sermonindex/bible

The Bible in 1,004 Languages

14,497,397 verses across 1,253 translations in 1,004 languages, every verse keyed to the same chapter-and-verse address so that any two languages can be aligned by joining on book, chapter and verse.

The Bible is the most widely translated text in existence, and for several hundred of the languages here it is the largest — sometimes the only — substantial digitised text. That makes this corpus unusually useful for low-resource machine translation, multilingual pretraining, and cross-lingual alignment research, on top of its obvious use for reading and study.

Published by SermonIndex.

Contents

206 translations are complete Bibles (66 books); 1,021 are New Testament or partial. Verse counts per translation range from a few hundred to 31,103.

Language coverage

1,004 distinct languages by ISO 639-3 code. English has the most translations (49), followed by Sanskrit (22), Ancient Greek (9), Spanish (8), Kaqchikel (7) and Romani (6). The long tail is the point: several hundred languages of Papua New Guinea, the Americas, Africa and South-East Asia, most of which have very little other digitised text.

Scripts and directions vary. text_direction in the translation config is ltr or rtl — respect it when rendering.

Schema

verse

translation carries name (native), english_name, short_name, language_name (native), language_english_name, text_direction, number_of_books, total_verses, website, `license_url`, license_notes and sha256.

Example — build a parallel corpus for any language pair

from datasets import load_dataset

v = load_dataset("sermonindex/bible", "verse", split="train")

eng = v.filter(lambda r: r["translation_id"] == "ENGWEBP")
arb = v.filter(lambda r: r["translation_id"] == "ARBNAV")

key = lambda r: (r["book"], r["chapter"], r["verse"])
a = {key(r): r["text"] for r in arb}
pairs = [(r["text"], a[key(r)]) for r in eng if key(r) in a]
print(len(pairs), "aligned verse pairs")

Source and licence

Mirrored from the Free Use Bible API (AO Labs) with the permission of its maintainer, who assembled it from eBible.org and other publishers on the principle that Scripture should circulate freely.

Licences are per translation. Every row in the translation config carries a license_url — check it for the translation you intend to use. Dataset assembly and metadata are CC BY 4.0.

Rights holders wanting a translation removed can open a discussion here.

Caveats

• —Versification differs between traditions. Joining on (book, chapter, verse) is right for most pairs, but Psalm numbering, the Hebrew versus Greek ordering of Jeremiah, and the treatment of disputed verses (Mark 16:9–20, John 7:53–8:11, the Johannine Comma) vary. Expect a small percentage of unmatched verses in any pair and do not assume a dense matrix.
• —Translation philosophy varies enormously, from interlinear literal to free paraphrase to oral-style translations built for languages with no written tradition. Two rows sharing a verse address are not necessarily sentence-level equivalents.
• —Footnotes and cross-references present in the API are not included; only verse text and section headings.
• —heading is the last heading seen before the verse in that chapter, so consecutive verses under one heading repeat it.
• —Quality varies. Some translations are scholarly, some are field translations in active revision, and a few have known transcription artefacts.

Related datasets

• —`sermonindex/bible-parallel-english` — 17 English translations plus Latin Vulgate and Syriac Peshitta
• —`sermonindex/bible-reference` — Greek and Hebrew lexicons, morphology, cross-references
• —`sermonindex/early-church-fathers` — 68,240 patristic passages keyed to verses

Citation

@misc{sermonindex_bible,
  title  = {The Bible in 1,004 Languages},
  author = {SermonIndex},
  note   = {Mirrored from the Free Use Bible API (AO Labs) with permission},
  year   = {2026},
  url    = {https://huggingface.co/datasets/sermonindex/bible}
}