diachronic-claim-typing
Use before literature search and before writing cross-period connection sentences in conceptual history, genealogy, or institutional-history papers, to type claims and audit section selection and over
SKILL.mdskills/en/diachronic-claim-typing/SKILL.md
GitHubDiachronic Claim Typing Skill
Purpose
For papers that connect norms, practices, or institutions across time, assign a claim type first, then select sources, design period-sections, and write connecting sentences. If the type outruns the evidence, shrink the claim. This is not a tutorial in writing survey history.
Genealogy here is selective connection of period-sections, not a complete chronicle. What is forbidden is presenting that selection as if it were the whole history.
Trigger Conditions
- When the research plan involves conceptual history, genealogy, intellectual or institutional history, reception history, or a formation/transformation arc
- When classics, critical editions, statutes, examination systems, or journal-primary texts likely sit outside paper APIs such as OpenAlex
- Before writing cross-period connectors ("from…", "adaptation", "continuity", "unique to the present")
- When cherry-picking, anachronism, or missing transmission evidence has been flagged
Do not invoke for single-period empirical work or cross-sectional surveys that do not claim temporal connection. Use literature-search and claim-evidence-gate instead.
Claim Types
Give each diachronic claim exactly one type. Split compound sentences.
| Type | What it asserts | Evidence required | Failure | Fallback |
|---|---|---|---|---|
| T0 existence | Source S contains text or record X | Locator (edition, transcript ID, catalogue number) plus a verified excerpt in literature/papers/ |
Invented pages; unread secondary | Do not cite |
| T1 interpretation | Inside that text, X can be read as Y | T0 plus acknowledgement of alternative readings | Unique-meaning lock | "a readable cross-section" |
| T2 institution | A practice was a branch of an institution or curriculum | Primary statute, textbook, or case | One case as nationwide | Keep it a case claim |
| T3 transmission | A was translated, imported, or named as B | Translation bibliography, introduction primary, or named citation | Similarity as causation | T3b |
| T3b juxtaposition | Structural similarity of A and B (no causal claim) | T0 on both plus a similarity map | Juxtaposition without similarity | Cut the section or the claim |
| T4 continuity | A method or norm persisted through period P | A primary that argues continuity, or primaries at multiple dates | Two classical points as "until…" | Cross-section only |
| T5 novelty | Conditions at period P differ in kind | Survives hostile reduction (quantity/time, degree) | The historian already states continuity or a precedent | Degree difference, or making a norm/practice contradiction visible |
Reception ("has been received as", "use as reception history") is a claim about someone else's interpretive act. Evidence requirements are T3. T1 is text-internal reading only. Without a reception primary, weaken to T1 "readable as".
Prohibition N1 (non-identification)
Not a type. Applies to every type. Do not write that a historical actor held this paper's coined concept. Exception only when that actor explicitly self-identifies.
Workflow
Step 1: Paper type and inventory
Read research-plan.md. Stop if the paper is not diachronic.
If it is, before literature search create docs/<paper-id>/design/sot/diachronic-claim-inventory.md (or docs/design/sot/ in a single-paper repo). Place intended connections with provisional types.
| Section | Claim (one sentence) | Type | Evidence (`papers/` + locator) | Fallback |
If paper-outline.md exists, copy the type ID as one line in the section note (inventory remains source of truth).
Step 2: Selection criteria and counter-lineage
Write one paragraph into the inventory before searching.
- Selection criteria (canonicity, institutional embedding, OA primary available, etc.)
- Counter-lineage: name the strongest line a reviewer would raise, and why it is out of scope (no straw alternatives)
- Sense of "formation": not encyclopedic origin, but identifying a recurrent starting point, etc.
Step 3: Primary layer and search channels
- Keep survey histories as Discovery, not citation grounds. Descend to the classics, statutes, or primary articles they rest on.
- If a primary is inaccessible, shrink that typed claim or omit the section. Purchase or subscription follows project policy (this skill does not order purchases).
- Channels outside paper APIs (classical digital libraries, J-STAGE, CiNii, institutional repositories) are grounded with WebFetch and manual-stubs. v1.0 adds no new APIs.
- Then (or in parallel) run
literature-search.
Step 4: Connection tests (T3 / T3b / T4 / T5)
- T3: translation, introduction primary, or named citation? Else T3b.
- T3b: T0 on both sides plus a similarity map?
- T4 / T5: does the historian already write continuity or a precedent? Survive hostile reduction (speed = quantity/time, necessity = degree, two points ≠ continuity)? If not, record the fallback and make it a start condition for the chapter.
If literature-gap-analysis novelty is "genealogical connection" and T3 has not passed, change it to juxtaposition/comparison.
Step 5: Pre-draft checklist
- Load-bearing claims have types
- T0 stays inside stub excerpts
- T3/T4/T5 passed or fell back
- N1: no historical actor = coined concept
- If the same analysis term (inside, memory, ability) shifts referent across sections, say it is an analytic construct
- Selection criteria and the strongest counter-lineage are recorded from before the search
- Teleology style: do fillers such as "thereafter", "thus it developed", "was inherited" hide selection? Does the chapter opening state selectivity?
Then continue to counter-argument-tdd.
Invariants
- Start the inventory before literature search (no post-hoc criteria)
literature-matrix.mdis an index; citation grounds areliterature/papers/- Similarity does not satisfy T3
- Do not print T0, N1, or other internal codes in
docs/chapters/
Out of scope (v1.0)
No survey-history tutorial, no field-specific historiography textbook, no CiNii/J-STAGE search script, no illegal access. Future lint candidates (teleology connectives, "has been received" with no citation, coined concept co-occurring with a historical actor) stay manual.
Outputs
docs/<paper-id>/design/sot/diachronic-claim-inventory.md(source of truth;docs/design/sot/in a single-paper repo)