Automatic Machine Translation Evaluation: A Qualitative Approach (2015)

Supervisor: Victoria Arranz i Irene Castellón

The present study addresses the problem of Automatic Evaluation of MachineTranslation (MT) from a linguistic perspective. Most of the studies performed in thisarea focus on quantitative analyses based on correlation coefficients; however, little hasbeen done as regards a more qualitative approach, going beyond correlations andanalysing data in detail. This thesis aims at shedding some light on the suitability,influence and combination of linguistic information to evaluate MT output, not onlyrestricting our research to the correlation with human judgements but also basing it on aqualitative analysis. More precisely, this research intends to emphasize the effectivenessof linguistic analysis in order to identify and test those linguistic features that help inevaluating traditional concepts of adequacy and fluency. In order to perform thisresearch we have focused on MT output in English, with an application to Spanish so asto test the portability of our approach.