‘The limits of my language mean the limits of my world’, as Wittgenstein notoriously puts it in the Tractatus.1 My interest in this paper is not so much to discuss that assertion itself, as to wonder what to make of the notion of language having limits at all. What sense can we make of the expression ‘limits of language’? And this question might itself divide into at least two sub-questions, namely: (1) what kind of limits does language have (assuming it has any)? and (2) does language even have limits?
It might seem at first as though language must have limits, and that those limits must be where it abuts onto ‘the world’. Language ‘over here’, or so the thought would go, is separated from, but in some relation to, the world ‘over there’. That relation would be essentially, if we are to follow the tradition, the relation of reference, or at least possible reference. On this view the world, which intuitively existed long before language, sits there waiting, as it were, for language to arrive on the scene and to refer to it. In referring (or at least in successfully referring), language both recognises its limits and transgresses them, in the sense at least that it points beyond those limits (maybe even puts out ‘feelers’) to reality or the world beyond. On this simple but powerful view of language, reality, whether construed as empirical, as in Locke, or of some higher order, as in Plato, not only comes first in the relationship, but ‘wears the trousers’, as J. L. Austin might have said. If I can quote David Pears characterising this view (as that of the Tractatus) in his magnum opus on Wittgenstein:
The idea is that in all our operations with language we are really running on fixed rails laid down in reality before we even appeared on the scene. Attach a name to an object, and the intrinsic nature of the object will immediately take over complete control and determine the correct use of the name on later occasions. Set up a whole language in this way, and the structure of the fundamental grid will inexorably dictate the general structure of the logical system.2
My intention is not to follow what happens to these questions in Wittgenstein, however, but to see how they fare in the work of Jacques Derrida. It might plausibly be thought, for example, and often indeed has been thought, that Derrida is proposing (however obscurely) a philosophy of language, and one that also attempts to produce a non- or anti-Platonist account of what holds language together in its relations with the world, once the Platonic presumption has been refused, albeit not an account that obviously turns to what Pears calls ‘the stability of our own practices’ to do the trick.3 It would also have in common with Wittgenstein that it gives rise to what Pears calls ‘the characteristic intellectual giddiness of withdrawal from Platonism’. On this view of Derrida proposing a philosophy of language, Derrida might be thought, and often has been thought, to argue for a kind of radical linguisticism and indeed a kind of ultra-linguisticism, as though there were really no world but only language. When Derrida notoriously claims in Of Grammatology that ‘there is nothing outside the text’, or ‘il n’y a pas de hors-texte’,4 it is difficult not to think that this amounts to a radical declaration of linguistic imperialism. In common with what is often taken to be the position of Saussure and the whole structuralist and poststructuralist movement inspired by him, reference stops even being a question just because, on this view, language constantly and endlessly refers only to itself. Here, the limits of my language are indeed the limits of my world not only because language gives me some kind of privileged access to the world, or maybe even allows me to ‘have’ a world in the first place, but because my world is in a significant sense entirely language anyway.
I want to argue that this is a mistaken reading of Derrida’s thinking. Far from imprisoning us and everything there is in language, or so I shall be suggesting, Derrida certainly questions any simple separation of language and world, certainly pays the closest attention to language in its most irreducible and idiomatic aspects, but in so doing does not so much bring the world into language as he de-linguistifies language and takes it out into the world. In other words, Derrida’s concern is to think about the supposed separation of language and world in a way that neither reinforces it nor simply denies it, but that places that separation into a situation of différance such that ‘language’ and ‘world’ at most name poles of attraction in a more general structure which is neither language nor world (nor their synthesis or amalgam). One way of saying this in Derridean language is that the traditional opposition of language and world is deconstructed. This means coming to some understanding of the claim that there is nothing outside the text that does not simply identify ‘text’ with ‘language’, and understands the thought that reality or world is ‘textual’ not to mean that reality or world is linguistic.
Derrida certainly seems to begin this process of deconstruction from the side of language, as it were – whence the (at least superficial) plausibility of the view I am contesting. But his derivation of the quasi-concepts of trace, différance and text entails that language, at least as usually understood, dissolves somewhat in the process of the derivation, which I shall now sketch out.
