By Ewing Y. Chinn
Philosophy East and West
Volume 44, Number 1
January 1994
(C) by University of Hawaii Press

P.55 The Buddhist philosopher and logician Dignaaga (A.D. 480-540) and the eighteenth-century Irish idealist Berkeley may look like strange philosophical bedfellows. However, the two have at least this in common: both were persistent critics of a theory of language that affirmed the existence of abstract entities. Berkeley is well known in Western philosophy for his attack on John Locke's theory of abstract ideas; and the Dignaaga-Bhart.rhari dispute over the existence of universals is one of the most important and interesting episodes in the history of Eastern thought. But just how far does this common seam run? Both reject abstract entities, but we must ask whether there are any significant similarities in their respective ways of dealing with a central question in the philosophy of language. This is the question: how do we account for the fact that certain kinds of words, namely general names, refer specifically to particular spatiotemporal objects? What exactly is the link between language and the world?(1) It is well established that Berkeley defended an anti-abstractionist approach to the problem of reference, maintaining in opposition to Locke that words do not require the mediation of any abstract entities in order to refer to the real existents of the world (which in Berkeley's case were concrete, particular ideas). But there has been some confusion and no clear consensus about Dignaaga's views on this matter. My aim in this essay is to arrive at a better understanding and appreciation of Dignaaga's theory of reference by conducting an experiment, the experiment of reading Dignaaga through Berkeley. We will see that Berkeley and Dignaaga make very compatible philosophical bedfellows indeed, in the area of the philosophy of language. This is not to say that there are no significant differences in their views on language and meaning. On the contrary, I will argue that while Berkeley's account of the meaning of general words lacks any satisfactory explanation of the semantic justification of our actual use of such words (what I will call the 'word-world mechanism'), this is not the case with Dignaaga. Although Dignaaga did not explicitly provide an account of the 'word-world mechanism', the essence of his answer is contained in his controversial notion of apoha (usually translated as 'exclusion'). Before embarking on our experiment, we must set the stage with a brief look at the way in which scholars generally view the Dignaaga-Bhart.rhari controversy. It is interesting that both Bhart.rhari and Dignaaga claimed as a point of departure for their treatment of the problem of reference a P.56 famous aphorism of Kaatyaayana, a grammarian of the second or third century B.C. [The occasioning basis for the use of a name is] that quality because of whose presence (bhaava), a name is applied to a thing. [The addition to the nominal base of the suffixes] tva or ta/[is taught], in the signification of this quality.(2) We can see that there is something puzzling about the statement that the quality of an object that is the basis for giving or applying a certain name to that object--for example, calling something `sukla (the white) or paacaka (the cook) --is signified and conveyed by abstract terms that result from adding the abstract suffixes tva or tal to those names. Examples of such terms are `suklatvam (whiteness) and paacakatvam ('cookness' or the essence of being a cook) , typical terms for universals.(3) Is Kaatyaayana conflating universals with concrete qualities of spatiotemporal objects? It is entirely unclear what Kaatyaayana took to be the 'occasioning basis for the use of a name', or as Radhika Herzberger puts it more precisely: What [exactly] is the basis on which names are given to things? Is the basis on which names are given a quality (guna), or a universal (jaati)? Is this basis located in words or is it located in the things which are named? Do names 'exceed over' (ati + ric or ati + vrt) their bearers?(4) Bhart.rhari's answer was that a general name cannot be directly given to a particular set of objects on the basis of some feature of those objects. It must refer to those objects indirectly on the basis of an abstract universal 'located in the word' and thus conveyed by the word (the `sabdajaati or 'word-universal'). We shall call this view the indirect theory of reference, the view that the meaning of a name (that is, the universal immediately signified by the name) determines its reference(s). Semantically speaking, a name has a referential relationship to spatiotemporal objects only because the universal it signifies has a referential relationship to spatiotemporal objects. For Bha.rtrhari, names do exceed over their bearers. Dignaaga, on the contrary, claimed that names do not exceed over their bearers, for, he said, a name directly applies and is non-distinct [from its object]. A name does not apply to its own object, having first presented it with another object.(5) and Even though a word has multifarious properties, it causes the object to be conveyed by means of that [quality] alone which does not exceed over its object; not by means of qualities etc. which belongs to words.(6) P.57 To say that the basis for reference is a guna and not a jaati would seem to entail that Dignaaga held a direct theory of reference, the view that a name denotes the appropriate spatiotemporal objects without the mediation of some abstract universal that is supposed to constitute the 'meaning' of the name. He apparently made no distinction between the meaning of the name and the references of the name, agreeing with Kaatyaayana that a name does not exceed over its bearers. Nevertheless Dignaaga has been represented by a tradition of commentators--beginning with Dharmakiirti--as endorsing the indirect theory of reference, the theory that a name refers to an object through the 'meaning' of the name. Where he differs from Bhart.rhari, according to the received view, is his conception of the meaning of general names, his advocacy of the much-discussed and maligned theory of apoha or `exclusion'. It seems that in order to resist reifying the meaning of names, to avoid any commitment to the positive existence of universals, Dignaaga is said to claim, for example, that 'this is a cow', is not to affirm the cowhood of the subject, that it exemplifies cowness, for the sole function of the word 'cow' is to differentiate the directly perceived object from other things, from non-cows. As a contemporary representative of the received view, Masaaki Hattori, puts it: [A] word indicates an object merely through the exclusion of other objects. For example, the word 'cow' simply means that the object is not a non-cow. As such, a word cannot denote anything real, whether it be an individual (vyakti), a universal (jaati) or any other thing. The apprehension of an object by means of the exclusion of other objects is nothing but an inference.(7) Well and good (perhaps), but how are we to understand the notion of a 'non-cow'? If the claim is that the perceived object is not a goat, and so forth, then we must have some understanding of what it is to be a goat and all those other things that the object is not. If the response reverts back to 'cow', we are in a perpetual circle of nonunderstanding, and can never explain why we use that name to refer to this particular object of our perception. This peculiar theory of 'negative (or apoha) universals' must seem to some critics like a desperate ploy to preserve a kind of phenomenalistic ontology. If Dignaaga actually held that the only thing real is the flux of momentary point instances (for example, sense impressions), he would naturally regard any attempt to describe this phenomenon in words as a distortion of the real. Thus we have the motivation for his attempt to minimize or even nullify the effects of this distortion by arguing that concepts are imaginative constructs and negative in content. The problem is that such a theory cannot account for reference. Furthermore, the phenomenalist, 'Humean' view of Dignaaga is, to say the least, highly controversial. P.58 David Kalupahana, for one, has argued eloquently and convincingly in several studies that Dignaaga is in the line of Buddhist philosophers faithful to the unique empiricism of the early discourses of the Buddha, an empiricism "that steers clear of the extremes of absolutely real and pure experiences on the one hand, and absolutely unreal mental constructions on the other hand."