From Early Starters to Late Finishers? A Longitudinal Study of Early Foreign Language Learning in School

two full cohorts of English learners in North Rhine-Westphalia, Germany (large-scale state study, 2010-2014)

German school students learning English as a foreign language, tested at the start of Year 5 (age ~10) and again in Year 7 (age ~12). The two cohorts differ in whether English instruction began in Year 1 or in Year 3.

Cohort comparison exploiting a state policy change that moved the start of English instruction from Year 3 to Year 1 - so assignment is by birth year and school policy rather than by family choice, which is stronger identification than the usual early-start literature. It is still not a randomised or discontinuity design, and the two cohorts differ in calendar time as well as in starting age. Outcomes are receptive skills (reading and listening comprehension) on the study's assessment battery.

The early-start advantage exists and then reverses. In Year 5 the early cohort outperformed peers with less and later exposure to English. By Year 7 the LATE starters had surpassed the early starters. This is the cleanest school-based counterexample to "start as early as possible": four extra years of primary-school English bought an advantage that had not merely faded but inverted within two years of secondary school.

Medium. Cohorts are defined by a policy change rather than by parental choice, so the usual selection story is weaker here than in most early-start research; the residual threats are cohort effects, differential primary-school teaching quality, and the transition into different secondary school tracks.

Directly contradicted by Porsch et al. (2023), which uses a nationally representative German sample of more than 30,000 Year 9 students and finds LONG-TERM ADVANTAGES for early starters in reading and listening. The two German studies disagree at different horizons and with different sampling; the honest reading is that the effect of an earlier school start is small and unstable in sign, not that either result is decisive.