Leveraging Speaker Context For Natural Language Processing