How many significant sounds are there in English speech? 46?

Speech cannot be fully reduced to graphic representation but we can capture enough information to enable speakers to reproduce it.  Some codes are good enough to help non-native speakers.

This page was originally developed to answer a question regarding how to map New Spelling and other notations to IPA.  It then became the starting point for an article on the phoneme inventory.

To work out the grapheme-phoneme correspondences, one must start with an inventory of significant speech sounds or phonemes.  These were worked out over 100 years ago by Ellis, Isaac Pitman,  Henry Sweet, and Daniel Jones

See Truespel

  The Phoneme Inventory
 The Jones-IPA  21 vowel matrix for RP

Any orthographic system for English should have a unique grapheme [symbol] for most of the 12 pure vowels and the important combinations of vowel phonemes.  The sounds that Daniel Jones considered to be essential for a full description of English speech are listed below. 

According to Harry Lindgren, many proposed reform notations for English fail by not having a unique symbol for schwa.  Schwa [an unstressed mid lax vowel] is one of the most frequent sounds in English speech. Roughly 10% of your utterances are [uhs] or schwas.  Lindgren considered this oversight to be sufficient to eliminate the proposed notation from serious consideration. 

Lindgren's system had both u /^/ and schwa ['].  Many systems will merge these phonemes.  Truespel uses u for both, Spanglish uses a for both. Since Truespel always marks stress, it is easy to determine if the u is stressed or unstressed. Thus Truespel does have a way of referencing the schwa [or mid lax vowel] sound.  about = /'baut/ = ubbout

According to Wijk, there are 46 different speech sounds in British English (RP):  21 vowels, 25 consonants (A. Wijk, p. 13).  According to Longman's Dictionary of American English, General American also has 46 speech sounds [21v  25 c]. Although there appears to be agreeement, it is very difficult to specify the minimum number of phonemes. 

Sixty symbols are normally used to represent the vowels.  Unfortunately, some are used for more than one sound. Most symbols in TO are polyvalent or multi-valued.  The codes overlap.
 

60 V-Markers found in TO: a, e, i, y, o, u, ar, er, ir, yr, or, ur, aa, ae, ai, ay, au, aw, ea, ee, ei, ey, eu, ew, ie, ye, oa, oe, oi, oy, oo, ou, ow, ue, ui, uy, aer, air, ayr, ear, eer, eir, eyr, eur, ew(e)r, iar, ier, yer, oar, oor, our, ow(e)r, uer, igh, aigh, augh, eigh, ough.

In TO, 44 symbols are used to represent 25 consonants:  21 single letters + 23 digraphs and trigraphs.  The 23 combined symbols  are shown below:

 

23 Marked consonants: ch, dg, gh, gn, gu, ng, ph, qu, sc, sch, sh, si, ssi, sci, ti, ci, ce, tch, th, wh, xc, and zi.   (ci, si, and zi are used for /sh/ and /zh/).

To represent 46 phonemes, English traditionally uses 104 different unigraphs, digraphs, and trigraphs.  Some are used more than once.

The most serious problem with the tradtional English orthography is its lack of predictability.  The chances that one can spell an unfamiliar word is 50% OR LESS.  This estimate is based on the fact that no phonemic notation will match TO more than 50% of the time.  The best attempt is about 40%.

Code overlaps are much more serious than using more than one letter or combination to represent a sound. Code overlap refers to the tendency for TO to use the same letter or combination to represent more than one phoneme. Each letter can refer to about 14different sounds.  The letters in TO are multi-valued or polyvalent.  (i.e., chaotic and confusing as opposed to alphabetic) 

Educators intent on teaching spelling and reading generally discard wh as a distinct phoneme, and make a few simplifications. 
 

From Traditional spelling			To Truespel
25 consonants	44 symbols	50 spellings	23 c	23 spellings
25 vowels	60 symbols	50 spellings	17 v	17 spellings
50 phonemes	108 symbols	108 spellings	40 phonemes	40 spellings

Orton phonograms  bibliography

  Truespel and Spanglish  asciibets [askee-bets] are used to indicate sound values below

The basic code:  Each phoneme represented by only one letter or digraph
 From sound to spelling.

25 consonants in English speech - 50 spellings in written English

24 Consonant Phonemes [ng missing from both TO and Truespel]

with one spelling by position b d h l p t v ng qu sh th tth

with one dominant spelling pattern f g j k m n r s w x z ch

The z phoneme is generally spelled with an s.  z is dominant only in the initial position

26 spelling alternatives for 12 consonants plus 73 consonant clusters - 99 

Truespel and Phonemic Spanglish notations are used to clarify the following vowel sounds:

    18 Vowel Phonemes [two phonemes are merged her/er and o/o:]

 

truespel    a   i   aar   oi	e   aa   u   ee  ae  ie  oe   yue  aw  oo   ue   ou  er  or
spanglish . a.  i.  ar    oy	e   o.   'u    i    ei   ai   o'     yu    o    u.    u    au  'r   or
spanglish.ae  y  aar   oy	e   aa    v    i    ei   ai   ow    yu    o    w    u    ou   r   or

Spanglish requires diacritics or markers to become a phonemic notation.  The alternative of recruiting r y w v as vowels [3rd line] can be visually disruptive  McGuinnes (1997), as above, ignores the schwa sound and most R-combinations.

