This study explores disaster recovery communication in the digital era. In particular, this study analyzes Twitter communication data corresponding to the 2016 Southern Louisiana flood recovery process and examines patterns and characteristics of long-term recovery communication. Based on network and sentiment analyses of the longitudinal Twitter data, the study identifies the dynamic changes in participants' numbers, dominant voices, and sentiments in social media communication during the long-term recovery process. From the additional content analysis of relevant news articles, in-depth contextual information is provided to support and supplement the findings. Findings show the weaning communication volume during the recovery phase, lacking local voices over the long-term recovery communication process, and prolonging negative sentiments over the recovery period. Based on the findings, the authors provide implications highlighting the need for investing in long-term recovery communication, better utilizing information from social media, and supporting local voices during disaster recovery.
