This document discusses research analyzing language patterns ("disaster scripts") on Twitter in response to 21 different disasters between 2012-2016. The research found: 1) Similar disaster response language was used across different disaster types, particularly references to victims, prayers, and social support. 2) The frequency of disaster scripts on Twitter correlated moderately with casualty numbers for earthquakes but showed low correlation globally for terrorism events. 3) The correlation between scripts and impact was strongest in North America and Western Europe, likely due to demographic and infrastructure differences in other regions.