Signage language: helping Rohingya refugees find their way

Bangla: সাইনবোর্ডের ভাষা: ক‍্যাম্পের ভেতরে রোহিঙ্গা শরণার্থীদের…রাস্তা খুঁজে পেতে সাহায‍্য করার জন‍্য

Signage in Rohingya Refugee camps feature photo Here is a question we asked Rohingya refugees in Bangladesh:

“Oney ki kemfor modotor ghoror sainbudhgin buzonne?”

Having some trouble understanding what this question means?

Well, you’re not alone. The majority of refugees we asked, over 80%, would also have trouble answering this question, but for a different reason.

We’ll revisit this question again in a moment. First, it is important to note the levels of illiteracy amongst Rohingya refugees. The low levels stem partly from the fact that Rohingya is a verbal language, with very limited use of it in written form in any alphabet. Even with English, the most popular written language amongst the community, only 31% of those we spoke to recently can read it. Their ability is mostly at a basic level, with almost half of those reading only numbers or recognizing letters. This makes communicating information using signs very challenging. Information must be provided in forms that are understandable to those who need it. But what if this illiteracy also applies to western forms of graphics as well as written languages? Where arrow symbols, emojis, and pictograms are as hard to interpret as French or Japanese to Rohingya refugees? How then do we create visual signs to show people the way without using established graphic standards and words?

What was the question?

The question put to you earlier – “Oney ki kemfor modotor ghoror sainbudhgin buzonne?” – translates to “Do you understand the signage for services in your camp?” The answer was a resounding “no!” for 65% of those surveyed (261 out of 404).

Finding your way… through the camps (or not)

This is the problem Translators without Borders was asked to look at within the sprawling refugee camps of Cox’s Bazar in Bangladesh. Since 2017 these camps are home to more than 850,000 Rohingya refugees. They also accommodate more than 130 national and international organizations and agencies offering assistance to the Rohingya community. Meeting the needs of a city-sized population of close to a million people is no easy feat. Like all “city” populations, the residents of these refugee camps need a range of services to meet their daily needs. Locating these services can be a challenge. There are no smartphones with Google Maps or street maps. Residents are not allowed to access the internet, nor are there detailed street directories or maps of the camps. So, maybe they should just follow the signs? Not if you do not understand these either.

Services such as medical clinics, food distribution centers and information hubs are scattered throughout tens of thousands of makeshift homes and buildings. Each must be found by navigating the maze of thousands of roads and alleys which snake through the camps. Making your way from point A to point B is not always straightforward, including when it comes to critical and even lifesaving services such as hospitals. While the camps are littered with signs, these come in a variety of format, style, color, and language combinations, many of which lead to confusion. The camp management staff who work in the camps had observed something problematic: many camp residents were getting lost. They saw a need for clear directional signage to guide residents to their destinations and, as it turns out, so did the community.

What did we do?

In January 2019, we set out to find out what was going on with signage in the camps. Our goal was to work with the community to develop signs they understand. The research focused on developing signs for the following key services and facilities:

health centers/hospitals/clinics
information hubs
women safe spaces
child-friendly spaces
nutrition centers
food distribution centers

The research aims to influence site management agencies to adopt a consolidated approach to the design of signage, based on community-identified needs and preferences. The research included observational field visits, pre- and post-design focus group discussions (FGDs), and a discussion group with camp managers and the site management sector working group. We tested comprehension of sign prototypes, and conducted a pre-pilot baseline study.

The FGDs explored the community’s color, content, and language preferences and their knowledge of forms, shapes, and logos. They also explored their understanding of time and distance, and their relevant cultural sensitivities around imagery. Throughout the research process, we worked with graphic designers to develop improved directional signage the community understands.

What do the numbers tell us about the need for better signage?

The Rohingya community confirmed the need for improved signage in the camps at various stages of the research. In January 2020 we asked 404 camp residents how often they used existing signs to locate services and facilities in the camps. We found that

45% answered “never”
12% answered “rarely”
15% answered “occasionally”
6% answered “some of the time”
6% answered “often”
16% answered “all of the time”.

We also asked the respondents if they had faced difficulties reading/following existing signage for services in the camp. We found that:

60% answered “yes”
5% reported knowing others who had experienced difficulties
89% indicated that better signage would make it easier for them to locate services in the camps
91% said that better signage would increase the likelihood of them using services.

In response to the same question during comprehension testing with 179 community members in October 2019, 97% responded the same way. During that survey, over 99% (178 out of 179) reported that better signage would make it easier or more convenient for them to access services.

The signs of the time

Our team observed and photographed a variety of signage formats and designs in the camps for the six service types targeted. The majority of signs provide information in text only, predominantly in English, sometimes accompanied by Bangla and/or Burmese translations. Our team immediately found some obvious trends:

English is the most popular language
Organization logos take prominence (way too much- some signs even contained 6 large logos)
Arrows give directions
Approximately 80% of the signs were in English and contained no icons or images (aside from the organization logos), even though only 30% of refugees can read them
Where there was directional signage, there was little follow-through on the directions at key intersections and crossroads, making the trail hard to follow
Icons or images are not used to signify specific services
There is little consistency in signage, including correlations between directional signage and the signs on the actual facilities.

Our observations in four camps confirmed most of the signs are not designed in a way that communicates information to the Rohingya community, especially those with low or no literacy.

What did the research tell us?

To develop our sign prototypes researchers from TWB worked with site management staff, members of the Rohingya community, and graphic designers. Through the focus group discussions and other consultations, we identified a variety of community preferences about how signs should be designed.

Color

Specific colors are often associated with specific organizations (e.g. blue for the United Nations and pink for BRAC).
For those who cannot read, color is often the best or only way of interpreting signage, especially when there are no pictorial aides and so at least some color is still preferred.
The community indicated a preference for designs with contrast – light-colored text/graphic content on dark backgrounds increases readability.
During final prototype testing, 100% of participants (179) indicated a preference for signs with a colored border around the central picture.

