Repository path: workshop/translations/yanfu-yili-yan/R04-v1/translation.md · rendered 2026-09-09

Yan Fu, Notes on the Method of This Translation (1898) — complete

Contamination: high. 信達雅 is the most cited formula in the Chinese translation tradition and is glossed in English textbooks with near-total uniformity as faithfulness, expressiveness, elegance. That gloss was in the lead's priors before this session began; the lead's task here included deciding whether to accept it, which it could not do naively. No published English rendering of the 譯例言 was fetched or opened, and the lead does not have a specific remembered rendering of the document as a whole — but the three key terms are contaminated at the level of a stock phrase.

Notes on the Method of This Translation

1. In translation there are three hard things: 信 xìn, fidelity; 達 dá, getting the sense through; 雅 yǎ, refinement. To seek fidelity is already very hard. Yet if you have fidelity and the sense does not get through, then though you have translated it is as though you had not — so getting through is what is to be prized. Since the seas were opened, men with the interpreter's talent have been common enough everywhere; but take up any book at random and hold it to those two standards, and few will answer. The reasons are three: shallow tasting, first; lopsided attainment, second; and few who can tell the difference, third.

Now what this book says is grounded in learning newly won by Westerners over the past fifty years, and it is besides a late work of its author. My rendering aims at making the deep sense plain, and so between word and phrase there is at times inversion and amplification. It does not fuss over matching character to character and clause to clause in order — but the sense does not depart from the original text. I have therefore headed it conveying the purport rather than rendering with the brush: it takes the liberty of expansion, and is truly not the orthodox method. Master Kumārajīva said: "Those who learn from me will be the worse for it." Many will come after me; I hope they will not take this book as their warrant.

2. In Western sentences the nouns and names are for the most part explained as soon as they are raised, like a side-branch in Chinese, and only afterwards does the thread reach far back to what went before, filling out the sense and closing the sentence. So a Western sentence may run to two or three words at the least, and to scores or hundreds at the most. Were one to translate in imitation of this, the result would surely be unintelligible; but to cut it down and take a short road risks leaking the sense. Here it falls to the translator to take the animating logic of the whole text and fuse it in his own mind; then, when he sets the brush down and lets the words out, the parts will naturally furnish one another. And where the original's diction and reasoning are deep in themselves and hard to share, he should draw in support before and behind, so as to bring the meaning out. All this contriving is done for the sake of getting the sense through — and getting the sense through is precisely the way to be faithful.

3. The Yijing says: "In cultivating your words, establish sincerity." The Master says: "Words get through, and that is all." And again: "Words without patterning do not travel far." These three sayings are the true course of composition, and they are likewise the model for translation. So beyond fidelity and getting through one seeks 爾雅 ěr-yǎ — refinement in the manner of the classics — and this is not merely in the hope of travelling far.

The truth is that where the principles are fine and the words subtle, using the word-usage and sentence-usage of the age before the Han makes getting through easy; using the handy vulgar writing of recent times makes getting through hard. One is forever bending the sense to fit the words, and an error of a hair's breadth becomes a thousand li. In weighing between these two there was something I truly could not help — how should it be angling after strangeness! This translation of mine has drawn no little mockery for papering over poverty with difficulty; in fact I took pains to seek plainness, and this is simply what that came to. Moreover, the arguments of the original book rest largely on logic and mathematics, on the investigation of things, and on all the learning of the calendar-masters. If a reader has never made his way to these, then even a countryman of the author's, sharing his tongue, will still fail to understand much — how much more when it reaches him at a second remove!

4. New principles come treading on one another's heels and the terms multiply. Search for them in Chinese and there is nothing at all to be found; and where something is forced into correspondence it jars in the end. Meeting this, the translator has nothing for it but to hold his own scales and fix the name by the sense.

