Research Registry · 研究登记
Hypothesis Lab
假说实验室 · Engineering uncertainty into testable structures
Version 1.1 · Methodology and Three Active Records · 方法论与三条正式记录 · 2026-08-18
Purpose
Hypothesis Lab registers hypotheses that extend beyond their existing historical or etymological evidence: testable but unconfirmed sound–meaning regularities, and high-risk deep-history proposals concerning cross-civilizational transmission, remote diffusion or common symbolic origins. A claim may enter before it is established, but its evidence, limits, review state and falsification test must remain visible.
目的
假说实验室收纳已有部分历史或词源证据、但进一步延伸的假说；可检验但尚未确认的声音—语义规律；以及关于跨文明传播、远古扩散或符号同源的高风险深历史假说。一个主张可以在尚未确立时进入系统，但其证据、边界、评审状态与证伪条件必须始终清楚可见。
01
Operating Principles · 运行原则
Homophony is evidence to investigate, not evidence to conclude.
同音是值得调查的证据，不是可以直接下结论的证据。
A hypothesis does not need to be established before it can be tested computationally. Its status must be explicit.
一个假说不必先成为定论，才有资格进入计算和工程测试；但其证据状态必须明确。
Unilanguage does not wait for certainty; it engineers uncertainty into testable structures.
Unilanguage 不等待确定性，而是把不确定性工程化为可检验的结构。
02
Four Evidence Tracks · 四条证据轨道
Track 1
Historical Relation · 历史关系
Inheritance, borrowing, attested transmission, dated forms and documented sound change.
继承、借用、有记录的传播、可断代形式与有文献的音变。
Track 2
Phonetic-Semantic Tendency · 语音—语义倾向
A sound–meaning association tested against controlled, genealogical and chance baselines.
把语音—语义关联与受控、谱系及随机基线进行比较。
Track 3
Cognitive Association · 认知关联
Embodiment, perception, spatial schema or conceptual metaphor that may motivate a mapping.
可能促成映射的具身、感知、空间图式或概念隐喻。
Track 4
Speculative Deep-History Hypothesis · 远古历史假说
A distant proposed relation that must be distinguishable from convergence, borrowing and chance.
必须能够与趋同、借用和巧合相区分的远古关系提案。
03
Hypothesis Status · 假说状态
01
Established facts · 已确立事实
Documented observations or historical relations accepted independently of the larger hypothesis.
不依赖更大假说、已有文献支持或独立确认的观察与历史关系。
02
Supported interpretation · 有支持的解释
An interpretation supported by converging evidence, while alternatives and limits remain visible.
获得多项证据支持、但仍保留替代解释与适用边界的解释。
03
Experimental hypothesis · 实验性假说
A specified claim ready for computational or engineering tests but not yet confirmed.
已经明确陈述、可进入计算或工程测试、但尚未确认的主张。
04
Speculative extension · 推测性延伸
A higher-risk extension beyond the evidence currently supporting the core claim.
超出现有核心证据范围、风险更高的延伸主张。
05
Falsification test · 证伪检验
The observations or results that would require the claim to be narrowed, downgraded, split or rejected.
能够促使主张缩小、降级、拆分或否定的观察与结果。
04
Initial Categories · 首批分类
Sound Hypotheses · 声音假说
M / M–N / Vowels / GKH / V
Controlled tests of recurring sound–meaning tendencies, including M—measurement—moon, M–N correspondence, vowel directionality, the G/K/H comparison group and V-related transformation proposals.
对反复出现的音义倾向进行受控检验，包括 M—measurement—moon、M–N 对应、元音方向性、G／K／H 比较组与 V 相关转换提案。
Script Hypotheses · 文字假说
Form, component and writing function · 字形、部件与书写功能
Tests of graphic form, phonetic components and possible semantic extensions without treating visual similarity as proof.
检验字形、声符及可能的语义延伸，但不把视觉相似当作证明。
Deep-History Hypotheses · 深历史假说
Sound before attested writing · 现存文字以前的声音
Proposals about sound–meaning layers older than surviving scripts, with explicit null models and deep-time limitations.
