| Robert |
Hi, I’m Robert Jelenic, Global Marketing Director at RWS Moravia, and I’m your host today on Globally Speaking. Today I’m joined by Stella Paris, the Head of Language Services at Translators without Borders. Stella is originally British but lives in Bergamo, one of the Italian cities hardest hit by the coronavirus outbreak. Translators without Borders has been on the very forefront of the fight against the infodemic. That is, the spread of dis- and misinformation on COVID-19-related topics all around the world. Stella will help us understand exactly what we mean by the infodemic and we’ll also explore some projects that the organization is working on. Then we’ll explore how industry partners and independent language professionals are helping the effort and even how you can get involved.
Stella, please, would you give listeners a little bit about yourself? |
| Stella |
Yes, sure. So, as you mentioned, I work for Translators without Borders. I’m currently Head of Language Services, which means I’m responsible for the part of our organization that manages all our language resources, whether those be our internal language resources or also our amazing community of over 30,000 linguists. And I’m also responsible for managing our language service partnerships with over a hundred non-profit organizations. |
| Robert |
Wonderful, thanks. I’m glad you mentioned the community and the partnerships because where I wanted to start today was with the so named infodemic and a little bit about how these community members and partners are helping TWB with the infodemic. Would you mind giving the audience a little bit about what the infodemic is? |
| Stella |
Absolutely. So, the infodemic has been a really fascinating phenomenon to watch and to witness. It seems a long time ago when everything really started to blow up, if you like, and the World Health Organization declared the COVID-19 outbreak as a pandemic. And accompanying that health pandemic there was also this phenomenon, global phenomenon of an infodemic that we believe is just as dangerous and spreading just as fast as the virus itself. And it was the Director General of the World Health Organization who highlighted this initially.
If we want to look at the definition of what an infodemic is, it’s a term that was coined during the SARS outbreak in 2003. And it means that there’s a surplus of information about a problem that is viewed as being a detriment to its solution. And so, this can take many forms, but the main two being misinformation and disinformation. |
| Robert |
Now as a layman, could you describe to me what the difference is between disinformation and misinformation? |
| Stella |
Yeah. Really, it’s quite a simple difference, and the difference lies in the intent of the person communicating the information. |
| Robert |
Ah, okay. |
| Stella |
So, misinformation is information that is unintentionally false. So, the person who is sharing that information may be doing so in good faith. They may believe that that information is gonna help somebody, or they may believe that it is true. Disinformation, on the other hand, is information that is intentionally false, and so it tends to be spread by individuals or groups who are wanting to create a situation of unrest or to actually damage, harm or disrupt the lives of others. |
| Robert |
Right. Is there a unique angle that TWB is taking to solve the problem? Or, I guess the bigger question is, how does language play a role in what you’re trying to solve specifically? |
| Stella |
Well, language plays a massive role because the reason why misinformation and disinformation spread so easily is due to a lack of information or a lack of understanding of key issues. So, if people are lacking information—reliable or trustworthy, fact-checked information in a language that they can understand—then they are much more susceptible to fake information, to disinformation and to rumours that may be circulating because they just don’t have that knowledge themselves.
And so, if we think about just our own contexts—and I’m going to imagine that the majority of your listeners are multilingual and indeed, if you’re listening to this podcast, I can assume that you’re also fluent in English. So, we have a massive advantage because, if I take myself, I speak English and Italian, and every day I read a British newspaper and an Italian newspaper. And then throughout the day, I read a lot of information from a lot of different sources from all around the world. And I then use that information to form my own opinions and to gain an understanding of different situations. But if you imagine that you speak a language that isn’t well-represented in the media and that isn’t easily available on the internet, then that’s where you start to have an issue. Because there just isn’t enough information for those people to access reliable information, if you like, coming from trustworthy sources.
