JONARDON

GANERI

AK&A7

AND OTHER NAMES

Accounts of P5ribhSsiki Terms in Nytiya and Vaiiesika Texts

Prasastapfida (c. 550 C.E.), the first major commentator on the VuiSesikus&-us, initiated a controversy about meaning whose repercussions would be felt for many centuries. He claimed that the usual model for the meaning of nominal referring expression (sum@?), according to which a term refers to an object on the basis (nimittu) of that objects possession of a certain descriptive feature, does not apply in the case of such terms as time, space, and &iSu. The main reason he gave for this claim was that the descriptive feature with which such a term would be associated, if the standard model is to apply must be singly instanced, contradicting the doctrine, accepted in Nyaya and Vaisesika, that a real or natural universal property (i&i) cannot have just one instance. He therefore significantly asserted that these terms belong to a new semantic category, which he called the class of paribhlisikt names:
In the absence of a [characteristic] real property, which [absence] is due to the uniqueness of &T&Z, space, and time, the three names [iikrisa, space and time] are [to be called] piiribhasiki [terms].

I would like, in this paper, to trace the historical development of Prasastapadas idea in later Nyaya and Vaisesika texts. Though the meaning of the term paribheiki was subject to a great deal of change, the effects of Prasastapadas recognition that such terms form a special class can be felt even in a text as late as Gadadharas &zktivlidu (c. 1650 C.E.). At the same time, we will see how systems of classification for nominal expressions evolved from relatively ad hoc listings of intuitively distinct nominal types to principled classificatory schemata. Prasastapada himself offered no positive account of the semantics of this new class of expression, nor did he give any clear criterion for deciding which terms belong to it. The terms cited seem, indeed, to have more than one unusual property. First, unlike ordinary generical nominals like [the] pot, whose. reference is token-reflexive in that the reference of such a term depends on the particular context in which it
Journal of Indian Philosophy 24: 339-362, 1996. @ 1996 Kluwer Academic Publishers. Printed in the Netherlands.

340

JONARDON

GANERI

is uttered, the reference of a paribhasiki name is context-independent. Each such term is semantically associated with a single individual, to which any utterance of that term refers. An apparent corollary of this is that such terms require no feature to delimit the domain of possible referents. Second, the individuals to which they refer are all homogenous, partless substances. In this, they differ from ordinary proper names, the names, for example, of people. Pataiijali argued, and in this he was followed by the majority of later philosophers, that ordinary proper names are semantically continuous with common nouns, for each utterance of a name like Devadatta refers to a temporal slice of Devadatta, the basis being the unifying feature Devadatta-hood. This was supposed to show that Vyadis theory of meaning, according to which the meaning of a noun consists entirely in its being associated with an object, could not account for such terms.2 It seems, however, that the application of Vyadis model to the terms distinguished by Prasastapada stands some chance of issuing in a successful semantic theory, since the objects to which they apply do not change over time. Third, and perhaps most significantly, the pc?iribh@ikt terms are theoretical terms, terms introduced into the Vaisesika philosophical discourse via specific reference-fixing stipulations. Take for example the term &&u. The Vaisesikas argue that sounds, which are considered to be qualities, must subsist in some substance. Yet no perceptible substance, nor any imperceptible substance whose existence is already acknowledged, can be the substratum of sound, since none displays the properties the substratum of sound must have. There has, then, to be some other substance which is, among other things, the substratum of sounds. A name for this substance is now introduced via the following stipulation: (Let the term &iiSu stand for that substance which is the substratum of sounds). Analogous chains of reasoning presage the introduction of the terms time and space into the Vaisesika discourse. In the next sections, I will review the discussion of the paribhasikt terms in the later literature, beginning with Udayanas PariLddhi, and also the Kiru~vaZ& his commentary on Prasastapada. I will look too at Vardhamanas Kira%vali-prakGa, Raghunathas Kiramivali-praldtdhiti, and Jagadisas Siikti, a commentary directly on Prasastapada. The views of Gadadhara, whose analysis of the term &iiSu in the &zktivcida is easily the most detailed we have, can then be examined. I believe that we can locate in these discussions a well developed theory of what are now called descriptive names, names introduced by a descriptive reference-fixing stipulation. The discussion of these terms

x%L&4 AND OIRJliR NAMES

341

in Nyaya-VaiSesika is interesting for another reason. It has become a platitude to say that the philosophers of classical India do not distinguish between the analytic and the synthetic, and, in particular, that they do not recognise the existence of a class of statements which are true in virtue of meaning. It might seem, however, that the definitional sentence &Yisa is the substratum of sounds, which the Nyaya call a combined application (sahapruyoga), is as good a candidate for analyticity as any. It is also widely held that, at least partly because of the empiricism of the prurr@u theory, Indian philosophers fail to reeognise the class of a ~r~~~~ truths. Again, the combined sentence would usually be regarded as true a priori. What the Ny2ya has to say about p&ibh@iki terms and their associated definitional sentences is, therefore, of some relevance to our broader understanding of Indian philosophical thinking.
UDAYANA ON THE CLASSIFICATION OF NOMINAL TYFES

In his PuriSuddhi (1911 a: 363), Udayana (c. 1050 C.E.) reports a view of some modem Naiyayikas who claim that there are five distinct kinds of nominal expression. In one of these kinds the term is said to pick out its referent directly ~~~gugr~i~y~, lit. by grasping by the horns), unaided by any indicating features (nimittopffZ~~u~ruhite). According to these theorists, proper names like Caitra and Maitra, and perhaps also the demonstrative pronoun that (u,~uu),~belong to this class. In a second kind, the referent of the term is indicated by a feature which, although not an essential property of the referent, serves to individuate it (tutasthopaluksuna). We might say that the feature is a syndrome of the object. The usual example is the term earth, where the mark odiferousness (gundhuvut~u) serves to indicate the referent, a piece of earth. The theorists who promote this view say that this category has a subclass, which they call the class of puribh~u~i~ terms, and of which 2&%&z considered to be an instance. The difference between is the two classes is that, in the former, the feature functions as the basis for referring, whereas, in the later, it does not. The two other kinds of reference mentioned here pertain respectively to terms whose reference is based on etymology (nimitte ntarbhiitopuluk~une nirvucunikaya . . .), like the verbal noun [the] cook (@cuka), and terms whose reference is based on a bundle of features, or one feature restricted by another (nimittusuizkocunikuy~ . . . J. The example given is heaven (isvarga), which picks out an object on the basis of its being the abode of goodness, as qualified by not being the abode of evil4