Derrida takes very seriously Saussure’s notorious claim that ‘language is a system of differences without positive terms’, first in order to deconstruct the very notion of the sign to which Saussure is still quite firmly attached, and to do so apparently by prioritising the signifiant over the signifié. If a view of language that we might think of as broadly ‘Platonic’ has a clear order of priority beginning with the thing, moving to the idea of the thing (roughly what Saussure calls the signifié) and then to the (usually spoken) signifiant (and just occasionally to one further remove, i.e. the written signifier of that spoken signifier), then Derrida uses Saussure apparently to invert that order of priority. Saussure, it will be remembered, arguing against his version of the ‘Platonic’ position (what he calls the view that language is essentially a nomenclature), argues that, once the perception that signs are ‘arbitrary’ is taken seriously and radically (i.e. beyond a merely conventionalist understanding involving a kind of aporetic fiction or fable whereby people must one day have agreed on what to call things), then the only thing that can be thought to hold language together in any kind of coherence (as a system, what Saussure calls langue as opposed to actual uses of language or parole) is the differential relationship of elements among themselves. Notoriously, Saussure’s view of language pushes ‘the world’ into the background, and draws attention to the way that linguistic elements relate to each other as a prior condition of their having any chance at all of referring to things. Signs, in Saussure, form a kind of self-supporting network, and it is that network and the intra-linguistic relations of its terms that determine what can be referred to, and perhaps even what can be said to exist. This is a kind of reversal the possibility of which already haunts the ‘Platonic’ view of language as a possible aberration or pathology to be avoided. I’m thinking here especially of Locke’s splendid indignation about ways in which ‘words’ rather than ‘things’ can sometimes come to the fore, for example in rhetoric, that ‘perfect cheat’, and his more deep-seated worry about the terms he calls ‘mixed modes’, where words can hardly fail to do so.5 The priority in Saussure clearly goes to sign over referent: and it seems as though the Derridean position, pushing Saussure a little further than the latter most obviously wants to go, must be that the priority goes not just to the sign, but within the sign to the signifiant, the so-called ‘play’ of which (i.e. its dynamic interrelatedness entailed by the differences-without-positive-terms view of language) generates what Derrida calls ‘effects’ of meaning or of signifieds. The signified becomes no more than a kind of sum of the differences between signifiers, always only, as Derrida puts it, ‘a signifier placed in a certain position by other signifiers’, or ‘a stratum of the signifier that gives itself as signified’.6 That signified could give rise, at least on occasion, to something like a ‘world’, which on this view would apparently always be in some sense the ‘poetic’ ‘product’ of the activity of language itself. On this very plausible and widespread reading of Derrida, he would then line up in a general kind of way with thinkers such as Merleau-Ponty and Heidegger and propose a version of what Levinas memorably characterises as a ‘disoriented’ ‘world’ of essentially lateral and fundamentally poetic significations intrinsically lacking any orientation or sens.7
Saussure himself draws back from extending the ‘differences without positive terms’ argument to the ‘sign as a whole’ – i.e. he thinks that it is valid at the level of signifier taken separately, and at that of the signified taken separately, but that at the level of sign ‘in its totality’, we are dealing, as Saussure says, with ‘something positive in its order’, and that the appropriate form to look for relationships in this order is that of opposition rather than difference. One imagines that the paradigmatic opposition for Saussure here is just the very opposition between signifier and signified themselves. Naturally Derrida’s reading does not (or at least would not – I don’t think he ever in fact makes this argument quite explicit) accept this consequence, just because he is concerned to undermine any thought that there could be a ‘sign in its totality’ and, apparently at least, to urge the idea that the functioning of the signifier is alone sufficient to account for the operation of language.
