Śabda eternal

There is a school of Sanskrit philosophy whose central claim is that nobody wrote the Veda.

Not that the author’s name was lost. Not that it was too old to recover. That there was no author at all, human or divine, and no moment of composition for an error to enter at.

Now look at what that school actually produced.

It produced a shelf of named men. Jaimini, who set down the ritual aphorisms it exists to interpret. Upavarṣa, whom this tradition and Śaṅkarācārya alike speak of as Bhagavān Upavarṣa, the reverend one. The commentator this chapter is about, who wrote the line-by-line reading of those aphorisms that the whole school has stood on ever since. Kumārila Bhaṭṭa and Prabhākara Miśra, who later split the school into two wings and gave the wings their names, and who both wrote on that commentator rather than around him. Every one of them is a person, with a position, with rivals, with a place in a sequence.

A tradition that held speech to be uncreated kept a careful roll of the people who explained it.

That is not an embarrassment, and it is worth being exact about why. The Veda has no author. The explanation of the Veda has authors, and Mīmāṃsā was scrupulous about which was which. The foundational commentary itself is thought to carry, embedded in it, a long stretch by an older hand, and the school did not absorb that hand quietly. It named him, and revered him.

So the man at the centre of this chapter is a commentator, and commentators, in this school, are allowed to have names.

Somewhere in India, before the year 400 of the Common Era, he sat down with Jaimini’s aphorisms and produced the book on which everything the school did afterwards turns.

We do not know exactly where he sat. We are not certain of his name in any form richer than a title.

He is called Śabarasvāmin. His book is the Śābara-bhāṣya.

Which means that the word at the centre of this treatise, śābara, the word for spells in barbarian speech, is also the name attached to one of the most rigorous defences the tradition produced of the proposition that Vedic speech must be Sanskrit, must be exact, and cannot be otherwise.

That is a coincidence. This chapter says so plainly and then sets it down, because the two positions deserve to be put in one sentence, and because a reader who searches the word will hit both and is owed an account of which is which.

The reason this school arrives here at all is a question chapter six left on the table. Every claimant to the śābara corpus was doing the same thing. The Nāth author routes it back through the Nātha, the eighteenth-century Sanskritist routes it into the Kāpālikas, the charter has the revelation handed down in five vernaculars by the teacher of Matsyendranātha. Not one of them says: we invented this. Every one of them says: this is ours to authorise.

People secure permissions where something stands in the way.

What stood in the way was a theory of language, worked out in Sanskrit across many centuries by some of the most careful minds the subcontinent produced, which held that the power of a Vedic word sits in its exact phonemes and nowhere else. Everything Movement I has described, the vernacular charter, the bilingual genre, the mantras that work in the mouths of people with no Sanskrit, occupies ground that theory declares empty.

This chapter builds the theory, and builds it at its strongest, because the contrast is only worth drawing if the thing being contrasted is formidable.

One word doing two jobs

Take the collision head on, and then leave it alone.

Śābara as this treatise uses it is an adjective formed from Śabara, the name of a forest people and, by long extension, of anyone whose speech the Sanskritic centre could not parse. In that sense it means “of the Śabaras,” and applied to speech, “in barbarian tongue.” Chapters two and four traced that usage from the lexicons back to a Prakrit narrative whose closing verses count out the Śaka year seven hundred less a day, which is 779 of the Common Era.

Śabarasvāmin is a personal name. The svāmin is an honorific. The man it belonged to wrote on the Mīmāṃsā Sūtras of Jaimini, and his commentary is the hinge between the sūtras and everything the school did afterwards. Kumārila Bhaṭṭa and Prabhākara Miśra, the two great systematisers who later split Mīmāṃsā into its rival wings, both wrote on Śabara rather than around him.

Now be exact about what the collision is, because a loose account of it runs in the flattering direction.

These are not two look-alike words. शबर is one word. शाबर is its ordinary vṛddhi derivative, the lengthened grade Sanskrit forms when it turns a noun into an adjective of belonging, and it is the same derivation on both sides of the collision. The Śābara-bhāṣya is the commentary belonging to Śabara. A śābara mantra is an incantation belonging to the Śabaras. One word, one regular derivation, two jobs that have nothing to do with each other.

