Repository path: workshop/translations/yingyi-jiejixing/R04-v1/translation.md · rendered 2026-09-09

「硬譯」與「文學的階級性」 §五 — Lu Xun on why he hard-translates

• Source: Lu Xun 魯迅, 「硬譯」與「文學的階級性」 ("Hard Translation" and the "Class Character of Literature"), Mengya 萌芽月刊 1:3, March 1930; collected in 《二心集》 (1932). §五 entire, the section in which he answers the objection that unreadable translations are the same as no translations. Chinese Wikisource traditional-character text, stored at ../source-zh-full.txt (10,752 non-whitespace characters) and the translated span at ../source-zh-sec5.txt (7 paragraphs, 1,999 non-whitespace characters). Public domain (Lu Xun d. 1936).
• Regime: R04 v1.0 (lead close translation, source-only). Draft frozen separately as T-yingyi-jiejixing-R06-v1 at commit a563d0b before the revision began, per R04 §Procedure step 2a.
• Translated by: the lead agent, in session. Cost: $0, never ledgered (charter §3, A4).
• Language pair: Chinese → English. The project's first translation from Chinese in either direction, and its eleventh source language.
• Contamination: high, on a measurement — the standard English translation (Yang Xianyi and Gladys Yang, Selected Works of Lu Hsun vol. III) is freely reachable, the gate ran on the frozen draft, and the pre-registered failure branch fired. Full basis: ../contamination.md, and §5 below.
• Evaluation: none designed. The log below is frozen before any is.

§5

At this point I can turn back again to my "hard translation".

By inference, this is a question that ought very properly to arise next: since proletarian literature lays its weight on propaganda, and propaganda must be intelligible to the majority — then these "hard-translated" and unintelligible theoretical "books from heaven" of yours, what after all were they translated for? Is it not the same as not translating them?

My answer is: for myself, and for a few people who set up as critics of proletarian literature, and for that part of the readership which is not out for "ease", is not afraid of hard going, and does want to make out something of this theory.

Since the year before last the attacks on me personally have been beyond counting; in every kind of periodical one is fairly sure to see the name "Lu Xun", and the writer's tone, at a rough glance, is mostly that of a revolutionary man of letters. But I read a few of the pieces, and found myself coming gradually to feel that there was too much idle talk in them. The dissecting-knife misses the seam between the tissues, and where the bullet strikes is no mortal wound. Take the class I belong to: it has not been settled to this day — now it is the petty bourgeoisie, now the "bourgeoisie", sometimes it is promoted to "feudal remnant", and again it is the same thing as an orang-utan (see the "Tokyo Letter" in Creation Monthly); on one occasion the abuse went as far as the colour of my teeth. In a society like this it is entirely possible for a feudal remnant to be in the limelight, but that a feudal remnant should be an orang-utan is set out in no "materialist conception of history" whatever, nor is any argument to be found by which yellow teeth are harmful to the proletarian revolution. So I thought: there is too little theory of this sort available to consult, and that is why everyone is somewhat muddled. Dissecting the enemy and chewing him up cannot, at present, be got round; but with a book of anatomy and a book of cookery, and the thing done by the book, the structure will come out clearer and the flavour better. People often compare the Prometheus of the myth to the revolutionary, holding that in stealing fire and giving it to men, and in not repenting though the Lord of Heaven tormented him, his greatness and endurance are just the same. But I stole the fire from other countries meaning to cook my own flesh with it, thinking that if the flavour were better the chewers on their side might get more good out of it, and my carcass would not have gone to waste either: the starting point is individualism through and through, and mixed in with it besides is the extravagance proper to a petty townsman, and the "revenge" of slowly groping out the dissecting-knife and running it, instead, into the dissector's own heart. Mr Liang says, "They want revenge!" In fact it is hardly only "they"; there are plenty of such people among the "feudal remnants" as well. Still, I am willing to be of some use to society, and what the onlookers see as the result is fire and light all the same. So the first thing I set my hand to was Literary Policy, because it holds the arguments of every school. Mr Zheng Boqi now keeps a bookshop and prints the plays of Hauptmann and Lady Gregory; in those days he was still a revolutionary man of letters, and in Literary Life, which he edited, he laughed at my translating this book as an unwillingness to go under, and thought it a pity that somebody else had got in ahead of me with the whip. That translating one book will float a man up, and that being a revolutionary man of letters is as easy as all that — I do not think so. A certain tabloid said that my translating On Art was a "surrender". Yes: surrenders are common enough in the world. But by that time Marshal Cheng Fangwu had long since crawled out of his Japanese hot springs and moved into a Paris hotel, so to whom here was allegiance to be tendered? This year the wording is different again: in The Pioneer and in Modern Fiction it is called a "change of direction". I have seen those four characters applied in some Japanese magazines to Kataoka Teppei of the old Neo-Sensationist school, where they counted as a good name. In fact all this confused talk is still the old malady of looking only at labels and refusing even to think. Translating a book about proletarian literature is not enough to prove a direction; and if the translation is crooked, it is on the contrary enough to do harm. My translations, then, are to be dedicated to these hasty critics of proletarian literature as well, since they are under an obligation not to covet "ease", and to bear the pain of studying these theories.

