ISCA Archive SpeechProsody 2014
ISCA Archive SpeechProsody 2014

Certainty and uncertainty in Brazilian Portuguese: methodology of spontaneous corpus collection and data analysis

Leandra Antunes, Véronique Aubergé, Yuko Sasa

This work presents a methodology used to collect some spontaneous social affect corpus and preliminary prosodic analysis of certainty and uncertainty in Brazilian Portuguese. The corpus was collected by a Wizard of Oz (Emoz)method, the scenario to induce certainty and uncertainty is based on the situation of a job interview, for which a companion robot (Emox) is supposed to be a trainer. The subjects were convinced to benefit of a free training of this “revolutionary” method to train to job interview. In this scenario the linguistic expressions are partially controlled, in order to focus the certainty/uncertainty expression mainly on paraphrasing and prosody. Data were preliminary analyzed for audiovisual prosody: videos analysis were made regarding eyes, mouth and face/head movements, while audio analysis were made about acoustic prosody parameters of fundamental frequency and duration. The first results show that using Emoz within such a scenario is an efficient way to induct spontaneous but comparable speech production. Prosodic results show that fundamental frequency and duration measurements, as well as eyes, mouth and face/head movements, are differently used in certainty and in uncertainty production in Brazilian Portuguese.


doi: 10.21437/SpeechProsody.2014-10

Cite as: Antunes, L., Aubergé, V., Sasa, Y. (2014) Certainty and uncertainty in Brazilian Portuguese: methodology of spontaneous corpus collection and data analysis. Proc. Speech Prosody 2014, 110-114, doi: 10.21437/SpeechProsody.2014-10

@inproceedings{antunes14_speechprosody,
  author={Leandra Antunes and Véronique Aubergé and Yuko Sasa},
  title={{Certainty and uncertainty in Brazilian Portuguese: methodology of spontaneous corpus collection and data analysis}},
  year=2014,
  booktitle={Proc. Speech Prosody 2014},
  pages={110--114},
  doi={10.21437/SpeechProsody.2014-10}
}