Who was allowed to learn Sanskrit?

Sometime in the fifth or sixth century, a Jain logician named Siddhasena Divakara made a proposal to his community. The scriptures, he suggested, should be translated into Sanskrit.

He was expelled for twelve years.

The story comes from the Prabhavakacarita, a hagiography rather than a chronicle, so treat the sentence as what the tradition remembers rather than as a court record. But a community does not invent a memory like that about nothing. In the same account Siddhasena loses a public debate because his Sanskrit is too grand for the people listening, while his opponent speaks plainly and wins them.

Twelve years, for wanting to put scripture into Sanskrit.

That is difficult to square with the idea most people carry about Sanskrit: that it was India’s universal language, the tongue everyone aspired to, the common inheritance. Something more complicated was going on, and it can be traced in writing, in three separate places.

The refusal in the Vinaya

The Pali Vinaya, the monastic code of the Buddhist canon, records an episode at Cullavagga V.33.1.

Two monks of brahmin birth come to the Buddha with a complaint. Monks are entering the order from every background, and each is reciting the teaching in his own way of speaking, and this, the two say, is corrupting it. They propose a remedy. Put the Buddha’s word into chandas, the elevated brahminical register of the Veda.

In the public-domain translation made by Rhys Davids and Oldenberg for the Sacred Books of the East in 1885:

“These corrupt the word of the Buddhas by (repeating it in) their own dialect. Let us, Lord, put the word of the Buddhas into (Sanskrit) verse.”

“How can you, O foolish ones, speak thus…? This will not conduce, O foolish ones, either to the conversion of the unconverted, or to the increase of the converted…”

“You are not, O Bhikkhus, to put the word of the Buddhas into (Sanskrit) verse. Whosoever does so, shall be guilty of a dukkata. I allow you, O Bhikkhus, to learn the word of the Buddhas each in his own dialect.”

A dukkata is a minor offence, the lightest grade in the code, cleared by confession. The refusal was not left as an opinion. It was made a rule with a penalty attached.

Two things in that passage are still argued over, and both are worth stating rather than smoothing.

The first is what exactly was refused. The Pali word is chandaso, and Rhys Davids supplied “(Sanskrit)” in brackets because it is not in the text. Franklin Edgerton rejected both of his readings, holding that chandaso means texts in a certain language or dialect, “and more precisely in the Vedic language (whether prose or verse).” A minority of scholars have read the whole episode as concerning recitation style rather than language at all.

The second is what was permitted instead. The phrase is sakaya niruttiya, and the weight falls on nirutti, a way of expressing or naming something. It can be read as “in his own dialect,” meaning the learner’s, or as “in his own idiom,” meaning the Buddha’s. Buddhaghosa, the great fifth-century commentator, took the second reading and glossed it as the Magadhi speech of the Buddha himself. Bryan Levman, surveying the field, lists Rhys Davids, Edgerton, Lamotte, Brough and von Hinüber on one side, and Buddhaghosa, Geiger, K. R. Norman and himself on the other. Geiger’s argument is the elegant one: saka, “own,” takes its referent from whoever is speaking, the way “speak in your own language and I will answer in my own” means two different languages inside one sentence.

So the refusal is certain and the remedy is contested. That is the honest state of it, and it is more interesting than either side winning. There is more on the wider relationship in Hinduism and Buddhism: the difference.

The Jain route to the same place

The Jain tradition arrived at the same refusal by a completely different road.

The Jain canon is in Ardhamagadhi Prakrit, and the Samavayanga Sutra explains why with a miracle rather than a rule. Mahavira preaches in Ardhamagadhi, and that Ardhamagadhi, as it is spoken, transforms itself into the language of every hearer present, Arya and non-Arya alike, for their benefit and welfare.

Notice the difference from the Buddhist text, because flattening the two into one story loses the interesting part. The Vinaya legislates against a prestige register and attaches a penalty. The Samavayanga makes no rule at all. It claims a vernacular that is already universally intelligible, which is a claim about what Mahavira’s speech does rather than about what monks may not do.

Both arrive at the same practical place. Neither language of revelation is to be Sanskrit.

And Siddhasena, twelve years in exile, is what the position cost somebody.

Both of them came back

Here is the part that undoes any tidy version of this.

Both traditions ended up writing in Sanskrit anyway.

Umasvati’s Tattvartha-sutra is the first significant Jain work in Sanskrit, and the reason is straightforwardly competitive: Jains needed to dispute with rival schools on ground those schools would accept. Buddhist literature drifted so far back toward Sanskrit that Franklin Edgerton had to invent a name for the result, Buddhist Hybrid Sanskrit, in the grammar and dictionary he published in 1953. Edgerton’s account of why is blunt. Some north Indian Buddhists abandoned their original principle of using genuine vernaculars and partially yielded to the prestige of the learned language of their brahmin neighbours. He notes that the texts kept changing in one direction only, always toward greater Sanskritisation.

And Ardhamagadhi did not stay a vernacular either. Paul Dundas points out that the surviving Ardhamagadhi was certainly not the speech Mahavira actually preached in, and that it evolved into a specifically Jain scriptural dialect, a sacred language of its own kind. He asks directly whether, from the beginning of the common era, anybody outside the learned monks still had the capacity to understand it.

The language chosen because ordinary people could understand it became a language only specialists could read.

That is the argument of this piece, not a failure of it. Prestige is what happens to a language that carries something people revere, and it happens whichever language is chosen. Refusing Sanskrit did not prevent it. It only delayed it, and produced a second guarded language in the meantime.

What the law books say

Now the harder half, and it needs care, because it is quoted carelessly in both directions.

