Decision making is an essential part of our life and we naturally strive to support decisions by arguments; today, the Internet is an obvious place to search for them. Unfortunately, general-purpose search engines often fail to represent search results in a suitable way (e.g., contrasting supporting and opposing arguments). This work analyzes a corpus of ca. 80 German web documents about current educational topics for argumentation patterns. Significance testing experiments explore the role of discource markers and discourse relations in argumentation and classification experiments point towards Argument Extraction.