MACHINE TRANSLATION: IT'S NOT THE END OF THE WORLD AS WE KNOW IT
(AND YOU SHOULD FEEL FINE)
by Diego Cresceri
The first time I heard about machine translation was around 10 years ago. I remember my former colleagues and I were stunned and skeptical about the idea of using software to translate automatically, then patching up the holes in a translation which would generally require massive edits (best case scenario), or would have to be rewritten from scratch.
The customer was one of the biggest global software companies, but they were absolutely not ready (neither were we) to manage localization projects using automated translation: their content was not MT-ready, they did not have a customized engine, their expectations in terms of time and quality were just too high. And of course, there was no training provided. Still, the team managed to deliver 400K words in a little over a month, and back then it was probably the biggest project the company had delivered in such a short space of time.
Since then, machine translation has come a long way, and neural machine translation is now able to produce impressive results for many language combinations and topics. Localization buyers have learned how to improve the translation generated by working on the source content, customizing their engines and training post-editors to produce the quality they want. Far from being reserved to IT companies, MT is now the solution of choice of many customers in different industries, including automotive, hospitality, and e-commerce, to mention just a few.
Although progress has been made, there is still a fair amount of skepticism about machine translation as an opportunity rather than a threat, and many are still convinced this is just another way for big companies to squeeze their vendors and, ultimately, the translators working on their projects.
I am not by any means saying that saving money is not one of the reasons why MT is now so widespread, but it’s certainly not the only one and probably not the main one. Below are two more reasons:
Machine translation speed varies depending on different factors, including language combination, engine size and document format. However, there is no denying that machine translation is faster than human, with some engines being able to translate up to 1 billion words per day. Even if you consider pre-editing, engine training and post-editing, this is still an impressive speed. This is why companies using machine translation have a major competitive advantage in terms of time-to-market, since they are able to produce more localized content in less time.
The introduction of new content types
Year after year, the volume of content produced grows at an impressive pace. As new content types keep growing, including patent, after sales and user generated content, companies must find a compromise between the need to localize a huge quantity of material and their limited resources. Machine translation is often a way for these companies to localize content that, otherwise, would remain stuck on their hard drives. For this reason, in many instances, machine translation is not a replacement product, but a different service altogether, for different content types.
This point is key, so allow me to repeat myself: in many instances, machine translation is not a replacement product, but a different service altogether, for different content types.
Every day, still, I hear people complaining about how machine translation is damaging the market, stealing jobs from humans, slashing rates and generally bringing about the end of the world as we know it. Most of the time, however, this is because they have absolutely no clue what it really is.
At Creative Words, we have been offering machine translation and machine translation post-editing since we opened in 2016, and this service now accounts for more than half of our business, and counting. This is why we are constantly enlarging our team of post-editors and offer ongoing post-editing training to our existing translators, too.
Due to the lack of barriers and prejudice, we find that recent graduates and young professionals in general are usually keener to accept machine translation post-editing jobs. This is not because, as many imply, they are desperate to enter the market and willing to accept a miserable rate. On the contrary, they are flexible enough to see that machine translation can be an opportunity and that rate alone is not the only yardstick when you evaluate a job offer.
What are the advantages of working on this kind of project for a newbie in the translation industry?
We all know how difficult it is when you’re starting out. When I was looking for my first job in the translation industry, I sent hundreds of CVs, and I got just one reply (which of course I will never forget). Agencies often look for professionals with 3-5 years of experience (in line with industry certifications), making it hard for newbies to find their way. Machine translation post-editing jobs can provide an opportunity for translators who do not yet have a well-established portfolio of experience to offer. This is particularly true for projects where light post-editing is requested, since they do not require extensive experience and are ideal for fresh graduates. Training will of course be needed, but these projects are a great opportunity for them to take their first steps on the market and issue their first invoices. And since one thing leads to another, cooperation can then move to other types of projects (it happens every day with our suppliers).
Although machine translation and post-editing is a common practice in the localization industry, few universities include MT and MTPE in their learning paths (at least in Italy). Based on a recent poll of our Facebook followers, around 67% of respondents (people that have graduated within the last 2 years) had not heard about it before they got in touch with us, and would therefore not be in a position to offer this service. Since demand for post-editing is constantly increasing, this could mean a missed opportunity.
More often than should be the case, translators-to-be only have a vague idea about what a CAT tool is and how they can use this technology to support their work. Again, there’s a big gap between academia and actually working in the industry, here. One of the reasons is the high licensing cost for most tools. However, there are plenty of alternatives right now to the desktop, license-based CAT tools and many tech companies would love the opportunity to collaborate with our universities. This is why I fear the true reason is to be found in ministerial programs that are stuck in the early ‘00s. For many Bachelor students, learning about a CAT tool only happens if they are lucky enough to come across a particularly active professor who is willing to integrate some classes on computer assisted translation into her/his program out of her/his own personal initiative.
Most CAT tools have a similar layout and the same basic features (concordance, terminology, and so on) therefore getting to learn how to use one of them effectively is a great way to start, possibly with no investment.
Applying for a job, going through the submission process, raising the right questions in the right way, adhering to specific instructions, getting feedback and responding to it without over-reacting: not easy tasks if you have little or no experience. And again, not something you learn when you study.
We invest a lot of time trying to fill this gap by advising linguists on how they can improve their applications, tackle a translation test or improve their communication skills. While of course giving our feedback on their deliverables, so that they can improve.
This post is getting way longer than I had imagined. Hopefully I did not go off topic too much and I managed to express my point of view and stimulate some discussion on this delicate issue.
I am convinced that machine translation offers an opportunity from many angles (not only for newbies, by the way), and that those who think it’s damaging the market should at least try and consider different perspectives as well.
Some might be upset and totally disagree, still, I would love to hear from them as well, so feel free to share and comment below!
Diego Cresceri - Founder and CEO of Creative Words, he does not deny his past but never looks back. An absolute lover of languages, he's an incurable optimist and cannot wait to see what the future holds.
"If you talk to a man in a language he understands, that goes to his head. If you talk to him in his language, that goes to his heart." Nelson Mandela