Data and sources

Download the dataset

CSV (spreadsheet)   JSON

201 records × 29 fields · generated 2026-09-10 · UTF-8. Compilation under CC BY 4.0: cite “Atlas of Languages (languages.ribeiros.click)” and the primary sources below.

What each record contains

id

fixed language number in the atlas (1–100 = position in the 2020 infographic; 101–201 = additional)

position_2026

current position by total speakers

total_speakers / total_year

total speakers and year of the figure (2026 or 2020)

native_speakers / native_year

native (L1) speakers and year; empty when no source reports it

total_2020_infographic / native_2020_infographic

the original infographic figures, for comparison

family_*, branch_pt

simplified didactic classification

countries

main countries, compiled by the atlas

script, origin, period

writing system, region and period of origin (100 languages with a history)

phrase, phrase_romanization, phrase_approximate

the phrase “Hello, good morning. Are you well?”, transliteration and flag for unreviewed translation

audio, audio_note, audio_url

audio origin (azure, mms or none) and remarks

Primary sources

1. Ethnologue 200 (2026), SIL International
   Public list of the 200 most spoken languages, with rounded totals. Source of the 2026 totals for 162 languages and of the selection of languages 101–201.
2. Wikipedia (en): List of languages by total number of speakers
   Tables compiled from Ethnologue (29th ed., 2026) with the L1/L2 split. Source of the 2026 totals and native figures for the 37 largest languages.
3. Wikipedia (en): List of languages by number of native speakers
   Cross-check of native-speaker values.
4. WordTips: The 100 Most Spoken Languages in the World (2020)
   The infographic that gave rise to the atlas. Its 100 entries, with 2020 totals and native figures, were fully transcribed and kept as the comparison baseline. It cites Ethnologue (22nd ed.) as its source.
5. Visual Capitalist (15 Feb 2020)
   Dated republication of the infographic, used to fix the 2020 reference date.
6. Glottolog
   Reference for the family and branch classification, simplified for navigation. Vietnamese and Khmer under Austroasiatic; pidgins and creoles in a separate contact-language group; Mande kept under Niger-Congo by convention.
7. ISO 639-3 (SIL)
   Three-letter language codes, used to identify varieties and match sources.
8. Ethnologue, Wikipedia and general encyclopaedias
   Basis for the short histories of the 100 original languages (origin, period, script, milestones) and the country lists. Dates are approximate; texts written for the atlas.
9. Azure AI Speech (Microsoft)
   Neural voices used to generate 122 audio files; in some cases the voice of the closest standard language (indicated).
10. Massively Multilingual Speech — MMS (Meta AI)
    Open model (CC BY-NC) used for 40 languages with no commercial voice. Lower quality, flagged on the site.
11. D3.js
    Visualisation library (ISC licence).

How the data were combined

1. The 100 entries of the 2020 infographic were transcribed and kept as total_2020_infographic and native_2020_infographic.
2. For each language, the 2026 total comes from the Wikipedia compilation (L1/L2) when available; otherwise from the Ethnologue 200 list (rounded). Two languages with no 2026 entry — Filipino and South Levantine Arabic — keep their 2020 value, with the year shown.
3. Native speakers: the 2026 value when Wikipedia reports L1; otherwise the 2020 infographic value, marked with its year. When no source reports it, the field is empty and the site shows “not reported”.
4. “Additional” = total − native, computed only when both values come from the same set of sources.
5. The displayed position (position_2026) orders the 201 languages by the most recent total; the id is fixed and identifies each record.
6. Families and branches follow Glottolog with simplifications; countries were compiled from Ethnologue and Wikipedia; the phrase and its transliteration were produced for the atlas.

Limitations

• Counting speakers is estimation: sources differ in criteria, collection dates and in what counts as “one language” (Standard Arabic vs. dialects; Filipino vs. Tagalog; Chinese varieties).
• Adding up totals does not give a population: people speak several languages.
• Ethnologue 200 values are rounded; Wikipedia values depend on the edition consulted.
• 96 of the 201 phrases are flagged as approximate and not yet reviewed by native speakers; histories cover only the 100 original languages, in Portuguese.
• Names in Portuguese, English and Spanish follow current usage, not a single standard.

Errors and contributions

Found an outdated figure, a wrong translation or a debatable classification? Contact channels are on the About page. Sourced corrections are incorporated in the next dataset updates.