Philosophy of Language
The role of language in Indian philosophy is great, and indeed, much of this philosophy remains unintelligible without an awareness of the role that language plays in it. It can reasonably be maintained that an important part of Indian philosophy is philosophy of language, to be understood in the double sense of “philosophy inspired by language” and “reasoned inquiry into language.” An exclusive concentration on the latter of these two, however, would not do full justice to the role of language in Indian philosophy. What is more, such a presentation would risk leaving out essential elements and remaining, to at least some extent, unintelligible. This article will therefore deal with both these aspects.
The point of departure is, and has to be, that the main actors in Indian philosophy held a number of views about language, which they considered either self-evident or too deeply anchored in their worldviews to allow for critical discussion. These views were therefore not normally objects of, but rather starting points for, inquiry.
These main actors were for a long time primarily Brahmans and Buddhists. Brahmans and Buddhists did not share the same views about language, and indeed, their positions were often radically opposed to each other. In spite of these divergences, a remarkable convergence took place in the course of time, which led to a situation in which Brahmans and Buddhists agreed on a number of far-from-obvious points. In order to understand this convergence, it will be necessary to discuss the different starting points and early developments of Brahmanical and Buddhist thought.
The language of the Brahmanical mantras was Sanskrit. Modern linguists may be tempted to point out that the language of many of the vedic mantras is an older form or even a predecessor of classical Sanskrit, which they may give a different name – for example, “Vedic” or “Old Indo-Aryan”; for the Brahmanical tradition, the Vedic language and classical Sanskrit were one and the same language, even though the Sanskrit of the Veda admittedly often used words and grammatical forms that were not commonly used in classical Sanskrit.
This conviction that Vedic and classical Sanskrit were one and the same language was important for the Brahmans. It allowed them to think of this language as unchangeable and even as eternal and without beginning. Indeed, in postvedic times, the vedic corpus itself came to be looked upon as eternal and beginningless. This raised questions that will be discussed below.
At this point I will consider two consequences of the belief in the efficacy of mantras. This belief was particularly important for Brahmans, for their livelihood depended on it. Brahmans owed their special position in society primarily to their knowledge of the Veda: they could recite portions of it and use its mantras at appropriate occasions. But to be efficacious, these mantras had to be recited in a phonologically correct manner. One of the consequences of the belief in the efficacy of mantras was therefore that major attempts were made to describe the phonology of those mantras, and secondarily of Sanskrit in general. A rich literature on the phonology of Sanskrit arose, much of which has survived until today (Deshpande, 1997, 31ff.; Scharfe, 1977, chs. 5, 7).
There was a second consequence of the belief in the efficacy of mantras, less immediately visible in the form of a surviving literature, but equally influential. The efficacy of Sanskrit mantras showed, at least to the Brahmans, that Sanskrit, their eternal language, was very close to objective reality. Sanskrit mantras could have an effect on the objective world because a close link connected Sanskrit with that objective world. Sanskrit distinguished itself in this regard from the dialects spoken by less eminent people. These dialects from our point of view were languages in their own right, and for the Brahmans no more than corruptions of the only real language, Sanskrit. No such thing as historical linguistics could therefore exist from the Brahmanical point of view. The one real language, Sanskrit, did not change and had not changed from beginningless time. All other languages were not real languages but corruptions due to people too lazy or ignorant to learn the one real language.
The conviction that Sanskrit is close to reality is central to Brahmanical thought. It finds expression in various ways. Most notable among these is the belief that the form of a word tells us something about the object it designates. This belief is behind the numerous semantic etymologies found in vedic literature, primarily in the vedic texts called Brāhmaṇas These etymologies have nothing whatsoever to do with the history of words, as in the case of historical etymologies studied nowadays by historical linguists. Semantic etymologies are something different altogether. They start from the assumption that the similarity between words reveals a common feature shared by the things designated. Semantic etymologies assume, at least in vedic literature, the presence of hidden links between objects, links that can only be brought to light through the analysis of the words that designate them.
Interestingly, the belief in the significance of semantic etymologies gave rise to an early attempt at systematization known as nirukta, “etymology.” Its classical treatise, also called Nirukta, was composed by a certain Yāska, who must have lived after Pāṇini but before Patañjali (see below), and therefore in or around the 3rd century BCE. This discipline, considered a vedic auxiliary science (vedāṅga), attempted to formulate the rules that govern semantic etymologizing. One of its principal rules, unsurprisingly, is that in establishing links between different words, considerations of meaning have to play a central role.
It is against this same background that we have to understand the interest in Sanskrit morphology and linguistics in general that characterized the late vedic period. This interest culminated initially in the famous grammar of Pāṇini (4th cent. BCE), which itself became the starting point for an extensive grammatical literature that consisted partly in commentaries on Pāṇini’s text, the Aṣṭādhyāyī, and partly in independent treatises that were yet profoundly influenced by Pāṇini’s grammar (Bronkhorst, 1981)
We will see below that the beliefs about the Sanskrit language just described came to exert an influence on Brahmanical philosophy. These beliefs themselves did not yet amount to a philosophical position, but they shaped future thinking.
Some scholars have made a somewhat different claim. They have drawn attention to the central significance of language study in Indian culture, and on the sophisticated way in which the grammarian Pāṇini in particular succeeded in dealing with complex linguistic issues. They have concluded from this that Pāṇini’s grammar played a role in India comparable to Euclid’s geometry in Europe in becoming a methodical guide for more recent philosophers. In support of this claim, they can point out that the study of Pāṇini, or one of his successors, was fundamental in the classical formation of scholars, so much so that all of them were acquainted with the grammatical method developed in these works. However, it is very hard, perhaps impossible, to give more substance to this claim. No one has yet been able to show in what way philosophical thinking followed a pattern derived from Pāṇini’s grammar. It is, however, beyond doubt that grammatical rules and arguments are frequently cited in certain philosophical developments, some of which will be considered below. The influence of the brahmanical presupposition as to the close connection between language (i.e. Sanskrit) and reality will become clear as the discussion proceeds.
But whatever the language they used for religious purposes, the Buddhists were not, at least not initially, inclined to believe that any of these languages – or any other language for that matter – had a particularly close connection with reality. However, developments now to be sketched led to a situation where a number of Buddhists came to believe that there was a close connection between language and our experience of the world.
This connection with language was initially not obvious in the philosophical developments that the Buddhist school, called Sarvāstivāda, in northwestern India underwent during the final centuries preceding the Common Era. For reasons that may be connected with the Hellenistic surroundings in which they found themselves, these Buddhists made an attempt to systematize the teachings they had inherited, ultimately from the Buddha, they thought. These inherited teachings contained lists of items that had been extracted from the sermons of the Buddha. These items frequently referred to mental states, some others to physical elements. The name that came to be used for these items was dharma, and the Buddhists of northwestern India developed the idea that these dharmas are the ultimate constituent elements of human beings, and by extension of everything else as well. For various reasons, these Buddhists also concluded that the dharmas must be momentary. On top of this, they stated that composite objects have no separate existence. In other words, a person is nothing but a collection and sequence of a large number of small and momentary entities; beside these numerous entities, there is nothing that one might call a person. The same applies by extension to macrocosmic objects in the external world: for example, there is no such thing as a chariot; there is only this vast sum of momentary constituent elements (Bronkhorst, 2009, 55ff.).
This remarkable vision of the world has but little in common with the original teaching of the Buddha, and the Buddhists of northwestern India had some difficulty anchoring it in the inherited Buddha words. However, as a philosophy, this vision came to flourish and exert a determining influence on most of the subsequent philosophical developments in Indian Buddhism. The question is: What has this vision to do with the relation between language and reality?
This relation enters into the picture in a manner that can be usefully illustrated with the example of a chariot. We see and believe in the existence of a chariot, when in reality there is no such thing. How are we misled into entertaining such an incorrect idea? As a result of language. We believe there is a chariot, because there is the word “chariot.” The same applies to all the numerous other things we believe populate the world. In reality, they do no such thing, but we are tricked by the language we speak.
Two observations have to be made at this point. The relation postulated by these Buddhists between language and experience does not concern reality as it really is. In deepest reality there are only vast numbers of momentary dharmas, as we have seen. The experience, whose parts correspond to the words of language, is not a reliable representation of reality. In other words, from this Buddhist perspective, language is not closely related to reality. Quite on the contrary, language tricks us into believing in a world that is ultimately not real. The parts of language that do the tricking are its words. Chariots, persons, and so many other things are ultimately unreal, but are believed to be real because of the words that supposedly designate them.
The systematizations of the Sarvāstivāda school of Buddhism appear to be the first manifestations of systematic philosophy on the Indian subcontinent. They were not primarily concerned with language, yet language played an important role in them, as we have seen. The words of language are responsible for the fact that we believe, incorrectly, that we live in a world in which there are chariots, persons, and other such things. However, this vision of the Sarvāstivādins had to face an obvious difficulty. Words, like everything else, are themselves accumulations – most specifically, successions – of dharmas. As such, they have no existence of their own, like all other accumulations of dharmas. This is problematic in that our incorrect beliefs in objects, such as chariots and persons, depend on words. If there are no words, what then is responsible for our mistaken conceptions?
It seems that the early Sarvāstivādins experienced this as a problem. This we may conclude from the fact that they included among their lists of existing dharmas a number of items that correspond to words and sentences. In other words, they postulated, beside the succession of auditory and other elements that make up the word “chariot,” the existence of a momentary dharma that somehow represents the whole word “chariot.” This went of course against their general principle of not recognizing the existence of wholes, but strictly speaking they managed to avoid this inconsistency. The “wholes” postulated in the case of words (and sentences) were, strictly speaking, not wholes, but entities (dharmas) in their own right, which somehow accompanied the succession of auditory elements that gave expression to that meaning. In this complicated way, the Sarvāstivādins had again words, and they could again claim that our illusion concerning the world was due to these words. These linguistic dharmas of the Sarvāstivādins might be looked upon as an idiosyncrasy of the philosophy of these Buddhists without deeper significance. Such a conclusion would not be justified, if for no other reason than that more recent brahmanical thinkers happily drew inspiration from these Sarvāstivāda inventions in order to deal with problems they had to face (Bronkhorst, 1987, ch. 3/7).
