John B. Buescher
Introduction 7
Ancient Indian Speculation on Language and Reality 11
The Beginnings of Grammar 11
Philosophical and Religious Issues Connected to Grammar 13
Early Buddhist Views on Language, Truth, and Interpretation 19
Denying the Preeminence of Any Particular Language 19
Searching for the Final Doctrine 21
Collecting and Standardizing the Teaching 23
The Growth of the Abhidharma 25
The Buddha's Word 31
The Truth behind the Multitude of Forms 31
The Buddha's Ultimate Word and Ultimate Truth 37
Definitive Sutras and Those Whose Meaning Must Be Drawn Out 45
The Quest for Interpretive Clarity 47
Two Truths 55
Statements or Objects? Existent or Nonexistent? 55
Two Truths and Four Truths 62
The Vaibhasika School and the Two Truths 66
A Gelukpa Presentation of the Two Truths in the `Srdvaka"Schools 85
Acknowledgments 133
Abbreviations 134
Bibliography 135
Notes 153
Index 173
UDDHISTS LONG AGO developed a doctrine of two truths-conventional truth and ultimate truth. They have understood this doctrine in two ways. The first-which appears to have been its original meaning-referred to the sorts of true statements that the Buddha made: his "worldly," sometimes ambiguous, conventional discourse, and his statements or discourse that plainly referred to the ultimate truth. The second way referred to the sorts of objects in the world.
The first way to understand this doctrine dealt with problems Buddhists encountered in their exegesis of the scriptures, particularly in abstracting the Buddha's highest teachings from the mass of his discourses. Each of the many early schools of Buddhism used the doctrine in internecine debates to distinguish what it saw as the Buddha's highest teachings from what it believed other schools had mistaken as ultimate truths, or had invented altogether. In this sense, the doctrine covered much of the same ground as another early distinction-between scriptures that were literal or definitive, and those that were figurative or required interpretation.
This early development of the doctrine of two truths is traced here briefly. Some is excavated from fragmentary evidence of the scriptures of the early schools, and most particularly in the scriptures and commentaries of the Pali canon, which preserve the traditions of the Theravada School, a close descendant of one of these early Indian schools, the Sthaviravada.
The second way that Buddhists have understood the doctrine of two truths was as categories of all the objects in the universe or as the modes of objective reality. Treating the doctrine of two truths this way, however, moved the Buddhists' internal debates about the status of scriptures and statements away from the doctrine of two truths (now seen as categories of objective phenomena), and framed them as debates about which scriptures or statements needed interpretation and which did not. This was how matters stood in the period after the earliest proliferation of schools, that is, the period of early classical Indian Buddhism, depicted in the Tibetan Buddhist histories as comprising four major schools-the two Hinayana schools of the Vaibhasika (or Sarvastivada) and the Sautrantika and the two Mahayana schools of the Cittamatra (or Yogacara) and the Madhyamika.
Who was most responsible for shifting the doctrine of two truths from a way to talk about scripture to a way to talk about the objective world is not clear, although it may have been the result of efforts to distill lists of the most important things discussed in the discourses (sutra) of the Buddha's dharma ("teachings"). These lists were the essentials of the teachings of the dharma and so were the abhidharma ("higher teachings"). The Hinayana Vaibhasika School developed what it regarded as comprehensive lists of ultimate "truths"-that is, objects that ultimately existed, as distinguished from those that existed merely conventionally. They practiced their analytical skills-otherwise applied to distinguishing reliable statements or scriptures from unreliable ones-to an exacting, detailed elaboration of what exists. Nevertheless, the schools of this classical Indian period of Buddhism regarded the doctrine of the two truths in this way, as referring to two classes of objects in the universe. Debates between the schools about the nature of reality were conducted in terms of distinguishing objects in the universe that existed conventionally from those that existed ultimately. Or, to put it in other words, which objects were in the class of "conventional truths" and which were "ultimate truths."
The Vaibhasikas' doctrine of two truths involved an elaborate atomic theory because the distinction they drew between the conventional and the ultimate was between compounds and simples. They argued that compounded objects were not ultimately real, even though they had a conventional reality. Their goal was to specify the objects that were elemental and so were ultimate "truths" in the sense that they were not susceptible to further analysis.