But I am confident that I have committed no deliberate crooked translation. When a passage hits the sore spot of a critic I do not admire, I smile; when it hits my own sore spot, I bear the pain; but I will not for that add or subtract a thing — and that too is one of the reasons for keeping to "hard translation" from first to last. Naturally there will always be better translators in the world, able to turn out writing that is neither crooked nor "hard" nor "dead", and then my versions will of course be weeded out; and I have only to fill in this space from "nothing at all" to "something better".

But there is still a great deal of paper in the world, while the members of each literary society are few; their ambition is large and their strength thin, and they cannot write over all the paper — so the critic whose office in the society it is to vanquish enemies, succour friends and sweep out the alien kinds, seeing others come and scribble over paper, heaves a deep sigh and suffers past bearing the misery of shaking his head and stamping his foot. The Shen Bao of Shanghai has gone so far as to call translators of the social sciences "Ah-Dog and Ah-Cat", and its indignation runs to that. Mr Jiang Guang-Z, whose "position in China's new literature has long been known to all readers", once went to Tokyo to convalesce, and meeting Kurahara Korehito talked of how many translations in Japan were so bad that they were simply harder to read than the originals... whereupon he burst out laughing and said: "...then China's translation world must be still more unaccountable. Lately a great many books in China have been translated from the Japanese; if the Japanese carry over a work of some European country with a little error and a little cutting, and it is then translated from the Japanese into Chinese, may one ask whether the work will not have changed half its face?..." (see The Pioneer.) This too is an expression of deep dissatisfaction with translation, and above all with relay translation. But Mr Liang at least cited the titles of books and their faults, whereas Mr Jiang merely gave a sweet smile and swept the field bare — which is indeed a great deal more universal. Kurahara Korehito has translated a good many works of literary theory and fiction directly from the Russian, and to me personally he has been of the greatest benefit. I hope China too may have one or two such honest translators from the Russian, who will bring out good books one after another, and not reckon the duty of a revolutionary man of letters discharged by swearing "Swine!" once on his own account.

But as things are, Mr Liang Shiqiu does not translate these things; the great man who calls people "Ah-Dog and Ah-Cat" does not translate them either; Mr Jiang, who has studied Russian, would have been the most suitable of all, but unfortunately since his convalescence he has brought out only A Week, and Japan has had two versions of that long since. China once made much of Darwin and made much of Nietzsche, and by the time of the European War made a great business of cursing them both; but of Darwin's works there is to this day only one translation, and of Nietzsche only half a book, while the scholars and great writers who have studied English and German have had no leisure to attend to it, or have not deigned to, and so it was let drop. For the time being, then, I am afraid there is nothing for it but to let people laugh and abuse as they will, and go on relaying from the Japanese, or take a copy of the original and translate it straight with the Japanese version alongside. I still mean to do this, and I hope there will be more people who do, to fill in a little of the emptiness inside the thoroughgoing high talk — because we cannot "burst out laughing" as Mr Jiang did, and ought not to go on, like Mr Liang, "waiting, waiting, waiting".

Translator's log

Frozen 2026-07-28, before any evaluation of this translation was designed. Decisions and alternatives, not quality claims. The artifact carries internal-judgment-only: true in its front matter rather than flagging sentences individually, because several entries below compare the draft against the revision (flatter, too genteel, heavier) and a blanket flag is the conservative reading of that. The lead never judges its own translation (charter §5); nothing here is a quality claim about the finished text.