But there are cases of the greatest difficulty. Take the dozen or so pieces headed 導言 in the first volume of this book: because the main argument runs deep, a plainer exposition is spread out first. I began by rendering it 卮言, "goblet words"; but Xia Suiqing — Zeng You, of Qiantang — objected that this was slack and bad, and said that the Buddhist canon already had such a thing and it might be called 懸談, "suspended discourse". When my elder Wu Zhifu — Rulun, of Tongcheng — saw it, he said further that 卮言 had already become a worn-out word and that 懸談 followed the Buddhists, and that neither is the act of a man able to stand on his own feet; better to follow the old practice of the philosophers and give each piece its own heading. Suiqing said again that if so, each piece becomes a composition in itself, and the original book's sense of establishing a single root is somewhat obscured; and that as for the names 懸談 and 懸疏, 懸 means 玄, dark and deep, and denotes words that gather up the essential purport — which does not fit here and must certainly not be used. So in the end I followed the original heading, translating it plainly as 導言, "leading words", and noted Wu's separate titles beneath each piece for the reader's convenience. This shows how hard it is to fix a name: though one would avoid the jeer of swallowing whole and skinning alive, there are times when it cannot be avoided.

Other names — 物競, the striving of creatures; 天擇, heaven's selecting; 儲能, stored power; 效實, power made actual — all began with me. To settle a single term has cost me ten days, or a month, of pacing. Blame me or know me: I leave that with the discerning.

5. The original book discusses at length the schools of thought since the Greeks, and everyone it holds up was a famous scholar of his day. On their lingering influence and the remnants of their doctrines hang the mind and the intelligence of the West these two thousand years, and no one who expounds Western learning can afford not to know them. I have therefore set down at the end of each piece a brief account of these gentlemen's lives and works, to furnish the student, roughly, with the means of knowing the man and judging his age.

6. Fathoming principles is like conducting government: both prize the pooling of thoughts and the widening of benefit. Wherever what the original argues agrees or differs with other books, I have set down in a note after the piece what my own poor learning knows, as material for comparison. Here and there I have added my own view as well, taking the sense of the Odes — the bird crying out and seeking the answering call of its friend — and of the Yijing: joined marshes, the noble man learning in company. Whether these views are right or wrong I leave to public discussion; I do not presume to hold fast to them. Were it said that I had set up a standard and hoisted myself upon it, that would betray the intention with which I carried the lead and gripped the tablets and toiled at this rendering.

7. As for the translating of this volume: Western books of natural philosophy are not easily turned about, and I took this one up deliberately, going through it day by day with my fellow students. When it was done, my elder Wu Zhifu saw it and was pleased with it; he took the axe to it and drew the lines straight, and his corrections and improvements were truly many. Yet learning that probes the abstruse and knocks at what is silent is not the most pressing business of the hour, and I did not want to put it before the world. But the manuscript was borrowed and copied out by Liang Rengong of Xinhui, Lu Muzhai of Mianyang and others, all of whom urged me to send it to the blocks soon; Muzhai posted it to his younger brother Shenzhi in Hubei, who also said it ought to be made public within the seas — and so calamity came upon the jujube and the pear, which was still not my own wish. When the cutting was done it was sent to Tianjin for re-collation, and I have accordingly issued these notes on procedure and set down its origins thus.

The 22nd day of the 4th month, the 24th year of Guangxu, the year wuxu [1898]. Recorded by Yan Fu at the Zunyi Academy, Tianjin.

Translator's log

Written at translation time, 2026-07-25, and frozen before any evaluation was designed and before any source page or synthesis was written. Decisions and alternatives, not quality claims.

