Canons within canons?

Anthology selections of US literature across series

Erik Fredner

2026-06-18

Follow along

https://fredner.org/mapping-canon

You can interact with many of the slides.

Question

How do editors reshape a literary tradition from one comprehensive anthology for another anthology that has less room for that tradition?

Hypotheses

  • Increasing selection pressure on existing anthology selections tends to favor the canonical core.
  • The greater the selection pressure, the greater the emphasis on the core.

Findings

  • Increasing selection pressure on preexisting anthology selections tends to favor the canonical core.
    • Mostly true.
    • A strong reselection record within an anthology series is necessary but not sufficient to survive a further cut.
  • The greater the pressure, the greater the emphasis on the core.
    • Partially true.
    • Canonicity is contextual, not absolute.

Project history

Figure 1: Author-edition relations in The Norton Anthology of American Literature from “Counting on NAAL,” co-authored with J.D. Porter.
Figure 2: Author-edition relationships across anthologies of African American literature from “The ‘Implicit’ Canon,” co-authored with J.D. Porter and Arthur Knight.

What’s different?

  • Previous analyses compared selections across anthologies with similar scales and mandates.
  • “Canons within Canons” compares comprehensive anthology selections to abridgements of their traditions.
    • e.g., US literature as represented in world literature.
  • Previous studies faced a problem: New selections were made from an unknowable pool of candidates.
    • Shrinking an existing anthology reveals cuts from a known candidate pool.

Approach

Norton anthology series analyzed

  • African American Literature (NAFAM) (4 eds., 1997–2025)
  • American Literature (NAAL) (10 full + 10 shorter eds., 1979–2023)
  • Western Literature (9 eds., 1956–2014)
  • World Literature (5 full + 5 shorter eds., 1995–2024)

Western vs. World Literature

  • The Western and World Literature series share a common origin: World Masterpieces (1956–1999)
    • Western Literature continues World Masterpieces
    • 1995 “expanded edition” of World Masterpieces is the first edition of World Literature
  • With respect to US literature, these series are continuous
  • I abbreviate the pair as NAWL

How large are the abridgments?

Table 1: Mean reduction of candidate authors and candidate works from source series to target series across all editions.
Source Target Reduction
NAAL NAAL shorter -41%
NAFAM NAAL -68%
NAAL + NAFAM NAWL -95%

Data

Key relational database records

  • Handmade!1
  • 986 authors + metadata
  • 5,787 works + metadata
  • 31,948 records of works in anthology volumes

Example: “‘Song of Myself’ by Walt Whitman appears on pages 23-66 of Volume C of NAAL (2022).”

Database principles

  • Represent selections as they appear in the table of contents (TOC)
  • But we also reconcile recurring entities when TOC representations differ across anthologies
  • We do not claim textual equivalence between reconciled works

NAAL full to shorter

*NAAL* full to shorter reduction per edition.

Figure 3: NAAL full to shorter reduction per edition.

Figure 4: Authors by their selections for NAAL full and shorter.
Figure 5: Authors by reselection opportunities for NAAL full and shorter.

Figure 6: Most NAAL shorter selections have always come from the first two editions.

Figure 7: Signature works surviving from NAAL full to shorter.

NAAL to and from NAFAM

What does NAFAM 1E include from earlier NAAL editions?

  • 4 NAAL editions predate the first NAFAM (1997)
  • First NAFAM selects 27 of 28 (96%) candidate authors who had appeared in in prior NAAL editions
  • But NAFAM only selected 45 of 136 (33%) candidate works
  • Different anthologies read the same authors in different ways, as suggested by author-work disparity

What do later NAAL editions pick up from NAFAM?

  • 32 of 155 authors (21%) who debut in NAFAM later appear in NAAL
  • 47 of 640 works (7%) that debut in NAFAM later appear in NAAL
Figure 8: NAFAM authors’ work selection records in NAFAM vs. NAAL.

NAAL + NAFAM to NAWL

95% reduction takes us to individual cases.

Who survives from NAFAM to NAWL?

  • 8 of 211 (4%) of NAFAM authors have ever been selected for NAWL
  • Only Frederick Douglass was selected for a majority of editions (13 of 19)
  • James Baldwin has the second-highest selection rate (7 of 19)

Who survives from NAAL to NAWL?

