Neural machine translation has revolutionised the translation and localisation industry. It’s time for translators – and clients – to adapt to this new landscape and evolve with it.
We are in a time where neural machine translation (NMT) cannot be ignored. LSPs, translators, and even clients are adopting this technology to speed up the translation process and to offer services that would not have been possible in the past. But to really reap the benefits of machine translation, we have to learn to adapt and acquire new skills.
MT solutions
In general, there are two main types of MT systems: On the one hand, we have commercial solutions, such as Google Translate. However, the available language combinations and fields can be limited, and often they cannot take into account special needs or requirements. In many cases, these solutions cannot always do the job and do not deliver the necessary quality.
On the other hand, there are companies that offer customisable solutions that can be specifically trained to meet individual requirements such as specific fields, language combinations, customer-specific data, etc.
And what about the translators?
Technical advances are important, they help us progress, but many times we are led to believe that these solutions can be integrated smoothly. But what about the human factor? Wouldn’t it be better to combine machines and humans?
Over the past few years, NMT has become a true turning point that requires the acquisition of new skills to operate it. NMT has meant that the roles and functions of translators, project managers, and others have had to evolve. For example, translators are no longer “only” translators, but rather they have become language experts in a much broader sense. In many cases, they contribute to the training and evaluation of MT engines. First and foremost, they have had to embrace post-editing and acquire new IT skills in relation to MT.
NMT can be good, and even great in some cases. Post-editing can achieve significant productivity boosts. Still, some translators are not very keen on it and might even react quite emotionally to this topic. They sometimes fear that it would be too much effort to work on the output. However, neural engines perform much better than their predecessors in terms of fluency, at least in some domains. Of course, they are not perfect, and there are some downsides such as terminological inconsistency, additions and omissions of content. But they can handle more complex content and more language combinations (albeit some better than others) than before.
Other times, translators might feel unsettled because they fear they will eventually be replaced by machines or feel a certain pressure on their rates and their productivity, as turnaround times are becoming shorter.
Conclusion
Perhaps it is time to redefine services, procedures and expectations. Translators ought to get out of their comfort zone, widen their skill set, and develop a positive attitude towards MT and adopt it to increase their productivity (augmented translation). We all should be open to new technologies and opportunities. Clients, in turn, should make an informed use of technology and not just assume that MT works for all kinds of content or that it will always save time and money.
I don’t believe that translators will be replaced by new technologies such as MT, but rather by other translators that adopt them and evolve with them. MT should be seen more like a productivity tool to assist translators as part of augmented translation workflows, and as an addition for some fields and sectors in which the only other alternative is no translation due to the amount of content. Whether we embrace it or not, MT is here to stay.
__
Image source: Gerd Altmann | Pixabay
