MeatballWiki | RecentChanges | Random Page | Indices | Categories

The goal for a MultilingualWiki is to integrate language minorities: We want to invite them to participate, and we want to make our know-how available to users of other languages.

This page discusses the mental model used for a a true multilingual wiki: One wiki with pages in different languages, and software tools in place to segregate the various languages from each other, if desired.

Target audience: single-language users

Assuming that the users only speak one language well, the importance of software tools increases significantly. We need RecentChanges filtered by language, we need integration with some online translation engine to machine-translate text available in only one language to the other languages.

Target audience: multi-language users

Assuming that the users speak most languages reasonably well (such as a Swiss wiki where most users can read at least two languages, even if they only speak one of them fluently), the need for software tools is not significant. Translation is not that important -- users can read most of the site in the various languages.

Write New Pages or Translate Existing Pages?

One of the first urges seems to be the translation of existing pages. It is not clear, however, whether that is the best approach both in terms of motivation and culture. Perhaps it makes sense to grow a translated set of pages organically. Maybe the sitemap is similar, maybe it is not. Maybe only certain topics on a page have been translated, others not. How to deal with such partial translations?

Setting the translation of existing pages as the goal of the translation effort may have unexpected consequences, such as cementing the language hierarchy -- after all, why write new pages in a language when the old ones haven't been translated yet? Encouraging diversity of sitemaps and topics on particular pages might even generate new ideas for the other languages.

Normally, translation deals with static documents; when the original is modified, the translation is updated to reflect the changes. Wiki pages, however, are dynamic, with contributors constantly adding, editing and reshaping. Thus, translation introduces a dilemma: should wiki contributors try to keep the translations in sync with the original, or let each document evolve separately in its own language? If the former is chosen, the translations become static mirrors of the original, nullifying the wiki advantage of collaboration for these versions of the document. On the other hand, the latter course leads to the forking of documents across the various languages of the wiki.

Avoiding the Big Wiki Hole

Assume a wiki for every language. Switching from one wiki to another to find some information on a topic requires a tiny bit of work such as clicking on a bookmark and searching for the topic or hacking the URL. Now assume that a lot of people can read a certain language which is not their native language, and the information they look for is available in this other language. This happens a few times. What will those people do? The hypothesis is that these people will switch to their non-native language, and therefore contribute less, since they can get the information they are looking for by reading their non-native language, and they will contribute less in their non-native language.

Nobody knows whether the assumptions or the hypothesis are true. In order to avoid this Big Wiki Hole -- the one wiki in the one ubiquitous language which attracts all contributors -- the MultilingualWiki approach tries to reduce the cost of switching languages. Integrating all the pages into one multilingual wiki makes searching and finding information in your non-native language easier, but encourages you to contribute in your native language.

Reader/Writer tension

Discussions about a MultilingualWiki primarily focus around the ability of readers, typically casual readers, to find content pertinent to them. The issue of organizing information on a wiki for the ease of consumption by casual readers is an ongoing problem regardless of how many languages are used, even one. While it is easy to organize static content that you can survey and analyze over time, it's much more difficult to organize dynamic content that may shift before you fully can comprehend it. Consequently, while it's much easier to consider organizing brochureware for multiple languages, including achieving parity across languages, this task is incredibly difficult for a living corpus. Just maintaining synchronization between pages is a sisyphean task.

Conversely, content generation isn't automatic. Someone has to write the content, and the ease of content production must be considered, as well as the commensurate feedback systems that encourage further writing. It's discouraging to join a seemingly pluralistic society only to discover that one subculture is overly dominant. For instance, as is often the case, the English (Japanese, German, etc.) version of pages may be where the interesting discussions and best ideas happen, encouraging those who even don't speak English as their first language to contribute to these pages. This leaves the counterpart pages in other languages to continuously lag behind, and the role of catch up editor is an unenticing one.

Essentially, the current discussion centres around culturally-neutral (CulturalDimensions notwithstanding) texts such as user manuals. This problem is alleviated as the corpus is written in the NeutralPointOfView for nearly equal consumption by all cultures. For these projects, the emphasis is really on the casual reader and not on the writers or the thinkers, and simple architectural methods (a TechnologySolution) for creating a MultilingualWiki are ideal. Conversely, any final answer shouldn't put too much emphasis on the software to create a pluralistic community, instead leaving that to a CommunitySolution as it is an entirely social problem.

Related projects

Babelingo [1] is a project by several community radios to support multilingual radio.

Cross Lang Wiki [2] is a project at the National Research Council of Canada [3] that is investigating features for supporting multilingual wiki content.

Contributors: BayleShanks, AlexSchroeder, StephenGilbert, SunirShah

CategoryWikiTechnology CategoryGlobalization


MeatballWiki | RecentChanges | Random Page | Indices | Categories
Edit text of this page | View other revisions