Translators Against the Machine #1: Good agency gone bad, by Laura Vuillemin

This article is by Laura Vuillemin, an English and Portuguese to French translator who raises awareness about the rights of translators in the face of AI and machine translation. You can learn more about her work at lvtranslations.com

This article is part of our ongoing Translators Against the Machine open call, where we call on fellow translators and other language professionals to tell us about their working lives. Our aim is to gather experiences and stories to paint a multifaceted, three-dimensional picture of what it’s really like to work in the translation industry. If you’re interested in submitting, you can read more about the open call here.

***

They were what you would call a perfect client. For years, they regularly sent me stimulating and varied work. I was in direct contact with a meticulous project manager who was a translator himself, understood the importance of queries, and praised the quality of my work. Payments were never late. What could possibly have gone wrong?

Sadly, the company this article is about progressively surrendered, like so many others, to the siren song of machine translation (MT). Initially “only” with certain projects deemed straightforward enough, but then gradually with all of them: training modules, press releases, subtitling scripts, highly sensitive user manuals, blog articles, white papers, and so on. As their own vendor manager told me a few months ago, 99% of their work is now done by MT. In her words, “only 1% of our projects involve translation and revision by a human. All others involve post-editing.”

This meant that lower rates were unilaterally imposed on freelancers, who started receiving emails that read as follows: “We would like to inform you that we have changed our processes, and now all project requests sent to you will be post-editing. This is our post-editing rate.” Take it or leave it, basically.

According to the company’s managers and executives, who, for the most part, had never worked as translators, the machine’s automated output was good enough to justify cutting linguists’ rates, often by half. Unsurprisingly to the translators, the output was poor, and they ended up fixing mistakes they would never have made themselves.

The freelancers who refused the agency’s poorly paid “post-editing by default” policy were blacklisted and not contacted again. I was one of them. The agency’s decision was made regardless of the quality I had provided, as if everything I had done in the past mattered less than saving a few cents. One day, projects simply stopped reaching my inbox.

This dramatic shift in the company’s strategy could have been “just” sad, but the problem is that it was also highly unethical. Remember the aforementioned 99% of projects? Well, they include projects from historic clients, some of them quite high profile, who have been collaborating with the company for years, if not decades. It is highly unlikely that they were consulted on the decision to throw their content into an MT engine, meaning these clients are still being charged for human translation, while post-editing is done behind their back.

There was no way I would agree to indirectly contribute to this, all while earning barely enough to cover my costs. That said, I refused to do machine translation post-editing (MTPE) at that time, and still do today, for two other crucial reasons.

First of all, the pervasive use of this technology is causing linguistic impoverishment, from both a syntactical and lexical point of view. Automated translations are based on statistics, with no care for discursive coherence. The output is produced by using the most frequent wording, and ignores all other variants, producing target texts that constantly sound the same. The horribly repetitive, literal, uninspiring and lackluster speech produced by machines has nothing to do with idiomatic language.

Secondly, translators are getting no compensation whatsoever, while doing MTPE, for helping to train and refine automated engines. Their translations are reused – or indeed stolen – ad infinitum, which poses serious questions in terms of copyright. The least reputable companies that aim to completely replace translators with machines are missing an essential point: if they ever manage to do so, it will only be because of all the corrections and improvements made by these same translators. Why should they benefit from it, while linguists don’t even get the crumbs from their table?

Writing this piece is no easy task for me. The client whose practices I’m denouncing here used to be an organization that I held in high esteem. How on earth could they have sunk so low that I don’t recognize them anymore? The end of our collaboration obviously had an impact on my income, but that’s not what prevails in my mind when I think about what a waste all this is. It also makes me question the logic of it all. When you have reliable translators you can trust, why would you choose to bypass them, ignoring their (valid) arguments and putting quality at risk? Above all, why would you shoot yourself in the foot by turning into a low-cost language service provider that no longer stands out from the competition, from whom potential clients and freelancers alike are now running away?

As demoralizing as this whole situation might look, my intention here is not to paint a desperate picture of the translation industry. I do believe that better days will come, provided we raise awareness and stop keeping quiet while unacceptable and harmful conditions are forced on us. What matters to me is defending translators’ rights and exposing shoddy practices, so that dependable professionals (and agencies) are no longer sidelined in favor of companies that shamelessly put profit before anything else.

Produced under a Peer Production License

– Written by Laura Vuillemin
– Edited by Guerrilla Media Collective
– Lead image by Harrison Haines, on Pexels