The Analects: a dialogue with Richard


QL: So if you take Line 2.16 of the Analects, for example ( 攻乎異端,斯害也已 ) you’ll notice the translators you compare don’t know what to do with it. Even where two translators come close to each other (Huang and Legge, Lyall and Pound) they nonetheless twist the line in significantly different directions.

  • Huang 1997: The Master said: “To apply oneself to heretical theories is harmful indeed!”
  • Lau 1979: The Master said, ‘To attack a task from the wrong end can do nothing but harm.’
  • Legge 1893: The Master said, “The study of strange doctrines is injurious indeed!”
  • Leys 1997: The Master said: “To attack a question from the wrong end – this is harmful indeed.”
  • Lyall 1909: The Master said, To fight strange doctrines does harm.
  • Pound 1951: He said: Attacking false systems merely harms you.
  • Waley 1938: The Master said, He who sets to work upon a different strand destroys the whole fabric.

How is it possible for competent translations of such a short sentence to differ so much?

Richard: Figurative language, maybe?

QL: Well, it’s more than that. It turns out there is too little semantic content in the characters to know exactly what Confucius meant. 

Richard: You mean the characters in 2.16 are too archaic to be intelligible?

QL: No, I don’t think that’s the problem here. To understand the problems these translators face, consider two interesting facts about classical Chinese:

(1) Classical Chinese was an inflected language.

  • So while modern Mandarin has additional (standardized) syllables for e.g. pluralization, possessive, and various other aspects of grammar, Old Chinese expressed these by modifications of the root-word itself (just like Latin and English).
  • These inflections aren’t reflected in the texts (not even the earliest surviving texts) because the text is only supposed to be an aid to memorization — memorization, that is, of the verbal version of the text which of course would include the inflections.
  • (And the memorized text itself was only a sort of key or outline to an oral doctrine, partially saved in the copious commentaries which exist for some of the texts, — and that “teaching” which was supposed to be “the meaning of the analects” assumed intimate familiarity with the Shujing (Classic of Documents), the Liji (Classic of Rituals), and other sprawling texts which no one has time to read anymore.)

(2) Classical Chinese had many fewer radicals than Middle Chinese.

  • Today, each Chinese character is composed of one or more (sometimes many more) simple characters, called radicals (“roots”).
  • Modern Chinese has a huge inventory of these radicals, which can be roughly divided into:
    • phonetic signs (which indicate how the word was pronounced, or at least what it rhymed with, when the writing system was put in its final form)
    • semantic signs (which hint at the meaning of the word the character refers to).
  • But this inventory was not always huge; it grew gradually in the Middle Ages.
  • Before this expansion, Classical Chinese generally didn’t make extensive use of compound characters (i.e., the kind that use a semantic radical to distinguish between two words with a similar pronunciation, or between two connotations of a single word).
  • The compound characters were mostly invented later, as the language changed (mostly in the direction of making many more words homophones) and the demands on the writing system became more complicated.

When you find early copies of these texts in archaeological digs, they often have all the key words written down ambiguously, using what is now the “phonetic root” with no additional radical to specify its meaning. Again, this created no problem because the original text was an aid to memorization, not a non-verbal means of instructing students.

The bottom line is that no one has any idea what some of these things mean. It’s a serious problem even for texts like the Dao De Jing, which are long and sustained and have some inner structure which allow you to — tortuously, hesitatingly — infer the original meaning of each line

The Analects is gone. No one will ever know what  攻乎異端,斯害也已  meant.

Richard: Yet, most translators substantially agree on probably 4/5 or more of the passages. I’m aware that parts of it are vexing for historians, linguists, and dedicated philosophers, but it seems clear that we can at lest get the gist of most of the work. With some guidance from those professionals and in light of other Confucian works, it’s still greatly valuable for the intellectually curious. I see no reason to hang caution tape around it with a “Professional Sinologists Only” sign.

QL: Well, I’m certainly no sinologist. (Though I do think context matters in trying to read an interpret old books.) Using the example I gave, Analects 2.16: do you think you know what Confucius meant by that? (Do you think it matters what Confucius meant by it?)

Richard: Again, I’m aware that some passages are obscure. Analects 2.16 is one of them. Yes, the Analects is a complex book. Some parts are difficult. Most of it is still within reach of those likely to have an interest in it in the first place (again: with guidance, and in light of other Confucian works).