Note what that regularity costs the argument elsewhere. The alternation between शबर and शाबर across chapters four and five is not evidence that the word had no stable home. It is grammar working correctly, and no weight can be put on it.

Nothing in the research for this treatise connects the commentator to the mantra corpus. No source consulted proposes an etymological or historical link between the commentator’s name and the mantra category, and none is manufactured here. The coincidence verdict stands on that point.

The wider question, whether the commentator’s name carries any ethnic or occupational sense at all, has an answer, and it sits in the very volume this chapter opens. Umesha Mishra’s critical bibliography, bound into Jha’s 1942 volume, records in its section on Śabarasvāmin a tradition that the commentator’s original name was Ādityadeva, and that he changed it to Śabara when he disguised himself as a forester for fear of Jain persecution. Mishra reports it as something that is said rather than as something established, and this chapter carries it at exactly that weight.

Read what it does and does not do. It attaches the forest sense of the word to the commentator’s name by a route that has nothing to do with the mantra corpus: a story about a man in hiding, taking the name of the people whose dress he borrowed. That is the ethnonym reaching the commentator by an independent traditional road. It leaves the corpus exactly where it was, unconnected.

What would still close the wider question: the Mīmāṃsā biographical tradition surveyed at source, and the Sanskrit onomastic literature on names of the Śabara type.

Dating a man by the people he could not have met

The Mīmāṃsā apparatus in this chapter rests on a single Indian scholar whose work has been in print for ninety years, read directly rather than through summary.

Ganganatha Jha translated the Śābara-bhāṣya into English in three volumes for Gaekwad’s Oriental Series, the first volume published at Baroda in 1933. In 1942 he published Pūrva-Mīmāṃsā in its Sources for Benares Hindu University, in the Library of Indian Philosophy and Religion whose general editor was S. Radhakrishnan, with a critical bibliography by Umesha Mishra. Both books have been read directly for this chapter. Jha is the standard English authority on this school, he is closer to the text than any Western secondary work this chapter could reach, and he governs here.

A word on quoting him. Under the term that governs in India, sixty years from the start of the year after the author dies, Ganganatha Jha died in 1941 and his work entered the public domain in India on 1 January 2002. He is quotable, and the printed copyright notice in the 1942 volume records a term that has expired. Everything taken from Jha in this chapter is nonetheless paraphrase, because a quotation needs the page in hand.

And a standing note on page numbers, which applies to this whole treatise and not only to this chapter. S. B. Dasgupta’s Obscure Religious Cults exists in a University of Calcutta first edition of 1946 and in a later edition from Firma K. L. Mukhopadhyay, and through the material chapters three, four and six use, the two paginate about thirty pages apart. The numbers in those chapters are the 1946 first edition’s own printed running heads, which is the edition all three chapters name. Lorenzen and Goudriaan both cite Dasgupta by the Mukhopadhyay numbers. A reader moving between this treatise and the standard literature will therefore meet two paginations of one book, roughly thirty pages adrift, and is owed the warning.

The principle behind that warning is general. A page number is meaningless without the edition it belongs to, and a citation that names one edition while paginating from another fails silently, because everything about it looks right. A page number in this treatise therefore carries two things a source line alone does not supply: a verification status, and a named edition. A source may be consulted and its page still unverified. Where a page has been checked against a running head or a printed page marker it is given plainly. Where it has not, the sentence says so. And where a book has been printed twice, the treatise says which printing it is holding.

Start with the date. No dating position on Śabara is attributed here to P. V. Kane, because no page of Kane has been read for one. Jha argues the question out in the 1942 volume, and his reasoning can be followed. The traditional dating goes first: it makes Śabara both a contemporary of Vikramāditya in the first century before the Common Era and the father of the astronomer Varāhamihira, whom scholars place in the fourth century after it. The tradition contradicts itself. What Jha puts in its place is a position after Upavarṣa and before Prabhākara, Kumārila, Maṇḍana and Śaṅkara, with a lower bound from inside the Bhāṣya, where under Mīmāṃsāsūtra 6.1.12 Śabara quotes a passage that occurs in the Manusmṛti.

