Research:Languages Onboarding Experiment 2024 - Executive Summary

Created
13:36, 1 April 2025 (UTC)
Collaborators
Objective co-owners

Runa Bhattacharjee, Fiona Romeo, and Ben Vershbow

Key Result co-owners

Mary Munyoki, Satdeep Gill

Hypothesis owners

Srishti Sethi, Tim Starling, Satdeep Gill

Subject Matter Expert

Amir Aharoni

Research Support

Caroline Myrick, Krishna Chaitanya Velaga

Program Management

Lauren de Lench

Technical contributors
Amir Sarabdani, Tim Starling
This page documents a completed research project.


The Wikimedia Incubator provides a development space for new language projects before they graduate to full production wikis. However, challenges such as outdated editing tools and lengthy approval times hinder progress. The Future of Language Incubation initiative tested whether moving incubating languages directly to production wikis with modern tools would increase editing activity.

Languages on the Incubator were grouped based on activity levels and randomly selected while incorporating feedback from the Wikimedia Language committee. The five language Wikipedias were selected for the experiment were Southern Ndebele (nrwiki), Iban (ibawiki), Obolo (annwiki), Tai Nüa (tddwiki), Pannonian Rusyn (rskwiki). More details about the selection process are available here.

These five languages were moved to production wikis on 14 October 2024 after approval from the Language Committee, with a three-month experiment period ending on 12 January 2025. In November 2024, we were able to interview contributors of all wikis except Tai Nüa, so that project is not represented in the qualitative analysis.

Key Findings

edit

Quantitative Analysis

edit
  • Editing Activity: Contrary to our expectations, most languages experienced a decline in editing activity after moving to production wikis, except for Southern Ndebele, which saw a slight increase.
  • Article Creation: Only Iban Wikipedia showed a significant increase in the number of articles created, while others either remained stable or declined.
  • Mobile Editing: Usage was generally low (below 5%).
  • Content Translation Tool: Low adoption, except for Southern Ndebele, where it was used for 25% of articles created or expanded.
  • Byte Difference Per Edit: Most wikis saw smaller edits post-graduation, except for Pannonian Rusyn and Obolo.

The overall finding was that moving languages to production wikis did not significantly improve editor productivity, compared to the Incubator control group.

Qualitative Insights

edit
  • Online Presence Matters: Languages with existing digital infrastructure (e.g., Pannonian Rusyn) had a shorter incubation period, while others (e.g., Southern Ndebele) took over a decade to graduate.
  • Incubator Experience: Some contributors found the Incubator useful for learning, while others struggled with unclear documentation and complex navigation. An Obolo contributor said, “In the Wikipedia world, you have to read a lot to figure out how things work, and sometimes even refer to Quora articles or external Wikimedia links.” A representative of Wikimedia User Group Malaysia said, “The criteria for becoming eligible [for graduation to a production wiki] are sometimes unclear. One contributor kept contributing, only to later realize it wasn’t enough.”
  • Technical Barriers: Content and Section Translation tools are hard to discover and not automatically enabled. Template and infobox issues require advanced technical skills. An Iban contributor said, “(I would like to receive) support fixing the Content Translation feature in Wikipedia. The references don’t show up (and there are) blank templates.” They also said that, “Since Iban is in Google Translate already, machine translation should be added to Content Translation.” (Further enquiry revealed that the machine translation support for Iban using Google Translate was deployed on 10th Feb 2025.)
  • Community Sustainability Challenges: Communities struggled to retain editors post-graduation, particularly in the Global South. An Obolo Wikipedian noted, “Since Wikipedia is a free encyclopedia, convincing people to join required some effort. Many people even asked if there was money involved!
  • External Support: Guidance from Wikimedia User Groups, Language Committee members, and Wikimedia Foundation staff played a crucial role in community success. A Pannonian Rusyn contributor highlighted that Wikimedia Foundation staff member, Amir Aharoni, supported the community by helping them navigate both on-wiki and off-wiki challenges, “Amir helped with everything. From writing on Incubator and the first translation on TranslateWiki to everything.” The contributor also mentioned the support received from Wikipedians in Serbia for registering their language with SIL International.
 
Average weekly edits per editor, Languages onboarding experiment 2024

Conclusions & Recommendations

edit
  • Technical Improvements Alone Are Insufficient: Simply moving languages to production wikis does not guarantee growth—more social and community-building interventions are needed.
  • Focus on Expanding Editor Base: Future efforts should prioritize recruiting and retaining new editors rather than assuming that existing editors will  increase their activity after moving to a production wiki.
  • Improve Support for New Wikis: Better documentation, onboarding resources, and automated tool enablement would ease the transition from the Incubator.

While the experiment provided valuable insights, further research is needed to refine strategies for sustainable growth of new language communities on Wikimedia projects.

Full report

edit

The full report is available on the following page: https://w.wiki/DfVy