In this work we attempt to understand the visual accents in Arabic facial expressions and create culture-specific facial expressions for a female multi-lingual, cross-cultural robot. This study will enable the investigation of the effect of creating such expressions on the quality of the human-robot interaction. Our work is twofold: we first identify the existence of accent variation in facial expressions across cultures, then we validate human recognition of these accents. Facial expressions embody culture and are crucial for effective communication; hence they play an important role in multi-lingual, cross-cultural human-robot interaction. Elfenbein and Ambady found that there are different accents in facial expressions, which are culture-specific, and that the differences in expressions between cultures can create misunderstandings [Elfenbein and Ambady, 2003]. Several studies compared American expressions with expressions from other cultures but none of them included Arabic facial expressions. There is no existing database for Arabic facial expressions. Consequently, we recorded videos of young Arab women narrating stories that express six emotions: happiness, sadness, surprise, fear, disgust, and disappointment. These videos were analyzed to extract Arabic accents in facial expressions. The expressions were then implemented on a 3D face model using the Facial Action Coding System (FACS). To evaluate the expressions we conducted a web-based, human-subject experiment directed at students and staff at Carnegie Mellon University in Qatar. Thirty-four participants were asked to choose the appropriate emotion for each expression and rate, on a ten-level Likert scale, the accuracy with which the expression represents the emotion. The cultural affiliation of the participants was recorded. Preliminary results show that Arabs are more likely to recognize the Arabic facial expressions over non-Arabs. To further support this conclusion the survey will be redistributed to a larger number of subjects from different cultural backgrounds and from different geographical areas.


Article metrics loading...

Loading full text...

Full text loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error