« All Events

Virtual Event
Event Series: AZTESOL Professional Development Series

Prosody and Pragmatics in AI Chatbots: Can ChatGPT Produce Ironic Tone?

January 14, 2026 @ 4:30 PM - 5:30 PM
Virtual Event

Event: Prosody and Pragmatics in AI Chatbots: Can ChatGPT Produce Ironic Tone?
Type: Paper Presentation
Presenter: Seiji Takahashi
Date: Tuesday, January 14, 2026
Time: 4:30 – 5:30 pm
Location/Format: Virtual

Description / Abstract:
The same utterance—like “Great”—can express either praise or irony depending on prosody. This study examines whether AI chatbots, such as ChatGPT, can convey pragmatic intentions through prosody like humans. Based on the findings, the presentation discusses the potential and limitations of using ChatGPT in teaching pragmatics.

Summary of Findings: AI chatbots hold promise for L2 speaking practice. However, if they cannot use prosody appropriately to convey pragmatic intentions (e.g., sarcasm), they may not effectively support L2 learners’ prosodic distinctions.

In this study, a native English speaker and ChatGPT-4o each produced 16 pairs of utterances differing in prosody to express pragmatic contrasts:

  • Irony vs. non-irony (e.g., “Great” spoken ironically vs. sincerely)
  • Certainty vs. uncertainty (e.g., “I think so” spoken with certainty vs. uncertainty)
  • Twenty native English-speaking participants judged the intended meaning of these utterances.
  • Results showed no significant difference in interpretation accuracy between ChatGPT and human utterances, though human irony was slightly more accurately interpreted.

Takeaways for Attendees:

  • Explore the potential and limitations of ChatGPT for pragmatic instruction.
  • Engage in live activities comparing human and AI utterances.
  • Analyze findings through visualized data.

About the Presenter:
Seiji Takahashi is a researcher in applied linguistics with a focus on pragmatics, prosody, and AI-assisted language learning. He investigates how both human speakers and AI chatbots convey pragmatic meaning through prosody and how these insights can inform L2 instruction.

Details