Currently the order is alphabetic so Albanian is the default. It would prevent many errors if French was the default.
This currently applies to the creation of a new dataset.
When adding a new expression to an existing dataset the default is English. Usually the English already exists, so the default should be French as well.
I already mentioned this and created a redmine (Sorry I didn't know where to do this at that time). The redmine item was closed...
Bertrand is looking into it, he will get back on us on the feasibility.
You won't be notified about changes to this idea.