Unlike some of the other post-Saussureans, however, Derrida conspicuously and explicitly does not decide to remain here with the signifier (which can be presented as ‘floating’ with respect to any signified, for example, or can, in some versions, be reclaimed as the ‘material’ part of the sign as against the ‘idealist’ attachment to the signified),8 in spite of the priority it seems to be given in this view of language. This is a slightly tricky point which I think has not always been taken in by the commentators: Derrida certainly casts suspicion on the very idea of a signified, and does so by prioritising the signifier. Look up in the dictionary a word you don’t know (i.e. a signifier without a signified), and the dictionary will never provide you with more than more signifiers, which may in turn send you elsewhere in the dictionary in search of their signified, which you will never find. It looks, however, as though the reason we do not seem (at least in any simple way) to get stuck in the dictionary, and typically return from it with a sense of having understood something, is that at a certain point, a signifier (placed in a certain position) counts as a signified or functions as a signified. Just because, as I mentioned earlier, and as Derrida says in a different section of the Grammatology, something of the signifier gives itself or gives itself out as a signified, secretes an ‘effect’9 of a signified, so, in a more general way the signifier /signifier/ cannot fail to secrete /signified/ as thesignifier-that-counts-as-its-signified. Or, put another way, even accepting the ‘differences-without-positive-terms’ argument and deciding that it finally entails the absence of a signified (or the thought, if I can put it in a more Anglo-Saxon idiom, that ‘words’ don’t need to have ‘meanings’), still, in the interrelatedness that marks language in the ‘difference’ model, the extreme proximity of the signifier /signified/ to the signifier /signifier/ in the network, and, more importantly perhaps, the millennial tradition of moving from signifier to signified (because the network is never, in spite of what Saussure might sometimes appear to suggest, simply independent of its history) tends to make almost inevitable that, given ‘signifier’, then ‘signified’. In other words, once the deconstruction of the sign proceeds to the point at which the signified falls away and the whole field seems to be covered by the signifier, then the very coherence of the signifier starts to be a question in its turn.
This is one of the reasons why Derrida moves cautiously toward the term ‘trace’ to capture what the ‘difference’ view of language entails, if one accepts its more radical implications and refuses the Saussurean ‘fall-back’ position on signs-as-a-whole and opposition.10 I want to argue that one advantage of this move (over the temptation to retain the term ‘signifier’ in the way that Lacan, among others, does) is that it allows a non-reductive extension of the ‘difference’ argument beyond what would still be, in Saussure at least, a quite subtle form of idealism. This is because (1) maintaining the thought of a ‘sign’ entails maintaining the distinction between signifier and signified: but the very thought of signified (i.e. of signified thought as signified, and no longer as signifier-giving-itself-out-as that we just saw) entails that the signified be ideally and in principle separable from the signifier, and this in turn rapidly entails that there be a transcendental signified, the most perspicuous traditional signifier for which is the name ‘God’ (this is why the metaphysical concept of sign is the concept of sign-vanishing-in-favour-of, and why, therefore, as Derrida puts it in Speech and Phenomena, ‘effacing’ the classical concept of the sign can involve insisting on it and maintaining it against that built-in effacement);11 (2) Deciding, on the ‘difference’ argument, that we can advantageously (and possibly even materialistically) replace signified with signifier, oddly produces a version of just the same problem. The traditional thought of signified cannot fail to generate a ‘transcendental signified’: the attempt to think signifier without signified cannot fail to generate a ‘transcendental signifier’12 which is really not so different from a signified –transcendentalised in this way, a signifier may just as well be a signified13 – whence the suspicion that it might be strategically preferable to use another term, i.e. trace. ‘Trace’, as Derrida derives it in these notoriously difficult pages of the Grammatology, is not the perfect nor even the right term for what is being thought here,14 but attempts to gather (from its semi-stable but never completely stable ‘place’ in the language system15) some of the possibilities of the elaborated and radicalised ‘difference’ account. ‘Trace’ attempts to capture something about that account: namely that faced with a ‘given’ element of the language system (what we’ll still always feel like calling a word or a signifier), and wondering what it is that makes that element the element that it is rather than something else (and this seems to mean wondering first, then, what makes it different from all the other elements that it is not), we are driven to the thought that any given element must in some sense bear the trace of all those other elements that it is not, so that (to take an almost trivially simple example) in the roman alphabetical system, what it is about the letter ‘b’, say, that renders it identifiable or recognisable as such (and thereby also repeatable as the same letter on another occasion) is the ‘presence’ in it of the ‘absence’ of all the other letters that it is not. On this view, ‘b’ is not simply present as itself (i.e. as a letter of the alphabet rather than just a line or set of lines), in that its ‘presence’ as ‘b’ emerges as the product, as it were, of the ‘absence’ of all those other letters. And those other letters are not simply absent (just because their absence is in some sense ‘present’ here and now in this ‘b’ that they define just by the trace of their absence). ‘Trace’ is the word Derrida proposes, then, to capture something of this set-up which defies simple description in terms of presence and absence, and which can, at the very least, be extended to the whole of what we typically call ‘language’.16 This radicalisation of the ‘difference’ view of language is also of course what generates Derrida’s infamous neologism ‘différance’, naming that which gives rise to the instantiable differences that Saussure is talking about. As Derrida puts it:
Now here the appearing and the functioning of difference presuppose an originary synthesis that is preceded by no absolute simplicity. This, then, would be the originary trace. Without a retention in the minimal unit of temporal experience, without a trace retaining the other as other in the same, no difference would do its work and no meaning would appear. We are, then, not here dealing with a constituted difference but, before any determination of content, of the pure movement that produces difference. The (pure) trace is différance. It depends on no sensory plenitude, audible or visible, phonic or graphic. On the contrary, it is their condition. Although it does not exist, although it is never a present-entity outside all plenitude, its possibility is de jure anterior to everything that is called sign (signifier/signified, content/expression, etc.), concept or operation, motor or sensory.17
So différance just is ‘trace’ when trace is thought along the lines we have indicated, as the condition of possibility for the particular differences that constitute the language system, and, as always in Derrida, the condition of impossibility that the sign ever quite settle into the fixed unity of signifier and signified or word and meaning. As is perhaps a little better understood than the general trace argument, Derrida will also call this condition ‘writing’ or ‘archi-writing’ or text, on the grounds that the tradition has massively determined ‘writing’ to mean ‘(graphic) signifier of a (phonic) signifier’, as opposed to speech which is thought to be directly the signifier of a signified. If we’re using the traditional terminology, language in general now looks better described as ‘signifier of a signifier’, for, according to the trace-argument, there is no signified.
Once this radicalisation of Saussure has taken place, and the sign has been disassembled, as it were (‘deconstructed’), then we no longer quite have a sign, of course, nor a signified, nor thereby a signifier. Language thought along the (generally anti-Platonic) lines proposed by Saussure leads to the description in terms of trace as complication of presence and absence that rapidly begins to render the ‘limits’ of language difficult to discern. This is not the same, I think, as annexing certain apparently non-linguistic elements to language by making them significant (so that non-linguistic objects can be counted as signifiers), nor even quite the same kind of thing that Wittgenstein does when, in the Investigations, he counts certain non-linguistic elements (colour-samples, say) as part of the language-game being described. Rather, on the basis of Saussure’s insight that if language is not a nomenclature then it will require the ‘differences-without-positive-terms’ characterisation, it seems as though a regular and coherent conceptual derivation has produced a description of language which has a kind of dissolving virtue with respect to the idea that language has limits, or at least that it is simply distinguishable from something other than it (such as the world, say).
For once we have ‘trace’, it seems, then we also have a way of thinking about more than just ‘language’, as Derrida claims in a famously dense and perhaps rather excitable page of the Grammatology, and which I am really trying to do no more than approach:
The ‘immotivation’ of the sign [i.e. what Saussure calls its ‘arbitrary’ nature] requires a synthesis in which the absolutely other announces itself [‘s’annoncer’ also means something like ‘to loom up’: we might want to say that in the trace the other looms without ever quite showing up in person or presence] as such – without any simplicity, identity, resemblance or continuity – in what is not it. Announces itself as such: that is the whole of history, from what metaphysics determined as the ‘not-living’, up to ‘consciousness’, passing through all levels of animal organisation. The trace, in which the relation to the other is marked, articulates its possibility on the entire field of entities, which metaphysics has determined as being-present on the basis of the occulted movement of the trace. The trace must be thought before the entity. But the movement of the trace is necessarily occulted; it produces itself as the occultation of itself. When the other announces itself as such [when it looms, then], it presents itself as dissimulation of itself. [So, in our alphabetical example, the ‘a’ and the ‘c’ and all the other letters loom in the ‘b’ but never appear in the ‘b’, which just appears as a ‘b’.] This formulation is not theological, as one might think somewhat precipitously. The ‘theological’ is a determinate moment in the total movement of the trace. The field of entities [i.e. ‘the world’ in terms of the standard opposition between language and the world], before being determined as field of presence, is structured according to the diverse – genetic and structural – possibilities of the trace. The presentation of the other as such, i.e. the dissimulation of its ‘as such’, has always already begun and no structure of being escapes this.18
‘Trace’, then, says something about what Derrida also sometimes calls an ‘originary synthesis’ whereby anything at all is what it is only in its relational difference from its others in general, of which it bears the non-present but also non-absent trace from the start. In terms of his later work, we can say that the trace entails a general ‘spectrality’: any apparently ‘present’ element is ‘haunted’ by the others of which it bears the trace, but this haunting also spectralises it, renders it less than fully present. This is what leads him to describe the trace as ‘originary trace’, and then point out that the very notion of the trace disrupts the concept of origin, so that if at the origin ‘there is’ trace (but then, the trace ‘is’ not), then it is no origin.19 Once what is, is an effect of the trace, then the trace is only ever the trace of (another) trace, and so on, without ever reaching any final or first ‘presence’.
