This round table aims at discussing different aspects of the importance of prosody in spoken corpora compilation and annotation.⠀

Mettouchi will use her experience of compilation of lesser described languages corpora (CorpAfroAs and CorTypo) to highlight some methodological issues and to focus on prosodic segmentation and granularity of transcription. ⠀

Mithun will use Indigenous North American languages to show how the written bias must be fully overcome for a correct analysis of speech, since we can understand it only on the basis of the acoustic information conveyed by sound, mainly prosody, especially when we observe natural, unscripted speech. ⠀

A. Panunzi starts from the consideration that prosody is the basic level for any annotation of spoken corpora, and will present a resource built to compare different criteria of prosodic segmentation, in order to observe the level of agreement among them.