In this situation Brahmanism developed two ontologies of its own. The one that is most interesting in the present context is called Vaiśeṣika. It is most interesting in that it takes over the Sarvāstivāda idea that the world of our experience has a close connection with the words of language, but adjusted to its own Brahmanical presuppositions. In Vaiśeṣika the world of our experience coincides with the real world, and the only language that is taken into consideration is Sanskrit. For Vaiśeṣika, then, language is not a source of confusion that makes us believe in the external reality of a world that does not in that form exist, as it is for Sarvāstivāda. Quite on the contrary, for Vaiśeṣika language (i.e. the Sanskrit language) is a source of information about the real world. If we wish to develop an ontological scheme, namely, a scheme of what there is, the Sanskrit language is the means par excellence to find it.
This fundamental conviction of the Vaiśeṣikas finds expression in their philosophy. Vaiśeṣika claims that all things (artha) in the world fall into three categories: substances, actions, and qualities. These three categories correspond to the three main types of words already distinguished by the grammarian Patañjali (2nd cent. BCE): nouns, verbs, and adjectives. In order to make this ontological scheme coherent, various Vaiśeṣikas added a number of further categories, arriving at a varying total of six, seven, ten, or even more. However, even in their own understanding of their system, these further categories were add-ons.
The three main categories had subdivisions. The fundamental rule followed to find those subdivisions was that they had to correspond to words. The list of substances illustrates this particularly well: it corresponds by and large to the nouns of the Sanskrit language. The list of qualities raises more difficult issues, for they are not all referred to by adjectives or other words; some of them, in particular, are designated by nouns. Questions about their existence are resolved by considering language use. The quality “number” must exist, because expressions such as “one pot,” “two pots,” and so on are common usage. The existence of the quality “dimension” (pramāṇa) is shown by the use of the word “measure” (māna). There are further examples of this kind in the classical surviving standard work of Vaiśeṣika, the Padārthadharmasaṃgraha of Praśasta (or Praśastapāda, c. 6th cent. CE). Vaiśeṣika is in this way not so much a philosophy of language, as it is a philosophy inspired by language, or rather: a philosophy inspired by certain beliefs about the Sanskrit language (Bronkhorst, 1992).
Vaiśeṣika ontology became preponderant in Brahmanism, especially in those schools of philosophy that developed ideas on language (Mīmāṃsā, Nyāya, Bhartṛhari). The other main early Brahmanical ontology, Sāṃkhya, was not inspired by conceptions about language in the way Vaiśeṣika was. This does not mean that it denied the close connection between language and reality. This will become clear in the discussion below of the “linguistic crisis” that hit all Indian schools of philosophy – Brahmanical, Buddhist, and even Jain – in the early centuries of the Common Era.
Most important among these assumptions was the claim that, like its language, the Veda itself, too, had no beginning: the Veda had always been there. This is of course only possible if the Veda had had no author. Being without author, the Veda is pure speech. This, from the Mīmāṃsā point of view, has important consequences. To understand these, consider first ordinary verbal communication between human beings. In ordinary verbal communication, there will be a speaker, a message expressed in language, and a listener. From the point of view of the listener, the received message will be reliable if two conditions are fulfilled: (1) the speaker is reliable, and (2) the message is correctly interpreted by the listener. In the case of the Veda, there is no speaker who formulated the message initially. The chance that the Veda is used to mislead us is therefore excluded, and we must conclude that the message of the Veda is reliable, if only we know how to interpret the text correctly. This is the self-assigned task of Mīmāṃsā.
How does one interpret a text correctly? Once again, it is a matter of eliminating sources of error. The aim is to get as close as possible to the text, letting it speak for itself. Mīmāṃsā developed a whole list of principles of interpretation, which are in the end nothing but concrete manifestations of this general aim and are justified in this manner.
The belief that the Veda is without beginning has some consequences that have to be taken into consideration. Being without beginning, it is not posterior to any event that has ever taken place in the history of the world. Yet many vedic passages give the impression of relating such events, usually in the form of what we might call myths. None of these stories can be taken literally, because the Veda had already existed for an eternity at the time when these events supposedly took place. Stories and other references to historical events must therefore be interpreted differently. They must be read in context and have their place in the Veda in order to illustrate or lend support to vedic contents that can be taken literally.
What are the vedic contents that should be taken literally? The process of elimination shows what they are. All statements of fact in the Veda must be interpreted metaphorically. This is partly due to the fact that the Veda cannot refer to historical events, for the reason indicated above. However, other, nonhistorical, statements of fact contained in the Veda have another disadvantage. They might turn out to be in contradiction with our sense experience. This would be serious, for it would imply competition between two different means of knowledge: the vedic word, and perception. This difficulty does not exist in the case of injunctions. Injunctions tell us what we must do. No sense experience can ever be in conflict with an injunction, because perception informs us about states of affairs, not about obligations. With regard to obligations, we have only one ultimate source of information, namely, the Veda.
Mīmāṃsā arrives in this way at an interpretation of the Veda in which injunctions are central. They are what the Veda is all about. Everything else is to be interpreted in connection with those injunctions. In practice, the task is complex, but the theoretical basis is relatively simple and straightforward. In the end it amounts to this – that the word par excellence, the Veda itself, tells us what to do (Bronkhorst, 1997).
The second half of the 1st millennium CE saw the rise into prominence of a school that claimed to be no more than an improved version of classical Mīmāṃsā, but which opened up a path for completely new developments. This school – which referred to itself by various names, among them Śārīrakamīmāṃsā and Brahmamīmāṃsā, later also Uttaramīmāṃsā, and one of whose most important early authors was Śaṅkara (c. 700 CE) – came to be known by the name Vedānta, or Vedāntism. It owes this name to the fact that the parts of the Veda known as vedānta (end of the Veda), better known as Upaniṣads, played an important role in its reflections. The importance of these portions of the Veda is, however, in a certain sense due to coincidence. The Brahmamīmāṃsā applied the same principles as classical Mīmāṃsā, but slightly improved them with the result that certain upanishadic statements gained center stage. Recall that classical Mīmāṃsā had maintained that only in the case of injunctions can we be sure that there will be no conflict with other means of knowledge. The Vedāntists added vedic statements that teach knowledge about brahman ; these vedic statements happen to occur in the Upaniṣads. They argued that knowledge about brahman can only be obtained from these vedic statements, not by any other means of knowledge. Since no conflict between different means of knowledge is therefore possible, these statements have to be taken literally (Bronkhorst, 2007).
The Nyāya school of thought was willing to agree with Mīmāṃsā in its claim that the word – and especially the linguistic expressions that are found in the Veda – is a reliable source of knowledge. Unfortunately, and for reasons that will be discussed below, it adopted the position that the relation between words and things was conventional. For them, therefore, the first of the two conditions specified above (a received message is reliable if the speaker is reliable and the message is correctly interpreted by the listener) retains its pertinence even in the case of the Veda. Unlike the Mīmāṃsakas, the Naiyāyikas had to argue that the Veda had a trustworthy author, and indeed that the reliability of the Veda depended on the trustworthiness of that author. In an important sense, they had to admit that the reliability of the Veda depends on the same factors that make an ordinary statement reliable, that is, the trustworthiness of the speaker. They solved this problem by attributing the Veda to the most reliable person there is: God himself. This did not change the fact that their position – to the extent that the word (namely, linguistic utterances) is a reliable source of knowledge – was bound hands and feet by the supplementary requirement that the speaker, who uttered these words, is himself reliable (NyāS. and NyāBh. 2.1.50–67; trans. Jha, 1939).
We saw in the preceding section that the Vaiśeṣika ontology, too, is ultimately based on the word as source of knowledge. The situation here is nonetheless different from the one described in connection with Mīmāṃsā and Nyāya. There (in Mīmāṃsā and Nyāya) I talked about the reliability of certain linguistic utterances: primarily those contained in the Veda, further those pronounced by reliable speakers. In the case of Vaiśeṣika, however, I considered the relationship between the structure of reality and the structure of the Sanskrit language; the Veda or other reliable statements did not play a role here. Interestingly, some Buddhist thinkers, most notably Dignāga (6th cent. CE), came to adopt an ontological position similar to that of Vaiśeṣika, ultimately for almost the same reason. Dignāga could not, of course, claim that language would tell us something about the world as it really is, for he, like most other Buddhists, thought that language is rather the source of our mistaken ideas about reality. However, language, being the reason why we experience the world the way we do, is capable of providing us with information about the world of our experience. In order to find out more about this, Dignāga undertook an analysis of the relationship between words and things, which I will discuss below. Here it must suffice to observe that he arrived at an ontological scheme similar to that of Vaiśeṣika, be it that in his case this ontological scheme was limited to the aspect of reality (ultimately unreality) ruled by language (Katsura, 1979).
This development is likely to have begun within Buddhism, and it is possible, though not certain, that Nāgārjuna (2nd cent. CE) was responsible for it. Buddhist philosophers maintained that the world of our experience is not real. They had made this the basis of their philosophies, but they had never been able to prove it in an objectively convincing manner. This changed when people started paying attention to the relationship, not just between words and things, but also between statements and what they signify. It was believed that statements describe situations that consist of the things denoted by the constituent words of the statement. Most ordinary statements can be thought to confirm this belief. The statement “the book lies on the table” consists of three words in Sanskrit and describes a situation constituted of three “things”: the book, the table, and the activity of lying. Most Indian thinkers looked upon statement in this manner, thus accepting, though often only implicitly, the “correspondence principle”: the words of a statement refer to the “things” that constitute the situation described by that statement.
A problem arises in the case of declarative statements that describe a situation in which something is produced or comes into being. The situation described by “the potter makes a pot” contains a potter and the act of making, but no pot. Most Indian philosophers considered this problematic. What is more, many were willing to concede that this problem tells us something important about the nature of the world.
The Buddhist Nāgārjuna used this problem in a variety of ways to show that something is fundamentally amiss. Many of his arguments are variants of the following question: If there is something that is produced, what is the need of producing it? If there is not, how can it be produced?
Identifying the problem is not the same as offering a solution to it. We will see that different thinkers came up with different solutions. Nāgārjuna himself concluded that difficulties of this kind show that nothing really exists: everything is empty (śūnya). His philosophy is for this reason frequently referred to as śūnyavāda (doctrine according to which the world is empty of essence), beside its other name, Madhyamaka.