‘Trace’, then, derived here from the reading of Saussure’s account of language, and radicalised as the quasi-originary possibility of any effect of identity whatsoever, has overrun any notional ‘limits of language’ with which we might have begun. ‘Language’ is an ‘abstraction’ with respect to the trace, and so, presumably, is the ‘world’. Derrida is perfectly explicit about this, writing on this same page that
the general structure of the unmotivated trace causes to communicate in one and the same possibility and without their being separable except by abstraction, the structure of the relation to the other, the movement of temporalisation and language as writing.20
Or else, writing in the slightly later essay ‘Signature, événement, contexte’:
I should like to show that the features that can be recognized in the classical and narrowly defined concept of writing are generalisable. [In my view] they would be valid not only for all orders of ‘signs’ and for all language in general but, beyond semio-linguistic communication, for the whole field of what philosophy would call experience, or even the experience of being: so-called ‘presence’.21
How, then, are we to account for the privilege that language still seems to have? For example, in ‘Cogito and the History of Madness’ Derrida says that language is no doubt the fact that always resists transcendental suspension or epoche.22 I imagine that Derrida would want to say that différance, implied by the originary non-simplicity or again ‘originary trace’ (where ‘trace’ crosses through ‘origin’) that is the upshot of his reading of Husserl as much as the reading we have followed of Saussure, gives rise to language (or perhaps ‘something like language’) just through the differentiation that it is from the start. In the order of ‘analysis’, at least, once the sign is deconstructed along the lines I have indicated – giving rise not to any finally-discovered nucleus or atom (and this is why deconstruction is not quite analysis, of course23) but to the originary and endless complexity that ‘trace’ or ‘text’ attempts to capture, but by definition never can quite capture – then the differential relationship which is the ‘principle’ of the ensuing ‘ontology’24 already contains – in the referral that is part of what the trace is, or that is just what the trace is – the possibility of a signifying relation which is continuous with what we would more normally be inclined to call language. If we saw him, in the difficult passage about the trace from the Grammatology, insisting on the whole field and history of mineral, vegetable and animal existence, this is not only to ask hard questions of the traditionally massive distinction that metaphysics (up to and including Lacan, Heidegger and Levinas at least) needs to draw between the human and the animal, with all that flows from it (so that on Derrida’s view the metaphysical need to deny the capacity of language to animals in general reflects an insufficiently critical approach),25 but also to suggest that the continuity that ‘text’ suggests extends across everything that is. In other words, the thought of the trace and of the text affirms that in the ‘first’ distinction between anything and anything, in the minimal referral (not yet a reference in the normal sense of the term) that the trace involves, the possibility of what we come to think of as language is already given.
