A complete computer simulation of a linguistic component like intonation would involve three rather different tasks. First, it would need to produce semantically appropriate selections from the available inventory of morphemes, on the basis of communicative intentions. Second, it would provide the phonological shapes of these morphemes, and combine them to form a complete phonological representation of the intonation of the sentence. And third, it would translates this phonological representation into correctly timed F0 events.