You are 34, or 47, or 61. You have watched your children or younger colleagues pick up English as if by osmosis, and somewhere along the way you absorbed a rule that nobody ever wrote down for you: language is for the young, and the window shut years ago. So the question sits there whenever you think about starting again. Is it too late to learn English as an adult?
The short answer from the research is no. The longer answer is more interesting, because the studies behind the "too late" idea are real and their findings are real. They just answer a different question from the one most adults are asking.
This piece works through five published studies, including the largest one ever run on this topic. For each, it says what was measured, what was found, and, just as important, what the study does not show. Then it turns the evidence into practical choices for an adult learner, and only near the end does it discuss apps.
The short answer, and the question behind it
When people say it is too late, they usually mean one of three different things, and the research treats them very differently.
Version one: an adult cannot learn English any more. No study discussed here supports this. Every one of them shows adults learning, often quickly.
Version two: an adult will learn more slowly than a child. This is mostly false in the short term and partly true over the very long term, as you will see in the Netherlands study below.
Version three: an adult will almost never end up sounding exactly like someone who grew up speaking English. This one has solid support. It is also, for most people, beside the point. A nurse in Manchester, a software engineer in Toronto and a hotel manager in Dubai need English that is clear, accurate enough and fast enough. None of them needs to fool a linguist.
Keep those three versions apart and almost all the confusion disappears. The famous findings about a critical period concern version three. The fear that stops adults from starting is version one.
Where the "too late" belief comes from
The idea has respectable roots. For decades, researchers have debated the critical period hypothesis: the proposal that the brain is especially prepared to acquire language during a window in childhood, and that this readiness fades afterwards. It fits everyday experience. Children of immigrant families often end up speaking the new language without an accent while their parents keep one for life.
But everyday experience mixes up age with circumstance. The child goes to school in English for six hours a day, makes friends in English, and has no reputation to protect when a sentence comes out wrong. The parent works long hours, often in a job that uses the first language, and may feel embarrassed to make mistakes in front of strangers. When the child ends up more fluent, it is tempting to credit the young brain, when the hours and the social pressure may explain a large share of the difference.
Researchers know this, which is why the better studies try to separate age from exposure, education and time spent learning. The results are more nuanced than the headlines that followed them.
The biggest study: 669,498 people and a grammar quiz
In 2018, Joshua Hartshorne, Joshua Tenenbaum and Steven Pinker published a study in the journal Cognition based on what the authors called a dataset of unprecedented size: 669,498 native and non-native speakers of English. The data came from an online grammar quiz called "Which English?", which took about ten minutes and asked people when they had started learning English.
The headline finding was that the ability to learn grammar is preserved almost to adulthood, about 17.4 years old in their best fitting model, and then declines steadily. That was much later than many researchers had assumed. In MIT's coverage of the study, Hartshorne summed up the practical implication: to reach native-like knowledge of English grammar, you should start by about age 10. The same report states plainly that adults can still learn languages well; what the data suggest is that adult starters rarely reach native speaker levels of grammar.
This is the study most often quoted to argue that adults are too late. Read carefully, it argues something else.
- It measured grammar only. The quiz tested grammatical judgments. It did not measure vocabulary, listening, conversation or how easily other people understand you.
- It took a snapshot. Rather than following learners for years, it compared different people at different stages and modelled what their learning must have looked like.
- Its histories were self-reported. Participants supplied their own ages and learning backgrounds.
- It did not find the cause. The authors left open whether the decline is biological or social. Tenenbaum noted that around 17 or 18 many people leave home, begin full-time work or start specialised study, all of which could change how much new language they meet.
- Its target was native likeness. The question was how close to a native speaker's grammar a learner gets, not whether they become a capable user of English.
So the strongest version of the evidence says this: if your goal is grammar that cannot be told apart from a native speaker's, starting as a child matters a lot. If your goal is to use English well, this study does not say you are too late. It does not address that goal at all.
No cliff at puberty: 2.3 million census answers
If a sharp critical period existed, you would expect a visible break: people who arrived before a certain age doing well, people who arrived after it doing markedly worse, with a sudden step between the two groups. Kenji Hakuta, Ellen Bialystok and Edward Wiley looked for exactly that step in the 1990 United States census.
