Opening translation

Brett NEILSON


  • fr
  • en
  • ar
  • Multilingual mode
  • Simple mode

What does it mean to open translation? The term implies the movement of translation beyond any closed or finite task. In widespread usage, it refers to a new participatory ecology of translation emerging on the internet. Practitioners of open translation operate in peer production networks, utilising free and open source software tools to produce translations that are freely available to as many people as possible. Theirs is a collective and distributed practice. For many it provides a platform from which to contest the status of translations as derivative works and the institution of individual authorship that underlies intellectual property rights. As such it can challenge an important building block of contemporary capitalism. This article supports and contributes to the project of open translation by asking how practitioners represent their practice. I am interested not only in how open translation is performed but also in the political motivations of its proponents. Drawing on studies that question the proposition that discrete languages exist before the act of translation (Sakai 1997), I investigate how open translation figures the relation between languages. I also ask if the collective subject constructed through such collaborative translation practices is a political figure adequate to the production of the common.

Instances of open translation are diffuse and do not conform to an ideal type. Examples include meedan.net and globalvoicesonline.org. The first uses tools of machine translation, machine assisted translation and distributed human translation to publish news stories and commentaries in Arabic and English. The second uses distributed human translation to publish in eighteen different languages content from a site that collates English language posts from over two hundred blogs and citizen media sources. For the purposes of this article, I draw information about open translation from one primary source: the Open Translation Tools manual. This was produced in June 2009 by Aspiration (an outfit that connects civil society organisations with software solutions) and FLOSSManuals (a platform for the production of manuals that explain how to install and use a range of free and open source software). Open Translation Tools is the result of a ‘book sprint’ held over five days in Amsterdam. About fifteen people gathered for twelve hours each day to develop the content. Their work is distributed on the FLOSSManuals site and freely downloadable as a PDF. Covering legal and political as well as technical matters, the manual is the most comprehensive guide to open translation available.

Open Translation Tools (2009: 11) defines open translation as a ‘nascent field of practice emerging at the crossroads of three dynamic movements of the internet and information eras’:

1.    Open content or the publication of knowledge resources under open licenses such as Creative Commons or Free Document License, which encourage redistribution, modification and broad reuse.

2.    FLOSS (free/libre/open source software) that provides an alternative to proprietary software and corresponding closed data formats by allowing users to study, change or improve its design through the availability of its source code.

3.    Open or peer production models that use the connected but distributed nature of the internet to bring collaborative human resources to bear upon the production of a specific outcome (11).

Open translation, then, is ‘the set of practices and work processes for translating and maintaining open content using FLOSS tools, and leveraging the open nature of the internet to make that content and those tools and processes freely available to the largest number of contributors and consumers’. These developments aim ‘to lower the barriers to participation in cross-language knowledge exchange, and help to avoid replication of the “expert culture” that permeates the professional translation industry.’ In so doing, they seek to effect ‘the transition of translation from an individual to a team sport’ (11).

It is worth dwelling briefly on the notion of openness that repeatedly animates these definitions. The opposite of open in this context is not closed but proprietary. As Kelty (2008: 177) notes, openness is at once a technical and a moral concept. On the one hand, it is about protocols, standards and implementations that ensure interoperability across different software platforms and thus counteract the monopolisation of computer networks by specific commercial interests. On the other hand, it expresses legal and social ideas about the production, distribution and modifiability of software and different kinds of content in contrast to the artificial scarcity created by intellectual property rights.

Openness is something more than publicness, since it is not merely constructed through discourse in the traditional sense of speech, writing and assembly. It involves the capacity to build, maintain, control and modify the technical and social infrastructures that make such exchange possible in the first place. Openness provides opportunities for collaboration and the construction of collectivities independent from traditional forms of constituted power. Potentially it offers all users the possibility to participate in developments regardless of their interests or capacities. Those ‘who disagree can simply “fork” and start their own development branch without losing access to the means of production’ (Schneider 2006). In this sense, openness is a precondition for the production of the common. But as a precondition, it is not necessarily sufficient to guarantee the continuation of this process. By examining the practice of open translation, this paper inquires into the limits of this sufficiency. It suggests the need for open translation to disinvest in the dream of building bridges between cultures. Instead it maintains that the production of the common might be furthered by a practice of open translation that seeks not the communication of information between language groups but the institution of relation at sites of incommensurability.

