A Theory of Philological Practice in Early Modern India

Abstract

Premodern Indian philology in the sense of ecdotics and interpretation begins late in the scholarly tradition, at the end of the first millennium CE. Although the knowledge form was entirely unsystematized, a philological theory can be derived from commentarial practices. These are reviewed and synthesized across the principal genres, and the implicit theory of the text reconstructed.

Note: The final draft of this essay was submitted in 2016, and (with one exception) could not take account of scholarship published after that date. The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007–2013)/ERC Grant Agreement No. 269804.

1 Sanskrit Philology in Practice But Not in Theory

Although India was among the most densely textualized cultures of the premodern world, we know less about the concrete textual practices of that culture than perhaps any other, certainly for the pre-Islamic period. We know, for example, next to nothing about the slow transition from oral to manuscript culture beginning in the third century BCE (the transition was never discussed in the tradition) or about the interactions between oral culture, which persisted well into the early modern period, and literate culture (works produced by literate writers as late as the seventeenth century could be orally disseminated, sometimes virtually without variation).¹ We know little about the creation of written texts, whether and, if so, how, they were dictated to scribes, for example, or inscribed by the author himself (and what difference this distinction might make). We know little about scribal culture itself, about the processes or social character of manuscript reproduction (whether by individuals or through what, in the case of northwest Buddhism, may have been professionalized scriptoriums), about the book market (which undoubtedly existed for secular books and, in a rather different way, for sacred works such as the Jain canon), about the practices of individual collectors or the organization or functioning of libraries (very few premodern manuscript catalogues have been preserved).² We have little idea how works once created were published, if that is even the appropriate word. All we do know is that publication, libraries, collectors, a market, established mechanisms of reproduction, scribal culture and all the rest did once exist.

This area of darkness extends to the very heart of the study of Indian textual culture, the discipline of philology. Although I am prepared to defend a conception of philology that embraces the widest possible number of its many senses—it should be understood, in the knowledge order of the contemporary university, as the discipline of making sense of texts and thereby includes everything from paleography, codicology, and textual criticism to the history, interpretation and comparison of all textualized language—I use the term on the present occasion to refer more narrowly to textual criticism: recension, emendation and other practices of text editing. Philology in that sense has neither a corresponding term in Sanskrit (or any other South Asian language) nor, consequently, was it ever the object of a corresponding form of discourse.

The absence of a cognitive category and disciplinary form of textual criticism is especially curious, given the development of other subdisciplines of philology and the general scope of systematization in Sanskrit intellectual history. In many ways, Sanskrit is the most philologized language in human history. The philological habit as such is fostered above all by a language’s time–space distance, and Sanskrit—never a language of everyday life but instead (according to the dominant language ideology) the language of the gods—was maximally distant from the human world. Accordingly, grammar, phonology, metrics, lexicography and hermeneutics all attained astonishing refinement. We need think only of the rules devised by Pāṇini in the fifth or fourth century BCE that reduce to order the apparent chaos of phonological and morphological transformations, and contrast those with the quite random way phonology and morphology in Greek and Latin have been understood and taught for centuries. But philology in the sense of textual criticism not only arose late in the history of Sanskrit culture; it was never codified as a practice, notwithstanding the codification of all its subdisciplines. Indeed, in Sanskrit India, virtually every human practice that could be reduced to the descriptive–prescriptive rigors of śāstra, or work of systematic knowledge, was so reduced, but not philology.

It may not be surprising that people can have a conception of the parts of a thing without having a conception of the whole.³ It may also be true that the textual practices of philology too thoroughly pervaded the Indian thought world for it even to be identified. At all events, the fact remains that we are left to gather the principles of Sanskrit philology from the raw evidence of the texts themselves; that is, from the practices of philological commentators. While my principal concern here is to give some coherent sense of those practices and to try to draw from them a more general theory, I also want to think about their emergence and consolidation. The lack of systematicity about those practices makes the first task difficult, and the relative lateness of their emergence presents something of an historical conundrum.

2 The Genre of Philological Commentary

What is striking about Sanskrit philology is not that it was embedded exclusively in commentary but that philological commentary arose so late, relatively speaking, in the history of Sanskrit textuality. Commentary as such is far older, of course, virtually coeval with the primary texts. But commentary of the sort we find in the early period is decidedly not philological. Thus, we have commentary on śāstra, especially the principal knowledge forms (vidyāsthānas) such as grammar, logic and hermeneutics, from the beginning of the Common Era, but here commentary is actually the form that the substantive conceptual development of the knowledge system took: one contributed to the system by writing commentary on the base text (the sūtra) or on one of its primary exegeses (the Sanskrit terms vṛtti, bhāṣya, vārttika, ṭīkā and so on all refer to different species of commentary or subcommentary). We occasionally meet with grammatical or semantic exegesis, but rarely, if ever, are questions of recension and emendation raised (techniques like yogavibhāga, or the splitting of sūtras, in the early grammatical tradition, are more interpretative than text-critical manoeuvers). By contrast, commentary on those textual forms where philology is central or shown to be central—namely epic literature, court poetry and (in a more hermeneutical sense of philology) scripture—is largely a phenomenon of the early second millennium and thus is centuries later (if not millennia later, in the case of Vedic scripture) than the texts to which such commentary is directed.

