Articles by alphabetic order
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
 Ā Ī Ñ Ś Ū Ö Ō
1 2 3 4 5 6 7 8 9 0


Nāgārjuna and the philosophy of language

From Tibetan Buddhist Encyclopedia
Jump to navigation Jump to search






Jan Westerhoff


The purpose of this paper is to examine some of the key points of N?g?rjuna’s discussion of problems relating to the philosophy of language. We will focus on two works from N?g?rjuna’s yukti-corpus that address these matters most explicitly, the Vigrahavy?vartan? (VV) and the Vaidalyaprakara?a (VP). The discussion will concentrate on four topics: N?g?rjuna’s views on semantics, the problem of empty names, the relation between language and momentariness, and the implications of Madhyamaka views on parts and wholes for the existence of language.1

1. The rejection of the ‘standard picture’


A very important part of N?g?rjuna’s application of his theory of universal emptiness is the rejection of what might be termed the ‘standard picture’ of semantics. It can be explained most clearly by examining the discussion around verse 29 of the VV.2 This verse states that If I had any thesis, that fault would apply to me. But I do not have any thesis, so there is indeed no fault for me. 3 This is certainly the most famous and also one of the most puzzling verses in the whole of N?g?rjuna’s VV. It is quoted frequently in the Buddhist commentarial literature, though the uses to which it is put are very diverse. I order to begin to understand how the verse fits in with N?g?rjuna’s thought in the VV more generally we need to consider his autocommentary.


If I had any thesis, the earlier fault you mentioned would apply to me, because the mark of my thesis has been affected. But I do not have any thesis. To that extent, while all things are empty, completely pacified, and by nature free from substance (svabh?va), from where could a thesis come? From where could something affecting the character of my thesis come? In this context your statement “there is precisely that fault for you, because the mark of your thesis has been affected” is not tenable.4

1 Our treatment does not aim to be comprehensive. There is much material relevant to the philosophy of language in N?g?rjuna’s other works that we cannot discuss here, though the points we raise in this paper are certainly central to Madhyamaka’s understanding of language. 2 Much of the discussion in this and the following section is based on my examination of the VV in Westerhoff 2010.


yadi k?cana pratijñ? tatra sy?n na me tat e?a me bhaved do?a? | n?sti ca mama pratijñ? tasm?n naiv?sti me do?a?. All quotations from the VV


are based on Yonezawa 2008.

4 yadi ca k?cin mama pratijñ? sy?t tato mama pratijñ?lak?a?apr?ptatv?t sa p?rvako do?a? | yath? tvayokta? bh?v?? tath? mama sy?n na ca mama k?cid asti


It is obvious that N?g?rjuna does here not make the false claim that he asserts no thesis whatsoever. After all, there are the M?lamadhyamakak?rik?, the VV ,and so forth, all of which are filled with philosophical theses and thereby contradict this way of understanding N?g?rjuna’s verse. What he wants to say here is that he does not have any thesis of a particular kind, that is, that among the theses one should assert there is none which exists substantially, none which is to be interpreted according to a fairly intuitive version of realist semantics.

What I mean here by ‘realist semantics’, and which is a theory we might reasonably ascribe to N?garjuna’s Naiy?yika opponent in the VV is one that assumes that there are mind-independent individuals and properties in the world, that they combine in different ways, and that it is their combinations that make sentences true. The M?dhyamika obviously has problems with such a notion of a ready-made world which does not depend on anything for its existence. And if his theory of universal emptiness makes him reject such a world, he will also have to adopt a different account of what makes sentences true. While the nature of a Madhyamaka-compatible semantics is not addressed very explicitly in N?g?rjuna’s writings, it seems plausible to suppose that a “worldless” theory based entirely on conventions between speakers might be what he had in mind.

In the context of the present discussion, it is evident how someone defending a semantics based on correspondence with a mind-independent world could criticize a convention-based semantics as producing only powerless statements, as N?g?rjuna’s opponent does. As the M?dhyamika has to regard his thesis of universal emptiness as empty too, he would have to supply it with such a convention-based semantics.

