<![CDATA[Artificial intelligence (AI)-powered chatbots have substantially reshaped the landscape of human-AI interaction, yet questions remain about the communicative behaviors users deploy in these exchanges and how they are shaped by AI system characteristics. This study investigates communication pattern typologies in human-chatbot interactions through qualitative thematic analysis of 150 conversation transcripts across customer service, mental health, and educational contexts. Grounded in social presence theory and affordance theory, and employing reflexive thematic analysis (Braun & Clarke, 2022) with an interpretivist-constructivist epistemological stance, this study identified five typologies of communication patterns: adaptive mirroring (prevalent across the majority of transcripts, 78.7%), emotional scaffolding (frequently manifested, 65.3%), contextual anchoring (emergent in a substantial portion of conversations, 58.0%), conversational repair mechanisms (present in a considerable minority, 42.7%), and trust-building narratives (identified in over one-third of transcripts, 37.3%). These figures reflect descriptive frequency counts indicating the proportion of transcripts in which each pattern was observed; they serve as organizational summaries, not statistical evidence, and patterns are not mutually exclusive. The distribution of these patterns varied across demographic groups and interaction contexts, with younger participants showing comparatively greater communicative flexibility. Results suggest that users deploy socially patterned communicative behaviors in AI-mediated contexts that extend beyond purely task-oriented exchanges. These findings carry implications for designing AI systems that are attentive to the relational and contextual dimensions of user communication in healthcare, education, and customer service.]]>
