Originally a talk about multilingual websites specially wikis at BarCampZurich 2006. ¹
Wikipedia is multilingual.
Three levels of users for the purpose of our discussion:
1. Pioneers develop something new (programming language, wikipedia) using their first language (eg. English around this time)
2. Early adopters are multilingual (eg. German and English) but stick to the language of the pioneers.
3. Main stream users using their first language which is not English.
There’s a mysterious transition between step 2 and 3, here: How did the first non-English Wikipedia get started?
Currently (October 2006) users of non-English languages have a problem:
1. Search Portuguese Wikipedia for Oddmuse and you’ll find no matches.
2. Search German Wikipedia for Oddmuse and you’ll find a single related page.
3. Search English Wikipedia for Oddmuse and you’ll find the page!
Search Portuguese Wikipedia for Oddmuse
Search German Wikipedia for Oddmuse
Search English Wikipedia for Oddmuse
Compare with Emacs Wiki, where all pages are in the same “namespace”:
1. Search for Brainfuck using the German Interface on Emacs Wiki. You’ll find English pages!
2. Search for Rekursiv Ersetzen. You’ll find German pages, too.
German Interface on Emacs Wiki
Compare with Stephanie Booth’s blog in English and French, where all multilingual features are conventions: She manually addes French and English abstracts to her posts.
Between total separation (example Wikipedia) and no separation (example Emacs Wiki).
For group 3 above, we want to filter languages in searches, feeds (or Recent Changes).
SisterSites introduces automation when browsing: Show pages with the same name on a few known related sites.
NearLink introduces more automation when linking: Switch linking targets between sister sites and the local site.
NearSearch extends the same mechanism to searching: Since the list of page titles is available from sister sites, these are always searched, and upon request, searches are run on sister sites and the results are merged with the local search results.
There’s more background on WikiNameSpace. When reading that page, just keep in mind that essential we’re talking about the same topic: It doesn’t matter whether a sister site differs in topic, community, or language used.
Eventually automatic translation might solve the problem. Right now translations are “poetic”: Original blog post in German compared to the automatic translation by Google.
automatic translation by Google
(Please contact me if you want to remove your comment.)
⁂
[fr] Une première localisation de ton billet a été posée http://wiki.crao.net/index.php/SitesWebMultilingues pour réactions éventuelles.
http://wiki.crao.net/index.php/SitesWebMultilingues
– ChristopheDucamp 2006-10-30 19:30 UTC