Repository path: workshop/translations/kachikachiyama/propositions.md · rendered 2026-09-09
Page metadata (front matter)
| type | manifest |
|---|---|
| id | kachikachiyama-r10-propositions |
| status | frozen |
| created | 2026-08-03 |
| updated | 2026-08-03 |
| links | workshop/regimes/R10-anchored-register.md, wiki/base/anchors/A-english-tale-register/A-english-tale-register.md, wiki/base/anchors/A-mchugh-presence/A-mchugh-presence.md, workshop/experiments/E-20260803e-purpose-index/design.md |
R10 target propositions — the matched pair on 楠山正雄「かちかち山」
Frozen 2026-08-03 (S098) before either translation was begun. R10 §2 requires the target
catalogue to be decomposed into numbered propositions and committed before translating. Two targets,
one source, one translator, one session: only the target catalogue varies (R10 "the pairing is the
point").
Declared departure from R10 §2, stated because it is a real weakness
R10 §2 says the propositions are frozen before the source is read for translation. They were
not. The source was read once, at selection, before A-english-tale-register existed — the tale had
to be read to know whether it was worth a session. So the ordering guard against tailoring the
catalogue to the source was not available for target T.
What limits the damage, and it is asymmetric between the two targets:
- Target C's catalogue is
A-mchugh-presence, frozen 2026-07-30 (S063), five sessions before this source was chosen. No tailoring is possible there, by date. - Target T's catalogue is
A-english-tale-register, written this session — but its content is determined entirely by two English corpora (Jacobs 1890, Kipling 1902) and its attestation counts are mechanical over those corpora. Nothing in it was read off the Japanese. - The asymmetry is the honest residual risk: T's catalogue could have been selected — not written — with this tale in mind. It was. The tale was chosen because its register is genre-marked in Japanese, which is the whole reason it can test the question. That is selection of materials to fit a question, declared here, and not a claim of blindness.
Target T — A-english-tale-register (purpose-indexed)
| id | proposition | opportunity in this source? |
|---|---|---|
| T1 | Open with an existential frame formula of the once upon a time / there was once / there lived family. | YES — 「むかし、むかし、あるところに、…がありました」 is the Japanese frame formula, exactly parallel. |
| T2 | Carry narration on marked temporal connectives from the repertoire (presently, by and by, whereupon, of a sudden, at last) rather than on subordination or on bare then. |
YES — the source runs on するとその, しばらくすると, そのうちに, やがて, とうとう. |
| T3 | Attribute speech with inverted or historic-present tags (said he, says she, quoth), not with a postposed modern tag. |
YES — the source's tags are all と言いました / と言って. |
| T4 | Prefer parataxis: sentence-initial And and So, coordination over subordination. | YES — the source is built on と / て-forms and すると. |
| T5 | Keep verbatim repetition where the source repeats, rather than varying for elegance. | YES — the three-mountain sequence, the two 「ああ、そうか。」, the twice-asked 「かち栗をくれないか」. |
| T6 | Use fixed epithets and flat moral labelling where the source labels. | YES — 古だぬき, the rabbit's むくい speech. |
| T7 | Preserve the assumption of orality; allow the teller's own voice where the source has one. | PARTIAL — 楠山's narrator does not address a listener directly; nothing may be invented. Recorded as an opportunity the source does not supply, per R10 §6. |
| T8 | Keep the lexicon small, concrete and non-Latinate. | YES — throughout. |
Target C — A-mchugh-presence (corpus-indexed: unmarked / literary-contemporary)
Propositions taken from that page's own catalogue, unchanged in substance.
| id | proposition | opportunity in this source? |
|---|---|---|
| C1 | No period lexis, no archaism, no learned or Latinate display. | YES |
| C2 | Dialogue attributed almost entirely with say, postposed and transparent; no retorted, quoth, no inversion. |
YES |
| C3 | Figuration sparse, short, unemphatic. | PARTIAL — the source has almost no figuration to place or to suppress. |
| C4 | Texture carried by named particulars rather than by style; particulars not glossed. | YES — 臼, きね, かち栗, 火打ち石, こうやく. |
| C5 | Standard contractions everywhere, in narration as well as speech. | YES |
| C6 | Short paragraphs, one-sentence paragraphs used structurally; rhythm managed by white space. | YES |
| C7 | Strong feeling and violence delivered flat and unframed, without emphasis. | YES — the killing, the burning, the drowning. |
| C8 | Sentence length varies widely and unpatterned. Recorded as the least discriminating property in A-mchugh-presence's own census and therefore expected to carry nothing. |
YES, and expected inert. |
Coding
Every logged decision carries R10 §5's target code — A (aimed, proposition ids), X
(contrary: a proposition bore and was not followed, with reason), N (none bore) — assigned at the
moment the decision is made.