But then, the opponent will object, his thesis will never get out of the domain of conventions and connect up with the real world, in the way the statements of the opponent, which are to be supplied with a realist semantics, do. For the opponent, N?g?rjuna’s thesis compares to his philosophical statements like money made playing Monopoly to money made in the real estate market: the first is perfectly functional in the context of the game but, unlike the second, cannot be used to buy anything outside the game. Monopoly money can only buy Monopoly houses, but real money can buy not just real houses, but all sorts of other things as well. If there were any statements which had to be supplied with a realist semantics, while others are only given a conventionalist semantics, then it is true that the “earlier fault” would apply. The second kind of statement would be powerless to refute the first kind.


pratijñ? | tasm?t sarvahh?ve?u ??nye?v atyantopa??nte?u prak?tivivikte?u kuta? pratijñ?lak?a?apr?pti? kuta? pratijñ?lak?a?at?pr?ptik?to do?a? | tatra yat bhavatokta? pratijñ?lak?a?apr?ptatv?t tavaiva do?a iti tan n?sti


Consider the following example. Some people believe that there are rules which are grounded in something distinct from a set of human conventions. They might think that some rules (the substantial rules) are reflections of norms existing out there in the world. For example, one might assume that our penal code contains a prohibition against murder because it is a fact about the moral world that murder is wrong. Even realists about moral norms, however, do not think that all rules are grounded like that.

There is no fact out there determining which side of the road is the right one for driving on, and it is not the case that either the English or the French have a defective understanding of the Rules of the Road instantiated in some abstract realm. We are dealing here with a purely conventional rule settled by a system of human agreements. Now it is evident that in a world in which there are substantial and conventional rules, the conventional rules can never overrule the substantial ones. When there is a conflict between the two, it is always the conventional rule which has to give way—because it is only grounded in human practices, but not in the way the world is.


The same happens when statements come in two flavors: one, those put forward by the opponent, deriving their meaning and truth-value from the world out there; the others, N?g?rjuna’s, deriving these from their location in a system of conventions. In this case, a convention-based statement would not be able to refute a substantial one.


However, when asserting that he has no thesis, N?g?rjuna rejects this twoflavor theory of statements. Once the nature of conventions employed has been seen through and is thereby “completely pacified,” it becomes evident that there are no statements which have to be supplied with a realist semantics. As all statements N?g?rjuna asserts (and therefore all statements the philosopher in general should assert) are to be supplied with the same convention-based semantics, the charge of powerlessness does not apply.

The mark of N?g?rjuna’s thesis, namely its emptiness, which requires it to be spelled out in terms of a convention-based semantics, does not render it impotent because of the claim that some theses have to be supplied with another kind of semantics, since the very existence of such a realist semantics is ruled out by the Madhyamaka theory of emptiness.


A similar strategy suggests itself to the global relativist in reply to the dilemma that global relativism either does not apply to itself, in which case it is not global, and hence false, or it does apply to itself, and then it undermines itself as a theory of the world. It is obviously unsatisfactory to embrace the first horn of this dilemma, arguing that the thesis of global relativism holds absolutely, in the same way in which it would have not been very convincing for the Madhyamaka to assert that all things are empty, apart from those which are not, such as the thesis of universal emptiness.

The objective is then to argue that embracing the second horn, that is, claiming that the thesis of global relativism itself is only relatively true, does not undermine its philosophical force. For it is of course true that if there are truths made true by the world, then truths made true only by a theory appear insubstantial in comparison. But the global relativist, like N?g?rjuna, will want to deny that such a two-flavor picture obtains, and has to offer arguments to this effect. If this is successful, that is, if he has shown that all truths are relative, then stating that a particular one, such as that of global relativism, is merely relative will not detract from its status. The question then becomes why this relative truth, rather than another one, should be adopted. This question, however, is one we have to address for any statement we want to adopt in the relativist framework. It does not pose a specific problem for the thesis of global relativism.


2. Empty names

Another issue in the philosophy of language that has particular relevance in the context of Madhymaka is the problem of empty names. The opponent raises this in verse 9 of the VV:

And if there was no substance (svabh?va), there would also not even be the nameinsubstantiality of things,” for there is no name without a referent. And if the substance of all things were not to exist, there would be the absence of substance. There would also not be the name “absence of substance.” Why? For there is no name whatsoever without a referent. Therefore, because the name exists, there is the substance of things, and because substance exists all things are non-empty. Therefore the statement “all things are without substance, because of being without substance they are empty” is not tenable.5


To understand this criticism, we have to take into account the Ny?ya theory of language the opponent has in mind here. The ontological theory this is?based on comes from the Vai?e?ika school and conceives of properties as separate entities over and above the individuals that instantiate them. We therefore regard an apple and the redness it instantiates as two distinct kinds of entities, neither of which is any more fundamental than the other. This ontological picture is combined with a realist semantics of the kind described above, a theory according to which there is a clear correspondence between the categories existing in the world, such as individuals, properties, and absences, and the expressions of language. The opponent refers to this when he claims that “there is no name without a referent.” This position should not be misunderstood as saying that any name we come up with will


5 yadi ca na bhavet svabh?vo dharm???? ni?svabh?va ity eva? | n?m?pi bhavet naiva? n?m?pi nirvastukan n?sti || yadi ca dharm???? svabh?vo bh?v?n?? svabh?v?n?? sadbh?v?c c???nya? sarvabh?v?? | tatra yad ukta? ni?svabh?v?? sarvabh?v? ni?svabh?vatv?t ??ny? iti tan na


be guaranteed to have a referent. The Naiy?yika is not obliged to believe in yetis just because our language contains the term “yeti.” The point is, rather, that when we continue to unpack such a term along the lines of “large apelike creature supposedly inhabiting the Himalayas,” we eventually end up with a description using only terms that have a referent. The Naiy?yika realism does not demand that each term has a referent, but merely that all of the simplest terms connect directly with the categories out there in the world.


