Abstract :
[en] In recent years, there has been a growing interest for the uses, applications and adaptation of translation technologies in creative domains. In this context, research on the interplay of machine translation (MT) and video games remains limited to this date (Zhang, 2022, p. 377; Jiménez-Crespo, 2024, p. 248), even though the topic is similarly and slowly gaining attention, both in academia (see, also, Brenner, 2024) and among language service providers (Moorkens, Way & Lankford, 2024, pp. 173—175).
Starting from a study aiming at assessing the possibility of customizing MT systems to this specific domain on a technical level (Hansen & Houlmont, 2022), this contribution sets out to explore in details the translations generated by such systems, based on a human evaluation and error annotation typology specifically devised to reflect in a more tangible way the improvements and limitations of MT in such creative domains. Incidentally, this custom typology will help, on the one hand, to better understand the complexities and translation strategies involved in the translation of video games, and, on the other hand, to illustrate the advantages of the MT adaptation process in those categories where the machine is precisely expected to fail.
Lastly, the contribution will put into context some of the challenges of video game localization that highlight the need to better understand translators current practices, difficulties and workflows, and where inappropriate use of technology could exacerbate these difficulties at best, but might also have a much larger and drastic impact. Although historically seen in the gaming industry as an end-of-chain component of the game development or even a way to cut costs, translation — and, therefore, the use of MT — has a direct and definitive effect on the overall quality of the game (Hansen & Houlmont, 2022, p. 266), on its reception by players and on their engagement (Moorkens, Way & Lankford, 2024, p. 183), or on the brand name as a whole (Jiménez-Crespo 2024, p. 71).