A binding rule adopted before the revision, and the reason for it. The contamination measurement (../contamination.md) ran on the frozen draft and between draft and revision. It reported four sites where the draft shares seven or more tokens with the Yangs' published English. Those four sites — L1 but there is still a great deal of paper in the world, L2 the colour of my teeth in a society like this, L3 the plays of Hauptmann and Lady Gregory, L4 to be of some use to society — were not altered in the revision on account of the measurement. Changing a rendering because an instrument says it matches is optimising against the instrument, and it would make every future measurement on this artifact meaningless. They stand exactly as drafted. This is nevertheless a priming event and is declared as one: after the measurement the lead knew that four short strings of its own draft coincide with the published version, which R04 §Procedure step 1 does not contemplate. It is limited to those four strings; no comparator prose was displayed at any point.

The pair, and what is new about it

1. First Chinese source in the project, and the first non-alphabetic source that is also a critical text. The project has read classical Chinese before (S-yanfu-yili-yan, 嚴復's 譯例言, 1898) and translated from it; this is modern vernacular Chinese, 1930, and it is polemic rather than preface. The register problem is therefore new: Yan Fu's classical prose is compressed and impersonal, Lu Xun's is a speaking voice under sustained sarcasm, and almost every difficulty below is a difficulty about tone rather than about sense.

2. The source is itself an argument about translation, and it argues for the opposite of what English wants. Lu Xun's 硬譯 is the deliberate retention of source syntax at the cost of target fluency. Translating him out of Chinese meant repeatedly noticing a choice between doing what he says and doing what the English sentence wants — and the honest record is that the two rarely conflicted, because his prose is not itself hard-translated. His polemical Chinese is fluent, idiomatic and fast. The doctrine governs his translations of Lunacharsky, not his own essays. That asymmetry was not obvious before translating and is the single most useful thing this limb produced (S-luxun-yingyi §5).

Register and voice

3. The essay quotes its opponent constantly, in scare quotes, and the quotes carry the sarcasm. 「爽快」, 「硬著頭皮」, 「等著，等著，等著」, 「他們要報復！」 are Liang Shiqiu's words being handed back to him. English quotation marks do the same job, so this transferred at almost no cost — the one place the two languages' apparatus lines up exactly. Retained everywhere, including the sentence-final 「等著，等著，等著」 that closes the section.

4. 「爽快」 — Liang's word, four appearances in the essay, two in this section. Live options: ease · a smooth ride · comfort · briskness. Chose "ease", held at both sites, because the noun has to work both as something a reader is out for (不圖「爽快」) and as something a critic covets (不貪「爽快」). Briskness is closer to the word and works at neither.

5. 「看客」 — onlookers. Lu Xun's loaded noun: the gawping crowd at a public execution, the word behind his own account of why he gave up medicine. English has no single word carrying that freight. Lost; onlookers is the flat reading and no scaffold was available inside the sentence.

6. 「嫣然一笑」 of Mr Jiang — a smile conventionally attributed to a beautiful woman. Applied to a male critic it is a sneer with a gendered edge. Chose "gave a sweet smile", which keeps the sweetness and loses the gender. Considered and rejected gave a simper (right register, wrong direction — it makes Jiang silly rather than affected).

7. 「混蛋」 — options Blackguard! · Swine! · Bastard!. The draft had Blackguard!, which is Edwardian and too genteel for a term of abuse thrown by 1930 Shanghai literary revolutionaries. Revised to "Swine!".

The two knives, and one thread held

8. 「解剖刀」 / 「解剖」 / 「解剖者」 run three times in one paragraph on a shared root. The draft had scalpel / dissecting / dissector, which breaks the thread at the first site. Revised to dissecting-knife / dissecting / dissector, holding it 3/3. This is the same handling the Beowulf anchor records for duguð/dugan (A-beowulf-ingeld), and it costs the same thing: dissecting-knife is heavier than scalpel.

9. 「腠理」 — the interstices between skin and flesh in Chinese medicine, from the 扁鵲 story about a disease that can still be reached there. Options: the vital spot (substitute) · the seam between the tissues (transparent calque) · footnote. Chose the calque, since the sentence's other half already supplies the sense (「子彈所擊之處，也不是致命傷」). The Bian Que allusion is lost with no trace.

10. 「不甘沒落」 … 「便會浮起」 — sink and float, a submerged pair three sentences apart. Kept as "an unwillingness to go under" and "will float a man up", so the pair survives.