32 spellings for 14 vowel sounds must be taught in TO (but not in a phonemic alphabet)
e.g., /ou/ -  tone, goat,  told,  low, though,  obey,   owe, oh

---

    Span -   to'n    go't    to'ld   lo'    tho'       o'bey     o'         [o' alternate ow]
 Truespel   toen   goet   toeld  loe  thoe      oebae   oe 

21 code overlaps must be taught for TO, e.g.,  ou is associated with over 7 sounds
<ou> -  out, soup,  touch,  soul,   though,  cough,  thought
OGD -   out,  suep,  tuch,   soel,   thoe,      kawf,    thawt
Span -   aut   sup    tu'ch    so'l      tho'        cof        thot 

According to McGuinnes (1997), most of the confusion in TO (the traditional orthography)  is caused by 26 alt. vowel spellings, 32 alt. consonant spellings, and 21 code overlaps.

Hanna also thinks that the sheer number of orthographic options makes the traditional writing system confusing and overly complex.  The obvious solution is to reduce the number of spelling options and eliminate the code overlaps.  Both Truespel, Spanglish, and most reform orthographies do this. 

While there are no spelling rules without exceptions, there are probabilities and these can be learned and used to increase the matches with traditional spelling.
Single Consonant Spelling Alternatives  (idea from McGuinness, p. 103)
Different ways the consonant sound is spelled at the beginning and end of words
Unlike the traditional orthography, Truespel always spells the same sound the same way
magic e endings usually indicate a long vowel:  exceptions have, give, ...

sound	key word	word beginning	word ending
b  d  f  g  h [silent in some words] j  k  l   [can be semi-vowel] m [can be semi-vowel] n  [can be semi-vowel] ng p  r  [can be semi-vowel] s sh t  v  w hw [semi-vowel] x  [ks, gs, sh/ch/s] y  [semi-vowel] z	boot, bib, tribe dog, did, mode fun, fife, enough  got, gig  hot, who, casbah job, judge, gem kick, cold, arctic  log, till, little man, comb, column  not, knot, gnome sing pig, pip, pipe  red, write, rare  sat, fence [confess] show, hush,  tot, tote, debt van, valve  win, when,  tax, xerox,  Xena yell, yacht zip, size, fizz	b  d  f  ph g  gu  gh h  wh (hw)  j  g c  k  ch l  m  n  kn  gn -- p  r  wr ur s  c  sc sh, ch t  v  w  wh  -- [foreign names] y z, x	b, be d  de f  ff  ph  gh  fe g  gue  gg  -- silent  terminal ge  dge k  ck  ic ke l   ll  el  le m  mb  mn n  gn foreign ne ng p pe r er re ce  se  ss  s  sh  [c in Saxon] t  bt ght te v  ve -- [ ow is a vowel] x  -no consonant endings s se  ze  zz  z
ch [tsh] [-k] ng  qu  [kw]  sh  sch  [sk-] [-sh] S'n  [shun] th [t, tth] th [d, dh, d, x]	chin  machine watch  singer/finger, bank quit, queen  shop, lush  school illusion, suspicion thin, breath [bretth] then, breathe	ch -- [Eng, Ing] qu  sh ch x sch, sk -- th  th  d	ch  tch ng  --  sh  - sion, cion, tion th  th the dth
silent h silent w,  w as vowel	honor, Allah who, write, wren - /u/ short u sound	h  silent w  silent - Welsh w [bwk]	h   silent marker low [silo] snow wow, cow,

The most likely spelling alternative listed first above.  More on silent letters
The options eliminated by most reform notations are in bold face. 
 

According to G. Dewey (1971), the basic problem is that each grapheme are used to represent not one but an average of 5 different sounds.  In a small abridged dictionary, one finds Dewey's 41 sounds represented 561 different ways.  The anomolies (irregular spellings) are reduced to 362 when one uses running word counts rather than the wide range of words found in a 42,000 word dictionary.  A sample of 100,000 words from various texts will include many repetitions and only about 7,000 different words.

OGD Restored English reveels a baisic sistematic coad for English spelling.    It is eezy tu reed and fwly predictabl in its speling.

