METAPHOR, ONTOLOGY, AND SCIENTIFIC TRUTH: Against Some Dogmas of the New Anti-Realism (link)
by Christopher Norris (link)

Introduction

Anti-realism is currently the prevailing trend across many schools of thought in epistemology and philosophy of science. There are, to be sure, some strong countervailing voices and some well-developed arguments in support of an alternative (ontological-realist or causal-explanatory) approach (1). But these latter have enjoyed nothing like the same degree of acceptance among workers in other disciplines - history of science, sociology of knowledge, cultural studies, etc. - where anti-realism is nowadays the orthodox line. Nor is it hard to understand why this should be the case. For those disciplines clearly have a large investment in the idea of scientific 'truth' or 'reality' as relative to - or constructed within - some culture-specific discourse, framework of enquiry, historical paradigm, conceptual scheme, or whatever (2). Hence the rapid diffusion of arguments from recent (post-Kuhnian) philosophy of science which are taken as lending powerful support to the anti-realist case. Other sources include W.V. Quine (on ontological relativity, meaning-variance, and the underdetermination of theories by evidence); late Wittgenstein (on language-games and cultural 'forms of life'); Heidegger, Gadamer, and other proponents of a depth-hermeneutical approach; Foucault's relentlessly sceptical 'genealogies' of power/knowledge; postmodernists such as Lyotard with their talk of paralogism, narrative pragmatics, and 'performativity' as the sole criterion of scientific truth; and the 'strong programme' in sociology of knowledge with its declared principle of according equal treatment to all scientific theories, whether true or false as judged by our present-day cultural lights (3). What they all have in common is the turn toward language - or some version of the socio-discursive constructivist argument - deployed as a counter to realist claims of whatever variety.

In so far as these arguments have found support from within the Anglo-American philosophical community it has come mostly from 'post-analytical' thinkers - Richard Rorty prominent among them - who seek to demote science (and philosophy of science) from its erstwhile position of high prestige (4). Of course there are others like Michael Dummett who continue the analytic project (broadly defined) but who none the less espouse an anti-realist position according to which we can have no grasp of verification-transcendent truth values (5). On this view - in short - it is strictly unintelligible that our present best notions of truth, method, observational warrant, theoretical adequacy and so forth might not correspond to the way things stand 'in reality'. That is to say, we could have no possible grounds for suspecting this to be the case given the fact that any reasons adduced would always be reliant on criteria derived from those same (for us truth-constitutive) standards of evidential reasoning. It is mainly under pressure from arguments of this sort that philosophers like Hilary Putnam have retreated from a strong realist position - such as Putnam took during the early 1970s - to a stance of (so-called) 'internal realism' which concedes most of the adversary case while hoping to avoid its more extreme relativist implications (6).

Still it is clear that the 'linguistic turn' in its various forms has done more to promote anti-realist than realist approaches to philosophy of science. That is to say, it has most often been enlisted in support of conventionalist, instrumentalist, or cultural-relativist doctrines, all of which manifest an elective affinity with the idea of truth as a discursive construct devoid of any real-world ontological grounding7. In what follows I shall take a rather oblique route by way of addressing these issues. My main concern is with Derrida's essay 'White Mythology: Metaphor in the Text of Philosophy', a work that is often read - mistakenly I think - as adopting an extreme anti-realist (or 'textualist') position with regard to philosophical truth-claims (8). Of particular interest in the present context are Derrida's lengthy and detailed discussions of the role of metaphor in scientific concept-formation and its bearing on matters of ontology and epistemology as raised by philosophers of science from Aristotle to Bachelard and Canguilhem.

Derrida and Benveniste on Aristotle

Let me give more substance to this generalized claim by quoting a further passage from 'White Mythology', this time at a length adequate to convey the complexity of Derrida's argument. The passage has to do with Aristotle's account of metaphor and its place within what Derrida calls 'the great immobile chain of Aristotelian ontology, with its theory of the analogy of Being, its logic, its epistemology, and more precisely its poetics and its rhetoric'. (p. 236) It is also concerned with Aristotle's attempt to distinguish human from non-human (animal) modes of being on the basis of a theory of language - a philosophical grammar - that would define man as a speaking-and-reasoning creature as opposed to a mere producer of pre-articulate, meaningless, unintelligible sounds. Hence the importance of the letter (stoikheion), the minimal distinctive element that possesses no meaning - no semantic content - in and of itself but whose role it is, within Aristotle's system, to create the possibility of articulate utterance by accomplishing the passage from 'sound without signification' (phone asemos) to meaningful speech-production (phone semantike). From this point, so Aristotle argues, one can go on to derive the entire inventory of human linguistic resources - from syllables, via nouns and verbs, to the highest, most complex levels of logico-grammatical structure - which set human beings decisively apart from the rest of the animal creation (9). Yet there is also another distinctive human attribute, one that enables us to acquire knowledge, to 'perceive resemblances', to apply words to objects in habitual or unaccustomed ways, or to reason analogically from one context of knowledge, usage or experience to another. Mimesis and metaphor are the two chief terms that Aristotle uses in connection with this range of capacities. And since they occupy such a crucial place in his system - since without them that system would fall apart - therefore it is wrong (a misreading of Aristotle) to assign them a subordinate or derivative role.

