Advancing corpus-based analyses of spontaneous speech: Switch to GECO!

We present the GECO (GErman COnversations) database, which was designed for investigating phonetic convergence in German and its correlation with personality traits and scores of mutual likeability and esteem. The data can more generally be used to investigate aspects of interlocutors ' behavior in free, spontaneous dialogs, including social aspects. The database will be freely available for non-commercial use and to our knowledge will be by far the largest German database of this type. GECO consists of 46 dialogs of approx. 25 minutes length each, between previously unacquainted female subjects. It further contains results of personality tests aimed at subjects ' extraversion, acting behavior, other-directedness and sensitivity to social cues (all collected once per subject), as well as participants ' mutual ratings in terms of competence and likeability (collected after every dialog). 22 dialogs took place in a unimodal (UM) setting, were participants were separated by a solid wall and could not see each other, while the remaining 24 dialogs were recorded with subjects facing each other (multimodal setting, MM). All dialogs were recorded in high quality (separate channels, 16 bit@48 kHz) in a sound-attenuated booth using AKG HSC271 headsets with rubberfoam windshields. In the MM setting, a transparent screen ensured sufficient speaker separation between the two channels. We had 12 speakers in the UM condition, and 8 (7 from UM, 1 additional speaker) in the MM condition, which was recorded 5 months later. All subjects were females between 20 and 30 years of age, mostly students. They were paid for each dialog they participated in. Subjects were naïve to the research questions; they were told that the purpose of the study was t