to represent a distinct, unstressed i sound, as in deprive, economy, perfect, prepare, and so on.  The natural thing to do is to spell these words with an i: dipriev, ikònomy, përfikt, pripair, etc.  But then the relationship to related words is obscured: dipriev/deprivaishon, ikònomy/ekonòmikal, përfikt/perfèkshon, pripair/preparaishon, etc.  So WMM now uses ê to spell a distinct, unstressed i when e is used in the current spelling: dêpriev, êkònomy, përfêkt, prêpair.  This rule is not applicable if the i sound is stressed, indistinct or schwa-like: the WMM spellings for English, forest, rivet and happiness remain Iñglish, fôrest, rivet and xapynes, not Êñglish, fôrêst, rivêt and xapynês.

There are also some related words where the rule of using the current spelling to represent the schwa obscures the relationship, such as kurij/kuraijus, konfërm/konfirmaishon, and eekwal/êkwòlity.  WMM changes the spelling of the schwa in such cases to agree with the stressed vowel in related words, so that the correct spellings of courage, confirmation and equal are in fact kuraj, konfermaishon and eekwol.


One unavoidable characteristic of highly phonemic spelling systems like WMM is that they merge homophones, that is, they force words which sound alike to be spelled alike.  Therefore, <so>, <sew> and <sow> are all in WMM, and <sent>, <scent> and <cent> are all sent.  The stock answer to this problem is that we all do perfectly well understanding which word is meant when English is spoken, and it should be no harder when it is written.  This is true enough.  Little confusion between <sent> and <cent> is likely, as they are different parts of speech, and even though <scent> and <cent> are both nouns, which one is meant can usually be determined by context.

I do feel, however, that it is a good idea for function words that could be confused with other function words to be spelled differently, due to the frequency with which these words are used.  This applies specifically to three sets of words: to/too/two, for/four and there/their/they're.  These are rendered in WMM as tu/tue/tú, for/fôr and dhair/dhár/dhá'r.  I justify these specific spellings as follows:

<to> is spelled as tu because it is almost always unstressed.  <two> is spelled as because this is more regular than tue and therefore functions better in compound words such as túsom and túfoald.  <too>, which to my knowledge is not part of any compounds, then gets the irregular tue.  Similarly, <for> is spelled for because the word is usually unstressed, leaving the spelling fôr for its homonym <four> (and <fore>).  Finally, the spelling dhá'r  for <they're> is quite obvious.  The spelling dhár for <their> was chosen to emphasize the relationship to the pronoun dhá.  This then leaves the regular spelling dhair for <there>.

I've considered having special forms for the function-word homophones <eye> and <owe>, since it can be hard to recognize single-letter words in compounds like ífûl and ísiet.  However, both of these words can be used as verbs, and I find the spellings íiñ and óiñ quite preferable to the alternatives ieiñ and oaiñ, for which reason I've decided not not make exceptions for these words.

Another sample

Here is another sample of WMM, the lyrics to the Dire Straits song "Industrial Disease" (words and music by Mark Knopfler).

Wôrniñ lietz âr flashiñ dàun at Kwolity Kontroal
Sùmbòdy thrú a spaner and dhá thrú xim in dhe xoal
Dher'z ruemorz in dhe loadiñ bá and añger in dhe tàun
Sùmbòdy blú a whisel and dhe wôlz kaim dàun
Dhair'z a meetiñ in dhe bôrdruem, dhá'r tríiñ tu trais dhe smel
Dhair'z leekiñ in dhe wòshruem, dhair'z a sneek in Përsonèl
Sùmwhair in dhe kôridor, sùmwùn wuz xërd tu sneez
"Gûdnes mé, kûd dhis bé Indùstrýal Dizeez?"

Dhe kairtaiker wuz kruesifíd for sleepiñ at xiz poast
Rêfeüziñ tu bé pasifíd, its xim dhá blaim dhe moast
Dhe wòchdôg got raibéz, dhe fôrman'z got dhe fléz
Èvrýwùn'z konsërnd abàut Indùstrýal Dizeez
Dhair'z panik on dhe swìchbôrd, tuñz âr tíd in notz
Sum kum àut in simpathy, sum kum àut in spotz
Sum blaim dhe manijment, sum dhe êmplôiéz
And èvrýbòdy nóz it'z dhe Indùstrýal Dizeez

