Part II: “Professionals into proletariats” *
Digital labour platforms and the platform economy do not simply affect pay and working conditions; they can reconfigure the entire architecture of a sector or profession. Yevgeny Morozov states that “Most platforms are parasitic: feeding off existing social and economic relations. They don’t produce anything on their own – they only rearrange bits and pieces developed by someone else”. Firat’s (2021) study on the impact of digital labour platforms on the working conditions of translators in Turkey finds that “working conditions have been rearranged and reorganized” for translation workers.
I posit that this parasitic relationship is symbiotic, whereby translation and digital labour platforms sustained by the labour of translators have become largely mutually dependent, but with the benefits moving in one direction only. The parasitic nature of this relationship is characterised in certain ways, some more specific to professions organised like translation and others more general to the platform economy:
1) Existing conditions in the translation sector make it an amenable host. Outsourcing and self-employment have long been common practice in this sector and much in-house translation work, in Western Europe at least, involves translation into and out of English, French and German only. International organisations also outsource most of their translation needs that do not involve a core of colonial legacy languages. One may question how many full-time permanent Kinyarwanda translators the International Criminal Tribunal for Rwanda hired, if any, in spite of the centrality of this language to its work. Translators working in-house for companies and organisations often do so on an outsourced contractual basis too.
Thus, individualisation, making the translator responsible for their own working conditions and terms of pay through self-employment status or similar arrangements, has long been standard. Many problems cited with working conditions in the platform economy are familiar to freelance translators: lack of employee status, remoteness and isolation, precarious working conditions, responsibility for own work tools and covering the costs of one’s holidays, sick leave, pensions, etc., usually incorporated into the fee negotiated by the translator and their client. A novel problem that has arisen in recent years with the growth of remote interpreting is the health and safety risks of such work and the need for suitable equipment to mitigate them.
Much outsourcing is handled by agencies, often run as local small to medium-size enterprises (SMEs). Low and falling rates of pay, particularly by larger agencies, have long been concerns. It is perhaps unsurprising that these large agencies were early adopters of the potential offered by digital platforms. The largest LSPs today have close ties to big tech firms, sometimes own other platforms, and have consolidated their position by buying out specialist and successful SMEs who offer better rates of pay to their translators. According to Lambert and Walker (2022): “the super-LSPs have a powerful marketing advantage in capturing clients and projects. Their vast size grants far greater purchasing power and leaves freelancers with a choice between LSP-mediated work at a lower rate and a higher-risk strategy of potentially intermittent direct-client work at higher rates”.
Small agencies and translators cannot compete with the power of data and algorithms. A lack of regulatory obstacles and low barriers to entry – potentially just the ability to speak more than one language and the necessary hardware and software – provide other easy access points. Translation, nonetheless, is highly-skilled work and demand in this field has long outstripped supply. Many translators hold some form of translation qualification or training. Rather than investing in education or promotion of the field, the situation has allowed tech solutions to flourish instead.
Digital labour platforms are just one type of platform translators have to contend with: machine translation provided as part of computer-aided translation (CAT) tools can demand more time from translators to edit and correct existing translations. CAT tools also allow surveillance. The growing use of machine translation post-editing (MTPE), where a human translator edits and untangles a computer-generated translation, is also more time- and labour-intensive. The data and algorithm-driven difference again ensures the translator is not paid for their time and labour.
The challenges of working with a human project manager at an agency are not the same as those posed by platform algorithms. In addition, existing payment issues such as fees for international payment transfers and free translation “tests” are compounded in the platform economy by new issues, such as unpaid time spent on platforms looking for jobs, the risk of wage and data theft, as well as wages being withheld due to subjective client ratings.
The result is a polarised profession that many qualified professionals choose to leave. The blame is placed on agencies and globalisation and there is a naïve assumption that the sole fracture line in the profession lies between professional (expensive) and unprofessional (cheap) translators.
