20th WCP: Professor Matilal’s Nåvya-Naive Realism vis-a-vis Dummett-Putnam-Mimamsa Anti-Realisms

The Problem

The vexed issue of the precise connection between words and things (or objects) has been a major preoccupation over the centuries, summoning the resources of metaphysics, philosophy of language, linguistics, ontology and increasingly semiological analysis, to solve this problem. Indian philosophy produced a number of different and often conflicting solutions, only to be rivalled by the even more bewildering variety of approaches and theories witnessed in the West, traditional and modern, relying largely on various model of the ‘word’ (natural, ideal, scriptural, semiotic, etc.). In this paper I want to suggest that there is an even more intricate relationship between the model of the word or language and the background view of the world. In other words, it is not at all as simple as sitting down one fine morning and asking, "Well, ol' boy, what is the connection between the word and the world?" as though it is a question simpliciter about some given or givens in our environment. Many theories proceed on the basis of this assumed dualism, if not a complete asymmetry between language and the world that it is supposed to describe, represent, or whatever its function is taken to be in relation to each other; for instance, the strict 'atomistic' and propositional theories of language assumed a transparent distinction between words and the world, such that a correct use of language would successfully represent or picture the structure of the world. Language is the `grammar of the world'. As Rorty has shown there has been an obsession with the linguistic end of the analysis. The world is elusively the backdrop to which language has to be re-connected, at least in our understanding and intellectual charting; for common sense tells us that there is a connection, otherwise we would fail to communicate with each other, relate newspaper blotches to events elsewhere, carry on our daily social commerce, fetch the glass of water when asked for one, and take formal promises made by our partners or friends or gurus impeccably seriously. But to ponder whether in some logical or philosophical sense language could precede the world, give us the world, and might even be the condition for the possibility of the world is too often considered to be an evocation of sheer sophistry, or merely of mythological significance. Although there are those that did ponder along these lines: Plato with his theory of Eidos, the Mimamsakas with their doctrine of apaurusheyatva, Bhartrhari who collapsed the ontology of the world with a higher speech form (sabdabrahman), and Wittgenstein in his more mystical aphorisms about language (continuing a Kantian insight into the apriocity of our categories of understanding). But I think there are even more radical critiques and divergent views of language and the world than these few idealistic tendencies might suggest. I want to bring to the forground the late Bimal K Matilal’s development of Nyaya-Vaisesika realist approach to the aporia, and occasionally interject the analysis with the dissident Buddhist voice, especially of Candrakirti. More significantly, it will be the living ghosts of Putnam and Dummett (or Putmett for short) that I will invoke as it were to haunt Matilal’s variation on metaphysical realism. The paper is dedicatory as it is programmatic, but I do hope that before long scholars of Indian philosophy will begin seriously to review Matilal’s work, as some of us have been doing with the Bhattacharyyas, Mohanty, Potter, Daya Krishna among others.

The late Professor Bimal Krishna Matilal wrote a number of works in which he attempted to get a handle on the problem we have raised. Steeped deeply in classical (Indian) scholastic philosophy, and inspired by analytical currents in Western philosophy, he skilfully analysed and debated several of the major Indian systems (Nyaya, Buddhism, Grammarian, Mimamsa). But his own view was not always clear. It would not be unfair to say that he veered closer to a realist ontology or metaphysical realism as he Christians it after Davidson (evident, e.g., in his masterful work on Perception), and this may have informed or impacted on his formulation of a theory of language appropriate to this ontology. In fact, I want to argue this to be the case, despite his enormous attraction to phenomenalist-constructivist theories (especially Buddhists ones), and his flirtations with Bhartrharian holism, even Saussurean semiology and lately Derridean deconstructionism as he completed his essay on The Word and The World. I am intrigued by all this and so I wish to trace this line of thinking in Matilal's work, which I believe awaits a fuller response from the Buddhist dialectical CRITIQUE OF LANGUAGE, which Matilal chastises as 'linguaphobia'! (ibid p 6) Again, it is for this reason that I choose to contrast Matilal on this problematic, as it would seem that at least one insight of Candrakirti's poses a serious challenge to Matilal's ontologico-linguistics, viz: 'there is no real connection between the reflected image and the face', or for that matter Dinnaga's supposed utterance: 'Language issues from imaginative construction and vice versa'.