The Dharmashastra texts set out who may study the Veda. They must be read for what they are: normative doctrinal literature stating what ought to be, rather than a record of what was practised or enforced. Scholars of the genre, Patrick Olivelle among them, are firm about that distinction, and any account that treats these texts as a description of Indian society is misreading them at the level of genre.

Within those texts, access to Vedic learning runs through the upanayana, the initiation that makes a man dvija, twice-born. Manu counts three twice-born varnas and places the Shudra outside them as ekajati, once-born. He directs that recitation not take place in the proximity of Shudras. Of women he says, in Ganganatha Jha’s translation, that the sacraments are to be performed for them at the proper time and in the proper order “but without the Vedic formulas,” and he states the rule flatly at 9.18: for women there is no dealing with the sacred texts.

The Gautama Dharmasutra, which is the earlier text, is harsher, and the difference is worth marking because the two are constantly conflated. Manu excludes by ritual status. Gautama prescribes punishment: at 12.4 to 12.6, a Shudra who intentionally listens to the Veda is to have his ears filled with molten tin or lac, one who recites it is to have his tongue cut out, one who remembers it is to have his body split.

Those mutilations are Gautama’s. They are not in Manu.

That correction matters and it cuts both ways. People attacking the tradition attribute Gautama’s penalties to the far more famous Manu, which overstates Manu. People defending it point out that Manu says no such thing, which is true and which quietly leaves Gautama unmentioned. The accurate statement is that one text excludes and an older one punishes, and both are in the corpus. What the tradition says about the wider structure is taken up in the caste system.

There is a connection here to something the site has covered elsewhere. A tradition that restricts who may hear a text, and then transmits it orally for centuries with no manuscript to fall back on, is doing two things at once: guarding the text and guaranteeing it. The apparatus described in how the Vedas were preserved is the second of those. The restriction is the first, and they are the same system seen from two sides.

The evidence from the stage

The same stratification shows up in the theatre, where it is a convention of craft rather than a rule of law, which makes it a useful independent witness.

The Natyashastra assigns Prakrit to women in feminine roles, to persons of low birth, to children and to the deranged, and Sanskrit to sages, to shrotriyas learned in the Veda, and to Buddhist monks. Andrew Ollett, whose Language of the Snakes is the standard modern study of Prakrit, notes that this is the one thing most people know about Prakrit: that it is the language conventionally assigned to women, children and the uneducated.

The rule bends. The Natyashastra allows queens and courtesans Sanskrit as the occasion requires, and religious affiliation cuts across social rank in a way that should complicate anyone’s simple picture, since Jain monks are given Prakrit and Buddhist monks Sanskrit.

But a dramatist writing for an audience has to give each character the language the audience will find plausible in that mouth. Which means the convention is evidence about expectation, and the default tells you where the prestige sat.

And a language held that way stops being a first language. Krishnamurti, following Madhav Deshpande, reports that by the time of Katyayana and Patanjali the first language of brahmins was Prakrit, and Sanskrit had narrowed to ritual use.

Read that again, because it is the fact that reorders everything. By roughly the third century BCE, in the account of the scholars who study this most closely, Sanskrit was no longer the everyday speech even of the community most identified with it.

So how did it reach Java?

That is the question the restriction makes urgent. A guarded liturgical language should stay where it is guarded. This one crossed the Bay of Bengal.

Sheldon Pollock, whose The Language of the Gods in the World of Men is the standard account of Sanskrit’s transregional career, is precise about the mechanism, and the precision is the whole point:

“The Sanskrit cosmopolis was also created by action, though not the actions of a conquest state. It was made instead by the circulation of traders, literati, religious professionals, and freelance adventurers. Coercion, cooptation, juridical control, even persuasion are nowhere in evidence. Those who participated in Sanskrit cosmopolitan culture chose to do so, and could choose to do so.”

He sets that against Latin deliberately. Latin travelled as the language of a conquest state and then of a conquering church. Sanskrit did not. It spread because people who were under no compulsion to use it wanted to, across an area running from Afghanistan to Java, and kings in places the Dharmashastra authors had never heard of took Sanskrit titles because Sanskrit was what political seriousness sounded like.

Pollock is not unopposed, and the objection is worth having. Johannes Bronkhorst argues that Pollock detaches Sanskrit too cleanly from the brahmins who carried it, since inscriptions across South and Southeast Asia testify to their presence, and that a spread with no agent and no motive is under-explained. David Gellner notes that Pollock himself concedes the mechanism of adoption is far from clear.

Take the criticism seriously and the picture is still not one of conquest. It is one of prestige, moving with people who had something others wanted.

Its vocabulary settled into Tamil, Telugu, Kannada and Malayalam as tatsama, words taken over unchanged, and tadbhava, words assimilated to the borrowing language’s sound patterns. It settled into Thai, Khmer, Javanese and Malay.

What the two halves add up to

Put the restriction and the spread beside each other and the shape of the thing appears.

Sanskrit was guarded at home and desired abroad. The same feature produced both: it was a language you had to be admitted to. That is exactly what made access to it worth something, and what made it travel.

A language nobody is excluded from is a language nobody has to acquire.

None of this touches what the tradition claims about Sanskrit in its own terms, which is a claim about revelation and about the nature of sound, and which is a different kind of statement entirely. Nor does it make Sanskrit the ancestor of anything; that question is taken up in is Sanskrit the mother of all languages. What it does is answer the plainer question underneath both.

Sanskrit was never everybody’s language. Two traditions inside India refused it in writing and were drawn back to it anyway. The law books shut most people out of the part of it that mattered most. And it crossed an ocean regardless, carried by people who chose it.

The reverence is real. So is the door.

Both belong in any honest account, and leaving either one out produces a story somebody wanted rather than the one the texts support.