On this view, what philosophers often take to be the ‘normal’ function of language, i.e.., reference to the world, is a derivative possibility of this generalised trace-structure. Reference in this ‘normal’ sense is a special case of referral or renvoi in the sense specified by trace and text. Language and world have, if you will allow me, the same general texture. This is why, in the infamous ‘Question of Method’ section of the Grammatology, in which the ‘nothing outside the text’ remark is originally made, Derrida, discussing the status of the apparently extra-textual referent of Rousseau’s Confessions, to which a reading might claim to gain some kind of access, says this:
The fact that reading should not be content to double the text does not mean that it can legitimately transgress the text toward something other than it, toward a referent (a reality, be it metaphysical, historical, psycho-biographical, etc.) or toward a signified outside the text, a signified the content of which could take place, could have taken place outside language, i.e., in the sense we are here giving to this word, outside writing in general. This is why the methodological considerations we are risking here around one example are strictly dependent on the general propositions we elaborated earlier, as to the absence of the referent or the transcendental signified [i.e. the arguments about the trace that I have been rehearsing]. There is no outside-the-text. [Il n’y a pas de hors-texte]. And this not because the life of Jean-Jacques does not primarily interest us, nor the existence of Maman or Thérèse themselves, nor because we have access to their so-called ‘real’ existence only in the text and have no means to do otherwise [I think that this last is often presented as though it were Derrida’s own position . . . ], nor any right to neglect this limitation. All reasons of this type would already be sufficient, of course [i.e. sufficient to render illegitimate the claim to transgress the text and get to the signified or the referent], but there are more radical ones. What we have tried to demonstrate following the guiding thread of the ‘dangerous supplement’, is that in what is called the real life of these existences ‘in flesh and blood’, beyond what one thinks can be circumscribed as Rousseau’s oeuvre, and behind it, there has only ever been writing; there have only ever been supplements, substitutive significations which could emerge only in a chain of differential referrals [renvois], the ‘real’ supervening or adding itself only by making sense on the basis of a trace and a call for a supplement, etc. And thus ad infinitum . . .26
And, returning to this slogan in the ‘Hors-Livre’ section of La dissémination, and modifying it a little, so that now we have, for example, ‘il n’y a pas de hors-texte absolu’27 Derrida wants to stress that this apparent refusal of the ‘outside’ does not amount to a simple generalisation of the ‘inside’: ‘If there is nothing outside the text, this implies with the transformation of the concept of text in general, that text no longer be the comfortable inside of an interiority or self-identity ( . . . ) but a different set-up of effects of opening and closing’.28 Which will lead (via a reading of Lautréamont’s Les chants de Maldoror, which for some reason I do not imagine is common reading among students of philosophy these days) to formulations such as ‘Il n’y a que du texte, il n’y a que du horstexte’, these now being the same thing, or differentiations of the same ‘thing’.29
This confirms the thought I have been urging upon you: in Derrida there is no outside, no inside, no limits or frontier between language and something else, rather a differentiated (by definition . . . I want to say more or less differentiated, differently differentiated) milieu of difference which Derrida also sometimes calls simply ‘the Same’, and which has something to do with his abiding interest in Plato’s Khora.30 ‘The limits of my language are the limits of my world’ not because, or not only or primarily because, as in the so-called Sapir-Whorf hypothesis, the language I happen to speak conditions and even creates the parameters of my reality (however true that may also be), but because language and world are the same, text and trace, which is neither simply language nor world.
1. Ludwig Wittgenstein, Tractatus Logico-Philosophicus, trans. D. F. Pears and B. F. McGuinness (New York: Routledge, 2001), 5.6. This chapter was given as a plenary lecture in April 2006 to the Boston College Graduate Student Philosophy Conference ‘On Language’, and was previously published in J. Burmeister and M. Sentesy (eds), On Language: Analytic, Continental, and Historical Contributions (Cambridge, MA: Cambridge Scholars Press, 2007).
2. David Pears, The False Prison: A Study of the Development of Wittgenstein’s Philosophy (New York: Oxford University Press, 2004), Vol. I, p. 10.
3. Pears, The False Prison, Vol. I, p. 12.
4. De la grammatologie, p. 227.
5. See my discussions in ‘The Perfect Cheat: Locke and Empiricism’s Rhetoric’, Legislations: The Politics of Deconstruction (New York: Verso, 1994), pp. 119–36. See also Sententiousness and the Novel (New York: Cambridge University Press, 1985), an e-book reprint at bennington.zsoft.co.uk.
6. De la grammatologie (Paris: Minuit, 1972), p. 229.
7. Emmanuel Levinas, ‘La signification et le sens’, in Humanisme de l’autre homme (Montpellier: Fata Morgana, 1972), pp. 17–63 (especially pp. 34, 42–5, 48–50).
8. Although Derrida would certainly agree with questioning any particular attachment to the signified (because that is just what he calls ‘logocentrism’), he more than once points out that the signifier can in no simple sense be thought of as simply ‘material’ – its intrinsic repeatability as the same signifier entails a measure of ideality that this self-proclaimed ‘materialist’ view cannot account for: see, for example, De la grammatologie, pp. 20, 45, 138: see also ibid., p. 32, n. 9: ‘The “primacy” or the “priority” of the signifier would be an untenable and absurd expression in that it is formulated illogically in the very logic that it wishes, legitimately no doubt, to destroy. The signifier will never precede de jure the signified, because otherwise it would no longer signify and the signifier “signifier” would no longer have any possible signified.’ See too the much more recent and playful remark about Hélène Cixous: ‘She knows just when to stop at the limit of crafty artfulness, when the signifier no longer signifies because it is merely a signifier’ (Jacques Derrida, Genèses, généalogies, genres, et le génie: les secrets de l’archive (Paris: Galilée, 2003), pp. 44–5).