Where and when

Arrows were not popular with focus group participants.
Participants preferred an image of a pointing finger to indicate direction.
In the final testing of the sign prototype, 93% of all respondents correctly determined the direction using the image of a pointing finger.
A finger pointing down was also preferred by 76% of participants to indicate arrival at a destination, as opposed to two open hands.
Communicating the number of minutes to arrive at a destination proved challenging – all variations tested poorly.
In line with participant literacy levels, 54% of respondents in final prototype testing could identify the time required to reach the destination.

Language

Not surprisingly, text/script is useful to those who can read.

Burmese and English are the most widely understood written languages.
The overwhelming majority of community members cannot read Bangla. There is also opposition from authorities to including Bangla in signs.

Format

format image

Unpopular image designs;
participants preferred realistic drawings

Realistic drawings of people using the service were by far the most popular image style.
83% of people consulted during prototype testing preferred signs that combined drawings/cartoons and text.
Less than 3% preferred signage with text (letters/script) only.
Photographs were not popular (89.4% did not prefer photos).
Simplified images (or emoji styles) of people caused confusion, with a large number of participants associating these with ghosts.
Ninety percent of respondents in final prototype testing reported looking at the picture to understand the meaning of the sign.

image designs

Finally, a sign of progress!

Women Friendly Space: one of our final sign prototypes

In response, our graphic designers working with TWB researchers suggested this sign. It is one of the six signs developed based on the preferences and needs of the Rohingya community living in the camps. The design incorporates community-tested pictorial communication (life-like diagram, pointing finger), as well as information in Burmese and English for those who can read these languages. Although this sign appears simple, it is specially designed for one community who, like all communities, have complex and specific communication needs.

All community members consulted during the testing of the final prototype of this sign said that it would be helpful (97% “very helpful”, 3% “helpful”) if signage in this format was used in their camps. Preparations for pilot testing of signs for the six services in four camps is under way, with findings from the pilot test expected in mid-2020.

Here’s to helping Rohingya refugees finally find their way through the camps!

Written by Peter Squires, Evidence and Impact Officer for the Rohingya Response, Translators without Borders

Countries with lower literacy levels need different COVID-19 communication strategies

People have a right to access the information they need during the COVID-19 pandemic. But the format and language of that information need to evolve as COVID-19 spreads to nations with lower literacy rates and more vulnerable groups of people.

covid-19 literacy rates communication strategies

The information should be easy for people to find, understand and use. It’s unwise to assume that written formats are always the most efficient way to convey information. As the disease rapidly expands into countries with lower rates of literacy, organizations involved in the response need to shift focus from written information to developing significantly more pictorial, audio, and video content.

That is the best way to ensure that older people, women, and other vulnerable people in those countries have the best chance of understanding lifesaving information.

It’s also a necessary adjustment where infection control limits in-person community engagement. Social media, SMS services, call centers, television, and radio will be essential communication channels. Formats need to diversify accordingly if the message is to get across.

Literacy dynamics are rapidly changing

COVID-19 is now rapidly spreading in countries with lower literacy rates. The average literacy rate in countries with confirmed COVID-19 cases on February 19 was 94%. One month later it was 89%.

The highest rates of change in new COVID-19 cases being recorded are predominantly in countries with lower literacy rates. Between March 16 and 22, the 15 countries with the highest percentage change of new COVID-19 cases had an average literacy rate of 85%. These include countries like Cambodia, Cameroon, Côte d’Ivoire, DRC, Nigeria, Tanzania, and Togo. All these countries saw increases in confirmed cases of at least 900% during this seven-day period.

Women’s literacy rates are often lower than men’s

In countries where UNESCO measures literacy, average literacy rates are 6% higher for men than for women. For example in Yemen, 73% of men and only 35% of women above the age of 15 can read or write a basic sentence about their life, a difference of 38%. The gender difference is also stark in Pakistan (25%), DRC (23%) and Mali (20%).

The map below highlights the gender difference in adult literacy in individual countries. Orange shading indicates countries where male literacy rates are higher than female literacy rates. Blue shading indicates the few countries where female literacy rates are higher than male literacy rates.

Older people often have lower literacy rates than people under 65

In many countries, older people are less likely to be able to read than younger adults. This limits their ability to access written information on COVID-19.

The average elderly literacy rate in countries UNESCO reports literacy data for, is 65%. UNESCO defines elderly people as those aged 65 or older. In countries with documented literacy rates from the same year, people aged between 15 and 64 have an average literacy rate 19% higher than people 65 years or older. The difference is greatest in Libya (63%), Timor-Leste (53%), Cabo Verde (50%), and Iran (49%).

Use data to design more inclusive communication strategies

To design effective COVID-19 communication strategies, responders need reliable data about language and literacy. As part of our COVID-19 response, we are making the necessary data openly available.

This is part of a Translators without Borders initiative to help make targeted information strategies more data driven. Language and literacy maps and datasets exist for DRC, Guatemala, Malawi, Mozambique, Nigeria, Pakistan, the Philippines, Ukraine, and Zambia.

Along with these existing maps and the interactive global literacy map above, we are also scaling up our efforts to release more subnational language and literacy data for countries affected by the COVID-19 pandemic. This week we released national and sub-national data for Thailand. We will release more datasets and data visualizations over the next few weeks and months, so stay tuned to our COVID-19 webpage or the Humanitarian Data Exchange for updates.

We derived most of those datasets from historical census data, typically available down to the Admin 2 (district or county) level. Such data is most useful when it is analyzed alongside up-to-date information on language and communication needs. To help us with our ongoing language data initiative, we urge organizations to include four simple language questions in needs assessments and surveys related to COVID-19.

