In this chapter, we focus on one component of an ongoing study conducted by the Faculty of Translation and Interpreting (FTI) and the Hôpitaux Universitaire de Genève (HUG): doctor–patient communication in emergency settings. The main aim of the paper is to compare translation quality obtained with two different translation tools for the language pair French to Arabic: Google Translate, an MT statistical system, and BabelDr, a flexible phraselator with speech recognition, developed by the FTI and HUG (Rayner et al., 2015; Bouillon et al., 2016). The comparison will be made from two different angles. We want to quantify (1) the impact of translation output quality on the transmission of information to patients; and (2) the impact of a type of incorrect translation in relation to the diagnosis. The chapter goes on to present a description of the two tools and data used in the experiment. It then provides an analysis of the translation output. The motivation behind the research reported here is to find a middle-ground solution that produces reliable output in Arabic and provides generative language with speech recognition.