Some remarks on sensitive periods and the methodological rigor of their examination

Abstract

The hypothesis that after a certain age the ability to learn languages suddenly diminishes has been equally prominent and divisive in language acquisition studies for more than seven decades. It has also become one of the fundamental questions about the nature of second language acquisition. However, no substantive consensus has been reached. Various objections are posed to existing studies, including small sample size, type of exposure (classroom instruction as opposed to naturalistic exposure), age range, homogeneity of language background, &c. To fit most existing objections, we tested a group of English learners (13-21 y.o.) with homogeneous language backgrounds (Ukrainian-Russian bilinguals) and comparable exposure to English over the course of three (Experiment 1) and six (Experiment 2) months of their exposure to a native speaker of English for approximately two hours a week. The interaction was limited to spoken discourse. The participants took a specifically designed and validated speaking assessment before entering the experiment, after three months (~ 26 hours of exposure), and after six months (~42 hours of exposure). Our results reveal that cumulative growth is not correlated with age. We thus show that both the hypothesized critical periods and sensitive periods for language acquisition lack a clear empirical basis. The populations were n=140 for Experiment 2 and n=280 for Experiment 1. Apart from running linear, quadratic, and polynomial (3) regression on age vs. overall growth and age vs. vocabulary growth/grammar growth/ as well as other subscales. All of them were a very bad fit for the model. We also ran Cohen's D and overall coefficient which showed that not only is growth not different depending on age, it is similar across all ages.

Keywords

second language acquisition; critical periods; sensitive periods