11. 「著了先鞭」 — idiom, from 祖逖's I fear only that Liu Kun will get in ahead of me with the whip. Rendered "got in ahead of me with the whip" rather than the plain beaten to it, on the ground that the whip is still legible in English as an image even when the story behind it is not — a copy of the vehicle without the tenor. The draft had the flat version; this is the one place the revision moved away from fluency, and it did so because the essay's own argument is that this is allowed.

12. 「填…空間」 (§P4) and 「填一填…空虛」 (§P6) — the same verb bracketing the section, once of a space and once of an emptiness. Held 2/2 with fill in this space and fill in a little of the emptiness; the reduplicated 填一填's casualness is carried by a little of.

The imported words, and an asymmetry worth recording

13. Three Roman-alphabet tokens stand in the Chinese source: Prometheus, Hauptmann, Gregory夫人. In Chinese each is visibly an import — the writing system changes mid-sentence. In English all three become ordinary words and the visible foreignness is annihilated. This is the reverse of the usual direction (a target-lacking category, TH-20260724 C1); here the source had a marking device the target simply cannot operate, because English cannot switch out of the Latin alphabet to signal foreignness.

14. But a fourth Latin-script token survives, and the difference is the point. 「蔣光Z」 — Jiang Guangci's name with its last character replaced by a bare Z, a redaction Lu Xun made in the Chinese. Rendered "Mr Jiang Guang-Z", and in English it is still visibly odd. So Roman-alphabet insertions that are foreign names vanish into English, while a Roman-alphabet insertion that is a redaction transfers intact. The variable is not the script but whether the target reader can still see that something has been done to the word.

15. 「布爾喬亞」 — a transliteration of bourgeoisie, standing beside 「小資產階級」, a calque, in the same clause: 「忽說小資產階級，忽說“布爾喬亞”」. The Chinese contrast is between a naturalised loan-calque and a raw phonetic import, and it is the phonetic one that is in scare quotes. In English both collapse to the same word. Chose "now it is the petty bourgeoisie, now the 'bourgeoisie'", with the italic doing what it can. Mostly lost; the sentence in English looks like a stutter where the Chinese looks like a jibe.

16. 「這四字」, of 「方向轉換」 — those four characters. Retained as counted characters rather than converted to that phrase, because English readers can still parse a count of characters as a fact about Chinese. Retained at the cost of a small opacity, which is the handling the essay's own doctrine prescribes.

17. 「阿狗阿貓」 — Ah-Dog and Ah-Cat. The 阿 prefix is a familiar personal marker (as in 阿Q); the phrase means any Tom, Dick or Harry while literally naming a dog and a cat. Options: substitute (Tom, Dick and Harry — loses the animals and the prefix) · copy-opaque · calque. Chose copy-opaque, twice, consistently. Recorded here because the choice was made under the essay's own doctrine, and a translator who did not hold that doctrine would very reasonably substitute.

Sense, syntax, and one editorial intervention

18. 「天書」 — literally books of heaven, i.e. celestial script, unreadable. Liang's original jibe was 「比天書還難」. Kept "books from heaven" in quotes, with the adjacent unintelligible doing the scaffolding.

19. 「猩猩」 — the draft had an ape; revised to "an orang-utan", which is what the word denotes and which makes the bathos of the accusation land. Considered a gorilla (also within the word's range, but the wrong kind of funny).

20. 「封建余孽」 — held as "feudal remnant" at all three sites for consistency. 孽 carries evil brood / accursed spawn and is gone; feudal spawn was tried in the draft margin and rejected because the phrase has to survive being used four times in a paragraph without becoming a joke of its own.

21. 「對於敵人，解剖，咬嚼，現在是在所不免的，不過有一本解剖學，有一本烹飪法，依法辦理，則構造味道，總還可以較為清楚，有味。」 — the sentence distributes: anatomy→structure→clearer, cookery→flavour→better. The draft collapsed the distribution (the structure and the flavour can be made rather clearer and rather better). Revised to "the structure will come out clearer and the flavour better", restoring the pairing. 「依法辦理」 became "the thing done by the book", which puns on the two books just named; the pun is not in the Chinese and is recorded as an addition made because it costs nothing and holds the sentence together.

22. 「元帥」 of Cheng Fangwu — "Marshal". Sarcastic in the Chinese by sheer inflation; Generalissimo was rejected as belonging to a different quarrel.

23. 「大談…大談…大罵」 — a triple 大 in one sentence. The draft carried two of three (a great deal … a great deal … a thorough cursing). Revised to "made much of Darwin and made much of Nietzsche, and … made a great business of cursing them both", holding 3/3.

