Topic Timelines for Enabling Close and Distant Reading of Discursive Shifts. A Pilot Case Using Periodicals of European Diabetes Organizations

In a project about the development of 20^th-century patient organizations and their ideas about disease, we are investigating discursive shifts in the organizations’ periodicals—mostly magazines and newsletters—using topic modelling. Since existing topic modelling techniques carry only limited value for historical analysis, we have developed a timeline extension of the topic modelling tool Topics2Themes that displays change over time and bridges distant and close reading. Our Topic Timeline tool generates a timeline visualization of when topics appear and how frequently they appear in a particular corpus—in our case, in member periodicals issued by European diabetes patient organizations. The user can zoom in to get a more detailed view of the timeline and also click on interesting sections of the timeline to access the sources in which the topics appear: in this way, the user can quickly shift between distant and close reading. In this paper, we describe how to generate topic timelines and how to use them to select sections of the corpus for close reading. We argue that this timeline extension makes it possible to analyse shifts in medical reasoning and organizational priorities across long time periods in large corpora. Hence, the Topic Timeline tool enables historical research that requires both overviews of large source corpora and close reading of individual texts.