The critical period hypothesis, what to actually tell parents who ask if it is too late.

Eric Lenneberg proposed the critical period in 1967. Joshua Hartshorne, Joshua Tenenbaum, and Steven Pinker tested it on two-thirds of a million speakers in 2018. The conclusions you can defend at a parent-teacher conference are not what the popular version suggests.

What Lenneberg originally claimed

Eric Lenneberg's 1967 book (Biological Foundations of Language, Wiley) proposed that language acquisition is biologically constrained to a developmental window that closes around puberty. Exposure before the window closes produces native-like competence. Exposure after the window closes produces a different, more effortful, less complete acquisition path. The mechanism, Lenneberg argued, was neurological: brain plasticity for language drops sharply with maturation.

The evidence Lenneberg cited was thin by modern standards but the proposal was reasonable. The clinical literature on feral children, aphasia recovery, and deaf children without early sign-language exposure all pointed toward some kind of developmental constraint. The claim was that L2 acquisition fell under the same constraint as L1.

The Johnson and Newport study

Jacqueline Johnson and Elissa Newport ran the study that defined the modern debate (Critical Period Effects in Second Language Learning, Cognitive Psychology, 1989). Forty-six Korean and Chinese speakers who had immigrated to the US at varying ages took a grammaticality judgment test in English. The earlier the age of arrival, the higher the score, with a steep drop after roughly age seven and a plateau in adulthood.

Johnson and Newport's curve became the canonical evidence for the critical period in L2. It was used to argue that adult language learners cannot reach native attainment, that the field should expect a ceiling, and that early start matters more than any other variable.

The Birdsong counter-evidence

David Birdsong (Ultimate Attainment in Second Language Acquisition, Language, 1992) ran the counter-study. Birdsong tested adult English speakers who had learned French as adults and lived in France for decades. A meaningful fraction (around fifteen percent in his original study) reached scores indistinguishable from native French speakers on grammaticality judgment tests.

The Johnson and Newport result said adults cannot reach native attainment. The Birdsong result said some adults can. Both could be true if the curve is a probability gradient rather than a hard wall. Birdsong's argument, expanded in his 2018 review work, is that the critical period is real in the sense that average attainment declines with age of exposure, and not real in the sense that adult acquisition is biologically prohibited.

The Hartshorne, Tenenbaum, and Pinker study

The 2018 paper (A Critical Period for Second Language Acquisition: Evidence from Two-Thirds of a Million English Speakers, Cognition) ran the most ambitious test of the hypothesis the field has seen. The researchers built a viral grammar quiz that collected demographic information from 669,498 respondents, including their L1, their age, and their age of first English exposure.

The result was a more precise version of Lenneberg's original claim. The grammar-learning rate stayed high until roughly age 17.5. After that age, the rate of grammar acquisition dropped sharply. Earlier proposals had located the cutoff at puberty. The big-data result located it later, in the late teens.

The mechanism Hartshorne and colleagues proposed was not neurological maturation per se. It was a combination of biological factors and the fact that adolescents and adults face different cognitive, social, and linguistic constraints than children. The cutoff at 17.5 lines up with the end of formal schooling for most learners and the cognitive transition into adult planning frameworks.

Grammar-learning ability remains high up to age 17.5, considerably later than previously assumed, and then declines steadily.

What the field actually agrees on

The debate has matured. The current consensus across the adult-attainment researchers (Birdsong, Hartshorne, DeKeyser, Bialystok) and the critical-period defenders (Hyltenstam, Long) has roughly these points of agreement.

First, age effects are real. Earlier exposure is statistically associated with better outcomes, especially for pronunciation and morphosyntax. This is robust across many studies.

Second, the effect is not a hard wall. Adult learners can reach high attainment, and a meaningful fraction reach native-equivalent scores on some measures. The wall version of the hypothesis is not defensible.

Third, the mechanisms are multiple. Some are biological (plasticity, motor control for pronunciation). Some are cognitive (declining working memory and statistical learning). Some are environmental (adults get less and worse input than children). The interaction of these produces the curve.

Fourth, the practical conclusion for instruction is that early start helps but late start works. High school is not too late. College is not too late. Adult learners with high motivation, sufficient input, and good instruction reach functional proficiency reliably.

What this means for the parent-teacher conference

Three common parent questions and the defensible answers.

Should my child start Spanish earlier?

Probably yes. The Hartshorne data suggest the grammar-learning window stays open until late adolescence, but earlier exposure produces better pronunciation outcomes and more comfortable bilingualism. If elementary Spanish is available, take it. If middle school Spanish is the entry point, that is still well within the high-acquisition window.

Is it too late for my child to become bilingual?

Almost certainly not. The age-of-exposure effects exist but the hard-wall version is not supported by the data. Teenagers and adults regularly reach high functional proficiency. The question is more about hours and quality than about age in any year before the late twenties.

Will my child speak with an accent?

If they start after roughly age six, probably yes, in the sense of perceivable accent. The intelligibility principle (covered in the intelligibility article) says this should not be a worry. Accent is not the same as intelligibility, and most adult learners are fully intelligible despite accents. Native-like pronunciation is the area where age effects are strongest. It is also the area where modern pedagogy has decided not to make native-likeness the goal.

The heritage learner question

Heritage learners present an interesting case for the critical period debate. They had early L1 exposure to Spanish, sometimes with reduced input volume compared to full Spanish-language environments. The result is often called incomplete acquisition, though that term is contested (Otheguy 2013 argues for divergent acquisition). Heritage students often have native-like pronunciation but variable morphological accuracy in low-frequency forms.

The implication is that age of exposure is not the only variable. Quantity and quality of exposure interact with age. A heritage student who heard Spanish from birth but attended English-medium school from age five may have a different profile than a heritage student in a fully bilingual home through age twelve. The critical period framework is too coarse to predict either case.

What this means for your Monday

The next time a parent asks you whether their child is too old, or whether starting Spanish in ninth grade is too late, here is the honest answer. The grammar window remains open through late adolescence. Your students are well within it. The pronunciation window has narrowed but intelligibility (not native-like accent) is the defensible goal at this age anyway. The variables that actually predict their outcome are time, input quality, motivation, and pedagogy. All of those are within your control. The age is not the blocker.

For your teaching: do not give up on accent or morphological accuracy with high school students because you believe the critical period has closed. The evidence does not support that defeatism. High school students can reach high attainment if the instruction is good and the hours are sufficient.