(8) Kalupahana thus places Dignaaga squarely in the Buddhist tradition that opposes absolutisms of any form, the middle way of the Buddha, Naagaarjuna, and Vasubandhu. We will see that the middle standpoint in Buddhism has the effect of taking away the (erroneous) terms of the realism-nominalism dispute concerning universals. So if one were to ask Dignaaga, "Do universals exist?" he might answer, "Do not say so!" And if he were then asked, "Is it the case that universals do not exist?" his answer would be, "Do not say so!" What this implies is that disavowing that names signify universals (of either the positive or negative variety) is not to be construed as embracing skepticism. Dignaaga, on the contrary, in refusing to be trapped in a philosophical myth, will offer us an explanation of the meaning and function of names that he believes is consistent with actual practice. I shall undertake to present an interpretation of this "Dignaga of the Middle Way" by utilizing Berkeley, who, as I said, is a fellow anti-abstractionist. But first, let us examine the indirect theories of reference of John Locke and Bhartrhari, and see what is problematic about such theories. II Recall the two questions raised by Kaatyaayaana's aphorism: (1) Is the basis on which names are given a quality of the object or a universal? (2) Do names exceed over their bearers? We find Bhart.rhari's answer to the first question in his rendition of Kaatyaayana's aphorism, where the phrase 'presence of a quality' (gu.nasya bhaava) is changed to 'essence of a quality (gu.nasya tattva). [The occasioning ground for the use of a name is] that quality because of whose essence (tattva) a name is applied to a thing.(6) (My emphasis) What Bhart.rhari is saying is that although our using a particular name to refer to an object is justified by a quality that is possessed by the object, it is the meaning of the name (that is, the universal signified by the name) that directs our attention to the property and thus to the possessor of that quality. It is in this manner that meaning determines reference, that the intension of a general name determines the extension of that name. Berkeley tells us that John Locke also held a version of the indirect view of naming: Tis thought that every name hath, or ought to have, one only precise and settled signification, which inclines men to think there are certain abstract P.59 determinate ideas, which constitute the true and only immediate signification of each general name. And that it is by the mediation of these abstract ideas, that a general name comes to signify any particular thing.(10) The only difference between Locke and Bhart.rhari with respect to universals, albeit a major difference, is that Bhartrhari claimed that universals are 'known a priori', whereas Locke, the empiricist, claimed that they are derived from experience, through a process that he calls abstraction.(11) The process sounds deceptively simple in the following passages. Words become general by being made the signs of general Ideas, and ideas become general by being separated from the circumstances of Time, and Place, and any other idea, that may determine them to this or that particular Existence. By this way of abstraction they are made capable of representing more Individuals than one; each of which, having in it a conformity to that abstract idea, is (as we call it) of that sort.(12) Abstraction is being depicted here by Locke as a process of isolation, isolating what is there in a particular experience, whereas Locke explains it elsewhere as a process of formation. It is unimportant to the purposes of this essay to determine what Locke really means or should mean by `abstraction', but it is important to determine what he means by a `general idea'. It is unclear whether what is being isolated (or formed) is a particular idea of a particular property or a general idea of an abstract universal. A 'general idea' in the former case is a particular mental image of a particular property put to general use. And the name that signifies that image applies to all those objects with a property that resembles the image. 'Conformity' means resemblance.(13) But there are two famous passages that support the view that Locke took an abstract idea to be the general idea of a universal: [T]hey make nothing new, but only leave out of the complex Idea they had of Peter and lames, Mary and Jane, that which is peculiar to each, and retain only what is common to them all.(14) Does it not require some pains and skill to form the general idea of a Triangle. (which is yet none of the most abstract, comprehensive, and difficult,) for it must be neither Oblique, nor Rectangle, neither Equilateral, Equicrural, nor Scalenon; but all and none of these at once. In effect, it is something imperfect, that cannot exist; an Idea wherein some parts of several different and Inconsistent ideas are put together....(15) What Locke says about the idea of a triangle in general applies equally to the idea of a man in general, for both are surely ideas of an abstract universal. This is why it must be the case that, as the passage says, (1) such an idea (of, for example, the universal man or triangle) cannot be identified with any particular idea of an instance of the corre- P.60 sponding universal, having no natural resemblance to any of them, but, at the same time, (2) all of its instances are in conformity with it. What we have is a classic statement of the relation between universals and particulars, two ontologically distinct kinds of objects. What about the second Kaatyaayana-inspired question: do names exceed over all their referents? It would seem that anyone who maintains that a name refer by virtue of the abstract universal it signifies must answer yes. But then we have a serious problem. If there is no resemblance (or any natural relationship for that matter) between an abstract universal associated with a name and the concrete bearers of that name--for this is why a name exceeds over its references--how do we know which objects are instances of a universal and which are not! The indirect theory maintains that an abstract universal is required to mediate between a name and the name bearers, but now it appears that something else is required to mediate between that abstract universal and the objects we seek to name, and we have the beginning of an infinite regress.(16) Following tradition (going back to the Greeks) , we shall call this objection to the indirect theory of reference the 'third-man' argument. Dignaaga must have been fully aware of this problem in Bhart.rhari's indirect theory of reference because he asked of that theory, "[How] [i]s the relation between a name and its spatial-temporal bearer teacheable?"(17) "The word which signifies a universal," he said, "cannot designate individuals, because of the infiniteness [of the individuals] and because of the deviation. Nor [can it designate] a relation [of the universal with the individual], nor a universal."