Make content available in multiple formats

Organizations responding to the pandemic should use improved data to develop communication strategies that are geared to the needs of the target population. Preparedness is a critical component of this. Organizations should develop content in as many formats as possible, recognizing that pictorial, audio, and video content is easier to access and absorb for many people. Additionally, older people often benefit from content that is easier to read. This requires incorporating design considerations such as larger fonts and good contrast. Plain-language principles also offer a useful model for creating clear and concise written and verbal content. The WHO proposes several key principles for improving understanding of health content.

In the rapidly evolving context of the COVID-19 response, organizations should complement written information with other formats. This is vital to ensure information is both believed and understood. We need to do this early to ensure people living in places with lower literacy levels don’t receive information too late to make a difference.

Written by Eric DeLuca, Monitoring, Evaluation, and Learning Manager, Translators without Borders

The project is funded by the H2H Fund, a funding mechanism for H2H Network members. The fund is a rapid funding vehicle for network members responding to humanitarian crises.

In the Democratic Republic of Congo:

Communicating in the languages of affected people is a priority for the latest Ebola response plan, and beyond

On 2 March, the authorities in the Democratic Republic of Congo announced that the last Ebola patient had been discharged from a treatment center. The epidemic isn’t over yet. But, after 18 months during which more than 3,400 people have been infected and over 2,250 died, the relief is palpable. Looking ahead, the Congolese government and its humanitarian partners turn their attention to implementing lessons from this 10th Ebola outbreak. In a country where more than 200 languages are spoken, prioritizing communication in the languages of affected people is one key lesson to help address the next emergency faster. The latest Ebola strategic response plan (SRP 4.1) points the way.

DRC Ebola response plan

The languages of affected people are finally a priority

The plan highlights the importance of improving risk communication and community engagement by using the languages and the formats preferred by people at risk. This includes developing communication tools and feedback mechanisms in appropriate languages, formats, and channels. The plan also emphasizes the need to equip health communicators to relay accurate information in local languages and with culturally acceptable wording.

For the first time since the beginning of this outbreak, the SRP mentions these issues. This is a key advance in adopting insights highlighted by health professionals, anthropologists, and communication specialists. It addresses three key factors TWB identified as critical to the effectiveness of Ebola-related communication: the languages that responders use; the content that responders deliver; and the way responders deliver that content. It also acknowledges the importance of feedback gathered from affected people by linking it to follow-up actions. People continue to have concerns and questions around Ebola and response efforts. Their concerns must be heard and their questions answered as the current outbreak draws to a close.

This is an important lesson that matters beyond Ebola

In multilingual DRC, to help people protect themselves responders need to listen, understand, and provide information and services in the languages of those at risk. Improving communication cannot alone guarantee better outcomes. But unless language is built into risk communication and community engagement strategies, response teams are unlikely to be effective.

Ebola DRC response plan

Recent actions by the risk communication and community engagement working group provide a case in point. Rumors and confusion have impeded efforts to contain the outbreak. So the group developed a multilingual tool to address the 25 most frequently asked questions. They collected the questions through the response-wide community feedback mechanism. The group members jointly drafted answers, and TWB supported with plain language editing to ensure accuracy and clarity. The group involved Ebola survivors to ensure the wording did not stigmatize them. Questions and answers were then translated into local languages for the widest possible reach and understanding. This tool equips responders to prevent the spread of misinformation and keep people safe.

Health professionals, social researchers, communication experts, and affected people worked together to provide and disseminate accurate, understandable information. This should be standard practice in mitigating the consequences of this outbreak, preparing for future health emergencies, and addressing wider humanitarian needs.

It is high time to turn evidence into action

The Congolese government and its humanitarian partners have a crucial role to play in implementing the latest response plan. And it seems they finally intend to give affected people’s languages and communication preferences the attention they deserve. TWB will work closely with those who are committed to a more language-aware approach. By proactively developing field teams’ capacity and resources, we can lift the language barriers to effective and accountable risk communication and community engagement.

Written by Mia Marzotto, Senior Advocacy Officer and Laure Venier, Community Engagement Program Coordinator for DRC, Translators without Borders.

Valérie travels the world and translates

Translators improve lives by translating potentially lifesaving information into languages spoken by vulnerable individuals. Those who volunteer as part of the Translators without Borders (TWB) Community have a range of experiences and skills. They share our vision of a world where knowledge knows no language barriers. We are grateful for all our translators, and we love sharing their stories.

Valérie Thirkettle is a multi-talented translator who has worked with TWB since 2018 and has donated almost 550,000 words of life-saving information. Her dedication and motivation to take on new projects and the care she puts into her translations make it an absolute pleasure to collaborate. Valérie is a lawyer who spent the majority of her career working for a prestigious intergovernmental organization dedicated to the exploration of space. Recently, she retired to pursue her passion for translation.

Valérie travels and translates — “How I feel when I sit down to face a big revision task” – Valérie.

A flexible working life

An avid traveler who divides her time between the Netherlands and South Africa, she enjoys the flexibility of TWB’s internet-based system. It gives her the chance to enjoy her other pleasures, studying literary translation, spending time with family and friends, golfing and enjoying nature, particularly in her beloved Africa. All the while, wherever she goes she can feed what she calls her “translation addiction.”

Valérie in Africa — Valérie enjoys the natural surroundings of Africa.

“I was attracted by TWB’s technology focus. I discovered how much language matters in humanitarian settings, so I hope my contribution can help people. And that it can improve the advocacy efforts of the organizations I translate for.”

Her ability to infuse her multi-sector knowledge into her translation work allows her to work on a number of different projects. “I am a trained lawyer and I have worked in international legal subjects and HR subjects. I like to make myself useful with the skills I have and contribute to the causes that resonate with me, and on a volunteer basis.”