Śabara lived before 400 CE. That is the conclusion, and it is a hard upper bound with the working shown. On where he lived Jha’s evidence pulls two ways, Ujjain against the north, and this chapter takes from that only what it said at the start: we do not know where he sat.

That is better than a range with a famous name on it.

The second thing Jha closes is an irony at the heart of the school. The Śābara-bhāṣya is thought to carry, embedded in it, a long section by an earlier hand, the Vṛttikāra. Jha sets out who that is likely to be. The oldest commentator on the twelve chapters known at all is Upavarṣa, whom Śabara and Śaṅkarācārya both speak of reverentially as Bhagavān Upavarṣa, and Upavarṣa has been generally identified with the Vṛttikāra whose views the Śābara-bhāṣya sets out in detail, though Jha records that doubts have been raised about the identification.

So the foundational book of the school that made authorlessness a virtue does appear to carry a stretch of somebody else’s argument inside it, and the school named him and revered him rather than hiding him. Jha supplies that from the Sanskrit directly, and Erich Frauwallner’s reconstruction is not needed to reach it.

The relation that was never agreed

Everything in Mīmāṃsā rests on a claim about how a word is joined to what it means.

The commonsense answer, then and now, is convention. A community agreed that a certain noise would point at a certain thing. Change the community and the noise changes, which is why the world has many languages and no way to guess a foreign one.

Mīmāṃsā denies this at the root. The statement sits early in the first chapter of Jaimini, and it is a sūtra of the school, a line of philosophical argument and not a liturgical formula: autpattikas tu śabdasyārthena sambandhaḥ. The relation of the word with its meaning is autpattika, inborn, there from the origin and not made. The line is Mīmāṃsāsūtra 1.1.5. Jha translates it in the first volume, rendering autpattika as inborn.

Śabara’s own gloss on the word is one sentence long and settles what it means. Jha renders it with the word constant: by autpattika, Śabara says, what is meant is constant, and what is figuratively spoken of as origin is really existence, and the relation between a word and its meaning is inseparable.

The word Jha prints there is constant, and the difference between that and eternal is worth holding on to, since this chapter turns on the term. Nitya in the Sanskrit will carry both. Where the stronger word is wanted below it is used of the varṇas, which is where the school itself uses it.

What the claim buys is enormous. If the bond between a word and its meaning were made by people, then people could have made it wrong, and any statement carried in words would only be as reliable as the people who set up the code. If the bond is original, then a correctly formed word cannot fail to mean what it means. Meaning stops being a social fact and becomes a structural one.

The rival schools saw the stakes and pushed back. The Naiyāyikas held the relation conventional and traced the convention to Īśvara, which let them defend scripture the ordinary way, by the reliability of its author. Buddhist and Jain logicians attacked the apparatus from outside. The debate stayed live for centuries and was never settled by agreement, only by the fading of the institutions that hosted it.

This treatise does not adjudicate it. It records that Mīmāṃsā took the harder road, deliberately.

The book with no author

The second pillar follows from the first and is the more famous.

The Veda is apauruṣeya. Not of a person. It has no author, human or divine.

The arguments are worth stating in their own terms. There is no tradition of an author, and a composition of that scale, transmitted with that much care, would have carried its author’s memory with it if it had one. The names attached to the hymns are the names of ṛṣis, and the tradition understands a ṛṣi as one who saw, a seer of what was already there, standing to the Veda roughly as an astronomer stands to a star.

Then the sharper move. Do not solve the problem by giving the Veda to God. A divine author is still an author, still a person, still a locus of intention, which means still something whose reliability has to be argued for from outside. Put God behind the Veda and you have not grounded it, you have made it the report of a witness, and now you must establish the witness.

