Qualitative methods allow us to work on stories, while quantitative approaches often cannot. But can we work on a story with quali-quantitative instruments? Can we find a story in a corpus just by the use of diagrams, without reading any of its individual documents? We will analyze one series of events that occurred in France in 2012: the accident that took place at the Fessenheim reactor and the subsequent political decision to close the plant. The data have been gathered from the OT-Media database. We will make use of textual statistics and narrative models to understand the constitution and evolution of this series of events. Our aim is to show how the use of textual statistics permits to realize narrative analyses on multimedia corpora.