Today marks the 59th year of Africa Day, a worldwide celebration commemorating the founding of the Organization of African Unity (OUA, from the French “Organisation de l’unité africaine”).
The OUA was founded May 25, 1963, largely as an effort to support the decolonization of African nations that were still under the influence of European colonization, like South Africa and Southern Rhodesia (now Zimbabwe). Today, the OUA’s role has been taken over by the African Union. At the OUA’s inaugural conference in Ethiopia, the organization called for an annual celebration of African Freedom Day on May 25, which has since been renamed Africa Day.
Most of the nations of Africa have been politically independent of colonial powers for the better part of half a century now. Still, discussions at the intersections of language and technology often overlook the continent. Africa is home to about one-third of the world’s 6,000 languages, but researchers in technology and linguistics alike have often focused largely on more dominant languages like English, Spanish or Chinese.
That said, recent times have seen major strides. In just the past month, Google Translate added ten African languages to its platform. This year on Africa Day, it’s worth exploring some of the recent developments in the field that have furthered the visibility of African languages in the tech sphere.
During the COVID-19 pandemic, it became especially clear that languages indigenous to the African continent were underrepresented on the internet. While individuals who spoke and read languages like English and French (languages that became prevalent in Africa during the colonial era) typically had access to accurate and up-to-date information on the pandemic, those who only spoke indigenous languages were often left in the dark.
Not only that, but many languages didn’t have native words for concepts like “hand sanitizer” or “face mask,” which made it difficult to render public safety information adequately. In response, activists working with WikiAfrica geared up to translate Wikipedia articles on the pandemic into indigenous African languages like isiZulu and Twi (the latter of which wasn’t even available on one of the most popular MT systems until just a few weeks ago).
More recently, the Lanfrica platform launched earlier this year, creating one of, if not the, first centralized databases of natural language processing (NLP) resources specifically for African languages. The platform hosts more than 2,000 languages spoken on the African continent.
Because many languages indigenous to the African continent are deemed “low-resource” — meaning that, in spite of their oftentimes quite large speaker bases, there are few NLP resources available in these languages — these languages lag behind others in terms of MT availability and other factors. Initiatives like WikiAfrica and platforms like Lanfrica help create a broader wealth of resources for researchers, thus improving accessibility for African languages.
In the words of Lanfrica’s developers, these types of programs “offer huge potential for better discoverability and representation of African languages on the web.”
“Such insight can lead to better allocation of funds, efforts, etc. toward bringing the more under-researched languages forward in NLP — thereby fostering the equal progress of African languages,” the team wrote in its announcement of the platform’s launch.