Jha’s summary of the Bhāṣya in the 1942 volume makes the same point at the level of the word: the relation between a word and what it denotes cannot be regarded as created by any person, and Jha’s phrasing is that it cannot be created even by God, because no such creator of word-relationship is vouched for by any means of cognition.

Take the author away entirely and there is nothing left to doubt. No composer, no intention, no possibility of error entering at the moment of composition, because there was no such moment.

That is not a piety. It is a piece of engineering, and it is one of the most economical solutions to the problem of scriptural authority any tradition has produced.

What “eternal” is doing

The school’s technical name for its central thesis is the eternality of śabda, and the word is doing something precise.

The claim is not that a particular recitation is eternal. A recitation begins and ends. The claim is that the varṇas, the phonemes themselves, are eternal, and that an act of speech does not produce them. It manifests them. The sound you make reveals a phoneme that was already there, in the way a lamp reveals a pot in a dark room without creating the pot.

The consequence is that the Veda is not a text in the sense a modern reader means. There is no original manuscript that later copies approximate. There is an eternal arrangement of eternal sounds, and every correct recitation is a complete instance of it.

Here Mīmāṃsā fought its neighbours as hard as its opponents, and the fight does not have to be chased as far as Kumārila’s Ślokavārttika. The debate is inside the Śābara-bhāṣya itself, on this very sūtra, and Jha prints it in the first volume of his translation. Jha himself, in the 1942 monograph, sends the reader to exactly that passage of his own translation.

The test case is the word gauḥ. The letters ga, au and the visarga constitute it, says the revered Upavarṣa, because what people call a word is what the ear apprehends, and those letters are what the ear apprehends.

Against him Jha’s text names the objector directly: the Sphoṭavādin Grammarian. On those terms, he answers, no cognition of the meaning is possible at all. The letters arrive one after another and never stand together, and above them there is no single composite whole for the meaning to follow from. So the word must be something other than its letters, and it is from that other thing that meaning follows.

The Bhāṣya’s answer is the load-bearing sentence of the whole school’s phonology. Each letter, as it is uttered, leaves an impression behind, and what brings about the cognition of the meaning is the last letter together with the impressions of all the preceding ones. The phonemes stay the units. Nothing above them is needed.

Two schools, both committed to the sanctity of correct speech, disagreeing about where in speech the sanctity sits. And one point of substance about attribution. Śabara’s opponent is not named in the Bhāṣya. He is the Grammarian who holds the sphoṭa, and Jha calls the position the Doctrine of the Sphoṭa without attaching a person to it. Bhartṛhari’s Vākyapadīya is the classical exposition of the doctrine, and it is a separate matter from who Śabara was answering.

Why grammar became a soteriology

If efficacy sits in the phoneme, correctly produced, then producing it correctly is not a matter of good manners. It is the mechanism.

This is where a set of disciplines that look, from outside, like technical philology reveal themselves as something else. Śikṣā, phonetics, stands first among the six limbs of the Veda, ahead of ritual procedure, ahead of grammar, ahead of etymology and metre and astronomy. It is first because nothing downstream works if the sounds are wrong. The Prātiśākhya literature, one manual per recensional tradition, specifies articulation, junction, quantity and accent at a level of detail that no descriptive linguistics would reach again until the modern period.

And the śikṣā texts state the stake without hedging. The doctrine, paraphrased here rather than reproduced, is that a sacred formula defective in accent or in phoneme, wrongly employed, fails to say the thing it was meant to say, and the misfired word turns on the man who commissioned the rite and destroys him. A verbal weapon, pointed the wrong way. The verse standardly cited for it is Pāṇinīya Śikṣā 52.

Patañjali’s Mahābhāṣya, the great commentary on Pāṇini, opens onto this ground almost immediately. The Paspaśāhnika, its introductory section, asks why one should study grammar at all and answers with a list of purposes, of which the first is the preservation of the Veda. Then it produces the classical illustration, which it draws from Taittirīya Saṃhitā 2.4.12: a compound whose accent alone decides which of two contending parties is destroyed, misaccented at a sacrifice, with the result falling on the sacrificer.