The three terms — the decision the document exists to force

1. 信達雅 was rendered "fidelity / getting the sense through / refinement", against the stock English "faithfulness, expressiveness, elegance". Each departure was made for a reason in the text, and each is reversible; I set them out so the choice is auditable rather than asserted. - 達 is the one I am most confident the stock gloss gets wrong. Dá is arrival — getting there. Yan Fu grounds it on Analects 15.41, 辭達而已矣, where the point is that words need only reach, and he says in item 2 that all his contriving is 為達, for the sake of getting through, "and getting the sense through is precisely the way to be faithful." That is a claim about transmission succeeding, not about the style of the prose. "Expressiveness" names a property of the writing; dá names an event in the reader. I chose a four-word verbal phrase over a noun and accepted that it reads clumsily beside two nouns, because a noun would have re-abstracted exactly what the verb is doing. - 雅 I rendered "refinement" and could not do better. "Elegance" was live and is not wrong; I rejected it because in English it sits on the ornamental side of the style/substance line, and item 3 puts yǎ on the other side — it is what makes 達 easier for subtle matter. Yan Fu glosses it himself as 爾雅, which is both "correct and refined" and the name of the ancient glossary, so the word carries "in the diction of the classics" as part of its sense. I kept 爾雅 in characters with a gloss rather than collapsing it into the English, because that double reference is the whole of his justification. - 信 → "fidelity" is the least controversial. I kept "faithful"/"fidelity" rather than "trustworthiness" although xìn is the trust word, because the sentence "getting through is precisely the way to be faithful" needs the English adjective to sit naturally.
2. Scaffolded borrowing, deliberately. Each of the three appears once as character + pinyin + English and thereafter in English only. This is the handling the project's own CJ-20260724-scaffolded-borrowing names (retain the item, manufacture its sense through context) rather than gloss or copy-opaque, and I used it knowingly on a text about how to fix names. Same handling for 爾雅, 導言, 卮言, 懸談, and the four coinages 物競 / 天擇 / 儲能 / 效實.
3. The four coinages were translated literally, not identified. 物競 is what English calls the struggle for existence; 天擇 is natural selection; 儲能 and 效實 are potential and kinetic energy. I rendered them "the striving of creatures", "heaven's selecting", "stored power", "power made actual" — which is what the characters say and what Yan Fu's reader saw — and did not supply the English scientific terms. Yan Fu's paragraph is about the cost of coining a name in a language that lacks the concept; putting the settled English term beside it would hand the reader the answer whose absence is the subject. The cost is real and I record it: a reader who does not already know will not learn from my English that 天擇 is natural selection.

Other decisions

4. 象寄之才 (item 1) → "men with the interpreter's talent". 象 and 寄 are two of the four offices of the Liji 王制 for interpreting to the four quarters. Rendering it "men of the xiang and ji talent" would have been opaque; a footnote would have been out of register in a document that footnotes nothing. Flattened, and the flattening noted.
5. 不佞 — Yan Fu's self-deprecating first person, roughly "your untalented servant" — is rendered as plain "I" throughout. Classical Chinese first-person self-abasement has no English carrier that is not either archaic-comic ("your unworthy servant") or invisible. I chose invisible, which loses a consistent register marker present at every occurrence. This is the same shape of loss A-garnett-vanka records for Russian T/V deixis and A-shaw-spider-thread for 〜でございます: a grammatical or formulaic politeness system with no target counterpart, dropped without trace.
6. The two canonical tags in item 3 — 修辭立誠 (Yijing, Qian, Wenyan) and 辭達而已 (Analects 15.41) — were translated, not identified by chapter. Yan Fu names only the books, and adding chapter references would tidy his argument into a scholarly apparatus he did not write.
7. 災棗梨 (item 7) → "calamity came upon the jujube and the pear". Printing blocks were cut from jujube and pear wood, and the phrase is the standard modest idiom for going to press. Live alternatives: substitute the sense ("and so it went to press"), or calque with a gloss. I calqued bare — transparent calque in the project's strategy list — because the joke survives the calque for a reader willing to think for a second, and because the modesty is doing work in a sentence that says publication was against his wishes.
8. 懷鉛握槧 (item 6) → "carried the lead and gripped the tablets", same handling, same reasoning.
9. 我罪我知 (item 4) → "Blame me or know me: I leave that with the discerning." This inverts the Mencian source, where Confucius says of the Spring and Autumn Annals that those who know him and those who condemn him will do so by it. Yan Fu compresses it to four characters. I compressed in English rather than expanding to the allusion, and lost the allusion.
10. 斧落徽引 (item 7) → "took the axe to it and drew the lines straight". I am not certain of this phrase; 斧 is the corrector's axe and the second pair reads as carpenter's marking-cord imagery, so I translated the image and kept both halves. Recorded as a place where I translated a construction I could not fully parse — the sense of the sentence (Wu Rulun revised it heavily and improved it) is not in doubt, but the figure may be.
11. Numbering. The seven items each open with 一、 — "one," repeated, the classical way of listing without ordinals. Rendering them all "1." would be unreadable; I numbered 1–7 and record that the numbering is mine.
12. Register. I pitched the English formal, slightly archaic, and paratactic — "so", "and", "yet" carrying joints where classical Chinese uses no connective at all. The live alternative was modern academic English, which would have made the argument easier to follow and would have made a text arguing that archaic diction is more capable read as though written by someone who did not believe it. The choice is legible as a choice and I do not claim it is the right one. internal-judgment-only.