  • 34 of 408 (8%) of NAAL authors exclusive of NAFAM authors have ever been selected for any edition of NAWL
  • NAAL authors appearing in most NAWL editions: T. S. Eliot (19 of 19), Walt Whitman (17), Emily Dickinson (16), Herman Melville (14), Frederick Douglass (13; also in NAFAM), William Faulkner (13), and Leslie Marmon Silko (12)
  • Of these, Faulkner is the only one who has been cut

Leslie Marmon Silko

Silko (Laguna Pueblo, born 1948) is the only highly reselected NAWL author who could not have been guessed correctly before 1979 (the first NAAL).

Figure 9: All NAWL works from NAAL + NAFAM candidates. Silko only has one NAWL work (“Yellow Woman”), which never appeared in NAAL.

Conclusion

  • Increasing selection pressure favors the canonical core
  • Strong reselection within one series is necessary but not sufficient for promotion to another
  • Anthology series tend to reproduce themselves rather than select across series, even within the same publisher
  • As editors shrink canons, they reshape them
    • Abridging NAAL from full to shorter is different than abridging American literature for NAWL, and the difference is not entirely explained by size

Next steps

  • Modeling the “test of time” with this data
    • Many authors’ and works’ canonicity predates these anthologies (e.g., most top NAWL selections)
    • Authors without prior canonicity isolate the effect of anthologies
  • To what extent does page allocation relative to literary form predict promotion?

Thank you!

Appendix

Figure 10: NAAL first edition debut authors and works are significantly more likely to be promoted from full to shorter editions than almost all other debuts.

Figure 11: Share of NAAL selections that have ever appeared in NAFAM increases over time.

Author NAWL Selections MLAIB subject records
T. S. Eliot 19 8,534
Walt Whitman 17 5,182
Emily Dickinson 16 4,609
Herman Melville 14 7,864
Frederick Douglass 13 1,175
William Faulkner 13 8,381
Leslie Marmon Silko 12 813
Figure 12: NAAL selection rate by birth year decile and NAAL debut year.

Database schema (scroll)

erDiagram
    data_series ||--o{ data_edition : "has"
    data_edition ||--o{ data_volume : "has"
    data_volume ||--o{ data_workinanthology : "contains"
    data_workinanthology }o--|| data_work : "selects"
    data_work ||--o{ data_work : "parent_id (excerpt of)"
    data_work ||--o{ data_work_authors : ""
    data_work_authors }o--|| data_author : ""
    data_work ||--o{ data_work_form : ""
    data_work_form }o--|| data_form : ""
    data_author ||--o{ data_author_countries : ""
    data_author_countries }o--|| data_country : ""
    data_author ||--o{ data_author_genders : ""
    data_author_genders }o--|| data_genders : ""
    data_author ||--o{ data_author_raceeth : ""
    data_author_raceeth }o--|| data_raceeth : ""

    data_series {
        int id PK
        varchar name
        varchar abbreviation "NAFAM, NAAL, NAWest, NAWL"
    }

    data_edition {
        int id PK
        int series_id FK
        varchar edition_number
        int year
        varchar variant "'' or 'Shorter'"
        varchar title
        int publisher_id
    }

    data_volume {
        int id PK
        int edition_id FK
        varchar volume_number
        varchar title
        int first_toc_page
        int last_toc_page
        varchar lccn
    }

    data_workinanthology {
        int id PK
        int work_id FK
        int volume_id FK
        int toc_page "start page"
        int toc_next "end page (guard toc_next > toc_page)"
        bool has_untitled_excerpt
        varchar reviewed
    }

    data_work {
        int id PK
        int parent_id FK "NULL for top-level; set for excerpts"
        varchar title
        bool is_editor_title
        text notes
    }

    data_work_authors {
        int id PK
        int work_id FK
        int author_id FK
    }

    data_author {
        int id PK
        varchar name
        int birth_year
        int death_year
        bool birth_death_circa
        varchar viaf_url
        varchar wiki
    }

    data_work_form {
        int id PK
        int work_id FK "canonical work id"
        int form_id FK
    }

    data_form {
        int id PK
        varchar name "poetry, fiction, nonfiction, drama, folk, song, screenplay, na"
    }

    data_author_countries {
        int id PK
        int author_id FK
        int country_id FK
    }

    data_country {
        int id PK
        varchar name
    }

    data_author_genders {
        int id PK
        int author_id FK
        int genders_id FK
    }

    data_genders {
        int id PK
        varchar name
    }

    data_author_raceeth {
        int id PK
        int author_id FK
        int raceeth_id FK
    }

    data_raceeth {
        int id PK
        varchar name
    }

AI Statement

  • Anthology selection data is handmade and validated by project contributors.
  • I used language models such as Claude to help write code for queries, analyses, figures, etc.