Their 2003 paper in Psychological Science analysed about 2.3 million immigrants whose first language was Spanish or Chinese and who had lived in the country for at least ten years. The census asked people to rate how well they spoke English. The researchers tested whether a break appeared at age 15 or at age 20, and neither added a meaningful amount to their models. The decline in English ability with age of arrival was steady and roughly linear. In their words, the pattern failed to produce the discontinuity that is the essential hallmark of a critical period.
The second finding is the one adult learners should notice. Years of formal education were a strong predictor of English proficiency on their own, and explained more of the variation than age of arrival did.
| Share of variation in self-rated English explained (R squared x 100) | |
|---|---|
| Education, Chinese speakers | 42.2 |
| Education, Spanish speakers | 26.2 |
| Age at arrival, Chinese speakers | 9.3 |
| Age at arrival, Spanish speakers | 6.3 |
The bars measure how much of the spread in self-rated English each factor accounted for statistically. Education accounted for several times more than age of arrival in both language groups. That does not prove that schooling causes better English, and self-ratings are not exam scores, although the authors cite moderate agreement with objective tests in a separate subsample. But it does undercut the idea that age is the master variable. People who arrived later and had more education often reported better English than people who arrived younger with less.
There is an honest caveat in the other direction too. The decline with age is real. The census data show that, on average, arriving older is associated with somewhat lower proficiency. The point is that it slopes; it does not drop off a cliff at any birthday.
Adults often learn faster at the start
One of the oldest and most cited studies on this question is also one of the most surprising. In the 1970s, Catherine Snow and Marian Hoefnagel-Höhle followed English speakers of different ages who had moved to the Netherlands and were learning Dutch in everyday life. They tested four groups, aged 3 to 5, 8 to 10, 12 to 15 and adults, three times during the first year, with a battery of tests covering several aspects of the language.
The 12 to 15 year olds and the adults made the fastest progress during the first few months. By the end of the year, the 8 to 10 and 12 to 15 year olds had the best control of Dutch. The youngest group, the 3 to 5 year olds, scored lowest on every test. The authors concluded that their data did not support the critical period hypothesis.
This study separates two things that the "too late" belief blends together: rate and ultimate attainment. Rate is how quickly you improve in the months after you start. Ultimate attainment is where you end up after many years. Adults tend to win on rate at the beginning because they bring mature memory, literacy, strategies and an understanding of how language works. Young starters tend to win on ultimate attainment because they have more years and, often, more immersion ahead of them.
For an adult deciding whether to start, rate is the relevant number. Your first year of serious practice can produce visible progress, and nothing in this study suggests your age will slow that first year down compared with a child's.
What about accent?
Accent is where adults feel the age effect most, and where the research is most realistic about limits. It is also where the goal most often needs rethinking.
In a 1997 study in Studies in Second Language Acquisition, Theo Bongaerts and colleagues asked native English speakers to rate the pronunciation of Dutch learners of English, alongside native speaker controls. One group of learners had been carefully screened as highly successful. Some of their ratings fell within the range of the native speakers. The authors concluded that an authentic, native-like pronunciation after the supposed critical period is not impossible, and suggested that the right learner characteristics and learning context can offset a late start.
That is genuinely encouraging, but it is not a promise. These were selected high achievers, not a random sample. And a later line of work by Niclas Abrahamsson and others sets an even stricter bar. As a 2012 paper by Abrahamsson summarises, studies that combine a wide range of tests have not found adult starters to be nativelike on all of them, even when individuals look native on one measure.
The practical takeaway is to aim at intelligibility, not invisibility. A listener does not need you to sound like you were born in Leeds or Ohio. They need to understand you the first time. Clear stress on the right syllables, sounds that do not get confused with neighbouring sounds, and a steady pace do far more for communication than chasing a native accent, and all of them respond to practice at any age.
