Regular English Pronunciation

Regular English Pronunciation is a suggestion for how English would sound if its pronunciation were logically connected to its spelling. Regular English Pronunciation is the antithesis of spelling reform as the solution to the problem of English spelling. Instead of changing our spelling, why not change our pronunciation? It wouldn't involve reprinting all books or breaking with the past. And pronunciation is in our hands, not in the hands of "authorities". All we have to do is to pronounce words the way they are spelled.


Background

The complex and often arbitrary quality of the relationship between English spelling and English pronunciation (Carney, 1997) has been blamed for the prolonged development period of literacy skills by English children (Thorstad, 1991) and the poor standard of their spelling at age ten (Spencer, 2002), the higher incidence of dyslexia in English-speaking nations (Paulesu et al, 2001), and the errors of second language learners of English (Cook, 1997). To get from the spelling of an English word to its pronunciation often involves a tortuous game of trial and error, where even adults make slips, as a reading of the poem "The Chaos" makes clear:
Dearest creature in creation,
Study English pronunciation.
I will teach you in my verse
Sounds like corpse, corps, horse, and worse.
I will keep you, Suzy, busy,
Make your head with heat grow dizzy.
Tear in eye, your dress will tear.
So shall I! Oh hear my prayer.
Gerald Nolst Trenite (1870-1946)

The Chaos is hard to read because it is full of words which share spelling patterns but do not share pronunciation patterns. Compare "creature/creation", "corpse/corps", "horse/worse", "head/heat". Such arbitrariness in spelling means that learners have to remember such a large number of rules and special cases that reading and spelling become a real challenge and source of difficulty for many.

But however convincing the scientific evidence for the reform of English spelling, and despite the advocacy of groups such as the American Simplified Spelling Board and the British Simplified Spelling Society, the last significant reforms in English spelling date back to the publication of the American Dictionary of the English Language in 1828 (Webster, 1828). It was this dictionary that popularised Noah Webster's proposals for the spelling of words like color, center, offense, traveled, organize, etc.

Why have there been no improvements in English spelling in the last 175 years? There are essentially three linguistic arguments against spelling reform, all somewhat weak: (i) that making the spelling follow the pronunciation would lose the spelling similarities between morphologically related words like nation and national; (ii) that no respelling by pronunciation could be consistent with every accent; and (iii) that competent adult speakers tend to read words and phrases as wholes (ideograms) rather than as phonetic components anyway (Smith, 1973).

However I believe it is the socio-political arguments against spelling reform that have prevented change. These include (i) that replacing existing printed materials is too expensive; (ii) that first-language English speakers treat a change to English as an assault on the cultural heritage of the English-speaking people; and, perhaps most important, (iii) that all respelling systems produce a written form that is perceived as child-like and fundamentally uneducated. Since adult speakers of English have battled through to achieve competence in English spelling, non-standard spellings are perceived as sub-standard.

Shaw Alphabet
The Shaw Alphabet was one re-spelling proposal

Regular English Pronunciation (REP) is an attempt to open up a second front in the battle to simplify the mapping between spoken and written English. Designed by Mark Huckvale in 2002, REP is based on the observation that if spelling can't be changed to match the pronunciation, maybe the pronunciation can be changed to match the spelling. Since pronunciation change requires little investment and can take place over a number of generations, it is likely to be more acceptable than spelling reform. REP is one suggestion for how English would sound if it were pronounced the way it was spelled. You can view REP as either a radical alternative to spelling reform (if you refuse to allow any spelling changes ever), or as a complementary approach (if you let pronunciation and spelling meet half way).


What is Regular English Pronunciation?

Regular English Pronunciation (REP) is an artificial accent of English - perhaps the first 'designer' accent of any existing language. The aims of REP are to provide a pronunciation of written English that is highly intelligible, easy to learn, and logically connected to the current spelling of English. Let's take these in turn:

Aim-1. Highly intelligible
REP aims to maintain as great a percentage of existing pronunciations as possible, compatible with Aim-2 and Aim-3. In fact over 75% of REP pronunciations of words in a typical running text are the same as standard pronunciations.
Aim-2. Easy to learn
REP aims to have as few pronunciation rules (letter-to-sound rules) as possible, compatible with Aim-1 and Aim-3. In fact REP version 1.01 consists of just 100 rules and 100 exceptions, well within the capability of a learner to remember.
Aim-3. Logical
REP will be as logically connected to the spelling as possible, compatible with Aim-1 and Aim-2. In fact REP is 100% logical but requires spelling to be parsed into morphological units before a pronunciation is assigned.