I will begin with some clues from Matilal's published works, and from a succinct paper entitled 'Some Issues of Nyaya Realism' which I think is not yet published. Now the very first sentence of the first chapter in his Navya-nyaya Doctrine of Negation, begins thus: "In the West logic has been primarily concerned with propositions or sentences. Navya-nyaya, like other Indian systems of logic, deals rather with what it calls jnana, by which it means something close to `particular instances of cognition'. An instance of cognition, it is true, can be shown to be ultimately related to some verbal form, namely, to a statement or sentence. In the case of determinate or qualificative cognition (savikalpa or visista jnana), with which Navya-nyaya is chiefly concerned, the relationship is very close. But jnana itself is not a form of language...." (NNDN p 7). Here then is a disavowal: jnana essentially has nothing to do with sentences or verbal forms; certain of its qualificative forms might involve verbal delimitors, but its primary connection is with the world of objects, directly and inexorably. So jnana-s "are always cognitions of or about..., and what is denoted by the expressions that fills the blank here is to be taken as the object or objects of such cognitions". Cognition always refers, beyond itself, to some object. As such the object is neither compromised nor ascribed a new property (such as jnatata a la the Bhattas). The world is as it is and our cognitions simply bring its varied furniture into our sharp conscious focus, no more and no less. There are two other features about jnana that we are urged to note.

The object of cognition, at least by the time of Navya-nyaya, has nothing to do with some supposed residue of sensory core, the primitive unrepresentable or unstructurable given 'on par with sense-data, percepts, or uninterpreted sensations' (SINR p 3). Some theories assume or posit this sensory core either to give legitimate support to an objective characterization of particulars in experience, or simply to force an argument towards an utter subjective, conceptual or indeed imaginative construction of a particular, conceding that there is no way of establishing any causal connection between this internal core of experience and what appears as the external object of awareness. Matilal attributes the latter motivation to Dinnaga and makes the curious following observation (I shall quote this long passage since it very intriguing): "These self-characterized objects", he writes, "which are similar to sense-data or percepts and which some philosophers in modern West have described in language by such verbless complex 'Red here now' are unanimously regarded by Dinnaga and other Indian philosophers [count Candrakirti in here] as virtually and actually ineffable. This obviously depended upon a view of language. Use of language depends upon shared properties, which are results of the synthesizing aspect of mental construction. For language is a social affair (as Quine had once said). The private sensory core, which are unique particulars, cannot therefore be expressed in language. This is also reminiscent of the argument against the notion of a private language a' la Wittgenstein in modern times.' [end quote].

But what does Matilal intend by suggesting that '[T]his obviously depended upon a view of language'? What is the referent of 'This' here: the claim that there is an ineffable core of sensory private experience? or its attempted description? I think he intends the former, but if so why is this metaphysical thesis about the unavailability of language made depended upon a prevailing view of language? Isn't this a claim that stands on its own right, on par with other similar logical and epistemological claims. So I detect a slight bias here. And of course the remark about language being a social affair (how would Quine, a sterile man of no social graces, know?), which Wittgenstein presumably also underscored in his denial of private language (for very different reasons though) is, I am afraid to say, not much developed here or elsewhere in Matilal's work.

Nevertheless, we get a second clue a few lines down when we turn to the Nyaya view proper. Nyaya apparently disputed the relevance of this residual sensory core, as this is neither error, nor jnana nor even foundation for knowledge. Again I quote, 'Later on in the same realist tradition, even its cognitive role was called into question. For there was a prevalent theory of language (ascribed to Bhartrhari) which maintained that language and thought (any cognitive experiences) are inextricably intertwined such that language anchors thought and vice versa. Being influenced by this theory [I really do wonder about this claim. anyway], the Nyaya realist claimed that any cognitive experience must be expressible in language. The ineffable sensory core, if there is any, should be little more than the physiological surface irritation, and hence non-cognitive. Besides, by its very nature, it would not be even introspectable or amenable to inner perception. Introspectable states have a judgemental structure' (pp 4-5 SINR).

