Research

Topics

Active CMC: When Communication Channels Alter What They Deliver

Today, computer-mediated communication (CMC) technologies have become a pervasive tool that people use to accomplish their collaboration needs. One argument is that CMC technologies are not passive channels that are neutral to communication itself. There are two observations supporting this argument. First, CMC tools are not all the same. Different media (text, audio, video) pose different constraints (e.g., whether there is visibility) and can shape communication behaviors differently. People carrying different cultural norms and different needs may also exhibit diverse behaviors when using the same tool. Second, because CMC tools are computing artifacts, CMC designs may listen to communication content, and actively involve in communication to shape its processes and outcomes. There is a broad space to apply computational techniques to endow CMC with functional properties beyond only passive mediation for supporting group activities such as teamwork and collaborative learning. In my research, I conduct behavioral studies to develop deep understanding around how social factors and technological properties jointly affect online teamwork. I further look at the active and participatory aspect of CMC, and consider using computing techniques such as machine learning (ML) and information retrieval (IR) to augment CMC channels with automatic feedback mechanisms for supporting online groups to attain their goals.

Publications: Wang, Fussell & Cosley (2011), Wang, Cosley & Fussell (2010), Wang et al. (2007)

Computer-Mediated Intercultural Collaboration

Different CMC tools such as video conferencing and instant messaging (IM) pose different constraints (e.g., no visibility in IM) on interpersonal communication. As intercultural collaboration often occurs remotely over CMC, it is important to understand how cultural and media factors interact to jointly shape styles and patterns of online communication. In my recent experimental work (Wang et al., 2009; Wang & Fussell, 2010), I have shown that communication media do not have universal effects on communication behaviors across cultures. Individuals from a more collectivistic culture (Chinese) tended to be more sensitive to media properties than those from a more individualistic culture (American). Chinese were more talkative when working over text-only chat than when they could see their partners through video, presumably because visibility introduces greater social presence and concerns of interpersonal conflict, and the Chinese cultural norm dictates that people maintain harmony and avoid talking too much in such a situation. Americans did not differ in their talkativeness over different media (Wang et al., 2009).

Interestingly, the cultural composition of a group can also influence how members of the group converse over different media. In three-person bicultural teamwork, when there was visibility and when Chinese were the majority in the group (i.e., a group consisting of two Chinese and one American), Chinese and American team members exhibited similar conversational styles as measured by talkativeness, responsiveness and distributions of conversational acts. However, when there was no visibility or when Americans were of the majority, the two cultures had distinguishable styles (Wang & Fussell, 2010). Informed by recent findings in cultural psychology, culturally specific behavioral patterns are switchable and subject to priming. Therefore, one plausible explanation of the present phenomenon points to the relative richness of cultural primes afforded by the combinations of media and group compositions. For example, when exposing participants to richer collectivistic cues available in groups where Chinese is of the majority and visibility is enabled by video, they may be primed to behave toward collectivism and converge their conversational styles. It’s noteworthy that seemingly subtle technical and compositional differences can have salient effects on intercultural communication.

Publications: Wang, Fussell & Cosley (2011), Wang & Fussell (2010), Wang et al. (2009)

Computer-Supported Creativity

Group brainstorming, or discussing to generate ideas, is a commonly practiced procedure to garner ideas collaboratively in workgroups. I take a socio-cognitive perspective positing that group brainstorming encompasses two distinct yet interrelated processes, the cognitive process of idea generation (i.e., to think of ideas individually) and the communicative process of idea sharing (i.e., to communicate ideas with other group members). Overhearing diverse ideas shared by other people may have the effect of cognitive stimulation and help to retrieve remote concepts from memory that are difficult to access independently. However, group brainstorming may also fail if shared ideas are too similar and fail to serve as cues to retrieve remote concepts. This theoretical perspective has its intellectual root in cognitive science, which considers it necessary to explain idea generation in terms of established cognitive mechanisms to avoid mysterious accounts of human creativity. I also contributed to this cognitive science perspective by using machine learning methods to model the cognitive process of idea generation from behavioral data (Wang, 2008).

Informed by the socio-cognitive model of group brainstorming, there is a potential to support brainstorming with computer-generated dynamic feedback. To develop the actual designs, I take multiple aspects of teamwork into account, including (1) the need to maintain people’s practice and preference of naturalistic communication, and (2) the need to support intercultural teamwork by benefiting from cultural diversity in knowledge and perspective while crossing the communicative and language boundaries between cultures. To support brainstorming, I proposed to use language-retrieved pictures, an interaction technique that uses relevant pictures to visualize concepts brought up in verbal conversations (Wang, Cosley & Fussell, 2010). In this design, the system retrieves relevant pictures by mapping keywords found in text inputs and keywords used to describe pictures. Retrieved pictures are displayed at the side of a text chatroom and updated in real-time to provide a visual context unobtrusively. The rationale is that pictures may present multiple concepts in a single scene, which can motivate new topics by leveraging people’s psychological processes of selective attention and multiple interpretations to diversify the concept space available to groups.

Publications: Wang, Fussell & Cosley (2011), Wang, Cosley & Fussell (2010), Wang (2008) , Wang et al. (2007), Wang et al. (2006)

People (in alphabetic order)

I am fortunate to be able to collaborate with and/or learn from researchers across various fields and institutions. They are great mentors and collaborators for my work. Some of them are listed below.

Claire Cardie (Cornell University)

Chun-Yen Chang (National Taiwan Normal University)

Hao-Hua Chu (National Taiwan University)

Tyng-Ruey Chuang (Academia Sinica)

Dan Cosley (Cornell University)

Susan Fussell (Cornell University)

Jeff Hancock (Cornell University)

Sara Kiesler (Carnegie Mellon University)

Tsai-Yen Li (National Chengchi University)

Chao-Lin Liu (National Chengchi University)

Carolyn Rosé (Carnegie Mellon University)

Research Communities