 

Between human and machine

For Walter Benjamin (1968: 80), ‘the task of the translator is to release in his own language that pure language which is under the spell of another’. Little did Benjamin know when he wrote these words in 1923 that the mystical notion of a pure language would also inform the vision of Warren Weaver, one of the pioneers of machine translation. Weaver closed a memorandum of 1949 with a metaphor that would become one of the best known in this field:

Thus may it be true that the way to translate from Chinese to Arabic, or from Russian to Portuguese, is not to attempt the direct route, shouting from tower to tower. Perhaps the way is to descend, from each language, down to the common base of human communication—the real but as yet undiscovered universal language—and then re-emerge by whatever particular route is convenient. (Weaver 1955: 23).

The dream of a universal language that overcomes the differences between languages as well as the internal opacities of mediation and signification is a very old one, and it takes a variety of forms. Often they are banal, sentimental or, as in the case of Benjamin, theological. Weaver’s vision of a ‘real but as yet undiscovered universal language’ is above all instrumental. By no accident was he also coauthor (with Claude Shannon) of The Mathematical Model of Communication (1949), which understood noise as mere interference in the transmission of information. In his memorandum on translation, he wrote: ‘When I look at an article in Russian, I say: “This is written in English, but it has been coded in some strange symbols. I will now proceed to decode”’ (1955: 18). But despite decades of research, this approach to language as code has yielded limited results in the machine translation of human languages. Most web based machine translation tools these days prefer corpus-based to rule-based methods, deploying statistical techniques and huge libraries of translated texts to translate between languages. In this context, the dream of a universal language, Benjamin’s as much as Weaver’s, has all but disappeared.

For Raley (2003: 293), the rise of machine translation accompanies ‘a renewed appreciation for the basic and easily translatable (the non-figurative, the non-literary); and a new economics and pragmatics of language and information exchange’. Machine translation posits the fixity of languages, ignores rhetoric and context, and disregards the influence of political struggles upon translation. As such, it ‘does not present us with a new theory of translation in the context of globalization, but rather with a further dimension and renewed rationale for it’ (293). In the case of open translation, however, machine translation is approached as a limited tool that can be used to ease the work flow in a collaborative process. The appropriate image is perhaps that of an assemblage of human and nonhuman actors (Latour 2005). The negative vision of machine translation, as destructive of the figurative and contextual dimensions of language, is replaced by one in which it acquires a circumscribed role in a wider process of distributed human translation.

Open Translation Tools notes that machine translation systems are ‘well below the quality threshold necessary to enable readers to participate in conversations and debates with speakers of other languages’ (2009: 1). The manual recommends ‘using machine translation where it is strongest, for quickly obtaining rough draft translations that in turn, can be edited or replaced by people’ (139). This requires the design and implementation of open source systems that can coordinate this collaborative process.  There are a number of options in this regard. The platforms can be ‘crowd sourced’ systems where anyone can contribute, managed systems accessible to only approved participants, or mixed systems where open submissions are moderated. They can also include social networking and scoring systems by which users can evaluate translations submitted by their peers. All work completed in these contexts is fed back into translation memory banks, from which it can be retrieved to aid future translations. The difficulty is that open translation ‘tool coverage is incomplete, and existing tools rarely interoperate or share standards for data exchange’ (12).

Advocates of open translation recognise that ‘regional and cultural issues in translation work cannot be overstated’ (13). But they do not focus on what translation teaches us about cultural dynamics, the role of borders or the textures of everyday life in a globalising world. The political moment of their practice is supplied by the opposition to the regimes of scarcity and control established by the dominant regime of intellectual property rights. The question is whether this stance accompanies or encourages a politics of translation that is adequate to the cultural dimensions of globalisation and the production of the common.

Open Translation Tools is not short on rhetoric that signals the role of translation in a globalised world. Consider the following instance:

Translation is the missing ingredient in a participatory global media ecosystem that potentially leads to a world with more complete and more nuanced understanding of the events that shape our shared circumstances. This becomes all the more true as dialog scales about our response to climate change and other global challenges … While participatory media and open knowledge networks offer global citizens better access to content, the free software movement is a foundation that enables emergent knowledge economies around the world (2009: 9).

