Learning patterns/Understanding language-specific challenges to translation on Wikipedia
What problem does this solve?
Translation has been, and continues to be, an accessible pathway for new Wikipedians to learn how to edit and gain an understanding of community-moderated guidelines while actively contributing to global public knowledge equity goals. The Content Translation Tool–a tool that allows users to translate in a side-by-side display that automates processes like links, categories, and templates across language Wikipedias–has become a critical aspect for creating a user-friendly pathway for new editors on Wikipedia. However, the tool contains many language-specific challenges. How do we begin to address these challenges and implement them into future iterations of the Content Translation Tool?
What is the solution?
To begin to address this problem, it is critical to begin documenting the challenges that language-specific communities are facing. With the support of the 2021-2022 CCCC Wikipedia Initiative Graduate Fellows, this learning pattern seeks to improve understanding of the challenges facing translation to and from Arabic, German, and Russian Wikipedia to cultivate an understanding of how to improve the Content Translation Tool. In the language-specific challenges section, the 2021-22 CCCC Wikipedia Initiative Graduate Fellow cohort describes the challenges they have faced while using the Content Translation Tool throughout the fellowship period. In the recommendations section, the fellows delve into how to address some of these challenges and offer creative solutions they have learned throughout the fellowship period.
Arabic Wikipedia (User:AbirWard): The Arabic Content Translation Tool seems to be programmed to translate to Egyptian Arabic which does not work well for other Arabic dialects. The tool needs to be in standard Arabic that is readable by all Arabic speakers. The second problem encountered is that the tool does not allow translators with few(er) edits to publish the translations which can be demotivating for translators who are not committed editors but simply wish to contribute to the translation rather than to the editing (writing, editing, and moderating content). Bots seem to have done a lot of translating which is problematic because there are many articles that have been mistranslated. Finally, mistranslation in the article title itself due to the fact that certain bots use machine translation. I tried to find and correct the spelling of names that can be misspelled in Arabic, especially if translated from English by someone who is not familiar with the pronunciation of the name in the native language. For example, Najat Aoun Saliba’s name was miswritten in Arabic because of how different the pronunciation is between one Arabic dialect and another (Tunisian, Egyptian, Kuwaiti, Lebanese, or Syrian).
Russian Wikipedia (User:Sashaalexa): Content Translation Tool does not always recognize specific terms and cannot translate some terms which can be applied only locally. It makes the translation process faster, but I had to add and edit a lot manually to convey the meaning. The second issue is that the rules of Russian Wikipedia differ from English WIkipedia, and what seems to be correct for one Wikipedia does not work for another. Moreover, the communication and feedback is badly developed on Russian Wikipedia, which makes it really challenging to get an appropriate feedback on the translation.
German Wikipedia (User:Katlett): The translation tool did help make the initial translation process easier for me to get started translating pages into German. It made it easier to set up and jump right into translating a page that had not yet been created. However, as I researched how others were using the tool, it became clear that the tool was not only supposed to help with the establishment of a translated page, it should also help with suggestions for translations from the original page that I was working from. This never happened for me and therefore I was left trying to work through my own translations as well as navigate the German side of Wikipedia. I think the biggest challenges that I faced was understanding the interface of German Wikipedia once I was done with my original translation and trying to figure out how/if sandboxes work the same way. I think this is partly because while my language skills were strong enough to work in fields I am comfortable in, the more technical side in German Wikipedia was hard for me to understand.
Arabic Wikipedia (User:AbirWard): The best way was to use Google Translate to translate one sentence at a time, especially since my MacBook does not have an Arabic keyboard. Another helpful trick is adding Arabic by going to system preferences/keyboard/input sources and adding any keyboard one needs. Then clicking on the language and selecting “show keyboard viewer” one can bring up the keyboard of choice. Also, one can use Yamli.
Russian Wikipedia (Sashaalexa): In general, the Content Translation tool worked perfectly for me, and I edited the translation easily when it was necessary. I especially liked that this tool informs about the percentage of the machinery translated text, which is important for moving it into the mainspace. However, Google Translate is another great way to do it. I did not find any solutions to receive feedback from Russian Wikipedia, but I think that building connections with Wikipedia editors all over the world from different countries would be important. Also, creating workshops to talk about the rules on different wikipedia would be very helpful.
German Wikipedia (Katlett): The Content Translation tool never worked quite right for me and so I used Google Translate to help me as I worked to translate. I didn’t use it as a copy and paste tool, but used it as a guiding suggestion tool. I also Googled different terms to make sure I was using the correct terms.
We have not found–to our knowledge–a centralized place to document these challenges. It may be useful to create a centralized place for inquiries related to different themes or genres of challenges for different constituents throughout the movement. Although this is a learning pattern, we feel this would be more suitable in a space dedicated to understanding challenges and providing recommendations. We hope by documenting these challenges and providing recommendations, it will help provide useful support for future projects aimed at improving the Content Translation Tool.