It stands to reason that philology in traditional India was focused in the first instance on works of literature (only secondarily on scripture). The tenth-century literary critic Bhaṭṭa Nāyaka introduced a typology, widely accepted by later thinkers, that sought to distinguish among three text genres: one where wording is predominant, a second where meaning is, and a third where wording and meaning are equal.⁴ The first category comprises scripture, since the Veda’s efficacy lies in how something is said, that is, in the particular sequence of its phonemes.⁵ The second category is occupied by śāstra, or science broadly conceived, for here (at least for Indian thinkers, who, of course, knew nothing of what has come to be called the ‘linguistic turn’) what counts is what is said, not how it is said. The third category pertains to poetry, where wording and meaning share predominance. While this typology does nothing to explain why the philology of epic literature, court poetry and (to some extent) scripture should manifest a substantial presence first in the early centuries of the second millennium, it does help us understand why philology was directed toward expressive texts and why, therefore, the philology of scientific texts in traditional India should be as underdeveloped, relatively speaking, as it appears to me to be.

3 Scriptural Commentary, Buddhist and Other

Recent discussions of early Buddhist scriptural commentary and exegesis, which are a very early phenomenon, may be thought to be inconsistent with the picture I have just sketched of what constitutes the relevant pool of data for a history of Sanskrit philology.⁶ Explaining why they are in fact consistent will help clarify some of the points I have just tried to make.

Buddhist commentators were essentially exegetes, not philologists. Vasubandhu’s important Vyākhyāyukti (Arguments for Exegesis, ca. fifth century; extant only in Tibetan), which defends the Mahāyānasūtras (early centuries CE) against their older opponents, the Śrāvakas (Theravada, late centuries BCE), argues for the acceptance of these sūtras as buddhavacana, ‘word of the Buddha’, the key term for authenticity and authority. Vasubandhu has been taken to be making a general philological argument against the Śrāvakas, that even such texts as the Mahāparinirvāṇa Sūtra were recited differently among the different schools and even ‘authorized editions’ (those prepared by Mahākāśyapa at the first Buddhist council and other early disciples) were gradually affected by variations, signaled by divergent chapter divisions and the like and even different passages and chapters. And if ‘authorized editions’ had degenerated, it was unclear, argued Vasubhandu, what exactly constituted buddhavacana.⁷

This comment aside, however, Vasubandhu and subsequent Buddhist tradition did not consider buddhavacana to be an object of either general historical or particular philological knowledge, and did not assess it on the basis of the sūtras’ language or textual form, let alone by invoking linguistic or textual archaism or particularity. Their basis for authenticity is instead the doctrinal truth of the sūtras. The Buddha was fully understood to have been a historical being, yet from the beginning of the tradition, his teaching was not restricted to his native tongue; on the contrary, there was a scriptural obligation to transmit it in any given local language (although one would think that precisely this obligation, and the concomitant prohibition against using Sanskrit, would have made the philological argument, in some sense of ‘philology’, even more compelling). For this reason, if for no other, buddhavacana was identifiable only by the quality of the truth it enunciated. Śāntideva’s Bodhiycaryāvatāra (eighth century) with the (ca. ninth-century) commentary of Prajñākaramati implicitly rejects anything resembling a philological–historical method in its defense of Mahāyāna as buddhavacana.⁸ A Mahāyāna text adduced by Prajñākaramati (the Adhyāśayasaṃcodanasūtra, not extant) moves far beyond philology to the outer edge of the hermeneutical with four criteria ‘for recognizing any insight as being buddhavacana’:

It must be sensible and not nonsensical; it must be in accordance with reality and not at variance; it must remove the afflictions and not increase them; it must convey praise for the virtues of nirvana and not those of transmigration … To whomever insight arises, Maitreya, or ever will arise according to these four criteria … should be named a Buddha; he should be named a teacher and his teaching learned as dharma. Why? Because whatever is properly spoken, Maitreya, is all the Word of the Buddha. Any hateful person who maligns these insights, Maitreya, saying they were not spoken by the Buddha, or shows them disrespect, maligns all the insights spoken by the Buddha.⁹

In such circumstances, where the authenticity (and hence authority) of a text is a function of its (a priori) truth—and not its truth a function of its (demonstrated) authenticity—philology has no role to play in the establishment or purification of a canon. It is therefore unsurprising that no philological apparatus was produced by Indian Buddhists until relatively late (with respect to Sanskrit grammar, for example, the mid-fifth century Candravyākaraṇa; with respect to Pali grammar, the twelfth-century Saddanīti).

