Saturday, September 03, 2016

Facebook, ISOC, and A12n

In his recent visit to Lagos, Nigeria, Facebook founder and CEO Mark Zuckerberg indicated that Facebook will add more African language interfaces. Meanwhile, at the African Peering and Interconnection Forum (AfPIF2016) in Dar es Salaam, Tanzania, the Internet Society (ISOC) released a report entitled "Promoting Content in Africa," which highlights the importance of internet content in African language for full access by Africans.

These two developments concerning on the one hand localization of the software for a popular social media platform, and on the other hand the creation of content, highlight the dual aspects of Africanization (A12n) of information and communication technology in/for Africa. As these processes develop, it would be useful for to find ways to integrate them as appropriate, and foster collaboration among organizations and individuals involved in either or both. (That was the intent of the African Network for Localisation, ANLoc, albeit with a focus mainly on the software and enabling aspects.)

It is possible, as the ISOC report notes, for content to be developed or translated in a language even when the software on which it is created is not localized in it. And that certainly would be the case for the less widely spoken languages, at least in the near term. However, the availability of software interfaces - whether for social media like Facebook or for production software - in at least the major African languages, would probably help even for the less-spoken ones.

Facebook sign-up in Hausa. (Source:
Facebook currently is available in the following African languages (links are to Wikipedia articles): Afrikaans; Arabic; Hausa; Kinyarwanda; Malagasy; Somali; Swahili; and Tamazight

One of the contributors to the ISOC report, Dawit Bekele, who is ISOC's African Bureau Director, was a participant in the PanAfrican Localisation Workshop in Casablanca, June 2005, and the Pan African Research on L10N Workshop & Localization Blitz in Marrakech, February 2007.


Ibraahiima Saar said...

Hi Don,
I have been reading a number of your articles including the one about "macrolanguages". The articles as well as the comments are really insightful.
You probably know that Facebook is being localized in Fulah. I contacted Facebook in 2014 asking them to add Fulah, of course with all convincing arguments you can imagine.
One year later, the great news arrived by email saying yes, Fulah is being added. Then I sent them basic locale information like the name of the language in Fulah, etc.
Now we are facing a huge problem. Most people translating have no experience in software translation and are not familiar with existing terminology. Just imagine everyone creating their own terms!
But the biggest issue we face is the difference between the dialects of translators. Most of translations in Pular Fuuta Jaloo are not intelligible by me for instance. And since the adoption is based on "voting", only very active translators and language communities will have their translations recognized.
Now we gave a very bad overall quality of translations on Facebook that everybody understands only partly.
I have not mentioned all the "direct translations" that are ridiculously inappropriate for a native speaker. Also those translations with the ɗ reolaced eith d, the ƴ with y' and ɓ with the number 6!!!
I wrote to Facebook asking them to create a different translation for fuc, ffm and fuh. Note that the literacy rate is by far higher in Fuuta Tooro.
Now my question would be, what are your impressions about those issues.
Is it a good solution to create translations for individual dialect that are not very close like fuc and fuh?
I'd really appreciate your help on his.
Ibrahima Sarr

Don said...

Jaaraama Ibrahima. You raise some important questions that deserve a longer treatment, but here are a few quick thoughts.

First, have you been in touch with anyone from ACALAN's Vehicular Cross-Border Language Commission for Fulfulde (the term they use which in this case includes Pulaar and Pular)? Prof. Fary Ka is the chair. Their input and guidance would hopefully be valuable to the localization effort in Fulah (ff or ful). Hopefully they have some experience with localization and terminology in this domain.

Second, I'm remembering some work by Alfa Ibrahim Sow. Although he was native of Guinea and came from a Pular background, he was one of the scholars of his era who worked on materials intended to be accessible across varieties of Fulah. None of that is directly applicable to the terminology issues you currently face, but could be an inspiration. It seems that speakers of the language who are deeper in it - by study, work, or just the experience of years of active use - have more facility in communicating across varieties. I recall for example once seeing a Pulaar (fuc) speaker from Mauritania interacting with a Fulfulde (fuq) speaker from the Wooɗaaɓe group of Niger in their common language with no apparent problems to understand.

Third, having said that, it is my impression that in the range of varieties of Fulah, the Pular of Fuuta Jaloo (fuf) and the Fulfulde of Aadamawa (fub) are kind of outliers. The rest of the varieties, from Pulaar (fuc) in the west to eastern Niger (fuq) and Nigeria (fuv) in the east seem like more of a spectrum. However some scholars divide the whole spectrum, including Fuuta Jalon and Aadamawa varieties into western and eastern. Looked at these ways it gets complicated again, but not as much as resorting to the nine (9) varieties classified as languages in Ethnologue & ISO 639-3.

Fourth, the general assumption is that on the level of terminology and dictionaries, it is easier to treat the language as a whole rather than 2, 3, or 9 separate languages, while on the level of text or discourse, it is maybe better to write/speak for the specific variety used. The work on localizing Facebook into Fulah is in a way testing that, but unfortunately without guidance of a team of experts(?) having a wider vision.

Fifth, another assumption is that even where different localizations or translations are needed for different varieties, it is possible to chose words and expressions that are more accessible to speakers across varieties than locally specific expressions would be. In the Fulfulde of Maasina (ffm) there is a unique counting system that is locally used instead of the common one known to its speakers and speakers of other varieties - but it would not be appropriate, by this thinking, if ever there were a localization in that variety. "Harmonization" in this case should be the guiding vision - so that any speaker of any variety of the language would not be totally lost in the localization or text in another variety.

Sixth and final, some years ago, Sonja Fagerberg-Diallo wrote on the diversity within Fulah, including a discussion of how a standard literary variety might emerge. That could be, as I recall, by the strength of production in one variety becoming widespread (that was the case for many of today's national/official languages of Europe, although those processes were largely top-down). Or perhaps it could be by another process (consider for example the N'Ko community's grassroots efforts at developing a literary standard). It may be that it is in the information technology environment that standard varieties of African languages like Fulah (however we write or call it) emerge.

Hope this helps. Mi yetti ma sanne.