The English as a Lingua Franca Pronunciation Core
As a follow up to our previous post on the ELF Core, here is more information on which English pronunciation features are the most important:
1. Consonant sounds
• All English consonant sounds are necessary EXCEPT /θ/ and /ð/ (for which most substitutions are possible, such as /f/ and /v/, but probably not /ʃ/, /ʤ/ or /z/).
• ‘Dark /l/’ (also written as [ɫ]) is not necessary. Speakers can substitute ‘clear /l/’ (possibly preceded by a schwa if the /l/ is syllabic, like at the end of ‘bottle’). Substituting /ʊ/ for /l/ at the ends of words might also be acceptable, but more research is needed to confirm this.
• /r/ should be pronounced as in General American pronunciation (technically called a “rhotic retroflex approximant” and written as [ɻ]. It should also be pronounced everywhere it occurs in spelling, as in American English.
• /t/ needs to be carefully pronounced between vowels (e.g. ‘Italy’) and in clusters in the middle of words (e.g. ‘winter’). It should not be ‘flapped’ (as in General American pronunciation, ‘Italy’ might sound like ‘Idaly’ or ‘latter’ might sound like ‘ladder’); and it should not be replaced with a glottal stop (like in Cockney ‘better’).
• The consonants /p/, /t/ and /k/ must be aspirated when occurring in initial position in a stressed syllable (e.g. the first /p/ in ‘paper’).
2. Consonant clusters
• Clusters of consonants at the beginning of words must not be simplified (e.g. learners mustn’t drop the /r/ at the start of ‘product’).
• Clusters of consonants in the middle or at the end of words are a bit more complicated. They can be simplified if it makes articulation easier, but only according to rules of elision (i.e. dropping sounds) that also apply to native English varieties (especially in clusters containing /t/ and /d/, like ‘postman’).
• If learners have trouble producing consonant clusters, it’s usually OK to insert a very short schwa vowel between consonants, providing they don’t then stress this syllable (e.g. ‘product’ could be pronounced more like [pә’rɒdʌkәtә] by Japanese speakers without damaging intelligibility).
• Similarly, learners can add a short sc
hwa at the end of a word ending with a consonant, provided this does not create another word which it might be confused with (e.g. ‘hard’ sounding like ‘harder’).
• Appropriate consonant cluster simplification means that adding a sound is better than deleting a sound. For example, if you pronounce ‘helped’ with two syllables instead of one by inserting a vowel sound between the /p/ and /t/ cluster, Jenkins’ data suggests you’re still likely to be understood in an ELF context. But if you miss out the /p/, for example, then ELF intelligibility is much more at risk.
• Length contrasts must be preserved, e.g. ‘pill’ versus ‘peel. However, the actual quality of vowels is less important, providing it’s consistent (e.g. don’t keep switching between different pronunciations of the vowel in ‘hat’ so sometimes it sounds like RP [hæt] and sometimes it sounds like New Zealand [het]).
• The length of diphthongs must be preserved but, again, the actual quality of the vowels is less important, providing it’s consistent.
• When a vowel occurs before an unvoiced consonant, it should sound slightly shorter than when it occurs before a voiced consonant. For example, the vowel in ‘right’ is slightly shorter than the vowel in ‘ride’, and the vowel in ‘kit’ is slightly shorter than the vowel in ‘kid’.
• The /ɜː/ vowel, as in ‘girl’ or ‘first’, must be pronounced accurately.
4. Word groups and nuclear stress
• The stream of speech should be divided into meaningful tone units (also known as ‘tone groups’, ‘word groups’ or ‘thought groups’).
• Nuclear stress (i.e. which word is stressed within a ‘tone group’) must placed appropriately, especially for contrast/emphasis. This means the difference in meaning should be clear between, for example, ‘Let’s meet NEXT Saturday’ and ‘Let’s meet next SATURDAY’.