How Many Languages Has a Book Been Translated Into?

Pick a book you love. Now tell me how many languages it's been translated into.

It sounds like something a catalogue should tell you in a second. It doesn't. The record of a book's translations is scattered across national libraries, ISBN agencies and commercial databases that don't talk to each other. A Turkish edition sits in the Turkish catalogue; a Georgian one in the Georgian catalogue; and no single source has ever held all of it. We wrote about why that is in an earlier post. This post is about what happens once you stop searching one catalogue at a time and correlate them all.

Kafka's The Metamorphosis turns up on Zenòdot in 38 languages. The multilingual picture book Am I Small? — 82. Not because we translated anything, but because the editions already existed — filed in different catalogues, waiting to be matched to the same work.

What Zenòdot does

Zenòdot is a free, independent platform that correlates records from more than 50 bibliographic sources — national libraries on every continent, Wikidata, UNESCO's historical index, and commercial aggregators — to answer one question: has this book been translated into this language?

Today the correlated corpus holds 61.9 million editions, organized into 12.9 million works and drawn from a base of 230 million records. It documents 638 languages across its sources, and it resolves an author's identity through more than 70 million name variants — so Dostoevsky, Достоевский and ドストエフスキー all lead to the same person, whichever script you type.

We didn't invent any of these sources. We connected them. And we built the connection to give a Catalan, Basque, Welsh or Amazigh edition the same documentary treatment as an English one — not as a feature added later, but as the reason the project exists.

What changed this summer

For a long time the hard part wasn't finding editions. It was recognising that two records — one filed under a work's original title, another under a translated title in a different alphabet — belong to the same book. When a catalogue leaves an edition's language blank, or spells a title a little differently, that edition drops out of the record. A blank language field is not an absence. The edition is there; the catalogue simply doesn't say what it is.

This July, Zenòdot completed a large-scale unification: over 108,000 duplicate work records were merged into their canonical families. Editions that had been scattered came home. And in the process, the correlation restored the language of millions of editions that catalogues had left unrecorded — by matching them, through their siblings, to a language we could identify.

More than 57.7 million editions now carry a resolved language — over 4.7 million of them recovered in that unification.

The languages that gain the most

Here is the part that matters. When you repair a fragmented catalogue, the languages that gain most are not English, Spanish or French. Those were always well served. The gains land on the languages the publishing market tends to overlook.

Occitan now passes a thousand editions with an identified language on Zenòdot — about fifty of them newly recovered, editions that every source held but none had labelled. Maithili and Sundanese — languages with tens of millions of speakers each — gained hundreds of editions apiece. These are not obscure tongues. They are large languages that commercial catalogues barely register, made visible by matching what the world's libraries already hold.

That is the whole point of the exercise: the infrastructure doesn't invent a single translation. It restores the language that fragmented catalogues had left mute.

See it for yourself

Search a title and watch the languages assemble: The Metamorphosis (38), Am I Small? (82), Hamlet (24 and counting).

Search an author in any script — Latin, Cyrillic, CJK, Arabic, Hebrew, Greek — and the same person answers.

Try it with a book in your own language. Especially a small one. That is where Zenòdot is trying hardest to be useful — and where it most needs your eye when it gets something wrong.

You can search it at zenodot.app.

What it still can't do

The same honesty as always. Zenòdot draws on the metadata of the world, and that metadata is incomplete: biased toward well-resourced languages, thin before 1990, blind to most self-published work. Some editions are flagged "Possible translation" — an algorithmic inference that can be wrong. A book missing from Zenòdot is rarely missing from print; more often the infrastructure that would connect it simply doesn't exist yet. No single tool — including this one — can claim totality. The data doesn't allow it, and pretending otherwise would be the easiest way to be useless.

Who's behind it

Zenòdot is built by one person, from the Valencian Country — no investors, and a public layer that stays free. The conviction behind it is plain: the infrastructure that documents the world's minoritized languages should not depend on the same market that made them invisible in the first place.

Frequently asked questions

Is there a single database that tracks all of a book's translations?

No. UNESCO's Index Translationum stopped reporting in most countries around 2013; commercial aggregators cover a fraction; national libraries record what is published inside their borders, not translations outward. Zenòdot correlates dozens of these at once, but no tool — including Zenòdot — can claim to hold them all.

When Zenòdot says a work exists in 38 languages, what does that mean?

It means Zenòdot has matched editions in 38 languages to that work across its sources. It reports what catalogues document, not what is currently in print or in stock.

Why do minoritized languages benefit most from correlation?

Because their editions are the ones most often left uncatalogued, mislabelled, or stranded in a single national database. Matching sources together recovers exactly the records that fall through the cracks — and those records are disproportionately in smaller languages.

Is Zenòdot a bookstore?

No. It's a discovery tool. It tells you where editions exist and links out to retailers; it doesn't sell books or guarantee availability.

What building Zenòdot keeps teaching me is this: a catalogue holds what has been connected to it, not what exists in the world. A translation missing from a database is almost never missing from print. What's missing is the wiring. This summer, we ran a little more of it.