(18) Bhart.rhari had an answer to this problem. He maintained (as the central thesis of his theory) that if meaning determines reference and if to understand a name is to know the bearers of that name, it must be that the content or meaning of the name coincides with the `complete content' of each individual bearer of that name. Bhart.rhari refers to the former, the content of a name, as 'word universals' (`sabdajaati) inhering in a name and to the latter as 'thing universals' (arthajaati) inhering in a bearer of that name. To say that they coincide is to say that there is just one set of universals (constituting the meaning of the name and the 'content' of each bearer of that name) looked at from two different points of view. It is obvious that, if universals exist, spatiotemporal individuals must be regarded as exemplifications or instantiations of universals. But Bhart.rhari seems to be saying more than that when he claims that universals constitute the complete content of an individual. This claim seems to imply that there is nothing more to the makeup of a spatiotemporal individual than a certain set of 'thing-universals'. An individual does not have a feature that we would call its 'particularity' or its 'individual essence', its haecceity. P.61 This conjecture is borne out by Bhart.rhari's treatment of the notion of an individual. "An individual," he says, is simply "that which is given as principal (pradhaana), as in 'this is that' ['that' being a name]."(19) This logical-linguistic definition of an individual as the subject of a proposition relativises that notion, for, as he says, "when a quality [that is, a universal] is being distinguished, it becomes principal," and "it is held that everything which is meant is an individual (dravya)."(20) Thus the only universal that cannot be conceived of as an individual is the highest universal standing at the apex of the hierarchy of universals, the all-encompassing Brahman, the being of all beings. It must be the case, then, that the spatiotemporal individual, the object of sensory experience, is conceived by Bhart.rhari as the fictional subject (a mental construct) of the sum total of universals that qualifies `it', the 'thing universals'. That same set of universals inheres in the name given to that fictional subject as 'word universals'. Thus, it is possible for Bhart.rhari to claim that to understand a name is to know the bearers of that name for, as Herzberger puts it, "the individual object 'qualified by all its qualifiers' is simultaneously given on a single utterance of a word."(21) But there is more to the story. In Bhart.rhari's theory, the object named is given or determined by the content of the name that Herzberger says is both analytic and antonymic. A name is defined, she says, by "the analytic content of a word [which is] represented by the set of expressions whose meaning is included in the word, and the antonymic content [which is] represented by the set of expressions whose meaning is excluded from the word."(22) So, for example, a rose is defined (in part) as a flower, vegetation, a substance that is not an orchid, pansy, weed, and so forth. The meaning or content of a name, in other words, is a cluster of particular universals related by the relationships of inclusion or inherence and exclusion (apoha). In fact, the full meaning of any name is the entire world of universals as conceived by Bhart.rhari. It turns out that no particular universal, with the exception of Brahman, has a being of its own, but is defined by the relationships of inclusion and exclusion to other particular universals, and ultimately by its inclusion in Brahman. In the last analysis, the crucial implication of Bhart.rhari's theory of universals is that the only meaning that a universal (and its name) has is Brahman.(23) All universals signify Brahman, but the only thing we can say about Brahman is what it is not; and this is equivalent to what Dignaaga said (surely about Bhart.rhari and not himself), that "in accordance with its own relations [to universals], the word conveys the object, through exclusion (apoha)." It would seem that Hattori had the wrong target in mind when he criticized (mistakenly) Dignaaga rather than Bhart.rhari for maintaining that "....a word indicates an object merely through the exclusion of other objects.... [T]he word 'cow' simply means that the P.62 object is not a non-cow." The problem with such a theory, Hattori rightly says, is that "a word cannot denote anything real, whether it be an individual, a universal, or any other thing" (see note 7). If Dignaaga held some kind of direct theory of reference in opposition to Bhart.rhari's indirect theory, what kind of theory is it and how does it work? We have seen that the indirect theories of Locke and Bhart.rhari, which postulate the existence of abstract entities as the mediator for reference, don't work. It remains to be seen whether Dignaaga's conception of names can explain how a name succeeds in referring to the proper bearers of that name for all those competent users of the language. A close look at how Berkeley tried to explain the relation between names and name bearers without postulating abstract ideas will provide the avenue to answering both questions. I will argue that the prevailing interpretation of Berkeley's theory of the semantic relation between a general name and its referents is mistaken, and present what I contend is his actual position. Let us begin with an important and somewhat puzzling passage from the Principles of Human Knowledge: The ideas imprinted on the senses by the Author of Nature are called real things: and those excited in the imagination being less regular, vivid and constant, are more properly termed ideas, images of things, which they copy and represent.(24) Berkeley is first of all talking about a widely accepted distinction in the theory of ideas of that period---especially prominent in Hume's philosophy--between the vivid ideas given to the senses (the objects of sense perception) and the less vivid, but more lasting 'copies' that are somehow 'excited' in the mental faculty known as the imagination by the occurrence of sensory ideas. The fact that Berkeley called the original sensory ideas 'real things' is certainly no surprise. The point of the passage however is his recommendation that we call only the mental objects of the imagination 'ideas'. The reason is clearly that an 'idea' in his time is a mental representation, in the sense of a particular copy or image, of some object or other; and the only thing that actually qualifies as such (Berkeley thought) is an idea of the imagination, it being a copy of sensory ideas. A fleeting sensory idea, on the other hand, cannot be a copy of other sensory ideas, and there is nothing else for it to represent. It is important to remember that an idea of the imagination for Berkeley is always a particular and not a general idea (as in the idea of a triangle in general), for sensory ideas are concrete particulars and an image or copy of such ideas must itself be a particular entity. It is also very important to note that Berkeley uses the word 'represent' in two senses. the natural sense of being a copy orimage of that P.63 which is represented and the conventional sense, we shall say, of being made a sign of (or made to stand for) something. In the conventional sense, anything is capable of representing anything. The former sense is what Berkeley means by a 'mental representation', but whenever Berkeley says that 'X is made to represent Y', he is using the term 'represent' in the latter sense of X being made to be a sign of Y. For the sake of clarity, I will hereafter use the terms 'represent' and `representation' exclusively to mean 'to be a copy or image of,' in contrast to the conventional relation of signification. 