Most of a foreign accent starts in the ear: learners hear the sounds of a new language through the sound filter of their mother tongue, so sounds close to Vietnamese are heard as Vietnamese sounds, and the mouth produces exactly what the ear heard. The longer the mother tongue has been in use, the firmer this filter becomes. That is why late starters often keep their accent however long they study, and why training the ear to tell sounds apart often does more for pronunciation than drilling the mouth alone.
The good news from research is that a foreign accent and intelligibility are two different things: someone with a clear accent can still be understood perfectly. The realistic goal for a child aged 8–17 is therefore not to erase the accent but to fix exactly those sounds that make listeners misunderstand. This article explains the mechanism and the key studies, and applies them to the particular difficulties Vietnamese learners face with English and Chinese.
The first year of life: the ear specialises in the mother tongue
In 1984, Janet Werker and Richard Tees published a much-cited study: babies growing up in English-speaking families could tell apart pairs of sounds from Hindi and from Salish (an indigenous language of Canada) that English-speaking adults could not. This ability declined clearly between six and twelve months of age. The phenomenon is called perceptual narrowing.
Patricia Kuhl described a complementary mechanism, the perceptual magnet effect: the typical, prototype sound of a category in the mother tongue pulls nearby sounds towards itself in the listener's perception. Two physically different sounds that both fall within the pull of one mother-tongue sound are heard as one. A familiar example is Japanese speakers' difficulty in telling English /r/ from /l/, because both are pulled towards a single Japanese sound.
Perceptual narrowing is not a loss of ability. It is an optimisation: the child's ear becomes very sensitive to the differences that matter in the mother tongue and ignores those that do not. For anyone learning a foreign language at an older age, this means relearning how to hear differences that the mother tongue has taught the ear to ignore.
Flege's Speech Learning Model: new sounds, similar sounds and age
Similar sounds are harder than new ones. James Flege, in the Speech Learning Model he set out in 1995 and revised with Ocke-Schwen Bohn in 2021, argued that the more a sound in the new language resembles a sound in the mother tongue, the harder it is to learn correctly, because the learner files it under the old category. A completely new sound, with nothing similar in the mother tongue, is more likely to be learned as a category of its own.
Age works through how settled the mother-tongue sound system is. According to Flege, there is no biological window that closes at a particular age; adults can still form new sound categories. But the more stable the mother-tongue system, the more strongly it pulls new sounds towards it. He stressed that the quantity and quality of input from native speakers, meaning years of hearing many real voices, is an underestimated factor.
Accent outcomes show no sharp break. In a 1995 study by Flege, Murray Munro and Ian MacKay of Italian immigrants in Canada, the degree of foreign accent perceived by native listeners rose gradually with age of arrival, along a fairly continuous line rather than breaking at puberty. Even people who arrived quite early could keep a slight accent.
A complementary model. Catherine Best's Perceptual Assimilation Model predicts how hard each pair of sounds will be according to how they are "assigned" to mother-tongue sounds: two new sounds assigned to the same old sound are the hardest, two new sounds assigned to two different old sounds are the easiest. This model helps teachers predict which pairs of sounds Vietnamese learners will struggle with.
Common difficulties for Vietnamese speakers
The table below brings together the difficulties that teachers and studies of Vietnamese learners mention most often. How much each one matters varies by region and by person; this is a list of things to watch for, not a diagnosis.
| Language | Difficulty | Why, in terms of perception | Effect on intelligibility |
|---|---|---|---|
| English | Final consonants and final consonant clusters, as in "desk", "asked" | Vietnamese syllables have few kinds of final consonant and no final clusters, so the ear skips them | High: losing a final consonant can change the word and erase markers of tense and plural |
| English | Word stress and sentence rhythm | In Vietnamese each syllable carries its own tone and the rhythm runs evenly syllable by syllable | High: native speakers rely heavily on stress to recognise words |
| English | Long and short vowels, as in "ship" and "sheep" | Two close sounds are easily assigned to a single Vietnamese vowel | Medium; context often helps the listener |
| English | The "th" sounds and some pairs of initial consonants | Absent in Vietnamese or different from dialect to dialect | Low to medium |
| Chinese | The four tones and the neutral tone | Vietnamese speakers are used to hearing tones, but Mandarin tones are easily assigned to the nearest Vietnamese tone | High: a wrong tone changes the word |
| Chinese | Telling initials such as zh, ch, sh from z, c, s | Many of these pairs have no matching contrast in Vietnamese | Medium, depending on the listener's dialect |
The last column is the one that matters most when deciding what to practise first. Studies of intelligibility show that stress, rhythm and final consonants usually affect understanding more than the "th" sound, even though "th" is the sound learners notice most.
Accent, intelligibility and comprehensibility: three different things
In 1995, Murray Munro and Tracey Derwing published a study that changed how the field looks at pronunciation. They had native speakers listen to learners and measured three things separately: accentedness (how different the speech sounds from a native speaker's), intelligibility (how much of what was said the listener could write down correctly) and comprehensibility (how much effort the listener felt it took to understand). The result: the three are related but not the same. Some speakers with a very noticeable accent were still understood almost completely.
Building on this, John Levis in 2005 distinguished two principles of pronunciation teaching: the nativeness principle, aiming to sound like a native speaker, and the intelligibility principle, aiming to be clearly understood. Most pronunciation researchers today lean towards the second, because it is more realistic for people who start after early childhood and because an accent is not an error. In their 2015 book Pronunciation Fundamentals, Derwing and Munro set out how to prioritise the pronunciation features that affect intelligibility.