And if you combine that with contexts where you may have a low level of trust in institutions and government, and you may also have multilingual contexts, then that becomes a recipe for disaster. And then if you want to look at that within the COVID-19 response, if you are in a context with a weak health system and you combine all those elements together, then you can see that very quickly the situation is going to escalate. |
| Robert |
Absolutely. I mean, I suppose it’s a bit of the sense of vacuum of reliable, verified information in the local language that really makes the environment ripe for mis- and disinformation to spread. A lot more than if you had, maybe, a critical mass of information in the native language that would outweigh or overwhelm, or at least give the users more access to something that they can reliably believe in. |
| Stella |
Yeah. And that’s exactly what we’ve been trying to do with some of our activities and actions. So, we have been working on social media monitoring. And the aim of that is to try and identify those rumours, but ultimately, we want to use that to identify the information gaps, because then that can help inform the communication strategies. And we’ve actually been invited to be a part of the global risk communication coordination group with other big global agencies such as the World Health Organization or the Red Cross or UNICEF, because they recognize the importance of language and language tools to help curb the spread. |
| Robert |
Fascinating, yeah. So, I mean, the questions are sort of piling up in my head. Two big ones that pop up are which sort of regions or which language groups are you concentrating on? And then as maybe a follow-up, what are you doing there? Could you share some interesting projects that you’re working on? |
| Stella |
In terms of languages, I mean, that’s always the million-dollar question, if you like, and especially with this crisis response. Translators without Borders, we’ve worked on a number of different crisis response situations over the last few years. This one is unique because it is truly global in nature. And that then leads to very difficult questions and decisions about what languages do we focus on.
We cannot cover all the languages of the world. So, what we’re trying to do is to prioritize those languages where we believe there is the greatest need. We’ve been looking at various different criteria and factors such as where the cases are increasing at any one time, the health systems of different countries—so, countries that have a weaker health system and also that may have weaker government—to try and make sure that the people who are living in those countries that do not have a government who is providing regular updates and has a defined communication strategy, we’re trying to focus on those areas and those languages primarily. |
| Robert |
Okay, right. So, I guess that means it’s not necessarily long-tail, marginalized languages. These could be mainstream languages, but just in places where there is a vacuum of verifiable, truthful information coming from official sources. |
| Stella |
Yes. And obviously the situation is changing very quickly. So, if we go back to February, which is when we first started to work on the COVID-19 response, we received a grant so that we could combat misinformation specifically to counter the infodemic.
And initially, we proposed and we worked on Asian languages, but that was because at the time, the outbreak was limited to Asia and nobody at that time really thought that it was going to go global as quickly as it did. And so of course, we mobilized this response with working in 12 Asian languages. And then very quickly as we progressed through this response, we realized that that didn’t really make sense anymore just to limit our actions to those 12 languages.
And so, then we started to take on more requests in other languages, whether they be also European languages, but also African languages. So, we had a whole range in there. And to date, I believe that we’ve translated into 97 languages so far on our COVID response. |
| Robert |
Wow. Yeah, that’s no small feat. Could you share with the listeners maybe, those who are contemplating getting involved, some of the kinds of projects that they’d be getting involved in? You mentioned social media listening in the past. You know, like what kind of projects are you working on that you can share with us? Especially the ones in which either industry partners or individual translators could be involved in. |
| Stella |
Yeah, sure. So, first of all, I mean, we’ve already received like, a huge amount of support from the language service industry and so we’re incredibly grateful for that. We have set up an emergency response fund and companies such as RWS Moravia, WeLocalize, Smartling have already stepped up and been incredible in terms of their generosity.
We have also been working with other language service providers who’ve been providing in-kind services, whether that be translation or also revision and quality assurance, as you said. And that is always, always very welcome for us because we are often working with our community-based model. I might tell you a little bit more about that in a minute.
So, in-kind sponsorships are incredibly helpful. And then obviously, our emergency response fund; that is also very helpful as it will just allow us to scale and to reach even more communities, vulnerable communities. In terms of other projects that we’ve been working on, we’re working on an interesting project called TICO-19, which is the translation initiative for COVID-19, and this is really a collaboration.
So, it’s combining ourselves but also respected academic institutions: the Carnegie Mellon University and Johns Hopkins University, and then the big tech players such as Amazon, Facebook, Google, Microsoft and then Translated and Appen from the language service industry. And the objective of this initiative is just to really translate data so that we can develop COVID-19-related information and improve the machine translation, the machine learning of that data in a wide range of languages. |
| Robert |
Interesting. So, really around building the corpus of data for the machine translation engine to try to reach these audiences faster and with more volumes of, you know, good information, as you said. |
| Stella |
Yes, absolutely. So, I mean, in a lot of the languages, there are already existing machine translation engines. For some of the other languages that we’re looking at, there is no machine translation or language technology currently available because of a lack of data in those languages. So, we’re really trying to look at this as a comprehensive project.
So, it’s covering all languages, but only focusing on COVID-19-related content, and the big tech companies, they are contributing to this. They’re working with their own translation vendors as well to get this standard set of standard corpus translated into, well, we have a list of almost 90 languages, I believe. But the aim is to get this content translated into as many languages as possible, just so that we will really have this data available and available and open to anybody.
