# Language atlas — sources and credits

## Names and rules

The launch set contains 787 language entries. Base English names come from
Unicode CLDR 48.0.0, extended with active named sign languages in the IANA
Language Subtag Registry dated 2026-08-08. There are 162 named sign languages.
Historical and non-fictional constructed languages count. Language families,
programming languages, fictional languages and unnamed sign languages do not.
This is a fixed game set, not a complete inventory of the world's languages.

Familiar aliases share one point. Examples include Persian/Farsi,
Bengali/Bangla, Chinese/Mandarin and Tagalog/Filipino. Cantonese is separate.
Norwegian includes Bokmål and Nynorsk. Unicode's documented language aliases
supply the other merges; near spellings are never guessed automatically.
The reviewed speech spelling "finish" also counts as Finnish, following a
player report. Current catalog: `languages-embedded-v2`; judge: `languages-exact-v2`.

- [Unicode CLDR 48.0.0](https://github.com/unicode-org/cldr-json/tree/48.0.0)
  and [supplemental language aliases](https://github.com/unicode-org/cldr/blob/release-48/common/supplemental/supplementalMetadata.xml).
  Unicode License v3: [retained licence](UNICODE-LICENSE.txt).
- [IANA Language Subtag Registry](https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry),
  under [IANA/IETF CC0 terms](https://www.iana.org/help/licensing-terms).

## What the heat means

[CLDR 48 territory-language information](https://www.unicode.org/cldr/charts/48/supplemental/territory_language_information.html)
provides approximate language-use shares for countries and territories. It
includes first- and additional-language use where available; sources differ in
age and coverage. There are estimates for 506 catalog entries. Small territories
may be absent from the map geometry. A language without an estimate still
counts and never receives an invented footprint.

The cumulative view adds each recalled language's population share in each
territory. Warmer colors mean more overlapping language use. It is not the
number or percentage of distinct people, nor a city-level speaker density map.
Inspect one recalled language to see its separate share and source estimates.
For script-specific records that may describe the same people, the largest
share is retained instead of adding them. Chinese uses CLDR's broader grouped
Chinese estimates. Country boundaries do not imply uniform use everywhere
inside them. Historical names are never painted as invented modern homelands.

When you name or select a language, the camera turns toward its largest
estimated speaker population among mapped territories. This uses population
times language-use share from the same CLDR release; manual dragging takes
priority. The combined heat and individual heat scales remain unchanged.

Geometry is [Natural Earth public-domain data](https://www.naturalearthdata.com/about/terms-of-use/)
from world-atlas 2.0.2, shared with Countries. French Guiana uses its separate
territory estimate. Borders are cartographic outlines, not claims about
sovereignty or language ownership.

## Earth imagery

**NASA Earth Observatory / Reto Stöckli.**
[Blue Marble: Next Generation, July 2004, with topography](https://science.nasa.gov/earth/earth-observatory/blue-marble-next-generation/base-topography/).
[Original 5400×2700 JPEG](https://assets.science.nasa.gov/content/dam/science/esd/eo/images/bmng/bmng-topography/july/world.topo.200407.3x5400x2700.jpg).
Used under [NASA's image-use guidelines](https://www.nasa.gov/nasa-brand-center/images-and-media/).
No NASA endorsement is implied.

The local Earth texture is a 4096×2048 WebP, resized with sharp at quality 88.
The Languages discovery card is a local renderer frame, cropped to the Earth
and resized to 1000×1000 WebP. It combines the same NASA texture, owned scene
code and the CLDR heat layer. No provider is contacted during a game; the
owned land/ocean rendering and flat map remain available if an asset or WebGL
is unavailable.

Original JPEG SHA-256:
`85a24885616097b5e06532b7d8e30fa5d749a782e70ea55971dd8539a0899206`.

## Spoken examples

Completed atlases and replays offer short human recordings for 102 languages,
with their original sentence and English translation. Recordings and text are
from Google FLEURS under CC BY 4.0. [Audio sources, licenses and individual
credits](/languages/audio/v1/ATTRIBUTION.md). Samples play only on request;
coverage has no effect on which language names count.

## Immutable launch projection

Catalog: `languages-embedded-v1`. Judge: `languages-exact-v1`.
Geography: `languages-cldr48-geography-v1`.

The source snapshot, projection recipe, reviewed aliases and executable audit
are retained in the project repository. A new source or rule requires a new
version; ordinary play does not download or regenerate the catalog.

- `public/languages/v1/earth.webp` — SHA-256 `afed1598ae134d8df7fa9bee02a190a545088421117eadcf7e002e02db37e58f`.
- `public/images/cards/canyouname-language-earth-v1.webp` — SHA-256 `e4d33634ec30151da441b2c4e841be47cf8db5c7c81b045562127aa81d877b91`.
- `src/data/languages-catalog-v1.json` — SHA-256 `d0f1e15245bb417b3c6bf67186f40be9f413e3de4451a23c7abbd617190d8f43`.
- `src/data/languages-geography-v1.json` — SHA-256 `e475ede1cbd8e4dd104f7ee1bd27fb8d27b39ec11f923cac8a5507ab8de8b407`.
- `scripts/data/languages-sources-v1.json.gz` — SHA-256 `24a4f5807d9d31a96aee99842bae5e52450805cd678f4ace5520107b0658a0ce`.

Additional camera projection: `languages-cldr48-populations-v1`.

- `src/data/languages-territory-populations-v1.json` — SHA-256 `c170d28372eeed58b4e95b63bd3c3c382c0cf12423d7c5a1e3e49fb488f9204e`.