Acquisition and Assessment of Semantic Content for the Generation of Elaborateness and Indirectness in Spoken Dialogue Systems