And so we are back to cognitions once again, although a more detailed concern with attributed predicate property, its prior knowledge, copula relations with univerals (of cowhood, e.g. in the judgement 'x is a cow') follows which need not detain us presently. What Matilal wants to stress is that Nyaya categorically denies the role, in any significant epistemological analysis, of the core sensory data: it is found to be almost superfluous, as structured thoughts or judgements are not (therefore) interpretations or organisation of these raw data, for we have no conscious verbal access to such data. (loc cit)

What is being said? It seems to me the following: the raw sensory core might well be there, we do not know as we have no linguistic access to it; it's perhaps only a 'surface physiological irritation' or physiogenic motions like our feelings and other sensations (indeed e-motions and like e-mail messages not terribly cognitive or intellectual in worth either). Language only comes into play or into the foreground when judgements are being formed or about to be formed: this is the structured perception; and language should be banished from having concerns with the pre-linguistic, uninterpreted and unstructured or pre-structured data. We should pause to note while the Nyaya theory of cognition admits an indeterinate phase, namely, nirvikalpa prior to the arising of the determinate judgement, which is savikalpa, the pre-lingustic, pre-structured inchoate state which Matilal is denying is not to be identified with nirvikalpa phase. For the nirvikalpajnana already is on its way to be being structured, or one might say that it registers the logical pre-condition for the determinate jnana. The difference is in degree rather than in kind (as Matilal once put it to me).

On this strong point of the rejection of the ineffable etc base, which Matilal attributes to Gangesa, he adverts to Donald Davidson's explosion of the myth of the core and inexplicable given which otherwise leads to a kind of conceptual relativism. So so he avers, "I find a support to Davidson's conclusion in Gangesa's clever move to direct our attention to the logical prerequisite of a propositional knowledge or judgement and omit our talk of the unstructured data or the pure pre-linguistic content of our perceptual experience'. At core, what Matilal actually wants to endorse is a version of metaphysical or external realism that comes through as an implication of Davidson's doctrine and which he believes is upheld by the Nyaya-Vaisesika philosophers, who were apparently in a head-on collision course with the 'idealists' and 'relativists' in India (pp. 1 & 8): It is significant he notes, 'that their mistrust in the availability of the pure, structureless data as well as their outright rejection of any form of representation, of intentional content or propositional entities intervening between us and the world seems not only compatible with but also conducive to their brand of realism (if we can omit some minor details).' This is a telling endorsement even as it underscores a motivation on Matilal's part. We must, at all cost, reject so-called idealist and relativist versions of the theory which rely precisely on language's interventionist role and pander to paradigms of representation or of intentionality, and so on. Before I say anything more on the preferred paradigm or context-independent schema, let me refer to a thought experiment which might frame the distinction between schema and context alluded to in this realist critique. This comes from Hilary Putnam and it is highly instructive as it has been provocative (though I will modify it a bit). Imagine there are two identical lokas or twin earths; the respective inhabitants of both lokas behave in exactly indentical ways towards a certain liquid substance which each side consumes when thirsty and use it in their cooking, and so on. Moreover, both the loka-dwellers call it 'water' in their respective (natural) languages. On earth X their scientist tell them that it is H₂O. Now is there any way in which we can say that on earth B their own scientist will also tell them that what they call and use as 'water' is actually is H₂O? Putnam says emphatically, 'No', for we have no access other than through what we are told from within the schema of each earth-community what the substance is: even if earth B scientists decide to call the liquid H₂O, and their propositional knowledge attests to this objective characterization, there is no gainsaying that it is exactly what is intended by use of the term H₂O in earth A and vice versa. In other words, both make judgements that look, on the face of it, to be identical, but there is no way in which we can say that each side shares the knowledge and therefore the linguistic contents involved in exactly identical terms. It so happens that their conventions or common-sense discourse coincide, but this gives no warrant to suppose that they are perceiving and are therefore speaking about the same reality qua reality. Of course Putman uses this imaginary example to bolster his own theory of internal or what he now calls common-sense realism which stops external or metaphysical realism on its tracks.

