Static and Dynamic Toxicity: New Approaches for Analyzing the Spread of Hate Speech in Online Conversations

This paper proposes a new forensic research framework for analyzing Hate Speech in German online conversations by introducing the two concepts of static and dynamic toxicity. While existing computer-based approaches focus on detecting individual toxic comments, they do not sufficiently consider the conversational context and the structural effects of toxic communication on discourse. Using two large-scale German-language datasets with more than 1.6 million messages, this study examines how toxicity develops within conversations and how hateful content influences their course and termination. The analysis shows that toxic posts occur disproportionately often at the end of conversations, where they correlate with the breakdown of interaction, revealing a disruptive functional role that goes beyond verbal insults. In contrast, temporal and structural variables such as thread length or response times show only marginal effects on the spread of toxicity. These findings suggest that Hate Speech should not be understood solely as a static property of isolated utterances, but as a dynamic interaction phenomenon that shapes and constrains discourse over time. By integrating linguistic, structural, and temporal dimensions, this research contributes to the development of forensic methods capable of contextualizing toxic behavior and strengthening the evidential value of digital communication in legal and investigative contexts.