342

JONARDONGANERI

Satikara Misra (c. 1450 C.E.) is the only post-Udayana NyayaVaisesika I know of explicitly who adopts and advocates this five-fold division. Some of the comments in his Ku@idu-ruhasyu help to clarify the intentions behind the classification. He raises the following question (1917: 26-27): how can it be said that the name Devadatta grabs its referent by the horns, when there is a feature, namely Devadatta-hood, which can be the basis for the reference? After all, there must be some criterion of distinguishing the referent of Devadatta from other objects, like a post or a pot5 But if this is so, then there is no justification for the claim that names like Devadatta belong to a different semantic category than ordinary nominals like [the] cow: both refer to objects on the basis of the objects possession of a certain feature. This is, of course, just the point made by those who accept Pataiijalis thesis about grammatically proper names. Sarikara M&as reply is interesting. He notes that this feature Devadatta-hood is presumably that property which collects together all the different temporal stages of Devadatta, and so at any given time, it is possessedby just one individual. Suppose then that there is someone who does not know who Devadatta is, and to enlighten her, we say This is Devadatta. Here, since the utterance occurs at a particular time, the feature Devadatta-hood cannot be doing any discriminator work. So it cannot be the basis of the association between the name and the individual Devadatta. Sarikara M&as point seemsto be that even if we associatea name with a diachronic criterion of identity, it will not help to distinguish the referent of the name at any given time. It is not clear, however, why he thinks this, since presumably at any given time the feature Devadatta-hood is possessedby at most one individual, and would suffice to discriminate that individual from all others present at that time. Sarikara MiSra also clarifies the category of pliribhiisikt terms like iik&~~. These are ones for which a syndrome (~~&ffs~~~~~ksu~~ such as being the substratum of sounds serves as an indicator (~~u~~~~) of the referent6 The word indicator is used here in that sense in which, to repeat a somewhat hackneyed example, the crow sitting on the roof indicates a certain house. That is to say, it is an adventitious qualifier, a convenient but accidental mark by which a certain house can be distinguished. As used in the context of the term &Gu, the idea seems to be that the feature being the substratum of sound has an evidential, heuristic role in the identification of the referent of the term, Although it is as yet far from clear what this role is to be, we will see later how another Naiyayika, Vardhamana, tries to make the idea more precise. What is in any case clear is that those who endorse the

%Kih

AND OTHER NAMES

343

five-fold classification refuse to reduce the semantics of the F~rib~~~ki terms to that of ordinary proper names: the former are tied, in one way or another, with a descriptive feature, but the latter refer directly. Although it was Udayana who mentioned the above doctrine, it is not one which he himself accepts. In the Kirqziivuli, he distinguishes just three categories of referring expression: nuimittiki, aupctdhiki, and pdribhlisiki. Both the first and second kind are ordinary nominals, which refer on the basis of a feature. In the first case, this feature is a generic property or natural kind (j&i), whereas in the second case it is an imposed or ~nomin~ kind (~~~dhi).7 Examples are the terms (the) cow and (the} beast respectively, where a beast is defined to be a thing having a tail, hair, and certain other properties. The notion of a basis (nimittu) for reference is here thought of as according to the following definition: a word w refers to an object x olt the basis of a feature just in case there is a property K, natural or imposed, such that w refers to x (if and) only if x possessesK. Udayana thus understands the distinction between imposed kind terms and natural kind terms as being derived from an ontological distinction among the properties on which they are based, but it is not clear why he thinks this will lead to any semantic differences between the two. However, in view of the fact that the majority of imposed kinds are composite properties (sa~~~~~dhi), it might be that the difference is between homophonic and non-homophonic meaning clauses. Thus, we have: ([the] cow refers to x only if x is a cow), but ([the] beast refers to x only if x has a tail, hair, etc.). It then remains an open question whether terms having an etymological analysis (e.g. ~p&zka, dhenu) and terms having a logical analysis (e.g., bhiita, svQ~~~)can both be assigned non-homophonic reference-clauses. Udayana now distinguishes a third kind or referring expression, to which he applies Brasastapadas label ~~ribh~~ik~ term. In his comments on ~~~tap~da, he says this (Uday~a, 1971: 70; 1956:
333-336):
The three pribhiisikr terms, &&a, time and space, are [terms which] refer directly, without a basis (nimittam antarepa @igagr&ikatay@. (Question:) What is the reason [for saying this]? (Answers) It is because of the absence of any universal such as &&a-hood, etc. (Question:) And why is that? (Answer) It is due to the fact that there is no plurality of particulars, since [ak&a etc.] are unique [entities]. That is to say, the [VaiSesika] definition of a universal, as that which is eternal, unitary, and occurring in a plurality of particulars, is not satisfied, the obstacle being due to the nature [of CkGa etc.] itself. (Opponent:) Even then, [the term ak&a] cannot be a ~~ri~~~ik~ term, for it is held

344

JONARDON

GANERI

to be imperceptible. What is said &owever] is that %Wu is that which is sounds substratum. So the propetty of being sounds substratum could be a complex condition (upiidhi). (Reply:) No! [That is not right] because[the function of] the property of being sounds substratum is to be an indicative mark (up&?k?;u~t~y~ tu@stha),just as is this when we say This is Devadatta. Otherwise, the combined sentence(sahupruyoga) &iJa is the substratum of sound would not be proper [i.e., non-tautologous]. (Opponent:) The addition of [the abstraction suffices] -tva and -tuZ to the word crkrisa would then [form a] senseless[word]. (Reply:) No. [The term &&.r-hood] denotesthe nature of &da, either metaphorically or [in the sense of its being] beyond distinctions [i.e., homogenous].

The ~~~~~~~~~~~ terms, he stipulates do not refer on the basis of any associateddescriptive clause.8Borrowing instead an earlier idea, he says that they grasp their referents directly (&-tigagriihi&~~yG). Unlike the Naiyayikas who endorse the five-fold schema, he includes Cati& in this group. Thus Udayana rejects their distinction between Prasastapadas ptirbh@iki terms and ordinary proper names, and considers both to refer directly to their referents. For him the meaning clause for &Cz is just

Udayana considers the suggestion that a term like d&a should be classified as an u~p~~b~ki, a term whose reference is based on an imposed kind, since the term-introducing stipulation {Let &Sz stand for the substratum of sounds) links &i&z with the composite descriptive clause the substratum of sound. This is just to analyse &Su as a descriptional name, a name whose reference is fixed by description. Its meaning clause would be
Mcdesctiption @x) (titi~u refers to x iff&& x is the substratum

of sound), and it is, as we shall see, explicitly advocated by a later Naiyayika, Jagadisa. Udayanas argument against this view is that it leaves us unable to explain why it is that the combined application (sahuprayoga) sentence, &L& is the substratum of sound, can be informative. If &&u is tied definitionally with the substratum of sound, then this sentence should be equivalent in meaning with the uninformative sentenceThe substratum of sound is the substratum of sound. Udayana claims that the view that c%i~a is a descriptional name misconstrues the role of the descriptive clause. This, he says, functions simply as an indicative syndrome (up~~a~a~y~ ta?a~tbff), which he compares with the role of the demonstrative that in That is Devadatta. The

i&&

AND OTITIER NAMES

345

name Devadatta is not associated with any descriptive clause, yet its reference can be indicated by means of some other expression like a demonstrative, functioning as au heuristic device. The feature being the substratum of sounds is, for Udayana, just one way of many to show others which object is the referent of the term &ii~a, and has no conventional tie with this term. Udayana also notes, as Prasastapadahad already, that &&u cannot be a n&&Sor term which refers on the basis of a natural kind, because of the doctrine, universally accepted in Ny$ya-Vaisesika circles, that a natural kind (i&G) must be plurally instanced, Thus, even if we can form gr~mati~~ly the abstract term &&z-hood ~~~~u~u), this term does not denote the essence of i&.&z, but simply the property of being distinct from other substances, i.e., the property (Xx)(Vy)(x # y iff y # &&z).~ In Udayanas three-fold schema, terms like &iSu belong to the same semantic category as proper names like Devadatta, both referring directly to their referents. We saw above that Satikara M&a, though substantially echoing Udayanas arguments, retains an earlier distinction between these two classes. His proposal is that the term is semantically tied with a descriptive clause, a tie which is not definitions but evidential. Let us now see how later Nyaya authors developed these ideas and, in particular, how they modified Udayanas schema.
LATER NAIYAYIKAS: VARDHAh,liiNA, RAGHUNATHA AND JAGADTtSA

