Navigation menu



In the following I sketch a new orthography for the English language. Although English spelling is a worsening mess, I rather doubt that any significant spelling reform will take place in the foreseeable future: the existing written standard has far too much inertia at present. But there will have to be fundamental reform at some point, or the written and the spoken languages will lose all contact with each other. Here I suggest a way they could theoretically be brought back together again at the current stage of the language’s development.

Comments on matters of fact or technicalities are welcome.

General Principles

Piecemeal reforms to gradually create and then maintain a harmony between the written language and the spoken word would have had to have begun several centuries ago to be effective. It is now no longer possible to tidy up English spelling via the requirement that the rules be followed consistently, because there are no consistent rules (although you can pronounce 80% to 90% of English words correctly with only 50 to 60 rules, some of them relatively simple). New rules are needed. But these rules cannot be devised completely arbitrarily. They will need to comply with a number of general principles. A new orthography should be:

  1. Simple: there must be a good reason for every rule.
  2. Consistent: there should be no exceptions.
  3. Able to cope with both British and American Standard English: that is, the number of divergent spellings should be kept to a minimum and everyone should be able to tell from its spelling how a word is pronounced in the standard version of his or her dialect.
  4. Typographically unproblematic: unless we decide to create a whole new alphabet, the creation of new letters is to be avoided, and the use of accents to be kept to a minimum.
  5. Not too far removed from other European languages in its use of letters: again, unless we create a whole new alphabet, we must respect the history of the one we have. Using ‘q’ as a replacement for ‘th’ would probably be a bad idea, for example.

Rather surprisingly, perhaps, it is possible to respect all of these principles at once. Let us start with the hardest task: squeezing the English vowels into the Roman alphabet.


The greatest difficulty in creating an orthography for English using the Roman alphabet is that English has twenty vowels, while the Roman alphabet has only five letters with which to express them. The situation is not as hopeless as might appear from this simple statistic, however:

  1. Eight English vowels are diphthongs and therefore easily and naturally expressible as digraphs.
  2. Two of the remaining twelve can be regarded as stressed and unstressed versions of the same phoneme.
  3. The eleven monophthongs divide into six short and five long vowels, whereby the long ones can be seen as ‘stretched’ versions of short ones, which can therefore naturally be written as double letters.

So in the end we are only one letter short:

Short VowelsLong Vowels
‘a’ as in ‘pat’:a‘a’ as in ‘father’:aa
‘e’ as in ‘pet’e  
‘i’ as in ‘pit’i‘i’ as in ‘Rio’ii
‘o’ as in ‘pot’o‘o’ as in ‘more’oo
‘u’ as in ‘put’u‘u’ as in ‘lute’uu
‘u’ as in ‘but’ and
‘u’ as in ‘upon’
?‘u’ as in ‘burn’?

To fill in the gap it is not necessary to invent a new letter, or to press an old one into unnatural service: the letter ‘e’ is underused in the schema above, the long sound in ‘burn’ is always stressed and followed by an ‘r,’ a doubled ‘r’ sound does not exist in English, and the sound of ‘u’ in ‘upon’ is never stressed, while the sound of ‘e’ usually is. So for the vowel in ‘burn’ we can write ‘er,’ and when an ‘e’ sound is followed by an ‘r’ the ‘r’ can be doubled, to avoid confusion. For unstressed ‘u’ in ‘upon’ we can use ‘e,’ and when an ‘e’ sound is unstressed we can write it ‘ë’ to avoid confusion. Such confusion could in any case only arise for children and foreign learners, and would even then usually not be important.

This leaves us with the stressed version of this sound: the ‘u’ in ‘but.’ One option would be to use ‘e’ for this and always write an ‘e’ sound ‘ë.’ This would take English out on a limb with respect to other languages, and its results are aesthetically unpleasing. It would also make the diaresis absolutely vital, which would be undesirable: accents may be unavoidable, but their rôle should be kept as insignificant as possible. Bearing in mind that most occurrences of this sound are descended from /ʊ/, and that this sound is now much more common than that in ‘put,’ I suggest therefore that we use a plain ‘u’ for it, and ‘ü’ for the less common sound. The diaresis is once again required to avoid a minor confusion that would not normally arise.

When long ‘a’ and ‘o’ are followed by an ‘r,’ they do not need to be written double. Where short ‘a’ and ‘o’ are followed by an ‘r’ the ‘r’ is will be doubled.

Short vowels do not occur before other vowels or at the ends of words. Long vowels in these positions do not therefore need to be written double. Since short ‘i’ sometimes follows long ‘i’ this could give rise to ambiguities (‘seeing’ becomes ‘siing’). To avoid this problem, and bearing in mind that a double ‘i’ is aesthetically not very pleasing and in handwriting easy to confuse with a ‘ü’, I suggest that long ‘i’ normally be written ‘iy.’ So we are left with the following schemata:

1. Simple Vowels
SoundNormal SpellingSpelling with rBefore vowels/
at end of word
ə:r: er  
ə e 
ər  er 
ɔ:ɑ:oo o
ɔ:ɔr oror
2. Diphthongs


  1. The dots on thi unstrëst ‘e’ and on ‘u’ ar eunli rieli necesserry for children and forrin lerners. Seu wi can ifektivli du withaut accents.
  2. Neut that this spelling is oolseu natrel for the commen rendering of this vaul as e long monephthong /ɛ:/.

