Karen's Linguistics Issues, December 2001 | This Month's Articles | Previous Months

This article was originally published in ENGLISHTEACHING professional, Issue 21, October 2001. © Copyright Robin Walker.

Pronunciation for International Intelligibility

by Robin Walker, December 2001

If you cannot read the phonetic symbols in this article, you will need to download the IPA phonetic symbols software.

The background. Until very recently few teachers or learners really questioned the idea that in order to be understood when speaking English, students would need to get as close as possible in their pronunciation to one of the dominant native-speaker accents, such as Received Pronunciation (RP), the standard British accent, or GA (General American), the USA equivalent. Nor did anybody really bring under scrutiny the idea that the measure for successful pronunciation should be the speaker’s degree of intelligibility as determined by a native-speaker. The last decade, however, has brought about such a significant change in the role of English throughout the world that it is essential to re-examine this situation. English is currently regarded as the world’s principal international language, as a result of which there are now more exchanges between non-native speakers of English (NNS-NNS), than between non-native speakers and native speakers (NNS-NS). In the immediate future at least, this situation is not going to change in favour of the minority of native speakers, and so suddenly the hegemony of their particular (and sometimes peculiar) accents is under fire. 

The demise of native-speaker accents. One of the first people to question the idea of a native-speaker accent as a model or norm was R. Macaulay. In 1988, in an article provocatively entitled ‘RP R.I.P.?’, he pointed out a simple but surprising truth about this supposedly prestige accent: less than 3% of the UK population actually used it at that time, and the percentage was falling. Macaulay also drew our attention to another forgotten reality of RP, namely that it was an accent which enclosed ‘unnecessary’ difficulties for learners of English, such as the ‘r’ sound or certain of its diphthongs. 

These arguments were reiterated in The Cambridge Encyclopaedia of the English (1995), where Professor David Crystal suggests that a standard Scots accent would actually be a better model for most learners of English. Moreover, he adds, RP is constantly changing, a fact that was confirmed only recently by John Wells, Professor of Phonetics at University College, London. A study he carried out in 1998 has shown how with words like chance  and one,  younger native speakers of British English have a clear preference for the non-RP vowels, making these rhyme with the vowels in can and gone, as opposed to those in car and sun. 

Realistic goals. If a native-speaker accent is an undesirable goal for our students for the arguments outlined above, it is at the same time a wholly unrealistic goal for the vast majority of learners, many of whom bear the scars of fruitless attempts to satisfactorily differentiate between ship and sheep, or between hat, hut, and hot. The dilemma, of course, is what we put in their place once we knock RP, GA and other native-speaker accents off their pedestals? 

Joanne Kenworthy, in Teaching English Pronunciation, puts forward the concept of comfortableintelligibility as a suitable goal for the majority of learners. The term is self-explanatory, but does not actually pin down which features of English pronunciation need to be learned in order to attain this intelligibility. With this problem in mind, perhaps, Bryan Jenner attempted to determine “what all native speakers of all varieties have in common which enables them to communicate effectively with native speakers of varieties other than their own”. The results of his analysis were brought together in the Common Core, which is a list of the features of English pronunciation Jenner considered essential for intelligibility anywhere in the world. 

The Lingua Franca Core. Whilst constituting a great improvement over the frustration and futility of attempting to gain native or near-native command of all of the features of English pronunciation, the Common Core still did not fully address the reality of English as an International Language (EIL): the listener for Jenner’s core continued to be the native speaker. Because of this, and on the basis of extensive data collected in multilingual EFL classes, Jennifer Jenkins modified the Common Core so as to take the reality of fully EIL into account. The resulting Lingua Franca Core identifies 7 areas in which it is essential to eliminate error in our students’ pronunciation: 

·     Vowel quantity: vowel quality varies widely from one NS accent to another. However, the length differences between the vowels of English feature in all accents, and the long English vowels are very long in comparison with average vowel lengths in other languages. Because of this, the distinction between long and short vowels is more important than exact vowel quality, and should be clear in speech. With diphthongs, just as with pure vowels, length should be our main concern rather than exact quality.

·     Consonant conflations: when a consonant of English does not occur in a learner’s mother tongue, the ‘missing’ sound is substituted with something similar from the speaker’s first language. The substitution of one consonant for another can cause serious confusion for both NS and NNS listeners. Substituting /p/ for /f/, a Korean speaker of English, for example, would produce paint for faint, or copy, for coffee. An exception is made regarding /T/and /D/, since these two phonemes do not occur in the majority of the world’s languages, including some native-speaker varieties. A speaker with an Irish accent, for example, pronounces dare and there in the same way as are tin and thin.