The Navya-Naiyayika VardhanGna (c. 1440 C.E.) wrote an influential commentary on Udayanas Kiragvali, the Kira+ivali-prak%a, in which he sets down, perhaps for the first time, the classic definition of a basis for reference ~~~v~ti-~imittu) or meaning invariant for an ordinary nominal. He defines it as a feature which (i) occurs in all the individu~s to which the term can refer, (ii) is itself a meaning relatum (viicya) of the term, and (iii) regulates the presentation of the terms referents (pravyttinimittuqz ca vticyaqz sadviicyavrttitve sati vkyopasthtipakum, 1933: 34). The first clause states that the basis delimits the domain of applicability of the term, while the second clause statesthat it gives the descriptive content of the term. The last clause states that the basis regulates the way the referent is grasped in the hearers understanding. Vardham%nauses this definition to isolate a category of referring expressions called indicator terms (~p~la~Q~~vat~* The features

346

JGNARDON

CANER1

attached to these terms satisfy only the second and third clauses of the standard definition: they guide the interpretation of the term, but are not themselves contained in its semantic power (tucca v&yupasthtipa~~amlitram nicttu Suktyanturbhiitum). In other words, the hearer must identify its referent as that which is f, but it is not the case that it simply abbreviates a description, the f. Vardhamana significantly claims that &&r is an example of an indicator term, saying that &iiSa refers to that which is indicated by the feature ~~subs~aturn-of-sound-hood. V~~~~a does not believe that the reference of ordinary proper names can be explained in this way, and he therefore reintroduces the category of terms which single out their referents directly, a category into which he places names like Caitra. Following Udayana, but contradicting the earlier nomenclative practice, it is to these terms that he applies the label paribhd@ki; The stipulation or convention governing such a term concerns a particular object, and such terms lack any link with a descriptive or heuristic basis (pravyttinimittopaZak.xqzayogdbhdve sariketavi~ayasya vyaktivi&;asya). Vardhamana adds the intriguing remark that the referent of such a term must be presented perceptually. This is in sharp contrast with tik&z, which is supposed to be imperceptible. Thus Vardham&nas~~ribh~~~k~ terms are something like ostensive names, and his indicator or ~~~~~~~~~v~~terms are rater more like tbeoretieal or technical terms. V~~~~a, then, arrives at a classification in which there are four, rather than three, nominal types. I have cited his definitions of the types in the Appendix. We have so far gathered the following facts about the Naiy;iyikas use of the terror pdribh$iki. The general practice was constrained by PraSastap$%das assertion that &?iSu is n pclribh&ikt term, and this led most of the authors to name accordingly the category into which they thought this term fell. This led to a certain verbal incongruity, at least for those authors who, following Udayana, did not regard this class as in any significant sense comprising defined or technical terms. V~dh~~a retains Udayanas use of the terrr~ ~p~~b~~k~ to label the directly referring terms, but has then to introduce a new term indicator name to label the class into which he considers &%a to fall. The inappropriateness of the label p&-ibh@ikf led later to a new nomenclative practice, namely to use this label to designate terms introduced by an explicit stipulation by some author (&fhunika-s&k&a), perhaps for theoretical reasons or for the sake of economical expression, in contrast with terms which form part of the natural language, or whose introduction into the language is untraceable (cf. ~~~n~k~-s~~ketu).Those who followed this new practice (especiafly Cadadhara see below) could

iiKiih~

AND OTHER NAMES

347

no longer regard the category of ptiribhi&iki terms as comprising only proper names, for there is no reason why general names (j&i-iabda) cannot be introduced by explicit stipulation. Let me now turn to Raghunatha (c. 1500 C.E.). Raghunatha, in his sub-commentary Kiranavali-pratiia-didhiti (1932: 47-49), discusses Vardhammas theory, and appears to have some sympathy with it. He begins by reporting Udayanas three-fold classification, but suggests that it should be drawn in a different way. He defines a ptiribh@ikC term simply as an expression which refers to a single object (eka-vyaktirn~~a-bodhi~ sa~j~~ p~rib~ik~, an aa~hiki term as an expression which refers (on different occasions) to a plur~ity of objects with the help of an imposed property (ap~~i-puras~re~~~e~-~akti-bodhi~ auptihiki), and a naimittikf term is an expression which refers (on different occasions) to a plurality of objects with the help of a natural kind property (j&i-purastirena bodhik naimittiki). The point for Raghunatha is that these divisions do not represent significant semantic distinctions; ontological differences in the basis (nimitta) do not affect semantic properties. Raghunathas claim is that a single semantic proposal can account for the behaviour of all nominal expressions. Let us see how this proposal applies to the case of the term &%a. Raghun~~a endorses V~dh~~as suggestion that the basis need not itself be a meaning relatum (na tu ~u~~~tarb~~arn iti>, but believes that this is true, not just for arestricted class, but for all nouns. He does not think that the basis loses all semantic significance, however. First, it still delimits the class of permissible referents for the noun (a&kyen@i Saky&zugamasa~havGt) (c.f., clause (i) in Vardhamtias definition). Moreover, it can still, as Vardharnma suggests, regulate the hearers interpretation of the term, by restricting the way in which the referent is represented in the hearers language-based thought (Stibdabodha). We are not forced to accept Udayanas contention, that the referent is grasped in a pure unqu~ified state. Specifically as regards the term &ti&, Raghun~~a mentions a view according to which there is no conventional restriction on the mode under which 2&&a is to be grasped. Someone who thinks of its referent under description the substance different from the other eight12 is just as correct as one who thinks of its referent under the description the substratum of sound. The only conventional constraint is that they all associate the same object Likii& with the word &LiSa. Raghunathas response to this proposal is interesting, He says that we should agree that there are variations in the descriptions to which different hearers attach the word tikti& (c.f. vyat~attivaicitr~t). Yet, for the sake of economy (~~ghav~t), we must