It is not difficult to come up with a consistent code.  A perfectly alphabetic or phonemic orthography would have one and only one symbol for each sound.  The cells in the  following chart indicate phonemes (significant speech sounds), the letters indicate how three different notational systems reference these phonemes.

The first two columns represent pure (or uncombined) phonemes.  There are 12 of these. Most linguistis can agree on the minimum number of pure phonemes required to cover the significant sounds in English speech.  There is some disagreement on the number of combinations that need to be considered as unique phonemes but general agreeement on the 12 pure or uncombined vowel phonemes and up to 6 combinations.

The English spelling system (and augmented roman alphabet) as developed by King Alfred (ca. 878) and his clerics was nearly perfect.  That is, it had few or no alternative spellings for the same sound as we have today (be, seen, seat, believe, deceive, baby, donkey) and no overlap in the code (out, soup, soul, tough) where one spelling pattern stands for different sounds. (cf. McG, p. 81). OGD represents an attempt to reconstruct and restore King Alfred's English

Many words in English are spelled as they were 1000 years ago, but are no longer pronounced as they were 1000 years ago.

You can't teach from letters to sounds.  Phonics only decodes and suggests that there are more than 46 sounds.

The sounds of English were worked out toward the end of the 19th Century.  Johnson and Webster had an incomplete concept of the phoneme inventory.   Both spoke of long and short vowels as if made takes longer to say that mad. It may take slightly longer but it has nothing to do with maaaad.  It is not a long a, it is a diphthong /ei/ which is long.  The long and short or duration distinction is largely useless because the critical factor is not the vowel but the trailing consonant. 

The old Saxon language had long and short versions of the same vowel sound.  There are simply 12 different vowel sounds related only because they share the same letters. 

New Follick
or=oar, basic difference from Spanglish, a=sch    'agow   o=aa
it w'az a coars cowst thaet wi faund.

New Follick

 Under construction:  Ve eev re (shuud be)=  ra' (schwa) ri=ry (schwi) 
nice knees = nys nees,  good new food= guud niu food 
The preference for w instead of uu is because /u/ is short.  hoos huuk wos puut in mye pool.  Deconstruct to TO: hoos hook was put in my pool. (cant get to whose)

Alternate Notations
 

Chekt Klipt Spel3 CKS in a nutshell	IPA IPA Notation ''=shwa, ò =sh, j =y h =ng, 3=Zh
W'ns  'apon 'a  tym x  biutifl  dotr 'v 'a gret  m'aji5n wantd mor p'rlz  tu  pu.t  'amu.3  h'r  tre2iurz."Lu'k thru x sentr 'v x  mun `we.n i.t i.z blu," sed h'r m'uxr in a.nsr tu h'r kweschn, "Yu myt fynd yur hartz dizyr."  [ x =crossed d or eth]	W'ns 'pon ' taim ð' bjutif'l do:t''v ' greit m'd3iò'n want'd mo:': p':lz tu: put 'm'h h': tre3ju':z."Luk thru: ðc sent': 'v  ð' mu:n hwen it iz blu:," sed h'r m'th' in æns': tu: h': kwestò'n, "Ju: mait faind j': ha':tz di:zair." ipa
ALC Fonetik (Ver. of New Spelling)	Nyu Romaji - WorldEnglish
Wuns upon a tiem, the buetiful dauter of a graet majishan wonted mor perls to puut amung her trezhers. "Luuk thru the senter of the moon when it is bloo,"  sed her muther in anser to her qestion,"You miet fiend yur harts dezier."	Wans  apon a  taim the  biutifel  dooter ov a greyt  majishn  waanted mor perlz  tu  put  amang  her treezhyurz. "Luk thruu dhe ceenter ov dhe muun hwen it iz bluu," sed her mather in aenser tuu her queeschen," Yuu mait faind yur haartz dizair."
Truespel truespel-short.html	OGD positional systematic
Wuns upaun u tiem, thu byuetifool dautur ov u graet majishun waantud mor perls to poot amung hur trezhers. "Luuk thru the senter awv thu muen when it iz blue,"  sed hur muther in anser to hur kwestion,"Yue miet fiend yur haarts dezier."	Wuns upon a tym, the biutiful dawter ov a grait majition wonted mor perls tu puut amung her trezhers. "Lwk thru the senter ov the moon when it is blu." sed her muther in anser tu her question, "Yu myt fynd iur harts desyr.
Bophon by Bob Boden special font required multiple letters indicate a given sound	Saxon-Spanglish spanglish a=uh/ah ae=at  e=eh/uh  i=ee  ey=ay ai=eye  o=awe  ow=ou u=oo r=uhr  n=un w=/u/
Spanglish is an initial spelling alphabet that tries as hard to retain the spelling dialect of traditional spelling as it does trying to stay with the GA base pronunciation. Starting with SP, the learner  is gradually moved to the inconsitent and irrational traditional writing system. One of the first changes would be to replace s with ce and the plural z with s.  Then unstressed shn endings would be replaced with tion.	Wans  apon a  taim the  byutifal  dotr ov a greyt  majishan  wanted mor  perlz  tu  pwt  amang  hr trezhyurz. "Lwk thru the centr ov the mun wen it iz blu," sed hr mathr in aensr tu hr queschan, "Yu mait faind yur haartz dizair."
RiteSpell	Iqliz 2
Wuns upon a time, the biutiful dauter ov a grate maajishun wonted mor perls tu poot amung her trezhers. "Look thrue the center of the mune when it is blue." sed her muther in anser tue her queschun, "Yu mite fined yure harts desire.	W'ns 'pan ' tam d' but'f'l dot'r 'v ' gret m'jiz'n w'nt'd mor p'rls t' put 'm'q h'r trez'rs {Luc tru d' sent'r 'v d' mun wen it is blu} sed h'r m'd'r in 'ns'r t' h'r cwesk'n {U mat fand yor harts disar} [condensed]
19/44 changed RITEspel Wunce upon a time, the butifuldauter of a grate majician wanted mor perls tu put amung her tresures. Look thru the center of the moon wen it is blu, sed her muther in anser tu her question "U mite fynd your harts dezire"	32/44 changed Wans  apon a  taim the  byutifl  dotr ov a greyt majishan wanted mor perlz tu  pwt amang her trezyurz. "Lwk thru the center ov the mun wen it iz blu," sed her mather in aenser tu her queschan, "Yu mait faind yur haartz dezair."
Wans apon a taim d'a byutifl dotr av a greit majishn waanted mor prlz tu pwt among hr trezyrz.  Lwk thru the centr av d'a mun w'en it iz blu."  sed hr mathr in aensr tu hr cweschn, " Yu mait faind yur haartz dizair.	Wans apon a taim d biutifl dotr av a greit majishn wanted mor prlz tu pu.t among hr trezyrz.  Lu.k thru the centr av d mun wen it iz blu." sed hr mathr in a.nsr tu hr queschn, " Yu mait faind yur haartz dizair.

