…but it stays open longer than expected.
The ability of native and non-native English speakers to learn English grammar is stable to age 17 then decreases with age, according to a landmark study by Hartshorne, Tenenbaum and Pinker from MIT and Harvard, US.
There has been a long-standing debate over the ‘critical period’ for language acquisition: does it exist, and if so, how and when does it begin and end? For adult learners of English the question is then: can adult learners expect to be as proficient as those who learn a second language in childhood?
A major challenge for any study hoping to answer these questions is getting the necessary data regarding both the age learners were first exposed to English and how many years they were exposed in order to predict test results – at whatever age they were tested. This is messy data and having confidence in the results requires a very big dataset, one way beyond the usual studies using a few classes of language students. When Hartshorne and colleagues estimated just how big this dataset would need to be, they realised they would need several hundred thousand participants.
To collect data on this scale, an internet quiz, in English, was designed, asking for information, such as ‘native languages learned from birth’ and including a wide-ranging multiple-choice test of syntax: using the passive, relative clauses, prepositions, pronouns, modals, tenses and many other targets. Pilot testing was used to choose the more informative items from a larger pool and the overall test time for the final 132 items was less than 10 minutes as any longer might risk losing the participants’ attention but any less and there might not be enough data.
A highly innovative part of the study design was the inclusion of a further set of items reflecting differences in responses that were influenced by the participant’s non-English native language or dialect of English, such as Irish or Canadian. This enabled immediate feedback estimating the participants first language, including English dialect, from their responses. This game-like part of the test was not used in the main academic analyses but was very popular with participants so that the test was well-circulated resulting in a final sample size of 669,498 native and non-native English speakers aged seven to 89 years.
Thirty-eight native speaker languages were reported as spoken by at least 1000 participants the top five being English, Finnish, Turkish, German and Russian. Despite the well-observed phenomenon of different native speakers tending to make different kinds of errors in English, there did not appear to be any significant influence of native language on the main study’s results below.
The analyses looked at how age of the learner influenced the rate of learning, the peak attainment for that individual, and how long it took to reach that peak proficiency. Overall, the ability to learn grammar remained steady through childhood and into adolescence until 17.4 years, but then declined. This finding helps to answer questions about how the critical period comes about since theories generally relate to known changes occurring in childhood. For example, the steadiness of grammar acquisition throughout childhood runs counter to arguments that neuronal changes known to occur in early childhood, such as neuronal death or pruning, could cause the window for language learning to close. Similarly, the effects of hormones during puberty does not fit the timescale.
Regarding how long it takes to reach personal best proficiency, this was around 30 years for both native and non-native learners who were immersed in English. The degree of proficiency achieved did not alter with age if the learner began being exposed to the language before age 10 to 12 but was lower for those beginning language learning thereafter.
Research is constantly expanding the timeframe for brain development and maturation so that it is now commonplace to talk about; for example, ongoing development of the prefrontal cortex throughout late adolescence and early adulthood. These changes have not been considered as potentially associated with language acquisition in particular so this paper has opened up new avenues of research to determine why and how the window closes and signals the end of peak language learning and what can be done to keep that window open for adult learners.
REFERENCE
Hartshorne, J. K., Tenenbaum, J. B. and Pinker, S. (2018) A critical period for second language acquisition: Evidence from 2/3 million English speakers. Cognition 117: 263-277.
