Expressed emotion (EE) measures have been created in English; adaptation into a foreign language is difficult. The aim of this study was to adapt the five minutes speech sample (FMSS), with a designed procedure ensuring optimum quality of the adaptation, and thus better trans‐cultural validity. A strategy for improving inter‐rater agreement comprised three phases: (1) phase of initial ratings (70 French samples), (2) experimental phase in two steps: ratings of 40 other samples in French, followed by analysis of differences between the French‐language ratings and English‐language ratings and (3) final rating phase of the initial 70 samples. For each phase, the κ coefficients measuring inter‐rater agreement were calculated and compared using a bootstrap procedure. The improvements between these scorings were significant at p < 0.05 (phase 2 initial versus phase 2 final and phases 1 versus 3). The French inter‐rater agreement significantly improved after this procedure. Copyright © 2010 John Wiley & Sons, Ltd and Eating Disorders Association.