348

JONAR~N

GANERI

regard one description as principal. Different hearers can associate the word with different descriptions without thereby excluding themselves from the linguistic practice in question, but one description, the one by means of which the reference of the term is fixed in the linguistic rule, has still a privileged status. Use of a particular description is this rule, the description being the substratum of sound, carries an implication, though nothing stronger, that this is the way the referent is to be thought of. Raghun~thas discussion had a considerable impact, especially on two later Naiyayikas, JagadiSa (c. 1600) and Gad~~~a (c. 1650). Jagadisa wrote a comments entitled Siikti (JagadiSa, 1927), directly upon the ~~~~u~~~~~-~~~~y~, one which is heavily influenced by later developments. He re-employs, at least formally, Udayanas three-fold classification of nominal types, but like Raghunatha believes that the principles on which Udayana draws it are unsound. In particular, he does not accept Udayanas claim that there can be referring expressions which pick out their referents unaided by any basis or descriptive feature. His reconstruction of the classificatory schema is as follows. A paribh@iki term is now defined as a term the linguistic rule for which is delimited by a singly instanced property (abhay~v~tti-dhar~acchi~nas~~etavat~a~ evu p~ribh~i~-su~j~~a~). An a~p~hi~- term is defined as a word whose rule is delimited by a plurally instanced imposed property (~nag~tu~~hyavacchin~a-). A naimittiki term is, as for Udayana, a word whose rule is delimited by a natural kind or universal. The difference between the first two categories is now purely numerical, for, as we have already seen, according to Nyaya-VaiSesika, a singly instanced property must be an imposed property. This way of drawing the tripartite distinction differs importantly from Raghunathas in that for Jagad%aevery nominal must have a descriptional basis. In fact, Jagadisa explicitly defends the view that &&z has as its basis the imposed property ~substratum-of-sound-hood, the desc~ption~ doctrine criticised by Udayana and encoded in ~cd~~~~ti~*. Udayanas objection to this view was that it renders the sentence &zi&z is the substratum of sound vacuous. Jagadisa replies that this is not correct, for the sentence should be construed as The word &i.&z refers to the substratum of sound (suhuprayogasyiZk&apadavticya~ iabdavtin ityarthakutventipyupapatter). That is to say, it should be read as metalinguistically asserting the linguistic rule governing the word zEk&a. He adds that Udayanas view is unsatisfactory in its own right, for it is contrary to experience (anubhavavirodhtit; counter-intuitive?) to suppose that the word i&&a presents &&a by itself, i.e., in a pure

~Kk%A

AND OIII-IER NAMES

349

unqu~i~~ form. This move reflects the ~~y~yik~ growing discomfort with the idea that objects can appear as unconcep~~iscd givens in thought.13 Jagadisa also criticises the indicative syn<j;rometheory, mentioned in the five-fold classification and supported by Satikara M&a. He says that since this view, which denies that terms like ~&Su have a descriptive content (~u~~t~vucch~d~~~, is equally applicable to ordinary nouns like [the] pot, it would have the u~acceptablc consequence that they too do not possess descriptive content. Unfo~ately, he does not elaborate on this remark, and as it stands, there seems no reason why we cannot follow Sarikara M&a in drawing a sharp line between the semantics of common nouns and that of the ~r~~~~~~ terms. Gadadharas con~ibution to the development of Ny%yathought about p&ibh@iki terms, contained in his tract $aktivxida, is best read as a critique of the views of his predecessors.Gadsdhara isolates five such views, traceable to some extent in the literature we have just reviewed. In the next section, I will examine Gadadharas criticisms of each of these theories, and conclude by considering his own view, which seems to be a modification of that of JagadiSa.
GADi%DHARAS ~O~IB~O~ TO THE DEBATEi

We have seen how a number of authors categorise the types of nominal expression. When the expression in question is a token-reflexive common noun, the classifications are based either on the ontology of the invariant descriptive basis, or on the existence of significant etymological or logical structure. A greater part of the discussion, however, has concerned those expressions which are not token-reflexive, expressions which, in virtue of the conventions which govern their use of inte~retation, are semantically associatedwith one and only one object. Among the authors we have considered, it is Raghunatha who most clearly isolates this class, using the label ~p~ribh~ik~ to reserve those referring expressionswhich designate exactly one object. The st~d~d examples of these expressions are Praiastapadas names, zi&%r, space (dik) and Yime (%&z). Some authors include grammatically proper names such as Caitra, but others deny that these are strictly proper names. Following Gadaclhara, we distinguish five primary theories about the semantics of names in this literature, which I will call the direct theory, the common core theory, the indicative syndrome theory, Raghun~~as impficature theory, and Jagadisas descriptional theory.

350
The Direct Theory

JONARDONGANEFU

Early Navya-Nyaya authors, Udayana, Vardhamana and Satikara MiSra, recognise the existence of a class of name which grasps by the horns (+zgagrtihikayti) its referent. Such a term refers directly - there is no descriptive feature, semantically tied to the term, on the basis of which it picks out an object. The proper names, like Caitra and Devadatta, as well as the demonstrative pronoun that (~~~}, are supposed to belong to this class, and Udayana, at least, includes even the p~~ibh~i~i names, &-i&.r and so on. The origins of this view in Nyaya-Vaisesika can be traced back at least to Jayanta (c. 850 C.E.), who remarked that
Simply a particular is meant common property. Thus in the proper name, the meaning is [r)itthahood]. We therefore call (ucyafe) by a word in whose referents there is no case of a word like pittha, understood to be a a particular, since we cannot speak of a universal such a word a substance-word.4

Associated with this theory is the slogan even from a word, an object can be presented in a pure, unqu~ified state ~~~ upi ni$vi~~p~~~~. Although we do not know who first said this, Gad~dh~a (1927: 65) claims that it characterised the view of the older Naiyayikas, In its crudest form this view seems to state that to understand such a term is be in a state of bare acquaintance with it. The influence of Vya$s theory here is unmistakable. There is also a certain similarity between the view of these Naiyayikas and Russells theory of logically proper names, which, though striking, should not be over-stressed. Russell claimed that to understand a logically proper name, one must be in acquaintance with its referent, and believed that we are acquainted only with sense-data, and perhaps also universals, but not with ordinary objects. The Naiy~y~~ who propound this view claim that it is not the names of sense-datebut names of ordinary objects which refer in this way. In this, we may note, they differ from those who would endorse Patafijalis thesis about grammatically proper names. Vardhamma, however, insists that the referent must be presented perceptually, and in doing so conformed with the general opinion in later Nyaya that if there are states of pure, construction-free awareness of things, they must be perceptual states. So the most likely candidates for the direct theory are names and demonstrative pronouns referring to objects in the common perceptual field of speaker and hearer. The central drawback to this theory lies in its commi~ent to a doctrine of pure acqu~n~ce with the objects referred to. On behalf of the advocates of this theory, Gadadhara notes that this doctrine has a strong and a weak reading (1927: 65). According to the strong