Here the foundational role of free software is clearly stated. Open translation systems offer enhanced opportunities for ‘understanding’ and making content more accessible to ‘global citizens’. But it is the sharing of code that supplies the material and intellectual infrastructure for ‘emergent knowledge economies’. This is a very different image from Weaver’s vision of universal language providing a ‘common base of human communication’ (1955: 23). Dyer-Witheford (2006) argues that the open source movement creates new political practices of social association. At stake is the production of a ‘networked commons’ that provides ‘the strategic and enabling point’ for the complex coordination of ‘other commons sectors’, such as those associated with natural resources and the welfare state. The attribution of such a pivotal role to free and open source software is a feature of many contemporary accounts of the production of the common. What needs to be questioned, especially in the context of open translation systems, is how these accounts sideline (or at least subordinate) the role of translation itself in the fabrication of common knowledge resources. To highlight this problem it is not necessary to remember theoretical formulations such as Bauman’s (1999: 201) contention that ‘translation is the common feature in all forms of life’. It is sufficient to recall the role played by translation in the history of intellectual property.

Translation as derivative?

It is a well known fact of legal history that the first copyright case in Britain involved the question of whether a translation constitutes a new work or merely a version of an existing one. In 1720, the estate of George Burnet (author of a Latin volume of natural philosophy published in 1692) filed against a group of publishers who wanted to publish an English translation of the work. As Greene (2005: 147) explains, the judge decided in favour of the booksellers, making ‘the Lockean argument that the translators had mixed their labor with the text in question and thus could be said at least in some degree to be the new work’s proprietors’. A similar decision was upheld in the United States as late as 1853. In Stowe v. Thomas, a Pennsylvania court decided that an unauthorised German translation of Uncle Tom’s Cabin did not constitute a copy under copyright law. Not until 1870 did Congress reverse this holding by granting authors a form of derivative right in translations. In the international sphere, such a right was not extended until the Berne Convention of 1886. Open Translation Tools (2009: 115) suggests that the absence of translation rights prior to this time ‘followed from the linguistically defined boundaries of the publishing industry of the day’. There was ‘no incentive to lobby for translation rights, because the publishing houses did not work across markets’ (115). By this account, the notion of a translation as a derivative work derives from the international expansion of the publishing industry.

The historical emergence of translation rights is a peculiar legal attempt to control the unruly capacity of translation to challenge intellectual property rights. That this occurs in a context of transborder expansion is significant. Apter (2009: 89) notes that translation reveals the limits of intellectual property because ‘it is a peculiar genre; one that, counter to Romantic values and myths of avant-garde originality, exalts the art of the copy, flaunts its derivativeness, and proudly bears the lead weight of predication on literary antecedant’. In the current global environment, where human and technological mobilities contest and multiply borders, this process has become more complex and intense. In the case of internet based translation services, the situation is pronounced. Hypothesising the arrival of high quality machine translation, Ketzan (2007: 206) predicts ‘massive copyright infringement on an unprecedented global scale’. Present legal arrangements cannot cope with this scenario. Ketzan suggests measures such as the introduction of a ‘do not translate’ tag that would allow web designers to identify content unavailable for translation. Building on these proposals, Open Translation Tools (2009: 114) calls for ‘research and legal advocacy’ addressing the translation of digital content.

Clearly the law governing the intellectual property of translation is once again in crisis as technological developments outpace its enforceability. It is as if the derivative status of translations has overshadowed the notion of originality. Something more is at stake here than the withering of the aura of the work of art that, for Benjamin, was the hallmark of technical reproducibility. This is because derivation is something other than reproduction. It does not replicate an original but introduces a new layer of meaning and value that occasions an encounter with the untranslatable. If Benjamin understood the reproducibility of art with respect to the capitalist production of commodities, perhaps it is now appropriate to analyse the dynamics of translation in analogy to the operations of the financial derivative.

Such a parallel is hedged with difficulties and must be approached with caution. It is important to avoid reducing translation to a machine that, like the financial derivative, converts difference to indifference and approaches all specific qualities of life as exchangeable attributes of risk. Understanding the derivation involved in translation in analogy to the financial derivative means tracing how the latter, which began its life as a specialist contract used to create commodity price certainty, has come to play a central role in global finance. No longer do derivates operate only within particular product categories – such as wheat, cotton or barley derivatives. From the 1980s, there was a rapid expansion of derivative products that were financially oriented from the start and whose value could not be understood to derive from the price of underlying commodities. Since 1990, these market-traded financial derivatives have eclipsed commodity derivatives in size and importance (Bryan and Rafferty 2006: 39-67). As became clear with the financial crisis that began in August 2007, such derivative products now play a role in determining the price of commodities themselves.