'Signify', 'to be a sign of', and 'to stand for' are all ways of talking about signification. We shall use 'refer' and 'reference' as philosophical terms for the undefinable semantic relationship between a linguistic name and things in the world. It is a neutral term, so one might maintain either the position that X refers to Y because X represents Y or the position that X refers to Y by virtue of being a sign of Y. Because of his emphasis on ideas of the imagination alone being mental representations, it is very tempting to attribute to Berkeley the view that a general name is capable of referring to sensory ideas (the only real things there are) because of the mediation of a mental representation of those ideas. It is in fact commonly held that Berkeley thought that a name is the outward sign for an inward idea, the idea being the meaning of the name. According to this view, the choice of the name is a conventional matter, but what that name refers to is not. Reference, then, is based on representation. A name refers to precisely those objects represented by the idea signified by the name. If this is right, the only difference between Locke and Berkeley is that reference for the former is mediated by an abstract general idea and for the latter by a concrete particular idea of the imagination. Both would then agree on a crucial point--that reference is indirect, that the meaning of a name, the idea that it signifies, determines the references of the name. Two passages from the introduction to The Principles of Human Knowledge are often cited to support this interpretation of Berkeley: But it seems that a word becomes general by being made the sign, not of an abstract general idea but, of several particular ideas, any one of which it indifferently suggests to the mind. By observing how ideas become general, we may the better judge how words are made so. And here it is to be noted that I do not deny absolutely there are general ideas, but only that there are any abstract general ideas... an idea, which considered in itself is particular, becomes general by being made to represent or stand for all other particular ideas of the same sort. To make this plain by an example, suppose a geometrician is demonstrating the method, of cutting a line in two equal parts. He draws, for instance, a black line of an inch in length, this which is itself a particular line is nevertheless with regard to its signification general, since as it is there used, it represents all P.64 particular lines whatsoever; so that what is demonstrated of it, is demonstrated of all lines, or in other words, of a line in general. And as that particular line becomes general, by being made a sign, so the name line which taken absolutely is particular, by being a sign is made general. And as the former owes its generality, not to its being the sign of an abstract or general line, but of all particular right lines that may possibly exist, so the latter must be thought to derive its generality from the same cause, namely the various particular lines which it indifferently denotes.'(25) The first passage doesn't clearly support the 'indirect' interpretation of Berkeley, for it doesn't say that a word becomes a general name by being made to signify one particular idea of the imagination--it says that a word is made the sign of several ideas. Is it possible that Berkeley inadvertently left out the words 'any one of' before the word 'several'? With the insertion of 'any one of' in that place, we can read Berkeley as saying that a general name is not made to signify literally the same particular idea (the same image, say, of a dog) for every user of that name, due to the variation in people's experience, but one of several similar ideas for each user. The different ideas thus signified are close enough to be considered the same idea (that is, they are tokens of the same type). The second, much-quoted passage seems to confirm this liberal reading of the first passage. In it Berkeley seems to be arguing that while there is no such thing as an abstract general idea, there is such a thing as a 'general idea'. A general idea he explains as follows: "an idea, which considered in itself is particular, becomes general, by being made to represent or stand for all other particular ideas of the same sort" (my emphasis). He points out that this is comparable to the fact that a geometrical proof of a truth for a particular one-inch-long black line counts as a proof of the same truth for all lines in general (presumably of any length or color) because it is an accepted practice in geometry to take that one particular line to stand for (that is, to signify) all lines. Notice, however, that Berkeley goes on to say that it is also a fact that (in common linguistic practice) we make the particular name 'line' the sign of "the various Particular lines which it indifferently denotes." But if this is the way we come to have general names, what is the need for a general idea? If he believed that names refer to things indirectly, somewhere along the line he would have said that a name is made the sign of things by virtue of being made the sign of a general idea. But nowhere in his writings do we find anything close to such a statement. On the contrary, we find Berkeley in the first draft of the introduction to Principles denying such a view and pointing out the false assumption that motivates that view. That which seems to me principally to have drove men into the conceit of general ideas, is the opinion, that every name has, or ought to have, one only P.65 precise and settl'd signification. Which inclines them to think there are certain abstract, determinate, general ideas that make the true and only immediate signification of each general name. And that it is by the mediation of these abstract ideas, that a general name comes to signify any particular thing.(26) Why, Berkeley asks, must we assume that a general name must have "one precise and settl'd signification," one distinct meaning, that determines its referents? I contend that with the exception of the beginning of the second of our quoted passages (where Berkeley explains what he means by a `general idea'), both passages indicate that Berkeley held the view that signification alone determines and in fact constitutes reference. Berkeley maintains that a name is given directly by convention to the bearers of that name without the mediation of any special idea of the imagination. In short, he holds the theory that names are nothing more than signs of the object named. The context of the first passage bears out my reading, for it is immediately preceded by Berkeley's posing the question "Since all things that exist are only particulars, how come we by general terms," and then giving Locke's answer: "His answer is,'Words become general by being made the signs of general ideas.'" We must then read Berkeley's response as contending that, on the contrary, the significance of a name in a language lies solely in the fact that it is a conventional sign for a set of particular ideas. General ideas are nowhere to be found. So a name, then, depending on the context, is used to refer directly to an occurrent sensory idea (as in declaring "that's a dog") or to signify the recollection of an idea of the imagination (as in saying "I found a dog yesterday"). The one question that remains far which I have no satisfactory answer is why Berkeley in the beginning of the second passage (and other places) talks about an idea used to signify ideas 'of the same sort' (meaning `resemble'), as if such 'general ideas' exist. I can only conjecture that Berkeley is merely claiming that a general idea is a logical possibility-that we could, if we had reason to, make some particular idea the sign of other similar ideas. After all, what he said was "I do not deny absolutely there are general ideas." He did not say that there are general ideas.