Social media are becoming increasingly heterogeneous in nature and usage as online communication evolves. Understanding how communities use and repurpose media is important to our understanding of technology and society. We present Landscape of Words, a visualization that uses topic modeling to statistically summarize the semantics of arbitrarily large groups of individual contributors. Common topics, linguistic styles, and memes are projected onto an interactive 3-D landscape, allowing users to explore the usage patterns of millions of individuals. Our first prototype examines Twitter, illustrating how trends in topics evolve over time.