I was in Eswatini from the 21-22 May 2025 on what turned out to be my last training event for SWiP. My stay nearby to the Residence of Ndlovukazi (the Queen Mother) at Lobamba, Eswatini was far to brief. Muzi tells me that this is the royal, spiritual, cultural and political heart of Eswatini (previously called Swaziland). I probably broke far too many cultural taboos. Crap!
Sadly, I could not visit Umsamo Wesive (the national museum), wander around Libandla (bi-cameral parliament) or watch the Sihlangu Semnikat (the Kings Shield) practice at Somhlolo Stadium (the national stadium). But I did have a fabulous three day workshop at King Sobhuza II Memorial Park. And then I created a map for the organisations in town.
Although these locations were in walking distance of my hotel, I had double-booked myself for the “Great Debate”. So instead of exploring, I spent my days working alongside aspiring Wikipedians (journalism students, academics, and educators) on this WikiProject: Eswatini. Nights were spent planning (and worrying) how I would manage the “Great Debate” Topic: LLMs in Higher Education, which was scheduled for Friday, 23 May, my departure day.
Unlike other SWiP events, the visit to Eswatini did not have a post-event workshop evaluation. So I thought I’d share my thoughts with SADiLaR colleagues and Wikipedians
- Code cannot completely compensate for poor language
- We need to create conversational designs
- An aggressive template – and quick cover-up
- Unanswered Questions
1) Code cannot completly compensate for poor language skills?
I have been making onboarding tutorials in vernacular languages (Afrikaans, Zulu and Siswati) – see https://ss.wikipedia.org/wiki/Bantfu:Derek_J_Moore
Now if you speak Siswati its possible that you are chuckling (or cringing) at my poor attempts at this language. I share the sense of embarrassment. I am bilingual, speak Afrikaans (average) and English (reasonably well). But unlike the majority of South African citizens, I don’t speak or write any vernacular languages and I often get things wrong.
I’ve noted before that language support for Siswati (and other South African vernacular languages) code on Wikipedia is minimal. But what I have not quite comprehended is just how much of the code underneath English Wikipedia templates and modules also needs to be installed onto the vernacular platform. Neither can I find too many clear vernacular guidelines on installing the code.
English Wikipedia has access to a wide code base and explicit policies and guidelines. What isn’t clear to me, at least here in Wikimedia South Africa, who are the people in charge of each language group and which policies do I fall under?
Wikipedia is available in almost 350 languages. Each language has its own community of volunteers. Efforts have been made to create translation resources, so that we can build on each other’s work. But the languages that have documentation and support are minority world languages. Regular communication is sent in English, not vernacular languages. For an organisation that intends to make all the worlds knowledge freely available, this is disappointing.
Vernacular Wikipedia has a partner in the form of Wikidata, a free, collaborative, multilingual, secondary knowledge graph. Wikipedia and Wikidata have the potential to bring transformation for in minority languages in the majority world. You can freely correct and improve what has been written in Wikipedia and Wikidata. For example, my poor attempts at Siswati, isiZulu and Afrikaans on Wikipedia can be corrected by any of you. But when I add to Wikidata, then that information on that platform can be algorithmically translated and placed in all language versions. How good are these translations?
I suspect that my coding across Wikipedia platforms and averagely translated sentences, simply creates technical debt? Code does not compensate for my poor language.
2) Conversational Design
( see by Erika HallMarch 15, 2018)
Accomodation was not at Mlilwane, but were were comfortably housed at Happy Valley Hotel (-26.440434,31.188795 | 26°26’25.6″S 31°11’19.7″E). During the day, the Knowledge Workers (my pet name for the SADiLaR crew) with Herschall and I ran sessions. Henry Count Evans had assembled a fine bunch of people for the workshop. I enjoyed talking to aspiring and interested Wikipedians more about ss.wikipedia.org. It was clear to me that they were all quite familiar with the platform. After the workshop had finished, I popped into every Wikipedians sandbox and there I left them a message {{New userpage}}. The code can be read below.
{{ombox |type = notice |image = [[File:Essay.svg|50px|link=]] |style = background:#f4f4f4;border: 1px solid #FFB2B2; |textstyle = color:#0A0A0A; |text = Hello, ”'{{SAFESUBST:Perhaps experiment in ”'{{#if:{{{list|}}}{{{a|}}}|a|the}} user [[Help:My sandbox|sandbox]] of [[{{NAMESPACE}}:{{{1|{{ROOTPAGENAME}}}}}|{{{1|{{ROOTPAGENAME}}}}}]]”’. (PLEASE NOTE: A user sandbox is a subpage of the user’s [[Wikipedia:User pages|user page]]. It serves as a testing spot and page development space for the user and is not an encyclopedia article.)
I did not get a great response. Later I went back to what English Wikipedia calls the Tea House. SS calls it Likhefi leinthanethi which (if Google translate is right), means Internet Cafe. Today, if you visit this page, you might note the interaction, it is all one way and in English.
Khuluma nebantfu
There is often an apology for not using vernacular. But no attempt to hire or appoint someone who might assist. Now Bobby Shabangu is from that part of the world and I saw once that he was translating the communication. But gradually it has become a broadcast instead of a conversation? But I suppose when you have been a radio DJ, then maybe broadcasting stories comes more naturally than interacting? Or perhaps the organisational structures of Wikimedia, who send out the same press release across all platforms. Go through the typical corporate publishing process where all has to be approved and stay on brand.
Erica Hall wrote a book about Conversational Design (its free to download). She encourages designers to be story listeners and interreact with real people, to make social connections. Our conversations involve give and take and our orality makes us human. She has a lovely quote there.
Orality knits persons into community
Walter Ong
3) An aggressive template
iThempleti:Simple box is a template that creates an infobox in SS Wikipedia.
The instructions are clearly given. “A simple infobox that can be used for basically anything. All parameters are optional. If you don’t need all the lines, just leave them empty or delete them, but you must keep the brackets at the end.” For anyone who wants to experiment with an info box. this is a great spot to begin.
https://ss.wikipedia.org/wiki/iThempleti:Simple_box
On the workshop, I found myself showing advanced users how to use this template. But I quickly had to make an adjustment. Because the template contained a very aggressive photo of Jonas Savimbi, who played a big part in the civil war in Angola. Now I regard Wikipedia as a safe space. But that photo that a user (Naudefj) had inserted was unnecessary, and as a guest in that country, I don’t think that Wikipedians should have pictures of divisive figures in their templates. I removed (deleted) it. But the incident left me with questions about the relationship between Eswatini Wikimedian, Afrikaans Wikimedian and who was keeping this a safe space. our UNESCO hosts.
4) Some Concerns
I am in favour of Wikimedia’s decision to professionalise the paid work that Wikipedian volunteers do. It creates the conditions of employment and leaves much less room for doubt and dispute and protects both parties. I’ve done a few pieces of educational work overseas. When I worked in other countries, an international third party contacted me to do the work. They were very clear about what I could do or could not do and took responsibility for keeping me safe. We had a contract and there were parties to contact if things went wrong. In this visit, I went over to another country with a piece of code that was problematic. Essentially, Herschall and I and the SWiP team were bringing in aggressive code. With the many high-profile people this was a symbolic space where the king proclaimed he had no enemies? This code was a challenge. The second concern was that this work seemed to be part of a Wikimedia research grant See https://openreview.net/forum?id=zAoMpLW78U I am very much in favour of action research, but ethical consent from participants is necessary? I don’t think consent was given.
In return for these question, I was removed from the project. I must admit. I’m pretty bummed about that.