A slight complication arises from the fact that the vowels in some words have more than one pronunciation: ‘the’ is usually pronounced /ðə/ but becomes /ði:/ when it is stressed or before a vowel, for example, while the vowels in ‘a,’ ‘an,’ ‘and,’ ‘that,’ ‘are,’ ‘of,’ ‘to,’ ‘at,’ and various other structural/grammatical words all become /ə/ when they are not stressed, which is most of the time. To avoid complication, I suggest that all words except the definite and indefinite articles be written as pronounced in isolation, their stressed form. ‘A,’ ‘an,’ and ‘the’ will be spelt as follows:

Wiyk FormStrong Form


The sounds currently represented by ‘c,’ ‘x,’ and ‘q’ can all be dealt with by ‘k’ and ‘s.’ This leaves ‘c’ available to represent the first consonant in ‘children’ without the help of an ‘h,’ while ‘x’ can be used to replace ‘sh’ as in ‘short’: ‘xort cildren.’ The sound of ‘s’ in ‘measure’ can be spelled ‘zh’: ‘mezher.’ To distinguish between ‘th’ in ‘thing’ and ‘th’ in ‘that,’ the latter can be written ‘dh’: ‘dhat thing.’ All other consonants can represent the sounds they normally cover at present, the only slight complication entering with the treatment of ‘ng’:

The ‘g’ in ‘ng’ is sometimes pronounced, as in ‘finger,’ and sometimes not, as in ‘singer.’ One way to deal with this would be to write the ‘g’ double where it is pronounced: ‘fingger,’ but this is aesthetically not very pleasing. Another possibility would be to use the otherwise redundant letter ‘q’ to replace the silent ‘g,’ using a ‘g’ only where it is pronounced ‘sinqer,’ ‘finger.’ But this would represent rather too large a departure from the customary use of ‘q.’ So I suggest that where the ‘g’ is pronounced, this be signalled in general by a following ‘h’: ‘fingher.’

Before ‘l,’ ‘r,’ ‘w,’ and ‘y’ the ‘g’ in ‘ng’ is nearly always pronounced. For these cases I therefore suggest that the ‘h’ only be inserted in those few cases where the ‘g’ is not pronounced.

This leaves us with the following schema:



  1. Bifor semi-vaulz and likwidz, dhe ‘g’ in dhe kombineixen ‘ng’ iz prenaunst, udherwaiz not. Tu rivers dhis ruul, en ‘h’ iz inzerted bitwiyn dhe ‘g’ and dhe foleuing letter. Dhus ‘finger’ → ‘fingher,’ ‘singer’ → ‘singer,’ ‘language’ → ‘langwij,’ ‘jugular’ → ‘jugyeler,’ ‘angry’ → ‘angri,’ ‘angle’ → ‘anghel,’ ‘angular’ → ‘angyeler.’ Dhe kombineixens /ŋr/, /ŋl/, /ŋj/ and /ŋw/ ar reer, ekering meinli if not ekskluusivli in kompaund werds and proper neimz: ‘ringworm’ → ‘ringhwerm,’ ‘Longworth’ → ‘Longhwerth,’ ‘Bingley’ → ‘Binghli’ etc.
    Bifor /k/, /ŋ/ iz riten ‘n’: ‘bank,’ ‘plank,’ ‘anker,’ etc.
  2. Konsenents ar eunli riten dubel if thei ar seu prenaunst: ‘krankkeis,’ ‘busstop,’ ‘unnatrel,’ etc. Dhi eksepxen iz ‘r,’ wic iz never prenaunst dubel and iz riten dubel after xort ‘a,’ ‘e’ end ‘o’ tu xeu that thei arn’t long: ‘marri,’ ‘berri,’ ‘forrist,’ etc.
    Dhe leter ‘r’ iz riten if prenaunst in Emerriken, Airix end Skotix Inglix. In Standerd Britix Inglix it iz not prenaunst after vaulz.
  3. Dhe leter ‘t’ iz riten weer /t/ iz prenaunst in Standerd Britix Inglix. Bitwiyn vaulz in Emerriken Inglix it iz prenaunst /d/.

Britix or Emerriken?

In jenrel, dhe difrensez bitwiyn Emerriken end Britix prenunsieixenz ov werds ar sistematik inuf for it tu bi irelivent wic iz teiken az e model. Dhe kompremaizez ai prepeuz on ‘r’ and interveukalik ‘t’ ridyuus dhe difrens stil ferdher. Sumtaimz, dhe diferens iz tu greit for e sistematik kompremaiz, and dheer wil hav tu bi tu formz ev e werd in dhe foleuing keisez:

  1. In sum werds weer Britix Inglix has ‘oo’ ‘yuu’ and ‘aa,’ Emerriken Inglix wil hav ‘o,’ ‘uu,’ end ‘a.’
  2. Weer e vaul prezent in wun verxen ov Inglix iz ridyuust or absent in dhi udher: dikxenri/dikxenëri, sekretri/sekretëri, misail/misel.
  3. Unsistematik difrenses: temaateu/temeideu, herbs/erbs, etc.

Ev kors, thingz küd bi ferdher simplifaid bai ignoring standerd Britix Inglix prenunsieixen, sins hardli eniwun yuzez it eni mor, iyven in Briten. Or raadher, espexeli in Briten.


Punkcuiexen kan bi left az it iz eksept for dhe haifen, huuz yuus iz kurentli iregyeler and keiotik. I prepeuz it bi yuuzd ekskluusivli for werd breiks at dhi end ov lainz and in dhe foleuing keisez in kompaund werds:

  1. Tu kiyp silebelz sepret weer thei wüd udherwaiz keueles: e.g. ‘taip-raiter,’ ‘tait-reup.’ (‘Taipraiter’ and ‘taitreup’ wüd bi prenaunst ‘tai-praiter’ and ‘tai-treup.’)
  2. Tu kiyp saundz sepret weer thei wüd udherwaiz keueles: e.g. ‘reud-haus.’
  3. Weer dhe kombineixen iz ad-hok or not yet esimileited intu dhe langwij.

Udher werds nau riten widh haifens ar aither tu bi riten az wun werd or sepretli.