REPRESENTATION OF PROSODIC AND EMOTIONAL FEATURES IN A SPOKEN LANGUAGE DATABASE