Common Kai v1.0-beta Standard
Status: public-beta development standard, September 2026
Common Kai is being prepared as a shared artlang: a language people can use in ordinary conversation, make art with, teach, translate, and adapt to real relationships. It is not merely a list of story sounds. It must remain clear enough that a stranger can ask for help, disagree, apologize, make a plan, or say something beautiful without first becoming a lore archivist.
This document governs the v1.0-beta line. Earlier working materials remain intact in the private development history, but they are not a public standard or a public site surface.
1. What is canonical now
Kai has two public writing systems:
canonical reversible analytic encoding.
- Roman Kai RK-1 is the universal teaching and exchange spelling.
- Lumin LG-1 / PUA-1 is the full syllabic visual script. LA-1 is its
Sacred seals are meaningful visual compression, not a third everyday script. They may accompany a reading but may not conceal one.
The creator is the sole canon editor. Speakers, beta readers, translators, and teachers are invited to propose usage and identify failures; their evidence matters, but a suggestion becomes canonical only through a recorded editorial decision.
2. The pre-beta to v1.0-beta boundary
The v1 beta dictionary and private migration ledger do not relabel pre-beta working material as a finished natural lexicon. They distinguish two honest starting points:
| v1 entry kind | Meaning |
|---|---|
lexeme-candidate |
A source word carried into beta. It still needs a v1 usage audit. |
derivable-candidate |
A form mechanically produced by an earlier expansion. It is evidence of productive morphology, not automatically an independently established word. |
Every v1 record includes a semantic scope, valency/distribution placeholder, collocation evidence, derivation rule, attestation label, frequency status, and review status. Empty fields are a validation failure. A placeholder is allowed only when it says exactly what has not yet been reviewed.
3. Source of truth and portable exchange
analytic recovery.
- Roman Kai is the source spelling for learners and ordinary interchange.
- LA-1 is the canonical source representation for Lumin glyph sequence and
- PUA-1 is an optional convenience encoding for typesetting with Lumin Text.
- SVG is a presentation/export format, never the only copy of an inscription.
If a document uses PUA-1, it must provide either the Lumin Text font or an LA-1 fallback. If a document uses a seal, it must provide an analytic reading.
4. Lexical evidence rules
A form can graduate from candidate status only when its record shows:
- a bounded meaning that does not silently duplicate an existing common form;
- an actual grammatical distribution or valency frame;
- at least three natural contexts from more than one domain;
- a register and derivation analysis that a learner can recover;
- a reason to teach it rather than merely generate it;
- a source decision record.
Frequency labels are corpus measurements, not vibes. Until a reviewed corpus rank exists, a form must say that it is unranked.
5. Corpus and grammar evidence
The beta corpus records provenance for every line: source collection, source record, register, grammar focus, rights status, and review state. Do not use a line solely because it is old or aesthetically satisfying. A line becomes teaching evidence when it is grammatical, natural, intelligible, and clearly licensed for the package.
The 250-case grammar stress suite is a diagnostic harness. Each prompt tests a choice that ordinary speakers need to make: aspect, reference, negation, questions, modality, relation, register, repair, causality, or technical clarity. It is not a set of magical correct sentences. Beta annotations should record where a prompt is ambiguous, culturally narrow, or unnatural.
6. Versioning
| Version class | May change | Must not change |
|---|---|---|
| Patch | typos, broken links, examples, metadata | canonical grammar or a published PUA assignment |
| Minor | accepted words, clarified distribution, new corpus or lessons | the recoverable reading of published material |
| Major | grammar defaults, encoding policy, register boundary | historical source files and migration records |
No released form is silently deleted. It can be deprecated, restricted to a register, or superseded with a migration explanation.
7. What “1.0” will mean
The beta is usable for feedback but is not a claim that Kai has already earned the stability of a mature community language. The release gates live at /reference/v1-release-gates.html. They require manual lexical audit, natural corpus work, grammar stress review, digital accessibility, human phonetic review, and a recorded creator decision before a final 1.0 tag.
8. Beta download package
The static Kai site publishes the current machine-readable beta package without requiring an account:
- /data/kai-v1-beta-dictionary.csv
- /data/kai-v1-beta-corpus.csv
- /data/kai-v1-beta-grammar-stress-suite.csv
- /data/lumin-pua-1.json
- /data/kai-v1-beta-manifest.json
These files describe a beta baseline, not a completed final language release.