Here is the relevant passage from 'White Mythology', omitting those extended citations in the original Greek which Derrida is careful to provide but which are not (I think) necessary here:

This is the difference between animals and man: according to Aristotle both can emit indivisible sounds, but only man can make of them a letter... Aristotle does not analyze this difference; he interprets it by teleological retrospection. No internal characteristic distinguishes the atom of animal sound and the letter. Thus, it is only on the basis of the signifying phonic composition, on the basis of meaning and reference, that the human voice should be distinguished from the call of an animal. Meaning and reference: that is, the possibility of signifying by means of a noun. What is proper to nouns is to signify something, an independent being identical to itself, conceived as such. It is at this point that the theory of the name, such as it is implied by the concept of metaphor, is articulated with ontology. Aside from the classical and dogmatically affirmed limit between the animal without logos and man as zoon logon ekhon, what appears here is a certain systematic indissociability of the value of metaphor and the metaphysical chain holding together the values of discourse, voice, noun, signification, meaning, imitative representation, resemblance; or, in order to reduce what these translations import or deport, the values of logos, phone semantike, semainein, onoma, mimesis, homoiosis... Mimesis is never without the theoretical perception of resemblance or similarity, that is, of that which will always be posited as the condition for metaphor. Homoiosis is not only constitutive of the value of truth (aletheia) which governs the entire chain; it is that without which the metaphorical operation is impossible: 'To produce a good metaphor is to see a likeness'. The condition for metaphor (for good and true metaphor) is the condition for truth. ('White Mythology', pp. 236-237)

Again this is couched in Derrida's mixed-mode style of direct or oblique citation from the source-text (Aristotle) combined with analytic commentary and critical exegesis. Moreover, as the context makes clear, it is offered by way of an illustrative statement of precisely those 'classical' presuppositions which have governed the philosophic discourse on metaphor and - beyond that - the predominant (post-Aristotelian) way of thinking about issues in ontology, epistemology, philosophical semantics, and kindred disciplines. So there can be no question of reading the passage as an affirmation of Aristotle's views or even as signalling partial assent to any link in the 'great chain' of Aristotelian argument.

Yet it is equally unjustified - here as elsewhere in Derrida's work - to suppose that a deconstructive reading is a priori committed to the disarticulation of all truth-claims and the undoing of any theory (such as Aristotle's) predicated on values of truth, reason, logical form, conceptual adequacy, empirical warrant, and so forth. For this is once again to mistake the whole purpose and argumentative tenor of an essay like 'White Mythology': namely, its critical questioning of such values in a manner that itself maintains the highest standards of analytical consistency and rigour while not taking anything for granted in the way that those standards have hitherto or traditionally been applied. In other words deconstruction carries on the critique of established (commonsense, naturalised, or consensus-based) modes of perception or conceptualization which has characterized philosophy in the tradition from Aristotle to Descartes, Kant, and Husserl (10). No doubt it does so through a method of analysis - the rhetorical close-reading of various cardinal texts in that tradition - which departs very markedly from other, more conventional ideas of what constitutes a proper philosophical critique. No doubt it raises issues - about truth, representation, the extent to which metaphors can be 'adequately' conceptualized or intuitions brought under 'adequate' concepts - that have provoked consternation (or outright dismissal) among many philosophers. Yet the above passage should at least give pause to anyone who is tempted to regard 'White Mythology' as a mere exercise in 'textualist' mystification or an argument devoted to such simplified (pseudo-deconstructive) slogans as that 'all concepts are metaphors', 'reality just a fictive or rhetorical construct', or 'truth just a product of the will-to-power vested in figural language'.

