The Analects is the book people mean when they say “Confucius said”. It is also a book Confucius did not write, did not compile, and never saw. The question of who compiled the Analects has an answer of a kind in the oldest library catalogue in China, a sharper answer in a clue that a Song dynasty philosopher noticed in the text itself, and a more interesting answer in a 2019 study from National Chung Cheng University, which asks not only who did the compiling but what kind of Confucius they made. This entry takes those three in turn, because the third depends on the first two.
The Han catalogue’s account
The earliest statement about how the book came to exist is in the bibliographical treatise of the 《漢書》, compiled by Ban Gu in the first century AD from the imperial library’s catalogue:
The Han library on the Analects
論語者，孔子應答弟子時人及弟子相與言而接聞於夫子之語也。當時弟子各有所記。夫子既卒，門人相與輯而論篹，故謂之論語。
“The Analects are the words with which Confucius answered his disciples and his contemporaries, together with what the disciples said among themselves and heard from the Master. At the time, each disciple kept his own notes. After the Master died, his followers gathered these together, discussed and compiled them; hence the name ‘Discussed Sayings’.”
班固《漢書・藝文志》 · ctext.org
Three things follow from this paragraph, and each is routinely lost in English summaries. The first is that the book is posthumous — 夫子既卒, “after the Master died”. The second is that it is a compilation of separate notes by different people, 各有所記, “each had his own record”, which the followers then 輯, gathered, and 論篹, sifted and arranged. The verb in the title is an editorial verb. The third is that the paragraph does not stop there. It goes on to say that under the Han there were three transmissions of the text — the Qi version, the Lu version, and a version in old script — with different chapter counts, and that the last editor of the Lu line, the tutor and chancellor Zhang Yu, produced the text that 最後而行於世, “came last and circulated in the world”. The Analects we read descends from Zhang Yu’s edition of the late first century BC. Between the disciples’ notebooks and our text stand at least two rounds of editors, four centuries apart.
The clue in the honorifics
The Han catalogue says “the followers” and leaves it there. A more specific answer came from noticing something about the book’s manners. Most disciples in the Analects are named by their given names or courtesy names: Zilu, Zigong, Yan Hui. Two are not. You Ruo and Zeng Shen are called 有子 and 曾子 — “Master You”, “Master Zeng” — the honorific otherwise reserved for Confucius himself. The second and fourth passages of the whole book are theirs. The Song philosopher Cheng Yi drew the inference, and Zhu Xi put it at the head of his edition:
Cheng Yi’s inference
程子曰：「論語之書，成於有子曾子之門人，故其書獨二子以子稱。」
“Master Cheng said: ‘The book of the Analects was completed by the followers of Master You and Master Zeng; that is why, in the book, these two alone are styled Master.’”
朱熹《四書章句集注・論語序說》 · ctext.org
You call your own teacher “Master”. If the people writing the book up called You Ruo and Zeng Shen “Master”, they were the students of those two, which puts the compilation a generation or more after Confucius’s death, in the hands of one particular lineage among several. The Tang essayist Liu Zongyuan had made a similar argument, from Zeng Shen’s dates, centuries before Cheng Yi; the Chung Cheng study I come to below cites both. It is not proof — it is an inference from a pattern of usage — but it is the kind of evidence the text itself gives, and it is why the careful answer to “who compiled the Analects” is not “his disciples” but “the school of two of his disciples, probably, edited again under the Han”.
What the compilers built
This is where the question stops being antiquarian. In 2019 李貴生 Lee Kwai-sang published a forty-page study in the 《中正漢學研究》, National Chung Cheng University’s sinology journal, under the title “The influence of the compilation of the Analects on the formation and interpretation of Confucius’s teaching”. His starting point is a warning:
《論語》這種高度的「透明性」很容易令人忽略了它自身的文本性質，「《論語》中的孔子」與「歷史上的孔子」幾乎可以劃上等號。
“The high ‘transparency’ of the Analects makes it easy to overlook its own nature as a text: ‘the Confucius in the Analects’ and ‘the historical Confucius’ come to be treated as almost equivalent.”
Lee declines to rule on the authorship question — 本文無意裁斷過去哪種說法更為可信, “this paper has no intention of adjudicating which of the past theories is more credible” — and asks a different one. Given that editors selected, trimmed and arranged, what does their Confucius look like, compared with the Confucius of other early texts that record him? He compares passages in the Analects with parallel material in transmitted and excavated sources and finds three consistent tendencies.
A worldly scholar, not a sage. The Analects goes out of its way to present Confucius as a learner: someone who “loves learning”, who practises “at due times”, who treats study as the precondition of teaching. Where other texts sanctify him, the compilers kept the passages in which he is ordinary and diligent. Lee reads this as a choice, and one that contrasts sharply with the treatment of the same man elsewhere.
Inner virtue at the axis. The book is organised, Lee argues, around 仁 as an inward quality that is prior to outward ritual — 以內在德性比外在的禮文來得重要，前者是後者的根本, “inner virtue is more important than outward ritual form; the former is the root of the latter” — and around a model of government that starts from the ruler’s own cultivation and radiates outward. He notes a quiet difference here between the Analects and the Kongzi Jiayu, whose Confucius has rather more to say about punishment and administration. Which of those is “the historical Confucius” is precisely the question the transparency of the Analects hides.