The collaboration came together pretty quickly. A conversation started probably around March. So, this has happened very, very quickly. And I think this is just showing that the, let’s say maybe positive, impact, if we can, of the COVID-19 situation, because getting such a collaboration with so many different institutions and corporations to come together and to work on this initiative together in such a short space of time, I’m not sure that that would have been possible previously. So, we did already have existing relationships with a lot of these companies, and I know they already work together amongst themselves.
In this case, there was a common objective and really that willingness and the desire to make something of this. So, that’s quite exciting to see, and I think we’re gonna see more of that as well as we move forward. Because then, after we get this initial dataset published, then we’re going to also extend to other organizations such as TAUS who were already working on similar initiatives. |
| Robert |
There must have been just a drastic spike of need to get the right information in the right languages in the right hands in front of the right eyeballs. |
| Stella |
We have a page which is updated in real-time where you can see the number of words that we’re translating on the COVID-19 response globally across the organization and it’s still going up and up. So, it’s been quite interesting that it was quite a small amount in March, and then it went up in April, up in May and it’s still going up.
We do try when possible to focus on community-facing content and content that is specifically directed at health workers in this case. And so that is just the tip of the iceberg, if you like, when it comes to COVID-related content, because then there’s all the scientific research and all the other stuff.
And in general, the content that we see is just a teeny tiny fraction of the amount of COVID-19-related content that is flying around. And luckily, there has been this trend and lots of different community, translation community initiatives that have been popping up all around the world and have been taking matters into their own hands, if you like, in certain contexts.
So, if people have been seeing, “Okay, my government’s not translating information into languages of, maybe, immigrant populations in my country. So, I’m gonna do that myself, you know, and I’m gonna set up an initiative to do that.” So, we’ve been seeing a lot of this really spontaneous collaboration and initiatives popping up. And we developed something called the Community Translation Program so that we could allow such organizations to then use our platform and have access to our community as well. So, maybe to get a bit of a head start. |
| Robert |
Fascinating. And I guess not just the community but any of the assets you’re putting out into the world, right? So, as we build things like glossaries, those should be available for someone who, you know, for lack of a better word, wants to take it into their own hands if they think that the local authorities maybe aren’t doing as good a job as they could be. They’ve at least got a reliable source from TWB of, you know, what does this term actually mean? Or like, how do we write this term in this language? It’s been officially verified so that we’re not misusing terminology or things like that. So, I guess just putting those things out in the public is a massive and scalable solution to what you’re trying to do. |
| Stella |
Yeah. And that’s a very simple solution as well. So, the glossary that we have on our website right now, I think it’s in 23 languages.
You did mention the verification, and that is an interesting aspect because we as an organization are not really the authority on health terms, if you like. So, when we do the glossary ourselves, of course we’re getting it approved, but we cannot verify those terms. So, we’ve been trying to get feedback from the World Health Organization, The International Federation of the Red Cross, so that they can verify the terms because they are the ones that know best. |
| Robert |
So Stella, you mentioned earlier on projects around social media listening and TICO-19, building up a machine translation corpus to power that in a faster method. Are there any other technologically powered projects you can share with us right now that TWB is working on? |
| Stella |
Yeah. This crisis has just really focused our attention on how multilingual chatbots can be a solution right now. Because what we’re seeing is that within the humanitarian sector, we’re having to come up with innovative ways to be able to communicate with at-risk communities, and often now at a distance because physically, you know, we had to actually evacuate our own in-country programs from Nigeria and the Democratic Republic of Congo and many other aid organizations have had to do the same. So, that leaves us with a very concrete problem that we need to address, which is how can we really make sure that we’re getting the right information to vulnerable communities and how can we make sure we’re communicating with them and also listening to them, because that’s a really important aspect.
It’s not just about blasting information out and hoping people understand, and that they’ll just get on with everything. You really also need to hear what people have to say in order to understand what their problems are or where the lack of information is so that we can identify the information gaps. So, the multilingual chatbots is what we really see as being a key solution to this problem.
So, the World Health Organization, they did launch a health alert platform on WhatsApp and Facebook Messenger providing information. And also, UNICEF came up with a similar solution. But what we’ve been seeing is that these solutions, they’re kind of using what you might call a conventional menu interface. So, you actually do have to go and you have to select a topic and then it’s kind of prompting you.
