This work was funded in part by NSF grant ``ITR: Prosody Generation for Child Oriented Speech Synthesis''. The opinions expressed in this paper do not necessarily reflect those of NSF.
