Each stem is sized by document frequency. Tick the ones you want and submit them as a vector-space query.