Computational Approaches to the Study of Literary Style

Jonathan Reeve

Computational Approaches to the Study of Literary Style

Jonathan Reeve
Literary Modeling and Visualization Lab
Group for Experimental Methods in the Humanities
Columbia University

Open source philosophy

  • always-already online
  • open-access
  • collaborative

Macro-Etymological Text Analysis

Consider these near-synonyms:

  • ask / question / interrogate
  • kingly / royal / regal
Latinate words, Brown Corpus, by genre
Latinate words, Brown Corpus, by genre
Latinate words, Caterbury Tales
Latinate words, Caterbury Tales
Latinate words, Caterbury Tales, narrative time
Latinate words, Caterbury Tales, narrative time
Latinate words, Paradise Lost, by speaker
Latinate words, Paradise Lost, by speaker
Latinate words, A Portrait of the Artist as a Young Man, by chapter
Latinate words, A Portrait of the Artist as a Young Man, by chapter
Latinate words, CLMET Fiction, by publication date
Latinate words, CLMET Fiction, by publication date

Character Voice

Stylistic Analysis

Clarissa, letters, labeled
Clarissa, letters, labeled
Clarissa, letters, predicted
Clarissa, letters, predicted
Ulysses characters
Ulysses characters

Semantic Analysis

Ulysses characters’ topics
Ulysses characters’ topics
Clarissa, letters, labeled
Clarissa, letters, labeled
Clarissa, letters, predicted
Clarissa, letters, predicted

Histories of Literary Criticism

Middlemarch literary critical quotations, by chapter, diachronic
Middlemarch literary critical quotations, by chapter, diachronic

Words Likely to Appear in Critical Citations, 20th C

life, light, world, woman, new, great, consciousness, live, eye, knowledge, long

Words Likely to Appear in Critical Citations, 19th C

life, Paris, old, live, time, sit, key, soul, prayer, consciousness, belief, marriage

Middlemarch literary critical quotations, by chapter and journal
Middlemarch literary critical quotations, by chapter and journal

Experiments in Procedural Corpus Creation

Problems

  • Difficult to download and manipulate large collections of text files
  • Book metadata is scattered

Solution

  • Aggregate texts from public-domain sources
  • Clean licenses, paratext
  • Aggregate metadata from Wikipedia, Goodreads, Etc.
corpus-db.org
corpus-db.org
Vector similarity of Project Gutenberg texts
Vector similarity of Project Gutenberg texts

Which are higher-rated on Goodreads?

Novels set in Paris, or London?

  • London: 3.35
  • Paris: 3.8

    • Which Austen and Eliot novels are stylistically similar?

    PCA of Austen and Eliot novels
    PCA of Austen and Eliot novels

    Chapters

    • Uses Chapterize
    • Uses corpus-db.org
    Average number of chapters by LCSH
    Average number of chapters by LCSH
    Chapter statistics
    Chapter statistics

    Words distinctive of first paragraphs of chapters

    morning, early, breakfast, afternoon, summer, autumn, winter, sunday, weather, october, arrival, june, september, saturday, awoke, situated, november, july, season, december

    Words distinctive of middle paragraphs of chapters

    replied, retorted, inquired, doesn’t, haven’t, mustn’t, shouldn’t, wid, fer, wi, em, yer, protested, nothin

    Words distinctive of last paragraphs of chapters

    kissed, farewell, bye, muttered, parted, disappeared, sank, page, asleep, strode, chapter, kiss, withdrew, homeward, sobbing, thanked, wept, murmured, prayed

    Allusion Detection

    Allusion Detection

    • Uses custom fuzzy text-matcher (plagiarism detector) over 500 novels
    • “Biblical”: quote/structural resonance from the KJV
    • “Classical”: mention of Greek/Roman diety
    Biblical and classical allusions, by decade
    Biblical and classical allusions, by decade
    Classical allusions, modernist/non-modernist, by decade
    Classical allusions, modernist/non-modernist, by decade

    Most-Quoted Book of the Bible

    • Matthew: 129
    • Luke: 83
    • Psalms: 70
    • Job: 50
    • Isaiah: 48
    • Mark: 46
    • Proverbs: 39
    • Genesis: 34

    Most-Mentioned Greco-Roman Mythological Figure

    • Venus: 229
    • Muses: 156
    • Apollo: 136
    • Jupiter: 126
    • Juno: 100
    • Pan: 94

    Resources

    Do it yourself

    Books

    Notebooks

    DHBox

    DHBox Notebooks

    DH-USB