The testimony of the multitude: toward a computational model of listening to Holocaust testimony

Abstract Decades of effort collecting Holocaust testimonies have created a critical challenge for scholars: how to listen to thousands of stories as a collection of voices rather than as fragmentary data. Unlike text-mining methods, which reduce text to its basic components, this article proposes a new mode of “listening” via advanced computing. It argues that technology is foundational to bridging the gap between mass atrocity and mass testimony. This article presents a theoretical and computational model of “distant listening” that engages an entire archive without compromising the singular human experience. By treating each testimony as an unfolding conversational narrative and analyzing all testimonies within a select group simultaneously, the model listens to testimonies in relation to one another. This approach uncovers hidden characteristics and deviations from typical narrative schemas, ensuring that these safeguarded voices do not remain unheard.