Annotation of Valence for Spoken Personal Narratives