研究早于现存文字的音义层，同时明确零模型与深时限制。
Cognitive Hypotheses · 认知假说
Embodiment and conceptual structure · 具身与概念结构
Cross-language patterns that may arise from shared perception, action, space or metaphor rather than common ancestry.
可能源于共同感知、行动、空间或隐喻，而非共同祖源的跨语言模式。
Transformation Hypotheses · 转换假说
Meaning across representational forms · 跨表达形态的意义
Tests of which semantic relations survive transformation among word, sound, script, geometry, image and motion.
检验意义在词、声音、文字、几何、图像与运动之间转换时，哪些语义关系得以保留。
Reserved Sound Records · 预留声音记录
Experimental · 实验性
UNI-SOUND-GKH-001
A reserved comparison record for possible G / K / H sound–meaning tendencies. No rule is asserted or confirmed at this stage.
为可能的 G／K／H 音义倾向预留的比较记录；现阶段不主张、也不确认任何规则。
Experimental · 实验性
Experimental · 实验性
UNI-SOUND-V-001
A reserved record for a possible V-related sound–meaning or transformation pattern. It remains an untested research prompt.
为可能的 V 相关音义或转换模式预留的记录；目前仅是尚未测试的研究问题。
Experimental · 实验性
05
UNI-SOUND-M-001 · M—Measurement—Moon
Known Evidence · 已知证据
Supported Core · 有支持的核心
moon ↔ month
English moon and month belong to a documented inherited Germanic word family. Historical dictionary evidence connects Old English mōna “moon” with mōnath “month,” reflecting the long-standing use of lunar cycles to organize time.
Historical Relation · Established / Supported
Reference: Merriam-Webster dictionary histories for moon and month. The reference supports the lexical relationship; it does not support the larger M semantic-field claim.
Experimental Extension · 实验性延伸
M → measure / meter / mete / mode / meal / mile / medicine?
实验性主张是：M 首音可能反复呈现与测量、循环、数量、边界或时间相关的语义倾向。measure、meter、mete、mode、meal、mile、medicine 等词只能作为待检验的候选样本；它们与 moon／month 的有据历史关系必须严格分开，也不能自行证明共同来源或普遍语义规律。
Phonetic-Semantic Tendency · 语音—语义倾向 · Experimental
Required Research Record · 必需研究记录
| Hypothesis ID · 假说编号 | UNI-SOUND-M-001 |
|---|---|
| Claim · 主张 | M-initial sounds may show recurring semantic tendencies involving measurement, cycle, quantity, boundary, or time. · M 首音可能反复呈现与测量、循环、数量、边界或时间相关的语义倾向。 |
| Author · 作者 | Jinkai Liu｜劉金凱 |
| Origin or System · 来源或系统 | Collanguage / Unilanguage |
| Theoretical Support · 理论支持 | Established moon–month lexical history supports the core only. Articulatory iconicity and sound-symbolic clustering are candidate mechanisms for the extension, not historical proof. · moon—month 的已知词汇史只支持核心；发音象似与语音象征聚类只是延伸假说的候选机制，不是历史证明。 |
| Supporting Cases · 支持案例 | Supported core: English moon / month and documented related historical forms. Experimental extension candidates: measure, meter, mete, mode, meal, mile and medicine, plus pre-registered multilingual M lists. These candidates have varied histories and are not treated as one established family. · 有支持的核心：英语 moon／month 及有文献的相关历史形式；实验性延伸候选：measure、meter、mete、mode、meal、mile、medicine，以及预注册的多语言 M 词表。这些候选具有不同历史，不被视为一个已确立的词族。 |
| Counterexamples · 反例 | Many M-initial words fall outside the proposed fields; many target-field words do not begin with M; genealogy, borrowing, phonotactics and researcher-selected categories can create artificial clusters. · 大量 M 首音词不在拟议意义场内，大量目标意义词也不以 M 开头；谱系、借用、音系配列与研究者自选分类都可能制造虚假聚类。 |