reading, it statesthat someune who ~de~t~ds a directly referring term grasps the referent as completely unqualified by any feature. The weak reading states that the referent is not grasped as qualified by any feature conventionally tied with the word, &W say. However, in any sentence in which the uninflected nominal stem &%W occurs, there will be other words, nominal affixes, etc. and someone who understands the sentence will grasp akasa as qualified by their referents.t5 For example, in the sentence &4%-~ is a substance, the word &Su is construed with the masculine sing&r affix from the first triplet (i.e., nominative). So C&&a will be qualified by singularity, etc. Gadgdhara notes two problems even with this weaker version of the direct theory. He points out first that the direct theory is technically inconsistent with well-established Nyaya doctrines. The Nyaya claim that in the specification of the content of a thought, a qualifier of something may only be itself unqualified if it is either a universal (j&i) or a non-composite imposed property (aI~@zriiopcidhi). &ii&z, however, is neither of these things, and yet becomes the qualifier in a sentence like [A bird] is a sky-dweller (titiSavdsin), in which it is not itself qualified by further elements in the sentence. Cadadharas second, and perhaps better9argument is as follows. The hearer is said to know that litiSa refers to &C&Z, where &%r is not singled out by a description. However, in order to acquire such an item of dispositional, memory-based knowledge, the hearer must have been introduced to akasu in some way, and, since alma is imperceptible, this is presumably by means of some description. So the proponent of this view must somehow explain away the natural idea (which Gadadhara ascribes explicitly to Murari M&a), that in our dispositional or memorybased knowledge of things we always grasp those things under the mode, in which we were first presented with thcm.t6 Gadadhara appears to think that this argument extends beyond the ~~r~~~~~~~terms, to ordinary proper names like If&ha, whose referents are perceptible (cf., ~~d~~~na). His point, perhaps is that even if one can learn a word by acquainting oneself with its referent, one would only be said to understand that word if the knowledge so acquired grounds a dispositional, memory-based knowledge of which object is its referent. To reach the conclusion that understanding cannot consist in a pure, qualification-free (nirvikzlZpakrx) grasp of the referent, he must supplement this observation with an argument to the effect that memorybased thoughts are necessarily descriptive, and this seems improbable. Russell famously &timed that memory is a mode of acquaintance, that we are ~mediate~y aware of what we remember, in spite of the fact

352

JONARDON GANERI

that it appears as past (1912, p. 26). And more recent literature points to the existence of non-conceptual memory states.17 We should note, however, that Gadadhara recognised the difficulty, and even wrote a short tract on the question, the Nirvikalpakasmaraqavada, in which he tries to supply arguments to support his contention that memory cannot present a past object in a qualification-free state. The existence of qualification-free memory states seems indeed to have been a source of great controversy among later Naiyayikas, presumably because on this question hinges the feasibility of a direct or realist theory of meaning for ordinary names. The Common Core Theory Gadadhara considers a view which like the first view denies that the public conventions governing a name tie it to an associated descriptive clause, but avoids that views dependence on a doctrine of bare, unqualified acquaintance. He calls this the indicator theory (upaZak$aSakti-vada), defining an indicator as a property whose sole function is to delimit the reference of the term, without entering the content of the thought. * The meaning clause for the term would be (3x: Mcindicator substratum of sound) (&iSa refers to x) where the sole function of the descriptive condition substratum of sound is to restrict the permitted values able to be taken by the quantifier, and hence is substitutable for any other co-extensive predicate. To use a modem phrase, it merely fixes the reference of the term, but is not part of its meaning. Any other property which picks out the same object as referent will serve the purpose equally well. According to this view, each language user identifies the referent of ktii~a under a mode or descriptive feature, but the particular mode may vary from person to person. One hearer might identify &%a under the mode the substance different from the other eight substances, another under the mode the substratum of sound. There is no hard and fast rule.20 The only restriction on the idiosyncratic descriptive features - _I is that they all have the same extension, viz. the object akasa. I think Wiggins has something similar in mind when he remarks that
the sense of a proper name simply consists in its having been assigned whatever reference it has been assigned: to know the sense of n is to know to which entity n has been assigned, a single piece of knowledge which may be given in countless different ways by countless different descriptions (1975, p. 11).

.%KJih

AND OTHER NAMES

353

Evans, citing this passage, remarks that it is difficult to hold, of ordinary proper names, that there is some particular description semantically associated with the name (1982, p. 48). One might imagine the linguistic community as comprising individuals each of whom identifies the referent of the name Cam-a in their own private fashion: for one, Caitra is the brother of Devadatta, for another he is the man last seen in the launderette, etc. (c.f., vyutpattivaicitrtit - Raghunatha). But what guarantees that they all understand the same name Caitra in a public language is that for each it is true to say that they know that Caitra refers to Caitra, and it is the statement of this common core fact which should be the meaning clause for the name. Thus, the function of the domain restrictor in the meaning clause is simply to fix the extension of the term. It is, in Gadadharas jargon, an indicator or upaZak?qza. As an account of what is involved in understanding ordinary proper names, this theory has much to recommend it.21 But as applied to the paribhasiki term &kzi~a, it is less plausible. The common core theory now says that &YSa refers to a certain object, which is indicated by the description the substratum of sound (SubdiiSraya~opaZuksitu - _I evakusupadu-Suktir). In this view, the only distinction which accrues to the description the substratum of sound is that it is the description associated by the name-giver with the name; it has no other semantic salience. Both the direct and the common-core theories are motivated by a desire to explain the apparent informativeness of the sentence &ii& is the substratum of sounds by analysing it as having a simple namepredicate form. Gadadhara, however, rejects this view, precisely because it does not accord a sufficiently distinctive status to the description the substratum of sound in the semantics of &iSu. The sentence &iiSa is the substratum of sounds seems to be a priori; on the current proposal, however, the truth expressed by it is a posteriori for anyone who does not identify akaSa under the description the substratum of sounds, and vacuous for anyone who does.22 The Syndrome Theory The traditional form of the indicator theory seems rather different from Gadadharas reconstruction. The advocates of the five-fold classification say that the substratum of sound is a syndrome (tufastha) for the referent of titiia. This suggests that it has a more significant role in the recognition of the referent. Vardhamana adds a remark which may give a clue to the nature of this role. He suggests that the feature is a sort of interpretational heuristic (vucyopasttipaka). Perhaps his

354

JONARDON

CANER1

idea comes to this. Someone who understands the name &&a knows that there is associated with it a descriptive feature the substratum of sound which she is able to draw upon as an aid in identifying the referent of the name. However, it is not required for understanding that the person should think of the referent under the description the substratum of sound, nor should this description matter to the hearers interpretation of the statement in which the name occurs, or to the truth-value assigned to it. The description is simply a communitywide dossier of information associated with the name.23 If this is right, then the view bears some resemblance to Donnellans analysis of the referential use of definite descriptions, according to which the descriptive clause is only a heuristic device for making a reference.24 Matilal seems to think that this is the way to understand the doctrine:
a distinguisher or qualifier may identify an object in a totally irrelevant way, or it may be very pertinent to the identification of the object in the context of the further attribution of properties of it. I can identify an object as the man with the whisky-glass in hand, and go on to say that he is a logician, where the distinguisher, viz. having a whisky-glass in hand, is totally irrelevant to his being a logician. But if I go on to say that he is going to get drunk very soon, then at least the audience may understand that the identifying qualifier is relevant. Nyaya underlines the distinction by calling the relevant distinguishers viSe;qa, and the irrelevant ones upaluk~a~a. One can in this way identify an object by an irrelevant qualifier, and assign a proper name to it (1985: 387).

