Open protocol · RFC stage · spec v0.3 · canon v15

Unicode for meaning.

NeuroGlyph is a semantic substrate for human ↔ machine communication. Sixty-five universal atoms of meaning — the same graph in any language, for any agent. The words you see are just a render.
Not a new language to learn — a semantic layer every message can carry.

Early-stage protocol (RFC). The spec, registry and grammar below are real and public. A “type anything” encoder is planned for Phase 4 — until then the demo runs on curated phrases.

Explore the atoms Watch the demo
65
semantic atoms
12
semantic macros
1
graph per meaning
languages render
The problem

Words travel. Meaning drifts.

Every translation is a negotiation. Four hops and the sentence is intact — while the image, the colour, the specific ache have quietly disappeared.

drift 0
English
“He’s feeling blue today.”
+1
Русский
«Он сегодня грустит.»
+2
中文
「他今天心情很沉重。」
+3
Deutsch
“Ihm geht es heute nicht gut.”
meaning preserved
meaning lost

The colour, the image, the weight of the mood — gone. Only the gist survives.

ŋ
one glyph graph
English
feeling blue
Русский
грустит
中文
沉重
Deutsch
nicht gut

NeuroGlyph carries the meaning once. Every language is a render, not a translation.

The vocabulary

Sixty-five primes. Everything decomposes.

After fifty years of cross-linguistic research, a universal inventory of semantic atoms has been identified — meanings that every human language encodes and that cannot be defined any further. Every idea, in any language, is a combination of these 65 atoms. (Chart of NSM Semantic Primes v15)

Hover an atom to see it render. Core atoms already carry RU / EN / ZH — the rest arrive with the full lexicon.

How it works

One graph. Any language.

The meaning is parsed into a language-neutral graph of primes. The words you read are a render — a font for meaning, applied at the last step.

01 · source
“I want this.”
a sentence in any human language
02 · parse
IWANTTHIS
semantic parsing into universal concepts — the words of the source are not carried over
03 · glyph graph
Arg0 Arg1 WANT I THIS
one graph, identical in every language
04 · render
я хочу этоI want thisこれが欲しい
the graph renders into any language — losslessly, in any direction
The demo

Watch meaning travel.

Four real phrases, encoded into glyph graphs by the tracer. Click one, render it into another language — the meaning does not change.

macro (compositional alias) prime coreference
A cultural macro renders as its nearest equivalent in each language — TOSKA → тоска / saudade / 怅惘.
The stack

From source to screen.

A message in NeuroGlyph is a small stack — each layer adds one kind of information, from what was meant to how it is shown.

L0
Source
the original message — text or speech
L1
Semantic graph
concepts and their relations
L2
Features
diacritics — valence, power, aspect, evidentials
L3
Pragmatics
intent and mode: assert, request, emote…
L4
Envelope
tone, register, urgency, politeness
L5
Provenance
source hash, annotator, confidence, glosses
L6
Render
the words your language shows
L7
Visual
glyphs your eye can read at a glance
Open

Built in the open.

The schema, the registry and the grammar are free and versioned. NeuroGlyph is a public good — a protocol, not a product.

Get involved

The most useful contribution right now is a render: a native speaker’s reading of a prime or macro in your language, proposed through the RFC process. Every render feeds the lexicon and tests the protocol.