| Test Method · 测试方法 | Pre-register semantic fields and sampling rules; compare M-initial vocabulary with matched non-M phonemes across balanced languages and corpora; control for cognate families, borrowing, phonotactic frequency and multiple comparisons; then test replication with independent semantic coding. · 预注册语义场与抽样规则；在均衡语言和语料中，把 M 首音词与匹配的非 M 音位比较；控制同源词族、借用、音系频率及多重比较，再由独立语义编码检验复现性。 |
| Mapping Type · 映射类型 | Historical Relation (core) + Phonetic-Semantic Tendency (extension) · 历史关系（核心）＋音义倾向（延伸） |
| Evidence Status · 证据状态 | Supported Core + Experimental Extension · 有支持的核心＋实验性延伸 |
| Review Status · 评审状态 | Core documented; extension author-proposed; corpus design, external review and replication pending. · 核心已有文献；延伸由作者提出；语料设计、外部评审与复现待完成。 |
| Falsification Test · 证伪检验 | If M-initial words are not significantly concentrated in the measure / cycle / quantity / boundary / time semantic fields, or if an apparent concentration is mainly produced by a few cognate families, the hypothesis must be downgraded, split or rejected. Failure against matched phoneme, borrowing and multiple-comparison baselines, or failure of independent replication, also counts against the extension. · 如果 M 首音词并不显著集中于测量／循环／数量／边界／时间语义场，或表面集中主要由少数同源词族造成，则假说必须降级、拆分或否定；若不能超过匹配音位、借用和多重比较基线，或无法独立复现，也构成对延伸假说的不利证据。 |
06
UNI-SOUND-MN-001 · M–N Correspondence Hypothesis · M–N 对应假说
Primary observation · 核心观察
man /mæn/ ↔ 男 nán
The forms show a strong semantic overlap and a compact syllable resemblance. The proposed initial m ↔ n relation is experimental; no historical relationship is inferred.
Secondary observation · 辅助观察
man ↔ 氓 méng / máng
词首 m、近似元音与词尾鼻音 n↔ŋ 构成可检验的结构相似；但表示“百姓”的读音主要是 méng，更近音的 máng 主要见于“流氓”。这项错位必须计入反证与限制。
| Hypothesis ID · 假说编号 | UNI-SOUND-MN-001 |
|---|---|
| Title · 标题 | M–N Correspondence Hypothesis｜M–N 对应假说 |
| Author · 作者 | Jinkai Liu｜劉金凱 |
| System · 系统 | Collanguage / Unilanguage |
| Claim · 主张 | In specified cross-language comparison sets, bilabial nasal /m/ and coronal nasal /n/ may participate in recurring phonetic-semantic correspondences more often than matched baselines predict. This is a statistical research question, not a historical sound law. · 在预先规定的跨语言比较集中，双唇鼻音 /m/ 与舌冠鼻音 /n/ 可能以高于匹配基线的频率参与反复出现的音义对应。这是统计研究问题，不是历史音变定律。 |
| Supporting Cases · 支持案例 | Primary candidate: man /mæn/ ↔ 男 nán, with high semantic overlap and whole-syllable resemblance. Secondary candidate: man ↔ 氓 méng / máng, where sound and sense support different readings and must remain separated. Future cases must be pre-registered rather than selected after inspection. · 主候选为 man↔男；辅助候选为 man↔氓，但其音义支持落在不同读音上，必须分开。未来案例须预注册，不能事后挑选。 |
| Counterexamples · 反例 | Most m- and n-initial words do not share person/male meanings; many person/male words have neither sound; similar short nasal syllables are common; unrelated languages can converge by chance. The méng/máng split in 氓 weakens any unqualified form-meaning match. · 多数 m／n 首音词不共享“人／男性”义，大量“人／男性”词也没有这些声音；短鼻音音节常见，非亲属语言可偶然趋同；氓的 méng／máng 分读削弱任何不加限定的音义匹配。 |