REP assigns a pronunciation to a spelled word using a set of rules and a list of exceptions. The rules are simple enough and the exceptions few enough that they could be printed on one piece of paper. Thus you would be able to replace your 500 page pronunciation dictionary with a single sheet of paper! The rules can easily be implemented as a computer program, indeed this web page contains a program for pronouncing words.

If by now you want to hear some REP accent spoken,  Click here!. Here is the full transcript.

The first thing you should notice is that the speech is highly intelligible with the vast majority of words having their conventional pronunciations. But here are some of the changed pronunciations in that passage. The pronunciations are given in SAMPA format, and I am comparing the REP pronunciations with standard Southern British English accent.

SpellingStandard British
Pronunciation
Regular
Pronunciation
oncewVnsQns
youngjVNjaUN
troubletrVb@ltraUb@l
mindmaIndmInd
friendsfrendzfri:ndz
askedA:skt{skt
only@UnlIQnlI
answerA:ns@{nsw@

From these you should begin to see that the REP forms are just more logical than the standard forms - why is "mind" pronounced /maInd/ in standard English when "wind" is pronounced /wInd/ and most other places where you have "in" in English you get /In/? Likewise "ou" has many different pronunciations in standard English: "out/soup/could/young/thought/cough/...", but in REP there are only two rules: "our" => /aU@/ (viz "hour"), otherwise "ou" => /aU/.

Although logicality is a main aim of REP, the need to maintain high intelligibility means that we need to treat very common words with idiosyncratic pronunciations as exceptions to the rules. Thus it would not make sense to change the pronunciation of the common grammatical words "could/should/would" with the standard REP rule for "ou", and these words are treated as exceptions, along with the peculiar standard pronunciations of "of/are/do/don't/...".

REP is an ongoing project, not a definitive set of rules. Thus like computer software and unlike real accents it comes in "versions". The current version of REP is 1.01. You can look at the list of rules and exceptions in REP 1.01.

An investigation into the the intelligibility of REP was conducted at University College London in 2002/3, and reported at the 15th International Congress of Phonetic Sciences. The study had two important outcomes: (i) it confirmed that REP was highly intelligible, having about 90% of the intelligibility of standard pronunciations on a difficult intelligibility in noise task, and (ii) it discovered that listeners adapted to REP, showing significantly improved intelligibility (compared to a control condition) even within the first hour of exposure .

There is still more work to be done with REP:

  • Refining the rules
  • Considering how to deal with "unpronounceable" transcriptions (e.g. "people")
  • Formalising the morphological analysis component
  • Studying the regularisation of English lexical stress
  • Studying the effect of REP on spelling accuracy (i.e. is it easier to take dictation when the speaker is speaking REP?)
  • Producing tutorials and teaching materials
  • Recording more demonstrations
  • Putting REP into computer speech synthesis systems.

Your contribution is welcome. If you add to the project be sure to build a web page and send us the address so it can be linked to this page.


Regular Pronunciation Tool and Examples

Pronunciation tool

Enter the word you want pronounced into this form and click on 'Pronounce'. The pronunciation will be calculated by rule and displayed in SAMPA pronunciation format:

English Spelling (e.g. UN+LIKE-LY):
Regular English Pronunciation : SAMPA format

You can see a list of the rules in Regular English Pronunciation version 1.01.

You can hear me reading some example transcriptions in Regular English Pronunciation version 1.00.


Pronunciation Tools and Resources

The development of REP involved the creation of specialised resources and tools. These are made freely available below for anyone interested in furthering the REP philosophy. If you make use of these be sure to acknowledge their source.