Where the source resisted

13. Item 3's central sentence is the hardest thing in the document and the shortest. 用漢以前字法、句法，則為達易；用近世利俗文字，則求達難 — "using pre-Han word-usage and sentence-usage makes getting through easy; using the handy vulgar writing of recent times makes getting through hard." Everything is in 利俗: 利 is convenient, handy, serviceable; 俗 is common, vulgar, of the age. The compound is dismissive but its first half concedes the very advantage he is refusing. English has no word that is simultaneously "convenient" and "beneath us". "Handy vulgar" is a two-word patch on a two-character word.
14. 艱深文陋 (item 3) is a compressed allusion — the charge, from Su Shi, of 以艱深文其淺陋, using difficulty to cover over one's shallowness, where 文 is the verb "to give a patterned surface to". "Papering over poverty with difficulty" keeps the verb and loses the pun that 文 is also the word for literary patterning that appeared two sentences earlier in 言之無文.
15. Classical Chinese has no tense and few pronouns; English demands both at every clause. I supplied "I", "he", "one" and a past tense throughout. This is the C1 shape the project already records — an obligatory target category forcing explicitation — and I record that it fired here on nearly every sentence rather than at scattered sites.

Second-reader verification — 2026-07-25 (S022)

The frozen body above is unchanged. This note records what an independent second reader found; it does not edit the artifact (charter §3, A4: translations are frozen with their logs).

The D-20260725-05 ratification vote made application of the typology changes conditional on an independent, original-language check of this translation and the reading built on it. That check ran blind — google/gemini-3.6-flash was given the classical Chinese and this English with the translator's log stripped, and asked neutral questions naming no sense and no proposal (E-20260725-slatef-verification).

Verdict: SOUND-WITH-QUALIFICATIONS. One defect, in a sentence no reading turns on.

The defect inverts the charge Yan Fu is answering: 文陋 is an accusation of crudeness, and the frozen English turns it into an accusation of bad faith. His reply — 刻意求顯, taking pains to seek plainness — answers the real charge and reads oddly against the rendered one.

Everything the ratified reading rests on was checked and held. Asked on what grounds the author defends his register choice — offering aesthetic, instrumental, both, or neither — the second reader answered instrumental, quoting the same sentence the lead's reading quotes (「用漢以前字法、句法，則為達易；用近世利俗文字，則求達難」), added that the text rules out novelty-seeking (「豈釣奇哉！」), found the three terms to be a hierarchy rather than independent goods (「為達即所以為信也」), and described 雅 as "a property of the translation's language." No part of that came from this translation's framing; the log was withheld.

Standing note for the project. This was the first independent check of any lead translation. Six existed; the first three checked returned two SOUND and this one. NEXT.md note (s) recorded that published translation is not checked translation — it applies to the project's own output too, and the cost of finding out is about two cents a source.