Is there a critical period for learning a language?

There is strong evidence that earlier is better, particularly for accent and grammar, but the popular version — that a window slams shut at a fixed age — overstates what the research actually shows.

The hypothesis. Proposed by Eric Lenneberg, the idea is that there is a biologically determined period during which language acquisition happens naturally and after which it becomes effortful and incomplete.

The strongest evidence:

Cases of extreme deprivation. Children deprived of any language input until late childhood have never achieved normal grammatical competence, despite intensive later teaching. These cases are rare, ethically harrowing and methodologically compromised — the children suffered severe abuse alongside language deprivation — so they support the hypothesis without proving it cleanly.

Deaf children acquiring sign language. This is the more informative evidence, because it isolates age of first exposure without the confound of abuse. Those exposed to a signed language from birth achieve full native competence; those first exposed later show systematic, lasting differences in grammatical processing.

Second language studies. Large-scale analyses consistently find age of acquisition predicts ultimate attainment, with a decline that is gradual rather than a cliff.

What the evidence does not support:

A sharp cut-off. The decline appears continuous, not stepped, which argues against a discrete window.

Uniformity across domains. Accent is affected most and earliest; grammar somewhat later; vocabulary hardly at all — adults learn vocabulary faster than children, which is why adult learners frequently progress more quickly at first.

Impossibility. Some adult learners achieve near-native ability, which a strict critical period should preclude.

The competing explanations, which are genuinely plausible: adults have less input and less immersion; they have stronger existing L1 interference; they are less willing to make errors and be corrected; and social identity affects how much a new accent is adopted.

The practical implication. Children learn accents better; adults learn faster initially and benefit from explicit instruction. Neither observation supports giving up as an adult, and adult learners routinely reach full functional competence.