UC Santa Barbara

In language, conventions play an important role in shaping communication norms, and digital platforms have emerged as new landscapes for exploring how these conventions are constructed and negotiated. Words follow shared conventions, where both their form and meaning are agreed upon by speakers of a language. In the digital space, emojis function similarly to words by following shared conventions, although their meanings can be more fluid, emerging through complex communicative interactions. This dissertation examines the communicative conventions surrounding emoji use and interpretation in digitally mediated and human-AI interactions, through a functional linguistic lens. This dissertation contains three core studies. The first two core chapters analyze emoji use and interpretation, while the third investigates how users respond to convention violations by AI chatbots. The first study presents a computational linguistic analysis of Thumbs Up Emoji using unsupervised clustering algorithms and a novel embedding method to cluster computational representations of emoji meanings, showing the range of pragmatic functions and demonstrating the range of context in which Twitter users employ Thumbs Up Emoji in online interactions. The second study explores variations in emoji interpretation across demographic groups, using survey methodology to show how message content, sender familiarity, and participant identity significantly impact emoji understanding. The third study extends this investigation by focusing on human-AI interactions from a psycholinguistic perspective, showing how violations of general and generation-specific communication patterns impact information processing and user experience.