2) The digital economy is part of digital capitalism. Manufacturing and sustaining crises is the lifeblood of capitalism. With the roots of the broader digital economy going back at least as far as the 1990s, the rise of platforms and apps came off the back of the 2008 global crisis. The threat to people’s homes and livelihoods, not to mention war and climate change, make platform work a necessity and not a flexible autonomous choice.
Although the roots and creeping dominance of digital labour platforms in the translation sphere have flourished over the past two decades, the reality only hit home for many following the 2020 pandemic: exploitative working conditions and poor pay were not some temporary glitch. During the pandemic, established translators, who like most other workers worldwide faced a downturn in work and income, were told to diversify the services they offer by LSPs and professional bodies, pushing many who had thus far evaded contact with digital labour platforms into their open arms.
At the same time, bilingual and polyglot individuals struggling to find work were encouraged by platforms to take on translation work: “Today, companies are leveraging bilinguals, trilinguals and polyglots to present digital materials across all mediums in various languages, allowing companies to expand their reach and position themselves as global enterprises. […] there is an increasing demand for their services, making the gig economy a lucrative and viable opportunity for anyone who is bilingual, trilingual or polyglot.”
According to one report: “The COVID-19 crisis has not only fuelled the demand for online translation and transcription services, but also increased worker supply on translation and transcription platforms due to the relatively low entry barriers for workers. The resulting heightened competition has raised pressure on workers in the sector to improve their productivity while maintaining or even lowering payment rates”.
The use of “colonial mechanisms of exploitation” (Casilli, 2017) in the platform economy cannot be ignored. The so-called refugee crisis, affecting more than 1% of the global population, and wars, have also provided fertile ground to push translation platforms onto a cheap labour force with no alternatives. The crisis in Venezuela since 2015, producing both refugees and cheap labour, has seen the country become a prime source of labour for digital platforms, including translation-related tasks.
3) The World Economic Forum finds “lack of definitional clarity” to be one of two issues holding back understanding of the platform economy. This lack of definition has allowed Uber and other location-based apps to toy with the employment status and rights of their workers, simply shifting the goalposts of vague meaning when threatened by regulatory or judicial oversight. The result is a situation worthy of Humpty Dumpty in Alice in Wonderland: “When I use a word, it means just what I choose it to mean — neither more nor less.”
LSPs and platforms like to refer to their translation workers using the nebulous term “linguist”, whereas to describe the actual work done, especially to clients, there is a whole plethora of meaningless terms to describe the same form of exploitation and hide the presence of the translator: gig work, cloud work, microwork, click work, crowd work, ghost work, all variations on the vague theme.
To confound matters, the Gigworker platform tries to confuse gig work with self-employment, listing translation and transcription as “some of the most popular types of gigs”: “At its core, the gig economy is the work model where individuals perform short-term tasks or projects for various employers rather than working full-time at a single company or organization”. Gig work mainly relates to location-based work but the everyday conflation with different types of work is real. Fumagelli et al. (2018) correct this to digital labour being described as “the labour-force of independent contractors who work on their own account and at their own risk for low wages and without social security, as in the case of many platform-based business models like Uber, Foodora or other work and logistic platforms”.
In 2022, Fairworks produced the first and only Translation & Transcription Platform Ratings, applying its ratings on cloud work platforms’ fair pay, conditions, contracts, management, and representation to nine platforms, which included industry big-hitters Lionbridge (and Gengo, owned by Lionbridge), Transperfect, Translated and SmartCat. The others, such as Rev and Scribie, are not necessarily closely associated with translation. Appen, one of the largest translation platforms in the world, ranked 8th by Nimdzi, instead features in its overall annual platform report, and is ranked 3/10 for fairness in 2023.
Fairworks defines cloud work as “remote work via online platforms that can be performed by workers irrespective of their location – as long as they have an internet connection”. Translated performed the best in the 2022 report, with Transperfect and SmartCat receiving a 0/10 rating. The report’s findings are interesting, particularly as concerns working conditions: “Platforms define the terms of exchange largely unliterally and often employ non-transparent algorithms to set prices or manage work allocation” and “The majority of surveyed workers on translation and transcription platforms regularly engage in unpaid tasks, such as checking or applying for jobs, taking unpaid qualification tests, curating their profile or communicating with clients and project managers. Unpaid labour on translation and transcription platforms contributes to worker precarity by reducing workers’ net hourly earnings.” While not all translators work through platforms, the top-down domination of digital labour platforms eventually affects working and pay conditions throughout the whole sector.
