Electronic communication media are a widespread means of interaction. They effect network relationships among people. Such networks provide connectivity but are often structured in clusters. Current cluster analysis in social corpora is mainly based on structural properties. This paper extends existing approaches with content-based cluster identification and community detection in social corpora. Following a design science methodology, we demonstrate our approach using a corporate e-mail dataset. After analyzing relationships between structural and content-based groups we conclude that our method contributes to detecting online communities, especially for large structural or smaller but dispersed topical groups.