# Language Book Entry Schema v1.0

Language Book is a **cross-language comparable semantic database**. It records mappings that can be searched, compared, reviewed and tested. It is not a proof that all languages share one historical origin.

## Editorial policy

> 词条可以发表，假说必须标级，文学可以自由展开，证据必须独立核验。  
> An entry may be published; a hypothesis must be graded; literature may explore freely; evidence must be evaluated independently.

> 一个英语词 → 一个主要汉语 Mapping → 独立历史词源 → 独立声音假说 → 独立实验验证。  
> One English word → one primary Chinese mapping → separate historical etymology → separate sound-semantic hypothesis → separate experimental validation.

`entry_status`, `mapping_status`, `historical_relation_status` and `literary_layer.status` are independent axes. `Published` means that an editorial object is public; it never means that a mapping, hypothesis, historical relation or experiment has been proved.

## Canonical files

- Dataset: `data/language-book.v1.0.json`
- JSON Schema: `data/language-book-entry.schema.v1.json`
- Deterministic migration: `scripts/build-language-book-v1.mjs`
- Validation: `scripts/validate-language-book-v1.mjs`
- Browser adapter: `js/language-book-data.js`

The v0.6 dataset and earlier package records remain in the repository as migration provenance. v1.0 copies each old entry into `legacy` so that no reviewed object is silently discarded.

Current Legacy Website Import release: `1.2.0` (37 records; Schema remains `1.0.0`). Legacy Batch 001 adds fourteen schema-compatible records, merges four old-site sources into existing records and defers one unclear item without changing the v1.0 canonical filename or core schema. Releases may add reviewed records and optional schema-compatible fields without conflating publication, mapping, historical evidence or literature. The Namcha Barwa record continues to demonstrate optional named-entity and multilingual literary fields; ordinary lexical records remain valid without them.

## Entry fields

| Field | Meaning |
|---|---|
| `id`, `slug`, `title` | Stable identity, URL/search slug and bilingual display title. |
| `languages` | Source, target and optional literary-transformation forms with language and pronunciation. |
| `primary_mapping` | Exactly one source word and one primary Chinese target, plus gloss, meaning, type and rationale. It does not contain historical etymology. |
| `entry_status` | Editorial lifecycle of the entry itself. |
| `mapping_status`, `mapping_level` | Independent classification and A–D research level of the cross-language mapping. |
| `historical_relation_status` | Independent judgment about a historical relation. `Not claimed` is different from `Unestablished`. |
| `evidence` | Four independent tracks: `Historical`, `Phonetic-Semantic`, `Cognitive`, `Speculative`; every track has its own `status` and `confidence`. |
| `phonetic_observation` | Modern-form observations and explicit limitations. An observation is not a regular sound law. |
| `semantic_structure` | Comparable conceptual units and their relation, such as `LOCATION → PRESENCE`. |
| `related_words` | Optional typed word relations. `relationship_type` and `family` distinguish genuine historical derivatives/borrowings from speculative semantic associations. |
| `semantic_associations` | Optional cognitive or literary concept links. `is_etymological` states explicitly whether a link is historical. |
| `hypotheses` | Individually graded consonantal, root-consonant, vowel or other candidate rules. |
| `experiments` | Plans or results linked to the hypotheses they test. No experiment is inferred from publication. |
| `literary_layer` | Proposition, prose, poem/lyrics, translations and archival manuscript references. `is_historical_evidence` must be `false`. |
| `media` | Typed media with bilingual alt/caption and source/provenance. Unknown values are recorded as `unknown`, never guessed. |
| `references` | Source records used by evidence objects. |
| `author`, `version`, `dates` | Editorial authorship and version history. The project author convention is `Jinkai Liu`. |
| `editorial_notes` | Bilingual scope and evidence-boundary notes. |
| `search_terms` | Canonical search aliases consumed by Search, Dictionary and Semantic Mapper. |
| `query_views` | Optional related-word standard-meaning views. Each has exact `terms`, a validated `primary_mapping`, and `featured_mapping: null`. The adapter preserves ID, page, evidence and count, and never overwrites the canonical record. |

Correction release `1.2.11` keeps 37 records and Schema `1.0.0`. The stable `abash` record now centers on BASH ↔ 拍. Following the existing featured-form convention, `primary_mapping` carries bash’s standard translation and `featured_mapping` carries the Candidate comparison with 拍. Queries for abash or 窘迫 display its related standard meaning and historical correction, using the same frozen Mapper renderer. Research level remains C; no rule is promoted. `HYP-BPMF-CONSONANT-GROUP` records the user’s existing grouping under a stable case-level identifier (no earlier dedicated b-p-m-f identifier was found in this checkout). Its linked experiment is planned, not run.

