💬 NLP

From Gentlemen to Frontiermen: Masculine Formations in English-Language Fiction (1771--1930)

By applying unsupervised structural topic modeling to 150 canonical novels from 1771 to 1930, this study traces the shifting landscape of masculine authority in English-language fiction, revealing a historical transition from inherited and sacred models toward achieved, commercial, and imperial forms of masculinity that are particularly prominent in works by male authors.

Original authors: Rong Wang

Published 2026-07-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Rong Wang

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you have a giant, time-traveling library containing 150 of the most famous British and American novels, written between 1771 and 1930. For over a century, literary critics have argued about how the idea of "being a man" changed in these stories. They've pointed to specific characters—the knight, the gentleman, the frontiersman—as examples. But this paper asks a bigger question: If we look at the entire library at once, how does the balance of power shift between these different types of men?

The author, Rong Wang, uses a computer to read these books not as whole stories, but as a collection of specific "windows" where men are the main focus. Here is how the study works and what it found, explained simply.

The Method: Tuning the Radio to the "Men's Channel"

Usually, when computers read novels, they get confused by everything in the story: the women, the children, the weather, the plot twists, and the men. It's like trying to listen to a specific radio station while a whole orchestra is playing in the background.

To fix this, the author used a smart tool (called "coreference resolution") to act like a sound engineer. This tool identified every time a character was mentioned (by name, pronoun, or description) and grouped them. Then, it sliced the books into 500-word chunks. If a chunk had more mentions of men than women, it kept that "window" for analysis. If it was mostly about women, it set it aside.

This ensured the computer was only analyzing the "men's channel" of the novels, not the whole broadcast.

The Discovery: Six "Scripts" of Manhood

The computer didn't know what a "man" was supposed to look like. It didn't have a list of words like "brave" or "strong" to look for. Instead, it just looked at which words appeared together often. From this, it automatically discovered six distinct "scripts" or "costumes" that men wore in these stories:

The Aristocratic Knight: The medieval hero, the prince, the sword-wielder. (Think King Arthur or old historical romances).
The Christian Man: The man of God, focused on sin, prayer, and the soul. (Think the Victorian preacher).
The Gentleman: The polite city-dweller, focused on letters, proposals, and social manners. (Think Pride and Prejudice).
The Country Squire: The landed farmer, the horseman, the man of the village and the estate.
The Professional Breadwinner: The man of commerce, money, offices, and railroads.
The Imperial Adventurer: The man tested by the wild, the sea, the jungle, and the frontier.

The Trend: A Great Shift in the Spotlight

The study tracked how much "screen time" each of these six scripts got over 160 years. The results show a massive reallocation of attention:

The Old Guard Fades: The scripts based on inherited status (being born a knight or a lord) and sacred authority (being a man of God) slowly lost their spotlight. The "Gentleman" script, which was the most popular in the early 1800s, almost completely vanished by the 1930s.
The New Guard Rises: The scripts based on work and adventure took over.
- The "Professional" (the businessman) grew, but not as much as you might expect.
- The biggest winner was the Imperial/Adventurer. The "Frontiersman"—the man facing the wild, the wolf, and the unknown—became the dominant type of hero in the later novels.

The Surprising Details

It's not just about money: While the "breadwinner" (the guy with a job) became more common, the "adventurer" (the guy in the wild) became the most common. The story of the century isn't just about men going to work; it's about men going to the frontier.
Who wrote them matters: These adventurous and commercial stories were written much more often by male authors. Female authors, meanwhile, kept writing more about the "Gentleman" and social manners, even as that style faded in the broader culture.
It's not just an American thing: Even though the study included many American books (which naturally have more frontiers), the trend of the "Gentleman" fading and the "Adventurer" rising happened in both British and American books.

The Bottom Line

This paper argues that English fiction didn't just swap one type of hero for another. Instead, it redistributed the spotlight.

In the early days, a man's value in a story came from who his father was (rank) or what his God was (religion). By the end of the 19th century, a man's value in a story came from what he could do (work) and where he could go (adventure/frontier).

The study proves that this shift wasn't just a feeling critics had; it is a measurable, statistical reality across hundreds of books. The computer, by listening only to the men in the story, showed us that the "Frontierman" replaced the "Knight" as the ultimate symbol of masculinity in the long 19th century.

Technical Summary: From Gentlemen to Frontiermen: Masculine Formations in English-Language Fiction (1771–1930)

Problem Statement
While gender studies and literary history have long established that masculinity is plural and constituted by competing scripts (e.g., the gentleman, the soldier, the clergyman), existing scholarship has rarely measured the relative weight of these scripts across a century of fiction in a reproducible manner. Traditional analysis often relies on selected canonical examples, making it difficult to observe systematic shifts in the allocation of narrative space to different forms of masculine authority. This paper addresses the gap by asking: when the unit of analysis is restricted to passages narrated about men, how does English-language fiction reweight competing masculine formations between 1771 and 1930?

