The politics of machine translation. Reprogramming translation studies

ABSTRACT

This paper develops a conceptual framework for analyzing the politics of machine translation. It contributes to ‘reprogramming’ Translation Studies as a discipline capable of dealing with the challenges posed by the socio-technological transformation often referred to as ‘digitization.’ It starts from the premise that the distinction between human and machine translation cannot be taken for granted and presupposed as unproblematic. Rather, it needs to be made an object of empirical investigation. The distinction between human and machine translation is not only part of the analytical vocabulary of Translation Studies. It is also part of the practical vocabulary of the social world: It is made by all kinds of social agents in all kinds of social situations and with all kinds of social meanings. These meanings, this paper suggests, can be considered political when the human/machine distinction becomes entangled with (antagonistic) us/them-distinctions – when the fact that a machine is doing the translating instead of a human becomes politically meaningful. By drawing on a series of examples, this paper demonstrates the analytical fruitfulness of such an approach and the multidimensionality of the politics of machine translation. It closes with a number of remarks on the ‘anthropolitics’ of Translation Studies.

KEYWORDS:

• Machine translation
• politics
• digitization
• sociology of translation

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Throughout this paper, the term ‘translation’ is used generically to encompass both ‘translation’ and ‘interpreting’.

2 That is why speaking of a ‘technological turn’ in Translation Studies (Jiménez-Crespo, 2020) is perhaps somewhat misleading. It would be more appropriate to say that technology has – after a hiatus during the discipline’s emancipation from linguistics – shifted its status from a goal to an object of research in Translation Studies.

3 Of course De Groote (2023, pp. 286–287) is right in arguing that the relationship between translation and media-technologies is much older than digitization. The history of this relationship is coextensive with the history of translation as no translation – and no translation theory – can exist without some mediatechnological infrastructure of communication (see also Littau, 2011, 2016; Cronin, 2003, p. 24). See Rozmysłowicz (2020, pp. 480–484) concerning the medium of writing as a condition of the possibility of machine translation.

4 See Grbić and Kujamäki (2018) for a critical discussion of the distinction between professional and non-professional translation in Translation Studies.

5 A helpful collection of anecdotes about machine translation use in everyday life can be found at https://mt-stories.com/.

6 For Germany, Hansen-Schirra and Maaß (2019) postulate a connection between machine translation, the public perception of the viability of the translation profession and declining student numbers in university translation programs. For the framing of machine translation in Chinese Media see Wang and Ping (2020).

7 Given that the focus is on social practices of distinguishing between human and machine (translation), the proposed approach differs from prominent ways of problematizing this distinction (and other related distinctions such as society/nature). Bruno Latour’s notion of hybrids or hybridization, for instance, overlooks the important fact that social agents often draw a line between them and technological artefacts. The salient point is that the ‘agency’ of things should be an empirically open question and not answered a priori on a theoretical-conceptual level.

8 The translation-sociological attitude towards the human/machine-distinction circumscribed by these questions is conceptually indebted to the emerging research on ‘human differentiation’ (Hirschauer, 2017; Dizdar et al., 2021). See also Dizdar and Rozmysłowicz (2023).

9 The relationship between translation and politics has been the topic of a vast and growing body of research in Translation Studies. For a good overview for the manifold ways in which this relationship has been understood and empirically investigated see Fernández and Evans (2018). For a conceptual reflection on ‘the political’ in Translation Studies with reference to Mouffe (and Laclau) see Sadler et al. (2023).

10 Although the concept of the ‘political’ – and thus its distinction from ‘politics’ – is central to the following analysis, the phrase ‘the politics of machine translation’ will still be used in certain places. However, it is meant to be read in terms of the ‘political’. While this might create terminological confusion, this phrase is nevertheless retained for its linguistic handiness. In this regard, it is superior to awkward constructions such as ‘the political of machine translation’ or to the constant repetition of ‘the political dimensions of machine translation.’ Also, it is reminiscent of typical phrases such as ‘the politics of translation.’

11 Choosing this memorandum as a historical point of departure for the analysis of the politics of machine translation is not entirely unproblematic. It runs the risk of reproducing a historiography of machine translation which locates its origins solely in the USA and can itself be seen as part of a politics of machine translation serving to secure the United States’ position as the dominant and technologically most advanced superpower over other states – such as the USSR where research on machine translation was also launched during the same time period (Gordin, 2020). Bearing this problematic in mind, Weaver’s memorandum is still a useful starting point as it not only reveals a particular politics of machine translation, but also the political origins of machine translation in a condensed way.

12 See also Lehman-Wilzig’s claim that machine translation is a prerequisite for world peace (Lehman-Wilzig, 2000).

13 An interpretation of Weaver’s texts as pure universalist rhetorics concealing power interests is certainly plausible and possible, given they were written at the dawn of the Cold War. However, such an interpretation would not be able to account for the fact that this rhetoric itself was taken to be plausible and possible by Weaver. The salient point here is that Weaver utilized images which have been handed down for millenia, constituting the social imaginary with regard to the significance of language(s). Perhaps the most familiar images (at least in the Western tradition) stem from the bible: The just mentioned confusion of tongues in the Tower of Babel narrative and the ‘miracle’ of Pentecost (through which the confusion is lifted and humankind reunited as a community). Weaver’s ‘Tower of Anti-Babel’ can be seen as a technological and secular substitute for this ‘miracle’.

14 According to Luhmann, trust, unlike confidence, ‘presupposes a situation of risk. You may or may not buy a used car which turns out to be a ‘lemon’. You may or may not hire a babysitter for the evening and leave him or her unsupervised in your apartment; he or she may also be a ‘lemon’’ (1988, p. 97). The solution to this problem of risk is trust. The argument here is that, unlike human translators, translation machines did not have to be trusted because they posed no risk of betrayal (this might have changed, however, with the advent of translation machines such as Google Translate and their associated privacy risks).

15 https://www.cnbc.com/2022/10/03/google-shuts-down-translate-service-in-china-.html.

16 The original Facebook-post does not seem to be accessible anymore. However, the Twitter-discussion referring to a screenshot of that post can still be accessed under https://twitter.com/OliveraStajic/status/1335942963248173059 [accessed on 31/01/23].

17 ‘Also der Herr Bürgermeister und sein Team versuchen es ja … aber offenbar mit Google Translate ' [The mayor and his team are trying … but apparently with Google Translate ; my translation]. https://twitter.com/OliveraStajic/status/1335942963248173059 [accessed on 31/01/23].

18 Such positions could be observed at the ‘Is Machine Translation Translation?'-panel at the 2022 EST-conference in Oslo. The name of the panel itself is already an interesting moment hinting at the current politics of Translation Studies.

Additional information

Funding

This work was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Grant SFB 1482/1–2021–442261292.

Notes on contributors

Tomasz Rozmysłowicz

Tomasz Rozmysłowicz is a postdoctoral researcher at the CRC 1482 Studies in Human Categorisation at the Johannes Gutenberg University in Mainz. His research interests include the sociology, history, and politics of translation, machine translation, and translation in the context of forced migration. He is co-editor of Chronotopos – A Journal of Translation History.