This report has drawbacks nonetheless: it fails to recognise the huge power imbalance between platforms on one side and workers and clients on the other, recommending that clients, who do not see the invisible translator in any case, advocate for better conditions, but also acknowledges “significant information asymmetries exist between platforms and their clients and workers, resulting in a power imbalance that ultimately affects working conditions for translators and transcribers”. In addition, 35% of the 213 translators surveyed for the study reporting “having a degree or an official qualification as a translator” is not so much an indication of low entry barriers to the profession as much as of the difficult professional and financial situation of many professionals in the current economic climate.
4) The remaining terms – microwork, click work, crowd work, ghost work – closely related to points 2) and 3) above reveal a layer of translator invisibility and exploitation that most people, including translators, are completely unaware of but without which our AI and platform-driven world could not operate: Digital or AI colonialism.
The two main translation players in this field are Appen and Lionbridge. Appen defines microwork as “One of the center points of technological innovation today is data. But, to gather and annotate that data takes thousands of hours of work. Work that can be done by anyone, anywhere in the world. The internet makes it possible to break up these tasks into thousands of micro-tasks which can then be distributed to workers around the globe. This is known as microwork”. The term was coined in 2011 by the CEO of Samasource (now Sama AI), a US-based company that “that seeks to alleviate poverty by outsourcing digital work from Walmart, eBay etc. to impoverished populations”. On the other hand, Jones (2021: 16) calls it the “creep of something like a refugee industrial complex”. Corporations are not the only ones involved; an important 2020 study by Mona Baker & Attila Piróth shows how not-for-profit Translators without Borders has used its work to benefit its corporate partners, like Microsoft.
Microwork “includes manually tagging photos and video, transcribing audio and categorising text, so when we use a search engine or speak to a voice assistant, the process feels effortless”. Platforms such as Amazon Mechanical Turk pay some workers in gift vouchers instead of cash and it is not unusual for platforms to exploit the existing precarious and informal working arrangements in the Global South. Data capture and extraction also mean that digital colonialism goes beyond working conditions.
Endorsed by the World Bank and the IMF, in partnership with big tech firms beneficiaries, since 2012, and seen as a way of helping to achieve the UN sustainable development goal of eradicating extreme poverty by 2030, microwork has been written into international development agreements, forcing this new form of colonialism onto populations and countries that lack the power to resist. Such work is not without its controversies. Earlier this year, it was revealed that Sama was paying Kenyan workers less than $2 to carry out work removing sexually explicit and violent content to sanitise ChatGPT. In addition, Google used microworkers from Figure Eight (later bought by Appen) to train CIA drone AI as part of Project Maven; workers had no idea of the end purpose.
Microwork is not just the future of work in the Global South but potentially the whole world. Since the pandemic, studies have shown an increase in such work in Europe, although there is no evidence to state that much of this work is translation-related. Nonetheless, almost all microwork platforms include translation tasks and it plays a large role in the boom of both digital platform work and the translation industry. Jones (2021: 41) states that “this repackaging of once prestigious, well-paid employment as ‘low-skill’ tasks displays in stark terms capital’s violent path through the ‘vocations’, turning professionals into proletariats”.
Microworkers, or ghost workers, are the invisible underpaid “humans in the loop” behind modern technologies and the huge powerful firms that own them. Some of the firms involved are some of the biggest LSPs in the world. The impact on the broader translation sector is inevitable. Take Appen, set up in Australia in 1996. With around 1100 staff, according to its latest annual report, and over 1 million “”flexible contributors” in more than 170 countries”, and worth almost $400 million, it has long been accused of non-payment and delayed payment by workers. In addition, over the past year, its profits have gone down “from above $38 per share in August 2022 to less than $4” in June 2023. Attracting huge investment thereafter in response to its plan to further develop AI and cut more costs, its plans and decisions as a heavyweight could affect those of other LSPs.