Now I am coming to my main argument: the central point to be made is not that, or not simply that, on the one hand there are no ineffable unstructured percepts and on the other hand that language anchors thought (or that all our cognitions are shot through with language), but to acknowledge that language anchors or might just well spread over and encompass the unstructured qua pre-structured content of our experience as well. I think Mark Siderits has made this plea more forcefuly in his book on Indian Philosophy of Language. To put it more cynically: Why privilege language's anchor only in respect of cognitive or structured experiences, and deny its trekking all the way down to the sensory core, or onto other kinds of sensations? This latter denial, when forced, leads ultimately to the denial of the pre-structured core as well, which is an ontological claim and not therefore the business of epistemological or linguistic analysis - a fault too readily committed by the Naiyayikas, despite cautionary reminders from the Buddhist quarters.

Returning to the question of the preferred paradigm, I want to suggest a reason for why I think the realists in India opted for this rather more simple-minded characterization of the role of language in relation to our awareness of things. Here I turn to the Nyaya theory of Perception, which again Matilal has best articulated for us. Details aside, we know that the theory moved from Gautama's cruder version of sense-organ contact with the object (indriyasannikarsa) to Gangesa's more elegant view of the directness of the presentation of the object to awareness (neither any sort of contact nor first-person ascertainment is deemed necessary). The mind directly 'touches' the object. A Lockean type of representationalism, Berkeleyean idealism, Humean sensa-data empiricism, Brentanoian 'intentional content' phenomenologism, are all set aside, and it would have no bar either with any kind of phenomenalism, constructivism or constructionalism. An F-attribute is all that one has in the initial grasping (i.e. n the nirvikalpa phase), which is absolutely unmediated and gets promptly completed when the F-attribute (which may be a simple universal or an unanalysable property) as the qualifier is conjoined with the qualificandum, x, the nebulous haze that floats into the awareness, whence immediately arises the cognition, G = Fx. (The relation that causally connects F and x, as aRb, are also said to be real and tacit in the awareness.) Such a theory of perception indeed is also most conducive to the doctrine of realism being harnassed, albeit direct or metaphysical realism. Now for those more committed to the economy of this doctrine, nothing should be allowed or supposed to stand in the way of the directness of the perceptual graps, the 'touching' of reality — language, i.e. words and concepts included. In other words, F-attribute neither presupposes a cognition of the thing of which it is a cognition (which is the preserve of the vaguely floating x or 'this...'); nor does the x presuppose a core sensory data to which it ascribes the attributes or features, for all such presuppositions would either lead to a paradox or be prolix and therefore are unwarranted.

To be sure the F-attributes are linguistic in a sense or linguistically-imbued, for otherwise it would make no sense to speak of universals, such as cowness, colour-sensations such as red patches, or other properties such as — blobs of coldness, or — very hotting, and so on, as featuring in the awareness. Qualifiers are property-instantiations in predicate form and as such they function is one of characterization. This must peforce involve them in quasi-linguistic behaviour, as it were. This admission, which Naiyayikas have no difficulty in making, is probably what Matilal counts as the modified adoption of the Bhartrharian thesis, according to which it is sensations that penetrate the linguistic level (not that the linguistic tropes penetrate the sensory level, as the Buddhists might have it). But the primacy of the process, we should note, is never given to the linguistic, it is given rather to the cognitive (recall Matilal's opening sentence earlier), and only secondarily to the linguistic. Hence the allusions to the Bhartrharian influence is a bit of a red-herring or a misnomer, for the Naiyayikas are no metaphysical bhasavadins (proper namesakes occur without the calling of the mother of necessity). Still, the assertion about all thought or judgement being expressible in words is needed in the Nyaya view so as to "avoid the reduction of the public concepts finally to the inexpressible phenomenal criteria. They are identified in terms of causal relations between agents and the world (there being nothing in between on this view)". But how much of the linguistic criteria has the Nyaya allowed to penetrate the cognitive process? This question is pertinent, for what is grasped or perceived is not a bit of language but an object qua object which is descriptively marked as possessing such and such a feature. Big deal, as the Americans might say.

