Languages
Every language in this sample. S is speakers, M is monolingual data, Bmax is the largest bilingual corpus with any partner, and Σ (ran_sum) is the sum of all components, a rough single-number proxy used in the RAN Paper. Each digit is an order of magnitude (see the overview); each page lists the raw counts and sources behind its score.
The set is deliberately diverse. From English at the top to Indigenous languages of the Americas (Quechua, Guarani, Cherokee, Owens Valley Paiute) that are usually flattened into one "low-resource" label despite differing by orders of magnitude.
All languages
| Name | RAN | S | M | Bmax | Σ | Speakers |
|---|---|---|---|---|---|---|
| English | 9/10/es-8/fr-8/de-8/zh-7 | 9 | 10 | 8 | 27 | 1132366680 |
| Chinese | 9/9/en-7/ja-7 | 9 | 9 | 7 | 25 | 1299877520 |
| Spanish | 8/9/en-8/fr-8/pt-8 | 8 | 9 | 8 | 25 | 485000000 |
| Arabic | 8/8/en-7/fr-7 | 8 | 8 | 7 | 23 | 422000000 |
| Hindi | 8/8/en-7/ur-6 | 8 | 8 | 7 | 23 | 341000000 |
| Vietnamese | 7/9/en-7/zh-6 | 7 | 9 | 7 | 23 | 76000000 |
| Korean | 7/8/en-7/ja-6 | 7 | 8 | 7 | 22 | 77300000 |
| Turkish | 7/8/en-7/de-7 | 7 | 8 | 7 | 22 | 82231620 |
| Nepali | 7/7/en-7/hi-6 | 7 | 7 | 7 | 21 | 24131000 |
| Sinhala | 7/7/en-7 | 7 | 7 | 7 | 21 | 15300000 |
| Mongolian | 6/7/en-7/zh-5 | 6 | 7 | 7 | 20 | 7000000 |
| Hausa | 7/6/en-6/fr-6 | 7 | 6 | 6 | 19 | 43900000 |
| Swahili | 7/4/en-7/fr-6 | 7 | 4 | 7 | 18 | 15437390 |
| Welsh | 5/6/en-7 | 5 | 6 | 7 | 18 | 977366 |
| Maltese | 5/3/en-7/it-6 | 5 | 3 | 7 | 15 | 570000 |
| Yoruba | 7/2/fr-6/en-6 | 7 | 2 | 6 | 15 | 40000000 |
| Guarani | 6/1/en-6/es-2 | 6 | 1 | 6 | 13 | 4500000 |
| Quechua | 6/0/en-6/es-2 | 6 | 0 | 6 | 12 | 8900000 |
| Cherokee | 4/2/en-4 | 4 | 2 | 4 | 10 | 12300 |
| Owens Valley Paiute | 1/3/en-3 | 1 | 3 | 3 | 7 | 12 |