Module 4. Modeling in the Humanities: Meaning, Context, and Ambiguity

Course 3. Modeling Meaning: How We Structure Humanities Data
Estimated Time: 35–40 minutes

🧭 Module Objectives

• Explain why modeling in the humanities involves interpretation and uncertainty.
• Recognize how context and cultural meaning shape data design.
• Describe how ontologies like CIDOC-CRM and TEI formalize humanistic knowledge.
• Discuss ethical and representational challenges in structuring human experience.
• Reflect on how data models both reveal and limit the stories we can tell.

The Challenge of Modeling Meaning

In the humanities, our data are not just facts: they are interpretations of lived experience, culture, and creativity. We deal in nuance: irony, metaphor, ambiguity, and contradiction.

When we model such data, we face a paradox: "How do we structure meaning without reducing it?"

A spreadsheet can list Jesse Welles' songs by title, year, and album, but it can't capture the feeling of "Horses" or the irony of "War Isn't Murder." Humanistic modeling therefore requires context, flexibility, and transparency about what we're representing—and what we're leaving out.

Every Model Is a Theory

Modeling always involves interpretation. When you decide that a song "expresses" a theme of war, you're already making a claim about meaning. As we saw in our previous course, Johanna Drucker writes that "data are capta": not given but taken: selected, framed, and defined through human intention.

A data model is therefore not a neutral map of reality, but a theory of how reality is organized. This is especially true in the humanities, where meaning is negotiated rather than measured.

Context Is Everything

Humanistic data gains meaning only within context:

• A song carries different implications when played at Farm Aid than when shared on TikTok.
• A word in a lyric changes depending on its cultural and historical moment.
• A place (like Ozark, Arkansas) means something different to locals than to distant listeners.

Contextual modeling means linking data to its situations of use: time, place, audience, performance, reception.

Capturing these contexts requires models that can connect across domains: not just what happened, but why and how.

Modeling Ambiguity and Uncertainty

Humanists often work with incomplete, conflicting, or subjective evidence. A traditional database prefers clarity: "true" or "false." But cultural data lives in shades of gray.

Digital humanists use several strategies to handle this:

• Qualitative tagging (e.g., "probable," "disputed," "approximate").
• Alternative relationships (e.g., "possibly influenced by," "attributed to").
• Temporal ranges (e.g., date: [1972-1974]).
• Contextual commentary stored alongside structured data.

Example: "The song War Isn't Murder is possibly influenced by Bob Dylan's Masters of War."

This phrasing preserves interpretive uncertainty while still enabling connections: a hallmark of ethical modeling.

Formal Frameworks: Data Ontologies and Standards

To share data responsibly, humanists often build on established frameworks that express relationships and context in a standardized way.

🏛 CIDOC-CRM (Conceptual Reference Model)

Used in museums and heritage projects globally, the CIDOC-CRM defines events, actors, objects, and relationships as conceptual entities. It's designed to capture process and context—not just static facts.

By aligning with such ontologies, we make our models interoperable—able to "speak"—with other cultural data systems.

✍️ TEI (Text Encoding Initiative)

For textual scholars, TEI provides an internationally-developed schema for encoding literary and linguistic features:

• Structural: <div>, <line>, <speaker>
• Interpretive: <note>, <interp>, <theme>
• Contextual: <date>, <placeName>, <persName>

A TEI-encoded lyric might mark a line as both ironic and biblical in reference: something a simple database row could never do.

These standards embody a long tradition of modeling interpretation without erasing ambiguity.

Ethics, Representation, and Responsibility

Modeling human experience carries ethical stakes. How we represent people, events, or ideas can shape how others understand them.

Ethical modeling asks:

• Who decides what gets modeled — and what’s left out?
• How do we avoid re-inscribing bias or oversimplifying identity?
• Are we clear about the limits of our model’s claims?

In the Wellespring Project, this means treating artistic expression and audience reaction as living, interpretive data: not reducing them to statistics, but situating them within narrative, community, and care.

The Productive Power of Incompleteness

No model can capture everything. And that's okay!

The goal is not perfection, but productive incompleteness: a model that invites revision, dialogue, and further interpretation. As your understanding grows, your model evolves: just like scholarship itself.

In the humanities, this openness is a strength: it keeps data alive, responsive, and reflective of ongoing meaning-making.

Key Takeaways

• Humanities data modeling is interpretive, contextual, and uncertain.
• Context gives data meaning; without it, we risk distortion.
• Ontologies and frameworks like the CIDOC-CRM and TEI formalize complex relationships without erasing nuance.
• Ethical modeling emphasizes transparency, responsibility, and humility.
• Incompleteness isn't failure: it's an invitation to continued inquiry.

Knowledge Check & Reflection

Suggested Readings & Resources

• D’Ignazio, Catherine, and Lauren F. Klein. Data Feminism. The MIT Press, 2020.
  • Directly addresses the module's ethical questions: "Who decides what gets modeled?" and "How do we avoid bias?" This book provides a powerful, modern framework for thinking about power, representation, and justice in data science and modeling.
• Drucker, Johanna. "Humanities Approaches to Graphical Display." Digital Humanities Quarterly 5 (2011).
  • This is the foundational text for the module's argument that data are not given but taken (capta). It directly explains why humanistic data modeling is an interpretive act.
• Scarpa, Erica. "Demystifying the CIDOC CRM: A Lightweight Introduction." Archeologia e Calcolatori 36 (2025): 481–92.
  • This article provides a concrete, practical example of using the CIDOC-CRM to handle the very challenges this module discusses: uncertainty, temporal vagueness, and interpretive claims in historical/archaeological data. It moves beyond theory to show how a model can formally represent ideas like "possibly dated to" or "might have been used for," making it a perfect real-world extension of the module's section on "Modeling Ambiguity and Uncertainty."
• TEI Consortium. "A Gentle Introduction to XML." In TEI: Guidelines for Electronic Text Encoding and Interchange, P5 Version 4.10.2. Text Encoding Initiative Consortium, 2025.
  • The module mentions TEI for marking up irony and biblical references. This guide is the perfect next step for students to understand how to implement those ideas using XML tags, connecting the theoretical to the practical.
• Trettien, Whitney. Cut/Copy/Paste: Fragments from the History of Bookwork. University of Minnesota Press, 2021.
  • While not about data modeling in a digital sense, this work brilliantly explores how historical "models" of books (like composite books made of fragments) were always incomplete, recombinant, and evolving. It offers a beautiful, concrete historical analogy for the module's concept of "productive incompleteness."