For parents, the practical point is this: do not judge your child's progress by whether they "sound like a native speaker". Ask instead: do native speakers understand your child the first time, do they have to ask again, does listening take effort? A student who speaks clearly, with correct stress and all the final consonants, is in a good position even if their accent is still recognisably Vietnamese.
High-variability listening training: a method that has been tested
One technique with a good deal of evidence behind it is listening discrimination training with many different voices (high variability phonetic training). Studies in the early 1990s by Scott Lively, John Logan, David Pisoni and colleagues had Japanese speakers listen to /r/ and /l/ pairs spoken by many people, in many words, with right-or-wrong feedback, and found that their ability to tell the sounds apart improved and held up after several months; a 1997 study by Ann Bradlow and colleagues showed that this improvement carried over into pronunciation as well. Applied step by step:
- Pick one specific pair of sounds your child often confuses, for example "ship" and "sheep", or two Chinese tones they mix up.
- Listen to each word spoken by many different people, male and female, old and young, from many regions, not just one voice.
- Each time, your child chooses which word it was and is told straight away whether they were right.
- Practise briefly and often over several weeks rather than in one long session.
- Only once the ear can tell the sounds apart, practise saying the pair and record it to listen back.
The principle of many voices is important: listening to a single voice teaches the ear the features of that speaker rather than the sound category. This is also one of the advantages of an immersion environment, where your child hears dozens of different voices every day.
Kuhl and the role of real people in learning sounds
A 2003 study by Kuhl and colleagues of 9-month-old babies learning Mandarin sounds found that learning from a live person was far more effective than learning from video. In 2007, Kuhl proposed the "social gating" hypothesis: social interaction opens the gate to sound learning in young children, possibly through joint attention, real-time feedback and the motivation to communicate. The hypothesis is about young children and does not transfer directly to teenagers.
Even so, a similar idea appears in research on older learners: pronunciation improves more clearly when learners need to be understood, have a listener who reacts and receive feedback. A student talking to an international friend, seeing that the friend does not understand a word and having to find another way to say it, is getting the kind of natural feedback that is hard to create when practising alone.
Devices and apps can support high-variability listening practice and self-recording. But research suggests that the step from knowing to using, that is, saying things correctly in the middle of a conversation, needs a real listener. That is where a period living in the language environment, together with teachers who can point out the pronunciation errors that affect intelligibility, adds to practice at home.
Rhythm and weak forms: why Vietnamese speakers' English often sounds "flat"
From the mid-20th century, phoneticians such as Kenneth Pike and David Abercrombie divided languages into two rhythm types: stress-timed, like English, where stressed syllables come at fairly regular intervals and unstressed syllables are shortened, and syllable-timed, like Spanish, where syllables are roughly equal in length. Later measurements, such as those by Franck Ramus and colleagues in 1999 or by Esther Grabe and Ee Ling Low in 2002, showed the division is not that tidy: languages sit on a continuum, and rhythm is tied to syllable structure and vowel reduction.
Although the classification is still debated, one feature of English is clear: vowels in unstressed syllables are often reduced, frequently to a short "uh" sound, and function words such as "to", "for", "and" and "of" have weak forms in ordinary speech. In Vietnamese, every syllable keeps its tone and its full vowel. Vietnamese speakers therefore tend to pronounce every English syllable with the same length and clarity, which makes sentences sound flat and the stress hard to pick out.
This is a place where practice pays off for intelligibility: practise hearing and using weak forms, shortening unstressed syllables and making the stress on the key words in a sentence stand out. It also helps listening comprehension, because fast native speech is largely made up of weak forms.
Frequently asked questions
Can a 14-year-old still fix their accent?
Yes, to the extent of fixing the features that reduce intelligibility, such as final consonants, stress, and long and short vowels. Reaching an accent indistinguishable from a native speaker's is rarer at this age, but that is not a necessary goal. Guided listening discrimination training gets results at every age.
Should my child learn a British or an American accent?
Research on intelligibility shows that what matters is speaking clearly and consistently, not which accent is chosen. Hearing many different accents helps your child understand speakers from many places. When speaking, your child will naturally lean towards the accent they hear most, and that does not affect how well they are understood.
Do Vietnamese speakers have an advantage with Chinese tones?
Partly: speakers of tonal languages are already used to paying attention to pitch to tell meanings apart. But because Vietnamese and Mandarin tones are different, Vietnamese speakers easily map a Mandarin tone onto the nearest Vietnamese tone. This is exactly the kind of similar sound that Flege describes as difficult.
Should my child imitate the voice of a favourite actor or singer?
Imitation helps your child notice rhythm and intonation, and it is motivating. But listening to only one voice teaches the ear that person's features. Combine it with listening to many different speakers so the ear picks up the general sound category, not just one person's way of speaking.
Why does my child understand well but still not pronounce clearly?
Understanding does not mean being able to tell every sound apart: context helps your child guess the right word even without hearing the sound clearly. When speaking, there is no context to make up for it, so the sounds they cannot yet distinguish show through. Listening discrimination practice on specific pairs of sounds usually helps here.
Are native speakers bothered by a foreign accent?
Research on comprehensibility shows that listeners have to work harder when an accent is very strong or when they have little contact with speakers of that accent. But an accent on its own does not make listeners think less of the speaker if the speech is clear. Most native speakers in multicultural countries are used to hearing many accents.