The Critical Period Hypothesis (CPH) claims that there is a maturationally constrained window during which language can be acquired with native-like outcomes; outside this window, acquisition is slower, less complete, and rarely reaches native attainment. The claim originated in first-language research and was extended, more controversially, to second-language acquisition. Current consensus has shifted from a hard "critical" period to a softer "sensitive" period of gradient decline.
Eric Lenneberg's Biological Foundations of Language (1967) proposed the hypothesis for first-language acquisition. Drawing on Penfield and Roberts, Lenneberg argued that language acquisition was tied to brain maturation, specifically to the lateralisation of language functions to the left hemisphere, which he believed completed around puberty. Before puberty, the brain retained the plasticity needed to acquire language naturalistically; after puberty, lateralisation was complete and acquisition became laborious and incomplete. Evidence came from feral and deprived children, late-signing individuals, and aphasia-recovery patterns showing better outcomes in children than adults.
Lenneberg's neurological argument has not aged well. Imaging research has since shown that lateralisation is largely complete much earlier than puberty, undermining his proposed mechanism. The behavioural pattern, however, has held up: extreme cases of language deprivation past childhood confirm that first-language acquisition is severely compromised when delayed. The best-documented case is Genie, a girl isolated until age 13 who, after rescue, never fully acquired the syntax of her first language (Curtiss 1977).
The hypothesis comes in two strengths. The strong version posits a hard cutoff around puberty, after which full native-like acquisition of any language feature is impossible. This version is largely rejected: too many post-puberty learners reach very high proficiency for an absolute cliff edge to hold. The weak version, usually called a sensitive period, posits a gradual decline in learning capacity rather than an abrupt loss, and it is what most contemporary SLA researchers accept. A true critical period in biology, like imprinting in geese, is a sharp window outside which the capacity is lost entirely; the language evidence fits the softer, graded picture better.
Crucially, different domains appear to have different sensitive periods. Phonology is the most age-sensitive: native-like accent is rarely achieved past early adolescence, and Long (1990) argued the phonological window may close as early as age six. Morphosyntax shows pronounced age effects but with more individual variation and higher achievable ceilings. Vocabulary and pragmatics show the weakest age effects; lexical growth continues into adulthood with no maturational ceiling, and adults often out-learn children on vocabulary in the short term.
Whether a critical period applies to second-language acquisition is the more contested question. Johnson and Newport (1989) tested 46 Korean and Chinese immigrants to the United States on grammaticality judgments in English. Among those whose age of arrival fell between 3 and 15, they found a strong negative correlation between age and ultimate attainment; for the 17-to-39 group they found no significant correlation but uniformly lower scores. This pattern, age effects within a window followed by a flat decline, became the canonical empirical signature of the CPH for L2.
DeKeyser (2000) replicated the design with 57 Hungarian-speaking adults and reached a stronger conclusion: among adult arrivals, only those with high verbal analytical ability scored within the child-arrival range. He argued for a Fundamental Difference Hypothesis: children acquire language through implicit, domain-specific mechanisms, while adults must rely on explicit, problem-solving cognition, and only those with high analytical aptitude can compensate. Against this, Birdsong (2006) and others have catalogued adult learners who reach native-like attainment by every measure. Sceptics argue that even a few such cases refute a hard cutoff; defenders respond that they are rare exceptions explained by individual differences in aptitude, motivation, or input quality.
The largest study to date used a viral grammar quiz to collect data from 669,498 native and non-native English speakers. Hartshorne, Tenenbaum and Pinker (2018) disentangled three usually-confounded variables: current age, age of first exposure, and years of experience. They concluded that grammar-learning ability is preserved nearly intact until about 17.4 years of age, then declines steadily, much later than Lenneberg's puberty cutoff. They argued that the apparent puberty offset in earlier studies was an artefact of insufficient data. Van der Slik and colleagues (2022) reanalysed the same dataset and argued it is equally consistent with a steady decline from early childhood, with no sharply defined offset at all.
The hypothesis sits naturally with nativist accounts and the notion of a Language Acquisition Device: if humans have innate, language-specific hardware, it is plausible that it runs on a biological timetable. It challenges purely environmental or general-cognitive accounts that predict no fundamental age constraint. The fossilisation of certain features in adult learners may partly reflect sensitive-period effects, though it is also shaped by communicative sufficiency, reduced motivation, first-language transfer, and input quality rather than maturation alone.
The hypothesis has been used to argue both for very early English instruction (start before the window closes) and against the assumption that earlier is always better. The empirical record is mixed. In immersion settings, earlier starters do tend toward more native-like outcomes, especially in pronunciation. In typical foreign-language classrooms with limited weekly hours, however, earlier starters do not consistently outperform older ones, because the young learners' usual advantage comes largely from sheer exposure, and older learners compensate with cognitive maturity, study skills, literacy, and explicit grammar knowledge.
The practical takeaways: pronunciation work warrants serious attention with young learners; adult learners should not be told their case is hopeless, since high competence remains attainable even where native-like phonology is unlikely; and the goal of approximating an idealised native speaker is itself questionable in a world where English functions as a lingua franca and intelligibility, not native-likeness, is what matters.