Their elaborate effort to specify these ultimate simples was important for the development of the discipline of philosophical analysis. But another effect they could not have anticipated or wished for was that their analysis of the objects of the world-their abhidharma-was the touchstone for the great protest that eventually identified itself as the Mahayana (the "great vehicle"), placing the Vaibhasikas with their analysis of the abhidharma in the Hinayana (the "little vehicle"). The Perfection of Wisdom literature, beginning about the first century c.E., mounted an assault on the detailed lists of the Vaibhasikas' class of ultimate truths. All of the truths that the Vaibhasika abhidharma regarded as ultimate-that is, as incapable of further division-were shown to be no such thing. They were, in reality, compounds, or were conditioned, or, we might say, defined not in themselves but only by their causes, conditions, or contexts. All the ultimate truths of the Hinayana were dissolved in the Perfection of Wisdom, and were therefore shown to be mere conventional truths. What then, according to the Mahayana analysis, was an ultimate truth? It was a conventional truth's emptiness of being what it was in and of itself-it was incapable of withstanding such analysis.
This marked the major divide within Buddhism, in its historical development and among the various schools. The doctrine of the two truthsconventional and ultimate-became, for the Mahayana, and especially the Madhyamika School within it, the essence of the Buddha's teaching, which revealed the nature of the world and all things in it. Understanding that doctrine constituted enlightenment, and the achievement of nirvana. The Madhyamikas viewed the entire body of Buddhist teachings through the spectacles of the doctrine of the two truths. The Tibetan scholastic schools inherited the Madhyamika viewpoint from their Indian Buddhist teachers and preserved the sources for studying it. They also developed their own commentaries on these sources.
The heart of this book is a translation of a portion of a Tibetan text on the subject of the two truths. The text is Ngawang Belden's (Ngag-dbang dPal-Idan, b. 1797) An Explanation of the Meaning of the Conventional and the Ultimate in the Four Tenet Systems (Grub lugs kyi kun rdzob dang don dam par don rnam par bshad pa legs bshad dpyid kyi deal dbyangs), which he wrote in 1835 in Urga, Mongolia. The first setion of his book is a compendium of views on the doctrine of the two truths in the first of the four "tenet systems" (as the Tibetans classify them), the Vaibhasika system. One might wonder how valuable it might be to study the "lowest" of the systems on the subject that is of most importance to the "highest." But the Mahayana's central doctrine of emptiness-expressed in the teaching of the two truths-precipitated, as it were, out of the storm clouds of disputes within the early Buddhist schools over the nature of truth and reality. That doctrine was sparked into a flame by the winds of controversy, whose currents can be seen in the material preserved in this text's description of the Vaibhasika views on the two truths.'
The doctrine of the two truths gave shape to questions throughout the entire range of scholastic Buddhist speculation on perception and knowledge. Questions about what ultimately existed, how it could be perceived, and how it could be expressed, and how the Buddha had done so in the scriptures, were all interconnected. The doctrine of two truths, therefore, was concerned with the interpretation of language, the fitting expression of the doctrine, and with the proper interpretation of the world itself-what exists, how it exists, and how it can be truly known.
Ngawang Belden's exposition of his material follows the Gelukpa tradition of the Gomang College of Drepung Monastery, and so follows the earlier teacher and writer Jamyang Shayba ('Jam-dbyangs-bzhad-pa Ngagdbang-brtson-`grus,1648-1721). Ngawang Belden's book conveniently collects together a wealth of material from the scriptures and commentaries. The publication of this particular section of his book is justified by the fact that it depicts the Buddhist debate out of which the doctrine of two truths, through the critique of the Perfection of Wisdom, emerged into the foreground of Buddhist philosophy. It therefore opens a window (although refracted through centuries of commentarial layers) onto the historical beginnings of this doctrine. This piece of Ngawang Belden (and, indirectly, Jamyang Shayba) also supplements the ongoing publication of other portions of their work.'
THE BEGINNINGS OF GRAMMAR
Y THE TIME of the Buddha (about 500 B.C.E.), the Indo-Aryan languages spoken on the Indian subcontinent had diverged so far from the earlier language of the Vedas that these sacred scriptures had become obscure, and interpretive commentaries had begun to appear. These commentaries were avowed attempts to clarify what had always been present in the Vedas, but had become hidden behind corrupt forms of speech, faulty oral transmission, and imperfect interpretation.
They were not meant to create something novel, but merely to preserve the original message of the texts and to insure that their transmission would continue without interference. What may seem now like original lines of thought in the commentaries were considered to have occurred solely through remembering (smrtr) what had once been revealed in full but had been forgotten until the commentaries had made them explicit. Although the commentaries were "remembered," the basic texts of the Vedas were "heard" (sruti), implying that they were transmitted and received intimately, from teacher's mouth to student's ear, and so without interruption or change from generation to generation.