Valerie keeps in contact with TWB’s Language Services Team by email. She is celebrated as a central, fun member of the community. The team recalls sharing many laughs with Valerie. With her varied experience, Valérie has seen the funny side of translation and mistranslation. She told us a story about a translation she once reviewed in which she noticed the section to sign and “date” the form mistakenly read “rendez vous d’amour.” “I loved it,” laughed Valerie, “filling in forms suddenly turned into something really exciting!”

Education for everyone

One of her favorite projects with TWB involved the revision and final linguistic sign-off of the Communicating with Disaster Affected Communities (CDAC) Network’s How-to Guide to Collective Communication and Community Engagement. This is essential for teaching better communication strategies on the ground. It helps inform people about their rights and situations in languages they understand.

Translators can often become emotionally involved in a project. When working with Street Child, for instance, Valérie says,

“I felt a strong resonance with the task, and, like with a good novel, the end came too early!”

Children in Bangladesh — Children learning in school, Bangladesh.

In fact, projects that assist young people tend to stand out for Valérie. Her time working with Think Equal also left an impactful and memorable mark. Think Equal has developed an early years education program for social and emotional learning. It was a large project in which Valérie took care of the entire revision. It included revising French versions of the program, an extensive set of books, lesson plans, and teaching materials. “The size and spread of this project made it complex, but an opportunity to develop new organizational skills for my translations.”

Overall, her translation experience has taught Valérie to appreciate the varied skills of other translators. She comments on how they build on one another’s strengths to deliver great work. She’s become increasingly involved in revising tasks and has embarked on qualifications in revising and proofreading. “My work with TWB gives me a great opportunity for continuous learning.”

One of her tips for other Kató translators is to “pay attention to the glossaries and be as consistent as possible with the terminology you use.” Valérie points out that you’re able to ask project managers for feedback throughout the process. “And of course, keep claiming more tasks, the humanitarian sector needs all the language help it can get!”

Get involved with the TWB translator community.

Written by Danielle Moore, Communications Officer for Translators without Borders. Interview responses by Valérie Thirkettle, Translator for Translators without Borders.

Language data fills a critical gap for humanitarians

Until now, humanitarians have not had access to data about the languages people speak. But a series of open-source language datasets is about to improve how we communicate with communities in crisis. Eric DeLuca and William Low explain how a seemingly simple question drove an innovative solution.

“Do you know what languages these new migrants speak?”

Lucia, an aid worker based in Italy, asked this seemingly simple question to researchers from Translators without Borders in 2017. Her organization was providing rapid assistance to migrants as they arrived at the port in Sicily. Lucia and her colleagues were struggling to provide appropriate language support. They often lacked interpreters who spoke the right languages and they asked migrants to fill out forms in languages that the migrants didn’t understand.

Unfortunately, there wasn’t a simple answer to Lucia’s question. In the six months prior to our conversation with Lucia, Italy registered migrants from 21 different countries. Even when we knew that people came from a particular region in one of these countries, there was no simple way to know what language they were likely to speak.

The problem wasn’t exclusive to the European refugee response. Translators without Borders partners with organizations around the world which struggle with a similar lack of basic language data.

Where is the data?

As we searched various linguistic and humanitarian resources, we were convinced that we were missing something. Surely there was a global language map? Or at least language data for individual countries?

The more we looked, the more we discovered how much we didn’t know. The language data that does exist is often protected by restrictive copyrights or locked behind paywalls. Languages are often visualized as discrete polygons or specific points on a map, which seems at odds with the messy spatial dynamics that we experience in the real world.

In short, language data isn’t accessible, or easily verifiable, or in a format that humanitarians can readily use.

We are releasing language datasets for nine countries

Today we launch the first openly available language datasets for humanitarian use. This includes a series of static and dynamic maps and 23 datasets covering nine countries: DRC, Guatemala, Malawi, Mozambique, Nigeria, Pakistan, Philippines, Ukraine, and Zambia.

This work is based on a partnership between TWB and University College London. The pilot project received support from Research England’s Higher Education Innovation Fund, managed by UCL Innovation & Enterprise. With support from the Centre for Translation Studies at UCL, this project was the first of its kind in the world to systematically gather and share language data for humanitarian use.

The majority of these datasets are based on existing sources — census and other government data. We curated, cleaned, and reformatted the data to be more accessible for humanitarian purposes. We are exploring ways of deriving new language data in countries without existing sources, and extracting language information from digital sources.

This project is built on four main principles:

TWB Language Data Initiative

1. Language data should be easily accessible

We started analyzing existing government data because we realized there was a lot of quality information that was simply hard to access and analyze. The language indicators from the 2010 Philippines census, for example, were spread over 87 different spreadsheets. Many census bureaus also publish in languages other than English, making it difficult for humanitarians who work primarily in English to access the data. We have gone through the process of curating, translating, and cleaning these datasets to make them more accessible.

2. Language data should work across different platforms

We believe that data interoperability is important. That is, it should be easy to share and use data across different humanitarian systems. This requires data to be formatted in a consistent way and spatial parameters to be well documented. As much as possible, we applied a consistent geographic standard to these datasets. We avoided polygons and GPS points, opting instead to use OCHA administrative units and P-codes. At times this will reduce data precision, but it should make it easier to integrate the datasets into existing humanitarian workflows.

We worked with the Centre for Humanitarian Data to develop and apply consistent standards for coding. We built an HXL hashtag scheme to help simplify integration and processing. Language standardization was one of the most difficult aspects of the project, as governments do not always refer to languages consistently. The Malawi dataset, for example, distinguishes between “Chewa” and “Nyanja,” which are two different names for the same language. In some cases, we merged duplicate language names. In others, we left the discrepancies as they exist in the original dataset and made a note in the metadata.

Even when language names are consistent, the spelling isn’t always. In the DRC dataset, “Kiswahili” is displayed with its Bantu prefix. We have opted instead to use the more common English reference of “Swahili.”