As stated above, there is no compelling reason to agree with Nāgārjuna’s conclusion, even if one accepts that statements like “the potter makes a pot” represent a major difficulty. Brahmanical thinkers in particular would hesitate to admit that nothing really exists. This position was close to the Sarvāstivāda Buddhist belief according to which the world of our experience has no more than conventional reality. Brahmanism was not ready to follow Buddhist thinkers in this regard. A relatively straightforward way out was to claim that the pot in “the potter makes a pot” is somehow present in the situation described. Since the potter who makes a pot is working with clay, one only had to assume that the pot was in some form or other already present in the clay. This position became known as the satkāryavāda (doctrine according to which the effect [kārya, i.e. the pot, in our example] is already present [sat] in its cause [the clay, in our example]). The Sāṃkhya philosophy adopted this solution. This, however, forced it to introduce some important changes into its system. Indeed, what does it mean that the pot is already present in the clay? This claim only makes sense if clay is considered to be the essential feature of a pot. Everything else – its shape, its color, and other characteristics – has to be looked upon as secondary. Only thus can one maintain that what is truly essential to the pot, namely, its substance, is already there before the potter has finished his job. However, Sāṃkhya had thus far believed that substances are no more than accumulations of qualities. This early position now had to be abandoned. This was done very thoroughly, so much so that one finds no trace of it in surviving Sāṃkhya texts. The texts by critics of the Sāṃkhya system, however, could not be so easily suppressed or “forgotten.” These texts belonging to other philosophical schools constitute, therefore, our most important source of information about the early history of Sāṃkhya, before the linguistic crisis obliged it to accept the satkāryavāda and introduce the modifications required by this new doctrine.
Jain philosophers followed a similar yet different way to solve the problem, adopting the so-called anekāntavāda (doctrine according to which it is one way in one respect, different in other respects). With regard to the pot, this means that, from the perspective of substance, clay and pot are the same; from the perspective of shape, they are different. This was considered sufficient to account for sentences like “the potter makes a pot,” for from the perspective of its substance, the pot is already there while the potter does his job.
The one school of philosophy that was not deeply affected by the linguistic crisis was Sarvāstivāda Buddhism. “Sarvāstivāda” means “doctrine according to which everything, namely past and future objects, exist.” The school had adopted this doctrine for reasons that have nothing to do with language, but once adopted, this doctrine came in quite handy in the new situation. These Buddhists believed that future pots exist as much as present and indeed past pots. In the statement “the potter makes a pot,” there is therefore a pot (a future pot) in the situation described. Unlike most other philosophers of their day, the Sarvāstivādins thus survived the linguistic crisis undamaged.
Nāgārjuna had drawn the radical conclusion that nothing exists. Others, most notably the mysterious author Gauḍapāda, drew the equally daring conclusion that nothing can come into being. His was the ajātivāda (the doctrine according to which nothing is produced). Gauḍapāda came to be claimed by Advaita Vedānta (Gauḍapāda came to be looked upon as the teacher of the teacher of the famous Śaṅkara), but parts of the text attributed to him (the Āgamaśāstra or Gauḍapādakārikā) are recognizably Buddhist.
It will be clear from what precedes that the linguistic crisis was responsible for many of the vādas (doctrines) that are characteristic of various Indian philosophies: satkāryavāda, śūnyavāda, anekāntavāda, ajātivāda, and to some extent even sarvāstivāda. Its impact, however, went further. Some schools refused to adopt any of the options so far considered and decided to draw conclusions that were not of an ontological nature, but were rather based on a reflection of the way words denote their objects. The problem so far considered depends vitally on the assumption that the word “pot” in “the potter makes a pot” refers to an individual object in the situation described. The Nyāya-Vaiśeṣika school of thought, having explored other solutions, ended up pointing out that the word “pot” in that sentence does not have to refer to an individual object, but rather (more correctly, also) to the universal that inheres in all pots. Indeed, all pots share that name because they have something in common, namely, a shared universal. Since the pot universal inheres in all pots – past, present, and future – it cannot but be eternal. Being eternal, it is always there, also in the situation described by “the potter makes a pot.” The new reflection about the way words denote things solved the problem (there was now something corresponding to “pot” in “the potter makes a pot”), and no further ontological conclusions needed to be drawn. In particular, it was not necessary to claim that the effect (the pot) exists already in its cause (the clay) before it is produced. In Nyāya-Vaiśeṣika, therefore, there was no place for satkāryavāda, and their solution came to be known by the name asatkāryavāda (the doctrine according to which the effect is not already present in its cause).
It is noteworthy that the two principal early Brahmanical positions, satkāryavāda and asatkāryavāda, can be read as adaptations of two positions proposed in Patañjali’s Mahābhāṣya, a major grammatical work in the Pāṇini’s tradition, dating from the middle of the 2nd century BCE, long before the linguistic crisis made itself felt. Patañjali discussed the question of the object denoted by a word. According to Patañjali, this object must be eternal, and he proposed two conceivable candidates: (1) words refer to forms, and (2) words refer to substances. In the first case, forms must be thought of as eternal, contrary to individuals, which are not. In the second case, substances are eternal. (The discussion becomes somewhat complicated on account of the fact that both “individual” and “substance” translate the same Sanskrit word dravya.) Different authorities held different positions in this matter; Patañjali himself thought that words signify both form and dravya. It is easy to see that the position according to which words refer to eternal substances is close to the satkāryavāda of the Sāṃkhyas. It is equally obvious that the alternative position, according to which words refer to eternal forms, is close to the one in which words refer to eternal universals, that is, the asatkāryavāda. These positions drew, therefore, inspiration from an old and venerable grammatical treatise, so as to arrive at a solution to a problem that had not yet been known to Patañjali.
The Nyāya school was not the only one to propose a semantic solution to the problem of the production of things. The Buddhist thinker Dignāga, mentioned earlier, came up with a solution of his own, which, however, shared with Nyāya the conviction that words do not refer to individuals. However, Dignāga had no place for universals in his ontology, so he could not simply copy the Naiyāyikas’ solution. The solution he proposed was inspired by his work in logical deduction.
Note, to begin with, that Dignāga sharply distinguishes between the world ruled by language, and the deeper reality that hides behind it. In his opinion, words do not apply to this deeper reality, which is the object of perception alone. Perception bears on the particular (svalakṣaṇa), language on the general (sāmānyalakṣaṇa). Ultimately, the objects of our perception are real, while the realm covered by language is not.
In the realm of language, words refer to their objects by way of exclusion (apoha): a word, say, “pot,” excludes everything that is not a pot. The word “pot” applies in this way to the pot, but does not denote it.
Dignāga assimilated this process of referring to the process of inference. He was among the first to have realized that an inference of the type “if there is smoke on the mountain, then there is fire” is logically equivalent to “if there is no fire on the mountain, then there is no smoke”; if B follows from A, not-A follows from not-B, and vice versa. Analogous applies in the case of referring: saying that all pots, and only those are referred to by the word “pot,” is equivalent to saying that all things that are not pots are not referred to by this word. Dignāga arrives in this way at an understanding of referring that does not postulate the existence of universals, or of a special link between words and individuals. The statement “the potter makes a pot” is no longer problematic for him, for there is no longer supposed to be a specific link between the word “pot” in this statement and the individual pot the potter is making (Bronkhorst, 1999).
Consider his ontological position in broadest outline. For Bhartṛhari, the totality of all that exists is the one encompassing reality, which he sometimes refers to as brahman. Though one and indivisible, this totality is yet divided in accordance with the words of language. These divisions are, however, less real than the encompassing whole from which they originate.
So far, Bhartṛhari’s philosophy is Buddhist philosophy turned on its head. In the Buddhist ontological scheme that we have come to associate with the Sarvāstivādins and others, composite entities were less real than their ultimate components; these composite entities owe their (relative) existence to the words of language. In Bhartṛhari’s philosophy, it is the other way round: composite entities are more real than their components, and these components owe their (relative) existence to the words of language.
Bhartṛhari illustrates his vision with a discussion of linguistic entities. Pāṇini’s grammar divides words into smaller entities, such as stems and suffixes. These smaller entities are the result of analysis and are artificial, in the sense that they are less real than the words from which they are extracted. A grammarian different from Pāṇini might identify different stems and different suffixes. Here, then, we find that the composite whole, the word, is more real than the elements into which it has been analyzed. The same reasoning applies to sentences and words. Words are extracted from sentences, but sentences are more real than the words into which they are analyzed. One can continue this reasoning and discover that the only really existent linguistic entity is the Veda, this huge totality of sentences. All other linguistic units are less real.
The total, undivided Veda corresponds, in Bhartṛhari’s view, to the totality of all there is, which he sometimes calls brahman. The world of our experience corresponds to smaller linguistic units, and like those smaller linguistic units, it is less real (Bronkhorst, 1992).
This ontological scheme had for Bhartṛhari more than just philosophical significance. He believed that its knowledge was vital for attaining the highest religious goal, liberation. Something similar had already been proposed by thinkers belonging to the Madhyamaka tradition, but Bhartṛhari gave it a completely Brahmanical and, what is more, grammatical twist. It is for this reason that Bhartṛhari could seriously maintain that grammar was the door to liberation. This claim has puzzled modern interpreters, who have tended to understand it as metaphorical exaggeration. It is, however, possible to take Bhartṛhari seriously in this regard. Bhartṛhari literally believed that the study of grammar was the door to liberation, because grammar was the means par excellence to understand the fundamental scheme of the universe, in which composite entities are more real than their constituent parts. Grammar shows this in the realm of words (with their constituent stems and suffixes) and sentences (with their constituent words). Once properly extended, this scheme covers the whole universe with all it contains; knowing this was for Bhartṛhari an essential step toward the highest goal: the door to liberation (Bronkhorst, 1995).
On a more day-to-day and linguistic level, Bhartṛhari owed his readers an explanation for his claim that words are more real than, and therefore different from, their constituents, and that sentences are different from their constituent words. Are words, and sentences, not simple sequences of sounds? Bhartṛhari thought they were not. According to him, a word is a different entity from its constituent elements. That is to say, beside the sequence of sounds used to communicate, say, a word, there is an altogether different “thing” that is the word itself. Bhartṛhari sometimes uses the word sphoṭa for this entity. He was not altogether original in this regard, for we have seen that the Sarvāstivādins had introduced a small number of linguistic dharmas for the very same reason. Beside similarities, there is, however, a major difference: the linguistic dharmas of the Sarvāstivādins were momentary (like practically all other dharmas), while Bhartṛhari’s sphoṭa is eternal. This may look like an important difference, but the two share an important feature: neither the linguistic dharmas of the Sarvāstivādins nor Bhartṛhari’s sphoṭa correspond to the common-sense notion of word or sentence. Common-sense words and sentences have a finite duration in time, corresponding to the time it takes to pronounce them. Bhartṛhari’s sphoṭa, like the linguistic dharmas of the Sarvāstivādins, do not exist for that length of time: they exist either for much longer (Bhartṛhari’s sphoṭa is eternal and has therefore no beginning or end in time), or for much shorter (the linguistic dharmas of the Sarvāstivādins are momentary).