In addition, Appen is proposing to pay workers in stablecoins, a form of cryptocurrency, to deal with costly international bank transfers, following a study carried out a consultancy that has provided similar studies for the World Food Programme. The success of this move could also have implications for how other LSPs and agencies pay their translators.
Jones (2021: 55) describes the impact of microwork on translation in the following terms:
Building this degree of flexibility first involves the carving up of existing jobs and projects into short tasks. Take the role of translator. In theory, a great deal of basic translation work can now be done by deep learning algorithms, though many tasks, like translating poetry or fiction, require forms of cultural sensitivity that are not yet programmable. For projects requiring less nuance, a platform like Lionsbridge [sic] can help break down larger texts into sections for algorithms and shorter passages workers complete as small tasks. These might include: ‘categorization of topics in a conversation, determination of emotions behind a statement, classification of intents and identification of parts of speech’. Rather than hire a few skilled, full-time translators or speech professionals, with rights, a proper wage and access to a union, companies are able to rent a transient team of fifty anonymous workers to fulfil the same role.
Much of the impact of the platform economy on the translation sector has been by stealth but is not unanticipated. Both LSPs and big tech companies rely on a lack of public knowledge of the products and services they use every day. While rates of pay are an overriding, long-term concern in the profession, the reality shows that rather than a two-track profession, the translation sector is in fact fragmented in many places. With the platform economy continuing to evolve, there is no indication that there is not worse to come.
According to Altenried (2020): “Crowdwork and digital technology more generally, are part of an ongoing heterogenisation of global space constituting fragmented, overlapping and unstable cartographies and questioning stable categories such as North/South or centre/periphery.”
Jones (2021: 57) further states: “there is no reason why more white-collar work – such as bits of accountancy, finance, copy, translation and so on – could not be carved up into petty tasks, particularly as these roles are automated ever more heavily. Such a scenario threatens to transform increasing numbers of professionals into wage hunter gatherers”.
* Jones, Phil. 2021. Work Without the Worker: Labour in the Age of Platform Capitalism. Verso Books. https://www.versobooks.com/en-gb/products/2518-work-without-the-worker
Altenried, Moritz. 2020. The platform as factory: Crowdwork and the hidden labour behind artificial intelligence. Capital & Class, 44(2), 145–158.
Casilli, Antonio. 2017. Digital Labor Studies Go Global: Toward a Digital Decolonial Turn. International Journal of Communication. 11.
Fırat, Gökhan. 2021. “Uberization of Translation: Impacts on Working Conditions.” The Journal of Internationalization and Localization 8(1): 48-75.
Fumagalli, Andrea, Stefano Lucarelli, Elena Musolino, and Giulia Rocchi. 2018. “Digital Labour in the Platform Economy: The Case of Facebook” Sustainability 10, no. 6: 1757
Lambert, Joseph and Walker, Callum. 2022. ‘Because we’re worth it: disentangling freelance, translation, status, and rate-setting in the United Kingdom’, Translation Spaces, 11(2): 277–302.
On 6 July, I spoke at the Translab 4: Translation and Labour International Symposium organised at the University of Westminster in London. My 20-minute contribution focused on the relationship between translation and the platform economy. I have expanded on this in a series of 3 blogs, of which this is part II. Part I can be read here: https://onesmallwindow.wordpress.com/2023/07/25/the-parasitic-relationship-between-translation-and-the-platform-economy-i/
one small window... 
[…] On 6 July, I spoke at the Translab 4: Translation and Labour International Symposium organised at the University of Westminster in London. My 20-minute contribution focused on the relationship between translation and the platform economy. I have expanded on this in a series of 3 blogs, of which this is part III. Part I can be read here: https://onesmallwindow.wordpress.com/2023/07/25/the-parasitic-relationship-between-translation-and-the-platform-economy-i/ and Part II here: https://onesmallwindow.wordpress.com/2023/07/26/the-parasitic-relationship-between-translation-and-t… […]