Nevertheless, problems arose from the disparity between the outward words of the ancient texts and the meanings attributed to them. The problem involved in textual interpretation was conceived conservatively, as purifying the Vedic language and expunging encroachments on the texts. This meant standardizing correct forms and excising deviant ones, which included those of non-Vedic dialects. Changes and novel interpretations, as well as unusual vocabulary, grammatical structures, or pronunciations, had to be avoided, for it was impossible to improve what had originally been perfect. The concern to uncover and preserve the words of the texts for their ritual recitation stimulated the development of the science of grammar, culminating in the work of Panini in about the fifth century B.C.E.
The Buddhists' attitude toward Panini was ambivalent, as was their attitude toward Sanskrit. Although he was not a Buddhist, his work was so important that some later Buddhists attempted to include him and his work, somehow, in their fold. In the Larikdvatdra Sutra, dated by modern scholars, based on historical and linguistic evidence, to the thirdfourth centuries c.E., the Buddha is made to prophesy that, "the author of [the treatises on] grammar will be Panini," and the fifth century C.E. Manjusri Mulatantra claimed that Panini attained a moderate level of Buddhist enlightenment-that of the "hearer" (srdvaka) level.
The Tibetan Buddhist historian Bu-don (Bu-ston, 1290-1364) wrote that Panini propitiated the deity Mahadeva in order to receive divine teachings on grammar. Mahadeva then uttered the sounds a, i, and u, and "by this Panini came to apprehend the whole of the grammatical science."' Nevertheless, a version of the story in the Manjusri Mulatantra had explained that Panini had propitiated the Buddhists' own Avalokitesvara instead of Mahadeva.
Grammarians elaborated their science to defend the language of the Vedas from a loss of meaning, but they also understood it to be their duty to keep the mother language itself pure. The yogic philosopher and grammarian Patanjali (second century B.C.E.) stated that the preservation of the Vedas in their exact form was the most important reason for studying grammar. As an example, however, he said that the explicit statement of a correct form, like also has the effect of implicitly negating corrupt forms, like gavi, goni, and gota, which happened to be dialect-based variations of the word.' Patanjali here expressed the traditional idea that the various Indic dialects were degenerate forms of a single, primordial, pure Sanskrit. Modern scholars, however, believe that the Aryans brought more than one dialect of Indo-Aryan with them into India, and point out that the language of the Vedas is itself not uniform but variable.'
The efficacy of the ritual sacrifice depended on the priest's absolutely correct recitation of the Vedic formulas and invocations. Some linguistic historians derive the primitive impetus for the development of grammar from this necessity, which was effected by a pronunciation technique called pada-patha, whereby each word was repeated separately, but "to do this correctly ... involved the beginning of grammatical analysis, and, since it involved the resolution of sandhi, phonetic
Sandhi is the combination or transformation of sounds at the beginning and end of words when they are joined to other words, characteristic of Sanskrit and other Indo-Aryan languages that are akin to it. The transformations can sometimes obscure the words, as they would occur by themselves, causing some confusion about their meaning. An often told, apocryphal story-not originally Buddhist but taken up and repeated by Buddhist historians Taranatha (b. 1575) and Bu-dondescribed the problem of sandhi when they told the story of how the study of grammar was supposed to have begun. One day a king, playing with his wives in a pond, said to one of them, "Ma udakam dehi"'"Don't splash water on me." Being from a different region, however, she understood him to have said, "Modakam dehr"-"Bring me sweets." She left the pond and returned shortly with sweets for her husband, but he became angry at the misunderstanding, and decided he would apply himself to studying how words should combine properly.-
PHILOSOPHICAL AND RELIGIOUS ISSUES CONNECTED TO GRAMMAR
When students entered religious training as priests in the Vedic tradition, they began with the study of grammar, a practice that was continued in the Buddhist scholastic centers of ancient India." They learned that grammar proceeded by analysis, breaking apart the verbal roots and stems from the prefixes, suffixes, conjugational and declensional endings, and from the sandhi, separating them from the networks of case relationships that formed the contexts in which they occurred.
