What Are You Really Asking For? A Comparative 5W1H Analysis of Learner Questioning in CPR Training with IVAs in Screen-based and Augmented Reality Environments

Paper Title

What Are You Really Asking For? A Comparative 5W1H Analysis of Learner Questioning in CPR Training with IVAs in Screen-based and Augmented Reality Environments

Publication Info

• Topic area: Learner questioning behavior with Intelligent Virtual Agents (IVAs) in different presentation modalities.
• Keywords: Intelligent Virtual Agents, Augmented Reality, Screen-based learning, Questioning behavior, 5W1H framework, Spatial presence, Social presence, CPR training, Learning motivation, Learning strategies.

Background and Problem

• Problem / challenge: Limited understanding of how presentation modalities (AR vs. screen-based) and learner characteristics shape questioning behavior with embodied IVAs. Prior research has focused on text-based interactions or system-level performance, neglecting naturalistic, spoken inquiries in immersive environments.
• Significance: Understanding these dynamics is crucial for designing effective IVA-supported learning systems, particularly in skill-based training contexts like CPR, where procedural and conceptual knowledge are both critical.
• Motivation and related work: Previous studies have explored AI-mediated learning and user experiences but lack systematic frameworks to analyze questioning behavior in immersive contexts. This study builds on the Cognitive Affective Model of Immersive Learning (CAMIL) and integrates the 5W1H framework to address this gap.

Solution

• Proposed approach: A 5W1H-based framework to systematically analyze learner questioning behavior across six dimensions: Who, When, Where, What, Why, and How. The framework is applied to compare AR-IVA and Video-IVA in CPR training.
• Novelty:
  1. Development of a 5W1H framework for analyzing questioning behavior in immersive educational contexts.
  2. Empirical comparison of questioning behavior between AR-IVA and Video-IVA modalities.
  3. Examination of the moderating effects of learner motivation and strategies on questioning behavior.
  4. Identification of design principles for adaptive IVA systems.
• Procedure and key techniques:
  1. Developed a real-time Q&A IVA system delivering CPR instruction in AR and screen-based environments.
  2. Conducted a user study with 40 participants, randomly assigned to AR-IVA or Video-IVA conditions.
  3. Collected and analyzed data on questioning behavior, presence, and learning outcomes using the 5W1H framework.
  4. Applied statistical and thematic analyses to evaluate hypotheses on modality effects, mediation by presence, and moderation by learner traits.

Results

• Concrete findings:
  • AR-IVA elicited higher spatial presence (M = 3.66 vs. 3.01, p < 0.001) and social presence (M = 4.96 vs. 4.46, p = 0.036) than Video-IVA.
  • AR-IVA learners asked more frequent (M = 6.60 vs. 4.20, p = 0.001) and longer questions (p < 0.001) but showed no significant difference in question complexity.
  • AR-IVA promoted conceptual deepening, while Video-IVA encouraged procedural refinement.
  • Presence acted as a selective filter, suppressing questions during action stages and increasing reflective inquiry during ending stages.
  • Practical performance scores were higher in AR-IVA (M = 11.65 vs. 9.70, p = 0.009), while knowledge quiz scores showed no significant difference.
• Advantage over baselines:
  • AR-IVA enhanced procedural learning outcomes and fostered deeper conceptual inquiry compared to Video-IVA.
  • Presence effects were more pronounced in AR, directing attention to core tasks and suppressing peripheral distractions.
• Experiments / evaluation:
  • Participants (N = 40) engaged in CPR training with AR-IVA or Video-IVA.
  • Data collected included transcribed questions, presence questionnaires, and learning assessments (knowledge quiz and practical performance).
  • Statistical analyses tested hypotheses on modality effects, mediation by presence, and moderation by learner traits.
• Limitations and future work:
  • Conducted in a controlled lab setting with a single learning task (CPR), limiting generalizability.
  • Short-duration study may not capture long-term behavioral patterns.
  • Excluded the "Where" dimension of questioning due to uniform attentional targets across conditions.
  • Future work should explore diverse contexts, longitudinal effects, and adaptive IVA designs tailored to individual learner profiles.

Summary

This study introduces a 5W1H-based framework to analyze learner questioning behavior with IVAs in AR and screen-based environments. Results show that AR-IVA enhances spatial and social presence, promotes conceptual deepening, and improves procedural learning outcomes compared to Video-IVA. Presence acts as a selective filter, shaping questioning timing and focus. Learner motivation and strategies further moderate these effects, highlighting the need for adaptive, learner-centered IVA systems. These findings inform the design of IVAs that optimize interaction strategies based on modality and individual learner traits.