Content translation

From MediaWiki.org
Jump to: navigation, search
Content Translation tool showing a warning when the user rely too much on automatic translation.


The Content Translation tool is aimed to improve multilingual contribution. Content Translation allows editors to create translations right next to the original article and automates the boring steps (no more hunting for links, categories, or initial translations).

Translating content present in one language version of Wikipedia to another Wikipedia is a process that involves different steps and tools. Creating a new Wikipedia page based on an existing one from a different language normally requires the use of automatic translation services, dictionaries, reformatting text, tweaking links and references, and a lot of tab switching. This is a process that can be streamlined a lot so that translators spend their time creating high-quality content that reads naturally in their language.

Content Translation is a Computer Assisted Translation (CAT) tool aimed to make translation of Wikipedia articles easier and better. It integrates other tools such as dictionaries or machine translation services, and it can be extended to include more. Content Translation complements the existing Translate extension: while text from Wikipedia menus and other user interface elements are translated and kept in sync by the community using Translate, Wikipedia content can be translated using Content Translation.

Most Wikipedia users edit only one language edition. However, just over 15% of users also edit multiple language editions. These multilingual users were found to be more prolific than their monolingual counterparts, making about 2.3 times as many edits on average.[1]

Try the tool[edit | edit source]


Screncast showing how to use Content Translation

Content translation is in a beta stage on some Wikipedias. This means you need to enable it first in order to try the tool:

  1. Log-in into Wikipedia (or create an account if you don't have one already)
  2.  Access the Beta settings of any supported Wikipedia (Afrikaans, Aragonese, Asturian, Bashkir, Basque, Bulgarian, Catalan, Chuvash, Danish, Esperanto, French, Galician, Gujarati, Indonesian, Kannada, Komi, Komi-Permyak, Kyrgyz, Lezgi, Macedonian, Malay, Minangkabau, Norwegian (Bokmål), Norwegian (Nynorsk), Occitan, Punjabi, Portuguese, Serbo-Croatian, Simple English, Slovenian, Spanish, Swedish, Tatar, Tuvan, Udmurt, Ukrainian, Uzbek, Vietnamese, Xhosa, Yakut, and Test-wiki) and enable Content Translation Beta feature.
  3. Access your "contributions" page, and select "translation" from the "New contributions" list. Pick an article missing in your language and translate it. When you are happy with the result, you can publish it as a new article.
  4. On some languages, translations are published to the user namespace so that you can review some possible issues, and you can move the page to make it a regular article anytime.
  5. Provide feedback.


What if my language is not supported?[edit | edit source]

Access Content Translation on our beta servers. You may need to:

Content translation is available in our testing servers for you to try it and tell us about your experience: Pick an article missing in your language and translate it. Don't be afraid of publishing the result since it will only be published in our test servers, but feel free to copy it to a real Wikipedia if you are happy with the result (and add it to the list!).

Currently only a few language pairs have automatic translation support (we'll support more over time). While Machine Translation speeds the translation process a lot, the tool can be useful for other languages too. You can adjust the URL parameters to select a specific article and set of languages. In the example below, article Australia is translated from English (en) to Burmese (my) :

http://en.wikipedia.beta.wmflabs.org/wiki/Special:ContentTranslation?page=Australia&from=en&to=my

This provides a translation side-by-side interface for you but you still can't receive machine translation for languages, that wasn't opted-in.

Purpose of the tool[edit | edit source]

Content Translation allows you to create an initial version of a Wikipedia page based on an existing version from a different language. So the main scenario we are addressing can be expressed as: A user creates a new page for language X based on the page for language Y, where languages X and Y can be any pair of different languages.

The tool is targeted to users that know two or more languages, and provides value for different kinds of users. For new users, the tool will provide a new way of contributing that is easier than creating a new page from scratch. For experienced editors, the tool will simplify the process of translating content.

