The Barbarian Tongue
Śābara means 'in barbarian speech'. Twelve hundred years of claiming power in languages that were never supposed to carry it.
Somewhere in the sixteenth century Tulsidas wrote four lines in Awadhi about ill-matched letters with no meaning and no japa, and the popular literature of four centuries has treated those lines as the charter of a whole genre of mantras. They are not a charter. They are a poet defending his right to write an epic in the language his neighbours spoke. This treatise starts there, with the most-quoted verse in the field read back inside its own sentence, and works outward: to the lexicon, where śabara is a people, a dialect and a kind of sorcery collapsed into one word; to a Prakrit novel of 779 CE in which Śabara-ness is already a costume that other men put on; to the Mimamsa doctrine of eternal sound and the Tantric apparatus of initiation that this corpus is said to be exempt from; to a frequency count on one Delhi paperback; to the Saora of southern Odisha, who are still there and whose name this is; and to a ledger of harm in which the word imposed from outside is still doing damage. It is a documentary history of a category, kept apart at every point from what the tradition holds, and it says on the page where the record runs out.
Start readingGoes deeper into The Initiate's Path. New to this ground? That path's 29 shorter readings come first.
Before the treatise begins
This is a documentary history of a word and of the mantra corpus that carries it. Śābara, in the popular Hindi literature of the last century and a half, names a class of vernacular incantations said to work without initiation, without Sanskrit, and without the long preparation the classical manuals demand. The claim is old, loud and almost entirely unexamined. This treatise examines it: where the name comes from, what the earliest datable witnesses actually say, what the printed corpus actually contains when it is counted, and what the social record shows happening around the people who use this vocabulary today.
It is not for a reader who wants to use any of it. There is nothing here to use.
That needs saying without softening, because the softened version is what everybody writes. The corpus is analysed and its history is reconstructed, and nothing in this book can be performed from. No working mantra is printed. No ritual procedure is described. There are no counts, no materials, no timings, no substitutions, no seed-syllables set out to be assembled. Where a source’s next line is an instruction, this book stops at the line before and says so on the page. Where a verse is quoted in half, the truncation is declared in the citation itself rather than left to look like an accident.
The reason sits at chapter 20, and it is not squeamishness. Chapter 20 is a ledger. It counts 2,468 murders recorded between 2001 and 2016 in which witch-persecution was the motive, on the National Crime Records Bureau figures as reported by ActionAid and the Odisha State Commission for Women. It reads the only statutory definition of ojhā, the village ritual specialist, in Indian law. It follows what the field studies say about how an accusation actually forms, which is that the specialist tends to confirm a quarrel that already exists rather than start one. A book about this vocabulary that also handed out usable procedure would be a book that had read its own last chapter and learned nothing from it. The operative line is the price of writing about this at all.
The thesis, in both of its forms
There are two claims here and only one of them is settled.
The weak form is a fact about a word. The lexeme śābara carries the sense “in barbarian speech.” The lexical and dramaturgical tradition puts it there: the Nāṭyaśāstra at 17.49 lists the vibhāṣās, the rough stage dialects, by the peoples they belong to, with the Śabara among them, and at 17.55 assigns that dialect by occupation. Monier-Williams glosses śabara as a mountain people and records the later slide into any savage or barbarian at all. This much is not in dispute, and chapter 2 shows why it is one leg of evidence and not two: a nineteenth-century Sanskrit lexicon glossing the word is downstream of the same dramaturgical tradition.
The strong form is a claim about this corpus: that these mantras bear this name because they are cast in the wrong register, in the speech of people whose speech was held to be no speech. That is contested, and the rival has a name. Teun Goudriaan proposed instead that the corpus’s name may run back to Saṃvara or Śambara, which would make it a lineage or deity name and not a statement about language at all. The suggestion sits in Goudriaan’s own catalogue, which is the source this treatise leans on hardest of all. Chapter 3 prints it, names it as a live rival, and sets out three tests that would decide between them.
