Initial toolkit: scripts, references, skills, and term database

This commit is contained in:
iacore
2026-07-10 21:01:02 +08:00
commit ac9e6e3d1f
51 changed files with 5192 additions and 0 deletions
+213
View File
@@ -0,0 +1,213 @@
---
name: translation-review
description: |
Unified review skill for Chinese-English Buddhist/Dharma translations.
Supports two modes:
- **self**: You translated the text. Edit target.dj directly.
- **other**: Someone else translated. Write review-comments.dj, do not edit target.dj.
What to review is identical in both modes — the same detection rules,
editorial standards, and terminology checks. Only the action differs: self
mode applies fixes directly; other mode records them for the translator.
---
# Translation Review (unified)
This skill replaces the previous `self-review` and `other-review` skills.
**What to review is the same no matter the mode.** The same detection rules,
editorial standards, and terminology checks apply whether you are polishing your
own translation or reviewing another translator's work. Only the action differs:
in self mode you edit `target.dj` directly; in other mode you write
`review-comments.dj` for the translator to apply.
## Modes
| Mode | When to use | Output artifact | Interaction style |
|------|-------------|-----------------|-------------------|
| **self** | You produced the current `target.dj`. | Edited `target.dj` + regenerated `bilingual.dj`. | Direct, decisive: you own the English. |
| **other** | Another translator produced the current `target.dj`. | `review-comments.dj` only. | Collaborative: ask before prescribing, 随喜 first, address the translator by name. |
A third mode — **direct edit** — is used when the user explicitly asks you to
apply your findings to `target.dj` even though it is someone else's translation.
Switch to self-mode output in that case, but keep a collegial tone.
## Core workflow (both modes)
1. Read `source.dj` and `target.dj` fully before changing anything.
(Other mode only: read the translator's note if present, and address them by name.)
2. Run a three-pass review using the detection rules below.
3. Apply the R1R14 editorial polish checklist.
4. Check terminology against the terms DB (see `terms-search` skill).
5. Verify that every source paragraph maps to a target paragraph with no missing
or truncated content.
6. Produce the correct artifact for your mode.
## Detection rules — three passes
Run these in order. Stop at the end of each pass and consolidate findings before
moving to the next pass. In self mode, fix immediately; in other mode, record the
issue with line/paragraph references.
### Pass 1: Accuracy and completeness
- **A1 Missing content.** Check for mid-paragraph truncation, skipped lines, or
entire paragraphs missing. Compare source and target line counts and paragraph
boundaries.
- **A2 Mistranslation.** Any word or phrase that contradicts the source or is
seriously off in meaning. Flag technical terms, numbers, negation, and modal
verbs especially.
- **A3 Terminology inconsistency.** The same Chinese term rendered in multiple
ways without justification. Check proper names, Dharma terms, and repeated
metaphors.
- **A4 Unnecessary addition.** English content not present in the source that
was added for fluency but changes the meaning. Common: adding "we should,"
"we must," "it is important to" where the source is descriptive.
- **A5 Number / time / person mismatch.** Chinese often omits plurality, tense,
and subjects. Ensure English reflects the intended scope and subject.
### Pass 2: Fluency and naturalness
- **F1 Calque / translationese.** Source grammar or word order carried over
literally: "as for X ...", "only then can ...", "more ultimate,"
"choice difficulty," "keen on" (as in "keen on greed"), etc. See
`references/translation-pitfalls.md` for a pattern table.
- **F2 Register drift.** The English is too formal, too casual, too academic,
or too sermonizing relative to the source. Match the source's register.
- **F3 Broken collocation.** English words that do not normally appear together
(e.g., "arise greed," "do delusion"). Fix to the usual collocation
("give rise to greed," "overcome delusion").
- **F4 Pronoun / reference chain error.** Missing subject, ambiguous "it" or
"this," or inconsistent names/pronouns across sentences.
- **F5 Sentence rhythm.** Overly long, convoluted, or monotonous sentences. For
oral talks, respect breath units and emphasis.
### Pass 3: Dharma and cultural fitness
- **D1 Buddhist term register.** Sanskritic terms (e.g., *samatha*, *vipaśyanā*)
vs. vernacular English ("calm abiding," "insight") vs. Chinese calques. Choose
consistently with the text's audience and the project's conventions. See
`references/buddhist-terminology.md`.
- **D2 Cultural anachronism.** Modern concepts projected onto classical material.