We expect that Content Translation will help to expand the sum of the human knowledge to more languages, facilitate the process to translators so that they can focus on the quality of content, and simplify the process to make new users to consider translation as a simple way to get started contributing to Wikimedia projects.

The tool was designed according to the following principles:

  • Save time. Help translators to create content quickly, without unnecessary copy&paste to external tools.
  • Provide assistance. Prevent errors, and make the user confident of their translations.
  • Encourage quality of translations. The tool should communicate properly the purpose of translations in Wikimedia context and help the user to avoid bad quality translations.
  • Don’t force the user. Since different translators follow different editing patterns, the aids provided should not intrude into the editing process.
  • Focus on content. Translation is more focused on content than text styling. Technical elements such as wikitext should be dealt in a way that do not make the translation harder.

More details about the analytics considered are available.

Development[edit | edit source]

The extension is being developed as Content Translation extension

Design ideas[edit | edit source]

During the design stage many ideas have been explored. We also conducted several tests with users who suggested ideas that we collected and added when creating the prototype. The main components envisioned to be part of the UX designs are the Translation Center, Translation View and the Entry points leading to them. The Translation Center would serve as a dashboard for users to track translations done by them. The Translation View would consist of the editing area along with translation tools that will be available for use. For the initial round of development, the Translation View has been given priority.

Initial exploration of design ideas for Content Translation

Translation Center[edit | edit source]

The Translation Center will provide the users a comprehensive view of their article translation activities and track progress. Users will be able to view article suggestions and initiate translations for new articles also from here. View prototype (Try to press New translation and search for "Mies van der rohe")

  • Initiate translations
  • List of translations and suggestions

Translation View[edit | edit source]

The Translation View will contain an editing area and translation tools like Translation Memory and Machine Translation suggestions, dictionaries, and glossaries. The editor will include smart features like link adaption using Wikidata that will help users get the language appropriate links for inclusion in the new articles.

View prototype
The key features of the Translation View are:

  • Three-column layout.
  • Automatic translation as template.
  • Warn automatic translation abusers.
  • Adaptation of links and references.
  • Information about words and suggestions for alternatives.

More design details for Translation View

Additional entry points[edit | edit source]

The entry points will indicate to the users that translations can be initiated for specific pages for languages that may be appropriate for the user (based on factors like geographic location) These entry points are:

More design details for entry points

The UX design ideas have been significantly improved and additional details added since development of the front-end has been initiated.

Translation Tools[edit | edit source]

Several translation tools are being considered for inclusion in the Content Translation tool. The primary objective is to aid the user with Computer Assisted Translation (CAT) tools that can significantly decrease their effort while creating articles from existing content. The tools in consideration are Machine Translation (MT) engines, Translation Memory (TM) systems, bilingual dictionaries, glossaries, language proofing tools such as spellcheckers. The usefulness of these tools may vary according to the degree of support for individual languages and user's familiarity and proficiency with the source and target languages.

Development and research notes about each of these tools can be found in this page

Workflow and Technical Architecture[edit | edit source]

Main page about the technical architecture and workflow

CX ArchitectureV1.svg Content Translation Sequnce Diagram.svg

Development Plan[edit | edit source]

The development milestones and upcoming plans can be found in Roadmap

Future plans[edit | edit source]

Some aspects are not part of the scope for the first iterations of the tool, but we have them in mind for the future. The include:

  • Real-time collaboration on a shared translation.
  • Add content to an existing translation.


How to participate[edit | edit source]

  • Try the tool
  • You can volunteer for our testing sessions. It may take some time for us to contact you since we select participants based on the languages that we are planning to enable next. In any case, we are interested in testing with as many users as possible and we'll appreciate your participation.
  • Take the new language survey where you can provide information on the support of existing tools for your language.
  • Provide feedback at the talk page.

All related pages[edit | edit source]

See Content translation/Documentation

References[edit | edit source]

  1. Multilinguals and Wikipedia Editing, Scott A. Hale. Dec 2013