Thus Derrida is not for one moment suggesting that just because Aristotle has recourse to metaphor - or to metaphor-related notions like resemblance, mimesis, the 'perception of similarity', etc. - at crucial points in his argument, therefore his entire ontology and epistemology (along with his logic, metaphysics, and conception of enquiry in the natural or physical sciences) comes down to nothing more than a series of figural tropes and substitutions, indifferent with regard to their truth-content or capacity for conceptual elucidation and critique. Nor is he committed to the absurd view that truth and reality just are what we make of them according to this or that favoured rhetoric, language-game, discourse, vocabulary, or whatever. Such a doctrine could be extracted [from] 'White Mythology' only by ignoring those many passages - among them the sections on Aristotle, Canguilhem, and Bachelard - that offer a precise and detailed account of the critical epistemology of metaphor and its role in the process of scientific knowledge-production. In short: '[m]etaphor, as an effect of mimesis and homoiosis, the manifestation of analogy, will be a means of knowledge, a means that is subordinate, but certain.' (p. 238)

This is not to deny - what is in any case quite evident - that Derrida is here paraphrasing Aristotle and drawing attention to the way in which metaphor has traditionally been treated as 'subordinate' to other (more directly reliable) means of acquiring knowledge. Thus: 'metaphor... is determined by philosophy as a provisional loss of meaning, an economy of the proper without irreparable damage, a certainly inevitable detour, but also a history with its sights set on, and within the horizon of, the circular reappropriation of literal, proper meaning'. (p. 270) To this extent Derrida is describing - and calling into question - an entire set of axioms (along with an implicit teleology) aimed toward the 'proper' understanding of metaphor as a detour on the path to adequate conceptual knowledge. This is why, as he writes, 'the philosophical evaluation of metaphor has always been ambiguous: metaphor is dangerous and foreign as concerns intuition (vision or contact), concept (the grasping or proper presence of the signified), and consciousness (proximity or self-presence); but it is in complicity with what it endangers, is necessary to it in the extent to which the de-tour is a re-turn guided by the function of resemblance (mimesis or homoiosis), under the law of the same.' (p. 270)

But again we shall mistake Derrida's purpose - or (not to beg the intentionalist question) the logic of his argument in 'White Mythology' - if we read such passages as opening the way to a wholesale metaphorization of philosophy, or a levelling of the metaphor/concept distinction that would view it as merely a symptomatic instance of this drive for the 'reappropriation' of metaphor by the philosophic will-to-truth. For there could then be no accounting for that other (often strongly counter-intuitive) process of conceptual 'rectification' that enables scientific metaphors, models, and analogies to bring about genuine advances in our knowledge of physical objects, processes, and events. Empson makes the point more succinctly when he remarks - in a review of E.A. Burtt's book The Metaphysical Foundations of Modern Science - that 'it is unsafe to explain discovery in terms of a man's intellectual preconceptions, because the act of discovery is precisely that of stepping outside preconceptions' (11).

One could go into a lot more detail with respect to Derrida's position on these issues of ontology, epistemology, and the logic of scientific enquiry. A crucial text here would be his essay 'The Supplement of Copula' where he argues - as against the linguist Emile Benveniste - that Aristotle's table of the categories (the various forms and modalities of predicative judgment) cannot be treated as mere products of a particular language, namely the ancient Greek, whose lexical and grammatical resources they erect into an absolute (quasi-transcendental) set of laws for the conduct of rational thought (12). Of course there is a very real question - much debated by philosophers, Kant among them - as to whether Aristotle's really was, as he thought, a complete and exhaustive (a priori deducible) listing of the categories concerned (13). Also there are problems, as scarcely needs remarking, with Aristotle's essentialist definition of 'substance' as that to which the categories apply but which cannot itself be qualified or modified in its essence by any such categorial predicates. However Derrida's argument has to do with Benveniste's more sweeping (and strictly unintelligible) claim that the very idea of 'categoriality' is one that could only arise within the context of a given natural language that provided the lexico-grammatical means for its expression. For this involves Benveniste in a confusion of terms, a failure to observe the crucial distinction between particular categories (which may indeed be language-dependent) and categoriality as the precondition for making any judgments whatsoever, including judgments with regard to the priority of 'language' over 'thought' or - as the issue presents itself here - linguistics over philosophy. The latter is a transcendental question, taking the term 'transcendental' (as Derrida specifies) 'in its most rigorous accepted sense, in its most avowed "technicalness", precisely as it was fixed in the course of the development of the Aristotelian problematic of the categories' (14). In this sense of the term, quite simply, 'transcendental means transcategorial', i.e. pertaining to the condition of possibility for thought and judgment in general. It literally signifies 'that which transcends every genre', every particular (as it might be linguistically-instantiated) mode of categorical predication.