·     Phonetic realisations: another strategy used when struggling to pronounce English is to use a sound from your mother tongue that is close to the required English sound. However, some such approximations may lead to unintelligibility, as with/B/, the fricative sound the Spanish use for the ‘b’ in cabin, or /F/, the sound they use for the ‘g’ in again or ‘a girl’.

·     Consonant cluster simplification: learners employ two radically different strategies to deal with consonant clusters. Of the two, deleting one of the consonants to simplify a cluster can affect intelligibility considerably, whilst the addition of a vowel seems to cause fewer problems. Turkish speakers of English, for example, will often insert a vowel before or after an ‘s’, so stone will sound like istone or sitone. However foreign this may sound to a native speaker, it is much less damaging to intelligibility than eliminating one of the two consonants, leaving tone or sone, for example.

·     Prominence and weak forms: on the continuum between stress-timed and syllable-timed languages, English, with its multiple weak-form words and its heavily-reduced unstressed vowels, lies well towards stress-timing. As a result of this, learners frequently deem native speakers as harder to understand than non-natives. Ability to deal receptively with weak-forms and other connected speech modifications is a goal for all who will come into contact with native speakers. However, in terms of NNS production, teaching should focus on achieving adequate prominence on the stressed syllables, rather than on attaining perfect weak forms or schwas for the reduced vowels. With correct prominence, even if totally lacking in weak forms or schwa, a learner’s English will be intelligible.

·     Tone groups: failing to use tone groups to divide the stream of speech into manageable, meaningful chunks has a serious effect on intelligibility. On the one hand, it may lead to breaks in speech in unexpected places, reducing intelligibility, whilst on the other, it reduces planning time for the speaker, which will inevitably lead to new errors of all types.

·     Nuclear/contrastive stress but not tone: one almost unique characteristic of English is the way in which it varies the most prominent stress in a tone group to create meaning. Thus, the utterance They rented a FLAT does not carry the same meaning as They RENTED a flat. Many other languages use syntactic changes to change meaning, and so learners often fail to pick up the significance of the tonic stress, and naturally fail to use it appropriately. Clearly, to put the main stress on the wrong word in an utterance, will direct the listener’s attention to the wrong place, leading to confusion, whether the listener is a native speaker or not.  

Incorrect word stress, which is widely regarded as a source of unintelligibility in spoken English, is probably the most surprising omission from the Core. However, breakdowns in communication attributable to word stress were not encountered in Jennifer Jenkins’ research data, which suggests that it is far more of a problem for NS listeners than for NNS-listeners. 

Comparative lists and the Lingua Franca Core. Traditionally as teachers we have referred to the widely available lists of indications as to where to concentrate our efforts when teaching pronunciation. These lists relate in detail the individual sounds or the connected speech features that will cause difficulties for a learner of a specific L1. Essentially they constitute a summary of what our learners cannot do, and in a great many cases, the resulting list is quite formidable. Table 1, for example, is for learners whose mother tongue is Spanish.

Table 1

Priorities for Spanish Speakers of English, based on Kenworthy (Teaching English Pronunciation), O’Connor (Better English Pronunciation) and Taylor (Pronunciation in Action).


1.        /i:/ and /I/ confused and a vowel more like /i:/ used for both (HP)

2.        /Q/ and /E/ confused and /E/ used for both (HP)

3.        /Q,,A:/ confused, a sound like //being used, except where ‘r’ occurs in the spelling, when /A:/ is replaced by vowel + /r/ (HP)

4.        /Å/, /´U/ and /ç:/ confused (if there is no ‘r’ in the spelling), a vowel intermediate between /Å/ and /ç:/ being used. Where ‘r’ occurs in the spelling /ç:/ is replaced by vowel + /r/

5.        /u:/ and /U/ confused with a vowel similar to /u:/used for both

6.        /Œ:/ is replaced by the vowel + /r/

7.        /´/ is usually replaced by the vowel suggested by the spelling (HP)

8.        /eI/ and /E/ confused (HP)

9.        //, // and // are replaced by the vowel + /r/                                   

   10.         No length variation - all vowels generally have the same length as the English short vowels, so long vowels seem too short (HP)


1.        Confusion between /b/ and /v/ - /B/ tends to be used for both, sometimes /b/ is used for /v/ (HP)

2.        /t/ is very dental in Spanish

3.        /d/ and /D/are confused and often used interchangeably (HP)

4.        /g/ is often replaced by a similar friction sound (/F /)

5.        /s/ and /z/ confused - /s/ used for both  (HP)

6.        /S/ does not occur in Spanish - /s/ used instead (HP)

7.        /Z/ does not occur in Spanish - /s/ used instead

8.        /dZ/ and /tS/ confused - /tS/ used for both, or the sound in the Spanish ‘yo’ is used instead