It might accordingly seem sensible to contrast the Buddhists’ ‘content model’ with the ‘linguistic model’ of the Vedas and to register, as a consequence, the importance of comprehension in the one case and accuracy in the other.¹⁰ But the distinction, surprisingly, carries no consequential differences for a history of philology as edition and emendation. While the peculiar character of Vedic language was acknowledged from the beginning of systematic philological reflection (being marked in Pāṇini’s grammar as chandas, ‘The Metrical’, i.e., the Veda, in contrast to bhāṣā, ‘the spoken’)¹¹ and was identified centuries later by Mīmāṃsā as key diagnostic of canonicity,¹² the unparalleled commitment—or ideology of commitment—to textual stability in the transmission of the Vedas completely excluded philological engagement except indeed at the level of content. Aside from the Vedic hermeneuts discussed below, none of the actual commentators on these texts (though they are vast and I cannot pretend to have examined them all) ever raises a question of textual variation.

The afterlife of these tendencies with regard to scripture across Hindu religious communities merits brief comment. Later apologists for the new scriptures (āgama) of communities devoted to the worship of Vishnu and Shiva that we begin to find from the early medieval period similarly held their texts to be in essential harmony with the Veda—indeed, in the eyes of some, as cognate with the Veda—and they were not in the least disturbed by, or even cognizant of, philological criteria. A good example of this indifference is offered by the defenders of the authority of the Pāñcarātra (Vaishnava) texts that began to be produced some time in the middle of the first millennium. While the tenth-century theologian Yāmunācārya argues that these texts were ‘composed by the Supreme Being himself’¹³ just like the Vedas (which he regards as created by God, unlike Mīmāṃsā, which holds them to be authorless), no attempt is made to justify this claim on the basis of philology. Notwithstanding what we might call a practical Vedicization in the creation of other Vaishnava texts of the period—the tenth-century Bhāgavatapurāṇa, for example, is remarkable for its archaizing language—Yāmunācārya offers no defense on the basis of language, nor accounts for the fact that while God created both the Vedas and the Pāñcarātra scriptures equally,¹⁴ their linguistic characters are so radically divergent. Instead, like the Buddhists, he is concerned only with Pāñcarātra truth, the harmony of that truth with that of the Vedas, the refutation of apparently contrary Vedic or quasi-Vedic texts that deny that truth, and the exclusion of other sectarian scriptures from the realm of Vedic truth.¹⁵ It is not until the seventeenth century that a kind of text-critical (if not language-focused) philology begins to challenge theology in the adjudication of scriptural claims, a point to which I return at the end of this essay.

Philology as recension and emendation in the world of Sanskrit, then, is essentially a set of practices found in commentaries on epic and court literature from the beginning of the second millennium, but for which we have no evidence of their ever having been systematized and theorized by the commentators themselves. It is on those practices that I shall concentrate in the remainder of this essay.

4 Sanskrit Textual Criticism

There exists no scholarly account of the origins of textual criticism in Sanskrit intellectual history; indeed, the very idea that textual criticism might have an origin seems to be unknown. If the contrast between philological and philosophical commentary is rarely drawn, at least we are finally beginning to get strong accounts of the long historical development of the latter^16; for the former, scholarship is still in its infancy. As already noted, philological commentary, both for epic and secular poetry, is an invention of the late first millennium CE (almost certainly in Kashmir), and in the following centuries, we can clearly observe an intensification of textual criticism.

Text-critical terms like ‘variant reading’ (pāṭha) and ‘interpolation’ (prakṣipta, later kṣepaka) that come into common use in the second millennium are hardly to be found before this period. One of the earliest discussions I have located is in neither epic nor court literature commentary but in Vedic hermeneutics. The seventh-century scholar Kumārila, when discussing a purity rule in the law books (smṛti), observes that a particular word in the rule ‘is not found to be used in the oldest pāṭha’ of the text in question,’ though no further detail is offered.¹⁷ It is only an eleventh-century commentator who raises questions more pertinent to our concern here: ‘It may be that someone interpolated the word and transmitted that pāṭha,’ he argues. ‘It certainly cannot be the pāṭha of the sages since it is not that of the bulk of learned scholars.’¹⁸ This commentator’s criteria for textual authenticity thus include, as we would expect, antiquity, but antiquity itself is determined by numbers and quality: the witnesses have to be credible and numerous.