QL: I asked you about 2.16 because I thought it might work as an example to illustrate a more general point. I picked that sentence, not because I think it’s the most important in the book (lol) or because you need to understand every sentence to get anything out of a philosopher, but because it clearly demonstrates the difficulties in the whole enterprise. You are willing to surrender a few obscure passages, to retreat to a more defensible position: but just as it is the contradictions that reveal the flaws in an argument, it may be the obscurities in the translations of the Analects that reveal just how indefensible the text really is.

Richard: You seem to be arguing that we should treat the exception as though it were the norm! If 4/5 of the passages I looked at were translated in substantially the same way, then the difficulties which translators face are peripheral — relevant only to the 1/5 of the Analects where the translators disagree — and curious students can study the 4/5 where the translators agree. Whatever peripheral issues historians and linguists are still debating don’t undermine this core.

QL: Are you sure? What if the passages whose translations are flatly incompatible were, like the tip of an iceberg, the only visible sign of the uncertainties and puzzles that overwhelm the scholar in every passage? What if the other passages posed similar puzzles, and you never even noticed that you had to grapple with them?

Richard: Sure, the Analects is a complex book, I acknowledged that. Even in the parts of it that are within my reach, there are bound to be places where technical historical or linguistic knowledge occasionally comes in handy.

QL: It is only when the professional translators are clearly struggling that ordinary readers think “Huh – how can you interpret one sentence in completely different ways? How can both options be reasonable?” The rest of the time, the puzzles the scholars grapple with sink out of sight.

Richard: But I don’t need to be able to understand the debates between historians and linguists to understand the Analects. I only need to understand the outcome of the experts’ debate, not how they got there. I never claimed the texts are accessible without guidance! If the sorts of puzzles you’re talking about exist throughout the Analects, there will be a commentary that can guide me around potential confusions and misunderstandings.

QL: But how much are you willing to rely on the commentary? When you reiterate your view about reading guides, it seems almost as though you thought you were staking out a moderate position between two extremes: perhaps you think I want you to master the whole scholarly literature before attempting to understand the basic texts, and have simply overlooked that there is a happy medium between becoming an expert and ignoring the experts.

Far from it!  In general, for classical works I say “Skip the introduction and the commentary, and dive straight into the text”. But if it is uncertain what the text said, you are reading an original philosophical work written by the translator, loosely inspired by the words of Confucius.

The commentary on the translation will not help you any more than the translation will, if the question is whether the translation actually conveys what Confucius said.

Richard: But if a bad translation doesn’t convey what Confucius said, better scholars will criticize it and make better translations. Remember, I’m not a sinologist, and neither are you. Do you think you can come up with a better translation? If all of the commentaries agree on many uncontroversial points, if all the scholars assign the same translations to their students, then I don’t see how you can presume to question whether it conveys Confucian ideas.

QL: Alas, I can see how what I have said might sound very presumptuous. But I hope you will bear with me, and perhaps by and by I will convince you that I do not imagine that I understand the Analects better than the translators do.

Now, if all the commentaries on the Analects agree on many points, that makes these points uncontroversial, certainly, but does it make the consensus a good one? I doubt it. I know you’ll call this presumption. But if the textual basis for a doctrine is flimsy and the consensus about the nature of the doctrine is strong, that only proves that the consensus comes from institutional and ideological pressures, rather than from the evidence.

Where there is voluminous and robust evidence, good research converges on the truth; where there is only scanty, brittle evidence, good research leads to the multiplication of hypotheses, as scholars discover each additional possible perspective on their inadequate data.

This doesn’t matter if you are studying the consensus itself rather than the evidence it is built on. When the Analects became important to the medieval Chinese literati, they built a tradition of commentary around it. If you care mostly about views of the literati under the Ming and Qing dynasties, none of my objections carry any weight: you can always fall back on the Neoconfucian interpretations as canonical. Understanding the thought of Confucius is difficult; but understanding how the Neo-Confucian interpretation of the Analects relates to other Neo-Confucian works is easy.

Richard: So what exactly is your position here, that no one should read the Analects? That we should read them only if we’re proficient in Chinese and other Confucian works? That we should only read what we can understand 100%, 90%, 80%?

QL: I would say that I would recommend the Analects primarily to people who are really, really, really interested in Confucianism; and ideally who are especially interested in Neo-Confucianism.

Richard: You mean because none of your objections are valid if I am willing to treat the Neo-Confucian readings as authoritative?

