A General Architecture for a Real-Time Discourse Agent
and a Case Study in Oral Reading Tutoring

Physical Events: Input

The user has two input devices, the mouse and a microphone. The keyboard is used only in restricted situations, such as entering a new subject ID, and never by children.

The system can recognize the following physical events:

Logical Events

System-Level Logical Events

User Logical Events

Listener Logical Events

Player Logical Events

Tutor Logical Events

Turn-taking events are instantiated with respect to the current context (student performance and discourse module) to yield specific interventions such as reading a word.

Physical Events: Output

The system can display text, play sound files (including sentence and word recordings), and display new sentences.