What importantly complicates the question of textual variation in the case of the Veda is the concomitant one of orality. I intentionally left untranslated the word pāṭha in the above discussion, for whereas it later comes to be used to mean a ‘variant reading’ of a literate manuscript tradition, its sense for Vedic culture is different. There, literate procedures were decidedly denigrated for scriptural transmission; as Kumārila puts it, ‘Something ceases to be regarded as revealed moral law (dharma) if it derives from the Veda that has been learned in ways contrary to reason, as, for example, through a written text’. The way that is not contrary to reason is the precise ‘after-pronouncing [i.e., reproduction] of what has been pronounced by the mouth of the guru’.¹⁹ And hence pāṭh/pāṭha must connote ‘recite/recitation’ as well, whereby Kumārila’s usage in the passage cited earlier (pāṭhe cirantane) comes to mean ‘original recitation’. Nor is there any reason to suppose that variants cannot be interpolated—or that Indians in the classical period did not believe that variants could not be interpolated—into oral traditions: here, numbers would seem to count.²⁰

Such early rarities set aside, much of our best data regarding practices of recension, or the examination of manuscripts in order to determine the earliest available state of the text, come from commentators on the epics, who often—and for the most part confessedly—functioned as editors.²¹ (I should note at once that there exists no term for ‘editor’, ‘edition’, or the like in any Indian language, outside of Persianate culture, before the modern period.)²² Commentators on court poetry, such as the earliest among them from tenth-century Kashmir, including Vallabhadeva, seem often (though not always) to have been editors as well. While fully aware of variation in manuscript copies, such commentators nowhere describe collation of manuscripts, though something like that may be implied by the evidence, direct or indirect and usually vague, that they compared them.²³

Epic commentators, on the other hand, often provide clearer testimony on the matter, especially with respect to the culturally foundational Mahābhārata, a work that was repeatedly edited and somehow—by a process as yet entirely unclear to us—published in the period 1000–1700. Thus, Nīlakaṇṭha Caturdhara, a Maharashtrian Brahman who worked in north India in the last quarter of the seventeenth century, gathered ‘many manuscripts from different regions’ and ‘critically established the best readings’. He occasionally evaluates manuscripts according to age (‘old,’ ‘recent,’ ‘damaged’, ‘good’) and readings by their frequency (‘common’, ‘occasional’, ‘rare’); frequently discusses variants of individual readings or in the sequence of verses, and even once admits failure (‘Only Vyāsa himself [the putative author] knows the true reading here’), though it is only rarely he tells us where, let alone why, he edited the text the way he did.²⁴ Nīlakaṇṭha’s explicit acknowledgment of the transregional dissemination of manuscripts, and his tacit recognition that these are all versions of the same text and must be compared with each other to attain the singular textual truth (which indubitably exists for him), are important markers of a theory of textuality in general as well as of the understanding of this particular text’s mode of being. And far from being peculiar to Nīlakaṇṭha, these beliefs were shared by every editor who bothered to explain, or at least allude to, his editorial procedures.²⁵

It seems very likely that this kind of transregional consciousness in editing was a phenomenon of the early second millennium, following a long period of the regionalization of recensions—no doubt affected, to some degree, by the regionalization of writing systems that we first begin to notice with the rise of vernacular or regional literacy in the last quarter of the first millennium—that is observable across the history of Sanskrit literature. Indeed, epic commentary itself is a phenomenon of that period, finding its origins around the beginning of the second millennium—there is no evidence that the earliest (Devabodha) had a predecessor—and experiencing a dramatic upsurge after about the mid-thirteenth century. It is then that commentary on the Vālmiki Rāmāyaṇa, the second great Sanskrit epic, commences, with a south Indian commentator arguing for the need to establish the ‘correct reading (samyakpāṭha) corrupted by scribes unskilled in the various scripts,’ by ‘examining multiple manuscripts from multiple regions’.²⁶

Indian scholars were fully aware that the textual condition required clear principles that needed to be followed in text editing, but again, given the lack of programmatic statements, we can discover them only by sifting the texts of our commentators.²⁷ Vallabhadeva, to return to our tenth-century literary scholar, wrote basically word-for-word commentaries (pañc[j]ikās), which required him to address very closely the textual state of a work, and he left us several dozen text-critical discussions on the various manuscripts that he compared.²⁸ These show that his criteria were multifarious: readings (or passages) could be judged as grammatically or contextually ‘correct’/‘reasonable’/‘proper’/‘right’ or ‘more correct’/‘more reasonable’; ‘authoritative’, ‘false’, ‘mistaken’, ‘corrupt’, ‘unmetrical’ and ‘ancient’; ‘interpolated’, ‘in need of emendation’, or ‘obscene’ and last but not least, ‘lovely’, ‘beautiful’ and ‘more beautiful.’²⁹ At least once, he adduces paleographical or at least graphical criteria, as when he notes that a variant is a result of confusing two similar characters (lipisārūpyamohāt) and he rejects it on the grounds that it would seriously contradict the narrative.³⁰ Vallabhadeva often evaluates a reading on the familiar principle of difficulty and the antiquity such difficulty implies: ‘This must be the ancient reading precisely because it is unfamiliar’. He sometimes combines principles of antiquity and aestheticism when asserting that ‘the old reading in this verse is more beautiful’. But antiquity can be too ancient, as it were, if it produces a grammatical (or lexical or metrical or rhetorical) irregularity such as a Vedicism. Here and elsewhere, like other commentators, Vallabhadeva was on occasion prepared to suggest a revision in order to save his author from a supposed solecism, but he typically hesitates to actually alter the text and winds up transmitting the offending reading.³¹