| Test Method · 测试方法 | Pre-register languages, semantic domains, transcription rules, alignment criteria and exclusions. Compare m↔n matches against other nasal and non-nasal pairs while controlling genealogy, borrowing, phonotactic frequency, word length and multiple comparisons. Use independent semantic coders and a held-out replication set. · 预注册语言、语义域、转写、对齐与排除规则；以其他鼻音及非鼻音配对为对照，控制谱系、借用、音系频率、词长与多重比较，并使用独立语义编码者及留出复现集。 |
| Falsification Test · 证伪检验 | Narrow, downgrade or reject the hypothesis if m↔n pairings do not exceed matched baselines, disappear after genealogical and borrowing controls, depend on post-hoc semantic categories, or fail independent replication. · 若 m↔n 配对不能超过匹配基线、控制谱系与借用后消失、依赖事后语义分类，或不能独立复现，则缩小、降级或否定假说。 |
| Mapping Type · 映射类型 | Phonetic-Semantic Tendency · 语音—语义倾向 |
| Evidence Status · 证据状态 | Experimental · 实验性 |
| Historical Relation · 历史关系 | Unknown / Unproven · 未知／未证 |
| Review Status · 评审状态 | Registered; corpus design and independent review pending. · 已登记；语料设计与独立评审待完成。 |
Non-generalization rule · 禁止泛化规则： man ↔ 男 and a few additional examples may motivate a test; they cannot, by themselves, license universal substitution of M and N. · man ↔ 男 与少量附加案例只能发起检验，不能据此把 M 与 N 普遍互换。
07
UNI-VOWEL-DIRECTION-001 · Vowel Directionality Hypothesis · 元音方向性假说
Open jaw posture and lower-vowel acoustics may motivate perceived expansion or outwardness in some tasks. This tendency is not assumed to be universal. · 较开放的口腔姿态与低元音声学特征，可能在部分任务中促成“扩张／向外”感知，但不预设其普遍性。
High, rounded or back vowels may pattern differently by articulation, acoustics, language and task. They must not be collapsed into one “inward” class before testing. · 高元音、圆唇元音与后元音可能因发音、声学、语言和任务呈现不同模式；检验前不得统并为单一“向内”类别。
Yin–yang labels are a Collanguage cultural-philosophical interpretation of perceived direction. They are not equivalent to acoustic measurements and do not constitute established linguistic law. · 阴阳标签是 Collanguage 对感知方向的文化／哲学解释，不等同于声学测量，也不是已确立语言规律。
The proposal may be tested in articulation, acoustic space, forced-choice perception, nonce-word meaning and corpus association as distinct experiments. · 可分别在发音、声学空间、强迫选择感知、无意义词配义与语料关联中测试。
| Hypothesis ID · 假说编号 | UNI-VOWEL-DIRECTION-001 |
|---|---|
| Title · 标题 | Vowel Directionality Hypothesis｜元音方向性假说 |
| Author · 作者 | Jinkai Liu｜劉金凱 |
| System · 系统 | Collanguage / Unilanguage |
| Claim · 主张 | Vowel articulation and acoustics may bias perceived spatial direction or openness: a is proposed as an outward/open candidate, while i, u, o may show inward, closed or other directional tendencies that require separate specification. Yin–yang is a cultural interpretive layer, not the physical claim itself. · 元音的发音与声学特征可能影响对空间方向或开合的感知：a 是向外／开放候选，i、u、o 可能呈向内、闭合或其他方向倾向，须分别明确；阴阳属于文化解释层，不等于物理层主张。 |
| Supporting Cases · 支持案例 | User-origin observations: a as outward / 阳; o, u as inward / 阴; i and other vowels as possible downward, closed or contrasting directions. The man ↔ 男 case records an open-vowel resemblance but does not score yin–yang interpretation. · 用户原始观察包括 a→向外／阳，o、u→向内／阴，i 等可能表示向下、闭合或对比方向；man↔男记录开放元音相似，但阴阳解释不计分。 |