The compound is not given here. This treatise describes the doctrine of correct utterance and does not demonstrate it, and an example that works is a demonstration. The reference is given so that a reader who wants it can find it; the words are withheld on purpose, and withholding the words is not a reason to withhold the reference.

But hold what the example means. Two utterances, phonemically identical, differing only in where the pitch falls. One produces the intended result. The other kills the man who paid for the rite. No charitable reading, no allowance for sincerity, no credit for intent. The accent is the operative part.

Which is why the transmission machinery is what it is. The recitational tradition developed a whole taxonomy of preservation schemes, each with its own name: pada, krama, jaṭā, ghana. How any of them is performed is a matter of technique and is left out of this account. What they amount to, taken together, is an error-correcting apparatus for a text held entirely in human memory, built by people with no writing anywhere in the loop.

Nobody builds that unless the phoneme is the thing.

Grammar, in this world, is soteriology with paradigms.

The consequence, named, and named at the right size

Now say the part that usually goes unsaid, and then say what it does not cover, because the second half is where this claim is most often stretched.

Mīmāṃsā debated eligibility, adhikāra, directly. The passage is the Śūdrādhikaraṇa, running from Mīmāṃsāsūtra 6.1.25 to 6.1.38, and it is held here in a named printed edition. Jha sets out the settled conclusion in Pūrva-Mīmāṃsā in its Sources, sending the reader on to the relevant stretch of his own Bhāṣya translation.

The argument has two legs and both are technical. The agnihotra and rites of that kind require consecrated fires, and the text enjoining the fire-installation names the Brāhmaṇa, the Kṣatriya and the Vaiśya and does not name the Śūdra; without the installed fires, the acts cannot be performed. And the text on initiation names the same three, so the learning that sacrificial performance requires cannot be acquired in the prescribed way. Jha records that Prabhākara, in the Bṛhatī, endorses the Bhāṣya’s summing-up on exactly the fire-installation ground.

Two exceptions follow immediately, and they matter because they show what kind of system this is. Under sūtras 6.1.44 to 6.1.50 and 6.1.51 to 6.1.62, the Rathakāra and the Niṣādasthapati are held entitled to perform sacrifices, the Rathakāra being defined by a specific parentage. A blanket ranking of persons would have no room for that. A technical system of eligibility, built out of what particular Vedic texts do and do not name, has room for exactly that.

Beside the jural lock sits a linguistic one, and it operates on its own. If power is in the phoneme, and the phonemes are Sanskrit, and Sanskrit is a language that must be learned formally over years from a teacher who has it, then most mouths on the subcontinent could not, in principle, utter an effective Vedic word. Not because they were insincere. Because the mechanism required a phonological system their communities had no access to.

Now the limit, stated as flatly as the claim. Mīmāṃsā is a school of Vedic exegesis, and its doctrine of ritual competence concerns Vedic ritual. It does not legislate on Purāṇic recitation, on bhakti devotion, on Āgamic or Tantric mantra, and it is largely uninterested in them. It is a short step from here to saying that the doctrine told most of India that what came out of their mouths could not structurally be a mantra. That is an inflation, and it is the inflation this book’s whole antagonist-structure would most like to be true. What the doctrine says is that what came out of their mouths could not be the operative speech of a Vedic rite. That is a large enough thing to be going on with.

In fairness to the Mīmāṃsakas: they were not ranking souls. They were describing ritual competence in a technical system, in the spirit of a surgeon describing who may hold the knife. Its authors would have been puzzled to hear the doctrine read as an insult.

The consequence is the consequence regardless. And the consequence is where the śābara corpus stands. Not as an answer to Mīmāṃsā, because nothing establishes that it was answering anything. As the thing that occupies the ground Mīmāṃsā describes as empty: potent speech, in mouths the Vedic system does not enfranchise, in languages it does not recognise.

Correct speech and barbarian speech are one system

The last turn is the one that makes the whole apparatus visible.

The same intellectual culture that specified correct speech to the level of the accent also specified incorrect speech, and it did so in the same registers of technical prose.

