Nikos Katsaounis

mvp live · seed corpus 65 proverbs · 13 clusters · 5 languages

Paremia

the open dictionary of the world’s proverbs

Type a word in any language; get every culture’s proverb for the same idea. The unit: the concept cluster.

Live MVPSearch the corpus →

01 · The idea

Donkey, camel, pot: same idea, different farm animal. Search another culture’s version: a listicle. The scholarship (8,300 proverb types in Helsinki, a yearbook since 1984, the Cervantes Institute’s Refranero) is walled.

Paremia is the missing middle. The concept cluster: the concept travels; proverbs are its five mouths. Launch languages (English, Greek, Spanish, Arabic, Mandarin), picked for maximum distance.

Small on purpose: 65 cited proverbs beat 6,500 in a file. LLMs invent proverbs; this corpus is the boring, correct alternative.

02 · The specimen · one concept cluster

Hypocrisy: the accuser guilty of the same fault

cluster pot-kettle · 5 languages · oldest form 4th c. BCE

Verbatim, from the seed corpus.

  • Greek

    Greece

    Είπε ο γάιδαρος τον πετεινό κεφάλα.

    Eípe o gáidaros ton peteinó kefála.

    Literal. The donkey called the rooster big-headed.

    Folk-traditional, mainland Greece and islands.

  • Arabic

    Levantine / Egyptian Arabic

    الجمل ما يشوف حردبته.

    Al-jamal mā yashūf ḥardabtah.

    Literal. The camel does not see its own hump.

    Widely attested across Arabic-speaking cultures.

  • Mandarin

    Mandarin

    五十步笑百步。

    Wǔshí bù xiào bǎi bù.

    Literal. The soldier who fled fifty paces laughs at the one who fled a hundred.

    Mencius, 4th century BCE.

  • Spanish

    Spain / Latin America

    Dijo la sartén al cazo: quítate allá que me tiznas.

    Literal. The frying pan said to the pot: get away from me, you’re smudging me.

    Don Quixote II, ch. 67.

  • English

    British / American English

    The pot calling the kettle black.

    Literal. The pot calling the kettle black.

    Shelton’s 1620 translation of Cervantes.

03 · How the search works

~40 KB static, no backend, client-side search. Weighted scoring: original, transliteration, theme, keyword, gloss. Diacritic-normalized: gaidaros finds γάιδαρος. Arabic renders right-to-left.

Around the bar: a proverb of the day; themes; an atlas: concepts, images, attestations 650 BCE to 50 CE. Results open the cluster.

Client-side holds to ~500 proverbs; then a real backend.

Back to the Lab← Lab