Speaker
Description
This research demonstrates how collaborative human-AI analysis builds pragmatic competence in EFL learners. Comparing native speakers, non-native speakers, and AI models on sarcasm detection revealed shared challenges (60% accuracy for natives, 51% for non-natives). These insights informed a pedagogical intervention where students learned through computational pattern analysis and peer collaboration. Results suggest that community-based exploration of AI-identified patterns, combined with metalinguistic discussion, effectively develops real-world digital communication skills essential for online communities.
Short summary
This research demonstrates how collaborative human-AI analysis builds pragmatic competence in EFL learners. Comparing native speakers, non-native speakers, and AI models on sarcasm detection revealed shared challenges (60% accuracy for natives, 51% for non-natives). These insights informed a pedagogical intervention where students learned through computational pattern analysis and peer collaboration. Results suggest that community-based exploration of AI-identified patterns, combined with metalinguistic discussion, effectively develops real-world digital communication skills essential for online communities.
Keywords
Human-AI Collaboration
Pragmatic Competence
Computer Assisted Language Learning
Digital Literacy
Abstract
This CALL SIG presentation explores how collaborative human-AI analysis builds pragmatic competence within learning communities. Our research addresses the conference theme by demonstrating how technology-mediated collaboration between learners, teachers, and AI systems creates inclusive communities of practice for developing real-world digital communication skills.
Initial findings revealed that sarcasm detection in memes challenges even native speakers (60% accuracy), while Japanese EFL learners performed at chance levels (51%). AI models achieved comparable performance (54-57%), suggesting shared processing challenges. These insights informed a three-phase pedagogical framework implemented with 34 participants across 7-8 collaborative sessions.
The intervention emphasized community-based learning: Phase 1 built metalinguistic awareness through group discussion of semantic-pragmatic incongruity; Phase 2 engaged learners as researchers, collaboratively identifying linguistic patterns using computational analysis; Phase 3 fostered critical thinking through peer debates about ambiguous cases where human-AI disagreement was highest. Students worked together to create "sarcasm detection frameworks," embodying the conference theme of building competencies through community engagement.
Preliminary results suggest that collaborative exploration of AI-identified patterns, combined with peer discussion and metacognitive reflection, effectively develops pragmatic competence for authentic online communication. The presentation demonstrates how CALL technologies can facilitate communities of practice where learners, teachers, and AI systems collaborate to develop practical language skills.
Participants will learn strategies for integrating computational tools into collaborative pragmatic instruction, receive our validated multimodal corpus, and understand how human-AI partnerships can strengthen professional development while building learner competencies for meaningful digital communication beyond the classroom.
| Scheduling preference | Anytime on Saturday |
|---|---|
| Title | Building Pragmatic Competence Through Human-AI Collaborative Learning |