Repository path: workshop/translations/wang-liulang/provenance.md · rendered 2026-09-09
Page metadata (front matter)
| type | note |
|---|---|
| id | wang-liulang-provenance |
| status | frozen |
| created | 2026-07-28 |
| updated | 2026-07-28 |
| links | workshop/translations/wang-liulang/source-zh.txt, workshop/translations/wang-liulang/R04-v1/translation.md |
〈王六郎〉 — source text, two witnesses, two emendations
Pu Songling 蒲松齡 (1640–1715), 《聊齋志異》 卷一 〈王六郎〉. Public domain. 1,533 characters in four paragraphs (291 / 592 / 510 / 140), narrative plus the 異史氏曰 coda.
Witnesses
- W1 — Chinese Wikisource,
聊齋志異/第01卷, fetched via the MediaWikiaction=parseAPI, 2026-07-28. Traditional characters,「」quotation, no editorial commentary in this volume's presentation. Base text. - W2 — Project Gutenberg #51828, 《聊齋志異》 (Chinese), fetched 2026-07-28. Traditional characters,
“”quotation.
What the comparison showed
Whitespace-stripped, W1 is 1,533 characters and W2 1,516; a character-level diff returns 166 opcodes. The great majority are not textual variants at all:
- Quotation marks. W1
「」『』, W2“”‘’throughout — ~50 opcodes. - Comma density. W2 drops ~25 commas W1 has (
一夕,方獨酌/一夕方獨酌), and swaps;for,or。at ~6 sites. - Mechanical 簡→繁 over-conversion in W2, which is what most of the remaining single-character differences are, and which is why W2 is the check and not the base: 云→雲, 獲→穫, 復→複, 別→别, 揚→颺, 旋→鏇, 為→爲, 只→隻, 沓→遝, 折→摺, 余→餘, 治→制, 於/于 normalised the other way. Each is the wrong member of a simplified-character merge, restored by rule rather than by reading. W2's text has been through an automatic converter; W1's has not.
Four differences are substantive and are recorded rather than resolved, W1 kept in each case: 飲則酹地 (W2 飲則酹酒於地); 要當以為長耳 (W2 …以為常耳 — W1's 長 contrasts with the 以為常 four lines earlier and is the harder reading); 許貨魚,益沽酒 (W2 許貨魚益利,沾酒); 家綦貧 (W2 家甚貧 — 綦 is the classical intensifier and 甚 the flattening).
Two emendations, both adopted, both from W2
W1 is corrupt at two places and W2 supplies the reading. Both are in the base text as used; nothing else was altered.
-
由此盯聚如初→由此相聚如初. W1's 盯 (to stare) is not a possible reading in the frame由此…如初; W2 has 相. A graphic or input corruption in W1. -
環如牆堵。許乃告曰:「數夜前,夢神言:淄川許友當即來,可助為資斧。祗候已久。」→…眾乃告曰:…. This is the one that changes the sense of a whole scene. Under W1 it is Xu who announces that a god appeared in a dream several nights ago saying "Xu of Zichuan, my friend, will be here shortly; help him with his travelling money" — but Xu has just arrived, does not know the villagers, and cannot be the one who had that dream or who has been 祗候已久 (waiting a long while). W2's 眾 puts the speech where the sense requires it: the villagers tell Xu. W1's reading is unconstruable; the emendation is minimal (one character) and is not a choice between two viable texts.
W2 additionally reads 環如牆堵。許益驚。眾乃告曰 — an extra sentence. Not adopted; the emendation is held to the single character the sense forces.
Comparator for the contamination gate
Herbert A. Giles, "The Fisherman and his Friend", in Strange Stories from a Chinese Studio, Vol. 2 (1880), Project Gutenberg #43628 — public domain and freely reachable, which is why this tale was preferred to the several others in the same collection whose English is not free.
Giles's English of this tale was not read at any point before the translation was frozen. The gate is computed by a script that prints counts and run lengths and never displays comparator prose (method note (abm)). See R04-v1/contamination.md.