## Enumerations

- Entry: `Draft`, `Reviewed`, `Published`, `Archived`
- Mapping: `Candidate`, `Reviewed`, `Supported`, `Rejected`
- Mapping level: `A`, `B`, `C`, `D`, `Unrated`
- Historical relation: `Established`, `Supported`, `Unestablished`, `Rejected`, `Not claimed`
- Literary layer: `Draft`, `Reviewed`, `Published`, `Not present`
- Evidence: `Established`, `Supported`, `Candidate`, `Interpretive`, `Under review`, `Unestablished`, `Rejected`, `Not claimed`, `Not evaluated`
- Confidence: `High`, `Medium`, `Low`, `Unknown`

Compatibility rule: reader-facing pages may retain older labels such as “B · Experimental” or `retain_without_promotion`; the adapter maps them to the v1 axes without rewriting the original page.

## Hypothesis object

Every hypothesis includes `hypothesis_id`, `type`, bilingual `claim`, `status`, `confidence`, `supporting_cases`, `counterexamples`, bilingual `testability`, `experiment_link` and `source_refs`. Supported types include Consonantal Semantic Skeleton, Root Consonant Semantic Stability, vowel relation and other candidate rules. A null `experiment_link` means “not linked”, not “disproved”.

## Media policy

Allowed types are `archival image`, `original manuscript`, `illustration`, `generated illustration` and `research figure`. `alt`, `caption`, `source` and `provenance` are required. When the creation method or source is not documented, use `unknown`. A generated illustration or literary visual is not research evidence merely because it appears on a published page.

## Complete compact example

The canonical AT record contains additional references and literary/media objects; this compact object shows every required field:

```json
{
  "id": "LB-en-at-003",
  "slug": "at",
  "title": { "en": "AT · 在 · 爱 | Love Is Presence", "zh-Hans": "AT · 在 · 爱｜爱在，世界就在" },
  "languages": [
    { "role": "source", "code": "en", "name": "English", "word": "at", "pronunciation": "/æt/" },
    { "role": "target", "code": "zh-Hans", "name": "Chinese", "word": "在", "pronunciation": "zài /tsaɪ̯/" }
  ],
  "primary_mapping": {
    "mapping_id": "MAP-at-primary-zai",
    "source": { "language": "English", "word": "at", "pronunciation": "/æt/" },
    "target": { "language": "Chinese", "word": "在", "pronunciation": "zài /tsaɪ̯/" },
    "gloss": { "en": "Location → presence", "zh-Hans": "位置 → 在场／存在" },
    "meaning": { "en": "Candidate semantic comparison", "zh-Hans": "候选语义比较" },
    "mapping_type": "Phonetic-semantic candidate",
    "rationale": { "en": "LOCATION → PRESENCE is the strongest relation.", "zh-Hans": "最强关系是 LOCATION → PRESENCE。" }
  },
  "entry_status": "Published",
  "mapping_status": "Candidate",
  "mapping_level": "C",
  "historical_relation_status": "Not claimed",
  "evidence": {
    "Historical": { "status": "Unestablished", "confidence": "High", "summary": { "en": "Independent histories; no common origin claimed.", "zh-Hans": "各有独立历史；不主张共同来源。" }, "items": [], "source_refs": [] },
    "Phonetic-Semantic": { "status": "Candidate", "confidence": "Low", "summary": { "en": "Limited modern-form observation.", "zh-Hans": "有限的现代词形观察。" }, "items": [], "source_refs": [] },
    "Cognitive": { "status": "Interpretive", "confidence": "Medium", "summary": { "en": "LOCATION → PRESENCE", "zh-Hans": "位置 → 在场" }, "items": [], "source_refs": [] },
    "Speculative": { "status": "Candidate", "confidence": "Low", "summary": { "en": "Untested.", "zh-Hans": "尚未测试。" }, "items": [], "source_refs": [] }
  },
  "phonetic_observation": [],
  "semantic_structure": { "concepts": ["LOCATION", "PRESENCE"], "relation": "LOCATION → PRESENCE", "status": "Interpretive" },
  "hypotheses": [],
  "experiments": [],
  "literary_layer": {
    "status": "Published",
    "is_historical_evidence": false,
    "proposition": { "en": "Love is presence. Where love is, the world is.", "zh-Hans": "爱就是在。爱在，世界就在。" },
    "essay_prose": [], "poem_lyrics": [], "translations": [], "archival_manuscript_media": [],
    "evidence_boundary": { "en": "在 → 爱 is literature, not etymology.", "zh-Hans": "在 → 爱属于文学，不属于词源。" }
  },
  "media": [],
  "references": [],
  "author": "Jinkai Liu",
  "version": "1.0",
  "dates": { "created": null, "modified": "2026-08-30", "published": "2026-08-30" },
  "editorial_notes": [{ "en": "The status axes are independent.", "zh-Hans": "各状态轴彼此独立。" }],
  "search_terms": ["at", "在", "爱", "love", "presence"],
  "page": "words/at.html",
  "legacy": null
}
```

## Editing and validation workflow

Edit the migration source or a future data-authoring source, regenerate v1.0, run `validate-language-book-v1.mjs`, then run the browser-data tests and link checks. Do not bypass a failed validator. Missing evidence, dates or provenance remain `null`, `unknown`, `Not claimed` or `Not evaluated` as appropriate.

