We propose the Cumulative Abnormal Tone (CAT) framework for tracking the normal and abnormal dynamics of textual tone around events. In our framework, we use text-mining techniques to derive the normal tone based on market and sector-wide news. We then focus on corporate events and track the abnormal tone dynamics around that event. This leads to a cumulative abnormal tone chart. We apply the CAT framework to the analysis of cumulative abnormal returns across earnings press releases. We find that the analysis of firm’s abnormal tone provides investors with relevant predictive information on the firm’s stock performance.