Open the app

The critical period: It gets more difficult with age (1959)

The critical period: It gets more difficult with age (1959)

The critical period hypothesis (CPH) is the claim that there is a limited window early in life during which a human brain acquires language naturally and effortlessly, and after which acquisition becomes slower, harder and — for some aspects of language — never quite complete. It is the idea behind the everyday intuition that children “soak up” languages while adults struggle, and behind the folk verdict that after a certain age you have simply missed the boat. The scientific picture is more interesting and more encouraging than that verdict: the window is real, but it is gradual rather than a cliff-edge, it closes much later than the popular version suggests, and it affects a native-sounding accent far more than it affects grammar or vocabulary. Most researchers now prefer to call it a sensitive period rather than a critical one, precisely to capture that it is a slope, not a slammed door.

Penfield, Lenneberg and the origin of the hypothesis

The idea grew out of neuroscience, not language teaching. In Speech and Brain Mechanisms (1959), the neurosurgeon Wilder Penfield and his colleague Lamar Roberts argued from clinical observation that the young brain has a plasticity the adult brain loses: children who suffered damage to the language areas often recovered speech by re-organising it into other parts of the brain, whereas adults with comparable injuries usually did not. Penfield concluded that there is an optimal biological period for acquiring language and used it to argue for starting foreign languages early in school.

The linguist Eric Lenneberg turned this clinical hunch into a full theory in Biological Foundations of Language (1967). He tied the window to lateralisation — the process by which language functions settle into one hemisphere (usually the left) — and proposed that the period ran from about age two to puberty, after which, he argued, the brain’s loss of plasticity made truly native-like acquisition impossible. The analogy that made the idea intuitive came from biology: just as a songbird deprived of adult song during a set early window never sings normally, and a kitten deprived of light during a critical phase of visual development never sees normally, a human deprived of language in the early years might never fully acquire it. The specific cut-off Lenneberg proposed — puberty — has not held up, but the shape of his claim, a biologically-timed window of heightened readiness, set the research agenda for the next half-century.

The evidence: deprived children and immigrant learners

Two very different bodies of evidence gave the hypothesis its empirical teeth.

Children deprived of early language. The tragic natural experiments are children who grew up without linguistic input. The most studied is Genie, discovered in Los Angeles in 1970 at the age of thirteen after a childhood of extreme isolation and abuse, during which she was kept strapped in a room and punished for making sounds. Once cared for, Genie learned a large vocabulary and clearly wanted to communicate — but she never mastered grammar. She could accumulate words yet struggled to combine them into structured sentences, producing strings like “applesauce buy store” rather than syntactically organised speech. Her case suggested that the lexicon (learning words) and the grammar (combining them) come apart: vocabulary may remain learnable throughout life, while the machinery for syntax appears to depend on early exposure. The evidence is real but must be read cautiously — Genie’s deprivation was so severe, and involved so much more than missing language, that no single conclusion can be cleanly drawn from one person.

Immigrant learners. Far stronger evidence came from healthy people learning a second language at different ages. In a landmark 1989 study, Jacqueline Johnson and Elissa Newport tested Chinese and Korean immigrants to the United States on their intuitions about English grammar, all of whom had lived in the country for years. The finding was clear and orderly: those who had arrived before about age seven performed like native speakers; from roughly seven to fifteen, scores declined steadily with age of arrival; and among those who arrived as adults, performance was lower and no longer tracked age in the same tidy way. This was the classic result for a sensitive period — not an abrupt switch that flips at puberty, but a steady, age-graded decline in the ultimate command of grammar.

The modern revision: a later, gentler window

The strong version of the hypothesis — a hard biological deadline at puberty, after which native-like learning is impossible — has not survived closer scrutiny. Two shifts define the modern picture.

The first is the move from “critical” to “sensitive” period. A critical period, in the biological sense Lenneberg borrowed, implies a sharp onset and offset and permanent consequences for what is missed. What the language data actually show is a gradual decline in attainment with age of first exposure — heightened sensitivity early on that fades slowly, not a gate that shuts. Calling it a sensitive period keeps the real phenomenon (younger is, on average, better for ultimate attainment) without the false implication of an all-or-nothing deadline.

The second is a dramatic sharpening of the timing, thanks to data at a scale earlier researchers could only dream of. In 2018 Joshua Hartshorne, Joshua Tenenbaum and Steven Pinker analysed a grammar quiz taken online by about 2/3 million English speakers (669,498 people) — orders of magnitude more than any previous study — and used it to disentangle three things that are usually confounded: a person’s current age, their age when they first encountered English, and how long they had been using it. Their conclusion was that the ability to learn grammar to a high standard stays remarkably stable through childhood and adolescence and begins to fall off only around age 17 or 18 — much later than puberty. But there is a catch that keeps the popular intuition partly alive: to reach fully native-like grammar you need to start young (roughly before age ten), not because learning ability collapses at ten, but because becoming native-like takes many years and the learning window, though open late, is not infinite. The precise interpretation of that 2018 result is still debated — a later reanalysis argued the data fit a steady decline from early childhood rather than a single sharp turn at 17.4 — but the headline correction stands: the window for grammar closes far later, and far more gently, than Lenneberg supposed.