The opponent argues that N?g?rjuna’s use of the term “absence of substance (svabh?va)” entails the existence of substance. Assuming that “substance” is one of the simplest terms in our language, it is guaranteed to refer to something out there in the world. When we speak about its absence, we can only mean its local absence, in the way in which blueness is absent from the red apple but present elsewhere. N?g?rjuna’s thesis of universal emptiness is therefore only true in a restricted sense, claiming that certain things do not have substances but that others do, or meaningless, if the term “substance” it employs does not refer to anything in the world. He replies to the opponent as follows in VV 57:


Where someone said “a name has a referent,” one would say “then substance exists.” You have to reply “we do not assert a name of this kind.” Where someone said “a name has a referent,” one would say “then substance exists.” You have to reply: “If there is a referring name of a substance it has to exist due to the substantial referent. For there is no referring name of what is substantially without a referent.” However, we do not assert a referring name. This is because a name, too, due to the absence of substance in things, is insubstantial and therefore empty. Because of its emptiness it is non-referring. In this context, your statement “because of being endowed with a name there is a substantial referent” is not tenable.6 N?g?rjuna here replies by rejecting the semantic theory the opponent presupposes. The opponent’s move from the use of the term “substance” to the existence of what the term refers to only works against the background of the specific theory of meaning assumed. For the opponent, if there is a referring name of a substance, this substance has to exist because the substantial referent exists. Something which is substantially without a referent, as for example the concept “horn of a hare” or “son of a barren

6 ya? satbh?ta? n?ma br?y?t sasvabh?va ity evam | bhavat? prativaktavyo n?ma br?ma? ca na vaya? sata || yo n?ma satbh?ta? br?y?t sasvabh?va iti sa bhavat? prativaktavya? sy?t | yasya satbh?ta? n?ma svabh?vasya | tasm?t ten?pi svabh?vena satbh?tena bhavitavya? | na hy asatbh?tasya svabh?vasya satbh?ta? n?ma bhavat?ti | na punar vaya? n?ma satbh?ta? br?mas tad api hi svabh?vasy?bh?v?n n?ma ni?svabh?vatv?t ??nya? ??nyatv?d asatbh?tam | tatra yad bhavatokta? n?masadbh?v?t satbh?ta? svahh?v? iti | tan na


woman” cannot be picked out by a referring name. If “substance” was an empty concept, we could not speak about it on the basis of the Ny?ya theory of language.


However, N?g?rjuna does not see any reason for coupling his theory of emptiness with a Ny?ya-style realist semantics. He does not have to assume that simple terms are guaranteed to have a referent in the way the Naiy?yika says they do, or indeed that names refer at all. By adopting a different theory of language, the problem the opponent described above in verse 9 of the VV can be circumvented. N?g?rjuna continues his response in VV 58: The namenon-existent”— what is this, something existent or again something non-existent? For if it is existent or if it is non-existent, either way your position is deficient.


And this namenon-existent,” is that existent or non-existent? For if it is the name of an existent, or if it is the name of a non-existent, in both cases the thesis is deficient. If in this context the namenon-existent” is the name of an existent, to that extent the thesis is abandoned. For it is not the case that something is now nonexistent, now existent. Moreover, if the namenon-existent’ is the name of a nonexistent—there is no name of something which does not exist. To that extent, the thesis “the name exists substantially, so there is a substantial referent of the name,” is deficient.7


This verse describes a dilemma for the type of realist semantics defended by the opponent. Consider the referent of the term “non-existent.” If it is an existent referent, then the statement we want to make encounters a problem, since we are now dealing with some object which ex hypothesi does exist and at the same time does not exist, since it is the worldly counterpart of the term “non-existent.” On the other hand, its referent cannot be non-existent, since the realist does not allow for (simple) terms in the language that have no referent. Whichever horn of the dilemma we adopt, it seems impossible to make sense of the meaning of the term “non-existent.” But since its meaning is perfectly clear to us, this indicates that there is something wrong with the realist semantics presupposed.