It must here further be noted that, contrary to a widespread misunderstanding among modern scholars, Bhartṛhari’s sphoṭa is not, or not primarily, a semantic entity, a meaning bearer. This explains that Bhartṛhari can present a completely analogous reasoning to justify the existence of sphoṭas corresponding to individual sounds, which have no meaning. The ontological problem is the same for sounds as for words: they are sequences of vibrations and would have no separate existence but for the postulation that they exist as sphoṭas. The confusion is due to the fact that more recent authors in the grammatical tradition, who were much more interested in semantics than in ontology, came to speak of the sphoṭa as a meaning bearer (Bronkhorst, 2005).
What seemed at first sight to be the simplest solution was adopted by the school of Mīmāṃsā, associated with the name of Prabhākara (c. 700 CE). This school maintained that a sentence expresses no more and no less than the accumulated meanings of its words. The objection that the words in a sentence are functionally related to one another, so that the meaning of the sentence goes beyond the mere meanings of its words, is answered by the claim that individual words also express their relation to other words in the sentence (anvitābhidhāna); as a result the accumulated meanings of the words amount to the meaning of the sentence.
The Prābhākara school was alone in maintaining this position. Prabhākara’s senior contemporary, Kumārila Bhaṭṭa, for example, observed that the primary meanings of words when joined up could not account for the sentence meaning; he and his followers, the so-called Bhāṭṭa school of Mīmāṃsā, as well as other thinkers, rather held that secondary meanings for words in a sentence had to be postulated so as to account for their mutual relation in a sentence (abhidhānānvaya).
A third position preferred to look upon the sentence as a separate entity that expressed its meaning independently of the meanings of the words that seemed to constitute it. We have seen that Bhartṛhari in particular advocated the separate existence of the sentence. It is therefore not surprising that those who followed him, primarily grammarians in the Pāṇinian tradition, also accepted the separate semantic role of the sentence.
All except the Prābhākaras were confronted with the need to understand how exactly the meanings of words and sentence are related to each other. This need was particularly acute in the reflections of the Mīmāṃsakas, the Vedic hermeneuts who had come to the conclusion that injunctions are the most essential part of the Veda, around which other vedic statements are to be interpreted. It is in this school that elaborate analyses of the sentence meaning, in terms of its constituent elements, began. Other thinkers, most notably those belonging to the Nyāya school (which had meanwhile absorbed the Vaiśeṣika school), followed later. Later again the school of grammar that derived its inspiration from Pāṇini and Patañjali joined the debate.
Concentrating on the vedic injunctions, the Mīmāṃsakas had to determine which part of any particular injunction expresses the injunction itself and arrange all other grammatical units around this central element as qualifying it. They found that, in a statement like svargakāmo yajeta (he who desires heaven should sacrifice), the bare injunction (“should”) is expressed by the verbal ending (-ta). Specifications as to the details of the injunction have to be derived from the other parts of the sentence. The Mīmāṃsakas introduced in this manner a hierarchy of meanings, centered around the semantic element that is qualified by all the others. This central semantic element they called bhāvanā (bringing into being). About this “bringing into being” we need to know its purpose, the means by which the purpose is brought about, and how it is brought about. The simple injunction svargakāmo yajeta (he who desires heaven should sacrifice) thus gives expression to a bringing into being by means of a sacrifice, with heaven as its purpose. (The “how” is not further specified in this example; most vedic injunctions provide information about the kind of sacrifice to be performed.)
The Naiyāyikas felt that they could not agree with the Mīmāṃsā analysis of the sentence. Not only did they have a different view as to the independence of vedic statements (unlike the Mīmāṃsakas, they did not look upon these statements as fully independent, but rather as utterances of God), but they were also committed to an ontology in which substances play a central role. Indeed, of the three categories discussed earlier (substances, actions, and qualities), substances are central, for actions and qualities can only exist if they reside in substances: substances are qualified by actions and qualities. A typical statement (such as “the blue bird flies”) indicates how a particular substance (the bird) is qualified by an action (it flies) and a quality (it is blue). The Naiyāyikas therefore proposed a hierarchy of meanings in which the subject of the sentence (typically a substance) is qualified by the other elements of the sentence.
The grammarians of the Pāṇini’s tradition were the last to join the debate. They felt that the analyses provided by Mīmāṃsakas and Naiyāyikas, which made abundant use of Pāṇini’s grammar, ignored the most important contribution that this grammar could make: its insistence that action is the central element of a sentence. The analysis proposed by these grammarians therefore put the action at the center of sentence analysis, action being the unit that is qualified by its other semantic elements (Diaconescu, 2012; Tatacharya, vol. I, 2005–2008).
One issue that sparked opposition among thinkers in the Brahmanical tradition concerned the question as to whether the relation between words and their meanings is natural. For the Mīmāṃsakas there was no question: this relation is natural and without beginning (Bronkhorst, 2012). The Nyāya-Vaiśeṣika school adopted a different position, maintaining that it is conventional: at some time in the past, meanings had been assigned to words. Most of the surviving texts of the school add that this had been done by God, the creator of the present universe. This postulated conventionality did not, at least not until a recent date, mean that these thinkers were willing to consider that people could follow their whims in assigning meanings to words. These thinkers, too, were Brahmans, and they, too, talked about the same fixed and unchangeable language as the Mīmāṃsakas, namely, Sanskrit. The conventionality they spoke about was related to the question of whether words had their meaning by nature, or rather had acquired it as a result of a decision, preferably one taken by God. Nyāya-Vaiśeṣika opted for the second alternative.
Whether or not the relation between a word and its meaning was natural or conventional, Brahmanical thinkers generally held that each word has one primary meaning; contextual factors may subsequently oblige an interpreter to resort to a secondary meaning. Considerations of this kind were central to the reflections of the Mīmāṃsakas in their attempts to find the correct interpretation of the Veda: these attempts can only bear fruit if one gets as close as possible to the text and lets it speak for itself; obviously the primary meanings of words are closer to the text than derived, secondary meanings.
The units of meaning smaller than words are the minimal grammatical elements known from Pāṇini’s grammar. Discussions about their meanings took Pāṇini’s analysis as point of departure, but could deviate from it with respect to the precise meanings of those elements. Discussions about these meanings attracted a lot of attention in the more recent literature of the schools concerned (Mīmāṃsā, Nyāya, grammar), and we can only consider one example: the meaning of the verbal ending. In a simple sentence like caitraḥ odanam pacati (Caitra cooks rice), the verb is pacati (cooks), the verbal ending -ti (which corresponds to English “-s”). The grammarians, following Pāṇini, assigned the meaning “agent” to this ending. The Naiyāyikas, more inclined to analyze the sentence in terms of a subject (in this case the individual called Caitra) qualified by various features, assigned the meaning “activity” (kṛti) or “effort” ( yatna) to the verbal ending, so that they could paraphrase the sentence along the following lines: “Caitra is characterized by the activity, or effort, of cooking rice.” The Mīmāṃsakas, finally, wanted to push their bhāvanā (bringing into being) to the fore and assigned this meaning to the verbal ending. The inevitable result was that their analysis of the above sentence took something like the following shape: “The bringing into being whose agent is Caitra and which leads to the softening of rice.” Even this simple (and simplified) example will make clear that the discussion of the meaning of the individual grammatical elements cannot be separated from considerations as to the meaning of the sentence as a whole.
This is what may have happened in the case of the claim (Ingalls, 1954; Staal, 1965/1988), according to which Pāṇini’s grammar provided methodical guidelines to science and philosophy in India, the way Euclid did in the history of Western science and philosophy. There can be no doubt about the appeal of this claim, but little has been done to test it against the evidence. An investigation into the possible influence of Pāṇini’s grammar on Indian geometry has not yielded conclusive results. It seems therefore safer, for the time being, not to exaggerate the effect that Pāṇini’s grammar has had on philosophical and scientific thought in India. The effect exerted by notions about language in general, however, is clear and wide, as the preceding pages have indicated (Bronkhorst, 2001).
The point of departure is, and has to be, that the main actors in Indian philosophy held a number of views about language, which they considered either self-evident or too deeply anchored in their worldviews to allow for critical discussion. These views were therefore not normally objects of, but rather starting points for, inquiry.
These main actors were for a long time primarily Brahmans and Buddhists. Brahmans and Buddhists did not share the same views about language, and indeed, their positions were often radically opposed to each other. In spite of these divergences, a remarkable convergence took place in the course of time, which led to a situation in which Brahmans and Buddhists agreed on a number of far-from-obvious points. In order to understand this convergence, it will be necessary to discuss the different starting points and early developments of Brahmanical and Buddhist thought.
Brahmanical Presuppositions and the Birth of Sanskrit Linguistics
Scholars are well informed about religious conceptions that were prevalent in Brahmanism at an early age, because this tradition has preserved for us a corpus of texts, at least part of which goes back to a time before any script was used in the Indian subcontinent. This corpus of texts, collectively known as the Veda, consists in part of sacrificial formulas, or mantras. Mantras had to be recited at special, mainly ritual, occasions and were believed to be vital for the efficacy of those rites. The belief in the efficacy of mantras had a determining effect on the Brahmans’ attitude toward language.The language of the Brahmanical mantras was Sanskrit. Modern linguists may be tempted to point out that the language of many of the vedic mantras is an older form or even a predecessor of classical Sanskrit, which they may give a different name – for example, “Vedic” or “Old Indo-Aryan”; for the Brahmanical tradition, the Vedic language and classical Sanskrit were one and the same language, even though the Sanskrit of the Veda admittedly often used words and grammatical forms that were not commonly used in classical Sanskrit.
This conviction that Vedic and classical Sanskrit were one and the same language was important for the Brahmans. It allowed them to think of this language as unchangeable and even as eternal and without beginning. Indeed, in postvedic times, the vedic corpus itself came to be looked upon as eternal and beginningless. This raised questions that will be discussed below.
At this point I will consider two consequences of the belief in the efficacy of mantras. This belief was particularly important for Brahmans, for their livelihood depended on it. Brahmans owed their special position in society primarily to their knowledge of the Veda: they could recite portions of it and use its mantras at appropriate occasions. But to be efficacious, these mantras had to be recited in a phonologically correct manner. One of the consequences of the belief in the efficacy of mantras was therefore that major attempts were made to describe the phonology of those mantras, and secondarily of Sanskrit in general. A rich literature on the phonology of Sanskrit arose, much of which has survived until today (Deshpande, 1997, 31ff.; Scharfe, 1977, chs. 5, 7).
