Type of Document	Dissertation
Author	Perez Galluccio, Roberto Gustavo
URN	etd-06112008-132429
Title	Animated Pedagogical Agents as Spanish Language Instructors: Effect of Accent, Appearance, and Type of Activity on Student Performance, Motivation, and Perception of Agent
Degree	Doctor of Philosophy
Department	Educational Psychology and Learning Systems, Department of
Advisory Committee	Advisor Name Title John M. Keller Committee Chair J. Michael Spector Committee Member Michael Leeser Committee Member Walter Wager Committee Member
Keywords	Pedagogical Agents Language Learning Animated Agents Animated Pedagogical Agent CAI Spanish Computer-Based Computer-Assisted Call
Date of Defense	2008-04-08
Availability	unrestricted
Abstract This past decade has seen an increase in the use of an interesting approach to facilitate the humanization of human-computer interaction (HCI): the use of animated characters, or agents, as interfaces that mediate between the human being and the machine. The design of animated agents has been, in general, based on human characteristics, such as gender, age, ethnic representation, and personality traits, as well as the use of gestures, facial expressions, and emotions. The challenge so far has been to identify what characteristics work better in what types of environment. In order to determine what kind of agents and what type of instructional contexts (ICs) would be more suitable for computer-assisted language learning (CALL) purposes, an experiment was designed to investigate the effect of agent accent (Anglo, Hispanic), appearance (Anglo, Hispanic), and type of instructional activity (Grammar-based, Communication-oriented) on student performance, motivation, and perception of agent. The study was conducted on a population of 212 college students enrolled in a basic Spanish class at a large southeastern university in the United States. Results indicated that there were no differences in performance, motivation or perception of agent between students exposed to the matching attributes (MA) agents (i.e., Anglo accent + Anglo appearance, and Hispanic Accent + Hispanic appearance) and students exposed to the non-matching attributes (NA) agents (i.e., Anglo accent + Hispanic appearance, and Hispanic accent + Anglo appearance). In addition, there were no differences in performance or motivation between students who worked with the Anglo (AA) agent and students who worked with the Hispanic (HH) agent in the grammar-based activity or the communication-oriented activity. Results did show a significant difference in motivation ratings between the AA and HH agents when the two types of instructional activity were merged into one dataset (HH > AA). Students felt more satisfied when they received positive feedback from the HH agent, and they were more willing to repeat the experience. Data analysis also indicated significant differences in student perception of agent. Participants who worked with the AA agent in the grammar-based activity rated that agent higher in terms of communicative ability (in spite of both agents using exactly the same script to deliver instruction), while students who worked with the HH agent in the communication-oriented activity thought that agent was more knowledgeable about the Hispanic culture, was better prepared to function in both Hispanic and Anglo cultures, had a more friendly attitude, and was more willing to answer student questions. These results seem to indicate that students tended to use their previous beliefs about language instructors to assign human-like qualities to the animated agents. As a consequence, some agent characteristics had either positive or negative effects on the students, in line with previous research on classroom interactions. Similarly, some motivational interventions had a positive effect on the students, as expected based on previous studies. Although more data are needed to strengthen the significance of these findings, these results have important practical and theoretical implications for the use of animated agents in language learning environments.
Files	Filename       Size       Approximate Download Time (Hours:Minutes:Seconds)     28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access    PerezRDissertation.pdf 1.87 Mb 00:08:38 00:04:26 00:03:53 00:01:56 00:00:09