This thesis describes novel computer science research on visual analytics methods for the detection and understanding of diverse phenomena of change that can be observed either within natural language text or based on it. The term change refers to the observable variation of features and patterns over time. In particular, two different kinds of phenomena are under research. The first part of the thesis deals with the diachronic change of linguistic features, namely language change. It includes pioneering work in the intersection of the disciplines of historical linguistics typological comparison of languages and visual analytics and contributes to the broader field of digital humanities or enhanced humanities (eHumanities). The second part of the thesis deals with visual analytics methods for the interactive detection and exploration of sudden unexpected changes in the information content conveyed by a large-scale text data stream. The research fills gaps in the previous work on time-related visual text analytics, demonstrates the commercial potential of such methods, and systematically outlines future research challenges for the live analysis and visualization of large-scale text data streams.