There was a second consequence of the belief in the efficacy of mantras, less immediately visible in the form of a surviving literature, but equally influential. The efficacy of Sanskrit mantras showed, at least to the Brahmans, that Sanskrit, their eternal language, was very close to objective reality. Sanskrit mantras could have an effect on the objective world because a close link connected Sanskrit with that objective world. Sanskrit distinguished itself in this regard from the dialects spoken by less eminent people. These dialects from our point of view were languages in their own right, and for the Brahmans no more than corruptions of the only real language, Sanskrit. No such thing as historical linguistics could therefore exist from the Brahmanical point of view. The one real language, Sanskrit, did not change and had not changed from beginningless time. All other languages were not real languages but corruptions due to people too lazy or ignorant to learn the one real language.
The conviction that Sanskrit is close to reality is central to Brahmanical thought. It finds expression in various ways. Most notable among these is the belief that the form of a word tells us something about the object it designates. This belief is behind the numerous semantic etymologies found in vedic literature, primarily in the vedic texts called Brāhmaṇas These etymologies have nothing whatsoever to do with the history of words, as in the case of historical etymologies studied nowadays by historical linguists. Semantic etymologies are something different altogether. They start from the assumption that the similarity between words reveals a common feature shared by the things designated. Semantic etymologies assume, at least in vedic literature, the presence of hidden links between objects, links that can only be brought to light through the analysis of the words that designate them.
Interestingly, the belief in the significance of semantic etymologies gave rise to an early attempt at systematization known as nirukta, “etymology.” Its classical treatise, also called Nirukta, was composed by a certain Yāska, who must have lived after Pāṇini but before Patañjali (see below), and therefore in or around the 3rd century BCE. This discipline, considered a vedic auxiliary science (vedāṅga), attempted to formulate the rules that govern semantic etymologizing. One of its principal rules, unsurprisingly, is that in establishing links between different words, considerations of meaning have to play a central role.
It is against this same background that we have to understand the interest in Sanskrit morphology and linguistics in general that characterized the late vedic period. This interest culminated initially in the famous grammar of Pāṇini (4th cent. BCE), which itself became the starting point for an extensive grammatical literature that consisted partly in commentaries on Pāṇini’s text, the Aṣṭādhyāyī, and partly in independent treatises that were yet profoundly influenced by Pāṇini’s grammar (Bronkhorst, 1981)
We will see below that the beliefs about the Sanskrit language just described came to exert an influence on Brahmanical philosophy. These beliefs themselves did not yet amount to a philosophical position, but they shaped future thinking.
Some scholars have made a somewhat different claim. They have drawn attention to the central significance of language study in Indian culture, and on the sophisticated way in which the grammarian Pāṇini in particular succeeded in dealing with complex linguistic issues. They have concluded from this that Pāṇini’s grammar played a role in India comparable to Euclid’s geometry in Europe in becoming a methodical guide for more recent philosophers. In support of this claim, they can point out that the study of Pāṇini, or one of his successors, was fundamental in the classical formation of scholars, so much so that all of them were acquainted with the grammatical method developed in these works. However, it is very hard, perhaps impossible, to give more substance to this claim. No one has yet been able to show in what way philosophical thinking followed a pattern derived from Pāṇini’s grammar. It is, however, beyond doubt that grammatical rules and arguments are frequently cited in certain philosophical developments, some of which will be considered below. The influence of the brahmanical presupposition as to the close connection between language (i.e. Sanskrit) and reality will become clear as the discussion proceeds.
Buddhist Systematic Philosophy and the Role of Language
Buddhist thinkers had no special attachment to the Sanskrit language, nor to any other language for that matter. They preserved what they believed were the words of the Buddha in different languages, often a language close to the one spoken in the region where they had settled. It is true that in the long run, some Buddhists came to think that the language in which they preserved the words of the Buddha was the language in which the Buddha had preached. Some went further and claimed, no doubt under Brahmanical influence, that the language of their canon was the original language from which all other languages derived; this happened in the case of Pali, a language that the Buddhists concerned call Magadhi, namely, the language of Magadha, the region where the Buddha had preached. Brahmanical influence took a different shape in northern India, where the Buddhists, having preserved the Buddha word in local languages for half a millennium, changed over to Sanskrit and even came to believe that the ancient dialect in which some of their texts had been preserved was a form of vedic Sanskrit (Bronkhorst, 1993).But whatever the language they used for religious purposes, the Buddhists were not, at least not initially, inclined to believe that any of these languages – or any other language for that matter – had a particularly close connection with reality. However, developments now to be sketched led to a situation where a number of Buddhists came to believe that there was a close connection between language and our experience of the world.
This connection with language was initially not obvious in the philosophical developments that the Buddhist school, called Sarvāstivāda, in northwestern India underwent during the final centuries preceding the Common Era. For reasons that may be connected with the Hellenistic surroundings in which they found themselves, these Buddhists made an attempt to systematize the teachings they had inherited, ultimately from the Buddha, they thought. These inherited teachings contained lists of items that had been extracted from the sermons of the Buddha. These items frequently referred to mental states, some others to physical elements. The name that came to be used for these items was dharma, and the Buddhists of northwestern India developed the idea that these dharmas are the ultimate constituent elements of human beings, and by extension of everything else as well. For various reasons, these Buddhists also concluded that the dharmas must be momentary. On top of this, they stated that composite objects have no separate existence. In other words, a person is nothing but a collection and sequence of a large number of small and momentary entities; beside these numerous entities, there is nothing that one might call a person. The same applies by extension to macrocosmic objects in the external world: for example, there is no such thing as a chariot; there is only this vast sum of momentary constituent elements (Bronkhorst, 2009, 55ff.).
This remarkable vision of the world has but little in common with the original teaching of the Buddha, and the Buddhists of northwestern India had some difficulty anchoring it in the inherited Buddha words. However, as a philosophy, this vision came to flourish and exert a determining influence on most of the subsequent philosophical developments in Indian Buddhism. The question is: What has this vision to do with the relation between language and reality?
This relation enters into the picture in a manner that can be usefully illustrated with the example of a chariot. We see and believe in the existence of a chariot, when in reality there is no such thing. How are we misled into entertaining such an incorrect idea? As a result of language. We believe there is a chariot, because there is the word “chariot.” The same applies to all the numerous other things we believe populate the world. In reality, they do no such thing, but we are tricked by the language we speak.
Two observations have to be made at this point. The relation postulated by these Buddhists between language and experience does not concern reality as it really is. In deepest reality there are only vast numbers of momentary dharmas, as we have seen. The experience, whose parts correspond to the words of language, is not a reliable representation of reality. In other words, from this Buddhist perspective, language is not closely related to reality. Quite on the contrary, language tricks us into believing in a world that is ultimately not real. The parts of language that do the tricking are its words. Chariots, persons, and so many other things are ultimately unreal, but are believed to be real because of the words that supposedly designate them.
The systematizations of the Sarvāstivāda school of Buddhism appear to be the first manifestations of systematic philosophy on the Indian subcontinent. They were not primarily concerned with language, yet language played an important role in them, as we have seen. The words of language are responsible for the fact that we believe, incorrectly, that we live in a world in which there are chariots, persons, and other such things. However, this vision of the Sarvāstivādins had to face an obvious difficulty. Words, like everything else, are themselves accumulations – most specifically, successions – of dharmas. As such, they have no existence of their own, like all other accumulations of dharmas. This is problematic in that our incorrect beliefs in objects, such as chariots and persons, depend on words. If there are no words, what then is responsible for our mistaken conceptions?
It seems that the early Sarvāstivādins experienced this as a problem. This we may conclude from the fact that they included among their lists of existing dharmas a number of items that correspond to words and sentences. In other words, they postulated, beside the succession of auditory and other elements that make up the word “chariot,” the existence of a momentary dharma that somehow represents the whole word “chariot.” This went of course against their general principle of not recognizing the existence of wholes, but strictly speaking they managed to avoid this inconsistency. The “wholes” postulated in the case of words (and sentences) were, strictly speaking, not wholes, but entities (dharmas) in their own right, which somehow accompanied the succession of auditory elements that gave expression to that meaning. In this complicated way, the Sarvāstivādins had again words, and they could again claim that our illusion concerning the world was due to these words. These linguistic dharmas of the Sarvāstivādins might be looked upon as an idiosyncrasy of the philosophy of these Buddhists without deeper significance. Such a conclusion would not be justified, if for no other reason than that more recent brahmanical thinkers happily drew inspiration from these Sarvāstivāda inventions in order to deal with problems they had to face (Bronkhorst, 1987, ch. 3/7).
Reality as Language Incorporated
The Sarvāstivāda philosophy constituted a challenge that Brahmanical thinkers could not leave unanswered. Buddhism possessed in this philosophy a coherent and well-thought-out system of thought, at a time when Brahmanism had nothing of the sort. It appears that Brahmanism had to come up with something comparable, so as to be able to defend itself in the public debates that sometimes took place at the royal courts. The outcome of such debates could be more than academic and might affect the court’s willingness to support one group rather than another (Bronkhorst, 2007).In this situation Brahmanism developed two ontologies of its own. The one that is most interesting in the present context is called Vaiśeṣika. It is most interesting in that it takes over the Sarvāstivāda idea that the world of our experience has a close connection with the words of language, but adjusted to its own Brahmanical presuppositions. In Vaiśeṣika the world of our experience coincides with the real world, and the only language that is taken into consideration is Sanskrit. For Vaiśeṣika, then, language is not a source of confusion that makes us believe in the external reality of a world that does not in that form exist, as it is for Sarvāstivāda. Quite on the contrary, for Vaiśeṣika language (i.e. the Sanskrit language) is a source of information about the real world. If we wish to develop an ontological scheme, namely, a scheme of what there is, the Sanskrit language is the means par excellence to find it.
This fundamental conviction of the Vaiśeṣikas finds expression in their philosophy. Vaiśeṣika claims that all things (artha) in the world fall into three categories: substances, actions, and qualities. These three categories correspond to the three main types of words already distinguished by the grammarian Patañjali (2nd cent. BCE): nouns, verbs, and adjectives. In order to make this ontological scheme coherent, various Vaiśeṣikas added a number of further categories, arriving at a varying total of six, seven, ten, or even more. However, even in their own understanding of their system, these further categories were add-ons.
