Content translation/Community Consultation June 2016

Content Translation is an article creation tool that has been in use since January 2015. Content translation helps users translate existing articles from one Wikipedia into languages on other Wikipedias where the article does not exist. It is used by editors to write many new articles, increasing participation by new users and also renewing active editing in many Wikipedias. Translated from existing articles, 88000 new articles with considerably rich content have been written using this tool. Content Translation is an opt-in beta feature. We are now preparing to take Content Translation to a bigger group of users by taking the tool out of beta feature on a few Wikipedias. We prepared a preliminary list with the following wikis: Arabic, Catalan, Chinese, Hebrew, Indonesian, Japanese, Norwegian (Bokmal), Persian, Portuguese, Punjabi, Russian, Spanish, Turkish, and Ukrainian. This set of wikis were selected to create variations in size, language groups, and usage patterns of Content Translation. The community consultation is being conducted on the 14 languages listed.

Based on an earlier consultation with the editors of the Catalan Wikipedia, the earliest adopter of this tool, the Content Translation team have prepared a preliminary set of requirements that we consider should be fulfilled before the tool is ready for use outside of beta. To focus on the features and concerns that are most important for users of Content Translation, we would like to get more feedback. We are hosting a consultation from June 8, 2016 to June 22, 2016 on this village pump.

As we work on the tool it is being improved every day, but we are particularly looking to better understand the issues in Content Translation that make the tool unusable or add extra workload for our editors. Please let us know on this topic thread. During the conversations, we may point you to Phabricator tickets for ongoing work, so that you can track the status of work. For new issues we will create new Phabricator tickets to help the development team keep the details documented. After June 22, 2016 we will go through all the feedback and follow up with a summary, and an updated plan for Content Translation’s move out of beta.

Consultation Period

Start: June 8, 2016 (local time)
Ends: June 22. 2016 (local time)

Discussion threads

Timeline

Date	Activity	Status
June 8, 2016	Announcements posted to village pumps	Done
June 8, 2016 - June 22, 2016	Consultation period	Done
June 23, 2016	Thank you notes posted to village pumps	Done
June 23-30, 2016	Feedback is reviewed by Language team	Done
July 4, 2016	Summary posted by the Language team (twitter link)	Done

Summary

General comments

Of the 14 wikis where the consultation was initiated, we received feedback from 5 wikis. In spite of the consultation announcement requesting for comments on the thread itself, several users asked where they could provide feedback. This may have been due to unclear messaging on our part or due to a deviation from the way earlier consultations were run. The team will review this shortcoming to avoid repeating this scenario in future.
The Norwegian Bokmal discussion used 3 categories for the comments and this structure can be used for similar consultations in future. The categories were:
- Functional deficiencies (defect) - i.e. the tool is not working as per expectations due to a bug
- Future wishes (feature requests) - i.e. new features to be added to the system
- Non-functional requests (general comments) - i.e. subjective requirements. Eg. ‘a translation should be good is a non-functional requirement.

Feedback

Editors emphasised on the need to reduce wiki text errors in the translated articles content so that they do not need to spend too much time cleaning them up after publication.
Better handling of templates were requested with several examples like incomplete adaptation, ordered lists, cite web to avoid translation errors.
Several issues related to reference adaptation were highlighted, including cases where the references were copied multiple times and left behind unnecessary HTML tags that had to be cleaned up. An example was also provided that showed the amount of effort that was needed to fix such an issue.
Miscellaneous other issues and comments:
- Section headings are sometimes observed to have not been translated.
- A translation proof-reading view was requested by editors of Norwegian Bokmal Wikipedia, so that the reviewer can identify errors and also leave remarks.
- Very wide tables do not fit inside the Content Translation columns. (Existing ticket)
- An editor requested for added levels of permission to use the tool, while another editor mentioned that editors should have adequate familiarity with both the source and target languages, and warnings should be displayed to users about this.
- An editor mentioned that the paragraph section divisions were not convenient for short sentences, while another editor mentioned including a wiki syntax editor.
- From the Spanish Wikipedia discussion, an editor mentioned that it would be good to have a way to import the previous history of the source article. Existing ticket
- Changes in the source article need to be better handled to be reflected inside Content Translation.
Tickets filed:

Please note: The above comments have been summarized from the discussions threads. The feedback received will be further evaluated and aligned with the development goals to improve and prepare Content Translation for wider use. More details on this page.