So it may be that the doctrine that a descriptive feature is a syndrome (tu~ustha) or interpretational heuristic (v~?cyopasth@aka) is a defensible one in some cases, such as the referential use of nominal descriptions. But once again, it does not seem satisfactory as an account of the relation between &iiSa and the description the substratum of sound. The reason is that this description is our only access to the referent of &i&z, and entity which is by nature imperceptible. Someone who used the term &ii&z has no other means than the associated description to identify the referent, so the role of that description cannot be merely evidential. Raghuniithas Theory Raghunatha has a quite different idea. His view is roughly as follows. The linguistic rule
MGndicator

(3x: substratum of sound) (&u&

refers to x)

- _I just states that iikiika refers to akasa. However, the use of the descriptive phrase the substratum of sound is not simply an arbitrary way of

&&M

AND OTHEtR NAMES

355

fixing the reference of the name: the linguistic rule non-sem~tic~ly implicates that L%Z%&Z should be identified under the mode the substratum of sound. In other words, this descriptive feature, as the mode under which &%a is presented in the public linguistic rule (sa~etiLavi~e~y~t~vacchedaka), regulates by implication the way hearers identify the reference of &i~a. Raghunatha nevertheless agrees that different hearers identify akasa in different ways, and are not thereby linguistically incompetent. The point is that each such hearer is in accordance with what is literally said by the rule, even if they do not respect its non-literal implication. Of the theories considered, ~ghun~thas is the first to assign to the descriptive clause a distinctive semantic, as opposed to evidential, role. This becomes clear when we consider the sentence tiESa is the substratum of sound. What this sentence literally states, on the current proposal, is that a certain object is the substratum of sound, and as such, has the same status as any contingent name-predicate assertion. But anyone who fully understands the word &i&z will know that the conventional implication of the linguistic rules guarantees that the sentence is true. So there would seem to be a clear sense in which the truth expressed by the sentence is a priori. This imaginative way of dealing with the peculiar semantics of descriptive names deserves further attentionF5 Gadgdhara criticises Raghunathas view in a terse and rather enigmatic passage (1927: 62). He suggests that there are certain terms whose linguistic conventions must be regarded as explicitly stating, and not merely implicating, the description or feature under which their reference is to be picked out. Gadadhara notes that ICanada,the author of the Vui&#u-sidru, gives the word substance (dravyu) a technical meaning by introducing it via the rule Let the word substance be construed as about an entity under the mode possessor-of-qualities.z6 The point, presumably, is that such a rule fixes, not just the reference, but the descriptive content of the term as well. According to Gadgdhara, however, there is no significant semantic difference between such terms, whose introduction can be traced to some explicit definition (c.f., lidhun~k;a-savketa), and those which cannot, since the same account is given of what it is to understand them: to understand such a term is to know, in any given context, which object is its reference and how the reference is to be identified. Gadadhara believes that his own proposal not only adequately explains the semantics of these noun phrases and singular terms, but can even be extended into an account of pronominal reference. If this is correct, then his account is to be

356

JONARDON

GANERI

regarded as preferable on the grounds of having greater explanatory power. The Descriptional Theory

The last doctrine to be considered is the theory that names are distinguished from ordinary nouns only in that their associated descriptive feature is singly instanced. This view is most closely associated with Jagadisa.27 On this view, it is a definitional truth that a name w refers to an object x iff x possesses a certain unique feature (or satisfies a definite descriptive condition) 7r. Views such as this are commonly called description theories of names, and names (if there are any) for which this theory is true are called descriptional names. The associated descriptive clause in the case of the name &&a is the substratum of sound, and its reference clause is: (Yx) MCdesxiption (kWa stratum of sound). According to Jagadisa, in the case of ordinary proper names such as Devadatta, there is the feature Devadatta-hood or the clause the thing x such that x = Devadatta. Jagadisa apparently goes so far as to say that for every individual, there is a peculiar property unique to it, the property of being that individual (tadvyaktitva; compare the medieval notion of haecceity). One need not, however, agree with him about this in order to accept a descriptional theory of names. It is a consequence of this view that the link sentence d&a is the substratum of sound is a tautology, but as we have already seen, Jagadisa explicitly recommends a meta-linguistic reading of this sentence, as asserting the term &Sa refers to the substratum of sound. This view sees names like &%a as theoretical terms introduced into the discourse to abbreviate a longer description. It should be born in mind, however, that the longer description is itself used to make a reference, and the description theory does not commit one to the Russellian thesis that such names are disguised quantifier expressions. The Nyaya, as already noted, claim that to understand such a name is to entertain a descriptive singular thought about its referent? There are, of course, special problems with the notion of having singular thoughts about imperceptible objects like &iiSa, or abstract objects like numbers, but if the notion of descriptive singular thoughts, thoughts about objects under descriptive guises (dharmittivacchedaka), is at all defensible, then it might well be extended to what is not perceivable. refers to x iffd&, x is the sub-

Gad&&-m himself te~tat~vely axepts Jagadsas view on the matter*

but notes that it requires us to attach a non-literal interpretation (~~~~~) to the term &-iSu in the combined* sentencei&i&r is the substratum of sound* (1927: 73), for we have there to construe it as synonymous with the phrase the referent of &&W, He therefore suggests, as an altermttive, a way in which this term can be regarded as occurring literally, In what seems to be a statement of the modified version of the indicator theory, he says that:
in C&&X t0 6+&R ftbe PfrsSibiiiq Off z?& ~t~~~~tiQR WXkF kh? fX-KKk %hl? SUbS&nC@

difFerem fmm the o&ex ei&t snhrances fmn the name Y&z&z, used ~~~~~~~, one &mild admit that the hu-e~ knaws the reference under a mode fm. viz. &e substance

diGkent firrm the atbe~ eight substances] wvi&21 is not a semm&c v&e, [and that this made] wguiates an iateq~tation {of the referencel under that mode [IR].~

When a term behaves in this way, he calls it a %~&%x%x&P rime, a name for which the basis of application is not a meaning relatum (c.f., Vardhamtias definition of an uphkppavuti name). The idea is that anyone who understands the term must think about its referent in a certain way? but that there is no one such way fixed by the linguistic rules, That Gadadhara is not comfortable with this solution is clear from his remark that it has to be accepted because there is nothing better, and his unease might be due to the fact that the d~sti~~t~v~ semantic role of the description the s~bs~~~rn of sound has once again dropped out of the picture. Gad$tdhara,in the end, leaves the matter open to further reseat&, In the h&tivL-lda, Udayanas three-fold distinction is dismantled, and replaced with a case-by-case examination of individual words. He does retain the term ~Wibh~~ik~, but defines it now as a noun whose convention is hovel (tidhunika, 1927: 5). The ptiribh@ikt terms, for GadMbara, are those whose introduction can be traced back to some explicit naming act. His discussion of the meaning of &&W reveals a fascinating debate that raged within Nyiya circles, and the ~rn~~~~e they attached to the topic. We have been able to trace some of the p~~ip~~ in this debate, and the views they espoused, and in doing so hope: also to have revealed something about the process by which schools like the Nyaya developed, a process which involved internal criticism and disagreement just as much as debate with philosophical adversaries,