What a contemporary semantic realist would want to reply here is that this problem can be solved if we regard “non-existent” as a second-order predicate. The semantics for a sentence like “yetis are non-existent” is then to be spelled out in terms of the second-order property of being uninstantiated applying to the first-order property of being a yeti. We can also seize the first

7 n?m?sad iti ca yad ida? tat ki? nu sato bhavaty ut?sata? | yadi hi sato yady asato dvidh?pi te h?yate v?da? || yac caitan n?m?sad iti tat ki? sata? asata? | yadi hi satas tan n?ma yady asata ubhayath?pi pratijñ? h?yate | tatra yadi t?vat sata | asad iti pratijñ? h?yate | na h?dan?? tad asad id?n?? sat ath?sata | asad iti n?ma y? pratijñ? asadbh?tasya n?ma na bhavati astitvasvabh?va iti n?ma tasm?t sadbh?ta? svabh?va iti | s? h?n?


horn of the dilemma in asserting that the second-order property of being uninstantiated is itself instantiated (since there is such a property). It is only applicable to first-order properties, but does not apply to itself; in fact, the theory of types which forms part of such a semantics usually excludes that properties can apply to themselves.


It is not entirely clear whether the realist will be happy with this, however. After all, this solution comes at the price of splitting up the property “is nonexistent” into infinitely many properties “is uninstantiated,” each of which only applies to objects at certain level in the type-theoretic hierarchy (firstorder properties, second-order properties, and so forth). If we want to hold that there is just one such property rather than infinitely many resembling ones, this solution will not be attractive to us. The discussion continues in VV 59 by N?g?rjuna once again emphasizing the point made in VV 57. The emptiness of all things was presented earlier. To this extent, this is a criticism of a non-thesis.


The emptiness of all things was presented here in detail by our earlier remarks. The emptiness of the name has been asserted above, as well. Having adopted the nonemptiness of things, you replied to this “if the substance of things did not exist there would be no name ‘non-substance.’” So far your criticism amounts to a criticism of a non-thesis, because we do not say that there is a referring name.8 As N?g?rjuna asserts a thesis of universal emptiness, it should be clear that the constituents of language are subsumed under this as well. It would indeed be problematic to combine N?g?rjuna’s theory with a Ny?ya-style realist semantics. But as N?g?rjuna does not want to do this, the difficulty described in verse 9 does not present a problem for him.


3. Language and momentariness


In his VP N?g?rjuna raises the interesting point of the consequences of the theory of momentariness (a theory all Buddhists accept in some form or other) for semantics. Prima facie the theory of momentariness appears to present a problem for the existence of language, as N?g?rjuna points out in VP 48, considering the example of the five-membered inference. Because they are asserted one-by-one, they do not exist one with one another.

8 sarve??? bh?v?n?? ??nyatvañ copap?dita? p?rvam | sa up?lambha? tasm?d bhavaty ayañ c?pratijñ?y?? || iha c?sm?bhi? p?rvam eva sarve??? bh?v?na? vistarata? ??nyatvam upap?ditam | tatra pr?g n?mano 'pi ??nyatvam ukta? sa?bhav? a??nyatva? parig?hya prav?tto vaktu? | yadi bh?v?n?? sv?bh?vo na sy?d asvabh?va iti n?m?p?dan na sy?d iti | tasm?d apratijñop?lambho ya? bhavata? sa?padyate | na hi vaya? n?ma sadbh?tam iti br?ma?


At the time when the thesis is asserted, the reason etc do not exist and if the reason is asserted also the thesis etc do not exist. Therefore the reason etc do not exist.9 The argument appears to be that because the elements of a semantic complex (in this case, an inference) are momentary, the complex, which depends on the existence of all the elements together cannot exist.10 The elements are produced sequentially, and because each ceases immediately after its production, when one element is produced, its predecessor has already ceased to exist. No two members of the sequence can co-exist, and for this reason the entire inference, which depends on all members existing jointly does not exist.

But since each element only acquires its role via its place in the inference ("because of smoke" is not a reason in itself, but only in conjunction with a specific thesis, a specific example and so forth) the non-existence of the whole, the inference also entails the non-existence of its constituents. This is why N?g?rjuna can conclude that "the reason etc do not exist.". As an example, consider as an example trying to build an arch out of blocks of ice produced in a freezer. This requires that we put all the blocks together in a particular structure.

