Define a proto-language and the stages after it, write the sound changes between them, and Lemula derives every word at every stage, with the history that got it there.
Name each checkpoint, give it a rough date, and attach the segment inventory the stage allows.
One rule set per stage, in the source, target and context notation the engine reads.
Forms, glosses and features. Type them in or import a CSV.
Every word is derived at every stage of every line, with the trace that produced it.
Each stage carries its own rules in source, target and context notation. The engine knows IPA, so classes like vowels or voiced stops are available without spelling them out. For any word you can see which rules fired and in what order.
Words also enter and leave the lexicon by other routes: compounding, derivation, semantic shift, irregular shortening, borrowing from a sister line, coining, and falling out of use.
Between two stages a word can be coined, split in meaning, rebuilt from another word, clipped, borrowed from a sister line, or dropped. Each of these is a recorded event in the word’s history.
A regular rule applied to every word that fits the pattern. p > b / V _ V.
A word enters the language at a stage. Before it, nothing.
One word becomes two, same form, meanings apart. The parent stays.
The word itself changes what it carries. Nothing splits.
A new word built from one parent by a recipe: affix, template, or a form the author sets.
Two or more words join into one. Order matters; the seam can run through the phonology.
A word loses a piece from one end where no regular rule explains it.
The word stops being used from this stage on. The reason is recorded.
A word crosses from a sister line at a chosen stage. It arrives as it stands there, then follows the new rules.
Attach a cited word list to any stage: attested or reconstructed forms, with source and confidence. On the first stage it seeds the line. On a later stage it scores what the rules produced.
In compare mode the rules stay the source of truth and the lexicon grades them. In replace mode the attested form takes over from that stage down. Either way, every word lands in one of four groups.
Named stages per line, a segment inventory attached to each one, and a feature schema for words: part of speech, gloss, anything else you define.
An IPA-aware phonology engine. Rules per stage as src > tgt / context. For any word, the exact rules that fired.
Cited word lists with attested or reconstructed forms, attached to a stage as seeds or as a scoring reference. Compare or replace mode; form and chosen features.
Daughter lines split off at the stage you choose. Borrowings cross between lines. Cognates are tracked across the family.
The app materialises every form at every stage and keeps the derivation history, so each result can be traced back.
Enter words by hand, or import and export CSV.
Family tree in two layouts, a searchable dictionary per line, a page per word with its history, compound origins, cognates and obsolescence notes. An author page and catalogue. Light and dark. Indexed by search engines.
Draft previews are visible to the author only. Publishing opens the pages to everyone with the link.
Drafts stay private to you. Publishing puts the pages on a public URL, indexed by search engines, with light and dark themes. Sign in with Google or an email and password.