Dhe wërk fôrs iz disgùsted, dàunz tuelz, wôkz
Inosens iz injurd, êkspeerýens just tôkz
Èvrýwùn seekz damijez, èvrýwùn agréz
"Dheez âr klasik simptomz ov a monetairy skweez"
On ITV and BBC dhá tôk abàut dhe kërs
Filòsofy iz eüsles, théòlojy iz wërs
Xistory bôilz oaver, dhair'z an ekonòmiks freez
Soasýòlojistz invènt wërdz dhat meen Indùstrýal Dizeez

Doktor Pârkinson dêklaird "Í'm not surpriezd tu sé eú xeer
Eú'v got smoaker'z kôf frum smoakiñ, brúer'z druep frum driñkiñ beer
Í doan't nó xàu yú kaim tu get dhe Bety Daivis néz
But wërst ov ôl yuñ man eú'v got Indùstrýal Dizeez"
Xé roat mé a prêskrìpshon, xé sed "Eú âr dêprèsd
Iem glad eú kaim tu sé mé tu get dhis ôf eür chest
Kum bak and sé mé laiter, nekst paishent pleez
Send in anùdher viktim ov Indùstrýal Dizeez"

Í gó dàun tu Speeker'z Kôrner, Í'm thunderstrùk
Dhá got fré speech, tueristz, polees in trukz
Tú men sá dhá'r Jeezus, wun ov dhem must bé rôñ
Dhá got a proatèst siñer, xé'z siñiñ a proatèst sôñ
Xé sez "Dhá wont tu xav a wôr só dhá kan keep us on àur néz
Dhá wont tu xav a wôr só dhá kan keep dhár faktoryz
Dhá wont tu xav a wôr tu stop us bíiñ Japaneez
Dhá wont tu xav a wôr tu stop Indùstrýal Dizeez
Dhá'r pôintiñ àut dhe enemy tu keep eú def and bliend
Dhá wont tu sap eür enerjy, inkârserait eür miend
Giv eú Ruel Britànya, gasy bir, paij thré
Tú weekz in Èspânya and Sùndá strip teez"
Meenwhiel, dhe fërst Jeezus sez, "Í'l keür it suen
Abòlish Mùndá môrniñ and Friedá afternuen"
Dhe udher wun'z àut on xuñger striek, xé'z díiñ bí dêgréz
Xàu kum Jeezus getz Indùstrýal Dizeez?

The evolution of WMM

WMM started life very simple indeed.  It began with the vowel representations, except that only the monographic forms for the long vowels were used.  My original proposal didn't even bother to suggest what to do with the consonants.

When I decided to make a full-fledged spelling system of it, I started by adding what I consider the usual consonant handling.  I used zh for the French j, ng for the soft ng and ngg for the hard one, and h for h.  I switched zh to jh simply because I liked it better.  I switched to ñ for ng because, overall, I preferred thiñ/thiñk/fiñger to thing/think/fingger, and also because it allowed me to distinguish uñkel and unkleen.  I switched to using x for the h sound because, as WMM became a more and more precise notation, I liked less and less the ambiguities of adhìr, mìshàp and fûthóld.  It had become clear that anyone who cared about familiarity was not going to appreciate WMM, so why compromise?

The biggest changes to WMM took place when I started using FEWL (see here) to build a WMM dictionary.  The biggest problem I saw was that there were simply too many diacritics.  Since one of the points of WMM was distinguishing the stressed and unstressed short vowels, I decided it might be interesting to do the same for the long vowels, and that the stressed ones would stand out better if they were digraphs.  This was an improvement, but I didn't like the way it looked when there was a digraph at the end of a word, or before another vowel, and so I decided to make these special cases, especially since these vowels are often unstressed anyway.  I found this helped the appearance of WMM a lot.  The only thing I didn't like was á at the end of a word, as in plá and tuepá.  I finally decided to leave this alone, even though I didn't (and still don't) care for it, since it had the advantage of being consistent with the way the other vowels were handled.

The original WMM dropped the acute diacritic for a long vowel at the end of a word, as in dépo and snàfu.  I changed my mind about this and decided to keep the diacritic here, so that these words became deepó and snàfú.  The main reason was that the diacritic was necessary for the vowels a and e (an ending unaccented a is probably a schwa, and a final bare e looks confusingly like the traditional magic e).  It seemed more consistent to leave in the accents for all the vowels (except for y, where leaving it out was quite natural and unambiguous), and I've decided it looks better as well.  With this change, you could be sure that a bare vowel was not long, unless it was part of a digraph.