QL: Basically, but it’s not an ironic recommendation. If you care a lot about the ideas of Zhu Xi and Wang Yangming, that is a genuinely good reason to read the Analects. It’s not only that you can substitute “trying to undertand Zhu Xi and Wang Yangming” for “trying to understand Confucius.” The Analects is important to their new conception of Confucianism, and part of their new conception of Confucianism was that the Analects needed to be treated as a very important book. You know that the Four Classics were the original textual basis of Confucian thought during its first millennium, right?

Richard: I know what you’re talking about. I’ve written about the Book of Odes.

QL: Those works were the were the original focus of the life of study that Confucius recommends. Rituals, Songs, Documents, Changes. Confucius was apparently very concerned to understand these texts, and to articulate why they were integral to ancient Chinese society. Many early Confucian works are commentary on these Four Classics, and the others are chock-full of allusions to and examples drawn from them. They were the focus of the original exam system for entering the civil service; the Han Dynasty enshrined this canon at the center of intellectual and political life.

Richard: Right.

QL: Pretty basic. But what most people don’t realize, perhaps even people who are interested in Confucianism, is that the four texts that were most central to Confucianism when the Portuguese and the Spanish reached China — the Analects, the Daxue, the Doctrine of the Mean, and the Mencius — were originally quite obscure. The Doctrine of the Mean is an extract from the one of the Four Classics; the Daxue, an extract plus a commentary on it; the Analect is of course a collection of saying of unclear origins, and the Mencius is the eponymous work of a disciple of Confucius. As Chinese culture degenerated, the Four Classics were replaced by these four mini-books which had originally been intended primarily as commentaries on the Classics. Do you see how this totally changed the nature, not just of “Confucianism” as an ideology, but the entire educational system and government erected on top of it?

Richard: I can see how it might be substantially different, at any rate.

QL: If you are primarily concerned to understand the Neo-Confucian philosophy that arose after China’s academic system switched to the Four Books, then naturally the Analects (one fourth of the new canon) is extremely important, and the translation problem disappears. In that case you’re really trying to understand the Neo-Confucian commentaries, and the Analects is foundational because it is a common point of references for the commentators. But understanding what the Analects means outside of the preconceptions and premises the Neo-Confucians bring to the text becomes unnecessary, and a consensus among the commentators really is adequate to resolve any textual difficulties (about what they thought about the Analects).

Richard: But you say that the Analects might be useful to someone who doesn’t care about Neo-Confucianism, if he is very interested in Confucius?

QL: Yes, very interested: and in particular, interested enough to invest a lot of time in other texts first. If you were truly eager to understand Confucius, then once you had already read a lot of Chinese philosophy you would eventually get to the point where reading the Analects (even if it’s fairly obscure) would provide more food for thought, and deepen your understanding more, than re-reading some other, clearer work.

This isn’t limited to Confucius. Once you’ve read straightforward texts which contain a position articulated in clear propositions, you can enrich your understanding of them by reading texts which contain no clear philosophical propositions at all. Histories contain nuggets of contextual information that flesh out an author’s references to contemporary life, for example. Poetry and drama will expose you to metaphors, proverbs, and rhetorical tropes an author has used (or abused). Likewise, poorly written, fragmentary, or second-hand accounts of a philosophical doctrine may, despite their drawbacks, give valuable information about some aspect of the doctrine. These sources would read like gobbledygook to someone unfamiliar with the actual structure of the doctrine they refer to, and studying them would be a painful and unprofitable use of his time; but once the framework is in place, many details that are absent in the original are unnecessary to figure out what details of the doctrine are being addressed.

Richard: But you think the Analects couldn’t be useful to someone who hadn’t already done substantial background reading in Confucian thought?

QL: Eh, never say never. Conceivably someone who was absolutely committed to a serious study of Confucianism might want to read the Analects early on, expecting his investment to pay off gradually as ideas he encountered in the Analects helped him get into all of the books he plans to read over the coming months and years. Or if after your initial introduction to Confucius you developed a special interest in a certain Confucian concept, you might consult a pertinent passage in the Analects, as a counterpoint to what you’ve read elsewhere, even if you wouldn’t profit from reading the whole book.

Richard: But as a general rule…?

QL: Yeah, as a general rule it’s not useful. Barring special cases. It shouldn’t be treated as though it were the gateway to Confucian thought.

Richard: And it’s not useful because you don’t think there is any way to know exactly what Confucius meant?

QL: Correct.

Richard: I don’t know exactly what Confucius meant. I don’t know exactly what every sentence of Plato’s dialogues mean, I don’t know exactly what every verse of the Bible means, I don’t know exactly what every line of The Divine Comedy means, I don’t know exactly what every line of Shakespeare means, et cetera. That’s why I re-read them, consult commentaries, and so on.

