A system for encoding English pronunciation

Alan Beale
December 31, 2004


This page proposes a new system for encoding the pronunciation of English.  I call it RESCUE (Readable Easy Sound Coding for Use with English - I was tempted to define the first two letters as Really Excellent, but I'll be realistic about this).  While the idea for RESCUE was my own, I'd like to express my appreciation to David Barrow for very useful discussions on how to improve it, particularly for use with British English.

The purpose of RESCUE is not to be an English spelling system, or an initial teaching alphabet.  Its purpose is to be a code in which the members of the Saundspel group can easily communicate how they (or those they hear) pronounce the English language.  It is not intended to compete with or replace IPA or SAMPA - its goals are much more modest.  I developed it due to significant opposition to the use of SAMPA by a number of prominent Saundspel members.

As David pointed out on the Saundspel group, using keywords is not a very efficient way of communicating pronunciation, but a code, such as SAMPA, is.

I proposed this code because of the need for some easy and well-defined code in which members could communicate pronunciation differences to each other.  (SAMPA/IPA is not easy, and because it is a system designed for and by experts, it seems that many of us non-experts cannot understand it or use it accurately.)  I've tried to set RESCUE up so that objections like 'it uses "j" for "y" - which is too unlike English to tolerate' will not apply.  It will not serve to discuss minute differences in pronunciation - SAMPA/IPA is needed for that - but an awful lot of the Saundspel discussions could be accommodated by it, I think.

I have promised that I will never attempt to promote RESCUE as a full-fledged alternate spelling system.  Group members can use it without concern that it will compete in some fashion with their own creations.  This is one reason that RESCUE eschews abbreviations and other conveniences that can make an orthography more attractive, but are of less utility in a pronunciation code.

In response to an earlier version of RESCUE, David suggested some conventions for more accurate representation of diphthongs. Such conventions could be added, but they are not part of this proposal. In essence, RESCUE defines encodings for each English phoneme, but does not attempt to represent more exact phonetic information about them. For purposes of RESCUE, /eI/ and /EI/ are the same thing.

Basic Principles

The basic principles of RESCUE are as follows:

First and foremost, it tries to be transparent.  This means that a literate English-speaking reader is likely to interpret it correctly, without having to study a detailed set of rules.  Wherever possible it tries to use familiar notations, even if there might be technically superior alternatives.  (For instance it represents the diphthong of cow by ow, even though au would be more phonetically accurate.)

It uses commonly understood digraphs to bring order to the English vowel system. Most of the digraphs can be replaced by letters with accents. 

It uses digraphs ending in h to add consonants which are not generally associated with a single English letter.

It uses the underscore to separate digraph elements when the occurrence of the digraph is misleading.

It does not allow abbreviations, even when a reasonable person would understand the meaning of the abbreviation. 

RESCUE is not a spelling system.  The idea is that you use RESCUE to describe a sequence of sounds, as perceived by the writer, and that there is no such thing as a "wrong spelling", if it accurately describes that sequence.

RESCUE is written in back-slashes, to avoid confusion with SAMPA.  (Alternately, for those with keyboards which place the backslash inconveniently, forward slashes can be used, preceded by "rs" to indicate RESCUE rather than SAMPA.  Thus, the preferred form for the normal American pronunciation of rescue is \reskyue\, but rs/reskyue/ may be used instead,)


RESCUE uses the normal English consonant letters, including y and w - I will save myself typing by not listing these here.

It uses the following consonant digraphs:

ch - /tS/ {ch}ip
dh - /D/  {th}en
ng - /N/  ri{ng}
sh - /S/  {sh}ip
th - /T/  {th}in
wh - /hw/ or /W/
zh - /Z/  vi{si}on


When the ng sound is followed by a k or g, the combination is written ngk or ngg.  An n followed by a g, as in "ungrateful", is written n_g (\un_graetf&l\).

The wh sound can also be written hw, if you wish to commit to this specific realization, excluding the voiceless w /W/ .

Often the letter h must be preceded by an underscore to avoid ambiguity, as in \mis_hap\, \not_hoel\ or \ad_heesiv\.

The digraph qh may be used to indicate the final sound of loch (/x/).  I have added this digraph because the sound occurs in a handful of English words - this is not intended to imply that RESCUE is useful for describing languages other than English.  (While the digraph kh is more familiar for this sound, using it would necessitate disambiguation of words like \singkhoel\.)