358

JONARDON

GANERJ

NOTES nkrlSakaladi&n ekaikatviid aparajiityabhrtve sati piiribh@ikyas tisrah sahj% bhavanti. Prasastapda (1927: 308). The term paribh@ikl is usually translated as

technical term, but in view of the wide range of theories about these terms, such a translation may easily become misleading. I prefer, then, to leave it untranslated. It would be interesting to investigate whether Pasastapadas of this term is in use any way influenced by the Patiinian term paribhiisa or meta-rule. I have discussed Vyadis theory in more detail in Ganeri (1995). 3 This is claimed by Satikam M&a, for whom see below. 4 A comparative list of this and the other nominai classificatory schemadiscussed pre is given in the Appendix. satyam acetem st~ha-ku~h~aa devada~a-sa~~~y~ t~dr~pye~himata~~t (1917: 26-7).
6 pdribwiki palak!anasya yathakBam iti iyam eva fafasthalaksayty tatasthatvm (1917: 27). ucyate Sabddrayatvasyo-

For this distinction, see Matilal (1968: 50) and (1986: 417-Q. a C.f. Jagad%a(1930: 309): niravacchinna-sahketa&litvam eva pmibh&ikatvam
. . . ity&XVyamatam.

The intention behind the use of . . . i&t . . . in such a clause is that no other co-extensive description can be substituted on the right hand side. This might be achieved better, though with greater theoretical commitment, by preceding the clause with a modal operator.An alternative (here I am indebted to Eberhard Guhe) would be to introduce some meansby which we can name intentionally individuated properties, sfch as that of Bealer (1982). Cf. JagadSa(1930: 309): na caivam ~~a~arn i~~diprayugo na sy& prakfyarthat~va~che~~-dhurmasym.va ~a~~dyartha~~ ~atal~din~ bh~u-pra~aye~hidh~n~t. iti vcicyae vy~vartaka-dh~~~uiva

It should be noted, however, that he does not explicitly mention ordinary proper names here. I2 There are nine categories of substancein the standard VaiSesikaontological scheme.See for example Annambhaffa (1983). :: For more on this point, see Matilal (1988). yeTarn arthefu sm-nyam na sambhavati taih punah ucyate kevala vyaktir
. . . evam ditthadifabdiin@t sa?jiiatvaviditatmanm abhidheyasya samiinyaSiinyatvi7d vyaktiv&& ata eva hi dravyahbda ity ucyate, (Jayanta, 1936: 298). I5 nirvikalpa~~a~ ca radarpie tatpadopasth~yaprakara~a~, na tu tadaqt+fe smiinyato ni~prak~a~~ (1927: 65). I6 tatra nirvi~lpa~r~asya s~ra~sy~napapatti~, anubhavasya svasam~aprak~aka-rn~~a-sm~ti-janaka~~~ (1927: 67).

:i See Evans (1982: esp. 239). C.f. Gadadharascommentator,Harinatba: upaluk?anatvaF padu-janya-bodhpavi~ayatrivacchedakah~acchinna-Sakti~i~ayat~-~unya~am. (1929: 91). This technical use of the term upalak~a~ in the Saktivada has to be distinguished from the use of the term in earlier literature as meaning a heuristic device, a use dating back at least to Bhartrhari.
I9 tatprakirraka-Bbdabodhe tadar@e natiriktaprasaktatvariipa-Sakyatavacchedakatvavag~hitvenopalak+e Saktijfiiinasya hetutv& (1927: 71). For details about restricted

quantifiers, as denoted by such formulations as (3x: F) (Gx), see Neale (1990: 4&43).
~ak~~k~~apad~ alga-dravy~tirikta-dravya~~in~ niyumata~ ~abd~raya~e~iva (1927: 70). py i?k&a-bodho bhavati na tu

-iiti!h

AND OTHER NAMES

359

* It is partially defended, for example, by Evans (1982: 399-400). However, the independencehe alleges to exist between using a name to fefer in a given practice and ~~erstu~d~~~ the nitme does not seem to be acknowledged by the Naiyayikas 22 See also V~~vabandbu (1999. 23 C.f . Evans (1982: 399), who notes that even if everyone now associatesthe name Homer with the description tire author of the Iliad and the Odyssey, it is still a f;oper name for Homer. RonneIIan (1966). Kaplan (1989) also reeognises the existence of a heuristic eiement in reference. Discussing Do~e~~ans usual example of a referential use of a definite description (Who is the man drinking martini?f, Kaplan remarks that according to my new view of what determines the referent of a demonstrative, the demonstration (here, the description) is there only to heip convey an intention [to refer to the perceived object] and plays no semantical role at all. He calls the description an off-the-record element in language. z The notion of non-litend implicature is developed by @ice; see his (1989)=
= dravya~d~ gu ~~v~~~~~- dravyaqt b~dd~v~m ity&@raka ~ha~~k~ kusya tit satiketo dravyalvc;idyavacchi-v~~~ff~-sanr gyki&s tatra ~~~a-sa~etagrahdd drav~a~e~a drama-bQdh~atter gunatvadi-prakaraka-b~~apapatte~ ca durviiratvti ( 1927: 62). ubhay~~~~d~arm~vacchjnna-sa~e#avat~~ p~ribh~~~ku-salltjiicrntam (1930:

309). 2* McCulfock (1989: 291-2) is one modem author who entertains such a theory of desc~p~ve names. B &i,%a@&z cfiaktyzistabravyiitirikta-dra~a~~i~~ b5dha-~~~,~~~ vi~~~?~v~ca~-ktiadipadasya prak@rci~& v~cya~~avag~~ta-~ak~~ruhasya~va tad-dharm~acchinna v&+yakasya tatprak&anvayabodhe hetutvam agaty&gz%ryum (1927: 73). Compare also the comment in the Maiijirsa: &&$a-[a~~adravy&riktadravyatva]~dharmaprak~aka-~a~ab~dh@ taddh~~~ dh~mit~vuccheda~~~a Dada-prak~a~-ja~ya~ ~~bdab~ha-viSaya~a-laissha-prakru-a hetur (ibid.),

REPERENCES Ann~bha~a. (1983). ~~kusa~rahad~jk~ on Tar~~~r~~. Ed, and trans. G. Bha~h~a. Calcutta: Progressive Publishers. Beaier, G. (1989). @aa& and Cmcept. Oxford: Oxford University Press. Donnellan, K. (1966). Reference and Definite Descriptions, Philosophical Review 75. Evans, G. (19g2). The Varieties of Reference. Edited by J. McDowell. Oxford: Clarendon Press, Adaria. Ed. Pt. Gadadhara Bha@carya. (1892). ~akt~v~~. With Su~&~~as Sud~~~a. Calcutta: Own Press. GadadharaBhat$%%rya. (1927). ~aktiv~a. With Krsna Bbagas Magjiiga, Madhava BhatmcsuyasVivytti and Sabitya RaxSanacaryas vinadini. Ed. Gosvami Damodara Sash-i.Benares: Kashi Sanskrit Series no. 57. GadadharaBhat@%rya. (1929). saktivada. With Harinatha Tarkasiddhmtas Vivrtti. Ed. Gosvami Damodara Sastri. Benares: Kashi Sanskrit Series no. 77, Gadadhara Bha~~c~a. ($933). ~irv~~~akasmar~v~. In V~av~~dbi. Ed. 3, M&a and Dh. Sastri. Benares: ~how~amba Sanskrit Series nos. 421, 446. Cane& J. (1995). Vy$i and the Realist Theory of Meaning, fownal of Hian
Philosophy 23: 403408.

