We look forward to receiving your signal!
Community across language barriers: Multilingual comments for swissinfo.ch
Can automatically translated readers’ comments foster exchange across language barriers? Together with SWI swissinfo.ch we examined the technical feasibility and social implications of this idea.
In a nutshell
- In this exploratory project, we investigated how automatically translated comments might work on SWI swissinfo.ch’s multilingual news website.
- We conducted user research and assessed the technical feasibility of the idea with regards to its specific context.
- Thanks to our prototype, SWI was able to make a low-risk, cost efficient evaluation of the idea.
SWI swissinfo.ch is the international service of the Swiss Broadcasting Corporation (SRG SSR in German/French). SWI offers global news with a focus on Switzerland in ten languages, helping Swiss citizens abroad to stay in touch with what is going on back home.
The comment function of the platform is key to fostering exchange. However, comments are currently tied to a single language, so that conversations only take place between people who speak the same language—despite the fact that SWI offers many articles in several languages. The goal of this exploratory project was to find out how automatically translated comments might facilitate discussions across these language barriers.
A comment function that allows for automated translations in ten languages is a complex undertaking. We had to consider both the technical context of the portal as well as the habits, attitudes, and needs of the readers.
For instance, what happens if a bilingual person sees a translated comment even though they understand the original language? How should we deal with general skepticism towards automated translations? How can we make sure that machine translations keep their original, nuanced meaning, don’t accidentally provoke, and enable real exchanges?
To answer these questions, we helped SWI clearly outline the problem context and created a prototype to explore how automated multilingual comments could work. In order to understand the readers’ attitudes, we evaluated statistical data, defined user groups with SWI, and conducted several user testings.
Based on the testing results, we then formulated recommendations for the development of the comment function. For example, we found out that it made sense to display comments from other languages by default in translated form. This makes it easier for readers to get an overview of all comments and to react to them.
To evaluate the technical feasibility, we examined how automated translations could be integrated into SWI’s current technical environment. Additionally, we looked into how the moderation of reader comments can be improved. Currently, all comments are manually checked within 24-60 hours. This ties up a considerate amount of resources in a small unit like SWI. In order to reduce both the delay and the time investment, we developed a plan for semi-automated moderation, where comments are automatically checked for spam and toxicity.
Ultimately, SWI was able to explore the idea and test its technical feasibility as well as its social implications. As this is an exploratory project, it is still unclear if and when the solution will be implemented. Either way, however, this project once again showed that prototyping and testing are incredibly valuable for exploring and evaluating an idea cost efficiently and with little risk.
Do you have further questions about the project or our offering in general? Get in touch with Gloria-Mona Knospe (firstname.lastname@example.org).