| Social cues shape our interactions and the way others see us. We convey a persona through our actions and words, but we are not always aware of how we are seen and interpreted. We use cues to coordinate conversation, determine speaker, and sense the mood of a group. However, online social spaces have expanded the ways in which we interact with others. Due to the nature of online communication, interactions persist in a manner that allows us to review our action. Otherwise devoid of traditional social cues, online places offer different mechanisms for mediation and feedback that are often not available in face-to-face conversation.;This dissertation applies affordances of computer mediated communication such as persistency, anonymity, large group backchannels, and archival into realized conceptions of a Social Mirror---a visualization to augment real-time conversation with additional feedback. The Social Mirrors capture interaction patterns and reflect visually aesthetic representations in real-time. The designs demonstrate additional cues alongside the traditional social cues one might expect by integrating this feedback directly into the conversation space. I show how this feedback influences the perception of self to change interaction, encourage self-evaluation and enable participants in conversation.;In the first chapters of this work, I establish that feedback from the social mirror affects people differently based on what the social expectations dictate. These mirrors, and those in related work, consistently demonstrate a balancing of conversation. This is not because the visualizations direct an individual to change; rather, the participants shape themselves into a perceived balance with each other through the mirror. In a mirrored conversation, the talkative back off and the quiet speak up while reporting their contribution was respectively dominating the social mirror and lacking from the social mirror.;More important than the individual projects, the concept of the social mirror provides a tool for group interaction. With social mirrors, groups gain access to new cues in real-time, but can continue to mediate interaction through self-directed means. The underlying theoretical construct I contribute (presented in Chapter i) defines the characteristics of social mirrors leveraged in this work: capturing interaction data, visualizing this data, and encouraging reflection on this data. The four example social mirrors---Conversation Clock, Conversation Votes, Fragmented Social Mirror, and Conversation Clusters---illustrate the social mirror as applied to conversation, however social mirrors can have broader applications that I hope contribute with future developments in other domains.;Specifically, this dissertation presents five contributions. First, I put forth the Social Mirror as a theoretical construct. I describe the necessary characteristics based in social visualization with influence from signaling theory, group dynamics, and accommodation theory. Social mirrors capture interaction, visually construct a cue, and feed that information right back into the current interaction and are not limited to my own designs. It is a feedback loop of self-evaluation that creates otherwise inaccessible social cues. Second, I present my visualizations for their design contributions. Each mirror combines abstracted audio visualization and augmented spaces with informative and aesthetic imagery. Each visualization leverages design to inform the viewer in an environment that is already visually demanding attention. In order to convey meaning, I rely on a well designed structure with minimal distractions. In this dissertation, I present the unique design challenges and visual solutions to ensure that the abstracted audio is a beneficial augmentation to conversation. Third, I demonstrate that the presence of a visualization encourages a "balanced" conversation. This is not because of the demands of the visualizations, but because the participants shape themselves into a perceived balance with each other through the mirror. I demonstrate this link initially with the Conversation Clock in terms of contribution: the talkative back off and the quiet speak up. This effect persists with the addition of approval/disapproval feedback in Conversation Votes. Most importantly, a greatly distorted visualization has relatively little effect on this balance. The presence of these social mirrors and reminder of a person's activity is more than enough to encourage "balance." Fourth, I demonstrate that explicit but anonymous feedback can empower individuals in conversation. Rather than refraining from participating, participants use the anonymous backchannel to express their opinion without the need to speak up. This channel leads to more speaking and more student initiated questions in a classroom setting. Finally, I present a social mirror that demonstrates how the context of conversation can be detected and incorporated into the social mirror. Accounting for poor speech recognition, the Conversation Clusters mirror allows a social mirror timeline to be automatically annotated by topic for longer archival and review. As a whole, this dissertation presents the social mirror as an interface to improve upon self reflection in conversation. Each project presented illustrates the use and benefit of such mirrors and opens the doors to additional mirrors by others. |
