Introduction
As topic modeling matured within digital humanities, a new generation of scholars began to move beyond large-scale thematic mapping and historical trend analysis toward more conceptually nuanced questions about literary form, genre, and the internal organization of texts. Among these scholars, Andrew Piper occupies a significant position.
Piper’s work—particularly in Enumerations: Data and Literary Study—explores how computational methods, including topic modeling, can illuminate not just what texts are “about,” but how they are structurally organized. His approach marks a subtle but important shift:
From themes as content → to patterns as formal and cognitive structures.
1. Corpus and Conceptual Orientation
Piper’s research engages with large corpora of:
- Nineteenth- and twentieth-century novels
- European literary traditions
- Translated and multilingual texts
Unlike earlier studies focused primarily on thematic clustering, Piper’s interest lies in:
- Enumeration (lists, repetition, accumulation)
- Distribution of elements within texts
- The internal architecture of literary works
Topic modeling, in this context, becomes a tool not just for identifying themes, but for detecting patterns of organization.
2. Topic Modeling Beyond Themes
Piper employs Latent Dirichlet Allocation in a way that expands its interpretive scope.
Conventional Use
- Identify clusters of words → interpret as themes
Piper’s Extension
- Analyze how these clusters:
- Appear across different parts of a text
- Interact with each other
- Contribute to narrative structure
Key Insight
Topics are not merely thematic units—they can function as building blocks of textual form.
3. The Idea of Enumeration
A central concept in Piper’s work is enumeration—the listing or accumulation of elements within a text.
Examples include:
- Lists of objects
- Catalogues of places
- Repetitive descriptive sequences
Using topic modeling, Piper demonstrates that:
- Certain topics correspond to enumerative patterns
- These patterns are unevenly distributed across texts
Literary Implication
Enumeration is not trivial—it reflects:
- Cognitive processes
- Narrative strategies
- Cultural modes of representation
4. Distribution Within the Text
One of Piper’s key innovations is shifting attention from:
- What topics exist
to: - Where topics occur within a text
This leads to a more granular analysis:
(1) Opening Sections
- Often dominated by descriptive or situational topics
(2) Middle Sections
- Increased narrative and interactional content
(3) Endings
- Resolution-oriented vocabulary
Structural Reading
Topic modeling thus enables:
A statistical mapping of narrative structure.
5. Genre as Patterned Distribution
Piper extends the concept of genre beyond thematic content.
Traditional View
- Genres defined by themes or conventions
Piper’s View
- Genres defined by:
- Distribution of topics
- Structural patterns within texts
Example
Two novels may share similar themes but differ in:
- How those themes are distributed
- How frequently they recur
- Where they appear in the narrative
Thus:
Genre becomes a matter of internal organization, not just subject matter.
6. Cognitive Dimensions of Topic Modeling
A distinctive feature of Piper’s work is its engagement with cognition.
He suggests that:
- Patterns detected by topic modeling may reflect
underlying cognitive habits of readers and writers
For example:
- Repetition aids memory
- Lists create order
- Clusters of words mirror conceptual grouping
Implication
Topic modeling does not merely analyze texts:
It indirectly models patterns of human thought.
7. Methodological Contributions
Piper’s approach introduces several refinements:
(1) Intra-textual Analysis
- Moving beyond corpus-level analysis
- Focusing on structure within individual texts
(2) Integration with Close Reading
- Using computational findings to guide interpretation
- Returning to specific passages
(3) Formal Analysis
- Emphasizing narrative and structural features
- Not just thematic content
8. Tensions and Critiques
Piper’s work also raises important methodological and philosophical questions.
(1) Overinterpretation of Patterns
- Do statistical distributions truly reflect narrative structure?
(2) Ambiguity of Topics
- Topics may not map cleanly onto formal features
(3) Limits of Quantification
- Subtle literary effects (tone, irony) remain elusive
9. Position within Digital Humanities
Piper’s work represents a further evolution of topic modeling in literary studies:
| Phase | Focus | Representative |
|---|---|---|
| Early | Thematic discovery | Matthew L. Jockers |
| Intermediate | Historical change | Ted Underwood |
| Advanced | Formal and structural analysis | Andrew Piper |
This progression reflects increasing conceptual sophistication:
From content → to history → to structure.
Conclusion
The work of Andrew Piper marks a significant deepening of topic modeling within literary studies. By shifting the focus from thematic identification to structural distribution, Piper repositions topic modeling as a tool for analyzing the internal architecture of texts.
Through the use of Latent Dirichlet Allocation, he demonstrates that literary form—traditionally considered resistant to quantification—can be approached through patterns of repetition, distribution, and organization.
The broader implication is both methodological and philosophical:
That the structure of literature, like its themes, may be understood as a system of patterns—patterns that reflect not only textual organization but also the cognitive processes underlying literary production and reception.