Social-ecological issues, crises, and disasters all demand organizational responses. The study of how organizations collaborate and communicate with one another in large-scale responses requires high-quality data that is difficult and costly to produce through traditional methods of data collection such as interviews, surveys, or manual coding of corpora. However, using natural language processing methods such as large language models (LLMs) allows for at-scale coding of social data from textual sources, such as messages and reports, produced "naturally" during an organizations' response. This project aims to test and develop pipelines for processing heterogeneous texts for extracting essential data necessary for analyzing organizational activities in the context of crises and social-ecological systems. Exploratory and statistical methods such as social network analysis will be used on extracted organizational activity data to study such questions as how causal narratives used by officials during a pandemic is structured as part of a larger discourse and how organizations discuss issues related to vulnerable social-ecological systems.