Some blocks are not yet frozen and therefore not ready to use, others have already melted away because we have taken them out so early. The only block we can really work with is the one we have taken out of the freezer right now, but this on its own is unable to constitute an arch. Alternatively, image we write a proof one a slab of stone with a brush dipped in water. By the time we reach the middle of the proof the inscription of the

9 gcig dang gcig brjod ba’i phyir gcig la gcig med do | gang gi tshe dam bca’ ba brjod pa de’i tshe gtan tshigs la sogs pa med la gtan tshigs brjod pa na yang dam bcala sogs pa med do | de’i phyir gtan tshigs la sogs pa dag yod pa ma yin no. All quotations from the VP are based on the edition in Tola/Dragonetti 1995. Compare also VP

10 We also find this argument in the Tibetan version (the relevant passage is missing in the Chinese version) of the Ak?ara?ataka attributed to N?g?rjuna in the Tibetan tradition and to ?ryadeva in the Chinese tradition: "When there is the thesis the reason does not exist because it has not been produced. Similarly, when there is the reason the thesis does not exist, because it is destroyed. Therefore, when the thesis is non-existent, of what is it a reason?"

(dam bca' ba'i dus su ni gtan tshigs yod pa ma yin te ma skyes pa'i phyir ro | de bzhin du gtan tshigs kyi dus su yang dam bca' ba yod pa ma yin te 'gag pa'i chos yod pa'i phyir ro | de'i phyir dam bca' ba'i dngos po med pa la 'di gang gi gtan tshigs su gyur (Peking 5235 160a).

See Gokhale 1930:20, note 31; Pind 2001:166.) See also the related discussion in ??ntarak?ita's Tattvasa?graha 872-877 (??str? 2006:240-242; Jha 1986: 508-513). Compare Jha's interpretation of V?tsy?yana's criticism of the n?stika in commenting on NS 1.1.1


(n?stika? ca d????ntam abhyupagacchann ?stikatva? jah?ti anabhyupagacchan ki?s?dhana? param up?labheteti (??str?


1998:8:4-5)): "There is yet another reason why importance has been attached to Example; it is through this that the position of the atheistic Bau??ha becomes doubly untenable. If the Atheist admits a corroborative example, he renounces his atheistic (Nihilistic) position (as by Nihilism, all things have merely momentary existence; and hence it is not possible for the Example, which must be in the form of something that existed in the past, to be present at the time that it is put forward); if, on the other hand, he does not admit an example, on the basis of what could he attach the position of his opponent?" (1984:1:46-47). first premisses will have evaporated already, so that we are never able to get the whole proof down on the slab at one time.

Of course the Buddhist has to explain that we are still somehow able to make and understand inferences. To deny this would not be in accord with conventional reality. One way of solving this problem could consist in appealing to memory, roughly along the following lines. At each moment, we assume, only one of the five members of a syllogism can exist. Yet in the present moment when e.g. member 3, the pervasion, is perceived by the mind there can be the memory of member 2 being perceived just before that, and of a memory of a memory of member 1.


The structure of this would look as follows:

t1: 1 t2: 2, mem(1) t3: 3, mem(2), mem(mem(1)) t4: 4, mem(3), mem(mem(2)), mem(mem(mem(1))) t5: 5, mem (4), mem(mem(3)), mem(mem(mem(2))), mem(mem(mem(mem(1))))


Even though the utterance of the thesis does not exist at time t2 anymore, there is still a substitute, the memory of the thesis coexisting with the reason at time t2. In this way substitutes for all the members of the inference can coexist at one point in time and the inference can be understood, even though the whole that is the inference never exists at any point in time. It is of course essential for this understanding that e.g. mem(1) is not identical with 1, since otherwise t5 would just be the cooexistence of the five members of the syllogism that N?g?rjuna rejects. This identity could only be supposed if we considered mem(1) and 1 be objects of the same (abstract) type.

Yet as N?g?rjuna’s discussion makes clear in several instances, the discussion of syllables, words, parts of an inference and so on is conducted in terms of concrete, spatio-temporal tokens of such syllables etc, not in terms of types. It is interesting to note, however, that a very similar argument would be available to N?g?rjuna even if the parts of the an inference and similar objects were understood as types, i.e. as abstract objects. This is because the doctrine of momentariness does not just apply to perceived objects, but also to the perceiving subject.

For N?g?rjuna there is no permanent self.11 But if the self is momentary too, there is no single self that could hold all the five members of an inference simultaneously in front of its mind's eye, for as soon as the reason is understood, the mind-moment that understood the thesis has already ceased to exist. And if there is nothing at the ultimate level that exists throughout the inferential steps there is also nothing at the ultimate level that could draw the required inference. As long as an inference is considered as something that establishes a conclusion in the mind of a particular person we can then conclude that no inference exists at the ultimate level.


4. Language and mereology


The Buddhist critique of the existence of the whole (sometimes referred to as “mereological reductionism”) is well known. According to its Abhidharmic formulation is asserts that the whole only exists as a superimposition on the parts. At the level of ultimate reality, there are no wholes. In VP 33 N?g?rjuna makes the interesting (and initially somewhat puzzling) assertion that there are also no parts:


Because there is no whole, there are no parts.