QL: I don’t want to compare inspired text to Confucius. Is that okay? Biblical exegesis is a separate issue, so many interpretative problems in the Bible aren’t analogous… and if I recall correctly, you’re a papist, so we’d get horribly side-tracked.

Richard: Regarding the Bible, fine. But that’s true of only one example I gave, not Plato, Dante, Shakespeare, or a number of other difficult authors.

QL: Agreed: so let’s consider the other authors you mentioned.

First, read any of these classics and you’ll find that the text is a complete work, an organic whole. Even a Platonic dialogue, written at the very dawn of Western philosophy, is not a compressed formula, or a litany used to memorize a broader doctrine. The text expresses a doctrine, in complete sentences. Heck, Plato’s dialogues express multiple doctrines, from different personae… each of which are scrutinized closely within the dialogue itself!

Second, even Plato is already writing in complete sentences with phonetically explicit words. You can look at the Greek text of Plato and read off the exact same dialogue that Plato’s students would have heard. (This is also true of the Septuagint, for example, while it is not true of the Hebrew text of the Bible, whose vowelless words posed no problem during the flourishing of the hereditary priesthood of Judaea, but which became ambiguous after its eclipse, creating the need for the vowel-markings with which the medieval Masoretes recorded their interpretation of the text.)

So there are very real problems faced not only by modern translators of ancient Chinese literature, but even by medieval Chinese editors who compiled the standard editions of the Analects and the other Confucian texts, which simply don’t exist for the Western classics.

Richard: And yet despite this, academic experts apparently do believe that a layman can get something from the Analects. I’ve benefited greatly from the Analects, as have many others.

QL: Let’s consider the consequences, looking specifically at Analects 2.16.

“The study of strange doctrines is injurious indeed!”


We have three elements:

(study) — (strange doctrines) — (injurious),

plus a particle (I think is the emphatic “indeed!” here, but maybe some of the translators give it a semantic meaning); but each of these three elements loosely indicates a whole field of terms. [ed. – Scroll back up to the beginning of the post to remind yourself how broad a field each of the three elements covers.]

  • Take the first element, the “study” of some work. Is Confucius talking about reading X? Reading X diligently? Discussing X? Taking X seriously? Giving X-adherents the benefit of the doubt? These are very very different possibilities within the semantic field of ”study”.
  • And then, “strange doctrines”: what is strange about them? are they wrong? new? peculiar? unnatural? inconsistent with orthodox doctrines of some sort?
  • is he talking about one particular doctrine (“that strange doctrine”), or one school of doctrines (“the Strange Doctrine”)? … or is he talking about a subset of all doctrines ( i.e. the ones that match the predicate “strange”)?

I won’t be pedantic and list the different possibilities for “injurious”, I’m sure you get the point. Take these possibilities, permute them, and you see the dizzying ambiguity of the sentence…

Richard: Your point about the structure and resulting additional ambiguity is important. And I have no way of forming an opinion on this one way or another. I can only defer to experts and my own observations about how the translations work. I do value your opinion, but it can only be one among several others.

QL: There is one question about hard translations, and which translator’s version to trust. This is the question which is valid for Plato, for Dante, and all the rest. There is another question about assessing the textual basis of a translation (including the very possibility of any translation at all!)

For example, you’ve probably heard of the Indus Valley Script — i.e., bunch of symbols found on pottery at Harappan archaeological sites.

Richard: Sure, I know a tiny bit about it.

QL: Would it make sense for me to propose my own translation of any of the ceramics shards found in the Indus Valley? lol, no.

Can I assess the quality of the available “texts” and assess whether there is enough available to make any translation at all? —— indeed, whether there is any evidence that the symbols are a language or a script at all?

That I think I can do. I’m not claiming to be 100% certain on the Indus Valley Script, but I think non-experts can read the evidence and, on the basis of existing evidence and scholarly analysis of it, come to their own conclusion. 

And that conclusion —in this case, that the Indus Valley Script is probably just artistic motifs or seal-marks and definitely cannot be shown to record a language — tells you a lot about what you can learn from any of the “rival translations” of the evidence. If the Indus Valley Script isn’t even a language, for example, then any purported translation of the script is just bogus.

Or consider those scripts which definitely record a language but which no one knows how to read, like the Khitan scripts for certain Central Asian languages. Do you know the ones I mean?

Richard: I don’t think so. Khitan, as in the Khitan dynasty?

