Why Translation Tools For Refugees Matters
For a brief period in 1999 I was in-charge of tent accommodation in Cegrane Refugee Camp in Macedonia (FYROM), which was the largest single camp in the Kosovo crisis, housing over 43,000 refugees. It was clear to me that the refugees who had fled the fighting lacked a lot of things, but within a few months in the camp their most basic needs were met. Or so I thought.
Within days of the conflict ending on 11 June, many refugees started to spontaneously return to Kosovo. Those that remained mostly lacked the means to return or perhaps were hoping for resettlement to a third-country. My job was to move families, who were living in parts of the camp left empty by the returnees, to new tents located closer to camp services and other families. When moving the families I quickly realized that I needed to be better at communicating with the refugees – in Albanian - about why they needed to move and the process for doing so.
My Kosovar staff were great, but their English was not the best. Without adequate translators and interpreters, the refugees and I could not communicate effectively. The ensuing communication barriers undoubtedly contributed to increased tension and stress for the people I was there to help. In the 17 years since then the need to improve how aid workers communicate with affected communities, including refugees, has not diminished. Until recently, communicating with communities was often an underfunded and neglected part of a response. Thanks to sustained advocacy work by the likes of the CDAC Network, Translators without Borders, and others, this is slowly starting to change.
In the following conversation, Imogen Mathers and Rebecca Petras discuss why it is important for refugees to be able to access information in their own languages. Imogen is producer/assistant editor at SciDev.Net; Rebecca is the Deputy Director of Translators Without Borders.
This edited transcript is used with the kind permission of SciDev.Net, “the world’s leading source of reliable and authoritative news, views and analysis on information about science and technology for global development.” The full podcast and accompanying article can be accessed at: Tech tools break down language barriers in emergencies
Imogen Mathers: For those who don't speak Farsi or Pashto, the messages we just played will be pretty meaningless, barring one or two global words like ‘please’ or ‘computer.’ In many countries prone to natural disasters or conflict, and where multiple languages are spoken, these kinds of linguistic barriers can be seriously dangerous. When emergencies strike information to assist people is often broadcast in the majority language, leaving those who don't speak it in the dark.
For refugees to poor to access to information and translation services can be perilous even life threatening. The messages we played just now include information for refugees arriving in Europe. How to register, where to photocopy documents and get medical help, where to get a cup of tea. These are simple vital messages, yet this kind of multilingual translation service is rare in emergencies and rarer still in the European refugee crisis. I spoke to the NGO Translators Without Borders to find out what they've been doing to address this problem and how technology can help.
Rebecca Petras: My name is Rebecca Petras, I'm the deputy director of Translators Without Borders.
Imogen Mathers: Please can you tell us about the Words of Relief project that you've recently set up?
Rebecca Petras: Yes, Words of Relief is a project that is specifically focused on crisis relief, and how translation and language can make a difference during a crisis situation. We developed Words of Relief as a pilot program really looking at what tools in translation and language we could use during crises. Our initial work in that was done in Kenya, and then we've taken it beyond Kenya to really apply it to various crises around the world in the last two years. The pilot project was about an eighteen month project in which we looked at three different areas that we really wanted to focus on. The first looking at what content to proactively be available before the crisis strikes. So in the case of Kenya we were looking at twelve categories of crisis and looking for the right kind of content that is helpful to affected population. Putting that into local language especially into Swahili and also into Somali.
Imogen Mathers: Somali because a large refugee population in Kenya is Somali?
Rebecca Petras: Yes, we did other local languages as well in Kenya. There are about twelve main languages in Kenya. The Swahili is a lingua franca for Kenya with Somali not having a lot of crossover, a lot of the Somalian population doesn't speak Swahili.
Imogen Mathers: What kind of information were those messages including?
Rebecca Petras: The kind of information we care about is basic information to help the affected population in the crisis, and really that's what we focused on simple messages. Public service announcements, audio, and basic health information depending on what the crisis was. So we were looking at everything from earthquakes to tsunami to drought to epidemic, and doing content in those areas that we could then translate and make openly available when a crisis strikes.
Imogen Mathers: So it's things like where people can access healthcare, food, water?
Rebecca Petras: Yes, it's what to do in an aftershock with an earthquake. What is an after shock and how to protect yourself? How to boil water if there's no clean water. We did a lot of work around various diseases including malaria outbreak, cholera, Ebola, basic basic information on what it looks like what to do.
Imogen Mathers: What role does technology play in this? What's the actual infrastructure that you're developing?