In this respect, because there is no whole for the parts of the thesis, reason, example, application, and the conclusion, there are no parts.12 This statement comes as a response to the opponent pointing out that without constituents or parts of an inference we cannot have a proof (such as the fivemembered syllogism) since proofs are essentially partite. But in this case how can N?g?rjuna express any logically established conclusions whatsoever? More importantly, unless there is something specific about parts of arguments, N?g?rjuna would presumably reject other, suitably similar parts, such as the parts of speech that constitute a sentence. In this case it is difficult to see how N?g?rjuna manages to say anything at all, given that meaningful speech consists of words connected up in the right way.

But of course if there are no items to be connected, no parts, there can be no connections, and therefore no meaning. It is then difficult not to see this position as selfrefuting. In order to understand what N?g?rjuna had to say up to this point in the argument we had to presuppose that his sentences were meaningful, and therefore composed of parts. If any of the preceding refutations of the Ny?ya categories has been successful it can only be so by having been presented as a piece of meaningful language. So if N?g?rjuna does not want to abandon these refutations it is not really obvious how he can hold on to his rejection of parts.

The M?dhyamika is caught in a dilemma: Either he denies the existence of parts, in which case there can be no lingustically expressed proofs, so that his arguments for universal emptiness cannot be expressed anymore, or, presupposing that this arguments do work, he will have to accept the existence of partite entities by implication and will thereby contradict his own position. VP 33 confirms that N?g?rjuna indeed wants to rejects the constituents of arguments, based on a general mereological point about the rejection of the whole and, consequently, the rejection of the parts. He specifically notes that the parts of the five-membered syllogism, the canonical form of inference in the Indian theory of reasoning, fail to form a whole. There are two questions we have to answer here in order to understand N?g?rjuna's point:

1. Why does he deny the existence of wholes?

12 yan lag can med pa’i phyir yan lag yod pa ma yin no | ‘dir dam bca’ ba dang gtan tshigs dang dpe dang nye bar sbyor ba dang mjug bsdu ba’i yan lag rnams la yan lag can med pa’i phyir yan lag rnam pa med pa yin no and

2. Why does the non-existence of wholes entail the non-existence of parts? In line with the very condensed nature of VP and its autocommentary N?g?rjuna does not give any argument for the rejection of wholes at this point. This is not very surprising, given that it is reasonable to expect that the justification for the rejection of wholes is known to an audience familiar with core Buddhist teachings.

Already in the Samyuttanik?ya we find teachings of the Buddha which are later synthezised in the famous simile of the chariot in the Milindapañha. There the monk N?gasena explains to king Menandros that the existence of the chariot is a merely nominal one, that of an entity superimposed on a collection of individual parts: the axle, the wheels and so forth. This mereological reductionism is a familiar tenet of Abhidharma philosophy, and one which plays a crucial role in arguing for the doctrine of non-self on the basis of its being composed by the five psycho-physical aggregates (skandha).


Yet N?g?rjuna's assertion that the non-existence of the whole also entails the non-existence of the parts that compose the whole13 would have appeared very puzzling to an ?bhidharmika. Even if the complex whole exists only as a name, if we break it down we will eventually have to hit some things that exists in a more substantial way. For the ?bhidharmika these are the ultimately existent dharmas. So how to we adress the second question, concerning the entailment between the non-existence of wholes and the nonexistence of parts? There are a variety of ways in which N?g?rjuna can justify his conclusion. In their commentary on the VP Tola and Dragonetti present two possible ways of spelling out the argument.14 Firstly, if the whole is regarded as non-existent it would then be strange to assume that it has parts -- non-existent objects such as a city in the clouds or square circles don't have parts.

Secondly, the parts of the wholes can be regarded as wholes which are nonexistent in relation to their parts, and so on. Neither of these strikes me as wholly satisfactory. The first fails to take into account that the M?dhyamika does not want to put wholes in the same category as cloud-cities or square circles, i.e. things that are not even real at the level of conventional reality. Wholes are not ultimately real, but conventionally real, and thereby able to have parts. The second appears to beg the question against the atomist, who makes precisely the claim that by a continued process of division of material objects we will eventually reach some objects that do not stand in the relation of


wholes to their parts, since they are impartite. To claim that the whole-part relation will continue all the way down is just a denial of that assumption, not an argument against it. A better way of understanding N?g?rjuna's argument here is to take note of the fact that the M?dhyamika agrees with the ?bhidharmika in rejecting wholes as ultimately real because they only exist in dependence, their existence is derived from the existence of their parts. The M?dhyamaka points out that the same holds for the objects the whole depends on, the parts. Even if there are impartite entities all matter is composed of, they fail to be ultimately real since the existence of each is dependent on that of some other. It can plausibly be argued that the best candidates for the ultimately existent dharmas of the ?bhidharmika are not material atoms but particularized property instances (sometimes called tropes or, more memorably, pins), such as the redness of this postbox or the shape of this apple.

