6. Speech to Sign Language Translation

At Geneva University Hospital (HUG), the largest hospital in Geneva, 52% of patients are foreigners living in Switzerland and more than 10% do not speak French at all (Bouillon et al., 2017). In the context of the current migration crisis, this situation is becoming more and more common: HUG medical staff are increasingly confronted with patients they do not share the same language with, especially in emergency rooms and in services for immigrants. Languages of particular importance to the Geneva hospital are Tigrinya, Farsi and Arabic.

Language barriers pose serious challenges to quality, safety and equity in health care (Flores et al. 2003; Wassermann et al. 2014). The BabelDr speech translation application, a partnership between the HUG and the Faculty of Translation and Interpretation, aims to solve these problems: it enables the translation of anamnesis questions into the patient's language. To ensure reliability, the system is based on a set of sentences that are humanely pre-translated. The system recognizes the oral question and then links it to the phrase closest to its database, using mixed methods.

In recent years, research in the area of sign language has highlighted the challenges that deaf patients face when they need access to medical information through sign language. Middleton et al (2010) in the United Kingdom and Lezzoni et al (2004) in Hong Kong collected data from deaf and hard of hearing people. Amongst the issues pointed out by the respondents, the lack of awareness of deafness among medical staff has been highlighted. In the United States, research has shown that Deaf people often have low literacy skills, and these same people have limited access to health-related training and information (Pollard and Barnett 2009), which may result in low health literacy in the deaf community (Hedding and Kaufman 2012). Generally speaking, deaf people are not experts in health and those who work in this field are rare (Earhart and Hauser 2008; Moreland and Agan 2012).

Following these premises, the main objectives of Research Area 6 are as follows:

• to extend the BabelDr platform for sign language
• to translate anamnesis questions into Swiss-French Sign Language
• to come up with specific advice on how to translate this kind of sentences into sign language
• to evaluate the application in emergency departments with standardized patients (HUG)