(27) IV It seemed perfectly obvious to certain philosophers like Locke and Bhart.rhari that an operative name in a language must be associated with something called the 'meaning of the name' that is known by anyone competent to use that name. How else did they think we could explain our ability to apply a name consistently to the same objects? Berkeley and Dignaaga rejected this opinion, this root assumption. That is, they criticized and rejected the belief that a word becomes a meaningful name by P.66 virtue of being made the sign of some one thing that determines the proper use of that name, the belief that the 'meaning' of a name is some kind of abstract entity. So Dignaaga says, in concert with Berkeley, that "... a name directly applies and is non-distinct [from its own referent]. A name does not apply to its own object [referent], having first presented it with another object." He and Berkeley are propounding an anti-abstractionist view of the meaning of general names. Because philosophers have gotten into the habit of talking about the meaning of a designating term as something other than its referent, it would be more appropriate to represent this view as contending that a functional general name in a language has no 'meaning', but it is a meaningful name by virtue of that fact that it signifies the same kinds of objects for all competent users of that name. Although there need be nothing more to a name, semantically speaking, than the fact that it signifies certain objects, there must be more to the explanation of how a name can have a 'right or meaningful use'. The crucial question for the direct theory of reference is this: if, as the theory implies, a name is initially arbitrarily assigned to certain objects and there is nothing more to the name than the fact that it denotes these objects, how is it that there is general agreement on future uses of this name, and how do we convey the use of the name to others? We are asking the question that Dignaaga asked of Bhart.rhari: "[How] is the relation between a name and its spatial-temporal bearer[s] teacheable?" We are asking, "How does the theory work?" Berkeley alludes to the fact that there is more to the story of how a word is made a sign of certain objects, in the following curious passage about the dog, Melampus: Suppose I have the idea of some one particular dog to which I give the name Melampus and then frame this proposition Melampus is an animal, where 'tis evident the name Melampus denotes one particular idea.... [But the word "animal" does not] indeed in that proposition stand for any idea at all. All that I intend to signify thereby being only this, that the particular thing I call Melampus has a right to be called by the name animal.(28) It seems clear what Berkeley is saying: although the word 'animal' does not stand for any idea (abstract or concrete), calling Melampus a 'dog' signifies (and presupposes) that there is something about Melampus that justifies referring to it by that name. Unfortunately, as George Pitcher points out in his book on Berkeley, nowhere in all of Berkeley's works do we find even a hint of the nature of that justification. Nowhere does he attempt to explain what Pitcher calls the word-world link. Pitcher gives a different interpretation of this passage, based on the presence of the word 'proposition' in the passage and on his opinion that "[Berkeley] sees no need for anything, idea or non-idea, to serve as a P.67 connecting link between general terms and their referents: general terms, according to him, simply denote their referents directly."(29) Pitcher maintains that in that key last sentence of the passage Berkeley could not be alluding to some justification for applying the name 'animal' to Melampus, notwithstanding what he said. On the contrary, Pitcher claims, Berkeley was merely making explicit the thought (that is, the proposition) expressed by calling Melampus an animal: that Melampus has a right to be called by the name 'animal'. In other words, to say that Melampus is an animal is to say that Melampus has the right to be called 'animal'. pitcher labels this kind of proposition a verbal proposition: [T]he thought corresponding to the sentence "Melampus is an animal" cannot...consist of an idea of Melampus combined with a (nonverbal) idea of animal in-general, since the latter idea does not exist. So it must consist of a single (nonverbal) idea of Melampus plus the thought, expressed in words, "[This] is an animal" or perhaps "[This] has a right to be called by the name 'animal'."(30) In fact, according to Pitcher, Berkeley believed that "... all purely general truths [that is, propositions] can only be verbal...," because of Berkeley's 'particularism' (the view that the only things that exist are particular entities, that is, particular concrete ideas). By a 'verbal thought', Pitcher means that "all, or part, of the thoughts will have to be in words"--and not just, as he said in the quote above, expressed in words.(31) The verbal constituent of the proposition expressed by the sentence "Melampus is an animal," is the word 'animal'. This is an astonishing explanation. Not only does it attribute to Berkeley an extreme conventionalist view of language that is unworkable, it saddles Berkeley with the dubious notion of a 'verbal proposition', for which Pitcher gives no textual support. Pitcher was probably misled by Berkeley's assertion that he "frame this proposition Melampus is an animal." He said he did so, however, after giving the name 'animal' to Melampus. But any proposition that Berkeley may have in mind must either precede or be simultaneous with the utterance of the sentence 'Melampus is an animal'. Furthermore, Pitcher should have been cognizant of the fact that the term 'proposition' did not have the meaning then that it has now. It meant essentially the same thing as what we now refer to as a 'statement'. I maintain that it is far more reasonable to read 'frame this proposition' as another way of saying 'think or utter this statement'. We should thus read the passage straightforwardly as I originally suggested: stating that Melampus is an animal presupposes that the speaker has some unspoken justification for making such a statement. But if the justification cannot be based on the 'meaning' of the name 'animal', some abstract entity signified by the word, where do we go for the answer? I believe that Dignaaga had an answer to that question, a question that Berkeley fails to address. P.68 The only place where we could go, Dignaaga advises, is to the things themselves, for Even though a word has multifarious properties, it causes the object to be conveyed by means of that [quality] alone which does not exceed over its object; not by means of qualities etc. which belong to words. (See note 6) That is to say, the reference of a name is conveyed by the name because of the fact that the name focuses our attention on a particular feature of the object named. It is not conveyed by the fact that it is related in some way to something else that the name is supposed to signify. Consider, for example, the difference between proper names and general names. A proper name is a mere label or 'pointer', Dignaaga would say, so it conveys nothing about the referent. Every strictly proper name operates like the indefinite pronoun 'this'. Not so with the various kinds of general names. Dignaaga says: In the case of proper names, a thing (artha) qualified (vi`sistat) by a name is designated as "Dittha." In the case of genus-words, a thing qualified by a genus is designated as "the cow." In the case of quality words, a thing qualified by a quality is designated as "the white." In the case of action words, a thing qualified by an action is designated as "the cook." In the case of substance words, a thing qualified by a substance is designated as "the staff-bearer," "the horn-bearer."