These pins then cluster together to form the familiar objects of our acquaintance. The important point to note here, though, is that the identity of each particular pin is not due to some intrinsic property had by it (otherwise it would in turn be a bearer of a properties, and hence not simple). Rather, it depends on what other pins it co-occurs with in a cluster, so that this red-pin is distinct from that because the former occurs in a cluster with a square-pin, while the latter occurs in a cluster with a triangular-pin.15

But if these pins are in this way dependent on other pins for their identity they cannot be ultimately real. So assuming that the opponent defends an ontology of pins N?g?rjuna is justified in saying that the wholes are not ultimately real (because they depend for their existence on their parts), and the parts are not ultimately real either (because they depend for their identity on other parts). The originality of N?g?rjuna's argument is worth noting. While part-whole considerations are usually applied to composite material objects (such as chariots) or to physio-psychological complexes (such as persons) the argument is here used to analyzed structures, in particular logical inferences and pieces of language. Such an analysis is certainly in accordance with the intention behind the more familiar versions of the part-whole arguments.

If there is any sense in which structures can be said to have parts, the same points that the Buddhist's analysis of material composites makes should apply to them. Yet it also raises the spectre of inconsistency, for now the very means by which one carries out the part-whole analysis, namely logical and linguistic complexes, are subjected to a part-whole analysis themselves. The opponent takes this issue up again in VP 49 when he remarks that "It is your thesis that 'all parts do not exist'. Therefore, because the thesis is asserted the remainder is also established for you.”16

15 See Siderits 2007:189-191. 16 khyod kyi yan lag thams cad med do zhes dam bca’ ba de lta bas na | dam bca’ ba khas blangs pa’i phyir khyod la lhag ma dag kyang grub pa yin no The opponent points out that N?g?rjuna aims to establish a philosophical position, that of the non-existence of parts, both generally and specifically in relation to the parts of the five-fold inference. As such he is aiming at establishing a thesis, and a thesis can only established by working through an inference that establishes the thesis as a conclusion.

But once N?g?rjuna has accepted the inference he has of course also accepted all the members of the inference, which makes his refutation of the parts of an inference inconsistent. Once more, the M?dhyamika seems to face a dilemma.17 He claims that since the M?dhyamika asserts that everything is empty, there must be at least one non-empty thing, namely that assertion.

If there is not, the assertion is empty too, and for that reason it cannot establish the Madhyamaka position. Since the only way to back up a thesis is by supplying the other four members of the syllogism, either N?g?rjuna contradicts himself, if he regards this thesis as established, because there must be one instance of a syllogism of which all the parts exist. Or, if his thesis is not established, he cannot say what he wants to say, at least not in a context in which truth-bearing assertions and inferential support play any role.

In response N?g?rjuna points out that the considerations that applied to the constituents of the syllogism (the "non-established parts") also apply to their subsentential components.

The syllables are also like this.

That is also seen concerning the syllables which are the constituents of the nonestablished parts. Just like because the syllables pra, ti, jñ? do not coexist, the thesis does in fact not exist.18

In the same way as all the constituents of an inference do not exist at one time, all the constituent syllables of any of these constituents do not exist simultaneously either, since when, for example, the second syllable of a word such as pratijñ? is uttered, its first syllable has disappeared and its third syllable does not yet exist. N?g?rjuna continues: Similarly also regarding the syllable pra, because pra etc. are uttered successively, the syllable pra does not exist.

The same then applies to the constituents of a single syllable. That "pra etc. are uttered successively" means that they, in turn have components, not all of which can exist at a single time. There is good evidence that this passage is to

be understood in this way. Consider the same argument being presented in the Tibetan version of the Ak?ara?ataka, immediately following the passage cited in the commentary to the previous verse:19 It is similar for the thesis as it is something that does not exist at a single time. When the syllable pra exists, the syllables ti and jñ? do not exist. Similarly the letters p, ra, a etc. do not arise simultaneously at one time either.20 N?g?rjuna finishes his reply by giving a list of the production-places of speech as understood in traditional Indian phonetics: When arising from conditions such as wind and space, the effort of tongue, teeth, palate, lips and so on they do not exist one with one another.21 His point is that since the individual syllables are produced by successive and distinct sets of actions of the various production-places they cannot all exist at the same time, but only one after the other. This of course implies that the entire syllable, word, or sentence can never be present at any one time. How is all of this a reply to the opponent who argues that because N?g?rjuna accepts a thesis, he also has to accept the existence of all the other members of the syllogism?