In the matter of emendation—or, in this context, perhaps better ‘correction’, since the restoration of the original text was not invariably the goal—a tension manifests itself that will mark the whole long history of Sanskrit philology. On the one hand, as manuscripts show, some scribes and editors were highly attuned to text-critical problems and fully prepared to ‘improve’ the text, whether to remove a grammatical deviation or to correct a supposed aesthetic or logical fault.³² On the other hand, some scholars explicitly rejected doing so, such as Mallinātha, who took care to assure readers that he was not transmitting anything not found in his manuscripts.³³ And generally, it seems, editor-commentators did seek to establish as coherent and authoritative a text as they could on the basis of received manuscript tradition (āgata) rather than conjecture (kalpita) (though the later proverb, ‘We must explain that text as we find it’, is not attested before the eighteenth century).³⁴ Yet even Mallinātha sometimes adopted a conjecture that his predecessors had only suggested while they themselves preserved the received text.³⁵

The work of the literary theorist Mahima Bhaṭṭa (ca. 1000, Kashmir) is instructive here and for his general approach to textual criticism. In his treatise on poetics (and this is common in such works), the phrase ‘correct reading’ (yuktaḥ pāṭhaḥ) often connotes, not what a study of the manuscripts indicates to be correct, but what in the view of the critic the text should be if the passage is to avoid some putative fault and conform to aesthetic norms. Mahima Bhaṭṭa will often assert that a given verse of some great poet ‘should more properly read’ such and such.³⁶ He does, however, occasionally note that the normative reading is actually found in some manuscripts,³⁷ and at times he directly addresses an acknowledged text-critical problem. But if visions of normativity did not prompt Mahima Bhaṭṭa (or any other poetician I know of) to actually emend a text, this was not always the case, as we have just seen, with commentator-philologists. Here, for example, is Mahima Bhaṭṭa’s discussion of a line in Kālidāsa’s poem Meghadūta, which, while dealing with trivial syntactical problems, (about which Mahima was nonetheless much exercised), illuminates larger issues:

Commentators, too, with their false air of learning, are often found to bring shame not only on themselves but on celebrated poets as well by their erroneous comments. For example, in the Meghadūta line jātā manye śiśiramathitā padminīvānyarūpā [v. 80 = 2. 16], the commentators are completely ignorant of the reading that offers the nominative phrase in conjunction with the verb ‘to think’ (‘I think the girl must be …’) and, failing to understand the poet’s true intention,³⁸ they reject the profound beauty of his own thought—and this, too, despite the parallel to the nominative phrase in conjunction with the verb ‘to think’ that is found elsewhere in Kālidāsa.³⁹ They thereby turn the nominative ‘the girl …’ into the direct object, simply inventing the word vā in their confusion when it is actually the word iva transformed by vowel sandhi, and so proceed with their commentary. This adds no beauty whatever to the meaning. Nor is it possible to find this construction in any work of the great poet Kālidāsa, so as to convince us that he would use the word vā in the sense of iva—a veritable canker upon this poem, which is, in fact, a treasure-trove of aesthetic emotion.⁴⁰

Notice the three criteria of text-critical judgment Mahima Bhaṭṭa uses: grammaticality, usage and beauty. The syntactic construction that he believes to be original is perfectly in keeping with Sanskrit grammar (and the putatively false reading can be explained by a failure to understand the euphonic combination of the original); it is found elsewhere in the poet’s work and the false reading nowhere; while the latter is, for this critic, inherently ugly. What Mahima Bhaṭṭa does not do here, however, is ground his argument by reference to actual manuscripts—yet he is not simply inventing an opponent. The reading Mahima Bhaṭṭa criticizes is, in fact, that of Vallabhadeva, perhaps a generation or two his senior and, if the latter, a typically conservative editor, did not emend it himself, some other commentator may well have done so.⁴¹ In other instances, however, Mahima Bhaṭṭa seems rather to be inventing than discovering text-critical problems.⁴²

The commitment to fidelity toward the received text that we see here, and that was clearly widespread, is corroborated in the treatment of interpolation. As a rule, commentators continue to transmit materials they consider to be interpolated (as they continue to transmit clearly corrupt readings) and do not expunge them. Indeed, they and scribes in general sometimes went out of their way to ensure that material they knew to be interpolated was preserved in their transmission. The quest for what is thus the maximally inclusive edition, as is evidenced in manuscript culture, persisted into the early print era.⁴³