Concrete and practical. The compilers favoured Confucius’s remarks in informal settings, about specific situations, and avoided abstract discussion of Heaven’s way and human nature — the very topics his followers would fight over for two thousand years. That may reflect the man. It certainly reflects an editorial taste.
What the format does to the words
The second half of Lee’s study is about the form of the book rather than its selection: the fact that it is a set of short, detached sayings. He identifies three effects, and each is something a reader of any English translation runs into without knowing why.
The first is universalisation. A remark made about one person on one occasion, once cut loose from its occasion, reads as a general truth: 脫離原來語境最明顯的效果是書中的內容不再受到具體場景所制約, “the most obvious effect of detachment from the original context is that the book’s contents are no longer constrained by the concrete scene”. Lee’s examples include the line about loving virtue as one loves beauty and the notorious remark about “women and small men” being hard to deal with — sayings that may have had a target and now have a species. The second is ambiguity: compressed classical sentences such as 先有司 or the description of Zilu’s courage exceeding the Master’s admit several construals, and the commentarial tradition is largely the history of choosing among them. The third is sanctification: what the compilers recorded as 善言, “good sayings” — sensible remarks worth keeping — later readers came to treat as 微言, “subtle words” concealing a doctrine. The aphoristic format invited that, and the format was the editors’ choice.
Why this matters for a reader in English
None of this makes the Analects unreliable. It makes it a text — a selection, shaped by a particular school, edited again under the Han, and arranged in a form that detaches sayings from situations. The practical consequences are modest and worth keeping in mind. When a translation renders a saying as a universal maxim, remember that the universality may be an artefact of the cutting. When two translators disagree about a sentence, it may be because the sentence, cut short, genuinely permits both readings. And when someone tells you what “Confucius believed” on a topic the Analects avoids, they are reading the editors’ silence as the man’s.
For a comparison of seven English versions and what each does with these problems, see which Analects translation should you read? The mirror-image question for the Daoist classic — how far back a text can be pushed and how little that tells you about its author — is in did Laozi live before Confucius? And the book Lee sets beside the Analects, the Kongzi Jiayu, has its own long argument about whether it can be trusted at all; that is in can the Kongzi Jiayu be trusted?
For the other Four Books text with an editorial history English readers rarely meet, see the Great Learning’s three principles and its eight steps.
FAQ
Who wrote the Analects?
Nobody in the sense of an author. According to the Han catalogue, disciples kept separate notes of Confucius’s sayings and, after his death, his followers gathered and compiled them. The received text descends from the edition of Zhang Yu in the late first century BC.
Which disciples compiled the Analects?
The text does not say. The Song philosopher Cheng Yi inferred from the fact that only You Ruo and Zeng Shen are styled “Master” that the book was completed by their followers — that is, by the second generation after Confucius. It is an inference from usage, widely accepted, not a record.
Is the Analects an accurate record of Confucius?
It is a selective one. Lee Kwai-sang’s 2019 study finds that the compilers consistently presented a worldly, learning-centred Confucius, organised the material around inner virtue rather than ritual or law, and kept concrete remarks rather than abstract ones; other early texts present him differently. The book records Confucius; it also constructs him.
Why do translations of the same Analects passage differ so much?
Partly because the compilers stripped sayings of their occasions and compressed them into a few characters. Lee shows that this produces both universalisation (a remark about one person becomes a rule) and ambiguity (several grammatical readings survive). Translators are choosing among readings the format left open.
The habit of quoting the Analects half a sentence at a time, and trusting the reader to restore the rest, outlived the Tang. The best-known instance is the ending of the Loushi Ming, which stops at 何陋之有 and leaves the conditional clause that carries the argument unspoken.
Sources
- 班固《漢書・藝文志》：「論語者，孔子應答弟子時人及弟子相與言而接聞於夫子之語也。當時弟子各有所記。夫子既卒，門人相與輯而論篹，故謂之論語。漢興，有齊、魯之說……張氏最後而行於世。」 — ctext.org (accessed 5 Sep 2026)
- 朱熹《四書章句集注・論語序說》引程子：「論語之書，成於有子曾子之門人，故其書獨二子以子稱。」 — ctext.org (accessed 5 Sep 2026)
- 李貴生〈《論語》編纂對孔子學說的形成及詮釋的影響〉，《中正漢學研究》2019 年第二期（總第 34 期），國立中正大學中國文學系，2019-12，頁 1–40 — PDF：「透明性」（頁 2）；「本文無意裁斷過去哪種說法更為可信」（頁 4）；世俗化學者形象（頁 8）；「以內在德性比外在的禮文來得重要」（頁 9）；「脫離原來語境最明顯的效果……」（頁 20）；善言與微言（頁 29–30）；引柳宗元〈論語辨〉與程子說 (accessed 5 Sep 2026)
- 《論語・學而》1.2「有子曰」、1.4「曾子曰」 — ctext.org (accessed 5 Sep 2026)