QL: Yes. They spoke a language related to Manchu, and when they were politically powerful in China they devised script based on the Chinese system to record their own language. But it quickly dropped out of use. Eventually the whole Khitan people disappeared, I believe.

Richard: Ah. So we have the texts, but we don’t have the language or any explanation of its script?

QL: Exactly. Maybe they’ve cracked it since I last read about it, but you can follow the logic: if an expert hazards a translation of one of those texts, his guess at what it means is certainly as good as anyone else’s… but it would be pointless to read it for insight into the literature, politics, or history of the Khitan people before there is any consensus on how their language was even written. At that point the “translation” needs to be interpreted as a linguistic hypothesis about the Khitan script, not an interpretative hypothesis about the meaning of a Khitan literary text.

So my conclusion is that even an amateur can form an opinion about the type of translation that is possible on the basis of existing evidence

Richard: I’ll have to look into this more, then; I’m open to any recommendations if there are any particular books or articles to look up.

QL: I think what would help most might be if you were to read about the way the Dunhuang scrolls have changed the interpretation of the Dao De Jing. There is a very accessible discussion of this in Moeller’s 2011 translation of Laozi. Would that interest you? Even if you find the DDJ boring, what you read about the DDJ will give you a much clearer picture of the history of the Analects (how it was preserved, what function it served, what difficulties arose later on from ambiguities and variants in the text).

Richard: It’s been a long time since I’ve read the DDJ, but I’d be interested, yes. I’ll trust your judgement that it’s relevant.  But for now I’m content to defer to later Confucian scholars, while noting that several points are necessarily controversial. I am slowly working my way through the rest of the canon, and from there to later Confucians, so my opinion may change down the road.

QL: As a general rule, it makes sense to trust authorities and trust traditions. But I think people who are professionally “Confucius translators” generally have distorting motives. Anyway, that’s a secondary issue and I won’t go into it right now if you think that’s special pleading.

Richard: It probably is special pleading, but I’ll take the bait: what motive are you talking about?

QL: If you’ve sunk half of your life into learning classical Chinese (which is not only it’s own language, nearly as different from modern Chinese as English is from proto-Indo-European, but has its own complex bibliographic and literary conventions), you are professionally capable of teaching and interpreting ancient Chinese philosophy and literature, and not much else. Are you really going to devote a lot of your time to arguing that we can’t really understand these texts? Or that it would be a waste of time for most amateurs to read the translation or commentary you spent a decade on?

Historians of China will often make asides about how different the paleo-Confucian interpretation of the Four Books is from the Neo-Confucian interpretation. One oft-noted controversy is the one over the opening of the Daxue: is the path of virtue to love the people (qin min) or to renew the people (xin min)? There was an entire dynasty (albeit short-lived) whose name is drawn from a word which current Confucian orthodoxy claims cannot be found in the Daxue; the textual disagreement would be extremely difficult to ignore entirely. But you’ll rarely see historians of Confucianism go one step further and point out that the reason for this confusion is almost certainly that the original text read (—) and there is no way to ever know whether students were taught to read it as xin or as qin. Nor will they point out that this is not an isolated anomaly, but a standard feature of how the Old Chinese texts were written and used.

Every translator and commentator has professional motives not to draw out these implications, regardless of his personal philosophical position. But in fact, most of the experts are themselves Neo-Confucian, became interested in Confucius via the Neo-Confucian doctrine, and have in many cases studied the texts at (Neo-)Confucian academies in China, which is a quasi-religious process akin to enrolling in a madrasa or entering a Buddhist monastery. Their understanding of some aspects of the Confucian canon thus becomes automatic, reflexive.


Postscript: A few days later we discussed the Presocratics, which brought Richard into closer sympathy with my point of view. The framework of my argument here, we agreed, can be easily extended to cover the difficulty of interpreting or translating the Presocratic philosophers (whose writings survive only in fragments).

Our conversation about the Analects raised a few interesting questions which we never came back to answer:

a., whether it is ideal that a beginner should have to rely on cross-references to other texts and the guidance of professionals (my bias is: people learn more when they ride bareback, without a commentary or key as intermediary),

b., to what extent context is central and necessary to the meaning of a text,

c., whether it is important to attempt to reconstruct the original meaning and doctrine of the author in the first place.

Also: d., differences between Biblical and secular exegesis, e. legitimate questions about the obscurity of classical texts and unknowability of their authors’ position, f. value of reading with and without commentary.

Maybe some day.

One thought on “The Analects: a dialogue with Richard

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s