The three main categories had subdivisions. The fundamental rule followed to find those subdivisions was that they had to correspond to words. The list of substances illustrates this particularly well: it corresponds by and large to the nouns of the Sanskrit language. The list of qualities raises more difficult issues, for they are not all referred to by adjectives or other words; some of them, in particular, are designated by nouns. Questions about their existence are resolved by considering language use. The quality “number” must exist, because expressions such as “one pot,” “two pots,” and so on are common usage. The existence of the quality “dimension” (pramāṇa) is shown by the use of the word “measure” (māna). There are further examples of this kind in the classical surviving standard work of Vaiśeṣika, the Padārthadharmasaṃgraha of Praśasta (or Praśastapāda, c. 6th cent. CE). Vaiśeṣika is in this way not so much a philosophy of language, as it is a philosophy inspired by language, or rather: a philosophy inspired by certain beliefs about the Sanskrit language (Bronkhorst, 1992).
Vaiśeṣika ontology became preponderant in Brahmanism, especially in those schools of philosophy that developed ideas on language (Mīmāṃsā, Nyāya, Bhartṛhari). The other main early Brahmanical ontology, Sāṃkhya, was not inspired by conceptions about language in the way Vaiśeṣika was. This does not mean that it denied the close connection between language and reality. This will become clear in the discussion below of the “linguistic crisis” that hit all Indian schools of philosophy – Brahmanical, Buddhist, and even Jain – in the early centuries of the Common Era.
The Word as Source of Knowledge
Before discussing the “linguistic crisis,” we have to pay attention to another important development in Brahmanism that shaped the way it came to think about language: the interpretation of the Veda, known as Mīmāṃsā, whose main classical text is the commentary of Śabara (c. 5th cent. CE). Mīmāṃsā did not initially present itself as a school of philosophy, but as a school of vedic hermeneutics. Recall that Brahmanism had preserved an enormous corpus of literature, the Veda, at least in part in the form of memorized texts. The vedic corpus contains mantras, as we have seen. It also contains texts globally referred to by the Mīmāṃsakas as Brāhmaṇas: texts in prose related to the solemn vedic sacrifices. The Mīmāṃsakas were primarily interested in these Brāhmaṇas, in which they expected to find matter useful or essential for the performance of those rites: injunctions, information as to how to proceed and resolve apparent contradictions, and so on. They based their endeavors on some global assumptions about the Veda.Most important among these assumptions was the claim that, like its language, the Veda itself, too, had no beginning: the Veda had always been there. This is of course only possible if the Veda had had no author. Being without author, the Veda is pure speech. This, from the Mīmāṃsā point of view, has important consequences. To understand these, consider first ordinary verbal communication between human beings. In ordinary verbal communication, there will be a speaker, a message expressed in language, and a listener. From the point of view of the listener, the received message will be reliable if two conditions are fulfilled: (1) the speaker is reliable, and (2) the message is correctly interpreted by the listener. In the case of the Veda, there is no speaker who formulated the message initially. The chance that the Veda is used to mislead us is therefore excluded, and we must conclude that the message of the Veda is reliable, if only we know how to interpret the text correctly. This is the self-assigned task of Mīmāṃsā.
How does one interpret a text correctly? Once again, it is a matter of eliminating sources of error. The aim is to get as close as possible to the text, letting it speak for itself. Mīmāṃsā developed a whole list of principles of interpretation, which are in the end nothing but concrete manifestations of this general aim and are justified in this manner.
The belief that the Veda is without beginning has some consequences that have to be taken into consideration. Being without beginning, it is not posterior to any event that has ever taken place in the history of the world. Yet many vedic passages give the impression of relating such events, usually in the form of what we might call myths. None of these stories can be taken literally, because the Veda had already existed for an eternity at the time when these events supposedly took place. Stories and other references to historical events must therefore be interpreted differently. They must be read in context and have their place in the Veda in order to illustrate or lend support to vedic contents that can be taken literally.
What are the vedic contents that should be taken literally? The process of elimination shows what they are. All statements of fact in the Veda must be interpreted metaphorically. This is partly due to the fact that the Veda cannot refer to historical events, for the reason indicated above. However, other, nonhistorical, statements of fact contained in the Veda have another disadvantage. They might turn out to be in contradiction with our sense experience. This would be serious, for it would imply competition between two different means of knowledge: the vedic word, and perception. This difficulty does not exist in the case of injunctions. Injunctions tell us what we must do. No sense experience can ever be in conflict with an injunction, because perception informs us about states of affairs, not about obligations. With regard to obligations, we have only one ultimate source of information, namely, the Veda.
Mīmāṃsā arrives in this way at an interpretation of the Veda in which injunctions are central. They are what the Veda is all about. Everything else is to be interpreted in connection with those injunctions. In practice, the task is complex, but the theoretical basis is relatively simple and straightforward. In the end it amounts to this – that the word par excellence, the Veda itself, tells us what to do (Bronkhorst, 1997).
The second half of the 1st millennium CE saw the rise into prominence of a school that claimed to be no more than an improved version of classical Mīmāṃsā, but which opened up a path for completely new developments. This school – which referred to itself by various names, among them Śārīrakamīmāṃsā and Brahmamīmāṃsā, later also Uttaramīmāṃsā, and one of whose most important early authors was Śaṅkara (c. 700 CE) – came to be known by the name Vedānta, or Vedāntism. It owes this name to the fact that the parts of the Veda known as vedānta (end of the Veda), better known as Upaniṣads, played an important role in its reflections. The importance of these portions of the Veda is, however, in a certain sense due to coincidence. The Brahmamīmāṃsā applied the same principles as classical Mīmāṃsā, but slightly improved them with the result that certain upanishadic statements gained center stage. Recall that classical Mīmāṃsā had maintained that only in the case of injunctions can we be sure that there will be no conflict with other means of knowledge. The Vedāntists added vedic statements that teach knowledge about brahman ; these vedic statements happen to occur in the Upaniṣads. They argued that knowledge about brahman can only be obtained from these vedic statements, not by any other means of knowledge. Since no conflict between different means of knowledge is therefore possible, these statements have to be taken literally (Bronkhorst, 2007).
The Nyāya school of thought was willing to agree with Mīmāṃsā in its claim that the word – and especially the linguistic expressions that are found in the Veda – is a reliable source of knowledge. Unfortunately, and for reasons that will be discussed below, it adopted the position that the relation between words and things was conventional. For them, therefore, the first of the two conditions specified above (a received message is reliable if the speaker is reliable and the message is correctly interpreted by the listener) retains its pertinence even in the case of the Veda. Unlike the Mīmāṃsakas, the Naiyāyikas had to argue that the Veda had a trustworthy author, and indeed that the reliability of the Veda depended on the trustworthiness of that author. In an important sense, they had to admit that the reliability of the Veda depends on the same factors that make an ordinary statement reliable, that is, the trustworthiness of the speaker. They solved this problem by attributing the Veda to the most reliable person there is: God himself. This did not change the fact that their position – to the extent that the word (namely, linguistic utterances) is a reliable source of knowledge – was bound hands and feet by the supplementary requirement that the speaker, who uttered these words, is himself reliable (NyāS. and NyāBh. 2.1.50–67; trans. Jha, 1939).
We saw in the preceding section that the Vaiśeṣika ontology, too, is ultimately based on the word as source of knowledge. The situation here is nonetheless different from the one described in connection with Mīmāṃsā and Nyāya. There (in Mīmāṃsā and Nyāya) I talked about the reliability of certain linguistic utterances: primarily those contained in the Veda, further those pronounced by reliable speakers. In the case of Vaiśeṣika, however, I considered the relationship between the structure of reality and the structure of the Sanskrit language; the Veda or other reliable statements did not play a role here. Interestingly, some Buddhist thinkers, most notably Dignāga (6th cent. CE), came to adopt an ontological position similar to that of Vaiśeṣika, ultimately for almost the same reason. Dignāga could not, of course, claim that language would tell us something about the world as it really is, for he, like most other Buddhists, thought that language is rather the source of our mistaken ideas about reality. However, language, being the reason why we experience the world the way we do, is capable of providing us with information about the world of our experience. In order to find out more about this, Dignāga undertook an analysis of the relationship between words and things, which I will discuss below. Here it must suffice to observe that he arrived at an ontological scheme similar to that of Vaiśeṣika, be it that in his case this ontological scheme was limited to the aspect of reality (ultimately unreality) ruled by language (Katsura, 1979).
The Linguistic Crisis
We have seen that a convergence between Brahmanism and Buddhism took place. Both Brahmans and Buddhists had come to believe that there is a close link between language and the world of our experience. This shared position hides some profound differences, to be sure. For Brahmans, only Sanskrit counted as language; the Buddhists imposed no such limitation. Second, the world of our experience was, for the Brahmans, the world as it really is, the real world. The Buddhists, however, admitted a close correspondence between language and reality only for the false reality that we believe we inhabit, but that has no ultimate reality at all. In spite of these differences, the points in common between Brahmans and Buddhists were responsible for a certain development in Indian philosophy, which affected them both.This development is likely to have begun within Buddhism, and it is possible, though not certain, that Nāgārjuna (2nd cent. CE) was responsible for it. Buddhist philosophers maintained that the world of our experience is not real. They had made this the basis of their philosophies, but they had never been able to prove it in an objectively convincing manner. This changed when people started paying attention to the relationship, not just between words and things, but also between statements and what they signify. It was believed that statements describe situations that consist of the things denoted by the constituent words of the statement. Most ordinary statements can be thought to confirm this belief. The statement “the book lies on the table” consists of three words in Sanskrit and describes a situation constituted of three “things”: the book, the table, and the activity of lying. Most Indian thinkers looked upon statement in this manner, thus accepting, though often only implicitly, the “correspondence principle”: the words of a statement refer to the “things” that constitute the situation described by that statement.
A problem arises in the case of declarative statements that describe a situation in which something is produced or comes into being. The situation described by “the potter makes a pot” contains a potter and the act of making, but no pot. Most Indian philosophers considered this problematic. What is more, many were willing to concede that this problem tells us something important about the nature of the world.
The Buddhist Nāgārjuna used this problem in a variety of ways to show that something is fundamentally amiss. Many of his arguments are variants of the following question: If there is something that is produced, what is the need of producing it? If there is not, how can it be produced?
Identifying the problem is not the same as offering a solution to it. We will see that different thinkers came up with different solutions. Nāgārjuna himself concluded that difficulties of this kind show that nothing really exists: everything is empty (śūnya). His philosophy is for this reason frequently referred to as śūnyavāda (doctrine according to which the world is empty of essence), beside its other name, Madhyamaka.