(32) The ability of a general name to convey a referent to anyone acquainted with the name must surely presuppose the existence of what we will call the word-world mechanism (WWM). A WWM was originally instilled in those present at the time of the introduction of a general name and passed on, along with the name, to all new users of the name. The mechanism that Dignaaga had in mind, I contend, is basically a set of shared or sharable general beliefs derived from the perception of the initial referent(s) of the name, on the occasion of the adoption of the name. To illustrate the basic point: suppose I see a white bird and decide to call it a 'swan' on the basis of its color. Although this means that I call it a swan in order to contrast this bird with birds of different colors (that I have presumably encountered), it doesn't necessarily mean that I take being white to be part of the meaning of the name 'swan'. Adopting the name merely instills in me the belief that there is a kind of white bird that I want to call 'swans', or the belief that all swans are white. To believe that all swans are white is to believe in an empirical fact about things that I choose to call 'swans'. It is not to believe that it is necessarily true that swans are white or that whiteness is a definitive, essential property of being a swan. It is a short step from that latter kind of claim to the claim that 'whiteness' is part of the meaning of the name 'swan'. P.69 The illustration above is, of course, a gross ovsersimplification, not only because a single belief is usually not sufficient to convey the referents of a general name, but more important because there must be a context or background that determines that it is that belief that is instilled and not some other belief. We might say that the WWM associated with name consists of background beliefs about birds and other related matters along with beliefs in the foreground concerning those qualities of the objects that are called 'swans'. To say that a word is a meaningful name, then, is to say that there we have the ability to use that word referentially and this ability is explained by the user's possession of a WWM associated with the name. A name, of course, can be passed on, for, as Dignaaga said, "In the case of a visible object, we may teach its name."(33) For example, we can show a particular swan to someone and point out the features that justify our calling it by that name. The same WWM is thus instilled in that person, assuming, of course, that he shares our background beliefs. Thereafter, anyone possessing the WWM for a name is competent to identify the referents of that name by virtue of certain properties that they possess. There is a temptation at this point to claim that what I call the `WWM for a name' is nothing more than a sneaky way of introducing the definition of a name (despite the disclaimer that properties singled out by the WWM are not essential properties of the objects named). It may be argued that the so-called WWM for a name constitutes the belief that, there exists a kind of object (a natural kind) that necessarily possesses certain properties. To possess that belief about the referents of a name is tantamount to having a concept of what it is to be the reference of that name. But this is a total misunderstanding of Dignaaga. To see that this is a misunderstanding, we must recognize and appreciate two important features of Dignaaga's thought; the first pertains to the notion of a belief and the second involves the doctrine of apoha. Let us begin with how Dignaaga views the process of perception. According to David Kalupahana, Dignaaga reiterates his idea that perception is devoid of metaphysical conceptual constructions. This is clarified by making the distinction, for example, between cognizing 'blue' (niilam vijaanaati) and cognizing something 'as blue (nila.m iti vijaanaati). The former represents the awareness of a colored object (arthe 'rtha-samj~nii) and the latter an object possessing the color (arthe dharmasamj~nii). The former is perception (pratyaksa) that involves the conception of color; the latter is metaphysical construction that assumes the color to be a characteristic or property ( of a really existing object.(34) Dignaaga is opposed to any views that take perception to involve the existence of abstract entities called by any name, universals, ideas, or P.70 concepts. He maintains that these terms refer to a metaphysical conceptual construction, a philosophical fiction. Kalupahana alludes to Dignaaga's rejection of the obviously erroneous view that in perception we are, for example, cognizing the abstract universal property 'blue'. But there is nothing confronting the perceiver except the particular thing itself, the thing that is blue. Dignaaga does not deny that, in order to perceive an object as blue (Or as a blue object), we have to have a conception of an object being blue. But this is not to be confused with a second view that Dignaaga rejects, the view that our perception of a blue object is a product of the imposition of a concept of 'blueness' on a sensory object, the Kantian view of perception. To have a conception of an object being blue is not to be in possession of an abstract entity. On the contrary, according to Dignaaga, it is to have the disposition to relate what we now perceive to certain previous perceptions and cognitions via memory. It is a bit misleading of Kalupahana to refer to this cognitive disposition as a metaphysical conceptual construction. It is not a construction at all, let alone a metaphysical construction. It is a natural tendency of the mind to associate similar or 'identical' elements of different experiences. This disposition or tendency is precisely what I take Dignaaga to mean by a belief (even though Dignaaga may not have used that very word) . Generally speaking, a belief is essentially a disposition to act (cognitively) in a particular way under certain circumstances. The same kind of WWM is at work in both seeing and naming. Our ability to see and call something a swan, for example, is not based on the possession of the concept of 'swan', but on the possession of a set of beliefs related to previous experiences that allows us at once to see it as a swan and to call it a swan. The crucial point is that the WWM associated with a name is essentially a complex disposition to relate elements of a present experience to the same (but not literally the same) or similar elements of previous experiences. A belief is thus first and foremost a dispositional property of all agents said to possess that belief, and, as it is with any disposition, it is not something of which we are totally conscious. But of course it could be brought to full consciousness, if someone were to ask, "Why do you call this a swan?" We may say that Dignaaga had a nonmentalistic, pragmatic view of belief. We might add that knowing in contrast to just believing is, in his view, just to have a firmly held belief or a well-entrenched disposition. The significant implication of this approach to belief and knowledge is that to believe, for example, that swans are white does not entail that we have anything like a mental concept of 'whiteness' or 'swanness'. Nevertheless, it might still be argued that the set of general beliefs that function as the WWM for a name must be recognized as having the special status of being definitive of the referents of the name, among all P.71 the general beliefs we have about these objects. If, for example, the white color of a swan is one of the qualities on which we base our calling something a swan, encountering a bird which lacks just this quality of whiteness, among all the other qualities that we believe to be characteristic of swans, would tend to cause us not to call it a swan. This would not be the case--we would still call that bird a swan--if the belief that all swans are white was not in the WWM of the name 'swan'. We would instead qualify or reject the belief that all swans are white. The mainstream tradition of Western philosophy has, until recent times, maintained the position that there are two intrinsically different kinds of general beliefs or statements. On the one hand there are statements that, if true, are necessary truths, statements that are thus immune to refutation by empirical facts. The special status of the beliefs in a WWM would fall into this category. The other kind of general truths are empirical generalizations, referred to as contingent truths because they are subject to refutation by the possible discovery of some new empirical fact. The obvious implication