THanks Mark.
We are using fields for attribute where we also have non_analyzed representation.
But if I go with the approach for analyzed content won't I add more indexing pressure/document given the case is each document represent only only language. Unnecessarily I will be indexing the same content against multiple fields with different analyzers. I was thinking of having different attributes for each language where the attribute not representing the document language can be ignored.
NOTE: I have many not_analyzed attributes other than the main content attribute
I also would avoid creating multiple indice for each language as English represent 80% of the content