Every dataset uses ISO 639-3 language codes and provides alternative names and spellings to alleviate some of the typical frustrations associated with inconsistent language references.

3. Language data should be open and free to use

We have made all of these datasets available under a Creative Commons Attribution Noncommercial Share Alike license (CC BY-NC-SA-4.0). This means that you are free to use and adapt them as long as you cite the source and do not use them for commercial purposes. You can also share derivatives of the data as long as you comply with the same license when doing so.

The datasets are all available in .xlsx and .csv formats on HDX, and detailed metadata clearly states the source of each dataset along with known limitations.

Importantly, everything is free to access and use.

4. Language data should not increase people’s vulnerability

Humanitarians often cite the potential sensitivities of language as the primary reason for not sharing language data. In many cases, language can be used as a proxy indicator for ethnicity. In some, the two factors are interchangeable.

As a result, we developed a thorough risk-review process for each dataset. This identifies specific risks associated with the data, which we can then mitigate. It also helps us to understand the potential benefits. Ultimately, we have to balance the benefits and risks of sharing the data. Sharing data helps humanitarian organizations and others to develop communication strategies that address the needs of minority language speakers.

In most cases, we aggregated the data to protect individuals or vulnerable groups. For each dataset, we describe the method we used to collect and clean the data, and specify potential imitations. In a few instances, we chose to not publish datasets at all.

How can you help?

This is just the beginning of our effort to provide more accessible language data for humanitarian purposes. Our goal is to make language data openly available for every humanitarian crisis, and we can’t do it alone. We need your help to:

Integrate and share this data. We are not looking to create another data portal. Our strategy is to make these datasets as accessible and interoperable as possible using existing platforms. But we need your feedback so we can improve and expand them.
Add language-related questions into your ongoing surveys. Existing language data is often outdated and does not necessarily represent large-scale population movements. Over the past year, we have worked with partners such as IOM DTM, REACH, WFP, and UNICEF to integrate standard language questions into ongoing surveys. This is essential if we are to develop language data for the countries that don’t have regular censuses. The recent multi-sectoral needs assessment in Nigeria is a good example of how a few strategic language questions can lead to data-driven humanitarian decisions.
Use this language data to improve humanitarian communication strategies. As we develop more data, we hope to provide the tools for Lucia and other humanitarians to design more appropriate communication strategies. Decisions to hire interpreters and field workers, develop radio messaging, or create new posters and flyers should all be data-driven. That’s only possible if we know which languages people speak. An inclusive and participatory humanitarian system requires two-way communication strategies that use languages and formats that people understand.

Clearly, the answer to Lucia’s question turned out to be more complicated than any of us expected. This partnership between TWB and the Centre for Translation Studies at UCL has finally made it possible to incorporate language data into humanitarian workflows. We have established a consistent format, an HXL coding scheme, and processes for standardizing language references. But the work does not stop with these nine countries. Over the next few months we will continue to curate and share existing language datasets for new countries. In the longer term we will be working with various partners to collect and share language data where it does not currently exist. We believe in a world where knowledge knows no language barriers. Putting language on the map is the first step to achieving that.

Eric DeLuca is the Monitoring, Evaluation, and Learning Manager at Translators without Borders.

William Low is a Senior Data and GIS Researcher at University College London.

Funding for this project was provided by Research England’s Higher Education Innovation Fund, managed by UCL Innovation & Enterprise.

TWB intern is recognized as a Young African Leader

Cédrick Young African Leader YALI — Cédrick Irakoze

At Translators without Borders (TWB), we are lucky to have extraordinary team members who are recognized worldwide. We are always grateful to have uniquely skilled members of the international community choose to be part of our cause. Today, we are proud to share the story of Cédrick Irakoze, Crisis Response and Community and Recruitment Intern for TWB. He was recently awarded a place to be part of the Young African Leaders Initiative (YALI) network. The YALI network invests in the next generation of African leaders, providing invaluable opportunities to connect and learn from experts. Learn more about the YALI network here.

Cédrick is a young Burundian language professional. He holds a bachelor’s degree in TESOL (Teaching English to Speakers of Other Languages) from the University of Burundi, and has years of experience as a professional translator. He believes that language can improve or even save lives in this global world. And interaction in the right language can be vital for everyone, no matter people’s language, culture, or the color of their skin.

“TWB is my professional home” – Cédrick

In 2018, Cédrick first featured in our blog as a volunteer translator from English and French into Rundi. This was his introduction to the world of language in humanitarian work: “When I joined TWB as an intern, I joined a community of like-minded individuals serving the global community. Now I call TWB my professional home.” Day-to-day, Cédrick engages and collaborates with our translator community to help create a world with no language barriers.

But in late 2019, he did something different. He successfully applied for the Young African Leaders Initiative program.

The Young African Leaders Initiative (YALI)

In 2016, Cédrick joined the Young African Leaders Initiative network with over 25,000 other young and talented individuals. In 2019, he met with 108 successful candidates from over 7000 applicants to attend a one-month leadership training course.

A group of YALI network members in Nairobi.

Energetic public officials, business owners, and local and international nonprofit leaders from all over Africa came together in Nairobi, Kenya. On hearing their stories, Cédrick reflected, “The way they are each committed to making their communities better inspired me.” The Translators without Borders team is delighted to have witnessed a team member take on such an exciting, formative challenge.

“Thank you very much. TWB showed me so much love and support before and during the program!” – Cédrick

Cédrick Irakoze, right, with TWB Kenya Manager Paul Warambo, left.

It’s all about communication

The course was about inspiring and equipping one another to become better leaders. Participants developed their communication skills and built solutions-oriented networks. These factors are central to the changes these young leaders want to see in society. Each member of the diverse group – native speakers of over fifty languages – played a vital part.

Cédrick Irakoze, left presenting to the YALI network members.

