Elamur Research⌕
☰
home / papers / 2026 / From Gentlemen to Frontiermen: Masculine Formations in English-Language Fiction (1771–1930)

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

Paper · 2026 · Rong Wang

Summary

The paper measures how the balance between different models of manly authority shifts in English-language fiction from 1771 to 1930. It does this reproducibly, instead of drawing conclusions from a few hand-picked canonical texts. The data are 150 British and American novels from the txtLAB Novel450 corpus. BookNLP coreference groups names, nominal descriptions and pronouns into character chains and infers each character's gender. The novels are cut into ~500-word windows, and only windows where male mentions outnumber female ones are kept. A structural topic model (STM) is fitted on this subset, with publication year and author gender as prevalence covariates and no seeded vocabulary. The model yields six masculine formations: one centred on Christian manhood, an aristocratic-chivalric type, a type built around the country squire, one of gentlemanly respectability, a professional-commercial type, and an imperial/adventure type. Formations based on inherited rank and religion decline, while those based on paid work and adventure grow. The steepest rise belongs to the frontier-wilderness side of imperial/adventure masculinity, not to the bourgeois professional. Adventurous and commercial formations are more common in books by male authors. The method shows how to anchor computational gender analysis in actual characters rather than in gendered word lists.

Key points

  • Corpus: the English subset of txtLAB Novel450, with 150 novels by 98 authors.
    • 108 British and 42 American novels.
    • 81 by authors recorded as male, 69 by authors recorded as female.
    • 107 third-person and 43 first-person narratives, spread over five periods (Georgian to Early Modernist).
  • Selection: 35,434 sentence-aware windows of ~500 words, of which 22,197 (63%) are male-referent.
    • A window needs at least two gendered mentions, with male mentions in the majority.
    • Every one of the 150 novels contributes windows.
  • Sanity check: the retained windows have a raw male-pronoun share of 0.75–0.77 in every period. This suggests coreference quality does not degrade on older prose.
  • Model setup:
    • Software: stm 1.3.8 with Spectral (anchor-word) initialisation and prevalence formula year + author_gender.
    • Settings: K = 12, seed 42, at most 75 EM iterations, final vocabulary of 13,195 terms.
    • Topics are labelled afterwards from their top and FREX words.
  • Uncertainty and robustness:
    • Period intervals come from a novel-level bootstrap with 4,000 replicates.
    • The main trends hold under other topic counts and when the most influential novels are left out.

Interesting ideas

  • I1. Define the unit of analysis by who a passage is about, using coreference chains and character gender, rather than by keyword lists; this keeps topics tied to referents instead of to vocabulary that merely sounds relevant.
  • I2. Drop proper names without a gazetteer or capitalisation, using one statistic: a term that is rare across documents (≤10% of them) but frequent inside one document (≥20 hits) is very likely a local name or coinage.
  • I3. Check a pipeline's output with an independent, cruder signal it does not reduce to, broken down by period; here, raw pronoun shares confirm the coreference-based filter on archaic prose.
  • I4. Resample whole documents, not individual passages, when building confidence intervals, so the uncertainty reflects how many independent sources there really are.
  • I5. Accept an explicitly stated recall-for-precision trade-off in data selection and name what it excludes (here, fathers and suitors in female-dominant passages).

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (frames): coreference chains in prag could select the sentences centred on one entity, so frames and roles get aggregated per character instead of per surface word.
  • I2 (world): the same document-specificity test could propose candidate character and place entities for a text's world, even in lowercased or noisy input where NER cues are gone.
  • I3 (learn): add negative-control checks where a simple unrelated metric must agree with the parser or coreference output per text era or genre, exposing silent degradation on older English.
  • I4 (learn): evaluation on fables, tales or treebanks should bootstrap by document, so that thousands of sentences from a few texts do not produce falsely narrow intervals.
  • I5 (learn): when filtering silver data by witness-parser agreement, record which constructions the filter systematically drops so later stages know their blind spots.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of TübingenTübingen · Germany75 works · 41,081 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 16,039 worksSocial Sciences · 1,772 worksLiterature and Literary Theory · 519 worksGender Studies · 52 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,263 works

