Sanskrit: Distant, Yet Familiar

Sanskrit seems about as far away as a language can be: another country, another age, an unfamiliar script. Look a little closer, though, and familiar words start to show through. Mother is mātṛ. Brother is bhrātṛ. Two, three are dva, tri. That thou art is tat tvam asi. None of this is coincidence. Sanskrit and English are distant relatives, and the kinship shows not only in words but in how the language is built.

This article covers what kind of language Sanskrit is, who Pāṇini was, why his grammar is famous, and where Sanskrit meets English, Spanish and Russian. You do not need to know any Sanskrit to follow it.

Russian words appear in Cyrillic with a Latin transcription in brackets, e.g. два (dva).

What kind of language it is

The word saṁskṛta means "put together, refined, perfected". The prefix sam- means "together", and the root kṛ means "to make". Something like "well-made" or "composed". It was contrasted with prākṛta, "natural": the everyday spoken dialects.

Sanskrit has two great periods:

• Vedic — the language of the Vedas. The oldest hymns of the Ṛg-veda are generally dated to the second half of the second millennium BCE. For centuries they were handed down orally, with astonishing precision.
• Classical Sanskrit — the language fixed by Pāṇini's grammar. It is the language of the Mahābhārata and the Rāmāyaṇa, the Purāṇas, the Bhagavad-gītā and the Bhāgavatam, of poetry and drama, and of treatises on philosophy, mathematics, astronomy and medicine.

Sanskrit is not a "dead" language. People still write poetry and scholarship in it, debate in it in traditional schools and chant it in temples every day. In India's 2011 census about 25,000 people named it as their mother tongue.

One more common misunderstanding: Devanāgarī is not Sanskrit — it is a script. Over the centuries Sanskrit has been written in Bengali script, in Grantha and in a dozen others. Devanāgarī is simply the most widespread today. On this site every verse is given both in Devanāgarī and in Latin letters (IAST, the scholarly transliteration with dots and bars over the letters).

One big family

In 1786 Sir William Jones, a British judge and orientalist in Calcutta, said the words that launched comparative linguistics:

The Sanscrit language, whatever be its antiquity, is of a wonderful structure; more perfect than the Greek, more copious than the Latin, and more exquisitely refined than either, yet bearing to both of them a stronger affinity, both in the roots of verbs and in the forms of grammar, than could possibly have been produced by accident.

Jones guessed that Sanskrit, Greek and Latin shared an ancestor. Today it is called Proto-Indo-European. Nothing was ever written in it; it is reconstructed by comparing its descendants. Linguists mark such reconstructed forms with an asterisk: *mātēr.

Many branches grew from this ancestor, among them:

• Indo-Aryan — Sanskrit, and from it Hindi, Bengali, Marathi, Nepali and others;
• Germanic — English, German, Dutch, Swedish and others;
• Romance (through Latin) — Spanish, Italian, French, Portuguese;
• Slavic — Russian, Ukrainian, Polish, Czech and others;
• Baltic — Lithuanian and Latvian;
• and also Greek, Iranian, Armenian and Celtic.

Sanskrit is not the parent of English, Spanish or Russian: they are branches of one tree. How can we tell? A descendant cannot recover distinctions its ancestor had already lost.

In Sanskrit the ancient vowels e and o fell together with a: daśa ("ten"), nava ("new") and mātṛ ("mother") all have the same vowel, a. Yet Latin keeps three different vowels in the same places — decem, novus, māter — and so does Russian: десять (desyat'), новый (novyy), мать (mat'). Had these languages descended from Sanskrit, they would have had no way of knowing where to put e, where o and where a. The same goes for l: Sanskrit has r in śravas ("fame") and śru ("to hear"), while Russian has l in слава (slava), "glory", and слух (slukh), "hearing", and English has l in loud. So each language inherited these words from the common ancestor and kept something of its own.

Some branches, though, grow closer than others. English and Spanish have lost almost all their old noun cases; Russian still keeps them, which is why Russian grammar often feels closer to Sanskrit than English grammar does. Of all living European languages, Lithuanian is probably the one that has kept the most archaic features.

