To prepare for the machine learning evolution, translators need to be human

Machine learning is unstoppable. As a professional translator and interpreter, I am not scared. Technology has been making our lives easier for decades, offering tools that help us polish our skills to perfection. Technology has been helping us learn, teach and produce content, and yet fearmongering and lack of information have been instilling dread among us all.

It is unsurprising. Technology has been changing people’s jobs since the agrarian revolution and, according to Oxford Economics, National Bureau of Economic Research, and Bureau of Labor Statistics, 47% of today’s jobs will be gone in 10 years. Uncertainty is scary after all, and we are all subject to it. My job, like every other job, will have to change and adapt. I will have to learn and adjust.

Translators, often forgotten by those reading our translated texts, eclipsed by the written word and subject to the industry’s predatory practices are already being affected by technology such as neural machine translation.

I personally think that one of the key tasks to tackle during the transition involves the education of our clients. This is not a new concept in the industry. Our jobs have been underestimated for decades by those who fail to see their complexities. In fact, only those with a superficial understanding of language could ever assume that what machines will offer is the whole package. More than ever before, both vendors and linguists need to work together so that those investing in language services can understand that machines make the process faster and simpler, but true value can only be drawn from human involvement in the process.

I also think that agencies will need to value proofreading and editing (as well as what is generally referred to as post-editing, which seems to be where change is leading us), and allocate fair rates. I dream with translation and interpreting agencies which do not prey on their freelancers, but rather work with them as a team, in constant communication, with utmost respect for their abilities. Editing a translation carried out by a machine is not a job that can be done lightly. Good rates would be paying for our creativity, our communicative and social skills, our empathy.

We tend to forget that our job is an intrinsically human one: we are the ones who allow for human connections beyond language barriers. Subtext, tone and linguistic nuances are human. Power dynamics, tentativeness and illocutionary forces go beyond the written words, and involve an engagement of background cultures and experiences. It is these finer details which have been shaping relationships between people and, therefore, business, education, economics, politics… the whole world. Machines cannot, at this stage, be trusted with these incredibly specific skills, especially when, often, two humans will not agree on the full meaning behind a single sentence.

There are plenty of formulaic texts which can already be automatically translated with surprising accuracy, but this is not always the case. Language is alive. It is and will always be creative, like humanity. Language evolves because we evolve. If anything, the speed of change has been accelerating as a consequence of world-wide connectivity, which in turn generates a constant need for us to keep up.

In order to deal with this speed, we have a multiplicity of programs. Technology and software developers have made our jobs easier in a lot of ways. We can now count on software for consistency, terminology management, research and professional development. The potential for growth excites, rather than intimidates me.

Of course, it can be a nightmare. A multiplicity of programs, having to work with them all, management software that has to be incorporated to deal with each one of the agencies we work with, powercuts, bugs and errors. The road is steep, but, ultimately, I want software that will work for me. That is what I hope for as a result of this machine learning evolution. As a layperson, I ask those in tech to involve us in the process in order to ensure this. I am optimistic because I can see how, every year, software design continues to improve, with more and more developers realising that collaboration with end users is crucial and needs to be a part of the process from the get-go.

I am a layperson when it comes to technology, but I work hard to be as involved and curious as possible because I believe that we need professionals who regard computers as their allies. We need professionals who can exploit everything that technology has to offer, so we can be the best linguists possible.

In 2013, I sat for my final undergraduate exams using pen and paper. As a professional translator, I have not carried out a single translation using either of these tools. I am originally from Argentina, where university education is publicly funded. A public university education is a luxury, but, because universities at home are immense public institutions, everyone involved has to face inordinate amounts of paperwork, bureaucracy and politics which slow down the pace of change and modernisation. My experience with translation software as a new graduate was almost inexistent, which is unacceptable after five years studying full time. Educational institutions need to work towards keeping up with the speed of change too.

The other side of my educational experience involves its holistic, all-encompassing approach focused on language learning. I studied English and Spanish grammar in depth, as well as the ways in which they can be compared and contrasted. Together with what I learnt about sociolinguistics, this knowledge has helped me understand the subtleties of communication and how it shapes human interaction. Many translation and interpreting programs take students’ knowledge and familiarity with their languages for granted, when it is this consciousness and an understanding of the languages’ dynamic and the relationship between the two which will ultimately help these graduates thrive alongside machines.

Learning a new language helps us question the ways in which our own languages work. We need to study our mother tongues. Choices which seem natural, obvious and unquestionable have in fact helped shape entire cultures, countries, people and communities. In order to see difference, we need to be able to question our own ways first. Only then we can analyse the contrast and work through it constructively. In a world of misunderstanding, our role remains as relevant as ever. As linguists and communicators, we bridge gaps in understanding and reconcile differences. These are the gaps that machines will not be able to fill.

For the machine learning evolution, we need to focus on doing what we do best: being human. We need to sharpen those sociolinguistic skills, and ensure we understand the factors that shape meaning beyond strictly textual aspects. We need to be as human as possible and let machines be machines.