9.        /j/ does not occur - the sound in ‘yo’ is used instead (HP)

10.         /h/ does not occur and is either deleted or substituted by /x/(HP)

11.         /N/ does not occur and is substituted by /n/ (HP in some cases)

12.         /l/ is always clear in Spanish

13.         /r/ in Spanish is a tongue-tip flap or roll

14.         /w/ does not occur and is substituted by /b/or /B/, or by /g/ if /w/ comes before /U/

15.         /p, t, k/ are not aspirated in Spanish (HP for /p/ and /t/)


1.        /e/ is inserted before /s+C/ or /s+C1+C2/ clusters

2.        Learners tend to add /s/ for plurals: ‘pens’ sounds like ‘pence’

3.        /s + C + s/ clusters difficult, with one of the /s/ being deleted

4.        /s/ sometimes deleted when final in a word-final cluster

5.        Final clusters with /t/ or /d/ are problematic, with deletion of /t, d/

      or the insertion of a vowel

Stress, rhythm and intonation

1.        Incorrect stress of compound words and ‘adj + noun’ combinations

2.        Speakers have an over-even rhythm. Stressed syllables occur, but each syllable has approximately the same length

3.        There are no weak forms in Spanish

4.        There is no equivalent system in Spanish to the system of nuclear stress of English

5.        Pitch range is too narrow and lacks high falls and rises

6.        Final falling pitch may not sound low enough

The rise-fall seems difficult

The shaded areas are those features that lie outside the lingua franca core, and what is immediately obvious once we ignore these areas, is just how much lighter the learner’s load has suddenly become. Gone are the bugbears (for learner and many non-native teachers alike) of those dreaded vowel sounds. And let’s be honest, just how often in genuine conversation will two people confuse a ‘ship’ from a ‘sheep’, or a ‘hat’ from a ‘hut’. Whilst not discouraging attempts to achieve good vowel quality, the core draws teachers’ and learners’ attention decisively towards the far more important issue of vowel length. 

There is a similar, significant reduction in the workload in terms of stress, rhythm and intonation. Once again, gone are those odious exercises on discriminating between fall and rise tones, which it turns out do not even reflect what is happening. In practice, it is both possible and correct to ask a ‘Yes/No’-question with a fall tone, or a ‘Wh’-question with a rise, as is so elegantly explained in David Brazil’s Pronunciation for Advanced Learners of English. Similarly, clusters come down to dealing with deletion of consonants (elision), except perhaps, where this would be natural among native speakers, as is the case for /t/ or /d/ when they come between consonants, as in can’t come or Hold the line. Here deletion could be openly encouraged since it facilitates speech. 

Good vowel length, good pronunciation of most of the consonants, good handling of clusters, the avoidance of incorrect deletions, prominence and good tonic stress - these are the focus of our work on pronunciation, together with one area which did not come up in any traditional list, but is a priority in the LFC, namely the appropriate use of tone groups. 

Monolingual groups. By rejecting native-speaker accents such as RP or GA as norms, and by applying the concept of the lingua franca core to the ‘difficulties’ our learners are claimed to have, we arrive at a much reduced, and much more achievable set of pronunciation goals. (If you work with a monolingual group other than Spanish, try the LFC ‘filtering’ exercise carried out above for the problems your learners are said to have. The result will almost definitely make you feel better about teaching pronunciation to your classes.)

However, this already bright picture is made even more so if we take a close look at how it is that precisely with monolingual groups, who make up the majority teaching situation around the world, the learners’ first language, instead of being ignored or, worse still, seen as an obstacle, actually provides access to an important number of our new pronunciation goals.

The consonant phonemes /z, S, Z/, for example, are described traditionally as not occurring in Castilian Spanish. However, a basic knowledge of Spanish phonetics reminds us that /z/is an allophone for /s/and is found in words like mismoor asno. The /S/, though not an allophone of Castilian, is common in a number of the other languages spoken in the Iberian peninsular, whilst the /Z/sound features in the pronunciation of Argentinean Spanish. In the same way, a Portuguese speaker of English having difficulty with the /tS, dZ/ sounds can be referred to the Brazilian Portuguese pronunciation of the words tia or dia, or to the accents from Sao Paulo city, Carioca or Mieiro. Another very English phoneme, /N/,does not ‘officially’ occur in French, yet once again a good working knowledge of French phonetics comes to our aid: the ‘n’ in en guarde or in dingue offers a very close approximation to the English velar /n/. 

