The Curious Engineer

All blogs / AI / ML

The Stone and the Shell

tedunderwood.com

Using large digital libraries to advance literary history

AI / MLEssays / personalMath

’10’15’20’25
Posts
126 from 19 Dec 2010 to 8 Feb 2026
Posting
now and then (1 in the last year)
Kind
personal blog
Feed
https://tedunderwood.com/feed/

2026 1

  1. The marionette theater of AI

2025 1

  1. A more interesting upside of AI

2024 3

  1. Will AI make us overconfident?
  2. Should artificial intelligence be person-shaped?
  3. Can language models predict the next twist in a story?

2023 3

  1. Liberally-educated students need to be more than consumers of AI
  2. We can save what matters about writing—at a price
  3. Using GPT-4 to measure the passage of time in fiction

2021 2

  1. Mapping the latent spaces of culture
  2. Science fiction hasn’t prepared us to imagine machine learning.

2020 1

  1. How predictable is fiction?

2019 2

  1. Humanists own the fourth dimension, and we should take pride in it.
  2. Do humanists need BERT?

2018 4

  1. Do topic models warp time?
  2. Remarks for a panel on data science in literary studies, in 2028
  3. New methods need a new kind of conversation
  4. A broader purpose

2017 5

  1. It looks like you’re writing an argument against data in literary study …
  2. We’re probably due for another discussion of Stanley Fish
  3. Finding the great divide
  4. Digital humanities as a semi-normal thing
  5. Two syllabi: Digital Humanities and Data Science in the Humanities.

2016 9

  1. The Gender Balance of Fiction, 1800-2007
  2. Notebooks? What good are notebooks?
  3. A distant reading of moments
  4. A more intimate scale of distant reading.
  5. The real problem with distant reading.
  6. Versions of disciplinary history.
  7. Three nasty problems.
  8. You say you found a revolution.
  9. The instability of gender

2015 10

  1. Emerging conversations between literary history and sociology.
  2. Can we date revolutions in the history of literature and music?
  3. Digital humanities might never be evenly distributed.
  4. A dataset for distant-reading literature in English, 1700-1922.
  5. Seven ways humanists are using computers to understand text.
  6. How quickly do literary standards change?
  7. Free research question about plot.
  8. Why it’s hard for syuzhet to be right or wrong yet.
  9. Syllabus for a graduate seminar.
  10. “Plot arcs” in the novel.

2014 6

  1. How to find English-language fiction, poetry, and drama in HathiTrust.
  2. Distant reading and the blurry edges of genre.
  3. A window on the twentieth century may be about to open.
  4. You can’t govern reception.
  5. How much DH can we fit in a literature department?
  6. New models of literary collectivity.

2013 16

  1. Measurement and modeling.
  2. The imaginary conflicts disciplines create.
  3. A half-decent OCR normalizer for English texts after 1700.
  4. Genre, gender, and point of view.
  5. Hold on loosely; or, Gemeinschaft and Gesellschaft on the web.
  6. One way numbers can after all make us dumber.
  7. Interesting times for literary theory.
  8. On not trusting people who promise “to use their powers for good.”
  9. Against (talking about) “big data.”
  10. A new approach to the history of character?
  11. On trolling.
  12. The long history of humanistic reaction to sociology.
  13. Distant reading and representativeness.
  14. Wordcounts are amazing.
  15. We don’t already understand the broad outlines of literary history.
  16. On the novelty of “humanistic values.”

2012 25

  1. What can topic models of PMLA teach us about the history of literary scholarship?
  2. Visualizing Topic Models with Force-Directed Graphs
  3. Visualizing topic models.
  4. Problems of scale.
  5. How everyone gets to claim they do DH.
  6. How not to do things with words.
  7. Where to start with text mining.
  8. Getting everything you want from HathiTrust.
  9. I like “open.” And I like “review.” But do they need to be fused?
  10. It’s the data: a plan of action.
  11. OT: politics, social media, and the academy.
  12. The obvious thing we’re lacking.
  13. Why DH has no future.
  14. Topic modeling made just simple enough.
  15. What kinds of “topics” does topic modeling actually produce?
  16. A touching detail produced by LDA …
  17. Etymology and nineteenth-century poetic diction; or, singing the shadow of the bitter old sea.
  18. Big but not distant.
  19. Literary and nonliterary diction, the sequel.
  20. The differentiation of literary and nonliterary diction, 1700-1900.
  21. Syllabus: ENGL581: Digital Tools and Critical Theory.
  22. Do humanists get their ideas from anything at all?
  23. Fish wins round two.
  24. MLA talk: just the thesis.
  25. A brief outburst about numbers.

2011 28

  1. Why digital humanities isn’t actually “the next thing in literary studies.”
  2. What no one tells you about the digital humanities.
  3. Exploring the relationship between topics and trends.
  4. Identifying diction that characterizes an author or genre: why Dunning’s may not be the best method.
  5. On transitive and intransitive uses of the verb “theorize.”
  6. LSA is a marvellous tool, but literary historians may want to customize it for their own discipline.
  7. The challenges of digital work on early-19c collections.
  8. A course description.
  9. Words that appear in the same 18c volumes also track each other over time, through the 19c.
  10. It’s okay not to solve “the crisis of the humanities.”
  11. Topics tend to be trends. Really: p < .05!
  12. For most literary scholars, text mining is going to be an exploratory tool.
  13. Digital humanities and the spy business.
  14. Why humanists need to understand text mining.
  15. The history of an association, part two.
  16. The history of an association.
  17. Should we model “topics” as associative clusters, or as statistical factors?
  18. Trends, topics, and trending topics.
  19. The key to all mythologies.
  20. Revealing the relationships between topics in a corpus.
  21. “… a selection of the language really spoken by men”?
  22. Using clustering to explore the relationships between trending topics.
  23. A bit more on the tension between heuristic and evidentiary methods.
  24. How you were tricked into doing text mining.
  25. Seriously geeking out.
  26. Why everyone should welcome the coming controversy over digital humanities.
  27. The Google dataset as an episode in the history of science.
  28. Identifying topics with a specific kind of historical timeliness.

2010 10

  1. How to make the Google dataset work for humanists.
  2. Several varieties of noise, and the theme to Love Story.
  3. On different uses of structuralism; or, histories of diction don’t have to tell us anything about “culture” to be useful.
  4. More reflections on the apparent “structuralism” in the Google dataset
  5. The rise of a sensory style?
  6. Colors
  7. Efficiency and pleasure
  8. On the imperfection of the Google dataset, and imperfection in general
  9. Using changes in diction to frame historical questions ≠ ‘culturomics’
  10. What I hope this blog will do