Example: translating 般若 as "wisdom" in a scholarly context may flatten the
term; in a popular talk it may be exactly right.
- **D3 Tone / voice of the teacher.** For oral talks, preserve the speaker's
warmth, rhetorical questions, and direct address. Do not flatten into essay
prose.
- **D4 Implicit meaning / implicature.** What the source implies but does not say
(e.g., irony, conventional politeness, Gricean maxims). Ensure the implication
survives or is compensated.
- **D5 Formatting fidelity.** Emphasis markers (`*...*`), paragraph breaks, list
structure, and soft-line markers (` `) are preserved. Do not introduce
Markdown `**` bold.
## R1R14 editorial polish checklist
After the three passes, run these final checks.
1. **R1 Spelling and punctuation.** Especially English em/en dashes (`---`, `--`)
and Chinese punctuation that should not survive.
2. **R2 Capitalization.** Sentence case, proper nouns, titles, Sanskritic terms
according to convention.
3. **R3 Articles.** Missing or surplus "a/an/the" after Chinese source.
4. **R4 Subject-verb agreement.** Chinese omission of subjects can hide errors.
5. **R5 Tense consistency.** Historical narrative vs. timeless teaching vs.
present event.
6. **R6 Voice.** Active/passive choices match the source and English register.
7. **R7 Prepositions.** Common Chinese-to-English transfer errors: *in, on, at,
of, to, for*.
8. **R8 Modifiers.** Adjective/adverb position, misplaced modifiers, stacked
nominalizations.
9. **R9 Parallelism.** Lists, comparisons, and repeated structures should be
grammatically parallel.
10. **R10 Redundancy.** Remove unnecessary repetition or filler introduced during
drafting.
11. **R11 Word choice.** Precision vs. overuse of generic words ("thing," "aspect,"
"level," "situation"). Replace with concrete terms.
12. **R12 Sentence openings.** Vary sentence beginnings; avoid a string of
"The..." or "It..." or "This..." starts.
13. **R13 Flow / transitions.** Logical connections between sentences and
paragraphs are clear.
14. **R14 Final read aloud.** Read the target text as if it were being delivered
to the intended audience. Fix anything that trips the tongue or the ear.
## Mode-specific instructions
### Self mode
- You are responsible for the English. Make the edit directly.
- Use `patch` or exact-string edits. Avoid speculative rewrites of entire
paragraphs unless necessary.
- Record only non-obvious or project-level decisions in
`translation-findings.dj`.
- After editing, regenerate `bilingual.dj` with
`../../scripts/gen-bilingual.py source.dj target.dj > bilingual.dj` and verify
line counts match.
### Other mode
- Do not edit `target.dj`. Write `review-comments.dj`.
- Address the translator by name if known.
- Deliberation protocol:
- Start with what is strong (随喜 / appreciation first).
- Phrase most issues as questions or options, not commands.
- Distinguish "must fix" errors (accuracy, missing content, serious
mistranslation) from "consider" suggestions (style, register, optional
polish).
- Include exact source/target snippets and line/paragraph references so the
translator can locate every issue quickly.
- When the user asks you to apply the findings, switch to direct-edit mode and
treat it as self-mode output.
## Output formats
### Self mode: edited target.dj
Make targeted changes. If you change a term, search the whole file for that term
and update it consistently. After editing, regenerate the bilingual file and
verify line counts.
### Other mode: review-comments.dj
Structure the file as follows:
```dj
# Review comments for <article>
Translator: <name>
Reviewer: <model / reviewer name>
Date: <YYYY-MM-DD>
## Summary
One or two sentences on overall quality and the main issue types.
## Must-fix issues
1. **Accuracy / missing content.** <source> <target> <why it matters>
2. ...
## Considerations
1. **Fluency / register.** <source> <target> <suggestion>
2. ...
## Terminology notes
Any DB queries or proposed term changes.
## Closing
appreciative, forward-looking note.
```
## References
- `references/common-issues-taxonomy.md` — structured accuracy/readability
checklist used in the three passes.
- `references/translation-pitfalls.md` — recurring calques and stiff renderings
for this project's genres.
- `references/buddhist-terminology.md` — register and convention notes for
Buddhist/Dharma terms.
- `../translation/SKILL.md` — the upstream translation skill that produces the
`target.dj` this skill reviews.
- `../terms-search/SKILL.md` — skill for querying the terms database before and
during review.