(Re-)calibrating the CEFR Descriptors: A Replication Study

Despite its significance, the research supporting the calibration of many of the CEFR illustrative descriptors has not been systematically replicated. This study addresses that gap, providing a partial replication of North and Schneider’s original research. Four hundred and seventy-one teachers from 45 countries joined an online rating exercise modelled on North’s (1996, 2000) methodology in which they assessed their own students using CEFR descriptors (337 of the original CEFR descriptors and 32 from the more recent Companion volume). To control severity, each teacher was also asked to assess samples of written work produced by students of other participating teachers. Data was analyzed using many-facet Rasch measurement (MFRM). Findings showed that, when anchored to North and Schneider’s original values, there was a strong correlation between judgements in the replication study and the original CEFR, though with some notable discrepancies. Implications are drawn for the validity and ongoing use of the CEFR descriptors.