 Other comparisons of text in different notations

The prinses laft doutingly becauz she wuud not acsept thees werds. Insted, she uezd her imajinaeshun, moovd into the fotografy biznes, and tuuk pikchers of the loonar serfis in culor. "I perseev moest sertenly that it allways apears hoely whiet," she thaut. She allso found that she cuud maek enuf muny in aet munths so that she cuud bi herself too enormus huej nue jooels too. 

alien = ah-lee-ehn.  Can usually be understood as ah.  get rid of ei.  
owvr throw the owld cowd

possible/passable    paasibl/paesabl/posibl
diaphones, diaphonic area  ven diagram showing overlap
Usually greater dif in spelling than in speaking [pronuniation].
bud/bed back/beck jam/gem
b''ud/bed, baek/bek, jaem/jem
staina [stone], stein-g, sten-danish, stan, stone, stun
bone alone
bein allein
ben allena
 

To convert strings of text into ALC Fonetic, Truespel, or Cut Speling, go to the
Automated Spelling Converter: http://www.ldc.upenn.edu/cgi-bin/sb/orthography/convert.cgi 

Substitutions: 

• y for /ai/, y. for yod
• a' for schwa or turned e,  (a' can be abbreviated with a schwapostrophe as in 'v instead of a'v)
• r for turned e:
• syllabic R for the consonant R, thus [prlz] is pronounced pc:lz or pa'rlz.
• syllabic lrmn for l.r.m.n. wns = wa'ns = once;    wnz = wa'nz = ones.  n = /u.n/ except in initial position. 'nacrl (unnatural)
• 5 is an s-form for sh. 5n for "tion"  mo'5n (motion)
• 2 is a z-form for zh. mei2ur (measure).    2n for /zhcn/ as in ilu2n (illusion).
• 3 is an n-form or yogh for ng.  fi3gr (finger)  ba.3k (bank) ba.3g but not ___
• z for /z/ in some plural endings (ca.ts, dogz),  t for/t/in some past tenses (spe.lt)
• x for dh (abbreviation:  x for dhc or x') x (the), xat (that), xo'z (those),  xy thy (thy thigh), with (with)
• c for ch (optional abbreviation) as in Italian ciao (chao')

Perhaps the only way to avoid "positional spelling" is to require that the schwapostrophe [ ' ] be used with all syllabic consonants.
Problem words: URN/ UNATURAL:  'rn,   u'rn   /  'nnac'r'l,  u.nnac'rl,  a'nnachrl
OGD: urn, un-atura'l  apex=aypex (eipeks)