The vowels are written as follows in RESCUE.  Example words were taken from Bob Boden's NBC English work:

a   - /{/    p{a}t
aa  - /A:/   f{a}ther
ae  - /eI/ or /EI/
aw  - /O:/   w{a}ter
e   - /E/    p{e}t
ee  - /i:/ or /i/

e&  - /E@/   b{ear} - mostly British
i   - /I/    p{i}t
ie  - /aI/   p{i}nt
i&  - /I@/   b{eer} - mostly British
o   - /Q/    p{o}t - British only
oe  - /oU/ or /@U/
oo  - /U/    p{u}t
ow  - /aU/   {ou}t
oy  - /OI/   v{oi}d
u   - /V/    p{u}p
ue  - /u:/ or /u/
uu  - alternative to oo
u&  - /U@/   t{our} - mostly British
&   - /@/    circ{u}s - the schwa
&&  - /3/    f{ur} - British
&&r - /3`/   f{ur} - American



An American pronunciation of pot would be written \paat\.  The British pronunciation would be written \pot\.  

uu may be used in place of oo by any who feel that oo is hopelessly unsuitable for use for the vowel of s{oo}t due to its association with the long u sound of s{oo}n and b{oo}t. 

ue never has an implied y sound.  The word unite is rendered in RESCUE as \yueniet\.

The y in sill{y} would be written as i in RESCUE if pronounced that way (a short i), or as ee if pronounced as a long e.  I would transcribe silly as \silee\ in RESCUE, while \sili\ might be more appropriate for a British speaker.

Though the e&, i& and u& diphthongs are shown for British English, they might be sometimes appropriate, before r, in American English as well, as in minority pronunciations of Mary \me&ree\ or theoretical \thi&retik&l\.

Technically, some vowel trigraphs are ambiguous, because the letters can be grouped in two different ways.  In practice, this does not create any problems.  For instance, \plae&r\ can only be \plae_&r\ (player) - \pla_e&r\ is not a combination of sounds which could ever occur in the English language.  Similarly, \dueet\ must be \due_et\ (duet), not \du_eet\.  It is always permissible to use the underscore in such cases for clarity, as in \boo_yaan\ (an American pronunciation of "bouillon").

The  tilde may be used after a vowel to indicate nasality as in (some pronunciations of) uh-huh \u~hu~\ or denouement \daenuemaa~\.


r's which are not pronounced are not written.  If my Longman's is to be believed about the British pronunciation of barber, it would be written as \baab&\ in RESCUE. 

If you wish to distinguish a syllabic consonant from a schwa-consonant sequence, indicate the syllabic by writing no vowel, and doubling the consonant, as in \taebll\, \matrr\ or \butnn\.

If you choose to write RESCUE with case distinctions (though there is no reason to do so), you should use @ as the upper case of &, as in \@merik&\.

You are not allowed to leave the ending e off of an ending long vowel: you cannot write \heero\ for \heeroe\ - RESCUE is a precise notation, not a spelling system.

RESCUE is intended to adapt to several theories about English phonemes held by members of the Saundspel group to which others may not subscribe.  If you believe that there is no schwa, and it's just an unstressed short u, you should write campus as \kampus\, and everyone else will write it as \kamp&s\.  If you believe that the two vowel sounds of murder are the same, you will write it \m&rd&r\ or \murdur\, depending on whether you think the schwa is the same thing as short u, while those who believe the two are different will write \m&&rd&r\ or \m&&d&\, depending on their rhotic preference.

To clarify the above a bit, different English speakers may perceive 1 to 3 different vowels in the words "supper" and "search".  Those who hear three distinct sounds will write \sup&r\ and \s&&rch\ (possibly without the r's).  Those who perceive both of the r sounds to be the same, distinct from the u, will write \sup&r\ and \s&rch\ (again possibly without the r's).  Those who find all three of the vowels the same may write either \supur\ and \surch\ or \s&p&r\ and \s&rch\, whichever the transcriber feels is more correct (and with or without the r's, as appropriate).  David notes that & is likely to be a more accurate representation of the sound in this case.

David also recommends that, if the same sound is used in all of cot, caught and core, as in Scottish English, the aw representation be used, as in \kawt\ and \kawr\.  If cot and caught are the same but core is different, as for many Americans, the representations \kaat\ and \kawr\ should be used.