As stated above, there is no compelling reason to agree with Nāgārjuna’s conclusion, even if one accepts that statements like “the potter makes a pot” represent a major difficulty. Brahmanical thinkers in particular would hesitate to admit that nothing really exists. This position was close to the Sarvāstivāda Buddhist belief according to which the world of our experience has no more than conventional reality. Brahmanism was not ready to follow Buddhist thinkers in this regard. A relatively straightforward way out was to claim that the pot in “the potter makes a pot” is somehow present in the situation described. Since the potter who makes a pot is working with clay, one only had to assume that the pot was in some form or other already present in the clay. This position became known as the satkāryavāda (doctrine according to which the effect [kārya, i.e. the pot, in our example] is already present [sat] in its cause [the clay, in our example]). The Sāṃkhya philosophy adopted this solution. This, however, forced it to introduce some important changes into its system. Indeed, what does it mean that the pot is already present in the clay? This claim only makes sense if clay is considered to be the essential feature of a pot. Everything else – its shape, its color, and other characteristics – has to be looked upon as secondary. Only thus can one maintain that what is truly essential to the pot, namely, its substance, is already there before the potter has finished his job. However, Sāṃkhya had thus far believed that substances are no more than accumulations of qualities. This early position now had to be abandoned. This was done very thoroughly, so much so that one finds no trace of it in surviving Sāṃkhya texts. The texts by critics of the Sāṃkhya system, however, could not be so easily suppressed or “forgotten.” These texts belonging to other philosophical schools constitute, therefore, our most important source of information about the early history of Sāṃkhya, before the linguistic crisis obliged it to accept the satkāryavāda and introduce the modifications required by this new doctrine.
Jain philosophers followed a similar yet different way to solve the problem, adopting the so-called anekāntavāda (doctrine according to which it is one way in one respect, different in other respects). With regard to the pot, this means that, from the perspective of substance, clay and pot are the same; from the perspective of shape, they are different. This was considered sufficient to account for sentences like “the potter makes a pot,” for from the perspective of its substance, the pot is already there while the potter does his job.
The one school of philosophy that was not deeply affected by the linguistic crisis was Sarvāstivāda Buddhism. “Sarvāstivāda” means “doctrine according to which everything, namely past and future objects, exist.” The school had adopted this doctrine for reasons that have nothing to do with language, but once adopted, this doctrine came in quite handy in the new situation. These Buddhists believed that future pots exist as much as present and indeed past pots. In the statement “the potter makes a pot,” there is therefore a pot (a future pot) in the situation described. Unlike most other philosophers of their day, the Sarvāstivādins thus survived the linguistic crisis undamaged.
Nāgārjuna had drawn the radical conclusion that nothing exists. Others, most notably the mysterious author Gauḍapāda, drew the equally daring conclusion that nothing can come into being. His was the ajātivāda (the doctrine according to which nothing is produced). Gauḍapāda came to be claimed by Advaita Vedānta (Gauḍapāda came to be looked upon as the teacher of the teacher of the famous Śaṅkara), but parts of the text attributed to him (the Āgamaśāstra or Gauḍapādakārikā) are recognizably Buddhist.
It will be clear from what precedes that the linguistic crisis was responsible for many of the vādas (doctrines) that are characteristic of various Indian philosophies: satkāryavāda, śūnyavāda, anekāntavāda, ajātivāda, and to some extent even sarvāstivāda. Its impact, however, went further. Some schools refused to adopt any of the options so far considered and decided to draw conclusions that were not of an ontological nature, but were rather based on a reflection of the way words denote their objects. The problem so far considered depends vitally on the assumption that the word “pot” in “the potter makes a pot” refers to an individual object in the situation described. The Nyāya-Vaiśeṣika school of thought, having explored other solutions, ended up pointing out that the word “pot” in that sentence does not have to refer to an individual object, but rather (more correctly, also) to the universal that inheres in all pots. Indeed, all pots share that name because they have something in common, namely, a shared universal. Since the pot universal inheres in all pots – past, present, and future – it cannot but be eternal. Being eternal, it is always there, also in the situation described by “the potter makes a pot.” The new reflection about the way words denote things solved the problem (there was now something corresponding to “pot” in “the potter makes a pot”), and no further ontological conclusions needed to be drawn. In particular, it was not necessary to claim that the effect (the pot) exists already in its cause (the clay) before it is produced. In Nyāya-Vaiśeṣika, therefore, there was no place for satkāryavāda, and their solution came to be known by the name asatkāryavāda (the doctrine according to which the effect is not already present in its cause).
It is noteworthy that the two principal early Brahmanical positions, satkāryavāda and asatkāryavāda, can be read as adaptations of two positions proposed in Patañjali’s Mahābhāṣya, a major grammatical work in the Pāṇini’s tradition, dating from the middle of the 2nd century BCE, long before the linguistic crisis made itself felt. Patañjali discussed the question of the object denoted by a word. According to Patañjali, this object must be eternal, and he proposed two conceivable candidates: (1) words refer to forms, and (2) words refer to substances. In the first case, forms must be thought of as eternal, contrary to individuals, which are not. In the second case, substances are eternal. (The discussion becomes somewhat complicated on account of the fact that both “individual” and “substance” translate the same Sanskrit word dravya.) Different authorities held different positions in this matter; Patañjali himself thought that words signify both form and dravya. It is easy to see that the position according to which words refer to eternal substances is close to the satkāryavāda of the Sāṃkhyas. It is equally obvious that the alternative position, according to which words refer to eternal forms, is close to the one in which words refer to eternal universals, that is, the asatkāryavāda. These positions drew, therefore, inspiration from an old and venerable grammatical treatise, so as to arrive at a solution to a problem that had not yet been known to Patañjali.
The Nyāya school was not the only one to propose a semantic solution to the problem of the production of things. The Buddhist thinker Dignāga, mentioned earlier, came up with a solution of his own, which, however, shared with Nyāya the conviction that words do not refer to individuals. However, Dignāga had no place for universals in his ontology, so he could not simply copy the Naiyāyikas’ solution. The solution he proposed was inspired by his work in logical deduction.
Note, to begin with, that Dignāga sharply distinguishes between the world ruled by language, and the deeper reality that hides behind it. In his opinion, words do not apply to this deeper reality, which is the object of perception alone. Perception bears on the particular (svalakṣaṇa), language on the general (sāmānyalakṣaṇa). Ultimately, the objects of our perception are real, while the realm covered by language is not.
In the realm of language, words refer to their objects by way of exclusion (apoha): a word, say, “pot,” excludes everything that is not a pot. The word “pot” applies in this way to the pot, but does not denote it.
Dignāga assimilated this process of referring to the process of inference. He was among the first to have realized that an inference of the type “if there is smoke on the mountain, then there is fire” is logically equivalent to “if there is no fire on the mountain, then there is no smoke”; if B follows from A, not-A follows from not-B, and vice versa. Analogous applies in the case of referring: saying that all pots, and only those are referred to by the word “pot,” is equivalent to saying that all things that are not pots are not referred to by this word. Dignāga arrives in this way at an understanding of referring that does not postulate the existence of universals, or of a special link between words and individuals. The statement “the potter makes a pot” is no longer problematic for him, for there is no longer supposed to be a specific link between the word “pot” in this statement and the individual pot the potter is making (Bronkhorst, 1999).
Bhartṛhari
Bhartṛhari is often referred to as a “philosopher of grammar.” He certainly was a grammarian, and he was also a philosopher. He used grammatical notions in his philosophy, but it is open to debate whether he was a philosopher of grammar.Consider his ontological position in broadest outline. For Bhartṛhari, the totality of all that exists is the one encompassing reality, which he sometimes refers to as brahman. Though one and indivisible, this totality is yet divided in accordance with the words of language. These divisions are, however, less real than the encompassing whole from which they originate.
So far, Bhartṛhari’s philosophy is Buddhist philosophy turned on its head. In the Buddhist ontological scheme that we have come to associate with the Sarvāstivādins and others, composite entities were less real than their ultimate components; these composite entities owe their (relative) existence to the words of language. In Bhartṛhari’s philosophy, it is the other way round: composite entities are more real than their components, and these components owe their (relative) existence to the words of language.
Bhartṛhari illustrates his vision with a discussion of linguistic entities. Pāṇini’s grammar divides words into smaller entities, such as stems and suffixes. These smaller entities are the result of analysis and are artificial, in the sense that they are less real than the words from which they are extracted. A grammarian different from Pāṇini might identify different stems and different suffixes. Here, then, we find that the composite whole, the word, is more real than the elements into which it has been analyzed. The same reasoning applies to sentences and words. Words are extracted from sentences, but sentences are more real than the words into which they are analyzed. One can continue this reasoning and discover that the only really existent linguistic entity is the Veda, this huge totality of sentences. All other linguistic units are less real.
The total, undivided Veda corresponds, in Bhartṛhari’s view, to the totality of all there is, which he sometimes calls brahman. The world of our experience corresponds to smaller linguistic units, and like those smaller linguistic units, it is less real (Bronkhorst, 1992).
This ontological scheme had for Bhartṛhari more than just philosophical significance. He believed that its knowledge was vital for attaining the highest religious goal, liberation. Something similar had already been proposed by thinkers belonging to the Madhyamaka tradition, but Bhartṛhari gave it a completely Brahmanical and, what is more, grammatical twist. It is for this reason that Bhartṛhari could seriously maintain that grammar was the door to liberation. This claim has puzzled modern interpreters, who have tended to understand it as metaphorical exaggeration. It is, however, possible to take Bhartṛhari seriously in this regard. Bhartṛhari literally believed that the study of grammar was the door to liberation, because grammar was the means par excellence to understand the fundamental scheme of the universe, in which composite entities are more real than their constituent parts. Grammar shows this in the realm of words (with their constituent stems and suffixes) and sentences (with their constituent words). Once properly extended, this scheme covers the whole universe with all it contains; knowing this was for Bhartṛhari an essential step toward the highest goal: the door to liberation (Bronkhorst, 1995).
On a more day-to-day and linguistic level, Bhartṛhari owed his readers an explanation for his claim that words are more real than, and therefore different from, their constituents, and that sentences are different from their constituent words. Are words, and sentences, not simple sequences of sounds? Bhartṛhari thought they were not. According to him, a word is a different entity from its constituent elements. That is to say, beside the sequence of sounds used to communicate, say, a word, there is an altogether different “thing” that is the word itself. Bhartṛhari sometimes uses the word sphoṭa for this entity. He was not altogether original in this regard, for we have seen that the Sarvāstivādins had introduced a small number of linguistic dharmas for the very same reason. Beside similarities, there is, however, a major difference: the linguistic dharmas of the Sarvāstivādins were momentary (like practically all other dharmas), while Bhartṛhari’s sphoṭa is eternal. This may look like an important difference, but the two share an important feature: neither the linguistic dharmas of the Sarvāstivādins nor Bhartṛhari’s sphoṭa correspond to the common-sense notion of word or sentence. Common-sense words and sentences have a finite duration in time, corresponding to the time it takes to pronounce them. Bhartṛhari’s sphoṭa, like the linguistic dharmas of the Sarvāstivādins, do not exist for that length of time: they exist either for much longer (Bhartṛhari’s sphoṭa is eternal and has therefore no beginning or end in time), or for much shorter (the linguistic dharmas of the Sarvāstivādins are momentary).
