Method & data
How Urim works
Urim records how the Church’s own published translations render Latter-day Saint vocabulary. It does not judge translations; it shows what was published, how often, and where.
Evidence
- Conference talks: 63,662 Portuguese and 61,988 Spanish paragraph pairs from official Church translations of General Conference talks, aligned paragraph by paragraph. Pairs whose paragraph counts do not match are excluded, as is any machine-translated text.
- Scriptures: verse-aligned Church editions of the Bible, Book of Mormon, Doctrine and Covenants and Pearl of Great Price (41,994 verses).
- Approved terms: for Spanish, the Church’s “Approved Terms for Interpretation” list.
Statuses
- approved term on the Church’s approved list.
- attested found in aligned Church text; counts and snippets link to the official page.
- inferred mined automatically, not reviewed. Never treated as verified.
Earliest attested use
- English: the earliest talk in the General Conference corpus (1830 onward, 12,198 talks) that contains the term, plus the earliest in a full-text source (verbatim or authoritative print). Early records are often minutes, notes or reconstructions and can include editorial notes, so each source’s grade is shown.
- Translations: the first appearance of each rendering in the Church’s official translations in this corpus. These begin around 1990, so the date means “first recorded here”, not “first coined”. Earlier translated usage is not represented.
- Matching is by word form, so an early hit may use the word in an ordinary sense rather than as a technical term.
Deseret Alphabet
The site itself is in English only. Deseret transliteration is offered as a tool: the hosted MCP has a to_deseret tool, the local translation server can transliterate text and files, and the proclamations are available as Deseret PDFs. All of it uses the engine from pulpit-vault’s Deseret edition (dictionary spellings, context-resolved heteronyms, 38-letter standard); words with no spelling stay in Latin script. The package is in deseret/.
Limits
- Counts show published usage, not correctness or doctrine. A rendering that is common in one era may be rare in another.
- Matching is by word form and ignores context, so a count can include uses that do not carry the technical sense of the term.
- Official translations begin around 1990 for these languages, so earlier talks are not represented.
- 367 of 481 seeded concepts currently have confirmed evidence in at least one language.
Data
Built from a snapshot of pulpit-vault (commit f57f3ff446, 2026-11-02).
Download the table as CSV or the full index as JSON. Source code and pipeline: github.com/Samuel-Baird/urim.