Sounds and script

The English alphabet runs A, B, C, D — an order inherited from the Phoenicians and Greeks that means nothing in particular. The Sanskrit alphabet is arranged by the anatomy of speech: vowels first, then consonants, from the throat to the lips.

Each row is one place in the mouth, each column one way of making the sound. It is a complete phonetic chart, drawn up more than two thousand years ago.

A few tips on pronunciation:

• An h after a consonant is a breath, not a separate sound. Bh in bhakti is a b with a puff of air. Th is a t with a puff of air — never the th of think. Ph is a p with a puff of air — never an f.
• English speakers have an advantage here: English p, t, k at the start of a word are already aspirated. The p in pin is close to Sanskrit ph; the p in spin is Sanskrit p. Spanish and Russian speakers pronounce p, t, k without the puff — exactly like Sanskrit p, t, k.
• ṣ is close to English sh; ś is a softer, more forward sh. c is like ch in church, j like j in jam.
• ṛ is an r that works as a vowel, rather like the er of American butter, or the l of Czech vlk ("wolf"). Czech vlk and Sanskrit vṛka are in fact the same word.
• Long vowels (ā, ī, ū) last exactly twice as long as short ones. The whole of Sanskrit verse rests on this: how syllables add up to a metre is explained in the section on verse metre.

Pāṇini: a grammar two and a half thousand years old

Now for the man who made Sanskrit "classical".

Pāṇini lived, by most estimates, around the 4th century BCE (dates range from the 6th to the 4th) in Gandhāra, in the north-west of the Indian subcontinent, near today's Attock in Pakistan. Tradition names his birthplace as the village of Śalātura. Beyond that we know almost nothing of his life. But his book survives.

"Eight chapters"

Pāṇini's grammar is called the Aṣṭādhyāyī — "[the book of] eight chapters". It has about four thousand rules, or sūtras (3,959 in the usual count). Together they describe the whole of Sanskrit: how words are built from roots, how they change for case and person, and how sounds change where words meet.

The sūtras are extremely short. Here are the first two:

• 1.1.1 vṛddhir ādaic — "the sounds ā, ai, au are called vṛddhi";
• 1.1.2 adeṅ guṇaḥ — "the sounds a, e, o are called guṇa".

Four words, and two technical terms are defined on which half the grammar rests. Pāṇini saved every sound he could. Later grammarians joked: "A grammarian rejoices at saving half a short vowel as much as at the birth of a son."

The Śiva-sūtras: naming a group of sounds in a word

Why it was needed. A grammar constantly has to talk about groups of sounds: "all vowels", "all consonants", "the sounds i, u, ṛ, ḷ". Spelling the group out every time would make the rules long. Pāṇini found a way to name any such group with two or three letters.

The list. Before the main text stand fourteen short lines, the Śiva-sūtras. Tradition says Pāṇini heard them from Śiva's drum as the god finished his dance. They contain the whole Sanskrit alphabet, not in the usual order but in a special arrangement. At the end of each line stands an extra letter (shown in bold). It is not a sound but a bookmark: it is never pronounced in words and exists only for counting.

1. a i u ṇ
2. ṛ ḷ k
3. e o ṅ
4. ai au c
5. ha ya va ra ṭ
6. la ṇ
7. ña ma ṅa ṇa na m
8. jha bha ñ
9. gha ḍha dha ṣ
10. ja ba ga ḍa da ś
11. kha pha cha ṭha tha ca ṭa ta v
12. ka pa y
13. śa ṣa sa r
14. ha l

The consonants are written with an a (ha, ya, va…) only so that they can be pronounced; what is meant is the bare sound h, y, v.

How to read an abbreviation. Take the sound to start from and the bookmark to stop at. The abbreviation means "every sound from this one up to that bookmark", skipping the bookmarks themselves. It is like saying "Monday to Friday" instead of naming each day.

Let us work out aC (the bookmark is conventionally written as a capital). Start at a and walk down the list until you meet the bookmark c:

• line 1: a, i, u — take them; skip the bookmark ṇ;
• line 2: ṛ, ḷ — take them; skip the bookmark k;
• line 3: e, o — take them; skip the bookmark ṅ;
• line 4: ai, au — take them; we have reached the bookmark c — stop.

