There are various phenomena that are characteristic for Conversational Speech. In spontaneous conversations, we often reduce articulatory effort or speak in dialect (“kanni net machn”), since sentences are planned on the fly, they often result to be grammatically imperfect (“so wie das Veranstaltung da”) and also disfluent (“also wir haben wir haben eines wir haben eines gekauft ein uraltes”). As humans, we are usually still able to decode (understand) such imperfect utterances. One specific type of disfluency is when we insert fillers in disfluent structures (e.g., aaahm, also, halt). The literature provides somehow contratictionary evidence, as some studies found fillers to be important cues for parsing utterances and that their useage reduces the cognitive load of the listeners, but others find that high filler-rates correlate with lower task performance (memorizing tasks). The aim of this project is to investigate whether the insertion of fillers is perceived as disturbing or rather as helpful for a human to parse the utterance. For this purpose, the tasks of this project are to select stimuli from a conversational speech corpus, to conduct a perception experiment with human participants, and to analyze which acoustic properties of the fillers correlate with how strongly disturbing the token was perceived.

Contact:

Barbara Schuppler (b.schuppler@tugraz.at)