I’m working with people in Morocco to get Amazigh contributions, any change to add it to the list in pontoon?
I’m not a language expert, but as long as languages (and not accents within a language) are requested on pontoon, there should be no problem.
Just flagging because according to wikipedia Tamazigh is a set of different Berber languages (so maybe individual ones should be requested).
I’ve spoken to some native Amazigh people today in Morocco and they’ve told me that Amazigh is the language and the rests are dialects. But they’ve also told me that some don’t understand each other, so a bit hard to tell. But I guess (because we anyway won’t be able to collect a very large corpus) it might make sense to define Amazigh as language and the reset as dialects.
E.g. finally, DeepSpeech can be selectively trained on the language or on dialects only, correct?
Before initiating any effort it would be wise to get the opinion of a linguistic about this. We should make sure we enable recognized languages with their official locale code.
@jbeatty Is there a resource that can help us here to take a decision?
You can train on whatever you inject it
I’m from Kabylia. I’m involved with Mozilla localization since years and now with Common Voice. We are a team of Kabyle localizers and activists who are working on the kab locale.
As I know, Morrocan Standard Amazigh Language is not yet fully standardized. We discussed about it last January in Tangers - Morroco with language activists from Morroco.
I recommand you to launch separately the corpus of the three major Berber languages Tarifit, Tacelḥit and tamaziɣt. Common Voice deals with natural.human language.
I recommand you also to use the latin Script as Kabyle. Please get in touch with people involved with linguistics studies in Morroco to know more about the deal.
If you are only interested with Common Voice, I think you have to choose one of these codes. You should not gather different language structures and phonetics within one Corpora, that’s why I’m asking you te get in touch with students/researchers dealing with Amazigh languages.
Thanks for the information @taqbaylitassa, really informative
By the way I’m Belkacem Mohammed, the admin of the kab locale on Pontoon. I lost the last access
thanks a lot @taqbaylitassa let me get 5K sentences in at least one dialect and come back to you, good point with the latin script since most of the contacts we’ve found can’t write or read the native script anyway…
In which variant? I have contacts from both Tachelḥit, Tarifit and Tamaziɣt speakers from Morroco which can help to validate sentences. Please would ask your contacts to see the kabyle localization of Common Voice? They could a lot of things to share with some of them, Tacelḥit for example.