A world of machine translation? Part III


The future of the professional translator

Faced with such encouraging prospects, what does the future bring for translators? Will we be replaced by robots or will we choke up the bottlenecks of automation? According to the study by Frey and Osborne quoted at the beginning of this series, translation is currently at a level of medium risk, with the same likelihood of automation as the work of manual packing workers, mechanical engineering technicians and land surveyors.

In this respect, some authors believe that machine translation will demote many translators to the status of post-editors, entrusted with reviewing and correcting the results of machine translation before publication, and they argue that this is not such a bad thing. Others, like Dorothy Kenny, qualify these predictions and argue that it is important to respect issues relating to intellectual property, while pointing out the paradoxical situation in which, in the struggle between machine translator and human translator, the most mechanical task is reserved for the latter. Nataly Kelly says that many translators are watching warily, or even with hostility, the intrusion of machine translation systems. However, the internet is full of resigned or favourable comments on the use of machine translators as professional translation tools.

In this last post in the series A world of machine translation? we look at what can happen in the different specialities within professional translation if machine translation systems are eventually implemented in a widespread way.

Sworn and legal translation

Sworn translation has been, at least in Spain, a refuge of reliable work during years of acute economic crisis between 2008 and 2014, a result of the business internationalization process and of the chilling exodus of our young people, the good and the bad of apparent recovery. What is going to happen in the future? Will we be able to continue certifying translations with a certain degree of independence? Or will we be expendable?

Automation does not leave the professions connected with legal and financial services unscathed, since complex algorithms are being developed that can process, extract and summarise the information contained in the legal instruments. The lawyer is not in danger, but their assistant is. Why? Because the lawyer makes use of creative and social intelligence in their work; the assistant, on the other hand, compiles and summarises the information, an easily automated task.

Can legal texts be machine translated? Up to a point, yes. When there is a program that can identify the lexical, phraseological and syntactic elements that make up the legal jargon of a language and find their closest, but not always exact, match in another legal system, most of the translation work will be taken care of. It will need to be checked by a legal translator or by a lawyer with linguistic skills. A possible obstacle to automation could perhaps be the deliberate lack of clarity of legal language, a feature that is also being simplified, at least in the English-speaking world, with the plain legal language movement initiative, which calls for the use of clear and understandable legal language.

The same cannot be said for the certification that accompanies a sworn translation. It is highly likely that over the coming decades there will still be a body of sworn translators, the only guarantors of the truthfulness and accuracy of a translation, just like there will be still be notaries and judges.

Website translation and localisation

The production of multilingual content is vital for the survival and visibility of websites, social media, forums, games, etc., where quantity is more important than quality in terms of texts, provided they comply with web style rules.

I would go so far as to predict that website localisation is the field in which machine translation techniques with human post-edition will be more likely to take root. Hordes of new translators, recently graduated from faculties of translation and interpreting, will end up in this flooded, poorly paid and undervalued market, while they are unable to find other work to do.

Advertising and commercial translation

Advertising, like poetry, has to separate itself from the well-trodden path to suggest and condense a lot of content in few words. I think it highly unlikely that a software program will ever be able to translate novel concepts using formulas that avoid the cliché. Translating publicity and advertisements will be done by advertising agents or very creative professional translators, who should demand a high price for the finely-tuned services of their human brain.

A certain type of commercial message could possibly be translated by machine followed by human post-editing, especially in multilingual mass campaigns like e-mail marketing.

Technical and scientific translation

Their similarity to formal texts means that many technical texts can be machine translated, almost with no human intervention, provided they have been written in accordance with certain guidelines. Simplified Technical English was invented precisely for that purpose. It consists of using brief sentences, structuring and using coherent and unambiguous terminology, avoiding the use of idiomatic expressions, repeating words without using deictics, avoiding the use of the gerund and writing instructions as simply as possible: the perfect recipe for machine translation. You don’t have to be a fortune-teller to speculate that user manuals, for example, will be translated like this with automatic quality controls and no post-editing.

Scientific texts for publication in specialist journals will be translated by researchers from the field or by highly specialised human translators. It is possible that some work will be pre-translated by machine, in the knowledge that the human translator will have to make many changes to produce an acceptable final text. This will demand in-depth knowledge of the subject being discussed, so translators working in this field will have to be extremely well-read.

Old-style translators will still be needed in the human sciences, where they can put their knowledge of the speciality and their stylistic skills to the test. At least for the time being, the humanities seem to be pretty much at odds with artificial intelligence and could offer a shot in the arm for the ablest translators on the edge of extinction.

Literary and editorial translation

I have deliberately left until last the archetypal translation that is beyond the reach of machine translation tools, the consolation that is always used to emphasise how irreplaceable we human translators are. We refuse to accept that a program could be capable of making a decent attempt at tackling the complex and perpetual challenge of literary translation, the last proud stand against the robot invasion. Those who have tasted it, know.

However, emboldened by achievements in other fields, some researchers go to great lengths to find the formula for storming this bastion as well. As they probe the formal aspects that confer identity on literary discourse, they disregard other factors that are no less important. The creative act of literature shows up the limitations of artificial intelligence.

My view is that no, they will not achieve this with true literature. If, as argued by neurobiologist Rafael Yuste, father of the BRAIN project, artificial intelligence has very little in common with the emerging properties of the human brain, automation can go far, but not to the same place as our creativity. So, it will not be machine translation systems that deliver the coup de grace to poor battered literary translation.

On a hopeful note

However machine translation systems evolve, one thing seems clear: the best way of digging our own professional grave is to fill both the internet and parallel translation corpora with poor, repetitive, inaccurate, literal and disjointed translations. If, on the contrary, we translate like humans, with creative intelligence, originality and attention to detail, and we specialise in fields less likely to be automated, maybe we can carve ourselves out a separate and persistent space, as well as doing our bit towards preventing the decline of culture in an increasingly automated world.

© Marta Pino Moreno.

Share now