The final change to the first version of WMM was the decision to use z for all plurals and d for all past tenses.  I really hadn't given this issue any thought at the start, and this treatment originally seemed to be contrary to the phonemic accuracy of WMM in other areas.  But I've come to the conclusion that using a consistent method for representing these inflections is more important than getting the pronunciation right, and also that z, not s, should be used for the plural unless one feels a need to stay close to traditional spelling, which I did not.

The second version of WMM mainly differed in adding the RP diphthongs.  While this significantly increases the complexity of both the description and the use of the system (at least for Americans), I felt the changes were worth making to illustrate how a spelling system can usefully work for both varieties of English.  An unexpected side-effect was that many words, such as stairz and feerful, actually became more familiar-looking due to the common use of the sequences air and eer in traditional spelling.  The second version of WMM also changed the digraph oh to oa (a clear improvement - compare chohk and choak) and modified the rules for use of é and ý.  It also added the flexibility of changing the representation of unstressed short vowels to elucidate word relationships. - however, the frequency with which i had to be replaced by e led quickly to the addition of ë (now ê) in version 2.1.  This in turn was followed rapidly by version 2.2, which differed only in its special spellings for <too>, <for> and <their>.  Version 2.3, which appeared after some time for reflection, split the eú digraph into eú/eü, similar to the other long vowel pairs such as á/ai and ú/ue.  Additionally, this version interchanged the letters ê and ë, so that ë and eü would have the same stress.

In retrospect, it is somewhat ironic that the main reason for the strangeness of WMM is the decision to use x for h, which I consider rather peripheral to the main points of the system.  I have toyed with the thought of changing this.  I could, of course, just use h for the h sound as well as in digraphs, but I'm convinced that this easy choice is wrong for this system.  One interesting idea is to use a capital H for the h sound, as in Hat, mìsHàp and bàthHàus.  It's amazing to me how natural this looks.  However, the use of capital letters with a different meaning from their lower-case is considered quite radical in spelling-reform circles, and would probably go over even less well than the x does.  A similar idea is to use the thorn in place of x, as in þapy, misþap and Þalowén.  This has the advantage of resembling an h more than an x does, but the disadvantage of introducing an entirely new symbol to the alphabet (and further, using it in an entirely non-traditional way).

WMM Variants and Relatives

Two variants of WMM are sanctioned.  WMM-h is WMM using the h in place of the x, a change which makes it far less scary-looking.  WMM-4g is WMM-h with the long vowel digraphs eliminated and replaced by the diacritic forms.  This form of WMM greatly resembles Bob Boden's SRS4g, differing from it mainly in its handling of the schwa.  Comparing a text in WMM-4g and SRS4g is a good way to get a feel for the overall effect and importance of their individual strategies for representing that difficult sound.  WMM-4g is actually superior to WMM in preserving the form of related words (fótogràf/fotògrafy rather than foatograf/fotògrafy), and it probably also seems better to those who want a spelling system, whatever else it may accomplish, to make words shorter.

Bob Boden has recently introduced his Bobdot system, which is a version of SRS4g that displays primary stress.  In response, I've introduced a system called Arbdot, constructed to be as similar as possible to Bobdot, while making some alternate design choices I find preferable.  Arbdot is very similar to WMM-h with additional diacritics instead of digraphs for long vowels, and with some smaller changes for Bobdot compatibility, such as using both c and k for the k sound.  I definitely prefer WMM, but the similarities of Bobdot and Arbdot help to focus comparisons of the two, or at least that is the theory.

Bobdot's emphasis on primary stress (rather than WMM's emphasis of simply stress or lack thereof) has led me to consider a WMM variant called WMM-ps, where the digraph form for a long vowel is used only when the vowel has primary stress.  In WMM-ps, we would write abreevýát and abrévýaishon, rather than abreevýait and abreevýaishon.  It is hardly logical to do this, since I keep the stressed/unstressed distinction for the short vowels, and the circumflexed vowels continue to be stress-independent.  So, in constrast to Bobdot, inspection of a WMM-ps word may or may not allow you to determine its primary stress.  Nevertheless, I find that of all the forms of WMM, WMM-ps (or WMM-ps-h) is the most readable.  I am continuing to explore this phenomenon.

To comment on this page, e-mail Alan at

Go to home page
Go to spelling system roster