Since many browsers send the list of preferred languages in Accept-Language HTTP header in every HTTP request, I would think there must be huge datasets somewhere from which one could get statistics such as:
- how many users actually list more than one language?
- what's the distribution of language counts?
- which language combinations are most common?
- how many users go into much detail with hand-crafted q-values and/or multiple regions per language?
But I did not find any.