Thus 'none of the concepts utilized by Benveniste could have seen day, including that of linguistics as a science and the very notion of language, without a certain "small document" on the categories [i.e. Aristotle's table]'. ('Supplement of Copula', p. 188) And again, more pointedly:

Philosophy is not only before linguistics as one might find oneself facing a new science, a new way of seeing or a new object; it is also before linguistics, preceding linguistics by virtue of all the concepts philosophy still provides it, for better or worse; and it sometimes intervenes in the most critical, and occasionally in the most dogmatic, least scientific, operations of the linguist. (p. 188)

I must refer the reader to Derrida's essay if he or she wishes to follow this argument in all its intricate and rewarding detail. Sufficient to say that it operates both through the mode of transcendental-deductive reasoning - as defined above - and through a critical exegesis of Benveniste's text alert to those items of empirical (i.e. natural-language) evidence that contradict his avowed thesis. Thus, as Derrida remarks, '[w]hat is not examined at any time is the common category of the category, the categoriality in general on the basis of which the categories of language and the categories of thought may be dissociated'. (p. 182) And as a matter of empirical evidence there is the case of Ewe - a language spoken in Togo - which according to Benveniste possesses no equivalent (no lexical equivalent) of the verb 'to be' in its jointly existential and copulative function, but which turns out, on Benveniste's own submission, to require that those resources be ascribed to it (in whatever lexico-grammatically distributed form) if Ewe is to make any sense in its various social and communicative contexts. Thus the transcendental argument receives confirmation at the level of empirical research, that is, through Benveniste's reflection on the evidence of how Ewe speakers actually communicate as distinct from his preconceived ideas about linguistic or ontological relativity.

There is another passage from Benveniste - cited at length by Derrida - which brings out the close relationship between these issues in the fields of linguistics, anthropology, philosophical semantics, epistemology, and philosophy of science. 'Surely it is not by chance', Benveniste suggests, 'that modern epistemology does not try to set up a table of categories. It is more productive to conceive of the mind as a virtuality than as a framework, as a dynamism than as a structure. It is a fact that, to satisfy the requirements of scientific methods, thought everywhere adopts the same procedures in whatever language it chooses to describe experience. In this sense it becomes independent, not of language, but of particular linguistic structures. Chinese thought may well have invented categories as specific as tao, the yin and the yang; it is nonetheless able to assimilate the concepts of dialectical materialism or quantum mechanics without the structure of the Chinese language proving a hindrance. No type of language can by itself alone foster or hamper the activity of the mind. The advance of thought is linked much more closely to the capacities of man, to general conditions of culture, and to the organization of society than to the particular nature of a language.' (15)

It is a remarkable passage for several reasons, not least its espousal of an extreme dualism between 'language' and 'thought', its casual (as if unnoticed) throwing-away of the linguistic-relativist thesis, and its distinction - equally fatal to Benveniste's arguments elsewhere - between language in general and 'particular languages'. What emerges through all these manifest contradictions is the acknowledgment that there must be some order of reality that thought can apprehend or that language (language-in-general) can articulate quite aside from all mere relativities of time and place. Thus the 'advance of thought' seems a process that can indeed occur - whether in quantum physics or Chinese philosophy - through a process of conceptual development that cannot be attributed to the formative influence of this or that 'particular language'. Rather it belongs to the 'capacities of man' as a knowledge-acquisitive agent, with perhaps some allowance - and here Benveniste slides back toward a relativist stance - for 'general conditions of culture' or the 'organization of society'. But these latter conditions are apparently conceived as belonging to an order more permanent - or at any rate of far longer duration - than anything on the scale (historic or geographical) that Benveniste associates with 'particular' localized languages or cultures. In short, the whole passage tends towards a transcategorial conception of thought, language, reason and truth which Benveniste cannot bring himself to endorse explicitly - given his linguistic-relativist credentials - but which comes through in his argument despite and against its overt professions of belief.

