Using machine learning for system-internal evaluation of transferred linguistic representations