Does embodiment in a corpus-based, immersive spoke-language interaction environment promote language acquisition better than embodiment-lower alternatives?

Summary

Outlines the design and outcome of an experiment created to understand adding embodied to NPC agents in an immersive spoken-dialogue virtual environment affect learning, and learning a second language. Controls against less-rich embodied agents.

To Do

  1. Analyse the observed interaction dialogue:
    1. Understand and quantify vocabularly used in the interactions
    2. Depict dialogue trees and calculate dialogue route probabilities
    3. Contrast observed dialogue with the previously meta-analysed online teaching materials, and search for distinctions
  2. Analyse the embodiment of the observed interactions:
    1. Gesture
    2. Spatial awareness
  3. Develop/deploy a method to codify these aspects in relation to the dialogue trees
  4. Develop a VR conversational agent informed by the above

Pre-experiment

To determine suitability of commerical ASR for experiement implementation: Speech recognition accuracy: comparative study between native and non-native Japanese speakers and ASR systems