This richness and diversity are reflected in TWB’s own community of translators and supporters, and in our way of working. We too rely on the power of teamwork to make change — to improve communications and access to information worldwide. Cédrick’s big takeaway is that when we come together we can innovate, we can flourish and we can make each other feel valued.

“Diversity is richness in professional life” – Cédrick

With the skills he’s learned through this course, Cédrick hopes to make a positive impact in his professional and social circles. “I can’t wait to contribute more and better to our common mission: to create a world that knows no language barriers.”

Cédrick Irakoze and friends at the YALI network meetup in Nairobi.

Start your own journey as part of the TWB community.

Written by Danielle Moore, Communications Officer for Translators without Borders. Interview responses by Cédrick Irakoze, Crisis Response and Community and Recruitment Intern for Translators without Borders.

Humanitarian work close to home: Irina Nosova

Shared philosophies

Irina’s philosophy fits right in with that of Translators without Borders: “In our information-packed society, it is essential to maintain access to vital information for everybody. So, my biggest motivation is helping people by delivering information to interested parties,” says Irina, former lawyer now turned English-Russian translator. Since joining as a volunteer translator in 2016, Irina has translated and revised a total of 110,220 words into Russian, one of our top ten most frequently requested languages.

Irina Nosova English Russian translator — Irina Nosova, translator

The projects touch on all sorts of information, including healthcare. Often, information and research in one language benefit speakers of another language – but it needs to be accurate and it needs to be available. So we asked about a particular project which made vital information available in Irina’s own country, in Russian, her mother-tongue. When tasked with translating an important anti-tuberculosis study she found it to be one of her most difficult projects to date. The translator translated study protocol and presentations to find out more, before later reading news articles and discovering the reality of the tuberculosis situation in Russia.

“I was shocked to find how high the burden of tuberculosis is in my country!” – Irina

Finding practical solutions

Although Irina focuses on the importance of language in Russia, she is also hopeful about how sharing information in many languages can spread helpful and life-changing information. “I hope that the study of novel tuberculosis treatment will speed up the registration of new drugs which are vital for successful treatment,” she told TWB.

As well as more academic pieces, Irina finds translating personal stories equally important. One of her projects involved translating patients’ stories for partner non-profit EURORDIS – Rare Diseases Europe.

“I realized that stories shared by patients with rare diseases and their families could inform people in similar situations in Russia about how to deal with those diseases.” – Irina

Tapping into your skillset

Irina’s desire to help as a volunteer translator has helped her tap into personal and professional skills. “Volunteering improved my time management: I have to calculate and allocate the time I can spend to complete the tasks before the deadline, alongside my other daily tasks. Before launching my own business, I worked as a lawyer and I volunteered with TWB at night, after work. Now, I can be more flexible and am able to contribute more time.”

Her advice for other TWB Community members is to constantly improve your skills, learn new terminology, and check your quality of translation. “Doing translation in Kató – TWB’s online translation environment – requires the same quality approach as any other project: the highest possible. So, before claiming the new task make sure you understand the topic and do your own research to provide the best possible translation.” It helps you understand the context and importance of the situation you are translating about – like in the case of Irina’s tuberculosis project. When Irina dug deeper into the topic she was translating about, she discovered a personal interest in medical translation and later, clinical research. Her projects opened new doors: “Volunteering with TWB improved my resourcefulness and research skills and pushed me to explore new horizons in translation, take new courses, and dig deeper.”

Join our translator community.

Written by Gloria Malone and Danielle Moore, Communications Officers for Translators without Borders. Interview responses by Irina Nosova.

Transfer Learning Approaches for Machine Translation

This article was originally posted in the TWB Tech Blog on medium.com

TWB’s current research focuses on bringing language technology to marginalized communities

Translators without Borders (TWB) aims to empower people through access to critical information and two-way communication in their own language. We believe language technology such as machine translation systems are essential to achieving this. This is a challenging task given many of the languages we work with have little to no language data available to build such systems.

In this post, I’ll explain some methods for dealing with low-resource languages. I’ll also report on our experiments in obtaining a Tigrinya-English neural machine translation (NMT) model.

The progress in machine translation (MT) has reached many remarkable milestones over the last few years, and it is likely that it will progress further. However, the development of MT technology has mainly benefited a small number of languages.

Building an MT system relies on the availability of parallel data. The more present a language is digitally, the higher the probability of collecting large parallel corpora which are needed to train these types of systems. However, most languages do not have the amount of written resources that English, German, French and a few other languages spoken in highly developed countries have. The lack of written resources in other languages drastically increases the difficulty of bringing MT services to speakers of these languages.

Low-resource MT scenario

In scientific literature for machine translation, there is no particular consensus on which corpus size constitutes a low-resource scenario. But we can say roughly that a low-resource condition is when the size of the parallel training corpus is not sufficient for reaching an acceptable result with the standard MT approaches. This is usually judged with a standardized automatic evaluation metric called BLEU, which correlates with human translation assessments.

Figure 2, modified from Koehn and Knowles (2017), shows the relationship between the BLEU score and the corpus size for the three MT approaches.

A classic phrase-based MT model outperforms NMT for smaller training set sizes. Only after a corpus size threshold of 15M words, roughly equivalent to 1 million sentence pairs, classic NMT shows its superiority.

Low-resource MT, on the other hand, deals with corpus sizes that are around a couple of thousand sentences. Although this figure shows at first glance that there is no way to obtain anything useful for low resource languages, there are ways to leverage even small data sets. One of these is a deep learning technique called transfer learning, which makes use of the knowledge gained while solving one problem to apply it to a different but related problem.

Cross-lingual transfer learning

Zoph et al. (2018) applied transfer learning in machine translation and proved that having prior knowledge in translation of a separate language pair can improve translating a low-resource language.

Figure 3 illustrates their idea of cross-lingual transfer learning.

