|
Audiovisual cues to a speaker's mental state: methodological issues| old_uid | 1769 |
|---|
| title | Audiovisual cues to a speaker's mental state: methodological issues |
|---|
| start_date | 2006/11/16 |
|---|
| schedule | 13h30 |
|---|
| online | no |
|---|
| summary | Audiovisual prosody can be viewed as the whole gamut of features that do not determine what speakers say, but rather how they say it. It refers to a combination of auditory features such as speech melody, tempo, loudness and voice quality, and visual features, such as eyebrow movements, head movements, hand and arm gestures, and posture. Previous work has brought to light that audiovisual prosody is useful in spoken interactions: it can signal various kinds of communicatively relevant information, such as accents, boundaries of speech units, feedback or a speaker's confidence level. In this talk, I will present the first results of ongoing research on how audiovisual prosody can serve as a window on a speaker's mental state. I will focus on methodological issues that come into play when wants to investigate possible correlates of the display of emotion (positive and negative emotions) and deception (truthful versus deceiving). In particular, I will describe some production experiments to elicit emotional and deceptive speech samples from speakers in a natural way. The former makes use of the so-called Velten method, and aims to investigate to what extent audiovisual correlates of "real" emotions differ from "acted" ones. The latter attempts to obtain truthful versus deceptive speech utterances from kids through an interactive game in which participants are invited to engage in a conversation with 2 different fairy-tale characters. |
|---|
| responsibles | Loevenbruck, Welby |
|---|
| |
|