Stress can optionally be shown in RESCUE.  A stressed vowel is indicated by following the vowel with an apostrophe: \si'lee\.  If you wish to distinguish primary and secondary stress, use " for secondary stress, as in \ra"tif&kae'sh&n\.  Note that if you choose to show stress with RESCUE, you must mark all stressed syllables, even if some of them are obviously stressed.  This is part of RESCUE's no-abbreviations policy. 

Similarly, syllabification may optionally be shown with a period, as in \kaa"n.sti.tue".sh&.na'.li.tee\.  Since two vowels not part of a digraph must be in distinct syllables, it is unambiguous to omit a period following a stress mark, as in \kaa"n.sti.tue"sh&.na'li.tee\.  If you indicate syllabification, it is unnecessary to also use the underscore to prevent ambiguities: \mis.hap\ is no more ambiguous than \mis_hap\.

As I have already said, RESCUE is not a spelling system.  If you pronounce "thinking" with two long e's, you should render it in RESCUE as \theengkeeng\, and no one can say you are wrong (unless they've heard you speak and believe you are misperceiving your own pronunciation).

RESCUE with diacritics

The above is the base RESCUE system. To improve readability and brevity, one may OPTIONALLY replace certain digraphs and symbols with other Latin-1 symbols, as shown below.  Note that you can mix diacritics and digraphs if you are so inclined, and that, even if you use all the possible Latin-1 characters, you must still use the underscore as needed to avoid ambiguity with the digraphs.

With these substitutions, RESCUE ends up looking a lot like


dh  - ð
ng  - ñ
th  - þ
aa  - â or ä
aa~ - ã
ae  - á
aw  - ô or ö
ee  - é
ie  - í or ý
oe  - ó
oo  - û or ü
o~  - õ
ue  - ú
uu  - û or ü
&   - ø (Both by itself and in the diphthongs eø, iø and uø)
&&  - ê or ë (øø may also be used)
&&r - êr or ër (øør may also be used)

(Allowing a choice between circumflexed vowels and umlauted vowels is just a way of accomodating input with different keyboard maps, for which one or the other may be easier to enter. Allowing ý as an alternative to í is for the benefit of those, like myself, who often have trouble distinguishing i and í.)


Here is a famous passage from Douglas Adams' Hitchhiker's Guide to the Galaxy rendered into RESCUE twice (once for my own brand of American English, and once for David's British sort), first with digraphs and then with the Latin-1 alternatives. I expect that no one interested in spelling reform should have any problems reading either one, but that many may notice differences from their preferred pronunciations.  Normally, of course, one would not expect to see such long passages in RESCUE, but it is an easy way of illustrating the system with a variety of words.

Digraph version (American)

In menee &v dh& mawr rilakst siv&l&zaesh&nz awn dh& Owt&r Eest&rn Rim &v dh& galiksee, dh& Hichhiek&rz Gied h&z awlredee s&plant&d dh& graet Insiekl&peedee& G&laktik& az dh& stand&rd ripaaz&tawree &v awl naalij &nd wizd&m, fawr dhoe it haz menee oemish&nz &nd k&ntaenz much dhat iz &paakrif&l, awr &t leest wie&ldlee inakyoorit, it skawrz oev&r dh& oeld&r, mawr p&destree&n w&rk in tue impawrt&nt rispekts. F&rst, it iz slietlee cheep&r; &nd sek&nd, it haz dh& w&rdz DOENT PANIK inskriebd in laarj frendlee let&rz awn its kuv&r.

(This transcription represents my own speech.  The spellings of \w&rk\, \f&rst\ and \w&rdz\ accurately reflect my pronounciation.  An American speaker with different vowels in the syllables of murder would denote them \w&&rk\, \f&&rst\ and \w&&rdz\.)

Diacritic version (British)

In mené øv ðø mô rilakst siviløzáshønz on ðø Owtør Éstørn Rim øv ðø galøksé, ðø Hichhýkøz Gýd høz ôlredé søplântid ðø grát Insýkløpédéø Gølaktikø az ðø standød ripozitré øv ôl nolij ønd wizdøm, fø ðó it haz mené omishønz ønd køntánz much ðat iz øpokriføl, ôr øt lést wýldlé inakyûrit, it skôs óvø ðø óldø, mô pødestréøn wêk in tú impôtønt rispekts. Fêst, it iz slýtlé chépø; ønd sekønd, it haz ðø wêdz DÓNT PANIK inskrýbd in lâj frendlé letøz on its kuvø.

(This transcription represents David Barrow's speech.  Many thanks to David for offering it to me.)

To comment on this page, e-mail Alan at

Go to home page
Go to spelling system roster