24. 「淘汰」, of his own translations being superseded — a word from the Chinese vocabulary of natural selection (優勝劣敗, 物競天擇), and Darwin is named two paragraphs later. "Weeded out" keeps a selection metaphor but not the Darwinian one; the echo across paragraphs is lost.

25. An editorial intervention on a defective source text, recorded because it is invisible otherwise. The Wikisource text reads 「（見《拓荒者》也就是深不滿於翻譯…」 — the parenthesis opens and never closes, and the citation runs straight into the next sentence. This is a corruption in the electronic text, not a feature of Lu Xun's prose. Closed it as (see The Pioneer.) and started a new sentence. No other emendation was made anywhere in the section.

26. 「他就笑了起來」 — he is ambiguous between Kurahara and Jiang, and is resolved only later by 「蔣先生卻只嫣然一笑」. English he is ambiguous in exactly the same way at the same place, so the ambiguity was kept rather than resolved; this is one of very few places where the two languages' referential machinery fails identically.

27. 「輸誠」 — a term for tendering allegiance in surrender. "To whom here was allegiance to be tendered?" The draft's pledged was revised to tendered for the transactional coldness.

28. 「不勝其搖頭頓足之苦」 — the frame is 不勝…之苦, cannot bear the suffering of. The draft dropped the frame into suffers past bearing from. Revised to "suffers past bearing the misery of shaking his head and stamping his foot", which keeps the noun.

29. 「克敵助友，掃蕩異類」 — two four-character phrases in series. Rendered as three parallel infinitives, "to vanquish enemies, succour friends and sweep out the alien kinds", trading the 4+4 shape for an English tricolon. The draft's overcome / assist / sweep out was flatter.

30. What the revision changed overall. Twenty-two of the thirty decisions above were settled in the draft and left alone. The revision's substantive moves were: the 解剖 thread (8), the distribution in the anatomy/cookery sentence (21), the triple 大 (23), the 先鞭 idiom (11), the 不勝…之苦 frame (28), the tricolon (29), orang-utan (19), Swine! (7), and 「從…到」 read as from…to rather than between in the "space between nothing and better" sentence. No sentence was rewritten for fluency; every change was toward a formal feature of the source that the draft had flattened, which is the direction the essay itself argues for and is recorded here as a possible influence of the source's content on the translator's practice rather than as a virtue.

What was not done

31. No published rendering of this passage into any language was consulted before the draft was frozen. The Yangs' English was downloaded, located by structural markers, and measured against — never displayed — with the four-string exception declared at the head of this log.
32. No dictionary of modern English usage, no style guide, and no target-side reference material of any kind (R04 §Known limitations).
33. The 1930 Shanghai literary quarrel was not researched beyond what the source text itself says. Identifications of 鄭伯奇, 成仿吾, 蔣光慈, 藏原惟人, 片岡鐵兵 and the journals rest on the lead's prior knowledge and are not verified against sources; where the log names a person or a title it is reporting what the translator believed while translating, which is the honest scope of a translator's log. S-luxun-yingyi flags the same limitation for the study limb.

5. Contamination — measured, and the failure branch fired

Full design and result: ../contamination.md. In summary:

The pre-registered control asked whether the runs are forced by the Chinese or recalled from the English, by giving three independent panel models the four Chinese sentences, blind, and scoring whether they reproduce the runs. The rule and its failure branch were frozen in git before dispatch.

All three independent models rendered 世間紙張還多 as "there is still plenty of paper in the world". None produced "a great deal of". The 12-token run — the whole of the headline number — is therefore not forced by the source, and the design's failure criterion binds: contamination: high.

What that costs, stated plainly. This artifact is a practice artifact and a source of process evidence. It is not an independent measurement of translating ability, it may not be used as an independent third translation in any baseline, and no claim about the lead's Chinese→English competence may rest on it (R04 §Known limitations, charter §5).

What it buys. The project's standing rule — the lead cannot serve as the independent third translator for any work whose standard English translation is canonical — has until now rested on the Turgenev/Garnett run of 21 tokens, declared but never tested. This is the first time the forcing defence has been tested and refused on a lead translation, and it is the second independent refusal of that defence overall, after RS-20260726-ovid-period-form refused it on Latin. Note (bcd) is amended accordingly and note (bcp) is opened.