The researchers first trained an NMT model on a large parallel corpus — French–English — to create what they call the parent model. In a second stage, they continued to train this model, but fed it with a considerably smaller parallel corpus of a low-resource language. The resulting child model inherits the knowledge from the parent model by reusing its parameters. Compared to a classic approach of training only on the low-resource language, they record an average improvement of 5.6% BLEU over the four languages they experiment with. They further show that the child model doesn’t only reuse knowledge of the structure of the high resource target language but also on the process of translation itself.

The high-resource language to choose as the parent source language is a key parameter in this approach. This decision is usually made in a heuristic way judging by the closeness to the target language in terms of distance in the language family tree or shared linguistic properties. A more sound exploration of which language is best to go for a given language is made in Lin et al. (2019).

Multilingual training

The path that was cleared by cross-lingual transfer learning led naturally to the use of multiple parent languages. The straightforward approach, first described by Dong et al. (2015), mixes all the available parallel data in the languages of interest and sends them into training as illustrated in Figure 4.

What results from the example is one single model that translates from the four languages (French, Spanish, Portuguese and Italian) to English.

Multilingual NMT offers three main advantages. Firstly, it reduces the number of individual training processes needed to one, yet the resulting model can translate many languages at once. Secondly, transfer learning makes it possible for all languages to benefit from each other through the transfer of knowledge. And finally, the model serves as a more solid starting point for a possible low-resource language.

For instance, if we were interested in training MT for Galician, a low-resource romance language, the model illustrated in Figure 4 would be a perfect fit as it already knows how to translate well in four other high-resource romance languages.

A solid report on the use of multilingual models is given by Neubig and Hu (2018). They use a “massively multilingual” corpus of 58 languages to leverage MT for four low-resource languages: Azeri, Belarusian, Galician, and Slovakian. With a parallel corpus size of only 4500 sentences for Galician, they achieved a BLEU score of up to 29.1% in contrast to 22.3% and 16.2% obtained with a classic single-language training with statistical machine translation (SMT) and NMT respectively.

Transfer learning also enables what is called a zero-shot translation, when no training data is available for the language of interest. For Galician, the authors report a BLEU score of 15.5% on their test set without the model seeing any Galician sentences before.

Case of Tigrinya NMT

Tigrinya is an Ethiopian language spoken by around 7.9 million people in Eritrea and Ethiopia. It is neither supported by any commercial MT provider, nor has any publicly available models. TWB is currently developing open datasets and MT for Tigrinya in cooperation with the Masakhane initiative.

Tigrinya is no longer in the very low-resource category thanks to the recently released JW300 dataset by Agic and Vulic. Nevertheless, we wanted to see if a higher resource language could help build a Tigrinya-to-English machine translation model. We used Amharic as a parent language, which is written with the same Ge’ez script as Tigrinya and has larger public data available.

The datasets that were available to us at the time of writing this post are listed below. After JW300 dataset, the largest resource to be found is Parallel Corpora for Ethiopian Languages.

Our transfer-learning-based training process consists of four phases. First, we train on a dataset that is a random mix of all sets totaling up to 1.45 million sentences. Second, we fine-tune the model on Tigrinya using only the Tigrinya portion of the mix. In a third phase, we fine-tune on the training partition of our in-house data. Finally, 200 samples earlier allocated aside from this corpus are used for testing purposes.

As a baseline, we skip the first multilingual training step and use only Tigrinya data to train on.

We see a slight increase in the accuracy of the model on our in-house test set when we use the transfer learning approach. The results in various automatic evaluation metrics are as follows:

Conclusion

Neural machine translation is a data-hungry technology. Although this severely reduces the possibility to expand it to the majority of the world’s languages, we can still apply various techniques to make it available to more people than if we limited ourselves to approaches tuned towards high-resource languages. Methodologies like transfer learning and linguistically informed data mixture have a role to play in helping everyone communicate in their language.

Written by Alp öktem, Computational Linguist for Translators without Borders

100 Translations to Prevent Sexual Exploitation and Abuse

Written by Alice Castillejo, Programme Advisor for Translators without Borders

For every new response, we need the right words to fight sexual abuse

I work for Translators without Borders, an organization that highlights the importance of language and clear communication. I am blessed to work with colleagues who expose me every day to subtle linguistic and cultural differences. Some of those differences result in hilarious misunderstandings, others are more challenging. They always point to the importance of choosing our words with care.

I was reminded of that again last week when I stood in front of a room full of my multicultural professional peers to discuss sexual abuse. I suddenly found myself acutely aware of the need to get my words right to avoid embarrassing myself or offending the audience.

100 Translations to Prevent Sexual Exploitation and Abuse

Understanding what sexual exploitation and abuse mean in a humanitarian emergency is not as simple as it seems. For local staff, who may never have worked for humanitarian organizations before, it can be even more complicated. Principles aimed at preventing sexual exploitation and abuse contain new ideas about power relationships, new terms to understand, and new rules and responsibilities to learn and put into practice. Providing the information in a language that local staff can understand is the least we can do and an important first step toward addressing the problem.

In collaboration with the UN’s Inter-Agency Standing Committee (IASC), Translators without Borders (TWB) developed a plain-English version of the key humanitarian messages on preventing sexual exploitation and abuse. The benefit of a plain-English version is twofold:

It promotes wider understanding, particularly among those with limited English proficiency.
It removes ambiguity and legal terminology, increasing the chances of an accurate translation into other languages. For example, we replaced legalese like “constitute acts of gross misconduct and are therefore grounds for termination of contract” with “humanitarian workers can be disciplined – even fired – for unacceptable behavior in relation to sex”.

Distribution of the 100 Translations

We then translated the plain-English version into more than 100 languages, and started to distribute them to communities around the world.