According to Martin (2007: 154), derivatives ‘multiply the means by which interconnection is possible, deepen the density and intensity of mutual contingency, and increase the scale of transaction devoted to interdependencies’. Perhaps it makes better sense, in this regard, to speak of derivates as translation than of translation as derivative. Since translation, by introducing difference at sites of incommensurability, does not necessarily seek, like the derivative, to commensurate across boundaries or ‘parcel’ all qualities in the ‘body of value’ (154). Unlike the reproducible artwork, it exceeds and bears back upon the logic of the commodity. In other words, it is not a dynamic of originality and replication but a means of blending different qualities and forms. As such, translation is not a secondary activity but a social practice in its own right. Indeed, it can play a powerful role in the contestation of intellectual property and the production of the common. This is no more the case than when it occurs in tandem with the development of free and open source software. But such a combination is not in itself sufficient to create an effective opposition to contemporary technologies of accumulation. Whenever or wherever translation is held to the ends of communication, the augmentation of understanding or the smoothing over of difference, it is susceptible to the capture of capital.

Heterolingual openness

In a series of important writings, Naoki Sakai argues that translation does not establish a bridge between languages but rather separates them. More specifically, he locates a radical disjuncture between the representation of translation as an exchange of equal values between distinct and fully formed linguistic communities and the practice of translation, which is potentially endless and creates continuity from discontinuity. Is it possible, he asks, to ‘isolate and juxtapose’ languages ‘as individual units, like apples, for example, and unlike water’ (2006: 72)? The widespread understanding of translation as the communicative transfer of a message between one language and another does precisely this. At stake is what Sakai calls the homolingual mode of address – ‘that is, a regime of someone relating herself or himself to others in enunciation whereby the addresser adopts the position representative of a putatively homogeneous language society and relates to the addressees, who are also representative of an equally homogeneous language community’ (1997: 4). In contrast to this, he develops the concept of the heterolingual mode of address, which describes ‘a situation where one addresses oneself as a foreigner to another foreigner’ (2006: 75). By seeking to engage with mixed as well as differing audiences, the heterolingual address works to produce the common ‘precisely by never assuming communality or taking comprehension for granted – across or within the borders of a nation-state’ (Morris 1997: xv).

Mezzadra (2007) extends this argument by relating the homolingual address to capital’s attempt to reduce heterogeneous forms of life to the abstract language of value. By contrast, he links the heterolingual address to social practices of cooperation and struggle that constitute a new political subject. What needs to be asked is how these dynamics of translation and address relate to the collective subject of open translation. Insofar as this is a subject forged through the production, distribution and modification of free and open source software, it is engaged in practices of collaboration and struggle that can create common knowledge resources. While some thinkers question the way in which free and open source software initiatives ‘presuppose a capitalist economy for their functioning’ (Shaviro 2008), these practices can also open spaces from which to initiate and widen struggles against this same system. Hardt and Negri (2004: 340) imagine ‘an open source society, that is, a society whose source code is revealed so that we all can work collaboratively to solve its bugs and create new, better social programs’. The problem is how to expand open source practices from computer networks into the wider social domain, Furthermore, there is the question of how to do this while avoiding what Pasquinelli (2008: 74) calls the impasse of ‘digitalism’, that is, ‘the belief in a pure symmetry of the technological over the social’. Open translation offers a possible path, since it moves from source code to the socially rich domain of language and translation, aiming to build ‘communities committed to bridging in a polyglot web, to preserving smaller languages and to making tools and knowledge accessible to a global audience (FLOSSManuals 2009: 2). Already, however, the use of the word ‘bridging’ above implies a drift toward homolingual address that endangers the emergence of a common based on production rather than property rights.

Open Translation Tools (2009: 11) describes social translation on the internet as a way for ‘translators to essentially become … de-facto bridges, deciding which content to move between language communities’. Similarly, there is mention of constructing a ‘global understanding index’ that ‘would be limited by the scope and rate of the transfer of knowledge and information across language communities’ (115). Here the homolingual understanding of translation as the transfer of a message between separate and fully formed language groups is fully in place. Indeed, it could be argued that such a model of translation is inscribed into open translation systems at the level of code, for instance, in the building of statistical databanks for translation between ‘language pairs’ or in the grouping of social translation site users according to membership in specific language communities. Such a representation of translation, I suggest, works against the production of the common by separating out language communities and placing the translator in a sovereign position from which to decide the content to move between such groups.