9. Derrida says that this use of ‘effect’ is not quite that of cause-and-effect, nor quite that of ‘illusion’ . . . See Jacques Derrida, Positions (Paris: Minuit, 1972), p. 90.
10. See, most recently, the remarks in the posthumously published seminar La Bête et le souverain I (Paris: Galilée, 2008), p. 181: ‘ . . . I substituted the concept of trace for that of signifier . . .’.
11. This is almost a rule that determines which concepts retain Derrida’s attention, at least in his earlier work: so metaphor, for example, and its two ‘deaths’ laid out at the end of Jacques Derrida, ‘La mythologie blanche’, in Marges de la philosophie (Paris: Minuit, 1972), pp. 320–21, or ‘Hors livre, préfaces’, in La dissémination (Paris: Seuil, 1972), p. 14.
12. Such as the phallus in Lacan, at least as Derrida reads him: but our logic here is that the signifier /signifier/ is already occupying this place as a paradoxical consequence of its being cut free from ‘signified’.
13. See La dissémination, p. 32, discussing the psychoanalytic interpretation of castration: ‘The void, the lack, the cut, etc., have here received the value of a transcendental signified or, what comes to the same thing, a transcendental signifier: self-presentation of the truth (veil/not-veil) as Logos.’
14. In general, there is never a ‘right’ term in Derrida, no true or proper name for his object, whence his series of more-or-less contextually motivated ‘non-synonymic substitutions’. This type of situation is why Derrida sometimes and rather mysteriously suggests the need to maintain a ‘non-classical’ separation between thought and language, as for example in the following perhaps rather obscure comments from a note to ‘Cogito et histoire de la folie’: ‘So it is necessary, it is perhaps time to return to the ahistorical in a sense that is radically opposed to that of classical philosophy: not in order to misrecognize but this time in order to avow – in silence – negativity [‘avow in silence’ to avoid the trap of the false or dishonest avowal of philosophy that claims to avow out loud a negativity that in fact it recuperates and thus forgets]. It is this negativity and not positive truth which is the non-historical fund of history. We would be dealing in that case with a negativity so negative that it could no longer even be named thus. Negativity has always been determined by dialectics – i.e. by metaphysics – as labour in the service of the constitution of meaning. Avowing negativity in silence means acceding to a dissociation of a non-classical type between thought and language. And perhaps between thought and philosophy as discourse: knowing that this schism can only be said, while effacing itself, in philosophy’ (L’Ecriture et la différence (Paris: Seuil, 1967), p. 55n.). See also ‘Write, He Wrote’, below.
15. Synchronic and diachronic, so Derrida is led in the Grammatology to motivate to some extent his use of this term in the light of other uses he finds around him: ‘If words and concepts take on meaning only through linkings of differences, one can justify one’s language and the choice of terms only through within a topic and a historical strategy. Their justification cannot ever, then, be absolute and definitive. It responds to a state of forces and translates a historical calculation. In this way, beyond those we have already defined, a certain number of factors belonging to the discourse of the time, progressively imposed this choice upon us. The word “trace” must of itself make reference to a certain number of contemporary discourses the force of which we intend to take into account. Not that we accept all of them. But the word “trace” establishes with them what seems to us the most secure communication and allows us to save time on developments that have shown up their efficacy in them. In this way, we liken this concept of “trace” to the one that is at the centre of the latest writings of E. Levinas and his critique of ontology’ (De la Grammatologie, p. 102).
16. Derrida of course thinks that it is no accident that graphic examples seem more promising that phonic ones to illustrate this argument about difference and the trace, and it is this that will in part justify the later extension of the term ‘text’.
17. De la grammatologie, p. 92.
18. ibid., p. 69.
