બેઠક A place to sit and hear them again.

A bethak is where people sit together and talk. That's all this is: 9341 Gujarati sayings, written down, with what they mean in English.

Why

Most people who grew up around Gujarati elders carry a few dozen of these around without ever having seen one written down. You know the shape of it, and roughly when it gets said. What you don't have is the spelling, or a way to look one up, or an answer when someone asks what it actually means.

The rest of the internet isn't much help either. Listicles of thirty sayings with no sources, and one good dictionary that's closed. Nothing you can search in Roman letters or check against anything.

Where they came from

1234 come from Gujarati Wikiquote, under CC BY-SA 4.0. That's a useful filter on its own. Somebody alive typed those in, so they're sayings people still say, not only ones that survived in old books.

The rest I read out of three public-domain printed collections. Four volumes, because one of them I did in both editions. All scanned by archive.org, none of them available as text until now.

• આશારામ દલીચંદ શાહ, ગુજરાતી કહેવતસંગ્રહ. Numbered entries, each with a stack of regional variants under it, footnote glosses on the hard words, and sometimes an English proverb Shah picked out as the match. I read both editions. They're the same book set in type twice, which matters more than it sounds like it should, for reasons below. Shah died in 1921, so this is public domain.
• કહેવત સમુદાય. 121 pages, and the worst scan of the four. On a long run of its pages the left half of the first letter of every line has been shaved off. Those entries are here with their Gujarati and no English, labelled as damaged rather than patched up.
• અરવિંદ નર્મદાશંકર શાસ્ત્રી, બૃહદ કહેવત કથાસાગર (1920). 507 pages that tell the story behind each saying and quote the Marathi, Hindi and Sanskrit versions next to it. I only took the sayings and the parallels. The stories aren't here because I couldn't get them out reliably: asked to transcribe a page of that prose, the reader handed back two paragraphs out of about a dozen and said nothing about the rest. A story that's silently 15% transcribed reads as finished and is quietly wrong. I'd rather have the empty field. 446 of the 507 pages are read; the last 61 are still to come.

Every entry says where it came from and links to the scanned page. The counts below don't add up to 9341, which is the good part: a saying found in two books is one entry with two sources.

2944

ગુજરાતી કહેવતસંગ્રહ (Gujarati Kahevat Sangrah) (1923)

2565

ગુજરાતી કહેવતસંગ્રહ (Gujarati Kahevat Sangrah) (1921)

2359

કહેવત સમુદાય (Kahevat Samuday)

1234

Gujarati Wikiquote

1087

બૃહદ કહેવત કથાસાગર (Bruhad Kahevat Kathasagar) (1920)

The one check worth anything

828 sayings turn up in more than one of these. That's the only check here that isn't circular. Two machines reading the same scan can agree on the same misreading, because they're staring at the same damaged ink. Two books, set by different compositors and scanned years apart, printing the same saying, can't agree by accident.

Shah also printed, under each entry, a count of how many variants he'd set beneath it. Where my count matches his, the shape of that entry is confirmed: no two columns merged across the gutter, no wrapped line split in half. It says nothing about whether the letters are right.

How to spell it in Roman letters

Every saying carries an ISO 15919 transliteration, generated from the Gujarati by a fixed rule set rather than by a model. A looser phonetic version drives the search box and the web addresses, so typing panch finds pāṁca.

What has not been checked

The English was written by a machine and nobody has reviewed it. 6651 of 9341 have a translation; the rest are waiting. A translation can read perfectly and still be wrong, and from this page you'd have no way to tell. Every entry links to its source so you can check.

There's a second problem, harder to see, and it only touches the entries from books. A saying typed by a person went through one uncertain step to get here. A saying read off a scan went through two: was the word read right, and then was it understood right. Only the second one is easy to catch. A misreading that still looks like plausible Gujarati comes back with a fluent English sentence attached, and nothing about it looks off.

1234 of 9341 entries have Gujarati that a person typed. For the other 8107 the Gujarati is machine-read and unconfirmed. Where a line was too broken to read, I left the English blank instead of repairing the text first. A guess buried inside an English sentence is invisible to everyone who comes after it, and that's the one thing I won't ship.

Take it

The whole collection is one JSON file in the repository. The material from the printed books is public domain, so take it and do what you like with it. The Wikiquote material is CC BY-SA 4.0: reuse it, credit Gujarati Wikiquote, share alike.