Chapter two established the evidence and this chapter recalls it. The Nāṭyaśāstra, in the chapter governing stage language, allots speech forms to character types: the polished dramatic languages to the high roles, and the rough dialects, the vibhāṣās, elsewhere. At 17.49 the śavara speech stands in the list of seven vibhāṣās. At 17.55 the Śāvarī is assigned on stage to herdsmen and forest-rangers. Two cautions, both of which cut against overreading it. The text spells the word śavara and śāvarī, with the palatal but without the long vowel this treatise otherwise uses. And no verse says that Śabara characters speak a vibhāṣā: what the verses do is put the dialect on the list and give it an occupational range on stage.

A numbering conflict a reader will hit. Hazārīprasād Dvivedī, in Hindī Sāhitya kī Bhūmikā, third revised edition 1948, cites the same material at 17.48: after Bharata’s seven Prakrits, Māgadhī, Āvantī, Prācī, Śaurasenī, Ardhamāgadhī, Bāhlīkā and Dākṣiṇātyā, Bharata separately names the speech of śabara, ābhīra, cāṇḍāla and the like. This is a recension difference and not an error by either party, and this treatise follows the numbering of the edition chapter two used.

Dvivedī’s point there is worth more than the numbering. He reads Bharata as describing literary language, and he brings Daṇḍin’s Kāvyādarśa 1.36 alongside it, where the speech of the Ābhīras and the like is called apabhraṃśa in poetry, together with Daṇḍin’s observation that Apabhraṃśa poems have their own structural unit, which shows that Apabhraṃśa poetry already existed in his time. So the list of rough dialects is a list of things a poet or a dramatist may write in, not a survey of how anybody spoke. That is an Indian scholar, writing in Hindi in the 1940s, putting the vibhāṣā material where this treatise needs it: in the history of literary registers.

Romila Thapar’s study of the image of the barbarian in early India, published in 1971, treats mleccha as a category for those outside the ritual and linguistic order, with unintelligible speech at the centre of the definition. Her wording is that the earliest distinction made by the Aryan speakers was a linguistic distinction. Her term is kept as she wrote it, because substituting “Sanskrit-speakers” for it changes what is being claimed. The support for this is the article’s abstract and publication record. The article itself has not been read by anyone on this project. The word mleccha is built on a root for indistinct speaking, and that Śabara runs in the same channel is this treatise’s inference, hedged in chapter two and hedged here.

So the technical vocabulary of correct speech and the technical vocabulary of barbarian speech are two ends of one measuring device. Śikṣā at one end, vibhāṣā and mleccha at the other, and the same culture holding the instrument.

Which is why the naming collision, coincidence though it is, is worth the page. Śabarasvāmin’s book defends the end where the phonemes are exact. The śābara mantras claim power at the end where they are not. One system, two poles, and one word that happens to name both ends of it.

What happens after

What happens next is a historical process rather than a philosophical one: the regional languages of South Asia become literary, and then liturgical, and take over functions Sanskrit had monopolised.

Sheldon Pollock, The Language of the Gods in the World of Men, University of California Press, 2006, is the standard work on that process.

What this chapter does have in hand is Dvivedī, who is describing the same shift from inside the vernacular tradition, in Hindi, sixty years earlier, and who has been read directly. He is not a substitute for Pollock and is not offered as one.

Chapter one has already shown Tulsīdās riding the process explicitly, in the opening pages of his own book, arguing that a vernacular composition can carry power, and reaching for the śābara mantras as he does it. Whether he knew he was answering Mīmāṃsā is not something any source consulted here establishes. What the verse shows is the argument. The target is a separate question, and it is open.

Chapter eight goes to the other great theory of mantric power, the Tantric one, which locates efficacy in the phoneme too and reaches an entirely different place with it.

The doctrine is built. It is coherent, economical, and defended by first-rate minds over many centuries, and Ganganatha Jha spent his working life making it readable in English without softening a line of it.

Within its own domain, it leaves almost nobody a way in.