4. The subject-predicate relation.
Nagarjuna’s assault on reason includes an attempted critique of verbal expression and the structure of language. For him, words are conventions devoid of deductively absolute or inductively contextual meaning or relationships to each other. That he himself engages in criticism by means of language does not bother him, because he grants that it functions somewhat on a practical level, in a “conventional” way, within ordinary consciousness. His goal is as usual to take us beyond words and the illusions he claims they create, into the higher mode of consciousness that puts us in contact with ultimate reality. His means is to demonstrate that language is illogical and futile, putting forward at least two arguments:
(a) He asks, “is the subject identical with or different from the predicate?” His answer is stated by Cheng as follows. “If the subject is the same as the predicate, they would be one and it would make no sense to call one a subject and the other a predicate… the sentence would be a tautology. If on the other hand, the subject is different from the predicate, there would be no particular connection between them.” In either case, predication is found redundant.
(b) Furthermore, “what is the status of the subject before predication? Does it already have predicates predicated of it or not?” (i.e. predicates “other” than the subject itself). “If a subject is without any predicate predicated of it, it is incomprehensible and non-existent. If a subject without a predicate is non-existent, to what does our predicate apply? If on the other hand, the subject does have some other predicate predicated of it before we ascribe a predicate, what further function would be served by ascribing an additional predicate since it already has something predicated of it? If it needs this predicate, then a second and a third can in principle be applied. This would lead to infinite regress.”
By such arguments, Nagarjuna seeks to give the impression that language is structurally unreliable and a stupid artifice. His arguments are shaped in such a way as to seem logically orderly and exhaustive, i.e. to consider all conceivable alternatives and eliminate them one by one, so that we have no leg left to stand on. He thus apparently uses some of the methodology of logic to convince us. But of course the descriptions of the nature and role of predication underlying his arguments constitute merely one particular view, so that his premises are not in fact exhaustive and only serve to show that his proposed view is faulty and to be rejected.
Thus, consider argument (a). Its first premise about tautology is obvious and trivial, being itself tautological. More important, the second premise is not at all evident. The subject may well be “different from the predicate” and yet have a “particular connection” to it. There is no logical basis for Nagarjuna’s proposed implication; the antecedent concept (“different”) and the consequent concept (“unconnected”) are quite distinct. If X equals Y in all respects, then ‘if X, then Y’ and ‘if Y, then X’ must both be true (though it does not follow that if they are both true, X = Y, since X and Y may well not be simultaneous). X and Y are different, means ‘X does not in all respects equal Y’, and so implies that X and Y are either non-simultaneous, or that ‘if X, then Y’ and/or ‘if Y, then X’ is/are false. Whereas X and Y are unconnected, means that ‘if X, then Y’ and ‘if Y, then X’ must both be false, as any lesser such relations between X and Y. Thus, the former concept is wider than the latter, and does not imply it.
The subject-predicate relation under discussion may and usually is posited as, for instance, a classificatory one – a relation between an individual and a class, or a subclass (species) and an overclass (genus), so that the former is included in the latter without being equal in scope to it. ‘Does not equal’ does not exclude ‘is greater than’ or ‘is smaller than’ or ‘exists before or after’, or any other non-equal relationship. Nagarjuna suggests that if the terms are not identical, they cannot be related by the copula ‘is’ – but this copula was never intended to mean total equation. Nagarjuna cannot change the convention that ‘is’ is different from ‘equal’; or if he insists on doing so and himself practices what he preaches, we can simply invent another word for what we mean by ‘is’.
Since Nagarjuna’s second premise is unwarranted, his attempted dilemma is dissolved.
Now consider argument (b). The first leg mentions a subject “without any predicate” and claims it “incomprehensible and non-existent”, so that eventual predication relative to it is senseless. The second leg therefore suggests that a subject can only have one predicate (if any, see earlier), and that ascribing more of them to it implies in each case that the preceding one did not fulfill its intended function (definition?) so that unending predication would be called for – an impossible task. But these arguments are worthless, because Nagarjuna clearly misrepresents predication; his view of it is a simplistic caricature.
What do we in fact mean by a subject or a predicate? Primarily, an object of consciousness – an individual concrete or an abstract ultimately known through comparisons of such concretes. This does not imply that we consider all existents as objects of consciousness, but only that as of the moment we think of something (as here) we must admit it as appearance and therefore as existent. Moreover, we need not and do not consider consciousness as invariably correct and all its objects as real – we may well conceive of an illusory object, which has no existence other than in the way of appearance. Secondly, this object (be it real or illusory) may be, and indeed has to be, cognized before we can name it and verbally predicate anything of it. Predication, like its terms, is an object of consciousness before it is put into words. Consciousness of terms and propositions about them may be wordless; words are merely useful concretizations of intended objects of cognition. Also, before terms are brought together in a proposition, the objects intended by the terms have to be known (or believed, verbally or not) somewhat; the proposition serves to add to that knowledge of the terms, by observing or hypothesizing a certain relation between them.
