Jan,
You bring up a good point. I feel like there has been a gradual shift
towards research across multiple language communities over the past few
years and that is starting to lead to some informal insights into this
question of transfer of findings across languages / cultures. First a few
examples in case you wish to explore yourself:
* Motivation / needs of Wikipedia readers across 13 different languages:
https://meta.wikimedia.org/wiki/Research:Characterizing_Wikipedia_Reader_Be…
* Motivation / behavior of new editors across Czech and Korean Wikipedias
(with some ongoing work in Vietnamese and Arabic Wikipedia as well I
believe):
https://www.mediawiki.org/wiki/Growth/Analytics_updates/EditorJourney_initi…
* Reading time across many wikis:
https://dl.acm.org/citation.cfm?doid=3306446.3340829
* Predicting aggregate page view in languages/regions where the takeaway
was that language was more important than geographic region when it comes
to predicting page views:
http://wikiworkshop.org/2019/papers/Wiki_Workshop_2019_paper_3.pdf
* Enabling page previews in English / German:
https://www.mediawiki.org/wiki/Page_Previews/2017-18_A/B_Tests
* Usage of the "Thanks" feature across a number of languages:
https://meta.wikimedia.org/wiki/Research:Understanding_thanks
* Effect on tourism of additional content about places in Dutch, German,
French, and Italian:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3046400
* Some data on the usage of blocks on various wikis:
https://meta.wikimedia.org/wiki/Community_health_initiative/Measuring_the_e…
* A bunch of data on the prevalence of anonymous editing on different
languages / projects:
https://meta.wikimedia.org/wiki/IP_Editing:_Privacy_Enhancement_and_Abuse_M…
* Scott Hale has also done some work on multilingual editing that might be
worth exploring:
https://arxiv.org/pdf/1501.00657v2.pdf and
https://arxiv.org/abs/1312.0976
* Statistics on content overlap across wikis:
https://wikitech.wikimedia.org/wiki/Wikidata_Concepts_Monitor#Wikidata_usag…
or
https://meta.wikimedia.org/wiki/Research:Expanding_Wikipedia_articles_acros…
In general, my personal views are:
* Language (presumably partially as a proxy for culture) is by far the most
salient aspect when it comes to understanding differences in behavior etc.
across wikis
* I hesitate to make broad statements about cultural differences, but there
are certain wikis that are more / less interconnected. For instance, there
is a good bit of overlap between Hindi Wikipedia and various other language
editions associated with India (Gujarati, Marathi, etc.). Same is true for
various languages in Spain (Asturian, Basque, Catalan, Spanish, etc.) and
Ukrainian / Russian Wikipedias.
* Obviously size matters a lot too in certain cases when it comes to
editing / maintenance workflows, though I would argue it's less of a factor
when it comes to reader behavior.
* Some wikis do have a reputation of being quite distinct -- for instance,
I would be hesitant to generalize anything to/from Japanese Wikipedia
because the statistics regarding interactions etc. there often look much
different than other wikis.
I would love to see some meta analyses that begin to look at similarities
in behavior or settings (e.g., AbuseFilters, rules around
ContentTranslation) across lots of different metrics to guide our
understanding of the similarities and differences between the language
communities. Until then, I would say there are going to be instances when
research on one wiki tells you little about how another wiki would react
(+1 to what Nemo says about even two is much much better than one language,
especially if they are much different languages/cultures). But there are
also often statistics you might pull up to make inferences around how
findings might transfer -- e.g., statistics on anonymous editing and
reverts might tell you something about how introducing new types of IP
blocks would play out in a new community.
On Wed, Oct 2, 2019 at 10:46 AM Federico Leva (Nemo) <nemowiki(a)gmail.com>
wrote:
Jan Dittrich, 02/10/19 14:35:
- How would you deal with such criticism,
particularly of the "if it is
not
about 'my' wp it is useless"-kind
[2]?
At a minimum, the research needs to have used methods which could extend
to multiple wikis. Being about 2 languages is ten times better than
being about 1 only, while being about 100 language subdomains is not a
hundred times more informative. Involving multiple languages helps go
beyond the language-specific and wiki-specific constructs (like
templates, workflows etc.).
Federico
_______________________________________________
Wiki-research-l mailing list
Wiki-research-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wiki-research-l
--
Isaac Johnson -- Research Scientist -- Wikimedia Foundation