Research

Methodological Challenge

The International Max Planck International Research School for Multimodal Digital Humanities (IMPRS-MDH) addresses a fundamental methodological challenge: computational approaches in art history and linguistics have traditionally analyzed images, texts, and spatial data in isolation; this monomodal paradigm fails to capture how humanistic research generates knowledge through the systematic interconnection of diverse media and modalities.

Aims

Recent advances in multimodal machine learning create unprecedented opportunities for computational humanities research yet require methodological reconceptualization rather than mere technical adoption. The IMPRS' research program examines how AI systems can model the nuanced argumentative structures characteristic of humanistic interpretation, while simultaneously subjecting these systems to critical cultural analysis.

Core Research Question

How can the interpretive frameworks and argumentative patterns of the humanities be computationally modeled using AI, and how might these models advance both our understanding of cultural history and our critical analysis of AI systems themselves across languages and cultures?

Fields of Enquiry

We pursue this across four connected pathways:

• Epistemic Modeling of Multimodal Reasoning: turning interpretive moves into models to inspect, contest, and test; linking claims to visual and spatial evidence
• Critical AI and Latent Space Analysis: reading multimodal models as cultural formations; probing what they normalize, erase, and privilege
• Multimodal Corpus and Network History: mapping how concepts circulate across media and communities over time, and what becomes visible when the corpus includes images and space
• Concepts and Cross-Cultural Multimodality: studying concepts as cross-modal phenomena, where words, images, gestures, and spaces co-produce meaning differently across cultures.

Multimodal Digital Humanities

The ambition of the IMPRS is to help establish multimodal digital humanities as a field in its own right. This requires methodological reflection beyond technological development as to capture the hermeneutically rich and historically grounded relations between text, image, and space.