At a glance · 概览
Curriculum over time · 每场集体学习按主题分行
Each row is a theme; each dot a session on its date. Shaded bands mark the three CC terms. A red-ringed dot = a published Xi speech exists (click → side-by-side compare); a plain dot = readout only (click → readout). Click a legend entry to mute/unmute a theme.
Theme matrix · 主题 × 届
Sessions per theme per term. Cell shading = that theme's share of the term's curriculum; the number is the session count. Rows sorted by total — the agenda's rise and fall at a glance.
Who teaches Xi · 讲解人来自哪类机构
Lecturers' home-institution type per term — the academic ↔ official ↔ military mix, and how it shifts. 每场集体学习按讲解人所在机构类型归类（同一场可计入多类）。
Top named institutions · 重点讲解机构 · click to expand each session
The specific organizations that supply Politburo study lecturers, ranked by number of sessions. Click any institution to open every session it taught — the lecturer, topic and date, with a link to that session's side-by-side readout / speech compare. 点击任一机构，展开其全部讲解场次：讲解人、议题、日期，并链接到该场的“通稿（读出）/ 讲话全文”并排对照页。
The lecturers · 讲解人全谱
Not every session has an outside teacher. Across all — sessions, — distinct outside experts briefed the — expert-briefed sessions; the rest were Politburo self-study or field study — members speak, no external lecturer. Only — experts were ever invited back. 并非每场都有外请专家：部分为政治局“自学交流”或“现场调研”。下方先按授课形式与单位类型统计，再列全部讲解人及其讲解议题（中英对照）。
Full roster · 全部讲解人 · name · institution · what they lectured on
Cadence · 节奏（相邻两场间隔天数）
Median = · Mean = · Max gap plotted at x.
Readout length · 通稿字数
Histogram of per-session body text size. Length proxies how much detail Xinhua released.
History-session deep dive · 18 sessions, 2013–2024
Companion to the allusion work. Full prose version: HISTORY_SESSIONS_ANALYSIS.html.
The only 10 classical formulas Xi quotes across all 18 sessions
| Formula | Source | Sessions |
|---|---|---|
| 民惟邦本、政得其民 | 尚书·五子之歌 | 018 |
| 礼法合治、德主刑辅 | Confucian legal tradition | 018 |
| 为政之要莫先于得人 / 惟在得人 | 贞观政要 (Tang) | 018 · 053 |
| 治国先治吏 | classical proverb | 018 · 053 |
| 居安思危、改易更化 | 左传 + Han reform | 018 |
| 蠹众而木折，隙大而墙坏 | 商君书·修权 | 005 |
| 善则赏之、过则匡之、患则救之、失则革之 | 国语·晋语七 | 053 |
| 择其善者而从之、其不善者而去之 | 论语·述而 | 018 |
| 法安天下、德润人心 | classical pairing | 037 |
| 育才造士，为国之本 | 欧阳修 (Song) | 053 |
No poetry. No 苏轼, no 王安石, no 张居正. No foreign attribution beyond Marx/Engels as framework. Contrast 用典 第一辑+二辑: 415 classical citations with Tang/Song poetry dominant.
Four-phase arc across Xi's three terms
History as method. Party history, historical materialism, imperial governance as model. Xi explicitly rules out 历史虚无主义.
History mobilized for policy: Silk Road → OBOR, WWII commemoration, law+virtue doctrine, patriotism.
Party / revolutionary history moves to front. Centenary prep. Offsite session at 北大红楼 + 毛泽东故居 (074). 党史学习 campaign.
探源工程, 中华民族共同体, 边疆治理 — history underwriting Xinjiang/Tibet/frontier policy via civilizational unity.
Repeat lecturers
| Lecturer | # | Sessions | Affiliation |
|---|---|---|---|
| 卜宪群 | 3 | 005 · 018 · 053 | CASS 历史所 |
| 李国强 | 2 | 031 · 101 | CASS 边疆所 → 历史研究院 |
12 of 18 lecturers are CASS or its 2019 spinoff 历史研究院 (which Xi himself established). The institutional pipeline Xi built now feeds his own curriculum.
