Veranstaltungen

Hier finden Sie eine Übersicht über aktuelle Veranstaltungen der Juniorprofessur.

Beyond Defaults: Statistical Paradigms in Digital Humanities Scholarship

Datum: November 19–20, 2026

Ort: TU Chemnitz, IdeenReich, Universitätsbibliothek - Straße der Nationen 33, 09111 Chemnitz

Statistical methods have become firmly established in Digital Humanities (DH) research practices. DH scholars routinely employ procedures whose theoretical assumptions differ fundamentally from traditional humanities methods. This uptake is often seamless because many software packages implement statistical paradigms by default. Yet statistical methods are neither neutral nor theory-free. Statistics encompasses distinct theoretical paradigms, such as frequentism and Bayesianism, each carrying different epistemological commitments. While humanities theories are subject to explicit reflection and debate, statistical paradigms typically remain invisible.This asymmetry has consequences for knowledge production in DH. When we model stylistic change, classify texts by genre, or infer networks of influence, we are not simply applying neutral tools. We are making epistemic commitments about what counts as evidence or how uncertainty should be represented. A frequentist framework treats probability as long-run frequency and asks whether observed patterns would recur under repeated sampling. A Bayesian framework, however, treats probability as degree of belief and asks how evidence should update our prior assumptions. These are not interchangeable perspectives. Making such commitments explicit seems to be a matter of methodological reflection.

The workshop approaches these issues from three interconnected perspectives:

1. Epistemic foundations: What assumptions underlie statistical methods? To what extent might DH not only apply statistical methods but also contribute to their theoretical development?

2. Statistical practices in DH: Where are frequentist and Bayesian models already in use, whether recognized or not? How do software defaults shape our methodological choices? And how does the choice of paradigm affect interpretation and evidence evaluation?

3. Perspectives on Large Language Models: How do LLMs, whose implicit inferential commitments remain largely unexamined, relate to classical statistical paradigms? What forms of epistemological scrutiny are needed when DH practice shifts from explicit modeling decisions toward assumptions embedded in training data?

In pursuing these questions, the workshop has two objectives. It seeks to foster methodological reflection by developing approaches for a more deliberate statistical practice. That includes examining how researchers can theoretically justify their choice of methods and interrogating the role that software defaults play in committing scholars to particular paradigms without their awareness. And by bringing together researchers from DH, philosophy of science, and statistics, it initiates an interdisciplinary exchange that has so far remained sporadic. The workshop thus offers first steps toward a new statistical literacy for DH scholarship.

The workshop will be held in English.

Programm

Registration: Die Teilnahme ist kostenfrei, um eine Registrierung unter rabea.kleymann@… wird gebeten