Imagining an open translation system adequate to the heterolingual mode of address is not simply a matter of purging this representation of translation from the programs and the manuals. It is also a matter of asking how collaborative social practices might displace the collective subject of open translation from a position of neutral and transcendent arbitration. For Sakai (2006: 75), the translator cannot occupy this site since she ‘is internally split and multiple, devoid of a stable position’. The ‘speaking I’ and the ‘I that is signified’ are not expected to coincide in translation (74). Thus the translator ‘cannot be an “individual” in the sense of individuum, the undividable unit’ (75). As Morris (1997: xi) explains, the capacity to translate ‘is not a property of individuals with a “talent for translation” but a complex and variable condition of sociality; for many people, learning multiple languages is not a gift, a pleasure, or a tool of trade but a forced process scored by painful legacies of history’. Open translation furnishes the potential to activate this sociality, with all its legacies, and to unleash it in a distributed and collaborative process that constitutes a new political subject, at once collective and multiple.

At stake is nothing less than a retooling of open translation. This is no easy task, either technically or politically. But the vision of a collaborative translation system that combines the work of humans and machines, forges new kinds of social relations and addresses mixed and linguistically heterogeneous audiences should not be impossible to realise. The open source, open content and peer to peer aspects of such a vision suggest a production of common resources in the digital realm. Similarly, the heterolingual questioning of the unitary and ethno-national grounding of languages involves the production of new kinds of commonality in the social domain. Added to this is the need to rethink the labour of translation in ways that pull away from volunteerism and confront the need to reassess and redefine the concept of exploitation under current global conditions. Within these parameters it is also necessary to construct a flexible program of translation: what is the content to be translated remains an unanswered and politically unresolved question. The path is difficult and no easy synthesis is possible. But in this task, thinkers interested in the political and cultural dynamics of translation and practitioners of open translation have something to learn from each other.                                                                                                                                                 

 

References

Apter, E. (2009) ‘What is Yours, Ours and Mine’, Angelaki: Journal of the Theoretical Humanities 14(1): 87-100.

Bauman, Z. (1999) In Search of Politics. Cambridge: Polity.

Benjamin, W. (1968) ‘The Task of the Translator’, in H. Zohn, ed. Illuminations: Essays and Reflections. New York: Schocken Books, 69-82.

Bryan, D. and M. Rafferty (2006) Capitalism with Derivatives: A Political Economy of Financial Derivatives, Capital and Class. Houndmills: Palgrave.

Dyer-Witheford, N. (2006) ‘The Circulation of the Common’, accessed 14 August 2009.

FLOSSManuals (2009) Open Translation Tools, accessed 12 August 2009.

Greene, J. (2005) The Trouble with Ownership: Literary Property and Authorial Liability in England, 1660-1730. Philadelphia: University of Pennsylvania Press.

Hardt, M. and A. Negri (2004) Multitude: War and Democracy in the Age of Empire. New York: The Penguin Press.

Kelty, C. M. (2008) Two Bits: The Cultural Significance of Free Software. Durham: Duke University Press.

Ketzan, E. (2007) ‘Rebuilding Babel: Copyright and the Future of Online Machine Translation’, Tulane Journal of Technology and Intellectual Property 9: 205-234.

Latour, B. (2005) Reassembling the Social: An Introduction to Actor-Network-Theory. Oxford: Oxford University Press.

Martin, R. (2007) Empire of Indifference: American War and the Financial Logic of Risk Management. Durham: Duke University Press.

Mezzadra, S. (2007) ‘Living in Transition: Toward a Heterolingual Theory of the Multitude’, Transversal, accessed 15 August 2009.

Morris, M. (1997) ‘Foreward’, in N. Sakai, Translation and Subjectivity: On ‘Japan’ and Cultural Nationalism. Minneapolis: University of Minnesota Press, ix-xxii.

Pasqinelli, M. (2008) Animal Spirits: A Bestiary of the Commons. Rotterdam: NAi Publishers.

Raley, R. (2003) ‘Machine Translation and Global English’, The Yale Journal of Criticism 16(2): 291-313.

Sakai, N. (1997) Translation and Subjectivity: On ‘Japan’ and Cultural Nationalism. Minneapolis: University of Minnesota Press.

Sakai, N. (2006) ‘Translation’, Theory, Culture and Society 23(2-3): 71-86.

Shannon, C. and W. Weaver (1949) The Mathematical Theory of Communication. Urbana: University of Illinois Press.

Shaviro, S. (2008) ‘Paolo Virno, Multitude: Between Innovation and Negation’, The Pinocchio Theory, accessed 16 August 2009.

Schneider, F. (2006) ‘Collaboration: Some Thoughts Concerning New Ways of Working and Learning Together’, accessed 12 August 2009.

Weaver, W. (1955) ‘Translation’, in W.N. Locke and A.D. Booth, eds. Machine Translation of Languages; Fourteen Essays. Cambridge: MIT Press, 15-23.