Going Forward

As I transition from Associate Professor to Full Professor, I look forward to new challenges locally in my involvement in leadership on campus, more broadly in my professional service, and especially in my research.

Locally, I have a history of service on campus at the department, school, and university level. In the past couple of years I have gotten involved in two areas of service I plan to expand upon in this new stage of my career. First is involvement in CMU’s Simon Initiative. Under this broad umbrella I have been actively involved in the Technology Enhanced Learning (TEL) writing initiative headed by Richard Scheines, Dean of CMU’s Dietrich School. My contribution was first as coordinator of its technology thrust, and now as co-leader in some of its early seed projects. Specifically, I am partnering on two different projects in which I am seeking to apply my own research in Computer-Supported Collaborative Learning (in collaboration with David Kaufer in the Rhetoric department) and Computational Discourse Analysis (in collaboration with Chris Neuwirth in the English department) to improve instruction on communication on campus at CMU. These projects have been in the planning phase during Spring of 2016 but will begin in earnest during Summer of 2016 and beyond. The vision of the TEL writing initiative is to start with these seed projects and expand to broader impact on campus across all of CMU’s schools. A second direction is involvement in the Language Technology Institute’s Governance Committee. The goal of this committee is to reflect upon the management practices of the department at all levels and to seek to improve its smooth functioning and morale. I believe my service to the department would be more effective if it were more focused and less diffuse. Thus, as we continue to work together as a committee to identify opportunities for making positive contributions, my goal is to seek a consistent area to focus my service contribution going forward. Some recent discussions in that context have focused on initiatives that will increase awareness and communication between research areas within the department, and contributing towards this effort fits my orientation towards bridge building and interdisciplinarity.

In my external professional service, at the same time as my Presidential term within the International Society of the Learning Sciences is winding down, I am transitioning from Associate Editor of the International Journal of Computer-Supported Collaborative Learning (ijCSCL) to Executive Editor. In this new leadership role, I will partner with the new Editor-in-Chief and the other two Executive Editors to revitalize the focus and scope of the journal. To start off this effort, Ulrike Cress (one of the current Executive Editors) and I are co-organizing a workshop at the International Conference of the Learning Sciences (ICLS) to be held in Summer 2016 entitled “Towards next steps for the CSCL Community: Advancing science and informing real world collaboration in Web 2.0”. Through a series of invited talks, feedback panels, and poster sessions in this full day workshop, we will work together to forge a fresh vision that will be offered to the community in a subsequent planned special issue. The goal is to strengthen and deepen the bridge between the CSCL community and the broader communities of research in Social Media Analysis and Computer Supported Cooperative Work. Ulrike Cress and I are also co-editing the in preparation Handbook of Computer-Supported Collaborative Learning, which will be published by Springer.

At the time of my tenure review, my research team was just transitioning into a push towards large scale dissemination and deployment, taking advantage of the rise of Massive Open Online Courses (MOOCs) as an opportunity to deploy our developed technology for dynamic support for collaborative learning at a grand scale. As highlighted above, this past few years have produced a whole series of deployments and positive demonstrations of impact and potential for even wider impact going forward in my continued collaborations with edX, the Smithsonian Institute, the Community Colleges of California system, etc. While continuing this work, new challenges have become apparent, many related to assumptions underlying current MOOC instructional design. First, the framing of MOOCs as courses that operate autonomously, with collaborative opportunities embedded as optional supplementary activities, limits the opportunity for collaborative discussion to provide the social support known to bolster student commitment, positive decision making, and learning. Into the future, my aim is to broaden and deepen my collaboration with leaders in the area of online learning communities, such as George Siemens and Dragan Gaesevic, as we develop a vision for bridging across formal learning settings such as programming MOOCs, informal settings for information sharing, such as Stack Overflow, and online production communities such as GitHub. This research is already getting off the ground in the context of my recently funded Big Data grant with Jim Herbsleb as my local Co-PI, and George Siemens as lead on a collaborative proposal at UT Arlington.

In addition, I have recently led a local team including Ken Koedinger, Geoff Gordon, Emma Brunskill, John Stamper, and Chinmay Kulkarni, in collaboration with Stanford University, to submit a proposal to the NSF Expeditions program to carry this vision further. In this broader framing of online learning settings, new problems arise with respect to supporting students on their own personal learning trajectories. The fields of learning analytics and educational data mining have produced models that are able to operate well within bounded decision spaces, usually operating at a single grain size: everything from regulating help strategies within a single problem solving step or pacing through a sequence of activities within a unit of a course to choices of courses within a degree program, or even choices between degree programs.  Multiple models at different grain sizes are needed in order to manage complexity, but this fragmentation limits the effectiveness of what can be accomplished.  For example, local models misattribute some important cases where students cease to participate in a learning opportunity because they have chosen to move to a different path that is better suited to their needs. Local models treat the discontinued participation on the current path as a lack of success rather than correctly rewarding the positive choice to move to a more appropriate path. Thus, students may fall through the cracks when trajectories cross the boundaries of individual models -- a student is in the wrong degree program, or where participation in an activity may be motivating, but may be distracting from more important goals. In order to adapt the state of the art in reinforcement learning to these broader decision spaces where random sampling is not feasible, there is a need for instrumental variables that relate to student states and dispositions. My contribution to this effort from computational modeling of these variables from observations of their interactions in a variety of forms of social engagement connected directly or indirectly with online learning communities.

With a dual research focus on Computational Sociolinguistics and Computer-Supported Collaborative Learning, my research has frequently afforded the opportunity to observe the extent to which social interaction through discussion makes salient very personal things like cultural identity, socio-economic status, or other power-relevant social identities that sometimes strategically position individuals for success, and other times hold people back. The idea of viewing linguistic choices in social interaction as currency within an economy that is on the one hand social but on the other hand has real implications for achievement in school and advancement at the career level is a well established idea. My joint research with Lauren Resnick and Sherice Clark on agency in urban school classrooms is one context in which these issues have raised questions in my past work. Though well established this idea encompasses many enduring open questions especially as they pertain to computational modeling. In my current work heading into the future, I am pursuing these ideas as Computer-Supported Collaborative Learning provides a setting in which to investigate questions in history courses related to historical causal reasoning connected with controversial events. For example, in joint work with Baruch Schwarz at the Hebrew University in Jerusalem and Kobi Gal at Ben Gurion University of the Negev, we are exploring how established paradigms for improving consensus building must be adapted and extended in discussions between Israeli and Palestinian students negotiating their different understandings of responsibility for past wars. In an analogous joint project with American History education in the Community Colleges of California system we are working to develop activities to be deployed across campuses in the system on topics such as the origins of the Constitution, especially as topics such as those touch on issues of racism and inequity, that continue to play a role in American politics today in ways that are very personal to the student population of this and similar community college systems.

In summary, my research has benefited from intense involvement in the School of Computer Science both in the language technologies community and in the learning sciences community as it fits within the human-computer interaction community. Because what drives my research is the goal of developing technology capable of both shaping conversation and supporting conversation to achieve a positive impact on human learning, I look forward to remaining active in both of these communities.

Carolyn Penstein Rose (cprose@cs.cmu.edu)/ Carnegie Mellon University