Local languages/Translation
 |
This is a draft document. Please feel free to make edits, and understand that content may substantively change. If appropriate, discuss at Talk:Local languages/Translation. |
Problem description
Translation of articles is one way in which Wikimedia projects in local languages can grow. To organize a team that translates any significant amount of articles to some or all languages would require an enormous team and amount of time. There is however things that can be done to simplify the translation process, increasing the likelyhood that anyone would volunteer to translate an article. And increasing the amount of volunteering translators by any percent would be a much more effective approach than to organize actual translation by a few translators.
Today translation of articles can be done by looking up the source of an article, locating or creating the corresponding article in the new language, copy the source from the source article to the edit box of the new article. Then actual translation as well as location and replacement of links can be done. Alternatively the source of the original article ain't copied, but instead the original article read and translated in the edit box of the new article. In each case the process puts a responsibility on the translator to handle links and formating of the article that might stop many people that are very able to translate but find the extra work of these other tasks to difficult or cumbersome.
If the process involved in translation can be simplified so as to put a minimum of other duties than actuall translation on the translator, the number of people willing to translate would probably grow. Thereby such a solution would also stimmulate the growth of smaller Wikimedia projects.
Another problem that arises when articles from large Wikipedias are translated into smaller language Wikipedias is that errors get copied. In large language Wikipedias the errors are quickly corrected but on the smaller ones it stays for a long time, because there are less people able to correct it. More effort into indicating the quality of large language articles could help translators judge the value of the article they translate.
Strategy for simplifying translation
Translation interface/toolkit
Translation could be simplified by automating some steps of the translation process and present translators with a simpler interface. Add an option in the left menu that reads "translate" that takes anyone into a page where which language to translate into can be choosen. When the language has been choosen, let the MediaWiki software replace all links that has a corresponding links in the new language by those corresponding links, and present to the translator a source that has the least amount of markup language possible. Some simple way of toggling between the new and original article and the source of the new article might be useful. Another possible feature for the translation interface could be to incorporate some machine translation toolkit that can be used by the translator if it seems to help.
Such a solution means a small addition to the source but could have very big effects even if it just rises the number of people that translates by a small fraction. It is a solution that every Wikimedia project would benefits from, but that still require a minimum of work organized by the WMF.
Some obstacles: Even though the solution would make it easier for translators to translate articles there are some obstacles that has to be considered. What about if there allready exists an article in the language that covers the subject in a different but less extensive way? Are there ways to incorporate the information from the existing article into this. And articles that are pure translations of other articles might not even be what is wanted, maybe the content actually should have some approperiate local twist. So policies about what kind of articles that actually can be translated with this method might have to be worked out. Further, the wording of sentences might often be so different that the placement of the links might have to be rearanged. Maybe two links have to appear in opposite order and the actual text of the link might also need to be bent in another way than the automatic machine replacement of links would do. The later problems does however only require that the interface is sufficiently advanced to allow to deal with such problems.
Quality indicators
The problem of copying of errors could be addressed by some kind of quality indicator for articles that makes it easier for translators to judge the value of a certain article. Also when a corrections is made to an article in one language there might be a good idea of having some way to notify the other articles about the change. However, the articles in different languages might have reasons to differ from each others and a notification of every change in every language may just create an irrelevant flood of such notification to all other articles. Some way of notifying other language articles about factual corrections or major changes or additions to article content might be a good idea. Could each article have a list of such corrections made in other languages, with possibilities to filter out changes made in given languages?
Note: The following proposals has considered creating a simpler editing interface in general:
The usability initiative (http://usability.wikimedia.org/wiki/Main_Page) might be an answer to this. From a local language perspective there is along the line of the bandwidth recommendation also important that such a solution takes into consideration that it should be quick to operate even on slow internet connections.