The question of interpolation seems to be one of the few where commentators on scientific treatises exercise text-critical judgment, for here it is precisely what is said rather than how it is said that can engage them philologically. One of the more telling cases is found in language analysis (vyākaraṇa). Beginning in the thirteenth century, with Haradatta Miśra in his commentary Padamañjarī (Bouquet of Words) on the Kāśikāvṛtti (Benares Gloss), a grammatical treatise of the early ninth century, we find a new attention to the integrity of the sūtra text of Pāṇini (fifth to fourth century BCE), the foundational work for the science. Haradatta often notes that the authors of the Kāśikāvṛtti inserted a given term into the sūtras, usually from another grammatical source, the later Vārttika (Exegesis) of Kātyāyana (third century BCE). Thus we frequently encounter statements such as ‘Since this is found in the Vārttika, we can conclude that it has been interpolated into the sūtra’; ‘this was interpolated into the sūtras by the author of the Kāśikāvṛtti’ and ‘present-day scholars have interpolated this into the sūtra’.⁴⁴ All the examples of this sort of editorial attention I have been able to locate date to the late medieval or early modern period.⁴⁵ The remarkably innovative seventeenth-century grammarian Bhaṭṭojī Dīkṣita offers a good number of text-critical remarks about the sūtra text of Pāṇini, Patañjali’s Mahābhāṣya (Great Commentary) on it and other works, as, for example, when he notes with respect to a given reading, that ‘according to older authorities it must have been interpolated by some scholar or other, since it is not available in the Mahābhāṣya and related texts. But it is, in fact, to be found in present-day manuscripts of both the Mahābhāṣya and the Vārttika.’⁴⁶ Such observations become commonplace by the eighteenth century, as in the work of the Thanjavur scholar Vāsudeva Dīkṣita.⁴⁷ It is in early modern grammatical literature, too, that scholars first begin to use the term sāṃpradāyika (saṃpradāya is attested in epic commentators in the sense of recension, here in the sense of a group of manuscripts related by script and region) in a text-critical context as the counterpositive to ‘interpolated’, and thus signifying ‘traditional’ or ‘original’.⁴⁸ The criterion of sāṃpradāyika is also found in the literary realm, but there it does not have the probative force it has in the domain of text transmission; on the contrary, it suggests (to borrow a phrase from the Islamic tradition of ijtihad), a closing of the doors of interpretation.⁴⁹

5 The Historical Moment

The data assembled above present two challenges to our understanding of Sanskrit philology: how to explain the moment of its historical emergence and how to understand the theory of textuality that underlies it. I deal with these in turn.

While Indian scholars clearly had some sense of textual variability prior to the early second millennium—they knew that readings could be corrupted, that interpolations could be inserted and that transmissions could be disrupted—it is only around 1000, rather dramatically and with increasing intensity thereafter, that text criticism became a self-conscious scholarly practice—something that cannot easily be explained away as a mere artifact of the survival of sources beginning in that period. This originary moment of text criticism may now be taken as a fact, but it is one we are far from being able to explain.

There was no sudden transformation in the material or intellectual or institutional context leading to the emergence of philological commentary; no new technical development or conceptual discovery or state-led innovation that lit the text-critical fuse. The introduction of paper early in the second millennium, for example, by no means displaced palm leaf and birch bark in book production. Thus its consequences in India were hardly comparable to the European and Islamicate experience, where, in providing a cheap alternative to parchment, paper vastly expanded communication practices.⁵⁰ The khagaz raj, or paper kingdom, of the Mughals marks a shift, but that is a seventeenth-century phenomenon. Nor can text criticism have been a response to some unprecedented estrangement from Sanskrit, for there had never occurred a moment of true rupture in Sanskrit cultural history. The arrival of new ruling groups from Central Asia beginning around 1000 certainly produced nothing of the sort, and in any case, the earliest stages of the philological revolution came several generations earlier.

The histories of other philological traditions suggests a range of causal factors. Philology developed in late imperial China—as ‘evidential research studies’ (kaozheng xue)—when scholars concluded that it was because of their failure to read the classics properly that the empire had been lost to barbarians (the Manchus) and they devised new philological principles to solve old problems of unintelligibility. Indians had no empire to lose except an imaginary one and if some felt even this was lost (to Mahmud of Ghazni in the eleventh century, the Delhi Sultanate in the fourteenth, or the Mughal Empire in the sixteenth), the event was rarely seen as a cataclysm and was never ascribed to some intellectual failure. The Arab scholars who invented the ‘art of philology’ (ṣināʿat al-adab) or ‘the literary arts’ (al-ʿulūm al-adabiyya) in early Abbasid lands were confronting the presence of an immensely important new text, the Qur’an, whose language they had, at all costs, to understand. New kinds of revelation did come to India in the course of the first millennium, whether Mahayana, Vaishnava, or Shaiva, but—as we have seen and, surprisingly, to my way of thinking—questions of their validity were typically not adjudicated on the basis of their language but rather on the basis of the doctrine: authenticity of a text was a function of its truth, and not its truth a function of its authenticity as proven by philology. As for the Greeks, they only seem to have stumbled upon ekdosis and diorthotes because they stumbled upon, or rather into, the Alexandrian library. The very fact that multiple written sources of a single work thereby became available made the need for those philological acts of edition and correction both obvious and necessary for the first time. In India, by contrast, where orality long persisted as a cultural value, the very idea of accumulating multiple copies of manuscripts of the same text would probably have seemed outlandish, and so too the idea of seeking out and explaining textual variation—until, for some reason, they no longer did. It is improbable that manuscript culture in India could have reached critical mass only at the beginning of the second millennium so as to provide the foundations for philological reflection, but it is then for the first time that Indian scholars begin to behave as if it did.