The technique we are employing in all of the above examples is sometimes known as association. We link the pronunciation feature we are aiming at with an equivalent or near equivalent feature in the student’s own language, or in related languages, dialects or accents. As a technique, it works especially well with consonants, although it also works with certain aspects of stress, rhythm and intonation. Elision, the deletion of certain sounds in rapid speech, may actually be a feature of your students’ mother tongue. In Spanish, for example, the /s/ is deleted in combinations with /r/ such as más rojo or más rápido

By approaching these sounds through the mother-tongue of monolingual groups, we are not only contemplating a reduced, achievable set of goals, but also switching the emphasis to what learners CAN do as opposed to what they supposedly canNOT do. This apparently minor shift in viewpoint could well prove critical to the success of our pronunciation work: we reduce the negative psychological effects of always stressing what is wrong, whilst at the same time stressing the value of the student’s own language as a tool for speaking English. This should prove especially positive for those students who explicitly or otherwise do not wish to lose their own national or regional identity on imitating pure ‘English’ sounds. 

Non-native teachers with monolingual groups. Monolingual groups may not be the ideal context in which to improve pronunciation skills, but they are a reality for most of us, and one, which will not change, in the immediate future. However, as Donna Brinton indicates, “... the task of the EFL pronunciation teacher is simplified by the homogenous first language background of the learners since knowledge of this language can generally be brought to bear in constructing the pronunciation syllabus.” 

Given such groups, who is best situated to deploy the detailed knowledge of the phonetics of both English and the students’ L1 that is needed to produce solutions of the sort seen in the last section? The answer quite simply would be fluent, bilingual speakers with a good knowledge of the two phonetic systems. But between two fluent, bilingual speakers, one from the UK/USA/etc. and one sharing the same nationality as the learners, who would be best? The answer to our question may lie in considering the following: 

·       The native-speaker of English can provide native-speaker modelling in class.  But since a native-speaker accent is not our goal, this advantage is of limited value, and competent non-native speakers can also produce perfectly adequate modelling.  

·       A good working knowledge of both languages’ phonetic systems is necessary, but many UK-trained EFL teachers qualify without even a good knowledge of the phonetics of English, and only a very few study the phonetics of their learners’ mother tongue. In contrast, in many countries a degree in English means the study of the phonetics of both the L1 and English systems. 

·       Who knows best, through personal experience, the physical and psychological difficulties in learning a second language pronunciation? Again, on average it is the non-native speaker of English who has greatest empathy with their students’ difficulties and diffidences. 

·       Who knows through personal experience which approximations to our pronunciation goals are sufficiently good to be intelligible to listeners, and which are not? Here non-native speakers have a clear advantage, since their native-speakers counterparts have seldom if ever experimented with adjustments in their pronunciation of English in order to be understood. 

The best teacher for monolingual groups, we conclude, is the fluent, bilingual speaker of either nationality, although the non-native teacher probably has the edge over the native-speaker colleague. The teacher worst situated, and by quite a significant distance, is the native speaker who is neither fluent in the learners’ mother tongue, nor fully understands the corresponding phonologies. 

Conclusion. By viewing English as a tool for international intelligibility, we establish a new perspective on pronunciation goals, with priorities that are both fewer in number and more realistic than those previously set. For monolingual groups the learner’s first language is a vital tool in achieving these new goals, and the bilingual non-native speaker teacher is an ideal instructor. 

Acknowledgements: My thanks to Dr Kevin Keyes of the Universidade Federal de Minas Gerais, Brazil, for his help with Brazilian Portuguese. 


D Brazil,Pronunciation for Advanced Learners of English, Cambridge Univ. Press, 1994.

D Brinton, Integrating pronunciation in the language syllabus, Speak Out! 16, IATEFL, 1995.

D Crystal,  The Cambridge Encyclopedia of the English Language. Cambridge Univ. Press, 1995.

B Jenner, Teaching pronunciation: the common core, Speak Out! 4, IATEFL, 1989.

J Jenkins  The Phonology of English as an International Language. Cambridge Univ. Press, 2000

J Kenworthy,  Teaching English Pronunciation, Longman, 1987.

R Macaulay,  RP RIP, Applied Linguistics Vol. 9 No. 2, 1988.

J D O’Connor, Better English Pronunciation, Cambridge University Press, 1967.

L Taylor, Pronunciation in Action, Prentice Hall, 1993.

J Wells,  Which pronunciation do you prefer? IAFTEFL Issues149, IATEFL, 1999.


Robin Walker has been teaching English in Spain for 20 years. He has been involved in teacher education since1983, and collaborates with the state teacher centres, Oviedo University and OUP España. He is a member of IATEFL and TESOL-Spain, and his specialist interests are pronunciation, ESP and learner independence.

Please feel free to leave a message in the guestbook at Karen's Linguistics Issues.

©Robin Walker2001. All rights reserved.