Quantum mechanics has become a standard topos in current debates about ontological relativity and the issue of linguistic representation (16). It has spawned a great range of philosophical positions - realist and anti-realist - with regard to the status of quantum phenomena and their implications for philosophy of science, epistemology, and interpretation-theory. Benveniste makes only passing reference to this debate in the above-cited passage. However it does lend support to the view that some degree of ontological realism is presupposed in any discussion of quantum mechanics that seeks an 'advance of thought' through the elaboration and testing of specific conjectures. This applies even to highly speculative thought-experiments - such as the famous series conducted by Einstein and Bohr - for which, as yet, there existed no means of observational or experimental proof (17). For those experiments would quite simply have lacked all probative force had they not presupposed certain realist postulates concerning - for instance - the space-time trajectory of photons or electrons, the effects of particle charge, the well-defined limits placed upon simultaneous measurement of location and momentum, the behaviour of waves and/or particles under certain specified conditions, etc. (18) Nor is this argument in any way refuted by their having given rise to heterodox ideas (complementarity, undecidability, Heisenberg's uncertainty-principle) that on the face of it would appear incompatible with a realist interpretation. For here again it is the case that these theories were arrived at only in response to certain deep-laid conceptual problems, problems that would not have arisen - or required such strongly counter-intuitive 'solutions' - except on the premise (the ontological-realist premise) that they captured something that was deeply and genuinely puzzling in the quantum-physical domain (19).

I should not wish to place too large a philosophical burden on Benveniste's brief reference to quantum mechanics or on Derrida's citation of it in support of his case - as against Benveniste - for the impossibility of relativizing truth to language. Still it is a passage of some significance in the present context of argument. This emerges more clearly if one considers Benveniste's suggestion that 'modern epistemology' has no need of anything like Aristotle's table of the categories since 'it is more productive to conceive of the mind as a virtuality than as a framework, as a dynamism than as a structure' (20). There is an echo here of Whorfian ethnolinguistics, more specifically of Whorf's much-debated claim that the ideas of relativity-theory or quantum mechanics could be better expressed in Hopi Indian than in any of the modern European (Greek-influenced) languages since Hopi manifested a different metaphysics, a world-view unencumbered by the subject/object dualism or the rigid categorical framework of predicative grammar and logic (21). What is curious about this claim, as many commentators have noted, is the fact that these theories were expounded and developed not - as it happens - in Hopi but in a range of other (mainly European) languages that on Whorf's account should have put up maximal resistance to their adequate expression.

All of which suggests that Davidson is right: that there is no making sense of the argument for ontological relativity if that argument is pushed to the point of denying the very possibilty of adequate translation from one language to another (22). What is involved here (once again) is a kind of thought-experiment, in this case an experiment with the idea of 'radical translation' conducted - as by Whorf or by Quine - with a view to establishing the incommensurability of diverse languages, conceptual schemes, ontological frameworks, etc. (23) But the result is rather to prove just the opposite: that any evidence adduced in support of such claims (whether ethnolinguistic evidence like Whorf's or hypothetico-deductive 'evidence' like Quine's) will always presuppose the possibility of comparing languages and hence, a fortiori, of translating between them. Thus, as Davidson remarks, Whorf is here attempting to have it both ways, on the one hand declaring that Hopi cannot be 'callibrated' with English, while on the other presuming to describe in English those various lexical and grammatical features of Hopi that supposedly render such description impossible.

My point in all this is that thought experiments may have various (positive and negative) kinds of result. In some cases - like the Einstein/Bohr debates and subsequent quantum-physical conjectures - they serve as a means of formulating and testing theories which cannot at present be physically verified but which none the less require that their terms be taken as referring to certain entities, processes, or events whose behaviour under given conditions is (to put it simply) what the experiment is all about. In other cases - as with Quine, Whorf, and Benveniste - what begins as an argument against ontological realism (and in support of the linguistic-relativist case) ends up by undermining its own thesis and thus showing such ideas to be strictly unintelligible. Nor should we be over-impressed by the fact that quantum mechanics has so frequently figured as a paradigm instance of ontological relativity in the thinking of Quine and others (24). For this is to ignore what emerges very clearly in detailed accounts of those original thought-experiments; that is, the extent to which they all necessarily relied on a realist understanding of quantum phenomena even if the results turned out to require some drastic modification to accepted ideas about the ontology of the microphysical domain, the limits of precise measurement, the wave/particle duality, or the distinction between observing 'subject' and observed 'object'. In this respect the Einstein/Bohr conjectures were on a par, ontologically speaking, with such previous classic thought-experiments as that by which Galileo proved the uniform (mass-independent) rate of gravity-induced acceleration for bodies in a state of free fall. He imagined the case of two such bodies, a cannon-ball and a musket ball, securely fastened together. 'Go figure!', as they say; the experiment is enough to demonstrate conclusively, as a matter of conceptual necessity, what Galileo would later put to the proof by his better-known series of empirical tests at the leaning tower of Pisa (25).

References

(White Mythology - J. Derrida)