The grammarian Bhartrhari (fifth century c.E.) explained that grammarians "know the connection of words" and that interpretation was the "science of dividing words into elements and bringing out their exact meaning." He called this systematic grammatical analysis smrti (the "remembered"), as opposed to the fruti (the "heard"), by which he meant the source texts upon which the grammatical analysis was done. His very use of these terms indicated that he believed that the legitimate interpretation of, or commentary on, a text, should, strictly speaking, stray no farther from it than a grammatical analysis will allow.' This conservatism in textual interpretation is consonant with Patanjali's statement in his Mahabadsya that a commentary should not add anything that is not already in the text, a directive which, if followed literally, would invalidate any comment whatsoever on the text.'°
With such a rigid restriction on commenting on a text, interpreters resorted to uncovering the etymologies of words in the text. The etymology of a word pointed to its origin and therefore was believed to reveal its real meaning divested of secondary meanings and connotations accumulated through time and circumstance. The commentators believed that the etymology of a word was a sure guide to its real meaning, so the "discovery" of a surprising or "hidden" etymology for a term in a philosophical dispute decisively secured one's doctrinal position when it could be linked to the etymology. The pursuit of etymologies (nirukta) intensified to keep pace with the linguistic drift that had occurred between earlier and later forms. The temptation to uncover hidden but (in truth) mistaken etymologies for word forms simply based on tortured similarity of sounds must have been great. Modern linguists have pointed to numerous instances when the ancient commentators asserted that words had certain etymologies, based more on their imagination than on the actual derivations of the words." If the obscurity of the language of the received texts made the development of a science of etymology necessary, it also encouraged its excessive application.
Underlying the excessiveness was the notion that all words could be analyzed into "substantives" (dravya) or into "nouns" or "names" (ndman). These names were supposed to be the true names of the constituents of the world, and they were ultimately built from grammatical roots. By distinguishing these names, one could discern the true names of real things. One might even conjure them, for the pronunciation of these names was often said to have an immense ritual and magical power. Nevertheless, even in the earliest example of a sustained work on etymology, Yaska's Nirukta, which he wrote about the turn of the sixth century B.C.E., a debate on this subject is described between two grammarians, Sakatayana and Gargya. Sakatayana held that all nouns could be derived through etymological and grammatical analysis from verbal roots, and Gargya held that all of them could not. Later Sanskrit grammarians reflected Sakatayana's position, for the most part. It is also evident in the willingness of the Nyaya-Vaisesika philosophical school to posit real substances (dravya) corresponding to every linguistically meaningful entity, even those not ordinarily regarded as substantives, such as words like "and" and "or," which they included in a category of phenomena they called "relations."12
The Grammarians, considered as a coherent philosophical school, and the Mimamsa ("inquiry," "exegesis") School, whose aim was the conservative defense of the Vedas, held that a close, ineluctable connection existed between the precise form of a word and its meaning. The Mimamsakas held, as one of their main doctrines, that, "Sound is permanent." In this, they expressed their belief that the words of the Vedas were eternal and immortal and were not created by human agency. In Indian society, where writing was still a novelty, words were regarded as essentially sounds.
This "sound" (sabda) was the sound of the Vedas, permanent because the Vedas were eternally true. Patanjali invoked this idea at the beginning of his Mahdbadsya by declaring that "the relationship (sambandha) between a word (sabda) and its meaning (artha) is fixed (siddha)."" He then carefully glossed siddha ("fixed," "established") as synonymous with nitya ("permanent," "eternal"), and as the opposite of kdrya ("produced"). He used siddha in this case, he said, with the same meaning that it had in the expressions "the eternal sky" (siddhd dyauh), "the eternal earth" (siddhd prthivi), and "eternal space" (siddham dkasam), as opposed to the meaning it had in the expressions "the rice is ready" (odanah siddhah) and "the pea soup is done" (supah siddhd yavdgur iti), which referred to products (karya).
Patanjali and other later grammarians and the Mimamsakas, even when asserting that sounds or words were permanent, attempted to avoid claiming, as a result, that when people utter words, either the sound they make has no beginning or end, or that they do not actually produce or create it. Patanjali distinguished two meanings of sabda-the first was, for example, the impermanent vocal tone (dhvani) that is the sound people make when they speak, and the second was the eternal and intermittently manifesting word that is the conveyor (sphota) of meaning.14 The relationship between these two was, however, admittedly difficult to describe, and other schools rejected the distinction as incoherent. The Naiyayikas and the Buddhists, in particular, emphatically declared that sabda ("sound") is anitya ("impermanent"). For the Buddhists anyway, this was consistent with their effort to distance themselves from the authority of the Vedas."
The Mimamsa School embarked on the work of standardizing Sanskrit, just as the Grammarians did. The Mimamsakas, however, always gave precedence to Vedic Sanskrit and were wary of the precedence gradually achieved by the form of Sanskrit described by Panini, which we now refer to as "classical Sanskrit." To the Mimamsakas, Vedic was the originally pure language, in which the god Brahma spoke and through which he created the universe. Another mythological explanation identified the goddess Vac as the speaker. She was later identified with the goddess Sarasvati, the consort of Brahma. She emitted sounds, and in doing so created the realm of discourse. Moreover, her emission of sounds actually created the world. This led to the notion that the universe was essentially a network of syllables or letters, each of which was a seed (bija) with a specific power (sakti) with which it could interact with others.