It must here further be noted that, contrary to a widespread misunderstanding among modern scholars, Bhartṛhari’s sphoṭa is not, or not primarily, a semantic entity, a meaning bearer. This explains that Bhartṛhari can present a completely analogous reasoning to justify the existence of sphoṭas corresponding to individual sounds, which have no meaning. The ontological problem is the same for sounds as for words: they are sequences of vibrations and would have no separate existence but for the postulation that they exist as sphoṭas. The confusion is due to the fact that more recent authors in the grammatical tradition, who were much more interested in semantics than in ontology, came to speak of the sphoṭa as a meaning bearer (Bronkhorst, 2005).
Sentence Meaning
In language, words and their constituent parts are joined into sentences. Sentences and their parts have meanings, but the relationship between the meaning of a sentence and the meanings of its constituent parts is not immediately obvious. Is the sentence meaning no more than the accumulation of the meanings of its constituents? Or are there supplementary meanings that have been added to those word meanings? If so, which ones? Or should we rather look upon the sentence as an entity that is altogether different from its parts, with a meaning of its own that is quite independent of the meanings of those parts? These issues attracted the attention of Brahmanical thinkers from an early date onward, and they came up with a variety of answers.What seemed at first sight to be the simplest solution was adopted by the school of Mīmāṃsā, associated with the name of Prabhākara (c. 700 CE). This school maintained that a sentence expresses no more and no less than the accumulated meanings of its words. The objection that the words in a sentence are functionally related to one another, so that the meaning of the sentence goes beyond the mere meanings of its words, is answered by the claim that individual words also express their relation to other words in the sentence (anvitābhidhāna); as a result the accumulated meanings of the words amount to the meaning of the sentence.
The Prābhākara school was alone in maintaining this position. Prabhākara’s senior contemporary, Kumārila Bhaṭṭa, for example, observed that the primary meanings of words when joined up could not account for the sentence meaning; he and his followers, the so-called Bhāṭṭa school of Mīmāṃsā, as well as other thinkers, rather held that secondary meanings for words in a sentence had to be postulated so as to account for their mutual relation in a sentence (abhidhānānvaya).
A third position preferred to look upon the sentence as a separate entity that expressed its meaning independently of the meanings of the words that seemed to constitute it. We have seen that Bhartṛhari in particular advocated the separate existence of the sentence. It is therefore not surprising that those who followed him, primarily grammarians in the Pāṇinian tradition, also accepted the separate semantic role of the sentence.
All except the Prābhākaras were confronted with the need to understand how exactly the meanings of words and sentence are related to each other. This need was particularly acute in the reflections of the Mīmāṃsakas, the Vedic hermeneuts who had come to the conclusion that injunctions are the most essential part of the Veda, around which other vedic statements are to be interpreted. It is in this school that elaborate analyses of the sentence meaning, in terms of its constituent elements, began. Other thinkers, most notably those belonging to the Nyāya school (which had meanwhile absorbed the Vaiśeṣika school), followed later. Later again the school of grammar that derived its inspiration from Pāṇini and Patañjali joined the debate.
Concentrating on the vedic injunctions, the Mīmāṃsakas had to determine which part of any particular injunction expresses the injunction itself and arrange all other grammatical units around this central element as qualifying it. They found that, in a statement like svargakāmo yajeta (he who desires heaven should sacrifice), the bare injunction (“should”) is expressed by the verbal ending (-ta). Specifications as to the details of the injunction have to be derived from the other parts of the sentence. The Mīmāṃsakas introduced in this manner a hierarchy of meanings, centered around the semantic element that is qualified by all the others. This central semantic element they called bhāvanā (bringing into being). About this “bringing into being” we need to know its purpose, the means by which the purpose is brought about, and how it is brought about. The simple injunction svargakāmo yajeta (he who desires heaven should sacrifice) thus gives expression to a bringing into being by means of a sacrifice, with heaven as its purpose. (The “how” is not further specified in this example; most vedic injunctions provide information about the kind of sacrifice to be performed.)
The Naiyāyikas felt that they could not agree with the Mīmāṃsā analysis of the sentence. Not only did they have a different view as to the independence of vedic statements (unlike the Mīmāṃsakas, they did not look upon these statements as fully independent, but rather as utterances of God), but they were also committed to an ontology in which substances play a central role. Indeed, of the three categories discussed earlier (substances, actions, and qualities), substances are central, for actions and qualities can only exist if they reside in substances: substances are qualified by actions and qualities. A typical statement (such as “the blue bird flies”) indicates how a particular substance (the bird) is qualified by an action (it flies) and a quality (it is blue). The Naiyāyikas therefore proposed a hierarchy of meanings in which the subject of the sentence (typically a substance) is qualified by the other elements of the sentence.
The grammarians of the Pāṇini’s tradition were the last to join the debate. They felt that the analyses provided by Mīmāṃsakas and Naiyāyikas, which made abundant use of Pāṇini’s grammar, ignored the most important contribution that this grammar could make: its insistence that action is the central element of a sentence. The analysis proposed by these grammarians therefore put the action at the center of sentence analysis, action being the unit that is qualified by its other semantic elements (Diaconescu, 2012; Tatacharya, vol. I, 2005–2008).
The Meanings of Words and Their Parts
As stated above, the interest in semantic questions grew over time, especially in the Brahmanical schools of Mīmāṃsā and Nyāya-Vaiśeṣika. This interest had old and venerable antecedents, to be sure. Pāṇini’s grammar assigned meanings to the different grammatical elements it introduced, most particularly suffixes and verbal roots. In spite of the holistic tendencies of authors like Bhartṛhari, or perhaps in part because of them, Indian thinkers continued to show an interest in the meanings of words and, ultimately, of minimal grammatical elements. Their concern with the analysis of sentence meaning only strengthened this interest, not least because certain ways of analyzing the sentence depended on certain ways of understanding the meanings of its elements. I will consider examples below.One issue that sparked opposition among thinkers in the Brahmanical tradition concerned the question as to whether the relation between words and their meanings is natural. For the Mīmāṃsakas there was no question: this relation is natural and without beginning (Bronkhorst, 2012). The Nyāya-Vaiśeṣika school adopted a different position, maintaining that it is conventional: at some time in the past, meanings had been assigned to words. Most of the surviving texts of the school add that this had been done by God, the creator of the present universe. This postulated conventionality did not, at least not until a recent date, mean that these thinkers were willing to consider that people could follow their whims in assigning meanings to words. These thinkers, too, were Brahmans, and they, too, talked about the same fixed and unchangeable language as the Mīmāṃsakas, namely, Sanskrit. The conventionality they spoke about was related to the question of whether words had their meaning by nature, or rather had acquired it as a result of a decision, preferably one taken by God. Nyāya-Vaiśeṣika opted for the second alternative.
Whether or not the relation between a word and its meaning was natural or conventional, Brahmanical thinkers generally held that each word has one primary meaning; contextual factors may subsequently oblige an interpreter to resort to a secondary meaning. Considerations of this kind were central to the reflections of the Mīmāṃsakas in their attempts to find the correct interpretation of the Veda: these attempts can only bear fruit if one gets as close as possible to the text and lets it speak for itself; obviously the primary meanings of words are closer to the text than derived, secondary meanings.
The units of meaning smaller than words are the minimal grammatical elements known from Pāṇini’s grammar. Discussions about their meanings took Pāṇini’s analysis as point of departure, but could deviate from it with respect to the precise meanings of those elements. Discussions about these meanings attracted a lot of attention in the more recent literature of the schools concerned (Mīmāṃsā, Nyāya, grammar), and we can only consider one example: the meaning of the verbal ending. In a simple sentence like caitraḥ odanam pacati (Caitra cooks rice), the verb is pacati (cooks), the verbal ending -ti (which corresponds to English “-s”). The grammarians, following Pāṇini, assigned the meaning “agent” to this ending. The Naiyāyikas, more inclined to analyze the sentence in terms of a subject (in this case the individual called Caitra) qualified by various features, assigned the meaning “activity” (kṛti) or “effort” ( yatna) to the verbal ending, so that they could paraphrase the sentence along the following lines: “Caitra is characterized by the activity, or effort, of cooking rice.” The Mīmāṃsakas, finally, wanted to push their bhāvanā (bringing into being) to the fore and assigned this meaning to the verbal ending. The inevitable result was that their analysis of the above sentence took something like the following shape: “The bringing into being whose agent is Caitra and which leads to the softening of rice.” Even this simple (and simplified) example will make clear that the discussion of the meaning of the individual grammatical elements cannot be separated from considerations as to the meaning of the sentence as a whole.
Language, Philosophy, and Science
The preceding pages have shown that thought about language exerted a profound influence on Indian philosophy in most of its manifestations. This observation should not too easily be reduced to the statement that in India grammatical thought deeply influenced philosophical thought. This last statement is no doubt true to at least some extent, but risks at the same time to put the cart before the horse. Sanskrit grammatical thought was itself the outcome of ideas and presupposition about language (and about Sanskrit in particular). It cannot be denied that those who wrote treatises about philosophy and science in classical India had received a thorough training in Sanskrit grammar, either Pāṇini’s grammar itself, or any of the later grammars that had been inspired by it. It is equally obvious that many of the discussions about the meanings of sentences, words, and their constituents, which I reviewed above, referred quite explicitly to Pāṇini’s grammar. The importance of this influence can yet be exaggerated.This is what may have happened in the case of the claim (Ingalls, 1954; Staal, 1965/1988), according to which Pāṇini’s grammar provided methodical guidelines to science and philosophy in India, the way Euclid did in the history of Western science and philosophy. There can be no doubt about the appeal of this claim, but little has been done to test it against the evidence. An investigation into the possible influence of Pāṇini’s grammar on Indian geometry has not yielded conclusive results. It seems therefore safer, for the time being, not to exaggerate the effect that Pāṇini’s grammar has had on philosophical and scientific thought in India. The effect exerted by notions about language in general, however, is clear and wide, as the preceding pages have indicated (Bronkhorst, 2001).
Comments
Post a Comment