The problem with pronunciation teaching in most ESL classrooms is a misdirected target. Most teaching still operates on a native-speaker model: students listen to British or American English, identify deviations from it, and practise correcting those deviations. The goal, implicit or explicit, is to reduce the L1-influenced features of their speech and approximate native-speaker norms.
The research challenge to this approach has been building for 20 years. The Global Englishes movement, ELF research by Jennifer Jenkins and Barbara Seidlhofer, and intelligibility research by John Levis have collectively produced a significant body of evidence that changes the answer to the basic question: what are we actually trying to achieve with pronunciation instruction?
YapYapGo is a classroom speaking practice tool for ESL and EFL teachers that creates the varied L1-to-L1 interaction conditions in which intelligibility - not native-speaker approximation - is what actually matters. Here's what the research now supports about pronunciation teaching.The intelligibility principle
The shift in pronunciation research can be summarised in one sentence: the goal of pronunciation instruction should be intelligibility to a diverse range of listeners, not approximation to native-speaker norms.
This distinction has significant practical implications. A Vietnamese student with strongly Vietnamese-accented English may be perfectly intelligible to a German colleague, a Japanese client, and a Brazilian partner - the three people they actually need to communicate with professionally. If that's true, spending significant classroom time training them to approximate RP British vowels is misallocated effort.
Conversely, certain pronunciation features genuinely do impede intelligibility across diverse L1 backgrounds. Getting these wrong costs communicative effectiveness. The question is which features matter, not whether pronunciation teaching matters.
What the research says matters most for intelligibility
Jennifer Jenkins' Lingua Franca Core identified the pronunciation features most critical for intelligibility in ELF (English as a Lingua Franca) communication:
High priority (teach and correct):- Initial consonant clusters (preserve them - "stream" not "steam")
- Final consonants in key words (preserve them - final /t/ in "cat")
- Vowel length distinctions (sheep vs ship; feel vs fill)
- Nuclear stress placement (the most important word in an utterance must be stressed)
- Consonant sounds where substitution causes misunderstanding (particularly /v/ vs /b/, /l/ vs /r/ in contexts where it matters)
- Th-sounds (θ and ð) - frequently substituted by L2 speakers without causing intelligibility problems in most ELF contexts
- Weak forms and connected speech reduction - important for native speaker-like fluency but less critical for intelligibility
- Intonation patterns beyond basic sentence-level features
- Vowel quality to match specific regional accents (teaching to sound "British" or "American")
- Assimilation patterns of native speaker-like connected speech
- Features that mark speech as non-native but don't cause misunderstanding
The exam context complication
A practical complication: IELTS, Cambridge, and TOEFL assessments are still normed against standard varieties of English. IELTS Band Descriptors specifically mention "features of accent" and "phonological control." Students preparing for these exams need to understand that the exam criteria may not perfectly align with the intelligibility research.
The pragmatic teacher position: teach pronunciation primarily for intelligibility, but note where exam criteria differ and address those specifically for exam-focused students. The two goals often coincide; where they don't, be honest with students about why.
What this means practically
Shift pronunciation feedback to intelligibility-based
When correcting pronunciation, ask yourself: "Did this cause misunderstanding?" If yes, correct it. If no, consider whether the correction is worth the interruption.
A student who substitutes /b/ for /v/ in a context where no misunderstanding occurred doesn't need immediate correction. A student whose nuclear stress placement causes their partner to misunderstand the point of their utterance does.
Prioritise nuclear stress practice
Nuclear stress - the main stressed word in an utterance - is one of the most impactful features for intelligibility and one of the least taught. "I didn't STEAL the money" (I borrowed it) vs "I DIDN'T steal the money" (someone else did) are different meanings made by stress placement alone.
Minimal pair activities with nuclear stress across different sentence meanings are highly practical and directly relevant. "Which word would you stress if you wanted to emphasise X?"
Use varied English audio
Most ELT pronunciation material uses native speaker audio from UK or North American English. Students who only hear these varieties may struggle with the diverse L2-accented English they'll encounter in real ELF communication.
Incorporate audio from Indian English speakers, East African English, Singaporean English, and other established varieties. This builds the broader phonological flexibility that genuine intelligibility requires.
Tool tip: YapYapGo pairs students across different L1 backgrounds in multilingual classes, creating the genuine ELF conditions where intelligibility is tested in practice. Students discover first-hand which phonological features matter for mutual understanding with a specific partner. A classroom countdown timer keeps pronunciation drilling activities focused without them running over time.
Teach minimal pairs that matter
Classic minimal pair work (ship/sheep, cat/cut) remains valuable for the vowel distinctions that genuinely cause confusion. The question is which pairs to prioritise.
Prioritise minimal pairs where:
- The distinction is commonly confused by your students' L1 background
- Confusing the two causes actual misunderstanding in context
- The words appear frequently in the communicative contexts your students need
Skip or de-emphasise minimal pairs where:
- Native speakers would understand both from context
- The substitution is rare in ELF communication
- The time could go to more impactful features
Develop self-monitoring rather than accent correction
The long-term goal of pronunciation instruction should be students who can monitor whether they're being understood and adjust when they're not - not students who have learned to sound like a specific accent they'll never authentically produce.
Self-monitoring strategies: "Did my partner understood my main point?" "Do I need to repeat or rephrase?" "Was I emphasising the right words?" These are more transferable than any specific phonological adjustment.
For the broader Global Englishes context in which this research sits, see our post on how Global Englishes changes what we teach in speaking class. A random student picker and activity timer are useful when running minimal pair drilling - the timer keeps each contrast to a focused two-minute slot, and the picker ensures all students practise the target sounds specific pronunciation contrasts for class feedback.
Sources:
- Jenkins, J. (2000). The Phonology of English as an International Language. Oxford University Press. - The Lingua Franca Core: which phonological features matter for intelligibility.
- Levis, J. (2005). Changing Contexts and Shifting Paradigms in Pronunciation Teaching. TESOL Quarterly. - The intelligibility principle as the goal of pronunciation instruction.
- Derwing, T. & Munro, M. (2005). Second Language Accent and Pronunciation Teaching. TESOL Quarterly. - Intelligibility vs comprehensibility vs accentedness: what actually matters.
- Seidlhofer, B. (2011). Understanding English as a Lingua Franca. Oxford University Press. - ELF phonology and its implications for classroom instruction.