Comparing Annotations of Non-verbal Vocalisations in Speech Corpora

Abstract

In this study eleven corpora of spontaneous and scripted speech (in English and in German) are analysed regarding their annotation inventories of selected highly frequent nonverbal vocalisations (NVVs). It appears that only one corpus considers all NVVs and that laughter is the only NVV annotated in all corpora. The findings lead to a discussion of possible reasons for this situation. In conclusion it is argued that a wider distribution and more consistency is needed with respect to the annotation of NVVs.