However, we explain this text-critical transformation, its effects were new and far-reaching. More than ever before, and in some ways never before, Sanskrit texts came to be mediated by a philological apparatus, one that with growing sophistication emphasized the dynamic changeability of transmission, the need for purification and the systematicity of reading, and whose growing density and wide circulation bespoke, or seemed to bespeak, new pedagogical needs and perhaps even correlated with new reading publics, of the sort we begin to perceive in other aspects of early modern culture (a good example is the growing popularity of introductory manuals, especially for grammar, hermeneutics and logic, the Sanskrit trivium). New institutional forms of pedagogy consolidating in the early centuries of the second millennium, whether in temples or Brahman settlements, and supported by wealthy courts, undoubtedly also had an important role to play.⁵¹ To the degree one is prepared to make something of it, the Indian date when philological commentary attains real cultural significance correlates with European and Chinese ‘early modernity’, if the story of that contested periodization is taken to start with the twelfth-century Renaissance and the Song (where the rapid expansion of print culture is especially pertinent) rather than with the date of 1500, which marks the beginning of global modernization, something quite different.⁵² That said, a later ‘early modernity’, beginning in the seventeenth century, seems also to be signaled by other new kinds of philological attention. The ordering and wording of foundational texts of the various knowledge systems (grammar, for example, or logic) come under systematic scrutiny for the first time in the seventeenth or eighteenth century, and the validity of sectarian scriptures—overturning more than a millennium of purely doctrinal adjudication—comes to be vigorously debated on the basis of recensions, quotations and other textual (if not linguistic) features.⁵³

6 The Theory of the Text

The theory of textuality that underlies Sanskrit philology can be described both negatively and positively, by its exclusions and its inclusions.

A wide range of questions were never addressed by the philologists of premodern India. Take the effects of a still-living oral culture on manuscript transmission, a phenomenon that crucially distinguishes the Sanskrit case from the Greek and Latin, and that the philologists of India did not and perhaps could not address precisely because they were inside that culture. First, because tradition was an oral-literate hybrid, textual transmission—and this pertains to scientific as well as literary texts—often shows the consequences of memorization and performance.⁵⁴ The manuscripts of the Śatakatrayam (The Three Hundreds), for example, of the seventh-century (?) poet Bhartṛhari present countless variants that are neither scribal mistakes nor learned corrections but oral variants in what by any standard was nonetheless fundamentally a literate culture. Second, because the tradition was active and not dormant, text-critical models that make sense for, say, Catullus or Propertius (where the descent of manuscript begins with unique copies of the ninth century) do not work for Bhartṛhari.⁵⁵ Manuscripts of his work were produced by the thousands well into the nineteenth century, and while, to some degree, these can be reduced to regional recensions (largely defined by script), they were always on the move and interacting with manuscripts from other regions. ‘Contamination’ in this world is therefore not the exception but often the normal state of affairs (and a phenomenon that clearly needs a new name).

Thus, the text as embodiment of an authorial intention—however much a value explicitly acknowledged by the participants in the literary culture—was constantly and in some cases irremediably destabilized by the messy business of bringing works to life in an oral world, whether in the classroom (where the set text was typically recited from memory) or in literary performance. There were some exceptions. The memory culture of the Vedic tradition ensured its invariant transmission, and this value seems to have been transferred to other quasi-sacred texts such as the sixteenth-century Rāmcaritmānas, a fully literate work transmitted with little variation despite, or perhaps precisely because, it was constantly presented in oral performance.

There was, however, a vast domain of questions the philologists of India could address, as we have seen, and if they never openly theorized their practices (a lack that can hardly be merely an artifact of the extant data), they provided materials enabling us to do so. Let me now lay out elements of such a theory of textuality.