Lexical Database

For rule development a database of the 10,000 most common words was built, using frequency data from the British National Corpus. This database contains (i) the normal spelling, (ii) the word frequency in written text, (iii) the word frequency in spoken text, (iv) the morphologically analysed spelling (according to REP), and (v) the standard pronunciation. This database is used by the PRuler program for rule development and evaluation. Download the REP lexical database (450kB).

PRuler Windows Application


PRuler application

The PRuler program was used to develop and evaluate REP. It allows the user to add, edit and delete rules and to evaluate them on the lexical database. See figure above. PRuler is written in Visual C++ and runs on Windows platforms only. You can download Pruler Vs 1.2. You can also read the help file.

REP Text to speech synthesis

The UCL ProSynth project synthesizer now speaks in Regular English Pronunciation. You can either try out our Web Synthesis Demonstration or Download Windows Application. In either case set the Input Format to "Regular English with diacritics" to get the REP accent.


ProSynth application


Frequently Asked Questions

Will I have to learn to speak "Regular English"?

I think it would be wonderful if Regular English became accepted as a legitimate accent of English, but that is not the aim. The project is not trying to change how any current speaker of English talks. If the REP accent starts being used (by computers for example), you will need to learn how to listen to it. One of the major design constraints of the new accent is that it should be highly intelligible to existing speakers of English, so even this task will not be difficult. One reaction of the subjects in our experiment was that REP sounded like a foreign speaker of English.

Won't REP lose the distinctiveness of some words?

All accents have words which sound the same but which mean different things. In my accent of English, the words PAW, POUR, PORE and POOR all have the same pronunciation. REP is no different, just that it will have its own idiosyncratic list of homophones. While we may lose the distinctiveness of some pairs, we will gain in the distinctiveness of others.

Won't REP lose some of the historical character of English? Won't we be losing something?

There is a natural conservatism among speakers of any language, and an understandable resistance to change. However languages themselves change all the time, particularly in accent and in word usage. It is a mistake to see all change as a priori destructive. To find a pronunciation of English that is closer to the spelling may well require that we return to earlier forms of English accent. So in fact we will be bringing back some historical characteristics of English rather than displacing them.

Won't REP mean that we will all end up sounding the same?

Technically REP defines a family of accents not a single accent. There will still be plenty of scope for geographic or sociolinguistic variation. Essentially REP will define the phonological transcription of English from spelling, but speakers will be free (as indeed they are now) to interpret the phonology according to their own vocal tract preferences and whims. Thus the word "now" in REP will sound just as different in Southern British, Scottish and Northern Irish accents as it does now.


Publications

M.Huckvale & M.Shaw, "The Intelligibility of a Spelling-Regular English Accent", in proceedings 15th International Congree of Phonetics Sciences, Barcelona, 2003. Download PDF.


Bibliography

  • E. Carney, English Spelling. London: Routledge, 1997.
  • V. Cook, "L2 users and English spelling", Journal of Multilingual and Multicultural Development, 18 (1997), 474-488.
  • E. Paulesu, J-F. Demonet, F. Fazio, E. McCrory, V. Chanoine, N. Brunswick, S. Cappa, G. Cossu, M. Habib, C. Frith, U. Frith, "Dyslexia: Cultural diversity and biological unity", Science 291 (March 16 2001), 2165.
  • F. Smith, Psycholinguistics and reading. New York: Holt. Rinehart & Winston, 1973
  • Spelling Reform Society www.spellingsociety.org
  • K. Spencer, "English spelling and its contribution to illiteracy: word difficulty for common English words", Reading Literacy and Language, 36 (2002) 16-25.
  • G. Thorstad, "The effect of orthography on the acquisition of literacy skills", British Journal of Psychology, 82 (1991) 527-537.
  • N. Webster, American Dictionary of the English Language, 1828.


Feedback

Please send constructive criticism and offers of help to M.Huckvale@ucl.ac.uk. Because of the volume of mail we receive, you may not get an individual answer.


Copyright

Prior to release, all data associated with the Regular English project remains the intellectual property of Mark Huckvale (© Mark Huckvale University College London).


© 2004 Mark Huckvale University College London Last revised: July 2004
Valid CSS! This site uses
Cascading Style Sheets.