Objective: The current study longitudinally examines college student Twitter patterns throughout initial phases of the COVID-19 pandemic. This study aims to better understand psychological impact and online personal communication during the pandemic. Participants: A dataset consisting of ~720,000 tweets posted by students from universities throughout the United States during the 2020 spring semester was analyzed according to structural and sentimental analysis. Methods: Using a data-driven approach, three time periods emerged which reflected the transition to online learning. Results: Significant changes in structure and sentiment of tweets were observed across phases. Conclusions: Changes in Twitter patterns revealed important features of this unprecedented transition to online learning for college students.