Analyse contrastive de deux techniques d'évaluation de traduction automatique.
Résumé
The aim of the paper is to compare two MT evaluation techniques. Results of the human evaluation of fluency and adequacy are compared with measurements of post-editing distance and post-editing time. It is shown that there is a correlation between fluency and adequacy scores, as well as between these scores and post¬editing time and distance. In other words, the number of modifications made to the MT and the time needed to post-edit appear to be indicative of the quality of the MT output.