Let me summarise the analysis and argument thus far. The Nyaya position is, as Matilal has it, that sensations have no core unstructured component: all cognitions worth their Howrah-salt arise structured; linguistic elements attach themselves like yokes, not ploughs, to the farming oxon, and their causal connection with the things of which they are cognitions is indubitably direct. This approach guarantees the conclusions of metaphysical realism. Or perhaps metaphysical (i.e. external) realism dictates just such a linguistically-compromised doctrine of knowing. Matilal says as much, thus, "All these, the Naiyayikas asserted, were needed to remain true to their view of metaphysical realism". But whether they are necessary is of course another, and indeed the moot question; but this remains to be shown; though I believe Sibajiban Bhattacharyya with Stephen Phillips in their work on Gangesa have been making good progress on this very issue.

As to language per se or sabda, which as a whole is another pramana since we obtain knowledge from using linguistic utterances, it too must operate in more or less the same way and conform to the same criteria or conditions of external, causal connectivity, otherwise the knowledge it delivers us will be much compromised or contaminated (not least by its own excesses). Worse still, if especially the tower of babel or babblers is admitted into the fray, we'd risk conceptual relativism; but most of all, if we are to remain committed to true blue (i.e. anticipating Davidson and Dummett) metaphysical realism, then we must find a way of talking about language that sustains this doctrine optimally. How can we achieve that? Well, there is one simple route: viz. use the model of sensory perception as the paradigm for linguistic knowledge, with all the devices, criterial requirements and descriptive categories opened up for us in the theory of propositional knowledge discussed thus far. Seek to ground this doctrine of word and knowledge in the 'direct grasp' modality that Gangesa has insisted upon for sensory perception; do not allow concepts (vikalpa) thrown up randomly or unconsciously by the mind, through karmic traces, or love's desires, or private reflective tropes, or other discursive items outside of the shared public language, etc. to interfer with and mediate the otherwise immaculate continuum between the word and the world. — This is not unlike the Kantian strictures on the categories of understanding which were to be strictly rational, i.e. universal, in form and were not to be mixed up with relative projections, artifacts of imagination, human intentionalities and so on. — Language, despite the temptations coming from the likes of Candrakirti and more recently echoed in the post-Sausurean semiological movement led by Rolland Barthes in France and Derrida more famously in North America, is not to be a regarded as a self-contained systems of signs which make reference only to internally-bifurcated or constructed signifieds, or that it is only by a wild stretch of our imagination that any real-term reference or what Frege called Bedeutung is acheived: otherise it is all Sinn.

If the latter were the case, we would not be able to distinguish between true and false sentences: the F-attributes must and do necessarily feature in the object of which it is a verbally-derived cognition for linguistic judgements that are true. That is how we know from words.

It is then little wonder that Arindam Chakravorty, following in the same tradition, would question whether it was even possible for someone to understand a patently false sentence, such as "Odourless green cheese utters Martian music", for one would be hard-pressed to make a logically coherent connection of Fx to G. This predeliction for a realist thesis of language and the epistemic moves undergirding it is further borne out in K C Bhattacharyya's wonderful description of how sabda functions in the context of a karana: the word, he says, reaches out to the thing, it touches the thing. (Professor Mohanty and I have commented in some detail on K C Bhattacharyya's immensely helpful insight, but again here I wish to point to the unargued but presupposed bias towards a realist schema which informs this insight.) Direct realism can only produce or issue in a theory of the direct relation between words and the world. Bhartrhari may well have believed that the world is linguistically given to us; but the Naiyayikas for their part, including their modern-day defenders it would be true to say, believe not that the world as a whole but only the public space and one-half-of-the-cognitive act (that too the predicate side) is linguistically riven. All else is thankfully free from the ravages and relativistic tendencies of a more general course of linguistics.

Why not, however, go all the way and take seriously the Buddhist dialectical argument that even this partial linguisticality, as conceded in the Nyaya cognitive theory, is an imaginative construction, that indeed language is a figment of our desires, intentionalities, which cover up the absence of real causal connections. In other words, we are far from having convincingly established a necessary causality in respect of the x and F-attributes which float into and feature respectively in our aawreness (as prakarata and visesana) Or, alternatively, consider that as a shared public or collective discourse, language is not devoid of its own relativities, constructions, self-referentialities, community or tradition-transmitted intentionalities and so on, that get in the way of authentic grasping of the world, which can only be 'wordless'.