A sudden-onset crisis often mobilizes people who have never thought about sexual abuse or power dynamics in their regular day jobs. The new local staff may come from a context where going to sex workers is commonplace, or where informal exchanges and bribes, including transactional sex, are part of getting things done. And often they live in a hierarchy where reporting one’s peers or seniors is dangerous.

We need to explain in clear language that these practices are forbidden in the humanitarian context, and that staff must report them. In translation, there are compromises in this process – is it better to find a word with no stigma or a word that will be more widely understood? For example, while the term “sex worker” is a more empowering term, we found that “prostitute” is more widely understood, despite its negative connotations. And if terms are gendered, have we chosen words that clearly indicate that sexual abuse may include sexual abuse of men? Are we sure the words we’ve chosen are neither so crude that they offend nor so euphemistic that they are incomprehensible?

Working with Partners to Ensure the 100 Translations are Effective

Since the launch of this joint IASC-TWB project in 2018, TWB’s team of translators and supporters has worked hard to produce accurate translations, which have then been reviewed and validated by local humanitarian staff from across the world. Our local reviewers have played an essential role, offering specific local terms, checking suitability, and adjusting translations to reflect local dialect. That is why, for example, we have several Arabic and Spanish language versions, as well as audio versions for Rohingya and Chittagonian.

We know that 100 languages is a drop in the ocean. But for each new multilingual humanitarian crisis, we hope to build the portfolio to meet the needs of newly engaged humanitarian staff.

How to start using the 100 Translations to Prevent Sexual Exploitation and Abuse

The key messages are just a start. Crisis-affected communities need to understand what behaviour from humanitarians is unacceptable. We may need to deliver the message in more hard-to-source languages and also in audio or pictorial formats. Staff training packages must be in languages that staff understand. And, of course, when someone is exploited or abused, they must be able to report in the language they are most comfortable in and receive support in that language.

You can help make sure that humanitarians understand what Protection from Sexual Exploitation and Abuse (PSEA) is and what it means for them. Please join us in this effort by distributing the translations to your colleagues, making sure your training is in languages they truly understand, and providing new translations in additional languages.

You can find more information about this project, and the growing number of translations, here.

The Core Humanitarian Standard Commitments are now available in plain English

Written by Kate Murphy, Plain-language editor for Translators without Borders, and Ellie Kemp, Head of Crisis Response for Translators without Borders.

Translators without Borders (TWB) helps its humanitarian partners apply plain language principles to written content. We worked together with the CHS Alliance to develop a plain-language version of the Core Humanitarian Standard’s Nine Commitments.

As humanitarians, we use tools like the Core Humanitarian Standard on Quality and Accountability (CHS) to hold ourselves accountable to the people we assist. We know that the best responses are those shaped by those directly affected. The CHS provides a great opportunity to institutionalise Accountability to Affected Populations in a way that effectively translates at field level for our work.

But ironically, many of the people we assist themselves aren’t yet aware of our commitments to them. Many don’t have the literacy skills to read, understand, or react to them. Others simply won’t have the time, motivation, or emotional energy to read through the full Nine Commitments of the CHS.

Teacher writing sentences in Rohingya Zuban (Hanifi Script). Kutupalong Refugee Camp near Cox’s Bazar, Bangladesh. Photo: Eric DeLuca / Translators without Borders

The question for CHS Alliance was how to make information on quality and accountability accessible to the affected population. Could plain language help everyone to hold aid organisations to account, in line with CHS Commitment 4? TWB jumped at the chance to help with this. For us, it was another opportunity to highlight the importance of communicating with people in an appropriate and usable language and format.

Everyone appreciates plain language

Plain language makes text easier to understand, easier to put into practice, and easier to recall later. For English materials, such as the new CHS onepager, this means using plain-English. Non-native English speakers appreciate it, especially those with limited education and limited English language skills. But highly literate speakers also benefit, especially those operating in high-pressure environments. And of course, it makes translation easier too.

That’s the thing about plain language: everyone who appreciates clear, concise information benefits from it.

Plain language reduces that reading effort for everyone.

To make the CHS Commitments easier to read, we helped the CHS Alliance apply three established plain-language principles:

Use personal pronouns, including “you” and “we.” That engages readers and adds certainty about who is responsible for different actions.
Rewrite sentences so they contain fewer than 20 words. Shorter sentences are easier to comprehend.
Replace uncommon or technical terms with alternatives that are familiar to all readers.

Photo credit TWB.

We used word-frequency data to identify words in the CHS Commitments that occur less commonly in English print and audiovisual media. The more frequently a term occurs in different media, the more likely it is to be familiar to readers. So, we replaced less frequent words with more common alternatives. For instance, we replaced ‘assistance’ with ‘support’.

Some words like “resilient,” “entitlements,” “competent,” and “efficiency,” are included in the original CHS Commitments, and experienced humanitarians are likely to comprehend them easily. However, word-frequency data suggests that they may reduce reading speed and therefore reader engagement for people who are less familiar with them. Because we wanted to reduce general reading effort, we replaced them.

Spread the word

CHS Alliance members and partner organisations can access the plain language English version of the CHS commitments. Over the coming months, the CHS Alliance expects to provide translated versions in an increasing number of languages and will keep you updated on progress.

In the meantime, CHS Alliance members who are part of the new AAP Community of Practice have suggested various ways we could use the plain-English version to communicate our commitments more effectively to affected people:

Provide an audio version of the plain-English commitments for less literate audiences.
Develop graphics to accompany the text and make it available as a poster or leaflet.
Develop a child-friendly version.
Provide staff with guidance on how to use the plain-English version as part of accountability to affected populations.
Agree a common way to track understanding of the plain-language versions, possibly as part of regular perception and satisfaction monitoring activities.

We’re excited to see how the new plain-language CHS will help make accountability a reality for all those we serve.

What do you think?

To find out more about how plain language could benefit your organization please get in touch with Kate Murphy, Plain-language editor at [email protected].