The Shape of Testimony: A Scalable Framework for Oral History Archive Comparison

A computational lens on survivor testimony

Researchers on a new arXiv preprint, "The Shape of Testimony: A Scalable Framework for Oral History Archive Comparison" (arXiv:2605.21623), propose a computational framework to compare styles across oral-history archives. The paper focuses on a long-observed distinction: the USC Shoah Foundation's interviews tend to follow a structured, interviewer-guided format, whereas the Yale Fortunoff Video Archive generally favors a more free-form, open-ended style. It has been reported that this stylistic split has influenced both scholarly research and archival practice, and the authors offer quantitative tools to measure those differences at scale.

Methods, scope and findings

The team describes a scalable pipeline that extracts temporal, lexical and interactional features from long-form interviews, then uses those features to cluster and compare collections. The approach is pitched at digital-humanities scholars and archivists who need replicable metrics rather than ad hoc impressions. The paper is a preprint on arXiv and, as such, presents methods and preliminary analyses rather than final peer-reviewed conclusions; readers should treat reported patterns as provisional.

Implications and cautions

Why does this matter? For Holocaust studies and other fields that rely on survivor testimony, computational comparison can reveal editorial and institutional effects that shape what audiences see and researchers cite. But there are ethical knots too: these archives contain sensitive, often traumatic material. It has been reported that the authors stress privacy-minded practices, yet the broader debate remains — can algorithms ever fully respect the human context of testimony? The paper opens a useful technical conversation while reminding readers that method and moral obligation must travel together.