Cite

Cite as Rong Wang. 2026. From Gentlemen to Frontiermen: Masculine Formations in English-Language Fiction (1771–1930). arXiv:2607.03323.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2607.03323 · arXiv 2607.03323

@misc{wang2026gentlemen,
  title = {{From Gentlemen to Frontiermen: Masculine Formations in English-Language Fiction (1771–1930)}},
  author = {Rong Wang},
  year = {2026},
  doi = {10.48550/arxiv.2607.03323},
  eprint = {2607.03323},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2607.03323}
}

References

37 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. James Eli Adams. Dandies and Desert Saints: Styles of Victorian Masculinity. Cornell University Press, Ithaca, NY, 1995. find
  2. Sanjeev Arora, Rong Ge, Yonatan Halpern, David Mimno, Ankur Moitra, David Sontag, Yichen Wu, and Michael Zhu. A Practical Algorithm for Topic Modeling with Provable Guarantees. In Proceedings of the 30th International Conference on Machine Learning (ICML), pages 280–288, 2013. ↗
  3. David Bamman, Ted Underwood, and Noah A. Smith. A Bayesian Mixed Effects Model of Literary Character. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (ACL), pages 370–379, 2014.
  4. Gail Bederman. Manliness and Civilization: A Cultural History of Gender and Race in the United States, 1880–1917. University of Chicago Press, Chicago, 1995. find
  5. Jonathan M. Bischof and Edoardo M. Airoldi. Summarizing Topical Content with Word Frequency and Exclusivity. In Proceedings of the 29th International Conference on Machine Learning (ICML), 2012. find
  6. David M. Blei, Andrew Y. Ng, and Michael I. Jordan. Latent Dirichlet allocation. Journal of Machine Learning Research, 3:993–1022, 2003. ↗
  7. Rachel Sagner Buurma. The fictionality of topic modeling: Machine reading Anthony Trollope’s Barsetshire series. Big Data & Society, 2(2):1–6, 2015. doi: 10.1177/2053951715610591. ↗
  8. Jonathan Chang, Sean Gerrish, Chong Wang, Jordan L. Boyd-Graber, and David M. Blei. Reading tea leaves: How humans interpret topic models. In Advances in Neural Information Processing Systems 22, 2009.
  9. R. W. Connell. Masculinities. University of California Press, Berkeley, CA, 1995. find
  10. R. W. Connell and James W. Messerschmidt. Hegemonic masculinity: Rethinking the concept. Gender & Society, 19(6):829–859, 2005. doi: 10.1177/0891243205278639. ↗
  11. Nan Z. Da. The computational case against computational literary studies. Critical Inquiry, 45(3): 601–639, 2019. doi: 10.1086/702594. ↗
  12. Leonore Davidoff and Catherine Hall. Family Fortunes: Men and Women of the English Middle Class, 1780–1850. University of Chicago Press, Chicago, IL, 1987. find
  13. Matthew J. Denny and Arthur Spirling. Text Preprocessing for Unsupervised Learning: Why It Matters, When It Misleads, and What to Do about It. Political Analysis, 26(2):168–189, 2018. ↗
  14. Nikhil Garg, Londa Schiebinger, Dan Jurafsky, and James Zou. Word embeddings quantify 100 years of gender and ethnic stereotypes. Proceedings of the National Academy of Sciences, 115 (16):E3635–E3644, 2018. doi: 10.1073/pnas.1720347115. ↗
  15. Mark Girouard. The Return to Camelot: Chivalry and the English Gentleman. Yale University Press, New Haven, 1981. find
  16. Andrew Goldstone and Ted Underwood. The quiet transformations of literary studies: What thirteen thousand scholars could tell us. New Literary History, 45(3):359–384, 2014. doi: 10.1353/nlh.2014.0025. ↗
