Download README.md from Lilambd/world-languages: direct link, hf CLI and curl.
- Browser
- Download file 6.07 kB
-
https://huggingface.co/datasets/Lilambd/world-languages/resolve/main/README.md
- Command line
-
hf download hf://datasets/Lilambd/world-languages/README.md
-
curl -L -o README.md https://huggingface.co/datasets/Lilambd/world-languages/resolve/main/README.md
license: cc-by-4.0
pretty_name: World Languages and Scripts
language:
- en
- ja
tags:
- linguistics
- languages
- glottolog
- cldr
- iso-639-3
- writing-systems
- endangered-languages
- tabular
size_categories:
- 10K<n<100K
configs:
- config_name: languages
data_files: data/languages.parquet
default: true
- config_name: language_country
data_files: data/language_country.parquet
- config_name: language_script
data_files: data/language_script.parquet
- config_name: scripts
data_files: data/scripts.parquet
- config_name: families
data_files: data/families.parquet
- config_name: languoid_tree
data_files: data/languoid_tree.parquet
- config_name: wals_values
data_files: data/wals_values.parquet
task_categories:
- other
World Languages and Scripts
Relational tables of languages, scripts and geographic identifiers. Sources include Glottolog, Unicode CLDR and UCD, ISO 639-3 and WALS. Update cadence: 2026-11-03 snapshot; no refresh interval stated. The dataset card lists CC BY 4.0 as its license. Keep the component-source terms.
One relational dataset of the world's languages and writing systems. It joins Glottolog, Unicode CLDR, ISO 639-3, WALS and the Unicode Character Database on shared keys (Glottocode, ISO 639-3, ISO 3166 alpha-2, ISO 15924). Every row says which source it came from. No value is filled in by guesswork: when a source has no value, the cell is empty.
Data as of 2026-11-03. Built by Lilambd.
What is inside
| Table | Rows | One row is |
|---|---|---|
languages |
8,925 | a language (8,618 Glottolog language-level languoids + 307 ISO 639-3 individual codes Glottolog does not map) |
language_country |
12,858 | a language in a country (Glottolog country list, plus CLDR share of territory population and official status) |
language_script |
1,001 | a language written in a script (CLDR, primary or secondary) |
scripts |
223 | an ISO 15924 script, with Unicode 17.0 character count, direction, origin country, and summed speakers |
families |
429 | a top-level family or isolate, with counts of endangered and extinct languages |
languoid_tree |
27,177 | a family, language or dialect node with its parent and full Glottolog classification path |
wals_values |
76,475 | one WALS typological feature value for one language |
Headline counts measured from the tables: 6,969 living languages by ISO type; 4,172 languages Glottolog rates as threatened to nearly extinct, and 1,239 as extinct; 238 genealogical families and 183 isolates; 171 scripts encoded in Unicode 17.0 (excluding Common and Inherited).
Key columns of languages
lang_id(Glottocode, oriso-xxxwhen only ISO has the code),glottocode,iso639_3,iso639_1,iso639_2bname_en,name_ja(CLDR Japanese name),autonym(the language's name in its own CLDR locale)iso_type,iso_scope,glottolog_category(spoken L1, sign, pidgin, ...)family_glottocode,family_name,parent_glottocode,tree_depth,is_isolatemacroarea,lat,lon,countries,n_countriesaes_level1–6 andaes_label(Glottolog Agglomerated Endangerment Status, which combines EGIDS, UNESCO and ElCat)description_level(Glottolog most extensive description: long grammar ... wordlist)scripts,cldr_code,cldr_locale,cldr_coveragespeakers_cldr_l1l2(sum over territories of CLDR population × population percent; includes second-language users),speakers_ohchr(OHCHR UDHR page estimate),speakers_best,speakers_sourcegdp_usd_reach_cldr(sum over territories of CLDR GDP (PPP) × population percent),n_official_countriesudhr_translations,wals_codes,wals_feature_values,wals_word_ordersources,data_as_of
Known limits
- Speaker counts exist for 897 of 8,925 languages. CLDR covers about 750 languages; OHCHR fills about 150 more. Ethnologue is not used (paid). The rest are empty, not zero.
- CLDR percentages count everyone who uses the language, so
speakers_cldr_l1l2is not a first-language count. - Glottolog groups some languoids in pseudo-families (Bookkeeping, Unclassifiable, Sign Language, Pidgin, Unattested, Mixed Language, Artificial Language, Speech Register). Filter them out of family statistics if you want genealogical families only.
- CLDR maps macrolanguages to one individual language (for example
zh→ Mandarincmn,ar→ Standard Arabicarb). - 621 languages have no coordinates: 314 Glottolog languoids that Glottolog itself gives no point for, and the 307 ISO-only codes.
language_countrykeeps CLDR territories whose codes are exceptionally reserved in ISO 3166 (AC, CP, CQ, DG, EA, IC, TA) and the user-assigned XK (Kosovo); thecountry_code_notecolumn marks them, and theircountry_iso3is null because ISO 3166-1 alpha-3 has no code for them.- 5 rows of
language_scripthave a nulllang_idbecause CLDR lists codes outside ISO 639-3 individual languages (isv Interslavic, tok Toki Pona, kro Kru, xdq Kaitag). - Wikidata is not joined in this release.
Sources and licences
- Glottolog 5 CLDF, glottolog/glottolog-cldf — CC BY 4.0. Hammarström, Harald & Forkel, Robert & Haspelmath, Martin & Bank, Sebastian. Glottolog. Max Planck Institute for Evolutionary Anthropology. https://glottolog.org
- WALS Online CLDF, cldf-datasets/wals — CC BY 4.0. Dryer, Matthew S. & Haspelmath, Martin (eds.). The World Atlas of Language Structures Online. https://wals.info
- Unicode CLDR (cldr-json) and Unicode Character Database 17.0 — Unicode License v3. https://cldr.unicode.org
- ISO 639-3 code table — SIL International, ISO 639-3 Registration Authority (via the MIT-licensed npm package iso-639-3). https://iso639-3.sil.org
- ISO 15924 code list via the MIT-licensed npm package iso-15924.
- OHCHR UDHR speaker estimates via the MIT-licensed npm package speakers; UDHR translation counts via the MIT-licensed npm package udhr.
This dataset is released under CC BY 4.0. Please credit "Lilambd, World Languages and Scripts" and the upstream sources above.