The other key refinement is that “language” is not one thing that ages as a block. Age affects its components very differently:

  • Accent (pronunciation) is the most age-sensitive of all. A native-sounding accent is the hardest thing to acquire late; measurable traces of a foreign accent typically appear even in people who started in later childhood, and are almost universal in those who start as adults.
  • Grammar (morphology and syntax) is age-sensitive but far more forgiving than accent — this is the domain the sensitive period is really about, and where late starters can still reach a very high level, just less reliably native-like.
  • Vocabulary and meaning show essentially no critical period. People go on learning words efficiently throughout life; a fifty-year-old can expand their vocabulary as readily as a teenager, and often faster, thanks to a larger conceptual scaffold to hang new words on.

Adults versus children: what the research actually shows

Here the evidence contains a genuine surprise that cuts against the folk wisdom. When you measure the rate of early learning rather than the final ceiling, older learners start faster, not slower. In a much-cited 1978 study, Catherine Snow and Marian Hoefnagel-Höhle followed English speakers of different ages learning Dutch naturally after moving to the Netherlands. Over the first months, the adolescents and adults made the fastest progress; the three-to-five-year-olds were the slowest of all. Only over a longer horizon did the younger children catch up and, in some respects, pull ahead. Older brains bring more to the table at the start — a mature memory, existing literacy, explicit strategies, and a first language to reason from.

So the honest summary is a split verdict. Children are not faster learners in the short run; what they have is a higher final ceiling, especially for accent, and the patience of years of immersion in which to reach it. Adults learn faster at first and can attain excellent grammar and unlimited vocabulary, but rarely shed every trace of a foreign accent. The popular belief that adults “can’t” learn languages confuses two different things — the near-impossibility of a flawless native accent with the very real possibility of full, fluent, high-level command of a language. The first is genuinely hard after childhood; the second is available at any age.

What this means for learning a language

The critical period is often quoted as an excuse — I’m too old, I’ve missed my window — and that is the one conclusion the science does not support. Three points matter for an adult learner. First, the window that closes is mostly the window for a perfect accent; the window for grammar closes late and gently, and the window for vocabulary never closes at all. Aiming for effortless communication rather than for passing as a native removes the only target that age truly puts out of reach. Second, adults start faster, not slower — the early weeks of a new language are, if anything, an adult’s advantage — so the discouragement most people feel is misplaced. Third, the ceiling that children reach comes from years of massive, meaningful exposure, and that is the ingredient an adult has to supply deliberately rather than getting it for free from a childhood surrounded by the language.

That last point is where method does the work that age no longer does automatically. Because grammar is internalised from encountering and producing whole sentences in context — the same input on which a child’s language faculty feeds, whether one accepts the nativist account of an innate device or a statistical-learning one — an adult’s best move is to reproduce that condition on purpose: lots of meaningful sentences, met and re-met until they stick, rather than rules memorised in the abstract. The neural specialisation the sensitive period reflects is anchored in the brain’s dedicated language regions described by Broca and Wernicke, and while those regions are most plastic early, they never stop learning. Structured, repeated exposure to full sentences with active recall — the principle behind sentence-based learning — is exactly how an adult supplies deliberately what a child receives by immersion.

FAQ

What is the critical period hypothesis in one sentence?

It is the claim that there is an early window in life during which the brain acquires language especially easily and completely, after which acquisition becomes harder and, for some aspects such as accent, rarely reaches a native level. Most researchers now call it a “sensitive” period, because the decline with age is gradual rather than a sudden deadline.

Am I too old to learn a new language?

No. The only thing age reliably puts out of reach is a flawless native accent. Grammar can still be learned to a very high level well into adulthood — the large 2018 study by Hartshorne, Tenenbaum and Pinker found grammar-learning ability holds up until around age 17–18 and then declines only slowly — and vocabulary shows no critical period at all. Studies even find that adults and adolescents learn faster than young children in the first months; children mostly win on the final accent, given years of immersion.

Why do adults keep a foreign accent even when their grammar is perfect?

Because pronunciation is the most age-sensitive part of language. The fine motor control and perceptual tuning behind a native accent are laid down very early, and after childhood the brain finds it hard to hear and reproduce every distinction of a new sound system. This is why someone can command a language’s grammar and vocabulary completely and still be identifiable as a non-native speaker — the two abilities have different developmental timetables.

Sources

Related