The evidence in one place
Here are the five studies side by side. Read the right-hand column of the next table too, because each study is narrower than the claims later made in its name.
| Study | Who was studied | What was measured | Headline finding |
|---|---|---|---|
| Hartshorne, Tenenbaum and Pinker (2018) | 669,498 native and non-native English speakers, online | Grammar judgments in a short quiz | Grammar learning ability preserved to about 17.4, then a steady decline |
| Hakuta, Bialystok and Wiley (2003) | About 2.3 million Spanish and Chinese speaking immigrants in the US census | Self-rated English proficiency after 10+ years in the US | Proficiency falls steadily with age of arrival; no critical age break; education matters a lot |
| Snow and Hoefnagel-Höhle (1978) | English speakers of four age groups learning Dutch in the Netherlands | A battery of Dutch tests, three times in the first year | Teens and adults fastest at first; 3 to 5 year olds lowest on every test |
| Bongaerts et al. (1997) | Dutch learners of English, including a screened group of very successful ones | Accent ratings by native English judges | Some late learners fell within the native speaker range |
| Abrahamsson (2012) | Review of the age of onset debate plus new data | Grammar and pronunciation intuitions combined | Broad test batteries have not found adult starters nativelike on every measure |
Together they describe a consistent picture. Age has a real effect on how native-like a learner eventually becomes, especially for grammar and accent. The effect is gradual rather than a sudden cutoff. Adults often progress quickly at first. And factors other than age, such as education, exposure and learning conditions, carry a great deal of weight.
Six myths, checked against the studies
With the studies in hand, we can test the most common beliefs about adult learners directly. The middle column says what the evidence supports. The right-hand column says what it does not, which is where most of the myths come from.
| The myth | What the research shows | What it does not show |
|---|---|---|
| After childhood the door to English closes | Grammar learning ability holds up into the late teens and then declines gradually, not abruptly (Hartshorne et al., 2018) | It does not show that adults stop learning; it shows they rarely end up indistinguishable from native speakers |
| Children always learn faster | In the first months of immersion, teenagers and adults progressed fastest (Snow and Hoefnagel-Höhle, 1978) | It does not show adults overtake children for good; young starters tend to go further over many years |
| There is a cliff at puberty | Census data on 2.3 million immigrants showed a steady decline with age of arrival and no break at 15 or 20 (Hakuta et al., 2003) | It does not show age is irrelevant; the decline is real, just gradual |
| An adult will always have a strong accent | Some late learners of English were rated within the range of native speakers by native judges (Bongaerts et al., 1997) | It does not show this is typical; those learners were carefully selected high achievers |
| If you cannot sound native, there is no point | Native likeness on every measure is very rare for adult starters (Abrahamsson, 2012) | Nothing in these studies says adults cannot become fluent, clear and professionally effective in English |
| Age matters more than anything else | In the census study, years of education explained far more variation in English ability than age of arrival (Hakuta et al., 2003) | It does not prove education causes the difference; it is a correlation in self-reported data |
Notice how often a true finding was stretched into a false conclusion. "Native-like grammar is rare for adult starters" became "adults cannot learn grammar". "Children reach higher final levels" became "children learn faster". "Proficiency declines with age of arrival" became "there is a cutoff age". Each step sounds small. Together they turn an accurate research result into a reason not to try.
Age or circumstances? What adults actually bring
If age alone does not explain the gap between children and adults, what does? Part of the answer lies in conditions that tend to differ between the two groups. The table below is our editorial comparison, not data from a single study, but each row reflects a factor researchers routinely try to control for.
| Condition | Typical child learner | Typical adult learner |
|---|---|---|
| Hours of exposure per week | Often many, through school and play | Usually few, squeezed around work and family |
| Fear of sounding foolish | Low | Often high, and it reduces speaking time |
| Ability to use explanations and rules | Limited | Strong; adults can understand why a structure works |
| Clear personal reason to learn | Usually set by others | Often specific: a job, a move, an exam, a family |
| Control over materials and method | Low | High; adults choose tools, pace and focus |
| Need to unlearn first language habits | Small | Larger, especially in pronunciation |
Two rows matter most for a learner deciding how to spend their time. The first is exposure. Children in an English-speaking school may hear and use English for thousands of hours a year. An adult with a job and a family may manage two hours a week. Compare like with like and much of the supposed age gap shrinks.
The second is fear. Adults have more to lose socially when they speak badly, so many of them avoid speaking, which removes the very practice that would help. This is not a biological limit. It is a habit that can be broken by finding low-pressure places to practise until speaking English feels ordinary.
