Research News

UB researchers working to improve understanding, trust of COVID-19 information for speakers of all world languages

Signs to help students navigate in the Crossroads Culinary Center, C3, in the Ellicott Residence Halls Complex in late August 2020.

UB linguistics researchers are leading a multidisciplinary project that trains native speakers to create original messages rather than producing translations of standardized public health messages — such as this sign at a dining facility at UB — because translation alone is not enough. Photo: Meredith Forrest Kulwicki


Published February 2, 2022

headshot of Jeff Good.
“Trusting a message depends upon more than translating a message. ”
Jeff Good, professor
Department of Linguistics

To be most persuasive and impactful, sharing knowledge about the COVID-19 pandemic with a global audience requires messaging in all the world’s languages, including many minority languages that are not associated with standardized writing systems, Western education systems or socioeconomic privilege. As of late 2021, preventative measures related to COVID-19 are available in about 700 languages, with roughly 630 of those translations targeted to speakers of minority or marginalized languages. 

But that’s only a modest beginning. There are approximately 7,000 languages currently spoken in the world, 6,300 of which are minority languages, including roughly 500 different languages across the five boroughs of New York. Is it possible to reach everyone? A multidisciplinary team, which includes researchers from UB, believes so. And they have designed a system to accomplish the task, not only for COVID-19, but for any critical public health information where the risk communication strategy requires reaching a linguistically and culturally diverse audience.

Their volunteer-run initiative, VirALLanguages, has organizational partners at UB, the Berlin-Brandenburg Academy of Sciences and Humanities, and the University of Hawaiʻi at Mānoa, with training in linguistics, social sciences and public health. Pierpaolo Di Carlo, a postdoctoral researcher in the UB Department of Linguistics, is VirALLanguages’ lead coordinator and first author of a paper published in the Bulletin of the World Health Organization that outlines the mechanics of the project, as well as its ambitious goal, which began as a conversation with Jeff Good, UB professor of linguistics.

“As the pandemic broke out, Jeff and I started wondering about what we could do to help. As it turned out, our findings on traditional forms of multilingualism in rural Africa provided key insights on how communication actually takes place in many marginalized communities,” says Di Carlo. “We were fortunate to find enthusiastic partners at the Community for Global Health Equity in Buffalo and further developed our ideas through feedback from our collaborators in Cameroon. Painstaking work from a number of volunteers across the world did the rest.”

VirALLanguages’ design prioritizes message production in minority languages featuring esteemed community members. The project also trains native speakers to create original messages, rather than producing translations of standardized messages, because translation alone is not enough, according to Good, who serves as a VirALLanguages project director and is one of the paper’s co-authors.

“Just because you’ve translated the institutional language of the World Health Organization into languages that people know doesn’t mean those messages will be effective,” says Good, an expert in linguistic diversity. “Trusting a message depends upon more than translating a message.”

As an example, Good points to the colonial legacy responsible for the continued presence of French and English in the Republic of Cameroon in West Africa, where he has focused much of his own research.

“French and English are the two official languages, but there are hundreds of other languages that are important instruments for building relationships in local communities,” he says. “Many speakers of minority languages understand a majority language, so someone in Cameroon might see and understand messaging in French that says ‘Wear a mask,’ but the message doesn’t resonate because layers of repression and colonialism get in the way of trust.”

Translation of health and safety information should involve communications that build upon relationships of trust. Linguists know that messages delivered in a primary language are more likely to be memorable and to produce a behavioral change than those shared in a person’s second language. From this perspective, behavioral change is critical. Many of the countries where minority languages are spoken have weak health care systems, and lack the resources necessary to roll out blanket vaccination programs and widespread treatment options. That leaves behavioral change as the most viable response to the pandemic.

Trust, in the case of effective messaging, is also about form. The primary domain for most of the world’s languages is not writing. Speakers of minority languages, which account for about one-tenth of the world’s population, often write in a majority language, but their local language is not written. Oral knowledge of a language is important for building connections with the community.

“There’s also consideration for how people want to receive a message,” Good explains. “In parts of Western Africa and sub-Saharan Africa, there’s this idea of a ‘greeting culture.’ You always start with a greeting. But these disembodied COVID-19 translations have no greeting. They’re institutional. And it’s hard to build trust if you inherently distrust the institution sending the message.

“We need trust-based, not text-based messaging.”

VirALLanguages accounts for these nuances by providing public health information to intermediaries in the community who know both the target language and the original majority language of the source material. Project volunteers then test intermediaries to be sure the information is understood. These intermediaries then produce messages in the local language that are subsequently reviewed by another community member as a final verification step prior to dissemination, which could be via social media, radio broadcast, or even a loudspeaker on a vehicle driving through the community.

After an initial period of intense activity in 2020, Good says that VirALLanguages is planning for a new phase of the work that will build community feedback into the model to refine messages to further community understanding.

“I don’t know of another collaboration of this type that has local speakers working with linguists, social scientists and public health experts to accurately disseminate public health information,” Good says.

Other co-authors of the paper are Bradley McDonnell, associate professor of linguistics at the University of Hawaiʻi at Mānoa; Mandana Seyfeddinipur, director of the SOAS World Languages Institute at the University of London; Katarzyna Kordas, associate professor in the UB Department of Epidemiology and Environmental Health; and Lisa Vahapoğlu of the UB RENEW Institute.

The work of virALLanguages was supported in part by the Community for Global Health Equity at UB.