Translators without Borders, empowered by MateCat

In the spring of 2014, an outbreak of the Ebola virus began to spread quickly throughout West Africa, in the region where Guinea, Liberia and Sierra Leone share borders. In the first weeks of the outbreak, cases were limited to outlying villages, but by early summer major cities, such as Conakry (pop. 2 million) and Monrovia (pop. 1 million), were reporting cases.

The violent symptoms and high mortality rate aside, what makes the Ebola virus particularly awful is the fact that it can spread through contact with an infected person’s bodily fluids — even sweat. So, family members burying loved ones who had succumbed to the disease quickly fell ill themselves.

In an epidemic like this, panic and misinformation can spread as quickly as the virus itself. Getting accurate and trusted information to affected populations — even to instruct them not to touch a loved one who has become infected — is very much a matter of life and death.

That’s easier said than done in this part of the world. Though the countries officially speak major languages — English in Liberia and Sierra Leone, French in Guinea — local populations actually speak any one of dozens of indigenous languages. Instructions written in French might be illegible to a Guinean who only communicates in Guerzé.

Translators Without Borders (TWB) responded to the Ebola crisis in late 2014 and early 2015, translating rapidly into many local languages as well as languages throughout Africa where fear of the disease ran high. The non-profit was able to work with a large number of professional and community translators to get the work done for responders desperately trying to communicate to the local people.

This is the kind of challenge Translators without Borders faces every day. Its team of 20,000-plus professional and community volunteer translators must mobilize to get information to people on the ground in a crisis where every second counts.

As part of the organization’s efforts to respond quickly and eliminate language barriers, the team upgraded its processes earlier this year, building out its Kató translation platform. Kató was built on top of MateCat with some important customizations developed by the TWB team, led by Mirko Plitt, TWB’s head of technology.

The results have been promising: Kató has proven effective at onboarding translators who aren’t very familiar with CAT tools, as well as quickly parsing projects and delegating translation tasks. We are incredibly proud to assist in the important work TWB is doing, so we reached out to Mirko for an update about how the Kató platform has helped translators work.

How did you find out about MateCat?
I’ve known about MateCat for several years, not least through my role as translation technology expert for the European Commission, which funded some of the development of MateCat through different programs. I had also used it for post-editing experiments and translation technology classes I teach, where the fact that it is open-source and web-focused comes in very handy. And finally, the MateCat team’s promotion of their CAT tool has definitely made it hard to go unnoticed.

Why did you choose to use MateCat to build the TWB Kató platform?
Although open-source technology may appear as a natural fit for a non-profit organization such as Translators without Borders, my initial plans were actually to select one of the other various strong commercial systems on the market. In the end, there were two or three factors that drove us to choose MateCat instead, both of which are ultimately linked to the TWB mission, which is of course different to that of LSPs or localization departments, and to which makers of commercial CAT tools mainly cater.

Most importantly, we felt that MateCat was particularly accessible to translators and project officers with little to no CAT tool experience, which is probably more prevalent in the TWB work context than at a typical LSP. The potential cost of the loss of sophistication was not relevant in the case of TWB because most of the more advanced features found in commercial systems — and which can of course make a big difference in a business setting — are not used in our processes.

TWB went through a major change in our “model” last year, specifically the relationship with our non-profit organization partners who use the services of our volunteer translators or other services we provide. This change required a fairly deep integration of our CAT tool with our non-profit-oriented customer relationship management system as well as with other systems that are central to our work, such as the TWB Workspace (i.e. our instance of the ProZ.com Translation Center) and more recently the Rosetta Foundation’s Trommons platform.

It was easier for us to tackle the type of integration work we had to do with an open-source system.

What have been the main advantages of building on top of MateCat?
Working with an open-source solution offers a particular advantage when working with organizations that prefer approaches that can empower local communities. Giving out free, widely used commercial CAT tool licenses can be a great model in some situations. However, providing local linguists with a tool stack that they can adapt to their community’s needs can be more suitable in others.

Another advantage for us has been MateCat’s web focus and its light project management model. MateCat’s teamwork functions are also becoming more and more important to us, such as the sharing of comments among translators, reviewers and project managers, or the review and quality assurance interface.

How does TWB use Kató?
We use Kató to deliver a wide variety of translated content. There is a lot of healthcare and nutrition-related content in there, but there are also a lot of fast-turnaround translations such as posters, brochures or daily news articles with critical information for populations affected by crises.

Kató is TWB’s newly improved and enhanced translation environment, which is essentially our customized MateCat instance. To this, we have added other additional features such as the integration of recorded spoken translations for low-literacy audiences or for use in crisis situations when information needs to be broadcast in audio format. We use the MateCat translation and review interface (which, unlike other parts of the system, we have not modified) as well as the API to integrate with the processes of our non-profit partners.

How did your processes change with Kató?
Since using this improved system, along with some other changes to TWB’s operations, we feel that the workflow of our volunteer translators has been made more efficient. The introduction of Kató means that translators no longer download files to be translated offline, using a CAT tool of their choice, or sometimes no CAT tool, and then upload the completed translated files. Instead, translators, now receive links to Kató and can directly translate online, using the dedicated translation memory we assign either manually or automatically.

This is especially good when working with teams of translators. Previously, we had to manually split documents into several files, and each translator would download and translate their respective files. In terms of process change, the ability to split translations and assign reviews across a team is maybe the biggest single gain we have made thanks to Kató.
What was the first campaign where Kató made a difference?
Our pilot project was a large healthcare translation spanning several months, and it also involved training a group of translators in Guinea, West Africa. In addition to working with the team in Guinea, we also had volunteers in other parts of the world, supporting the project as reviewers and mentors. The team translated about one million words of content developed by the Open University specifically for health workers in sub-Saharan Africa.

The Guinea team had no CAT experience, and it was particularly encouraging to see the benefits of MateCat’s relatively flat learning curve. One constraint we had not foreseen worked out in favor of a web-based solution like MateCat. We provided trainees with Linux laptops, as shipping other configurations would have been impossible within the project timeframe. As it turns out, Linux would have ruled out the commercial desktop CAT tool we had initially considered for this project.

I can safely say that we certainly would not have been able to complete this important project without a modern translation tool such as MateCat.

How many translators are working on Kató?
The original TWB community of translators had over 4,000 volunteers. However, that number has now more than tripled since TWB’s recent merger with The Rosetta Foundation. Trommons, the open-source Rosetta translation platform, is now also being integrated with Kató so that over the next few months we expect the number of Kató users to continue to grow quickly and to approach or even exceed 20,000 professional and community translators.

What are the top languages?
The main languages we translate into are French, Spanish, English and Arabic. But maybe more significant is the “long tail” of languages we have translated into, as the MateCat team knows because you generously keep adding support for languages we request! We’ve translated more than 200 language pairs in Kató this year alone, and I would suspect that even many of our readers in the language industry wouldn’t have heard the names of some of them.

What are the main goals you have achieved with Kató?
In general terms, I would say that Kató has laid the ground for scaling our operations in a variety of ways while professionalizing our services to non-profit partners by following industry best practices, such as around quality assurance.

Kató also helps us protect the effectiveness of the crucial work done by our volunteers — achieved by the use of a centralized translation memory which addresses the very real challenge of content redundancy within and even across non-profit organizations. In the past, volunteers would sometimes have translated content that had already been translated in a different project by another volunteer — instead of translating other potentially life-saving content waiting in the queue.

Another important goal was to best support those of our translators who don’t have access to CAT tools, and on the other end, also the vast majority of our partner non-profits who don’t have translation tools. Kató has therefore really become a central part of how we fulfill our mission as Translators without Borders.