Xi's signature history-prose (non-classical, recurring)
| Slogan | First | Recurs |
|---|---|---|
| 五千(多)年文明 | 018 | 007 · 029 · 057 · 066 · 082 · 093 |
| 讲仁爱、重民本、守诚信、崇正义、尚和合、求大同 | 013 (2014) | 082 (2022) — verbatim, 8yrs later |
| 创造性转化、创新性发展 | 013 | 029 · 066 · 082 · 093 |
| 中华民族共同体 | 029 (embryonic) | 082 · 093 · 101 |
| 历史虚无主义 (as enemy) | 018 | 074 |
| 大历史观 | 057 | — |
Vocabulary shifts across three terms · 三届话语演变
Keyword rates normalized per 10,000 characters so terms of different volume compare fairly; signature phrases via weighted log-odds (boilerplate suppressed). Toggle between the third-person readout corpus (all 107 sessions) and Xi's first-person speech corpus (the sessions with a published speech). 按每万字归一化的词频；可在第三人称“读出/通稿”语料与习近平第一人称“讲话全文”语料之间切换。
Thematic bucket trajectories · phrase-rate · 主题词频轨迹
Sensitivity check. Where the two views agree, the trend is robust; where they diverge (e.g. Party/ideology's CC18 peak shrinks under per-document), the level is driven by a few long documents. Read the slope, not the rank — bucket heights also reflect how many phrases a bucket contains (6–28). 敏感性检验：两种归一方式一致则趋势稳健；若分歧（如意识形态CC18峰值在“按篇”下回落）则说明被少数长文档主导。看趋势而非高低。
Which phrases are in each bucket? 各主题桶包含哪些词条？ — every phrase is verified to occur in the corpus; counts shown are total occurrences
Top 24 phrase shifts · by rate-range across terms
Numbers in each cell are raw counts. Bar height and Δ are normalized per 10K body-text chars (CC18≈70K · CC19≈92K · CC20≈53K) so differences in how much text Xinhua published each term don't distort the comparison. Hover a cell for the exact rate.
Signature 4-grams per term · log-odds distinctive (readouts)
Character 4-grams most distinctive to each term by weighted log-odds (informative Dirichlet prior) — characteristic of this term vs. the other two, with meeting-template boilerplate suppressed. Reads as fragments of each term's agenda.
Readout vs. speech · 通稿与讲话全文的语言差异
For the — sessions that have both a People's-Daily readout (third-person summary, 通稿) and Xi's later-published speech (first-person, 讲话全文), language is compared by weighted log-odds (Monroe, Colaresi & Quinn) and curated-phrase rate per 10K chars — on the same sessions, so this isolates register: what Xi emphasizes vs. what the summary foregrounds. Caveat: only ~half of sessions have a published speech (skewed toward ideology / economy / Party topics), so the cross-term speech grams further down are suggestive; this paired diff is the rigorous comparison.
Register signature · marker rate per 10K, paired sessions
First-person & directive markers. Readouts attribute in the third person (强调 = "[he] stressed"); speeches speak in the first person (我 / 我们).
Xi emphasizes more · than the readout conveys
Readout foregrounds more · than Xi's own speech
Speech vocabulary by term · most distinctive 4-grams (log-odds) — read with the selection caveat
Sessions
One acknowledged gap. Across all 108 numbered sessions, CC20 session 13 is absent from the People's Daily index (the upstream source for this corpus). The official numbering jumps from 第十二次 (2024-02-29 · 新能源技术与我国的能源安全) to 第十四次 (2024-05-27 · 促进高质量充分就业) — an 88-day gap vs. CC20's 54-day median. Most likely a session held but never publicly reported. No other CC term has a numbering gap. A placeholder row appears at the bottom of the table below.
| #序 | Date日期 | Term届·次 | Theme主题分类 | Topic → readout议题 → 读出通稿 | Compare ⇄对照（通稿/讲话） | Lecturer · affiliation讲解人 · 单位 | Chars字数 |
|---|