Working draft. The text and tools are still being checked, and corrections are welcome.

How to read these numbers

What is counted

Every count here is a count of names printed in a convention program. A program is a plan. It lists what CCCC accepted and scheduled, so it says who was invited onto the stage, and it does not say who attended, who showed up to speak, or who stayed to listen. The 2020 program describes a convention that was cancelled.

Where the series breaks

  • No program in the corpus (2000–2003)
  • 2020: program of a cancelled convention
  • 2021–2022: virtual conventions
  • 2024: source changes from PDF to eShow
  • 2027: listing before the convention

The charts shade these years. A change across a break may come from the source. For example, the step between 2023 and 2024 may come from the move from PDF programs to eShow, and the jump at 2027 comes from a listing that is still filling in.

Where the programs come from

1998 comes from the convention's web program in the Wayback Machine. 1999 comes from the PDFs that survive there, for ten of fourteen letter sections. 2004–2023 come from NCTE's program PDFs, and 2024–2027 from NCTE's eShow listings. 1994–1997 survive online only as front pages, and 2000–2003 are not online at all. The years before 1998 need the printed programs.

The PDF parser was scored against eShow for 2024–2026, the years both exist: it finds the listed names with precision 0.94–0.97 and recall 0.92–0.95. For six earlier years (2004, 2005, 2007, 2008, 2010, 2011) the program ends with an index of participants that lists each person with the codes of their sessions. Against that index the parser's (person, session) pairs reach precision 0.87–0.97 and recall 0.82–0.94. Two known gaps account for most of the misses: the Research Network Forum (code RN), a single event with 90 to 110 presenters a year that the parser does not read, and 2004, whose special-interest sessions are printed without codes in the program body. A human check of 40 randomly chosen sessions from 2004–2023 (Justin Lewis, October 2026) found no errors, which supports the figures without proving them: it is a small sample.

Names

People are joined across years when the printed names match exactly (case, punctuation and accents are folded). A reviewer (Justin Lewis, October 2026) also checked about 1,600 pairs of near-matches (a middle initial, a nickname, a one-letter typo, part of a hyphenated surname) and the 318 names that appear at two institutions in a year or return to an earlier one, and all but one pair were joined as one person. Names that differ in other ways stay separate, so counts of first listings run a little high and counts of returners a little low. Two different people with the same name still become one, and names that look like two careers are marked on their pages.

Institutions

A printed affiliation is matched to a record in ROR and, for US colleges, to IPEDS. IPEDS says whether a college is two-year (IPEDS level 2, or Carnegie basic class 1–14). Shares of institution types use people whose institution matched a US college in IPEDS. About 3% of printed affiliations match nothing: independent scholars, schools, publishers and a long tail.

Proposal areas

From 2004 through 2026 (except 2018) the programs print an area above most sessions. We read those lines as CCCC printed them. They agree with eShow's own track tags for every 2025 and 2026 session that has both. Workshops, special interest group meetings and many special sessions carry none, so shares are of sessions that have one. The families that join a renamed area across years (for example, Practices of Teaching Writing in 2005 and Approaches to Teaching and Learning in 2023) are listed in data/area_crosswalk.csv. I drafted them and one reader (Justin Lewis, October 2026) checked all of them and made no changes. They remain a judgment call, not CCCC's own list. The lists of areas also differ by era (first-year composition and social justice become areas only later, and institutional-and-professional ends in 2019), so a change in a family's share can reflect CCCC's headings and not the field's interests.

Notes by year

  • 1998: The convention's web program, replayed from the Wayback Machine. Sections A and J are cut short in the captured pages.
  • 1999: Ten of fourteen letter sections survive as PDFs. Sections E, F, K and M have session titles only, from the web program's session pages.
  • 2008: Read from NCTE's complete program (321 pages). An earlier build used the day files, which lacked Saturday.
  • 2018: NCTE's 2018 program is laid out differently from the neighboring years, and the parser has less to go on there.
  • 2020: This is the printed program of a convention that was cancelled. It shows what was scheduled, not what happened.
  • 2021: A virtual convention. The program is split across several files, including on-demand sessions.
  • 2022: A virtual convention. The program is split across several files, including on-demand sessions.
  • 2024: Read from NCTE's eShow listing, with paper titles copied from the PDF program.
  • 2025: Read from NCTE's eShow listing, with paper titles copied from the PDF program.
  • 2026: Read from NCTE's eShow listing, with paper titles copied from the PDF program.
  • 2027: A listing taken before the convention. It changes as NCTE adds and cancels sessions, and is re-captured monthly through May 2027.

Terms

The terms tool counts a word wherever it appears in a title, with word boundaries (“labor” does not match “collaboration”). A match shows that the word was in a title, not what the session argued. Open the titles before reading a trend.