As more of our conversations and communities take shape online, understanding how people connect and communicate in these digital spaces has become increasingly important. Social networks influence not just what we say, but who hears it and how that information travels. My work attempts to investigate these dynamics—particularly how new communities form, how ideas move across them, and how people use language to express and shape those connections. Through network and linguistic analysis, it explores the interplay between structure and speech in complex, ever-evolving digital environments. The dissertation begins by investigating audiences: how communities emerge, fragment, and reassemble in response to social and global events. The early pandemic is used as a case study to examine “social tectonics”—the rapid, often chaotic formation of new communities. Furthering these ideas, patterns of user and information flow across community groups are examined to challenge common assumptions about radicalization and highlight the influence of alternative media ecosystems. Finally, new k-core-based null model is introduced to offer fresh tools for understanding network structure. On the linguistic side, the project analyzes the subtle biases encoded in binomial word pairs and how those choices reflect cultural norms, including the ways such patterns are absorbed by the technologies we use. It also examines whether large language models can mimic the quirks of human randomness—and when we might want them to.