Entry · 词条
Draft / Reviewed / Published describes editorial availability only.
Language Book Entry Schema v1.0
Data Foundation
A cross-language comparable semantic database with independently evaluated layers. · 各层独立评估的跨语言可比较语义数据库。
Schema records · 统一词条
13 unified records, including lexical records and the Namcha Barwa Named Entity / Literary Entry.
Separate review queue · 独立审核队列
19 条 Package E/F/G 候选审核队列继续独立保存，不因 schema 迁移自动晋级。
Schema v1.0 · Four independent status axes
Mapping / Hypothesis · 映射／假说
Candidate / Reviewed / Supported / Rejected and A–D level are graded independently.
Evidence / History · 证据／历史
Historical, Phonetic-Semantic, Cognitive and Speculative tracks each carry their own status and confidence.
Literature · 文学
Prose, poetry, translations and manuscripts remain publishable and searchable, but are never historical evidence.
Editorial policy · 编辑原则： An entry may be published; a hypothesis must be graded; literature may explore freely; evidence must be evaluated independently. · 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。
Read schema documentation · 阅读 Schema 文档 → Open canonical JSON · 打开统一 JSON →
Lifecycle · 数据生命周期
1 · Candidate · 候选Sources prepared; review pending. Calibration candidates may be displayed only with a visible Candidate identity.
2 · Reviewed / Rejected / Needs EvidenceA named reviewer checks sources, boundaries, alternatives, conflicts and status.
3 · Published · 已发布A separate release gate admits a validated record to the published evidence layer.
Promotion rule · 晋级规则： AI extraction or source preparation alone cannot raise confidence, review status, experiment status or publication status. The 19-record queue remains outside Mapper; WATER, LANGUAGE, ADVANCE and LIGHT appear only as explicitly labelled Candidates. ADVANCE and LIGHT experimental validation are Not Tested; LIGHT is retain_without_promotion. · 单凭 AI 提取或来源整理不能提升置信度、审核状态、实验状态或发布状态；19 条队列仍不进入 Mapper，WATER、LANGUAGE、ADVANCE 与 LIGHT 仅以明确标记的 Candidate 身份展示；ADVANCE 与 LIGHT 的实验验证均为 Not Tested，LIGHT 为 retain_without_promotion。
Package F · Fixed review queue
Canonical published layer · 正式发布层
Unchanged: 6 entries
Package F does not alter Package E’s published dataset or frozen experiment assets.
Candidate review foundation · 候选审核基础
Exactly 19 records, each with source locators, support/non-support boundaries, four evidence tracks, counterevidence and a human checklist.
Open Package F review queue · 打开审核队列 → Open published Dictionary · 查看正式词典 →
Package G · Decision and publication gates
Formal human decisions · 正式人工决策
19 Pending · 0 Reviewed
Needs Evidence, Rejected, and Reviewed require named reviewers, explicit reasons, evidence checks, and append-only audit events.
Publication eligibility · 发布资格
19 Not Eligible · 0 Published
Reviewed is necessary but never sufficient for publication. A separate release checklist and explicit human publisher approval are required.
Package G.1 · Seven independent layers
Mapping is not etymology · 映射不等于词源
Each English word has exactly one primary Chinese mapping. Author rationale, historical lineage and unclassified notes are separate objects.
Hypothesis is not result · 假说不等于结果
Sound, consonant and letter-symbol hypotheses carry their own status/evidence/confidence. Experiments cite hypotheses and retain their tested condition and result.
Review per object · 逐对象审核
Source Verification and AI Review are separate for every mapping, rationale, etymology, hypothesis, note and experiment.
WATER calibration · WATER 校准 → LANGUAGE calibration · LANGUAGE 校准 → ADVANCE calibration · ADVANCE 校准 → LIGHT Candidate · LIGHT 候选 →
Package G.2 · Calibration source boundaries
Supported history · 有来源的历史
LANGUAGE's lexical lineage is source-backed and remains separate from the author-proposed 朗 mapping.
Disputed sound statement · 有争议的声音表述
The W↔F/H idea is preserved, but official phonetic classifications do not support calling W and F a voicing pair.
Method before decision · 先方法，后结论
The object-level AI rubric is active as a method only. No calibration object has received an AI review decision.
Package G.3 · ADVANCE boundary
Primary research mapping · 主要研究映射
advance ↔ 往 remains an author-proposed Candidate; 推进 and 前进 remain secondary lexical mappings.
Two analyses, side by side · 两套分析并列
Author Analysis preserves ad.van.ce / ad.vance.ment and the vance observation. Source-backed etymology instead records avancer → *abantiāre → abante and avancer + -ment.
Future validation · 未来验证
van/vance ↔ wang/往 is Untested, low-confidence and has a separate Not Tested experimental plan with no fabricated result.
Package G.4 · LIGHT → 籁 boundary
Status lock · 状态锁
The LIGHT entry and literary layer are Published while light ↔ 籁 remains Candidate / retain_without_promotion.
Independent history · 独立历史层
English Germanic/PIE, French/Latin and Chinese 籟 histories are stored independently; no English–Chinese cognacy is claimed.
Deduplicated hypotheses · 去重假说
Four European surface positives collapse to one deep PIE *leuk- family. Cross-modal and L-Light hypotheses remain Untested/low; the experiment remains Not Tested.