Positive Preferences: The Emotional Valence of What an Avatar Says Matters.

Abstract

This study examined preferences and ascriptions of human control for avatars associated with positive and negative language. An avatar is a graphical representation of an individual user in a virtual world. Users form judgments of other virtual world users based on the appearance and behavior of their avatars in the absence of physical cues. In particular, users have previously displayed sensitivity toward anthropomorphism and verbal behavior of avatars they encounter within virtual environments. Thus far, investigations of language in online and virtual spaces have been limited to specific contexts. University student participants and Amazon Mechanical Turk workers were shown two avatars. Each pair consisted of two out of three possible appearances: a tiger, a male, or a female. Each avatar was aligned with a different text introduction: one containing words of positive emotional valence and the other of negative emotional valence. Participants from both samples preferred avatars associated with positive language, regardless of appearance, but participants did not ascribe human control of an avatar based on either emotional valance or appearance. Significant differences for reported preferences and likeability demonstrate the salience of emotive language as a social cue even in short text introductions for avatars. These findings suggest that those we like and want to be around in virtual environments might not necessarily align with whom we view as human.