Abstract
Large text corpora are a main language resource for the human-driven analysis of linguistic phenomena. With the ever increasing amount of data, it is vital to find ways to help people understand the data, and visualization techniques provide one way to do that. Corpus Clouds is a program which pro-vides visualizations of different types of frequency information dynamically de-rived from a corpus via a standard query system, integrated with a standard KWIC display. We apply established principles from information visualization to provide dynamic, interactive representations of the query results. The se-lected design principles and alternatives to the implementation will be discussed and a preview on what other types of information connected to corpora can be visualized in similar ways are provided. Corpus Clouds can thus be seen as an-swer to the call by Collins et al. [1] to design in a principled way new visualiza-tion tools for linguistic data.