## Dataset Expansion v1 workflow

New reviewed records are authored one per file in `data/entries/*.v1.json`; `scripts/build-language-book-v1.mjs` sorts and merges them after the deterministic legacy migration. Each record must preserve the author's source/raw note, independently verify historical claims, use the standard lexical equivalent as `primary_mapping`, and keep phonetic observations, hypotheses and literature on their own status tracks. Search aliases belong in `search_terms`, including accented and unaccented forms when both are reasonable queries.

When a record compares look-alike words from different histories, use `related_words` to state the relationship explicitly. A direct derivative can be labeled `Etymological derivative`; a shape, spelling or literary link must be labeled `Speculative semantic association` (or an equally explicit non-historical type). Similar spelling alone never creates an etymological family. The Mapper renders the two categories separately.

Before adding a batch: compare stable IDs/slugs, leave unknown fields null or unclaimed, run the builder and validator, test English plus at least one non-English alias through `UnilanguageData.lookup`, run the full test suite and link checker, then regenerate the product manifest/checksums. A record may be Reviewed without being Published; a literary proposition may be Reviewed while its phonetic hypothesis remains Candidate or Unestablished.

### Active Association revision 1.2.12

The existing `related_words` typed relationship, `semantic_associations` (`is_etymological: false`), Cognitive track and bilingual `editorial_notes` encode the Active Association Principle. Schema remains 1.0.0; count remains 37. No new entry, adapter metadata or frozen UI changes are needed. Query-specific `primary_mapping.meaning` labels distinguish “Active association: bash” from “Cross-language candidate: 拍”. Historical evidence and the obsolete bash homograph remain independent. Author: Jinkai Liu.

ABASH/BASH status correction: optional `translation_status` and `translation_source_refs` apply to the standard meaning in each query view; `mapping_status` and `mapping_level` continue to grade only bash ↔ 拍. `active_association_status` is independent of `historical_relation_status`. Query views retain the same record ID and count. The renderer uses these optional fields only on this record; other records retain their existing presentation.


### ABASH / BASH separate candidates · release 1.2.14

The existing 37 records and stable ABASH ID remain unchanged. A query view may now carry a schema-validated featured_mapping for its own source, while primary_mapping retains the Supported standard translation. ABASH/怕 resolves to abash ↔ 怕 pà (psychological, low-to-moderate semantic strength); BASH/拍 retains bash ↔ 拍 pāi. Both are Candidate / C / Low, Historical Relation Not claimed. The existing Active Association link is Cognitive / Pedagogical · NOT etymology.

Translation Status ≠ Mapping Status ≠ Historical Relation ≠ Active Association Status ≠ Experiment Status.

Etymological-family deduplication｜词源家族去重: the existing experiment plan holds two candidate_observations with distinct IDs and a family_independence gate. Pending verification means eligible_as_two_independent_observations=false. Shared spelling or an Active Association does not establish family identity or independence. Count one root family once; retain positives, negatives, random/matched controls and sensitivity analyses. The medial /b/ in abash must not be silently treated as word-initial /b/. No experiment has run and no evidence level is promoted. Original raw notes and literary content are preserved.


### ABANDON meaning-first correction (dataset 1.2.17)

Semantic Path studies how one word’s historical meanings develop. Consonant Group Mapping compares an English/French historical word or root with the pronunciation of a semantically selected Chinese candidate. These are independent analyses. The b-p-m-f, d-t-n-l, g-k-h and z-c-s groups predict cross-language sound correspondences, not membership of every Chinese word in one path.

Semantic Path 研究一个词的历史语义怎样发展。Consonant Group Mapping 比较英／法历史词或词根与按语义选出的汉语候选词读音。两者独立；b-p-m-f、d-t-n-l、g-k-h、z-c-s 等组预测跨语言声音对应，不要求汉语路径中的每个字属于同组。

`diachronic_semantic_mapping` stores only historical stages and semantic development. `consonant_group_mapping` stores paired source/target units, meaning-first review, consonant comparison, independent scores, evidence and seven-step workflow. The schema rejects the removed Chinese path-constraint fields.

Dialect & Diachronic Chinese Evidence is a separate evidence track. Record dialect/location/date and transcription separately from Middle Chinese or Old Chinese reconstructions, with named sources and reconstruction systems. Modern dialects are not frozen copies of ancient Chinese. No dialect or reconstructed pronunciation is used to raise this candidate’s grade; these comparisons remain pending.

Dialect & Diachronic Chinese Evidence｜汉语方言与历时音韵证据独立记录。方言须注明地点、年代与转写；中古音／上古音须注明来源与重构体系，不能混用。现代方言不是古汉语的冻结复制品。本候选未以方言或重构音提高等级，相关比较待核验。

### Citation labels and metadata

Follow [Citation / Lexicographic Source Policy](standards/evidence-and-references.md#citation--lexicographic-source-policy). Existing reference `title` contains the specific work name; `url` identifies access; `provenance` records known edition, institution, access date and any original-work/platform distinction. Preserve unknowns and stable reference IDs. Schema remains 1.0.0.