Issues of trust in machine translation
Résumé
Following Marcello Vitali's analysis of the digital ecosystem of publication (On Editorialisation, 2018), I would like to address the various issues of trust in and through a range of translation technologies. In terms of machine translation (focusing here on NMT), the study will include both online platforms dedicated to translation (Google Translate, DeepL, Lingvanex) and embedded translation tools (Microsoft, Amazon, Netflix...). The study of their affordances and interfaces should reveal how they are designed to feed into our attention mechanisms and override misgivings (Citton 2017). Being anonymous, instantaneous and ubiquitous, these tools “invisibilise” not only the translator but also the process of translation. Beyond the production of target texts, one should also examine how translation data is retrieved, processed and stored. Finally, one should scrutinize the accountability policies associated with those tools. These findings call for a more assertive stance of human translators, who may usefully enlist the notion of trust as a selling point (Pym 2014). As per the market itself, since NMT falls within the wider range of AI powered technologies, one can note the absence of codes of conduct and oversight bodies in the field of online translation services, underlining the need for auditing structures (Villani 2017) and calling for a wider discussion of the value of translation as a commons.
Origine | Fichiers produits par l'(les) auteur(s) |
---|