Method & data
How it works
Deseret writes sounds, not spellings: one letter for each sound of English. So every word is first turned into its sounds, then each sound into its letter.
Spelling English in Deseret
- Pronunciations: the CMU Pronouncing Dictionary, 126,052 words with their sounds in ARPAbet, pinned to one published version.
- Sounds to letters: the same one-to-one table the Deseret Wiki uses for its own pages, so a word is spelled the same way here and there. Y + oo is the single letter 𐑏 and oi is 𐑎, the two letters added in 1859.
- Hand corrections: the wiki’s own (and, not, Deseret…) and Book of Mormon names after the Church’s pronunciation guide.
- Words the dictionary lacks are built from a known word and a regular ending or prefix (unfaithfulness = unfaithful + -ness). These are listed as built from parts, so you can check them, and can be left in Latin letters instead.
- Anything else stays in Latin letters and is listed by name with how often it occurs. Nothing is guessed letter by letter.
- Words with two senses such as read and live are decided by the word before (have read, will read) and reported.
How well it covers English
On the Deseret Wiki’s encyclopedia article, 7,741 words dense with names and technical terms, 97.0% are spelled from the dictionary, 0.6% are built from their parts and 2.4% are reported as unknown. Ordinary prose does better. Against the wiki’s own word table the spellings agree on every word except four with two senses (lead, read, use, wind), where this server picks the common reading.
Reading Deseret back
The dictionary is reversed. Deseret records sound, so words that sound alike are spelled alike (𐐺𐐴 is by, buy or bye); the server picks the likelier one and names the others. Spellings that match no dictionary word, such as the older conventions of the 1860s books, are respelled letter by letter and flagged.
Answering questions
Answers come from a snapshot of every source on deseret.wiki, and every result links back to the page it is on, and to the original where the wiki links out:
- the encyclopedia article on the Deseret Alphabet, in 23 sections (text from Wikipedia, CC BY-SA 4.0, credited with every answer);
- the library: 28 hosted PDFs, sermons and conference addresses, and further reading;
- 10 resources, 9 videos, 35 fonts, 40 books & prints and 15 quotes;
- the 38-letter chart with each letter’s name, sound and key word.
Snapshot of 2026-11-17, wiki commit d249d5818a.
Limits
- Spellings follow one modern American pronunciation, not a regional accent or the conventions of the 1860s printings.
- Only a short list of two-sense words is decided from context; the rest use the dictionary’s first reading.
- The wiki snapshot is rebuilt when the wiki changes, so a page added today may not be searchable yet.
Privacy
The connection keeps nothing: text you send is transliterated and returned, not stored or logged by the server. No account, no sign-in.
Source and credits
Source code, tests and the data files: github.com/Samuel-Baird/deseret-mcp. Pronunciations © Carnegie Mellon University (BSD licence). Method, letter table and corrections from the Deseret Wiki. Sister project of Urim, whose layout this site shares.