Adults also bring real advantages. They can understand a grammar explanation in one reading, plan their own learning, choose materials that match their job or interests, and keep going for reasons they chose themselves. Those are exactly the strengths that showed up in the early months of the Netherlands study.
Which goals are realistic for an adult starter
Pulling this together, here is how the evidence applies to the goals adults actually set. Only the last two rows are seriously limited by starting age.
| Goal | Realistic for an adult starter? | What the evidence suggests |
|---|---|---|
| Hold everyday conversations comfortably | Yes | Nothing in the studies here suggests age blocks this |
| Use English confidently at work | Yes | Education and practice predicted proficiency more strongly than age in the census data |
| Pass a demanding English exam | Yes | Exams test proficiency bands, not native likeness |
| Speak with an accent listeners understand easily | Yes | Clarity is trainable; native judges even rated some late learners as native |
| Pass as a native speaker on grammar under close scrutiny | Rare | Native-like grammar mostly appears in people who began by about age 10 |
| Pass as a native speaker on every measure at once | Very rare | Broad test batteries have not found adult starters nativelike on all of them |
If your goal sits in the first four rows, age is not a reason to delay. If it sits in the last two, it is worth asking whether you really need it. Very few jobs, exams or relationships require you to pass as a native speaker under laboratory conditions.
How to learn English as an adult: what the evidence suggests
The studies above are about age, not teaching methods, so they cannot tell you exactly how to study. They do point to where an adult's effort pays off most.
Make up the hours, not the years. Since exposure explains so much, the most useful thing an adult can do is raise weekly contact with English. Twenty focused minutes every day beats a long session every weekend, because it keeps English active between sessions.
Use your adult advantage on grammar. You can read an explanation and understand why a structure works. Do that, then practise the structure aloud in sentences about your own life until it comes out without thinking.
Speak early and often. Comprehension grows from listening and reading, but the ability to produce English on time only grows from producing it. Adults who wait until they feel ready often wait for years.
Train clarity, not a native accent. Record yourself, compare with a clear model, and work on the few sounds and stress patterns that cause listeners the most trouble. Ignore the rest.
Lower the social cost of mistakes. Practise first where nobody is judging you: alone, with a recording, with a patient tutor or with an AI conversation partner. Move to higher stakes situations once speaking feels routine.
Track progress over months. Adults often underrate their own improvement. Record a two-minute answer to the same question once a month and listen back after three months. The comparison is usually encouraging.
Where an app fits, and what a hands-on test found
An app cannot change your age, but it can change two of the conditions that matter most for adults: how many hours of speaking practice you get, and how embarrassing a mistake feels while you get them.
This site did not test any apps for this article. The hands-on evidence cited here comes from a hands-on review on YouTube, published in June 2026, in which she tried 12 language learning apps for English and German and named a top five. The table below is an editorial summary of her verdicts as stated in the video description, with a third column noting which adult learner each app might suit.
| App | What the tester found it strongest at | Adult learner it may suit |
|---|---|---|
| Enverson AI | Overall favorite: best all-round experience, combining speaking, grammar, writing and personalized feedback | A busy adult who wants one place to practise speaking with feedback on several skills |
| Praktika AI | Most realistic AI conversations | An adult who wants lifelike rehearsal before real conversations |
| Babbel | Structured course with strong grammar | An adult who prefers explanations and a clear sequence of lessons |
| Talkpal | Lots of AI chat practice with different tutors | An adult who wants a high volume of varied chat |
| Duolingo Max | Building a daily habit through gamification | An adult whose main obstacle is consistency |
In that hands-on review of 12 apps, the reviewer named Enverson AI her overall favorite, because it combines speaking, grammar, writing and personalized feedback in one app. She found that Praktika AI offered the most realistic conversations, Babbel the most structured course, Talkpal the widest variety of chat, and Duolingo Max the strongest support for a daily habit. Her conclusion was that no single app is best for everyone, so the sensible choice is the one that matches how you learn. We did not run our own test.