The result is a, i, u, ṛ, ḷ, e, o, ai, au — all the vowels of Sanskrit. Two symbols instead of nine sounds.

Other abbreviations work the same way:

A rule at work. Now we can read one of Pāṇini's real rules, sūtra 6.1.77: iko yaṇ aci. It has three words, each in its own grammatical case, and the case tells you the word's job in the rule:

• ikaḥ — genitive of iK: "in place of a sound of group iK" — what is replaced;
• yaṇ — nominative: "a sound of group yaṆ comes" — what replaces it;
• aci — locative of aC: "before a sound of group aC" — when it happens.

Together: "i, u, ṛ, ḷ before a vowel become y, v, r, l". Which sound becomes which goes by position, as in two columns:

Let us test it:

• iti + api ("thus" + "also"): the final i stands before the vowel a, so i → y. Result: ity api.
• su + āgatam ("well" + "come"): u before the vowel ā, so u → v. Result: svāgatam, "welcome" — the word still greets guests all over India.
• prati + eka ("per" + "one"): i before e → y. Result: pratyeka, "each".

You can hear it yourself: say "iti api" quickly and your tongue slides into "ityapi". Pāṇini simply wrote it down as a law. Three words, and a whole layer of sandhi — the joining of sounds where words meet (more on this below) — is described.

The rule has three parts: a condition ("before a vowel"), what to replace, and what to replace it with. Rules in programming languages are built the same way, which is why programmers are so fond of Pāṇini.

The last sūtra. The whole book ends with rule 8.4.68, just two letters long: a a. Throughout the other rules, short a is treated for convenience as an "open" sound, the partner of long ā. In reality short a in Sanskrit is more closed, rather like the u in English but. The last rule gives it back its true sound: "a [the working one] is a [the real one]". The grammar begins with the sound a and ends with it.

Why Pāṇini still amazes people

• Pāṇini's grammar is generative. It does not just list forms; it gives rules by which any correct word is built, step by step, from roots and suffixes. Modern linguistics arrived at this idea only in the 20th century.
• It has meta-rules — rules about how to apply rules: which one wins, which applies first, which one cannot "see" the output of another.
• When a notation for describing programming languages appeared in the 1960s — Backus–Naur form — the computer scientist Peter Ingerman proposed calling it Pāṇini–Backus form, because Pāṇini had used a very similar device.
• The American linguist Leonard Bloomfield called Pāṇini's grammar one of the greatest monuments of human intelligence.

The Aṣṭādhyāyī comes with a list of about two thousand verbal roots (the Dhātupāṭha) and lists of words that behave in special ways. Pāṇini's rules were refined by Kātyāyana and Patañjali; together with Pāṇini they are called the "three sages" of grammar.

A grammar made of holy names

Pāṇini's is not the only Sanskrit grammar. Some two thousand years after him, in 16th-century Vṛndāvana, Jīva Gosvāmī — whose Ṣaṭ-sandarbha, Gopāla-campū and other works are on this site — wrote a grammar called the Hari-nāmāmṛta-vyākaraṇa. In it the dry technical terms are replaced with names of Kṛṣṇa and His associates. A student memorising the rules of declension is at the same time repeating holy names. The language of description becomes a prayer.

Words you already know

Let us start with the simplest thing — numbers. Numbers change slowly, so kinship shows at once.

"One" is the odd one out: Sanskrit eka is formed differently from one, uno and один (odin).

Next, the family. Words for relatives are among the oldest layers of any vocabulary.

Body, nature, animals:

The Himalayas are himālaya, "the abode of snow": hima ("snow") plus ālaya ("abode"). The same hima became Russian зима (zima), "winter", and hides in Spanish invierno.

Verbs are the most rewarding, because one root lives on in dozens of words:

That last line deserves a word of its own. Yoga comes from the root yuj, "to yoke, to join". It is the same word as English yoke, Spanish yugo and Russian иго (igo). Spanish cónyuge, "spouse", literally means "yoked together". Yoga is a "yoking": the joining of the soul to God.

And Veda, "knowledge", comes from the root vid, the same root as English wit and wise, Latin videre ("to see", hence video) and Russian ведать (vedat'), "to know".