  17. Alexander Hoyle, Pranav Goel, Andrew Hian-Cheong, Denis Peskov, Jordan Boyd-Graber, and Philip Resnik. Is automated topic model evaluation broken? the incoherence of coherence. In Advances in Neural Information Processing Systems 34 (NeurIPS 2021), 2021. ↗
  18. Matthew L. Jockers. Macroanalysis: Digital Methods and Literary History. University of Illinois Press, Urbana, IL, 2013. find
  19. Matthew L. Jockers and David Mimno. Significant themes in 19th-century literature. Poetics, 41 (6):750–769, 2013. doi: 10.1016/j.poetic.2013.08.005. ↗
  20. Michael S. Kimmel. Manhood in America: A Cultural History. Free Press, New York, 1996. find
  21. Phillip Mallett, editor. The Victorian Novel and Masculinity. Palgrave Macmillan, London, UK, 2015. doi: 10.1057/9781137491541. ↗
  22. David Mimno, Hanna M. Wallach, Edmund Talley, Miriam Leenders, and Andrew McCallum. Optimizing semantic coherence in topic models. In Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing, pages 262–272, Edinburgh, UK, 2011.
  23. John W. Mohr and Petko Bogdanov. Introduction—Topic models: What they are and why they matter. Poetics, 41(6):545–569, 2013. doi: 10.1016/j.poetic.2013.10.001. ↗
  24. Franco Moretti. Distant Reading. Verso, London, 2013. find
  25. Andrew Piper. txtLAB Novel450: A multilingual data set of novels, 2016. Dataset. find
  26. Margaret E. Roberts, Brandon M. Stewart, Dustin Tingley, Christopher Lucas, Jetson Leder-Luis, Shana Kushner Gadarian, Bethany Albertson, and David G. Rand. Structural topic models for open-ended survey responses. American Journal of Political Science, 58(4):1064–1082, 2014. doi: 10.1111/ajps.12103. ↗
  27. Margaret E. Roberts, Brandon M. Stewart, and Edoardo M. Airoldi. A Model of Text for Experimentation in the Social Sciences. Journal of the American Statistical Association, 111(515): 988–1003, 2016. doi: 10.1080/01621459.2016.1141684. ↗
  28. Margaret E. Roberts, Brandon M. Stewart, and Dustin Tingley. stm: An R package for structural topic models. Journal of Statistical Software, 91(2):1–40, 2019. doi: 10.18637/jss.v091.i02. ↗
  29. Maarten Sap, Marcella Cindy Prasettio, Ari Holtzman, Hannah Rashkin, and Yejin Choi. Connotation frames of power and agency in modern films. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 2329–2334, 2017. find
  30. Jason D. Solinger. Becoming the Gentleman: British Literature and the Invention of Modern Masculinity, 1660–1815. find
  31. Palgrave Macmillan, New York, NY, 2012. doi: 10.1057/9780230391840. ↗
  32. Herbert Sussman. Victorian Masculinities: Manhood and Masculine Poetics in Early Victorian Literature and Art. Cambridge University Press, Cambridge, UK, 1995. find
  33. Timothy R. Tangherlini and Peter Leonard. Trawling in the sea of the great unread: Sub-corpus topic modeling and humanities research. Poetics, 41(6), 2013. ↗
  34. John Tosh. A Man’s Place: Masculinity and the Middle-Class Home in Victorian England. Yale University Press, New Haven, CT, 1999. find
  35. Ted Underwood, David Bamman, and Sabrina Lee. The transformation of gender in English-language fiction. Journal of Cultural Analytics, 2018. doi: 10.22148/16.019. ↗
  36. Norman Vance. The Sinews of the Spirit: The Ideal of Christian Manliness in Victorian Literature and Religious Thought. Cambridge University Press, Cambridge, 1985. find
  37. Hanna M. Wallach, Iain Murray, Ruslan Salakhutdinov, and David Mimno. Evaluation Methods for Topic Models. In Proceedings of the 26th Annual International Conference on Machine Learning (ICML), pages 1105–1112, 2009. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wang-2026-gentlemen-frontiermen-masculine-formations-english/