Nagarjuna tries to suggest the opposite, that we only know things in the framework of predication (and perhaps, of prior verbalization), and that predication merely elucidates or restates knowledge (or belief) already present in the terms. But we may reply that something can well exist without/before being thought of, and be thought of alone without/before being verbalized; and even if/when named, it remains comprehensible without/before being made the subject of any non-verbal or verbal predications; and furthermore that predications are themselves objects of consciousness and that most of them enrich the meanings of both subject and predicate rather than merely redundantly repeating meanings already in them. Nagarjuna also apparently confuses predication with definition, when he considers that a single predication must suffice. In truth, any number of predicates may be ascribed to a subject; predicates are numerous because they are not tautologies of the subject; every term is a complex with a potential positive or negative relation to every other term. Even definition has no ambition to tell us everything about something, but merely claims to focus on one set of predicates, which seemingly abide invariably and exclusively with the subject; and a definition may turn out to be erroneous.
In conclusion, Nagarjuna’s above arguments prove nothing but the incoherence of the particular view of discourse he presents, and do not succeed in invalidating all discourse. The superficial form of his arguments is usually logical enough. But it is not enough to give logical form to our rhetoric, i.e. that the conclusion follows from the premises – the premises themselves have to be first be found obvious or reasonable. It is the premises of Nagarjuna’s arguments that I above contest as naïve and misleading; and my conclusion is merely that his conclusion is not convincingly established. The theory of predication and underlying processes that I rebut his theory with may not answer all questions about these issues, but it is certainly more thought-out and closer to the truth.
To the objection that his use of language to communicate his ideas and arguments implies an assumption (which he denies) that language contains knowledge of some reality, Nagarjuna replies that language is “conventional”. This vague accusation of divorce from all reality has little content, so long as it leaves unexplained just how – in convincing detail – such convention functions otherwise (for language evidently does function, as his using it admits). We can also point out that although words are in principle mere conventions, it does not follow that knowledge is “conventional”.
First because that proposition, as a factual assertion, claims to know something beyond convention about knowledge; and as regards content, it claims the impossibility of any non-conventional linguistic knowledge (including, presumably, the knowledge the proposition itself imparts); whence, to assert that linguistic knowledge is conventional is self-contradictory. Secondly, all conventions imply factual knowledge: you have to know that there is a convention and what that convention is supposed to be and how to apply it correctly! You cannot have a convention about a convention… ad infinitum – it has to stop somewhere factual. Knowledge of conventions is also knowledge; a convention, too, is a reality in itself. It cannot float on an infinity of empty conventions, it has to finally be anchored on some real appearance.
Thirdly, because the conventionality of words is misunderstood. Affixing a label on something, arbitrarily or by agreement, does not imply that the ‘something’ concerned need not be previously known. We can be aware of things, and even think about them, without words. Words merely help us record rational products; giving us a relatively tangible instrument to manipulate. The value of words is not in making conceptual and logical thought possible, but only in making it easier (facilitating memory, classification, communication). Convention is therefore a secondary aspect of words; what counts is their meaning. A language composed only of meaningless words, each entirely defined by others, would have to be infinite in size, and would anyway communicate nothing outside itself. If the language is finite, like ours, it is bound to be based on some undefined prime words, and thus (since content is only verbal, here) be devoid of content, incommunicado. It could not even communicate its own conventions.
Thus, Nagarjuna’s dismissal of language as such is an incoherent thesis, which upon closer scrutiny proves inconsistent with itself.
 See Cheng, pp. 117-118. He there refers to MT V:1-5, and TGT V:1
He there refers to MT V:1-5, and TGT V:1 and VI:1.
 A view reminiscent of Kant’s and other Western philosophers’, incidentally.
 See Appendix 1: fallacies D and A.
 By ‘concrete’ I mean an experienced or perceived object, a phenomenon. By ‘abstract’, an object of reasoning or conception. A third class of object I do not mention here (so as not to complicate the issues) – objects of self-knowledge or ‘intuitions’; suffices in the present context to say that, in relation to abstracts, they have the same position or role as concretes (namely, given data).
 See Appendix 1: fallacy G.
 See Appendix 1: fallacy D.