Finally, pronouns — small words that barely change over thousands of years:

Why "ten" and not "dasha"

If the words are related, why do they sound so different? Because over thousands of years each language changed its sounds — but by rules. Sounds do not change at random; they shift in whole series. These series are the linguist's main tool: they separate true kinship from chance resemblance.

English shifted its own series. Where Sanskrit and Latin have p, t, d, English often has f, th, t. This is Grimm's law, described by Jacob Grimm — the same Grimm who collected the fairy tales.

Once you know the rule, pitṛ and father stop looking like strangers.

Sanskrit ś, Russian s, English h. An ancient sound *ḱ (a palatal k) became ś in Sanskrit and s in Slavic, but stayed a hard k in Latin (written c) and turned into h in English:

Here Sanskrit and Slavic went down the same road together, while Latin and English went down another. Linguists call the first group satem languages (from the Avestan word for "hundred") and the second centum languages (from the Latin).

The RUKI rule. After the sounds r, u, k, i, an ancient s became ṣ in Sanskrit (a sound like sh) and kh in Slavic. It is one of the clearest rules that Sanskrit and Russian share. Sanskrit snuṣā, "daughter-in-law", is Russian сноха (snokha); and the Russian locative plural ending -ах (-akh), "in …", goes back to the same ancient -su as Sanskrit phaleṣu, "in the fruits". The word phaleṣu occurs in a famous verse of the Bhagavad-gītā, 2.47: mā phaleṣu kadācana — "but never to the fruits".

Grammar: where the languages agree

Words are borrowed easily; grammar almost never is. That is why shared grammar is the best proof of kinship.

Cases

Sanskrit has eight cases. Here is vṛka, "wolf", beside Russian волк (volk), which still has six:

The locative matches almost sound for sound: vṛke — (o) volke. In Sanskrit prayers God is always addressed in the vocative: Kṛṣṇa!, Govinda!, Hare!

English has lost these endings on nouns, but it keeps cases in its pronouns: he — him — his, who — whom — whose. That is exactly the Sanskrit system, worn down to a few words. Spanish too keeps cases only in pronouns: él — le — lo.

Three numbers

Sanskrit has singular, dual and plural: vṛkaḥ, one wolf; vṛkau, two wolves; vṛkāḥ, wolves. Old English had a dual in its pronouns (wit, "we two"; git, "you two"), and Old Russian had it throughout; modern Russian still uses leftover dual forms for paired things such as eyes and ears.

Three genders

Masculine, feminine and neuter: deva (god), devī (goddess), phala (fruit, neuter). Adjectives agree with nouns in gender, number and case — as in Spanish and Russian. English has kept gender only in he, she, it.

The verb "to be"

Here the kinship is at its clearest. Put Sanskrit beside English, Latin (the ancestor of Spanish), Spanish and Old Russian:

The famous phrase from the Chāndogya Upaniṣad 6.8.7, tat tvam asi, is rendered in English almost word for word: That thou art. Tat — "that", tvam — "thou", asi — "art". In Spanish: Eso eres tú.

Verb endings

Sanskrit verbs take the endings -mi, -si, -ti for "I, you, he". Compare the verb "to give":

Spanish das ("you give") and Sanskrit dadāsi share both root and ending.

Participles

• The suffix -ta forms past passive participles: mṛta, "dead". Latin mortuus, Spanish muerto, and the English past participle in -ed / -t (kept, burnt) belong to the same family.
• The suffix -ant forms present active participles: bharant, "carrying". From it come Spanish -ante / -ente (cantante, "singer", literally "singing"), and English friend, originally an Old English participle meaning "loving".

Prefixes

Sanskrit verbs take prefixes, and many are shared: pra- ("forth") is English fore- and Latin pro-; ud- ("up, out") is English out; negative a-/an- is English un- and Spanish in-. And prapitāmaha, "great-grandfather", is built exactly like Russian прадед (praded).

