Babel · a window onto all of human language
we were never really scattered
The old story says language broke us apart. But every word's hidden connections tell the opposite. 10 million words·4,816 tongues·woven into one shape of meaning.
or wander into a word · a random one of ten million
real records, counted from every entry we hold
The un-Babel
No language is an island. English begged, borrowed and stole from hundreds of tongues; water answers to itself in three thousand languages; a muscle is a little mouse and a disaster a bad star in every European mouth. Pull on any word and the whole web of human contact comes up with it — five thousand years of meeting, trading, praying, falling in love. We were never really scattered.
Every page here is computed from the full
wiktextract extraction of
English Wiktionary — 10.6 million entries across 4,816 languages — loaded into a local
DuckDB. A few honesties: borrowed % counts entries with a borrowing etymology over all
entries (a relative signal, not a share of the lemma stock); translations live mostly on
English entries, so cross-lingual data is English-hub; and the Semantic Universe's meaning
embeddings come from a local qwen3-embedding model.