The goal: a universal etymology dictionary

Etymologically aims to be a universal etymology dictionary: one place that gives the history of the words of every language, traces each word back as far as the evidence allows, cites the scholarship for every step, and links each step to the entry for the word in its own language.

Why a universal dictionary

Words do not respect the borders between dictionaries. English orange came through French, Italian, Arabic and Persian from a language of India; Hungarian, Romanian and the Slavic languages have traded words for a thousand years; Latin and Greek words live on in almost every language of Europe. Yet the scholarship on these histories is split into hundreds of separate works, each covering one language or family, written in German, French, English, Russian and many other languages, and many of them long out of print and found only in specialist libraries.

A reader who wants to follow a word across languages has to move from one of these works to the next, and often cannot read the next one. A universal dictionary brings those works together, so that the whole path of a word can be followed in one place, in one language, with every step sourced to the work it comes from.

What “universal” means here

Every language. The dictionary is not limited to English or to the major languages of Europe. Living and dead languages, standard languages and dialects, languages with a long written history and languages recorded only recently are all in scope.
Every word with a history worth telling. Not only common words, but rare, old, regional and technical ones, together with their earlier forms and descendants.
As far back as the evidence goes. Each entry follows a word from its present form through its recorded stages, to the language it was borrowed from or inherited from, and on to the reconstructed root where comparative evidence supports one. The chain stops where the evidence stops: when the origin is unknown, the entry says so.
Each language from the scholarship on it. A word is written up from the specialist works on its own language, in whatever language they were written, rather than from what an English dictionary happens to say about it. For French that means the French dictionaries and also Wartburg’s Französisches etymologisches Wörterbuch, written largely in German; for Latin, de Vaan in English, Walde–Hofmann in German, Ernout–Meillet in French and others in Spanish and Italian. The dictionary brings these together and presents them in English.

One connected record

Entries are not isolated articles. When an etymology says a word was borrowed from French, the French word links to its own entry, which carries the history on from there; the etymology tree on each page shows ancestors and descendants across languages. As entries are added in more languages, these links join into a single record of how the world’s vocabularies are related, in which any word can be followed backwards to its origins or forwards to everything that came from it.

Built on published scholarship

A universal dictionary is only useful if it can be trusted, so its breadth is never bought with invention. Every entry is built from published dictionaries and scholarly works, and every factual statement is footnoted to the work it comes from, following the editorial policy. Where specialists disagree, the dictionary shows each account with its source. Speculation that no serious scholarship supports is left out, however popular it may be (see how sources are chosen). The works cited are listed on the sources page.

Work on this scale is done with the help of AI tools, so how they are used matters. Entries are not generated from what an AI model absorbed in training, which can be unreliable and prone to inventing plausible-looking facts. Each claim is read in the source it is cited to, and the finished entry is verified against those same sources before it is published.

The dictionary does not replace these works. It points to them, and a reader who wants the full argument for an etymology can follow the footnote to the dictionary or study it comes from.

How sources are chosen

Covering every language means drawing on a very large body of writing about words, and not all of it is scholarship. Sources are chosen for the method behind them, not for how many words they explain or how appealing their stories are.

The standard works of each language. The core of every language’s entries is its established etymological and historical dictionaries, together with studies of its loanwords and older stages, written by specialists and reviewed by other specialists.
The comparative method. An etymology is accepted when it rests on regular sound correspondences, a plausible development of meaning, and a historical setting in which the word could have passed from one language to another. Chance resemblances between words, of which every pair of languages has many, prove nothing.
No fringe theories. Theories that the field has examined and rejected are not used as sources and are not presented as alternatives. Hungarian is a Uralic language, related to Finnish, Estonian, Mansi and Khanty; the claim that it derives from Sumerian rests on look-alike words and is not given a place beside that evidence. The same applies to etymologies that trace large parts of Romanian vocabulary to Dacian, a language known from only a few names and glosses, and to other attempts to link a language to a prestigious ancient people without regular correspondences to support it.
Open questions stay open. Genuine disagreements among specialists, and hypotheses still under debate, are reported with their sources and labelled as such. Rejecting fringe theories is not a way of hiding real uncertainty: when a word’s origin is unknown, the entry says so rather than filling the gap.

How far along it is

A universal dictionary is a goal, not a description of the site today. Etymologically currently has 26,457 entries in 10 languages, with English the most complete, and new entries and languages are added continually. The languages page shows what is covered so far, and recent changes shows what has been added lately.

Many words are mentioned in entries before they have an entry of their own. Those are the next to be written, and each one added links more of the record together.

How you can help

If you find an error, a missing word or a source that should be cited, please write to [email protected]. Corrections are made promptly and recorded in the page’s public history.