My research focuses on pragmatic language understanding in Human Machine Interaction, with an emphasis on how multimodal cues (e.g., textual, audio, and visual) can be utilized to unreveal meaning beyond literal content. Sarcasm serves as a central case study in my research. Drawing on linguistics, cognitive science, my work investigates how pragmatic cues like prosodic variation, semantic incongruity, and facial expressions can be systematically modeled using multimodal fusion strategies. The broader aim is to move beyond surface-level prompt-style language processing, and toward systems that understand language as it is used in real human interaction: emotionally charged, culturally embedded, and shaped by dynamic social context.

Key words: multimodal sarcasm detection, pragmatic computing, human-machine-interaction, affective computing.