It looks like we retain multiple native language/ language exposure columns in the combined demographics. See if we can unify this in 002b