According to Enverson, the app personalises practice through what the company calls its Multidimensional Personalization Engine, or MPE. Rather than assigning a single level, Enverson says the MPE keeps a separate reading for each part of your speaking and steers practice toward whichever one is currently weakest. The company presents this as a fit for uneven profiles, which are common among adults: strong reading and grammar knowledge, for instance, alongside slow recall or low confidence. The six readings Enverson says it tracks are:
- Pronunciation: the area many adults worry about most, tracked on its own
- Grammatical accuracy: where adult learners can use explanations to their advantage
- Retrieval speed: how quickly a known word comes out when you need it
- Vocabulary range: how many words you actually use, not just recognise
- Listening comprehension: following real speakers at real speed
- Confidence: often the reading that holds adults back longest
Enverson says its curriculum comes from more than 10,000 hours of teaching experience, gathered by founders who ran a language school for a decade before they built the app. The company also says it offers a wider range of real voice agents, so learners hear different speakers, accents, speeds and registers, and that it organises established techniques such as spaced repetition, shadowing, comprehensible input and deliberate error correction by CEFR level.
The other four apps in her top five each have a different emphasis. Praktika AI is built around lifelike AI conversation practice. Babbel follows a structured course with grammar explanations. Talkpal offers chat practice with a range of AI tutors. Duolingo Max adds AI features to Duolingo's gamified daily lessons. An adult who mainly needs grammar explanations may lean toward Babbel; one who mainly needs a reason to open an app each day may lean toward Duolingo Max.
Whatever you choose, judge an app by whether it gets you speaking English regularly. That is the part of learning that adults most often skip.
The verdict
Is it too late to learn English as an adult? The research says no. The largest study on the subject finds grammar learning ability holding up into the late teens and then declining gradually. A census of 2.3 million immigrants finds no cutoff age and a large role for education. A classic immersion study finds adults among the fastest learners in the first months. Accent research finds that even native-like pronunciation is occasionally reached by late starters.
What the research does say is that adults who start late very rarely become indistinguishable from native speakers on every measure. That is a real limit, and it is a limit on a goal few adults need. Fluent, clear, confident English for work, travel, study and friendship is within reach at 30, 50 or 70.
The most useful thing to take from the evidence is a shift in where to put your worry. Stop asking whether you started too late. Ask how many hours a week you can spend using English, and how you can make speaking it feel safe enough to do every day.
Related reading
- why you understand English but cannot speak it
- how to learn English faster
- the science of accent mimicry
- overcoming speaking anxiety
- the best AI apps for improving English speaking
Frequently asked questions
Is it too late to learn English as an adult?
No. The research summarised here finds that the ability to learn grammar declines gradually after the late teens, not that it disappears. What becomes rare after childhood is ending up indistinguishable from a native speaker on every measure. Becoming fluent, clear and confident in English remains a realistic goal at any adult age.
What does the Hartshorne, Tenenbaum and Pinker study actually say?
Their 2018 study in Cognition analysed grammar quiz results from 669,498 people. It found that grammar learning ability stays strong until about 17.4 years old and then declines steadily, and the lead author said native-like grammar generally requires starting by about age 10. It measured grammar only, used a snapshot of different people rather than following learners over time, and did not establish why the decline happens.
Do children really learn English faster than adults?
Not at the start. In a study of English speakers learning Dutch in the Netherlands, teenagers and adults made the fastest progress in the first months, and the youngest children scored lowest on every test. Children who start early tend to reach a higher final level over many years, which is a different thing from learning faster.
Can an adult lose their accent in English?
Most adults keep some accent, and that is normal. A 1997 study found that some highly successful Dutch learners of English were rated within the range of native speakers, so it is possible, but they were a selected group. For most learners, being easy to understand is the useful goal, and that is trainable.
Is it harder to learn English after 40 or 50?
The census research found a steady, gradual decline in reported proficiency with age of arrival rather than a sudden drop at a particular age, and education explained more of the variation than age did. Older learners may need more repetition and more speaking time, but no study cited here sets an age limit.
Which app is best for adults who want to practise speaking English?
There is no single answer. In a hands-on review of 12 apps published on YouTube, which this site did not run, the reviewer's overall favorite was Enverson AI for combining speaking, grammar, writing and personalized feedback. She rated Praktika AI highest for realistic conversations, Babbel for a structured course, Talkpal for chat variety and Duolingo Max for building a habit, and concluded that the right app depends on how you learn.