Help the world read code in its own language.
NADA shows real, working code in a person’s own language. That requires the words to be right: natural to read, faithful to the language, and trusted by the people who speak it. That work is done by people, and we welcome your help.
Of 156 languages mapped, 130 are usable today, and only 79 are strong.
A language can be mapped long before it is good to read code in. Closing that gap, moving a language from a rough first mapping to terminology a developer trusts, is the work contributors do. One person who speaks a language well can move it forward by years.
You do not have to be a programmer.
The hardest part of this work is human judgement about language. These are the roles that move a language forward. Most people fit more than one.
Native speakers
You read your own language fluently. You can judge whether a translated term reads naturally, not just whether a dictionary allows it. No programming background required.
Linguists
You know how your language forms words, borrows, and resists borrowing. You help decide when to translate a programming term and when to leave it, and you record the reasoning for each word family.
Developers who speak the language
You read code and you read the language. You catch terms that are linguistically correct but wrong in practice, where a developer would stumble.
Reviewers
You do not start from scratch. You confirm, correct, or reject proposed terms so coverage moves from “mapped” to usable.
How a single term is decided.
Terminology is not crowd-voted and it is not left to a machine. A first mapping is proposed, an AI council screens it for nonsense and disagreement, and then people who speak the language make the final decision. The council saves reviewers’ time; it does not replace their judgement.
- 1
Terms are collected
Programming terminology is collected across ecosystems: the words that appear repeatedly in real code.
- 2
A first mapping is proposed
Each term gets a candidate translation in your language, recorded with the reasoning for the choice at the word-family level.
- 3
An AI council reviews
Several models read each proposal: one restates it, one checks for nonsense, one confirms. Terms that disagree or read awkwardly are flagged before any person reviews them.
- 4
People confirm
Native speakers and linguists confirm, correct, or reject. Human judgement is final. The council only narrows what people review.
- 5
Coverage ships free
Confirmed terms compile into the language data the free app reads. The data stays open under CC-BY for everyone.
A machine can flag a word that reads wrong. Only a person can confirm it reads right.
Linguistic decisions are recorded per word family as you go, so the reasoning behind a term is not lost and the next contributor inherits it instead of starting over.
Open data and open tools, developed in public.
The language data stays free for everyone under CC-BY; the pipeline that builds it is AGPL. Work happens on GitHub.
GitHub
github.com/nadalang — NADA’s repositories. [TO CONFIRM: some repositories are private ahead of the public announcement and will open then.]
The app
nada.build — download the free app and the VS Code add-on to see the data in use.
Contribution guide
A step-by-step guide for proposing and reviewing terms. [TO CONFIRM: link to the public CONTRIBUTING guide once repositories are open.]
Tell us which language you can help with.
One email is enough to begin. Tell us the language, whether you read code, and how much time you have. We will point you to the terms that most need a person.
We ask for your language, not your credentials.
chris@nadalang.org · we read every message.