| Counterexamples · 反例 | The same vowel carries many unrelated meanings; vowel realization varies across languages and speakers; roundedness, height, backness, duration, tone, coarticulation and orthography may produce conflicting directional judgments. · 同一元音承载大量无关意义；不同语言与说话者的元音实现不同；圆唇、高低、前后、时长、声调、协同发音与拼写都可能造成冲突判断。 |
| Test Method · 测试方法 | Define acoustic and articulatory variables before cultural labels. Run preregistered perception and nonce-word experiments across languages; randomize symbols and order; compare within- and cross-cultural groups; model vowels separately and report null results. Test yin–yang labeling as a distinct cultural coding task. · 在文化标签之前定义声学与发音变量；跨语言开展预注册感知与无意义词实验，随机化符号和顺序，比较文化内外群体，分别建模各元音并报告零结果；阴阳标签另作文化编码任务。 |
| Falsification Test · 证伪检验 | Narrow or reject a directional subclaim if effects fail preregistered thresholds, reverse across replications without a predicted moderator, vanish when orthography and culture are controlled, or cannot distinguish the target vowel from matched alternatives. · 若效应达不到预注册阈值、在无预测调节因素时跨复现反转、控制拼写与文化后消失，或不能把目标元音与匹配对照区分，则缩小或否定相应方向子主张。 |
| Mapping Type · 映射类型 | Phonetic-Semantic Tendency + Cognitive / Cultural Association · 语音—语义倾向＋认知／文化关联 |
| Evidence Status · 证据状态 | Experimental / Cultural · 实验性／文化性 |
| Review Status · 评审状态 | Registered; operational definitions and experiments pending. · 已登记；操作性定义与实验待完成。 |
Layer boundary · 分层边界： A measurable articulatory or acoustic tendency would not, by itself, validate a yin–yang interpretation; a meaningful yin–yang interpretation would not, by itself, establish a universal phonetic law. · 可测量的发音／声学倾向本身不能证实阴阳解释；有意义的阴阳解释本身也不能建立普遍语音规律。
08
Review Workflow · 评审流程
Status rule · 状态规则： Computation can begin before consensus. Publication status changes only when evidence and review change. · 计算检验可以先于共识开始；只有证据与评审发生变化时，发表状态才随之改变。
09
Unilanguage Hypothesis Evaluation Framework · Unilanguage 假说评测框架
Reusable method
The framework defines hypotheses before testing; uses random or frequency sampling and controls; applies blind annotation; preserves counterexamples; deduplicates evidence families; reports effect size with 95% confidence intervals; follows transparent decision rules; and publishes negative results.
可复用方法
该框架要求检验前预先定义假说；采用随机或频率抽样并设置对照；实行盲标；保留反例；进行证据家族去重；报告效应量与 95% 置信区间；遵守透明裁决规则；并公开负结果。
Read the Evaluation Framework · 阅读评测框架 Reference case · 参考案例
Open Experiment 001A · 打开实验 001A Frozen public record · 冻结公开记录
View Full 31-Section Preregistration · 查看完整31项预注册规则
10
Relation to the Unilanguage System · 与 Unilanguage 系统的关系
Rules · 规则
Protocol Book · 协议书
Defines the mapping rules, evidence tracks and testable structures used to formulate hypotheses.
定义构成假说所需的映射规则、证据轨道与可检验结构。
Cases · 案例
Language Book · 语言书
Provides word-level observations and documented cases without automatically turning resemblance into historical relation.
提供词汇层观察与已有案例，不把相似性自动写成历史关系。
Registry · 登记
Hypothesis Lab · 假说实验室
Separates known evidence from extensions, records uncertainty and states what could falsify each claim.
分离已知证据与延伸主张，登记不确定性，并明确每项主张的证伪条件。
Evaluation · 评测
UTP Lab · 翻译实验室
Tests whether proposed semantic structures remain useful when meaning is translated, compared or transformed.
检验拟议语义结构在意义被翻译、比较或转换时是否仍然有效。