What's being said near "Martha"? Exploring name entities in literary text collections

A common task in literary analysis is to study characters in a novel or collection. When dealing with large documents or collections automatic entity extraction, text analysis and effective user interfaces might facilitate the exploration of the topics discussed or the vocabulary used in the neighborhood of the characters. Using our interface, called POSvis, the scholar uses word clouds and self-organizing graphs to review the vocabulary in the vicinity of one or more entities, to filter by part of speech, and to explore the network of other characters in that vicinity. Visualizations show word usages within an analysis window (i.e. a book chapter), which can be compared with a reference window (i.e. the whole book). We describe the interface and report on an early case study with a humanities scholar.