Methodology
The study employs a computational approach combining character-level coreference resolution with Structural Topic Modeling (STM) on a curated corpus of 150 British and American canonical novels from the txtLAB Novel450 dataset.

Corpus Construction and Filtering: To avoid conflating masculine discourse with general novel themes (such as domestic life or female characters), the authors do not model whole novels. Instead, they process each text using the BookNLP pipeline to cluster character mentions (names, pronouns, nominal descriptions) into character-specific reference chains. The novels are divided into non-overlapping, sentence-aware windows of approximately 500 words. A window is retained for the modeling corpus only if male-character mentions exceed female-character mentions (with a minimum of two gendered mentions). This results in a "male-referent corpus" comprising 22,197 windows (63% of the total), ensuring the model operates on passages where men predominate narratively.
Preprocessing: The pipeline is deliberately conservative and dictionary-free. Tokens are lowercased and normalized; stopwords are removed. Crucially, no seed vocabulary for masculinity is used. Character names are removed via a data-driven statistical criterion (terms appearing in ≤10% of documents but ≥20 times in a single document) rather than an external gazetteer.
Model Estimation: An unsupervised Structural Topic Model (STM) is fitted to the male-referent corpus. Unlike Latent Dirichlet Allocation (LDA), STM allows publication year and author gender to enter the model directly as prevalence covariates. The model is estimated with $K=12$ topics, a choice made on interpretive grounds to ensure theoretically distinct formations (e.g., separating the "country squire" from the "gentleman") emerge as separate topics rather than fused entities. Topics are labeled post-hoc based on their highest-probability words and representative passages, not solely on frequency-exclusivity (FREX) words.
Validation: The study employs multiple robustness checks, including varying the number of topics ( $K$ ), removing the top five novels most heavily loading on each formation, and splitting the corpus by nationality (British vs. American) to ensure trends are not artifacts of compositional shifts.

Key Results
The model identifies six distinct masculine formations, revealing a clear historical trajectory of reallocation from inherited/sacred authority to achieved/commercial/adventurous authority:

Declining Formations:
- Gentlemanly Respectability: The most prevalent formation in the Georgian period (0.226) drops to near-absence by the Early Modernist period (0.015).
- Aristocratic-Chivalric: Shows a significant decline, moving from medieval/dynastic lexicons to near irrelevance.
- Country Squire: The rural estate masculinity declines significantly, though less sharply than the urban gentleman.
- Christian Manhood: Drifts downward but not significantly.
Rising Formations:
- Imperial/Adventure: This formation exhibits the steepest rise (0.049 → 0.244). The increase is driven primarily by the land/frontier-wilderness register (e.g., "wolf," "jungle," "sergeant"), rather than the maritime register.
- Professional-Commercial: Shows a significant rise (0.022 → 0.111), particularly in American fiction, reflecting the ascent of the breadwinner/clerk archetype.
Author Gender and Nationality:
- Adventurous and commercial formations are more prevalent in novels by authors recorded as male.
- Gentlemanly respectability is more prevalent in novels by authors recorded as female.
- The decline of inherited rank and the rise of the frontier/adventure formation hold true within both British and American subsets, indicating the trend is not merely an artifact of the corpus's increasing Americanization over time.

Significance and Claims
The paper claims to offer both a historical account and a methodological contribution:

Historical Contribution: The study provides a long-run, reproducible account of how fiction reorganized masculine authority. It confirms that the shift is not a simple replacement of one ideal by another (e.g., the bourgeois professional replacing the aristocrat) but a reallocation of narrative space. Notably, the paper argues that the most significant rise belongs to the imperial/adventure (specifically frontier) formation, complicating the standard historiography that treats the disciplined bourgeois professional as the sole emblem of modern manhood. The shift represents a move from authority based on inheritance and sacred status toward authority based on paid work, commerce, and the conquest of the wild.
Methodological Contribution: The paper demonstrates a reproducible method for measuring gendered authority that relies on character-level text selection rather than gendered vocabulary lists. By filtering for passages where male characters predominate, the study avoids the pitfall of inferring masculinity from whole-novel themes or vocabulary that merely sounds masculine. The fact that the model, seeded with no masculine vocabulary, spontaneously recovers categories established in independent scholarship (chivalric, Christian, gentlemanly, commercial, frontier) validates the approach as a tool for "distant reading" that remains grounded in specific narrative agents.

The authors maintain modesty regarding their claims, acknowledging that the results describe masculinity within "male-centred narrative space" and may underrepresent relational forms of masculinity (e.g., fatherhood, husbandhood) that appear primarily in female-dominant passages. However, the robustness of the trajectories across topic numbers and influential novels suggests the pattern is a property of the corpus itself.