Pedagogical Agents: Pedagogical Interventions via Integration of Task-Oriented and Socially-Oriented Conversation
- a math game, focus on training basic concepts with spatial representation
- student can play individually, learn how it works
- train a teachable agent
- job is to teach agent how to play game
- learns by watching students
- also poses questions (MC) - asks student to explain how the game works, rules etc
- when agent has watched a few times, it can play autonomously
- student is still helping agent (agree with choice or not)
System architecture
- math game
- teachable agent
- on-task conversation
- visual representation
- social (off-task) conversations (chat with agent, abt music, family)
- off-domain conversation
- on-domain conversation
Possible reasons for adding socially oriented communication
- increase overall engagement and receptivity
- improve recall of learning material through emotional engagement
- social experience helps cement newly learned associations
- promote trust and rapport building
- make students feel more at ease with learning task or topic
Research questions
- can adding social conversation
- enhance student's experience of game
- influence students' self-efficacy and math attitudes
- will social conversation module influence
- game experience
- learning outcomes
Some researchers would question adding off-task conversations - takes away focus from task
Result
- high achievers have more positive experience and better learning outcomes
- this did not apply to low achievers
- low achievers chat more, but high achievers like it more
- possible explanation
- in situation of disengaged chat
- low achievers keep chatting, high achievers do something about it
- can engage learners in voluntary on-domain chat (1/3 initiated chat conversations about the math game)
Next step
- Closer integration between task-oriented vs socially-oriented communications (related to the last result)