Nune Ayvazyan (URV); Anthony Pym (URV)
Machine-translation links appear regularly in official administrative information for culturally and linguistically diverse (CALD) communities in Catalonia, supplementing the translated and post-edited versions variously provided in Catalan, Spanish, English or Aranese. In the case of COVID-19 directives, raw machine translations were provided, resulting in errors that would be comical if they did not concern healthcare (“wash the hands regularly with ice”). Even when such links do not appear, younger users in many communities resort to machine translation in order to comprehend official information.
Here we report on the effectiveness and inclusivity of machine translation in COVID vaccination information in English, Russian, Arabic and Chinese, as indicated in a series of eye-tracking reception tests for a short informative text translated in three different ways. We focus in the first place on the way raw machine translation is received and how end-users activate degrees of machine-translation literacy when negotiating clear translation errors (cf. Bowker 2009, 2019; Bowker and Buitrago Ciro 2019; cf. Ayvazyan and Pym 2016, 2022 for Russian-speaking communities in Catalonia). We then consider the reception effects of human post-editing as a second kind of machine translation literacy that requires specialized training to be carried out effectively. Finally, we test the reception effects of pre-editing, understood as the writing of official start texts in such a way that the typical errors are avoided before they occur. In all three cases, we evaluate both comprehension and trust in the translated text.
Much as any use of machine translation may compromise users’ rights to full and clear healthcare information, we hypothesize that pre-editing in particular enables a series of trade-off positions (Grin 2022) where receptive literacy overcomes comprehension problems, trust in the text is not fatally compromised, machine translation aids in engagement with official languages, and the long tail of minority languages may be efficiently included in public communication.
Keywords: machine translation, inclusion, minority languages.