So, the solution we’ve been looking at is a way to incorporate natural language processing—so, using natural language so that we can also better understand and respond to questions. And going back to the topic of misinformation, this is just gonna be really key in understanding where those information gaps are so that then we can feed the bot, if you like, with the right information. |
| Robert |
Ah right. So, I suppose when you prompt the user with pre-configured choices or drop downs or buttons to lead them to a certain path, you may not be exposing the implicit misinformation that they’re carrying with them. So, I suppose in a chatbot setting, allowing for that natural language interface with a two-way communication, there’s probably a lot more chance for you to glean some insight into what people may have heard or seen elsewhere that may not be correct. So, that sounds like a way that you can actually extract some insights from that conversation at scale, which I think is the real key here given the size of the pandemic. |
| Stella |
Yeah, no, you hit the nail really on the head there. Yeah, because that that’s really our model. We have built a prototype so far in four languages: Hausa, Comorian, Congolese, Swahili, because we do have a local presence in Nigeria and the Democratic Republic of Congo. So, we’re building on the partnerships that we already have there.
And, yes, what we want to do is to create this kind of circular model so that we start off with a relatively simple bot, but then based on the questions that are asked, we are then curating the content and enriching the content and we’re combining the World Health Organization content but also content from local organizations, country-level organizations. Maybe also data about local statistics.
So, if we hear that somebody keeps asking, “Well, how many cases are there in Goma,” then we can upload that information. So, we’re really starting to create a conversation. And as you said, that way we can gather insights and we can understand the concerns of the people, and we can pass that information then on to our partner organizations so that they can better understand the situation and then develop more solutions to address those issues. |
| Robert |
Interesting. And that’s under development at the moment and will be released at some point in those countries? |
| Stella |
Yes, yes, we hope so. |
| Robert |
Amazing. You mentioned a bit earlier the community-based model that you work with. What does that mean? How do you work with groups that you call a community? What is a community in the TWB world, and can you tell us a bit more about that? |
| Stella |
Yeah. So, yeah, the TWB community, we always say it’s just an amazing and awesome group of over 30,000 individuals. So, in terms of who they are, it’s incredibly difficult to say because we just have this amazing diversity in terms of the geographies and the languages, but also the backgrounds. And I think that’s the interesting thing: we have a whole range of people who sign up to contribute to our mission.
And what we always say is the thing that unites everybody in our community is just the desire to get information into local languages, so really, to share TWB’s mission. So, within the community we have students, we have maybe newly qualified translators who are taking their first steps. We have maybe more senior translators. We also have translators coming from large LSPs who have actually been just helping us lately. They’ve been sharing the information for signups with their vendor databases.
And then we also have a lot of other professionals who might not be linguists in their daily lives, but they might be lawyers or doctors or engineers who just want to help people who speak their language. And so, we’re working with a diverse background and of course, we’re also working in contexts where there may not be translators—as in, there isn’t a translator profession per se for certain languages.
So, yeah. So, managing this community is a massive job as you can imagine. And what we’ve been working on a lot over the past year or two is just to really try and understand better who our community are and understand why they contribute to TWB and also what we can do to help them. So, we recognize, for example, that students and junior translators, they come to TWB.
Yes, they come to help. Of course, everybody does. But they also want to gain experience, and we can provide that experience because they can work on the content of international organizations and they can start to build up a portfolio, if you like. At the same time, we have more senior translators who are an amazing resource because then they can help, and they can mentor the junior translators. And we’re hoping to work on mentoring more in the future.
Beyond the community, we also have, for example, a role called the Language Associate. So, a Language Associate is like a language lead. These language leads can provide quality assurance. And so, what we’re doing a lot of at the moment is, we have community-based translation plus the review or quality assurance by a language lead, if we can manage that because of time constraints or also budget restraints as well.
Especially on the COVID-19 response where the turnaround times have been really tight and also there is a very definite need to ensure that we are providing quality, accurate information, the support, again, from the language service industry is really useful and amazing because if we can get in-kind support also to help with that quality assurance, just to make sure that we’re getting the information out there quickly and we are confident that it is accurate, then that’s more vital, let’s say, than ever before. |
| Robert |
Absolutely. It sounds like no matter where you get involved, you’ll make impact right away. And I also think it’s a chance for the language industry because it’s certainly one that is impacted by all this and certainly one that can help. This is our chance to step up and do something about it.
So, yeah, it was really great to talk to you today to learn a bit more about what you and the rest of Translators without Borders is doing and how we can all get involved. |
| Stella |
Thank you so much. And thank you for inviting me here and for giving space to talk about our work. And as always, we’re just really grateful for the support that we receive, and we just look forward to collaborating more in the future. |
| Robert |
Great. Thanks a lot for joining us, Stella. |
| Stella |
Thank you. |