of this two-truths theory is that there are two ways of knowing, for if the former cannot be refuted by the experience of new facts, it surely cannot be known entirely by empirical means. Dignaaga, however, would give an entirely different analysis of the special status of the beliefs in a WWM, for consider what he has said about the means to truth: Two means to truth have been declared [namely, perception and inference]. There are some who believe that knowledge derived from language is yet another means to truth. In that respect, we hold: Knowledge derived from words is not a separate means to truth from inference; for the name signifies its own object (svartha) by excluding what is other, in the same way as (the Reason) 'being an artifact' [establishes what is to be proved]. A word, which is applied to an object, illumines that aspect [of it] with which it is invariably connected, by excluding other objects, even as (the Reason) 'being an artifact' [signifies the property 'being impermanent']. Therefore, language [as a means to truth] is not separate from inference. (My emphasis)(35) It is not taking unwarranted liberties, I believe, to compare Dignaaga's statement that "knowledge derived from words is not a separate means to truth from inference" with Quine's famous rejection of the dogma that there is a distinction between analytic and synthetic truth. Remember Quine's (and the logical empiricist's) definition of an analytic statement: "A statement is analytic when it is true by virtue of meanings and independently of fact."(36) The analytic-synthetic distinction and the notion of analyticity was, in Quine's time, the most current theory or program to defend and justify the two-truths theory. It tried to explain in one stroke what made a statement necessarily true and how we came to know that it was true. Therefore the real target of Quine's attack on the P.72 analytic-synthetic distinction is the theory of two distinct kinds of truths. Dignaaga would wholeheartedly endorse Quine's conclusion that for all its a priori reasonableness, a boundary between analytic and synthetic statements simply has not been drawn. That there is such a distinction to be drawn at all is an unempirical dogma of empiricists, a metaphysical article of faith.(37) Dignaaga and Quine are undogmatic empiricists with respect to their conviction that there is only one kind of truth, contingent truth, and one basic means to truth (perception and inference being its two facets). It should have been obvious from the beginning, for if 'true' fundamentally means 'true of the world', then what we find in the world dictates what is true and what is false, and nothing accepted as true is immune from future refutation. Dignaaga must then accept the fact that the beliefs in a WWM of a name are just as subject to refutation as any other belief, and the WWM for a name is subject to revision. Nevertheless, he could also maintain, I contend, that this is an extremely remote prospect, because such beliefs concern a set of properties that is, in a sense, definitive of the referents of a name. That set is not 'definitive' in the absolute and metaphysical sense of being essentially and exclusively true of all past, present, and future references of the name. It is definitive in the relative and epistemological sense of being found to be invariably and exclusively true of all past and present referents of the name, and thus presumed to be true of all future referents. (Note that none of the member properties of a set need to be definitive in order for the set to be definitive.) Definitive properties are the means by which 'an object is conveyed by a name' as the reference of that name. But we cannot claim of a given object that it is an N on the basis of a set of characteristics just from the observation that all previous Ns (things that we call 'N') possessed those characteristics. A second observation is necessary: the observation that no other objects (things that are not references of the name) have all those characteristics. In short and in general, to apply the name 'N' to an object is tantamount to claiming that 'this is N', and that claim must be the conclusion of a kind of inference unique to Indian logic. In Dignaaga's theory of inference, it is possible to conclude that the subject of the claim (the has the 'inferable' or 'provable' characteristic-in this case, that this object is N--if and only if we have knowledge of a second characteristic of the subject called the 'li^nga' ('inferential sign') or 'hetu' ('indicator reason'). This second characteristic for N is the definitive set of properties of all Ns. The soundness of the inference is dependent on the trairuupya, the 'triple character' of the li^nga or hetu, which is: P.73 (I) It must be seen to belong to the pak^sa. (II) It must belong to all previous objects that we call N. (III) It must not belong to any previous object that we do not call N. Our disposition to call something a swan then must be based on an inference like the following: (1) This is a swan (the proposed name for the pak^sa). (2) This has P (the first characteristic of the hetu). (3) All (previous) things that have P are swans (the second characteristic of the hetu). (4) All (previous) things that are not swans do not have P (the third characteristic of the hetu). (5) This is a swan (the conclusion that the proposal to name the object a swan is correct). Dignaaga's statement that "the name signifies its own object by excluding what is other," taken out of context, can be used to support the interpretation that Dignaaga held what I called the theory of 'negative' or apoha universals.(38) But in my view, Dignaaga did not have a theory of meaning based on the notion of exclusion, for a name does not have a meaning at all. It is merely a sign for certain objects. What Dignaaga actually said (the full context) clearly shows that he is not talking about the meaning of the name at all. What I quoted above was preceded by "Knowledge derived from words is not a separate means to truth from inference," and the complete statement was: "for the name signifies its own object by excluding what is other, in the same way as (the Reason) 'being an artifact' [establishes what is to be proved.]" The remainder of the passage explains, as I have illustrated in the example above, that apoha, exclusion, is simply part of the process of inference by which we come to know the references of a name: A word, which is applied to an object, illumines that aspect [of it] with which it is invariably connected, by excluding other objects, even as (the Reason) 'being an artifact' [signifies the property 'being impermanent']. Therefore, language [as a means to truth] is not separate from inference. Dignaaga claimed that in actual fact (and contrary to the speculation of overly rationalistic philosophers) that which does the work in conveying the referents of a name is a set of properties of the referents and not some abstract universal that "exceeds over the referents." He explains how this is done with a theory of conception and inference characteristic of a philosophy that is both empiricist and pragmatic. NOTES 1 - Individual or proper names will be completely ignored in this essay, since the debate in the Indian tradition is solely over the question of P.74 what determines the reference of a general name. I should add that philosophers in this tradition generally treat any word that is applied to an object (descriptively, we might say) as a name. So a predicate like 'yellow' is just as much a name as the substantive noun 'cow'. 2 - Quoted in Dignaaga's Pramaa.nasamuccaya. The translation is from Radhika Herzberger, Bhart.rhari and the Buddhists: An Essay in the Development of Fifth and Sixth Century Indian thought (Dordrecht: D. Reidel Publishing, 1986), p. 30. I will use Herzberger's translations in this book for all of my quotes from Dignaaga and Bhart.rhari. I will also cite in these notes the original sources of the quotes, using the following abbreviations: PS Dignaaga, Pramaa.nasamuccaya VP Bhart.rhari, Vaakyapadiiya JS Bhart.rhari, Jaatisamudde`se 3 - The examples are from Herzberger, Bhart.rhari, p. 24. 