By concentrating on this network and its structure, revealed through grammatical analysis, or even by reciting the alphabet (varnapatha) or another set of letters, one could mentally retrace the world's evolution back to its source, dissolving the elaborated complexes of speech into their primary units-that is, Brahma's speech.' This suggests why studying language and grammar was so important. Bhartrhari wrote that, "Vac reveals herself to those who analyze speech."'' Grammatical analysis was therefore an analysis of the structure of reality itself.
Bhartrhari developed a theory that the world was produced by sound, and evolved (vivarta) or was elaborated through a process that mirrored grammatical transformations.18 The single sound that Brahma (or Vac) was said to have spoken and which elaborated itself into the universe was Aum, the syllable uttered at the beginning and end of the recitation of the Vedas. Buddhists had already theorized a similar process consisting of the periodic evolution (vivrtta) and dissolution or contraction (samvrtta) of the universe. Nevertheless, they explicitly rejected Bhartrhari's idea insofar as the Mimamsakas and, later, the Vedantins (who called it sabdadvaita, "non-duality of sounds") formulated it into a theory of transformation that explained the material world. The Buddhists said that the theory assumed the existence of a permanent essence that passed through merely superficial changes, contradicting the Buddhist principle of selflessness (nairatmya), according to which no real essence could maintain itself from moment to moment, even though subject to superficial "transformations" (parinama).
The Buddhists opposed the Mimamsakas' notion of language, according to which words-that is, sounds-were essentially immutable, that they had a reality of their own apart from their use. In this respect, all Buddhists were nominalists. Not only did they assert that, "Sound is impermanent," but they argued against any inexorable or permanent connection between a particular word (or its form) and its meaning. The connection, they said, was merely conventional-established simply by agreement (tacit or not) among the word's users.
This eventually led the Buddhist philosopher Dharmakirti (seventh century c.E.) to argue against the Mimamsakas' belief that substantially existent universals imparted a measure of their substance to particulars, and led him to explain the phenomenon of linguistic reference in a different way. The meaning of a term does not come from some pre-existent positive correspondence between it and its referent, he said, but from a kind of exclusion principle in which a person learns first what the term excludes-what one might call its context-and then concludes what it refers to. The Buddhist philosopher Dignaga (fifth century c.E.) quoted Bhartrhari's Vakyapadiya in his Hetucakradamaru in order to point out that the grammarian himself had admitted that, "People talk sentences, not words, and `word' is an abstraction artificially analyzed out by grammarians."19
The Buddhists and the Hindu grammarians did not always advocate diametrically opposed positions. Some schools of Buddhism, for example, notably the Mahasamghika and the schools of the Mahayana, developed a theory about the nature of the Buddha's speech (Buddhavacana) which in some respects resembled the non-Buddhist theory about the permanent "word of Brahma." One difference between the two was that the non-Buddhists ultimately traced words and letters back to the eternal Brahma, while the Buddhist philosophers of the Mahayana traced them back to emptiness (sunyata).Z° On the other hand, the Hindu grammarians sometimes admitted a kind of authority to the conventional meanings of words. Helaraja (late tenth century c.E.), for example, wrote that, "as a system of study entirely devoted to words and their popular signification, grammar does not care so much for strict adherence to reality or agreement of thought with truth, but takes words and their meanings as they are popularly used."2' And, to complicate matters further, even philosophers who did not defend the Vedas and their language could use the eternalist view of language that the grammarians had originally expounded as a way to defend Sanskrit.
The Jains, for example, used a similar argument to demonstrate the preeminence of Ardhamagadhi, the language of their own scriptures, and the language in which their founder, Mahavira, delivered his teachings. They considered all other languages-human and non-human-to have been derived from it: "The Lord propagated the law in the Ardhamagadhi language; this Ardhamagadhi, which gives peace, happiness, and bliss, undergoes modifications when it is spoken by the Aryans, the nonAryans, the bipeds, the quadrupeds, the wild and the tamed animals, the birds, and the worms." And then, "We salute Vac who is fully Ardhamagadhi and who modifies herself into all the different languages, and who is perfect and omniscient."22 Ardhamagadhi is one of the many Indo-Aryan languages and dialects that are closely related to Sanskrit and that are together known by tradition-not Jain-as the Prakrits (prdkrta, "unrefined"), as opposed to Sanskrit (samskrta, "refined").
...
Sharmik