Classical Indian philologists understood texts to be unitary creations embodying authorial intention, even for texts that we today consider paradigms of composite authorship, such as the epics and purāṇas (repositories of myth and legend). This intention, they felt, could be recovered by the judicious assessment of manuscript variants—one that was never, however, conceived of as genealogical—and they developed criteria of textual criticism in harmony with that fundamental principle. If variants could be adjudicated on the basis of antiquity, as we have seen, it was only because of the implicit conviction that the older the reading, the closer it brought us to that authorial original. When Indian philologists took cognizance of the problem of regional variation, as they did from an early period (the thirteenth-century Rāmāyaṇa commentator Uḍāli Varadarāja),⁵⁶ it was out of the same implicit conviction that a single text underlies variation, and variation therefore constitutes deviation—no ‘éloge de la variante’ here. It is only because texts were viewed as coherent wholes that the notion of interpolation could ever have developed into the widespread text-critical principle that it became. Not only were they taken as wholes (so that higher-order criteria such as non-contradiction across the narrative could be invoked), but so was an author’s complete oeuvre: thus the eleventh-century critic Mahima Bhaṭṭa could appeal to Kālidāsa’s works in their entirety to rule out a given usage.⁵⁷

Other principles of the philologists of India derived from other presuppositions, which sometimes worked in tension with those just discussed. Since any text in Sanskrit was a part of Sanskrit culture, it was expected to adhere to the rules that defined that culture. The conflict here between two values, authorial intentionality on the one hand and normativity (in grammar, metrics, rhetoric, and the rest) on the other, was rarely discussed as such, but the checkered history of emendation in service of those rules demonstrates its consequences. A similar problem, if in some ways more elusive because nowhere discussed in the tradition, concerns large-scale textual expansion, no doubt closely related to the scribal conservatism already noted. Editors clearly believed that fidelity was a virtue, omission a sin, and, in general, bigger texts—where bigger texts were available—were better than smaller texts. A tendency toward agglomeration can thus be found in almost every genre, epic (Mahābhārata), scientific treatise (Yogaśāstra of Patañjali), or court poem (Śiśupālavadha).⁵⁸ Other kinds of text-critical interventions, about which we know far less, are visible in, say, the re-edition of the Nāṭyaśāstra in medieval Kashmir, where the very structure of the work was altered to make way for new—and sometimes disruptive—material, such as the addition of a ninth aesthetic emotion (rasa) to the canonical eight.

Aside from such large axioms, and the tensions that could arise when these conflicted, we can identify a range of other more restricted principles at work in Sanskrit philology. Scribes then were as prone to make mistakes as we are now and ‘copyist’s error’ (lekhakapramāda) was a small principle that found wide application. Finer discriminations of such error (haplography, dittography, and the like) were not made, but script confusion—as when southerners misread ligatures in the Śāradā script of Kashmir—could be invoked to emend a passage. Aesthetic criteria for text-critical judgment are in evidence, too, for belles lettres, where ‘the more beautiful reading’ was determined less on the basis of subjective taste than by the invocation of principles from the neighboring science of poetics. It is here, at the intersection of text criticism with other forms of knowledge— evaluation of evidence, modes of reasoning, forms of proof—which both shaped and were shaped by philology, that a deeper, altogether unexplored, realm of intellectual history opens before us.

7 A Dead or a Living History?

The attempt to reconstruct a theory of philological practice in premodern India—indeed, the empirical investigation itself—raises an obvious but difficult question, with which I close: Are such investigations and reconstructions merely a chapter in intellectual history, or do they have continuing relevance to the practice of philology today? In other words, do traditional Indian notions of textuality have any proper role to play in the present-day study of Indian texts?

Both the practice of philology in the wide sense and the philological study of Indian texts in the narrow concern a search for truth. The truth of the text, however, is not singular and unique. This is very clear in the case of meaning, which can usefully be seen as triadic: a work’s meaning for its original audience; its meaning for the many subsequent generations of its traditional readers; its meaning for us today. These meanings, I have long argued, are all are equally true, however irreconcilable with each other they may be, since meaning will always be meaning for different readers and cannot be reduced to any one of them. We must strive, difficult though it will be, to hold these three truths in balance, for it is only in their combination that the notion of the ‘true meaning’ of the work makes any sense.⁵⁹ The question for us here is whether the same logic holds in the case of the text itself.

Just as we present-day scholars want to know, in a historicist sense, what a fifth-century author meant, so we want to know—and this is, of course, prior—what he wrote. This is the foundation of modern textual criticism. But just as we also want to know what tenth- or fifteenth-century readers thought the author meant, we also want to know what they thought the author wrote.⁶⁰ These two different text-critical goals, now typically associated with the names Lachmann and Bédier, need not be mutually exclusive, as they are always represented as being. For me, the point is not that we cannot know or should not bother trying to know the original text (or that in some cases there cannot have been an original), nor that all versions of a work can be ‘just as good’ in some aesthetic sense. It is rather that textual variation embodies variable human consciousness, and that we need to take all such variation seriously if we are to seriously understand the history of consciousness. Indian philologists themselves, at least, seemed to accommodate all such options, not incoherently but pluralistically, for example complementing their views of authorial intention with the desire for maximally complete texts. Such pluralism—again, to the degree we have the intelligence (and now the technology) to hold these competing claims in equipoise⁶¹—may be one important lesson that Indian philologists of the past can teach today’s philologists of India. And that therefore makes their theory of philological practice essential for our own.