Tola and Dragonetti22 suggest that N?g?rjuna replies to the charge of inconsistency claiming that he does, after all, hold a thesis (pratijñ?) by pointing out that there is no pratijñ? since its syllables do not coexist. How could he therefore hold a thesis? We should note in addition, however, that this reply does not imply N?g?rjuna getting caught on the other horn of the dilemma, that of the impossibility of making any substantial reply in a debate. For N?g?rjuna will accept, his arguments in this verse notwithstanding, that we manage to communicate using syllables, and words and sentences built from these.

The doctrine of momentariness does not refute this truth about conventional reality, but stresses the contribution conceptual superimposition makes in order to create this reality. In order to understand the word pratijñ? on hearing the syllable ti we must contribute the syllable pra from memory and the syllable jñ? by anticipation. In the same way, even though considerations of momentariness apply to all semantics complexes, including inferences, we are able to communicate inferences from one person to another, even though no more than one part of an inference is ever present at one time. The opponent is thus mistaken to assume that N?g?rjuna could not

19 This passage is also included in the Chinese version, see Gokhale 1930:9. 20 de bzhin du dam bca' ba la cig car med pa'i chos can gyi phyir te | gang gi tshe yi ge pra yod pa de'i tshe yi ge ti dang yi ge jñ? yod pa ma yin no | de bzhin du dus gcig tu pa yig dang ra yig dang yi ge a la sogs pa rnams kyang cig car 'byung ba ma yin no (Peking 5235 160a-b). 21 rlung dang | nam mkha’ dang | lce dang | so dang | mgrin pa dang | rkan dang chu dang ‘bad pa la sogs pa’i rkyen la byung ba’i tshe gcig la gcig med pa yin no 22 1995: 132-133.

communicate a thesis, word, or syllable even at the level of transactional reality, and would therefore be condemned to either inconsistency or silence. From investigating these few passages of N?g?rjuna’s works it has become evident that the discussion of problems connected with the philosophy of language we find even in quite early Madhyamaka sources is complex and concern intricate issues at the very heart of Buddhist philosophy. Madhyamaka philosophy of language is a largely unexplored field that deserves far more investigation than has been undertaken hitherto. I hope that the discussion presented in the previous pages, cursory though it is, might present a first small step in this direction.


Bibliography


Vasudev Gokhale: Ak?ara-Çatakam. The Hundred Letters, A Madhyamaka Text by ?ryadeva, After Chinese and Tibetan Materials, Harrassowitz, Heidelberg, 1930.

Ganganatha Jha: The Tattvasasa?graha of Sh?ntarak?ita: With the Commentary of Kamalash?la, Motilal Banarsidass, Delhi, 1986.

Xuezhu Li: “Madhyamaka?vata?ra-k?rik? Chapter 6”, Journal of Indian Philosophy 4, 2015, 1-30.

Ole Holten Pind, "Why the Vaidalyaprakara?a cannot be an authentic work of N?g?rjuna", Wiener Zeitschrift für die Kunde Süd-und Ostasiens 25, 2001, 149-172.

Sw?m? Dw?rik?d?s ??str? (ed.): Mah?muni V?tsy?yan's Ny?ya-Bh??ya on Gautama-Ny?ya-Sutras with Prasannapad? Commentary of Pt. Sudar?an?c?rya, ??str?, Bauddha Bharati, Varanasi, 1998.

Sw?m? Dw?rik?d?s ??str? (ed.): The Tattvasa?graha of ?c?rya ??ntarak?ita with the 'Pañjik?' Commentary of ?c?rya Kamala??la, Bauddha Bharati, Varanasi, 2006.

Mark Siderits: Buddhism as Philosophy: An Introduction, Ashgate, Aldershot; Hackett, Indianapolis, IN, 2007.

Fernando Tola, Carmen Dragonetti: "N?g?rjuna's conception of 'voidness' (??nyat?)", Journal of Indian Philosophy 9, 1981: 273-282.

Fernando Tola, Carmen Dragonetti: "Dign?ga's ?lambanapar?k??v?tti", Journal of Indian Philosophy, 10, 1982, 105-134.

Fernando Tola, Carmen Dragonetti: N?garjuna’s Refutation of Logic (Ny?ya): Vaidalyaprakara?a, Motilal Banarsidass, Delhi, 1995.

Jan Westerhoff: The Dispeller of Disputes: N?g?rjuna’s Vigrahavy?vartan?, Oxford University Press, Oxford, 2010.

Yoshiasu Yonezawa: Vigrahavy?vartan?: Sanskrit transliteration and Tibetan translation, Journal of the Naritasan Institute for Buddhist Studies, 31, 2008,



Source


[1]