摘要：The aim of this article is to present some methodological remarks about how theAMPER (Multimedia Atlas of Romance Prosody Space) acted-speech corpusshould be designed in the case of languages that, like most of the Romanceregional languages of the Italian Peninsula, do not have a standard oral form.Thanks to the experience gathered while collecting prosodic data for Friulian, ithas become clear that for this type of languages the preparation of the acted-speechcorpus requires the researcher to take into account various aspects of diatopicvariation (basically lexicon, segmental phonetics, morphology and syntax). Inaddition to a solid prior knowledge of such aspects in the different dialects, theresearcher will also need the co-operation of local informants (native speakers ofthe dialects to be studied) that can help him/her to prepare the sentences of thecorpus in question.
