Linguistic dissimilarity increases age-related decline in adult language learning

56,024 adult immigrants with 50 different first languages

Adult immigrants to the Netherlands who came for study or work and sat the Dutch state language proficiency examinations. ADULTS ONLY - outside this archive's 4-18 age scope, and included because it is the best large-sample test of the age gradient on real proficiency exams rather than a quiz.

The measurement contrast with the internet-quiz literature is the reason this source matters: the outcome is a REAL STATE EXAMINATION with separate speaking, writing, listening and reading scores, not a researcher-designed grammaticality quiz. Linguistic dissimilarity between each L1 and Dutch is measured separately for morphology, vocabulary and phonology. Observational; age of arrival is not randomised and the sample is restricted to people who chose to sit a state exam.

Performance on all four language skills declines MONOTONICALLY after an age of arrival of about 25 years - a steady decline resembling ordinary cognitive-ageing trajectories, not a sharp adolescent breakpoint. Linguistic dissimilarity between L1 and Dutch increases the age-related decline across all four skills, and most strongly for speaking. The authors read the constraints as reflecting both language-learning ability and acquired L1 proficiency.

High as an observational design; age of arrival, motivation and the decision to sit a state exam are all self-selected. The finding that decline tracks general cognitive-ageing curves is itself a hint that the mechanism is not language-specific maturation.

Part of a programme of work on the same Dutch state-exam dataset by the same group, so internally consistent rather than independently replicated. It is the strongest available check on the internet-quiz findings because the instrument is a real proficiency exam.