360

JONARDON

GANERI

GangeSa Upadhyaya. (1901). Tattvacint&nuni, Sabdakha@a. With Mathutanatha Tarkavagitis ~~~syu~ and Jayadeva M&as Aloka. Ed. Kama~y~a~a Tarkavagisa. Calcutta: Biblio~eca Indica. &ice, P. (1989). Studies in the Way of Words. Cambridge, Mass.: Harvard University Press. Jagadisa Tarkalatikam. (1930). Sri!&. In Prasastapada (1930). Jayanta Bhaffa. (1936). Nyayumafijuri, Ed. S. N. Sukla. Benares: Kashi Sanskrit Series no. 106. Kaplan, D. (1989). Demonstmtives: An Essay on the Semantics, Logic, Metaphysics and Epistemology of ~mons~tives and Other Indexicals. In J. Almog, J. Perry, and H. Wettstein, eds., Themes From Kaplan. Oxford: Oxford University Press. McCullock, G. (1989). The Game of the Name. Oxford: Clarendon Press. Matilal, B.K. (1968). The Nuvyunyiiya Doctrine of Negation. Cambridge, Mass.: Harvard University Press. Matilal, B.K. (1982). Logical and Ethical Issues in Religious Beliefs. Calcutta: University. Mat&l, B.K. and Shaw, J. L. eds. (1985). AnuZytjcal Philosophy in comparative Perspective. Dordrecht: Reidel. Matilal, B.K. (1986). Perception: An Essay on Classical Indian Theories of Knowledge. Oxford: Oxford University Press. Matilal, B.K. (1988). Some Issues in Nyaya Realism, addressed to the Oxford Radhakrishnan Centenary conference Realism and Non-Realism. Stephen Neale. (1990). Descriptions. Cambridge, Mass.: MIT Press. PraSastapada. (1927). Pra~astap~~~a. With Jagadisas St&i, Padman~ha M&as Setu and Vyomasivac-aryas VyomavatiI Ed. G. Kavimja and Dh. Shastri. Benares: Chowkhamba Sanskrit Series no. 374. PraSastapda. (1971). PraSustapifdabh~ya. With Udayanas Kiraniivalr; Ed. J. S. Jetly. Baroda: Gaekwad Oriental Senes 154. Raghunatha Siromani. (1930). Prutyakp-mani-didhiti (pramanyavuda). With Gaddhams G~~d~r~. Ed. P~dv~dibhaya.~~ Ananutcnrya. Raghun~tha Siromani, (1932). K~ra~va~~prak~u-d~hi~i. ed. Gopinatha Kaviraja. Benares: Princess of Wales Saraswati Bhavana Texts no. 38. Satikara Misra. (1917). PraSustap?idubhusyatfkrTsamgraha, including Satikara Mist-as Karufdarahasyam. Ed, V. P. Dvivedin. Benares: Chowkhamba Sanskrit Series 231 & 255. Udayana. (1912). Ny~yuv~tikka-t~tpurya-pari~Mddhi, with Vardhamanas Nyayanibandha-pra~a. Ed. V. P. Dvivedin and L. S. Dravida. Calcutta: Biblio~eca hrdica, no. 192. Udayana. (191 lb). Kiranavuh, with Vardhamanas Prukasa and Rucidatta Miltas Vivrtti. Eds. S. Satvabhauma (Fast. I-III) and N. C. Vedantatirtha (Fast. IV, 1956). Calcutta: Bibliotheque Indica 190. Reprinted in 1989 by the Asiatic Society, Calcutta. Udayana. (1967). Nyayuvartikka-tatpurya-purisuddhi. See Gautama (1967). Udayana. (197 1). Kirqzizvuh. See P~~as~pada (197 1). Vardhamana Upadhyaya. (1911). Kjra~va~pra~u. See Udayana (191 lb). Vardhamana Upadhyaya. (1933). Kiraniivali-prukusa. Ed. G. Kaviraja. Benares: Princess of Wales Saraswati Bhavana Texts no. 45. Visvabandhu Tarkatjrtha. (1994). Proper Names and Individuals, in B. K. Matilal and A. Chakrabarti, eds., Knowing From Words. Dordrecht: Kluwer Academic Publishers.

iiK&k

AND

OTHER

NAMES

361
SYSTEMS

APPENDIX:

TABLE OF CLASSIFICATORY

The five-fold schema mentioned by Udayana: Name 1. 2. 3. 4. 5. direct etymological pmibha@ki syndrome-based bundle-based Definition nimittopalaksana-rahite @riga-gr&ikaya nimitte ntarbhtitopalaksane. nirvacanikaya . nirnimitta eva tafasthopalak+ne paribhtianikayd . . sanimitte taFasthopalak;aF eva?prabikaya viSe?avannimitte nimittasarikocanikaya Example Caitra piicaka ok&a ppythivr svarga

Samkara MiSras five-fold schema: Name 1. 2. 3. 4. 5. direct paribha$ki etymological basis-possessing bundle-based Definition @riga-gr&ikaya tafasthopalaksav antarbhiitopalaksana. . sanimitta nimittasarikocaniya Example asau, Devadatta ak&a ppircaka go svarga

Udayanas three-fold schema: Name 1. paribh@kF 2. auptihikf 3. naimittiki Definition nimittam antarena Syfigagrtihikaya up&hi-nimittaka jati-nimittaka Example aka$a, kala, U&k pa& Wipa, gandha

Vardhamanas four-fold schema: Name 1. 2. upalaksanavati 3. piiribh@iki Definition jatiriipapravfTtinimittavati vdcyopasth@akamama pravrttinimittopalak~a~yog~h~ve sa+etavi?ayasya vyaktivisefasya pratyaksenaupasthiipaka, SyrigagrAikaya pacikriycintarbhnvena prTkanukiilakrtiyogyatamatre Example go nkclSa Caitra

4. aupadhikt

piicaka

Raghunathas three-fold schema: Name 1. piiribh@ikf 2. aupadhikl 3. naimittiki Definition ekavyaktimcrtrabodhika upiidhi-purask&ep anekavyaktibodhika jati-puraskare@ bodhika Example

362
JagadiSasthree-fold schema: Name 1. piitibh@iki 2. aupadhiki 3. naimittikt Definition

JONAFXXINGANERI

Exampie
~k&z

ubhayavyttidharm&xxzchinna-sahketavat anugatopadhyavacchinna-sarnj&i (anugatajatyavacchinna-sanzjfia)

Department of Philosophy, Kings College Lmdon, Strand, Lmdon WC 2R2LS, United Kingdom