19. Derrida here writes ‘aucune structure de l’étant’ rather than ‘de l’être’. Trace needs to be thought before the ‘étant’ (the entity, or a being with a lower-case ‘b’, das seindes), but perhaps not before ‘l’être’ (Being, Sein). The reference to Heidegger, unmistakable here, would refer us to the earlier discussion in the Grammatology where Derrida suggests a kind of ambiguity in Heidegger’s Sein, whereby on the one hand it still has at least a foot in the metaphysics of presence and functions as an ‘originary word’ and a transcendental signified, even as it also questions and even deconstructs any such understanding. Derrida’s gesture in these pages, curiously like the one he makes with respect to Husserl in the Saussure chapter, is that it is absolutely necessary to pass through Heidegger’s thinking of the onticoontological difference (the difference, then, between ‘être’ and ‘étant’), even as he asserts that what he, Derrida, will call ‘différance’ is ‘more originary’ still: ‘To get to the point of recognizing, not short of but on the horizon of the Heideggerian paths, and still in them, that the meaning of being is not a transcendental or trans-epochal signified (even were it always dissimulated in the epoch) but already, in a properly unheard-of sense, a determinate signifying trace, is to assert that in the decisive concept of ontico-ontological difference, everything is not be be thought at once: entity and Being, ontic and ontological, “ontico-ontological” would be, in an original style, derived with respect to difference, and with respect to what later on we shall call differance, an economic concept designating the production of the differing and the deferring. Ontico-ontological-difference and its ground (Grund) in the “transcendence of Dasein” (Vom Wesen des Grundes, p. 16) would no longer be absolutely original. Differance tout court would be more “originary”, but one could no longer call it “origin” or “ground”, these notions belonging essentially to the history of onto-theology, i.e. the system that functions as effacement of difference. However, difference can only be thought up close to itself on one condition: that one begin by determining it as ontico-ontological difference before crossing this determination out. The necessity of the passage through the crossed-out determination, the necessity of this turn of writing is irreducible. This is a discreet and difficult thought that, through many unnoticed mediations, ought to carry the whole weight of our question, a question that we are still provisionally calling historial. It is thanks to it that later we’ll be able to try to make differance and writing communicate’ (Derrida, De la grammatologie, p. 38). Compare on the use of the concept of experience and the need to ‘pass through’ transcendental phenomenology and ‘leave a track’ [sillage]: ibid., pp. 89–90.
20. De la grammatologie, p. 69. See also twenty pages later: ‘Because archi-writing, the movement of differance, irreducible archi-synthesis, opening together, in one and the same possibility, temporalisation, the relation to the other and language, cannot, as a condition of any linguistic system, form part of the linguistic system itself, be situated as an object in its field.’ (ibid., p. 89). The ‘relation to the other’ part of this description is what justifies Derrida in saying that his later interest in more apparently ethical and political questions is not the result of some ‘turn’ on his part.
21. Marges de la philosophie, p. 377.
22. ‘Because this difficulty or this impossibility must rebound on the language in which this history of madness is written, Foucault indeed recognizes the necessity of maintaining his discourse in what he calls a “relativity without recourse”, i.e. without support from the absolute of a reason or a logos. Necessity and at the same time impossibility of what Foucault elsewhere calls a “language without support”, i.e. refusing in principle if not in fact to articulate itself on a syntax of reason. In principle if not in fact, but the fact here does not easily let itself be bracketed. The fact of language is no doubt the only one that in the end resists all bracketing. “Here, in this simple problem of elocution”, says Foucault, “the greatest difficulty of the undertaking was hidden and expressed”’ (L’Écriture et la différence, p. 60).
23. See Jacques Derrida, ‘Lettre à un ami japonais’, Psyché. Inventions de l’autre (Paris, Galilée, 1987), p. 392, and especially Jacques Derrida, ‘Résistances’, in Résistances de la psychanalyse (Paris: Galilée, 1996), pp. 10–53.
24. ‘Ontology’ in quotation marks: cf. La dissémination, p. 65: ‘The beyond of the totality, another name of the text insofar as it resists all ontology . . . ’. See too the remarks on ontology in response to Antonio Negri, in Michael Sprinker (ed.), Ghostly Demarcations: A Symposium on Jacques Derrida’s Specters of Marx (London: Verso Books, 1999), pp. 257–62.
25. See especially the texts collected in the posthumous volume L’Animal que donc je suis (Paris: Galilée, 2006).
26. De la grammatologie, pp. 227–8.
27. La dissémination, p. 42, my emphasis.
28. ibid.
29. ibid., p. 50.
30. See especially Khôra (Paris: Galilée, 1993).