Rebecca Petras: Technology plays two main roles. The first is as a work flow process in order to improve rapid response. So what I was talking about before with proactive information that's readily available. But the fact of the matter is early on in the crisis, especially in the first seventy two hours, you're really just trying to get a feel of what's going on where it's going on and getting as much information from the affected population to the aid organization. So going the other direction. In that sense we really want to bring technology to bear by figuring out ways to connect using data aggregators who are getting information straight from the affected population via social media, going through them and making sure they are only focused on the one percent or two percent of messages that matter. Taking all of the english messages coming in translating them very quickly and providing it back with geo location so that people can be found, this is for search and rescue. The best example of what we did with that work was in Nepal.
The second place that technology is really important, it's really quite interesting, is with machine translation. We generally know it as Google translate or Bing translate. And for the high languages or the world languages such as French Spanish English, these tools have worked really well. You'll get like an 85 percent correct rate. But in a lot of crisis languages machine translation either doesn't exist or it's really really poor. Poor in the sense that it's actually detrimental to translation process. One of our big goals is to make machine translation useful in those languages and in those cases where we often see crises strike. The goal being if you have that tool you really in many cases can help the aid worker and the affected population communicate better, many without our help without our human translation needs.
We then have taken it on ... Actually working with Microsoft, they gave us an inkind donation to strengthen those engines. We took the Swahili engine from a 1 percent correct, what they call a blue score which is basically how accurate it is or how helpful it is. We took it from one percent to over fifty percent. Our translator then with a thousand words, instead of taking four to five hours to translate it can do it in about twenty to twenty five minutes. Which is a huge improvement.
Imogen Mathers: You mentioned that you develop a database of messages preemptively, before a crisis might strike. What extent do you think yo will then be adjusting this response according to the demands of the situation? I imagine it's going to be very contact specific as well.
Rebecca Petras: Absolutely. A lot of content is specific to that specific situation, but a lot is not. Cholera is a great example. Cholera hits so many places in so many different languages we have really strong content in our repository in a lot of local languages that absolutely for the affected population dealing with cholera. Because we had it already done in Swahili we were able to do it very quickly in Colundi when cholera struck the refugee camp for the Balindi refugees in Tanzania. You can either grab it and say right here it is in Arabic, here it is in Haitian Creole, here it is in Colundi that's what we're going for. There's no question that in many cases you have to modify or do new content, but we have shown where content available can then be grabbed then used very quickly in another part of the world.
Imogen Mathers: Does radio come into your work at all?
Rebecca Petras: Absolutely. Radio is key certainly in parts of the world where literacy rates are low we very much want to work with radio. One of the main things we try to do is provide public service announcements that are then available for local radio stations. We always will do that through partners. Inter News is always a big partner, we did some work in Kenya in Somalia I believe. We used radio in West Africa as well with the Ebola crisis. In the European refugee crisis radio is not used a lot, however audio is being used a lot. Registration centers, used on buses, and in those cases we're actually doing the audio.
Imogen Mathers: The Ebola outbreak ad now with the Zika crisis women have been affected in quite specific ways. In Ebola they were often the caregivers and now with Zika obviously with issues around pregnancy. How did gender considerations come into your work?
Rebecca Petras: It's a big issue, it comes in a number of ways. Number one typically if there's an english speaker in the community it's not a woman, so they tend to be the least informed. They get their information second or third or fourth hand. So to really make sure that you're talking to the woman in a way hat she understand is tricky, and you need to understand is she literate. Is she literate in her local language or not? How do we reach her because she may not have that mobile phone, she may not have you know been going to the community meetings. So we do try to tailor messaging to be very simple. We try to make sure there is iconography as well, and making sure that audio and video is used where possible.
Imogen Mathers: Also with children, when you're dealing with a whole other set of communication issues and ethics issues. How do you go about responding to the needs of kids?
Rebecca Petras: It's an excellent question. It's not easy. With girls in particular you still have the literacy issues so we try to get around that with audio as well. One of the things we do is we try to always work with protection teams, and we try to make sure their information is easily understood and in the right language. That is a real big issue in the migration crisis, we really try to coordinate with an organization who are focused on protection, but to be totally honest I can't say that that has been resolved. That we clearly understand how best to do it, but it is very much top of mind.
This transcript is used with the kind permission of SciDev.Net, “the world’s leading source of reliable and authoritative news, views and analysis on information about science and technology for global development.”
To learn more about what is #CommIsAid watch this great video from the CDAC Network.
