ArticuLab

HCII ~ School of Computer Science ~ Carnegie Mellon University

Other Projects: JR Summit | NUMACK | Collaborative Storytelling | Virtual Peers and Autism | Scaffolding Science Achievement in the Culturally-Diverse Classroom | Connection Machines

Scaffolding Science Achievement in the Culturally-Diverse Classroom

Alex playing with childThe "achievement gap" between children of color and their European American peers remains a critical unsolved issue in our educational system, especially for children who do not share upper-middle class Mainstream American English literacy practices, nor come to school speaking the Mainstream American English dialect.

In our work, by examining the linguistic practices of particular communities - both inside and outside school - and how language is used in the cultural activities of a community, we come to understand the role that dialect and language may play in culture and identity, as well as the role they may play in the elementary school classroom. On the basis of this examination, we build virtual peers who engage with children in academically successful and personally satisfying peer collaborations in school and informal learning contexts. Our research therefore allows us to combine the sociocultural norms of language practices with innovative technology that provides effective learning environments for diverse student populations. Click here for publications.

Our current research goal is to design culturally authentic agents that bridge the gap between language skills practiced outside the classroom setting and those language skills required in the classroom. As designers of effective learning environments, we propose that virtual peers and other educational technologies should be designed to engage in authentic cultural practices. We argue that culture and ethnicity include more than physical appearance and encompass verbal and nonverbal behaviors as well. To reach this goal, we observe verbal and nonverbal discourse behaviors of African American children as they participate in the social activities of play and the learning activites of the science classroom. We extend our existing realization and cognitive architectures of virtual humans to support an embodied conversational agent - a virtual peer named Alex - that models verbal and nonverbal behaviors of African American children who speak African American Vernacular English (AAVE). 

Alex is a virtual peer who engages children in the science activities of the third grade classroom, interspersed with the getting-to-know-you activities that all children engage in both inside and outside school. While the physical appearance of Alex is designed to be race and gender ambiguous, the verbal and nonverbal discourse behaviors are based upon human models of African American children between the ages of 8 - 10 years old - African American children who speak AAVE only, who speak MAE only, or who "code-switch" back and forth between the two dialects depending on social context (for example, speaking AAVE when playing and MAE when working on a science task in the classroom).

Our research has demonstrated that children emulate the speech patterns of the virtual peer, as well as the virtual peer's scientific ability, demonstrating that technological partners such as these can serve powerful roles as learning partners. However, children who worked with a virtual peer that spoke AAVE only used speech that was more fluent and less hesitant (see Finkelstein et al., 2012) and they also produced more strongly reasoned scientific arguments - talk that was more like the virtual peer's model - than children who worked with a virtual peer that spoke MAE (under review). These results call for a re-examination of the cultural assumptions followed in the design of educational technologies, with a specific emphasis on the way in which we index culture and identity, and the ways in which we ask culturally-underrepresented groups to participate in learning activities.

(Thanks to the National Science Foundation and the Heinz Foundation for generous funding of this research