Any preventive methods/tools for languages under attack?

Common Voice
#1

I’ve been in a meeting with language activists working on endangered languages in Turkey. Some of these languages are in preparation phase in CV.

They are afraid that their affords will be systematically vandalized, such as adding sentences mixed with other languages, hate speech etc.

On the other hand, the collaborative nature of the projects does not provide required tools for such occurrences. For example, addition of sentences does not require a login, so deletion of such sentences from a person/persons is not possible; reporting them using the report button (e.g. as “not in my language” or “inappropriate” for hate speech) will not be preventive etc. The only solution I can think of is 7/24 watch-dog groups trying to downvote them, which is impossible of course.

Are there any communities under such oppression? How do you deal with this?

1 Like
Dataset 17 Release
#2

I still believe that it will be much more ideal to give local core volunteers some level of reviewing before it’s out there for recording.

After we add “propose sentences” to the main site. I ended giving up reviewing them or add new sentences.

It’s just too troublesome and not possible to maintain the same quality as before, when we took full responsibility to commit or batch add sentences.

2 Likes
#3

This happened in the early phase of the Esperanto dataset. People added elephant trumpet sounds and similar stuff.

There was a lot of manual work necessary. The good news is that trolls normally do not have enough energy to keep this up for a long time and they do not systematically validate their donations.

4 Likes