Things Sanskrit shares with Spanish rather than English

• No word order needed. Because cases show who does what, word order is free. Rāmo vanaṁ gacchati, vanaṁ gacchati rāmaḥ and gacchati rāmo vanam all mean "Rāma goes to the forest". In English, The forest goes to Rāma means something else entirely.
• "It pleases me". Sanskrit says mahyaṁ rocate, "it pleases me" (literally "it shines to me"), where the person who likes stands in the dative — just like Spanish me gusta and Russian мне нравится (mne nravitsya). English once said methinks and it pleases me, but now says I like.
• "Doing it for oneself". Sanskrit verbs have two sets of endings: one for acting on behalf of another, one for acting for oneself. The grammarians' classic example: yajati, "he performs a sacrifice" (a priest, for his patron), and yajate, "he performs a sacrifice for himself". Spanish does similar work with se (lavar — lavarse).

And one thing Sanskrit shares with Russian but not with English or Spanish: no articles. There is no "a" or "the".

Where Sanskrit is unlike English

There are real differences too:

• Many tenses and moods. Aorist, perfect, two futures, an optative mood for wishes.
• Vowel length matters. a and ā are different sounds and can change the meaning.
• Words fuse. Sounds change where words meet, and in writing words merge into long chains. More on that below.
• Very long compounds. Sanskrit loves words built from five or six roots.
• The verb usually comes last in the sentence.

Sandhi: words that merge

Sandhi means "joining". When two words stand side by side, the sounds where they meet affect each other:

• na + asti → nāsti ("is not");
• su + āgatam → svāgatam ("welcome");
• iti + api → ity api ("thus also");
• rāmaḥ + gacchati → rāmo gacchati ("Rāma goes").

It sounds exotic, but every language has sandhi — the others just do not write it down:

• English writes an apple but a pear; want to becomes "wanna", did you becomes "didja";
• Spanish writes a + el as al and de + el as del, and in verse la amiga is read with one syllable across the join;
• Russian не есть (ne yest'), "is not", fused into нет (net) — exactly as Sanskrit na asti fused into nāsti.

Sanskrit simply writes down what everyone says.

Compounds

Sanskrit builds words from several roots, as English does with rainfall or redhead. The grammar sorts compounds into types, and each is familiar:

The word bahuvrīhi is itself an example of its type: bahu-vrīhi means "one who has much rice", that is, "a rich man". Just as a redhead is not a head but a person who has red hair.

Sanskrit in our languages today

Sanskrit has given the world's vocabularies not only cognates but direct loans, often through Persian, Arabic or Hindi.

Caturaṅga means "four-limbed", after the four arms of an ancient army: infantry, cavalry, chariots and elephants. Its first part is catur, "four". Spanish ajedrez comes from it through Persian and Arabic; English chess has a different name, but the game itself came from India.

English juggernaut, "an unstoppable force", is a distortion of Jagannātha, "Lord of the universe" — the British were overwhelmed by the huge chariots of the Ratha-yātrā festival in Purī.

And karma, yoga, mantra, guru, nirvana are now at home in almost every language on earth.

Try it yourself

Take half a verse of the Bhagavad-gītā, 2.20:

na jāyate mriyate vā kadācin

• na — not;
• jāyate — is born (related to Latin gens and English kin);
• mriyate — dies (related to mortal and murder);
• vā — or;
• kadācit — ever.

"[The soul] is never born and never dies." The same verse ends na hanyate hanyamāne śarīre, "it is not slain when the body is slain", where śarīre, "in the body", is in the locative case.

On this site every verse comes with a word-by-word translation: each word of the original is explained beneath it. Open any verse and look for familiar roots. You will be surprised how often you find them.

The language in which God is spoken of

Sanskrit is called deva-bhāṣā, "the language of the gods". The reason is not just its age. The Vedas are in Sanskrit, and the tradition holds that the sound of the Vedas was not invented by people but heard — which is why the Vedas are called śruti, "that which is heard" (from the same root śru as English loud). The holy name, the mantra and the verse are spoken in Sanskrit because what they speak of is present in the sound itself.

But for an English, Spanish or Russian speaker there is also a simpler joy. When we read tat tvam asi — "that thou art" — we are not hearing a foreign tongue but a very old voice of our own speech. Sanskrit is no stranger. It is an elder relative who has kept what the rest of us lost along the way.