Kural GraphBack to the Kural

About Kural Graph

The Tirukkural of Tiruvalluvar is 1,330 couplets in 133 chapters, on right conduct, wealth and desire. This is a way to read it, and to see how its ideas connect.

How the links were made

The links between couplets and ideas were written by an AI (Anthropic’s Claude), not by a scholar. Every link shows its reason, and the words in the plain reading that support it. Links marked “inferred” are an interpretation, not something the couplet says outright.

A second, different AI model checked them independently, and disagreements were judged again. These links are still in draft and need review, so treat them as a reading of the text and not as authority. The text itself always outranks them: read the couplets.

The texts

The readings on this site were written by an AI (Anthropic’s Claude), not by a scholar or translator. Each couplet has two readings. The plain reading says what the Tamil says, in today’s words, and adds no fate, god or setting of its own. parimēlaḻagar’s reading is his 13th-century Tamil commentary, paraphrased into English. It keeps what he says, including his views on caste, karma and women, which are of their time.

V. V. S. Aiyar’s English translation of 1916 is also offered wherever you choose a reading. It is in his own words, from the Standard Ebooks edition, which is believed to be in the United States public domain. The readings don’t draw on it, and it uses words of its time, such as “love” where the plain reading says “loving-kindness”.

The paraphrases are not his exact words, so each one opens onto his Tamil, from Tamil Wikisource. Two key words follow the project’s own reading: அன்பு is loving-kindness, the mēttā of Buddhism, and காமம் is desire. parimēlaḻagar himself defines அன்பு more narrowly, as fondness for your family and those connected to you.

The Tamil starts from the thirukkural dataset (Apache 2.0). It was checked line by line against Project Madurai and two editions on Tamil Wikisource, and 105 decisions were reviewed, including 63 repairs of damaged text. Notices and credits.

The romanised Tamil

How to read the romanised Tamil

This is a revised IAST, not the standard scheme. t and ṭ are swapped, as are d and ḍ, and the hard consonants are spelled as they sound.

TamilWord start, or doubledBetween vowels, or after a nasal
கkg
சcs, or j after a nasal
டtd
தṭḍ
பpb

e and o are short (எ, ஒ); ē and ō are long (ஏ, ஓ). A hyphen marks a vowel letter inside a word, as in aṉbu-īṉum.

Your privacy

The site works offline after your first visit. Your reflections and settings stay in this browser. Nothing is sent anywhere unless you connect a language model, and then only the text you choose goes to the provider you chose, directly from your browser, using your own key.

The walks

A walk starts from a situation, goes through what it can feel like, then to what the Kural sets against that, and ends at the couplets that say so. The feelings are our suggestion and are marked as such: the Kural doesn’t know your situation. Every later step cites a couplet. Where the Kural offers no way through, the walk says so rather than inventing one. Couplets that could hurt someone in distress are never part of a walk.

Walking through a text this way comes from GitaViśva, Himanshu Mohanty’s knowledge graph of the Bhagavad Gita, which also inspired this site.

The interface and code

This site is a fork of GitaViśva by Himanshu Mohanty. Its interface and code come from that project, used under the MIT licence. Its name, its paintings and its Gita data are not used here.

Licences

See the notice for the sources of the texts, and the third-party notices for the software, fonts and search model this site ships.

Not advice

This is a book from two thousand years ago, read by software. It isn’t medical, legal or mental health advice. If you are struggling, please talk to someone who can help.