4 - lbid., pp. xvii-xviii. 5 - PS 5.36; trans. Herzberger, p. 117. 6 - PS 5.12-13; trans. Herberzger, p. 111. 7 - Masaaki Hattori, Dignaaga, On Perception: Being the Pratyaksapariccheda of Dignaaga's Pramaa.nasamuccaya (Cambridge: Harvard University Press, 1968), p. 12. 8 - David Kalupahana, "Essay Review of Radhika Herzberger's Bhart.rhari and the Buddhists," History and Philosophy of Logic 9 (1988): 228. Kalupahana mentions the history of misinterpretation of Dignaaga in this review, but for a more thorough discussion, see the chapter, "Dignaaga's Epistemology and Logic," in his most recent book, A History of Buddhist Philosophy (Honolulu: University of Hawaii Press, 1992). 9 - JS 7, trans. Herzberger, p. 31. 10 - George Berkeley, The Principles of Human Knowledge, in vol. 2 of The Works of George Berkeley, Bishop of Clyne, 9 vols., ed. A. A. Luce and T. E, Jessop (London, 1948-1957). The page references will be to the Luce and Jessop edition. 11 - I grant that the epistemological distinction between the a priori and the a posteriori is conspicuous for its absence in Indian theories, so the claim that universals are known a priori to Bhart.rhari may seem highly dubious, to say the least. However, we find the notion of intuition or pratibhaa in Indian philosophy, and isn't this recognized as an a priori way of knowing? As Bimal Matilal points out, "For [Bhart.rhari], intuition is different from perception and inference P.75 and is a means by which we understand the undifferentiated meaning of a sentence as a whole.... [I]t comes from within [and]... can arise in all sentient beings, for its root cause is the Word-principle which is an integral part of sentience and hence present (potentially) in all such beings.... This is how they learn a language" (Matilal, perception: An Essay on Classical Indian Theories of Konwledge [Oxford: Clarendon Press, 19861, p. 33). Although it is beyond the scope of this essay to take up this highly controversial issue, I am inclined, like Matilal, to read Bhart.rhari as treating intuitive knowledge as nonempirical and a priori. 12 - John Locke, An Essay Concerning Human Understanding (Oxford: Clarendon Press, 1975), III.ii.6. 13 - John Mackie, in his Problems from Locke (Oxford: Clarendon Press, 1976), maintained that Berkeley badly misunderstood Locke on this key point. Mackie claimed that what Locke meant by a 'general idea' was an idea which is, "like everything else,'particular in its existence', is 'general' in its signification" (Mackie, p. 110). Locke did indeed say, as Mackle pointed out, that the general nature of words and ideas is "nothing but the capacity they are put into, by the understanding, of signifying or representing many particulars.... [T]he signification they have is nothing but a relation that, by the mind of man, is added to them" (III.iii.11) . If Mackie is right, Berkeley is attacking a straw man. It is beyond the purposes of this essay to determine who is right about what Locke meant by a 'general idea'---Berkeley or Mackie--for we are solely interested in 'Berkeley's Locke', in the view against which Berkeley sets his own position. Notice, however, that the concept of abstraction suggested in this passage--what I call isolation and what John Mackle calls selective attention--is only presented with examples of simple ideas, ideas that pertain to a single quality of an object. The most prominent example is the following: "Thus the same colour being observed today in chalk or snow, which the mind yesterday received from milk, it considers that appearance alone, makes it a representative of all of that kind; and having given it the name whiteness, it by that sound signifies the same quality wheresoever to be imagined or met with; and thus universals, whether ideas or terms, are made" (II.xi.9). It is with complex ideas signified by nouns that we seem to encounter abstract general ideas. For example, Locke said, "Of the complex Ideas, signified by the names Man, and Horse, leaving out but those particulars wherein they differ, and retaining only those wherein they agree, and of those, making a new distinct complex Idea, and giving the name Animal to it..." (III.ii.15). 14 - Locke, Essay, III.iii.7. P.76 15 - Ibid., IV,vii.9. 16 - Berkeley's famous criticism of Locke that the very idea of an abstract idea is incoherent can be seen as just the reverse of the third-man argument, the reverse of the problem of determining which particulars are instances of a universal. Locke claimed that while the abstract idea of a triangle resembles none of the triangles in actual existence, it must be instantiated by all of them. Berkeley thought that this is absurd, for "It is...a received axiom that an impossibility cannot be conceived. For what created intelligence will pretend to conceive, which God cannot cause to be! Now it is on all hands agreed, that nothing abstract or general can be made really to exist, whence it should seem to follow, that it cannot have as much as an ideal existence in the understanding" (from the first draft of the introduction to the Principles, vol. 10 in Works, pp. 134-135). 17 - Herzberger, Bhart.rhari, p. 26. 18 - PS 5.2; trans. Herzberger, pp. 146 ff. 19 - VP 1.64, p 122.5; trans. Herzberger, p. 23. 20 - JS 13; trans. Herzberger, p. 37. 21 - Herzberger, p. 45. 22 - Ibid., p. 18. 23 - Consider what Bhart.rhari said: "Divided into cows and so forth through distinctions present in those things which are its relata, [this] Being is called the [Supreme] Universal; and all words are fixed in this Universal" (JS 33; trans. Herzberger, p. 35). Dignaaga's statement (quoted in this paragraph) that "the word conveys the object through exclusion" is from PS 5.2; trans. Herzberger, p. 111. 24 - Berkeley, Principles, 33, p. 54. 25 - Ibid., Introduction, 11 and 12, pp. 30-32. 26 - Berkeley, First Draft, 18, pp. 134-135. 27 - Also recall the passage in which Berkeley claimed that "a word becomes general by being made the sign... of several particular ideas, any one of which it indifferently suggests to the mind." It is possible that Berkeley was equivocal about the notion of meaning On the one hand, to know the meaning of a name is simply to know that it signifies certain objects. But on the other hand, the name also causes a particular idea (of a reference of the name) to arise in the mind. Such an idea might be thought of as an example of the references of the name, and this is what Berkeley meant by a general idea. In short, it may be that Berkeley thought of a word as both a conventional sign and a natural sign. P.77 28 - Berkeley, first Draft, p. 136. 29 - George Pitcher, Berkeley (London: Routledge & Kegan Paul, 1977), p. 90. 30 - Ibid., p. 86. 31 - Ibid. 32 - PS 1.3; trans. Herzberger, Bhart.rhari, p. 120. 33 - PS 2.4; trans. by Richard Hayes, in "Dignaaga's Views on Reasoning (Svaarthaanumaana)," Journal of Indian Philosophy 8, no. 3 (1980): 252. 34 - Kalupahana, A History of Buddhist Philosophy, p. 197. 35 - PS 5.1; trans. Herzberger, Bhart.rhari, p. 145. 36 - W. V. Quine, "Two Dogmas of Empiricism, in From a Logical Point of View (Harper & Row, 1961), p. 21. 37 - Ibid., p. 37. Hilary Putnam, in "'Two Dogmas' Revisited, " in Realism and Reason: Philosophical Papers, vol. 3 (Cambridge University Press, 1983), argues convincingly that, contrary to the traditional (and superficial) understanding of "Two Dogmas of Empiricism," the real significance of the essay was its attack on the metaphysical and epistemological distinction between a priori/necessary and a posteriori /contingent truths, and not simply the analytic-synthetic distinction. Putnam points out that this would account for the historical importance and influence of the essay. 38 - I discussed this interpretation in part one.