This treatise commits to the speech reading. It also says plainly what would overturn it: a manuscript or citation attaching the corpus to Saṃvara or Śambara earlier than the register usage is attested; a pīṭha list or Śaiva source in a named recension that fixes the name to a place or a deity; or any pre-modern text that uses śābara of these mantras while glossing it by descent rather than by tongue. None of those has been found here. Any of them would change the book.
What the chapters establish
So, before the reader starts, what the twenty-two chapters reach.
The name is older than anything that now carries it, and it arrives already borrowed. Śābara-vidyā is in the record in Uddyotanasūri’s Prakrit Kuvalayamālā of 779 CE, and the people practising it there are Vidyādharas dressed as Śabaras. Śabara-ness is a costume in the earliest securely dated attestation, put on by men who are not Śabaras, roughly two centuries before the earliest fixed enumeration of the six acts that anyone can point to. That pattern never breaks. Jain narrators use the name for a disguise, a Nāth charter uses it for a claim about language, a poet in Awadhi uses it to defend his own book, the Bombay and Banaras presses use it to sell cheap grimoires, and a marketing industry uses it now. Every one of those repurposings was profitable for somebody who was not a Śabara.
The definition the whole field runs on is a colonial gazetteer sentence about a different population. R. V. Russell and Rai Bahadur Hira Lal wrote in 1916, of the Savars of the Central Provinces, that their charms are especially intended to appease the spirits of persons who have died a violent death. G. V. Sitapati quoted that paragraph in 1943 as a tradition come down to the present generation, marking it as hearsay and glossing the term quite differently three lines below. Verrier Elwin reproduced the definition in a footnote of 1955 with no citation of its own, having said in the same footnote that Russell and Hira Lal are not describing the people he studied. Chapter 14 follows the chain link by link, and sets beside it Elwin’s own statement, hundreds of pages later in the same book, that there are no standard Saora prayers, formulas or fixed invocations at all.
The corpus can be counted, and the counting is the most concrete evidence in the book. Chapters 10 to 13 work through a popular Delhi compilation and eight independent printed compilations by the same published method, and chapter 10 prints the tables. Two words carry the register between them. Duhāī is the outcry made in another’s name, ān is the binding invocation, and the Hindī Śabdasāgara glosses each of the two by the other, so in the standard dictionary of the language the appeal and the oath are one thing rather than two. Both run through every one of the eight compilations, and the book this research began from turned out to sit at the bottom of that range rather than in the middle of it, which is the opposite of the way a case study usually fails. Two controls from the same bazaar shelf, an Indrajāla compendium and an Uḍḍīśatantra, score at or near zero on the same terms. So the register belongs to these compilations specifically and not to Hindī occult printing in general, and it is the register of a speaker with no court to go to, who cries out in a name that is not his and binds by it anyway.
The classical grid does not organise any of it. Chapter 21 lays the ṣaṭkarma, the six acts, against the vernacular material and finds two literatures sorted on different axes: the six classify by intended effect on a third party, and this corpus classifies by whose warrant is being claimed. Ṣaṭkarma vocabulary appears in 22.4 per cent of entries as adjectives and never as architecture.
The scholarship on the corpus is thinner than its readership by orders of magnitude. Teun Goudriaan and Sanjukta Gupta gave it two pages in 1981 and that remains close to the whole of what a careful reader can stand on. David Lorenzen, the field’s principal authority on the Kāpālikas, does carry Śabara-tantra in his index and discusses it at pages 37 and 42 to 43, where he allows that the tradition may have some factual justification. Raphaël Rousseleau establishes what Alexander Cunningham did in 1871, when he identified the Śabaras of the Sanskrit and Greco-Latin sources as one scattered people and turned a category into an ethnic name. Beyond those, the discoverable literature is a shop front: uploads, storefronts, priced services, and nothing a reader can check.
And the harm in this world does not attach to a book. Chapter 20 assembles the ledger and finds the documented injury attaching to a role and to an accusation, with the ritual specialist confirming a conflict already under way rather than starting one.
Where the record has an edge, the chapters say where the edge is, and where two serious researchers disagree the divergence is named and left standing. That is not modesty. It is the shape of the evidence.
The tradition does not need this treatise. The Saora, whose name this is, need it least of all.
It is offered anyway, with the record kept open.