Aller au contenu
Accueil » Lacan’s version of the Signifer

Lacan’s version of the Signifer

Lacan, Krauss & the unconscious

The unconscious is structured like….

Language is central to Lacan’s work from about 1950 on, before the Seminars and all but the earliest of the Écrits. The precise formulation that « the unconscious is structured like a language » first occurs (I think) in Seminar III from 1955-56 (The Psychoses), as a development of the argument from the 1953 « Function and Field of Psychoanalysis », that « the unconscious is the discourse of the Other ». This will be the heart of what Lacan does until about 1970. Seminars XVIII (1970-71: D’un discours qui ne serait pas du semblant, On a discourse that would not be one of semblance) and XIX (1971-2: …ou pire; …or worse) will introduce the idea of lalangue. Lalangue is language not as communication, representation or even meaning, but as embodied jouissance, enjoyment. The dimension of lalangue is profoundly senseless, an obdurate core which resists meaning. As a limit to meaning, and thus to any analysis based on interpretation, lalangue takes up that problem of the potential interminability of analysis which troubled Freud late in his career – and which we catch a glimpse of at they very beginning of psychoanalysis, when Freud declares in The Interpretation of Dreams that every dream has its navel, the point beyond which interpretation cannot go.

From the early 1970s on, then lalangue will lead us though to the famous 1973 seminar on sexuality, Encore (where Fink translates lalangue as llalangue) and then the 1975 seminar on the sinthome, where Lacan’s reading of James Joyce serves to crystallise those ideas of the limits of meaning. The sinthome is an element which, though it makes no sense itself and is thus unanalysable, is what allows there to be sense (and analysis), though at the same time it marks the limit of these, beyond which they cannot go. All of this is why the Lacan of those last few years moves away from language-based models (an unconscious which would be structured like a language) and towards topological models (in particular, knot theory).

From the outset, that dictum of the unconscious structured like a language was not at all the endorsement of Saussurean linguistics it’s easily read as, but a thoroughgoing critique. We can see what Lacan does from about 1950 on as a three-decade-long investigation of just how those classical Saussurean ideas are not an adequate description of the state of things.

We could point out to begin with how Lacan’s reading of Saussure provides a description of what Lacan will call the Symbolic. Saussure (at least in one of his modes) describes the sign as differential. That is, the sign is what it is, not because it contains a positive kernel of signified meaning, or because it represents the world, but simply because it exists within a system of elements from which it can be differentied. (Signs can represent, of course – they do that all the time. The point is that this representing is an effect to be explained rather than the basic mechanism of signs.) From this bare property of difference alone, Saussure claims, one may deduce all the properties of language.

Nevertheless. as Krauss points out, there’s a very strong residual nominalism in Saussure (the famous diagram of the tree in the Course in General Linguistics, where the signifier « tree » in effect represents a signified idea of « treeness »). There’s also a very strong reliance on naive comunications theory (which sees language as thought coded into language, thrown across space to a hearer who then decodes it into thought again – which, if you think about it, describes at best the experience of a speaker and hearer who don’t know the language very well: it implicitly the experience of the incompetent user as its model).

If for Saussure all of that is bound up with the concept of the sign, then – as Krauss says – Lacan will argue against the very concept of the sign. For Saussure, the sign is made up of two parts, signified and signifier. The function of the signifier, as the name suggests, is to signify the signified meaning. That is, the sign is a unity, held together by that relationship between signifier and signified. Saussure will, of course, famously argue that the relationship between the two is « arbitrary »: this is a recognition of the fact that different languages obviously give different names to similar concepts, that some concepts can be signified in a number of ways, and so on. But Lacan’s position is more radical: he talks in the 1957 « Agency of the Letter » (Écrits) about « the illusion that the signifer answers to the function of representing the signified, or better, that the signifier has to answer for its existence in the name of any signification whatsoever ».

Krauss’s opening suggests the « antinarrative, nonobjective » nature of much painting, as if there is a battle between the image and the word, and goes on to make a series of disciplinary claims and counterclaims. Richard Boothby suggests, in his Freud as Philosopher: Metaphysical after Lacan, that the image and the symbolic are the flipside of each other, so that the symbolic [is] the un-imageable background to the image. If that’s so – and we’ll come back to that – then even the « most austerely antinarrative, nonobjective paintings » Krauss invokes can be so only on the basis of, if not language, then at least a background which is not itself visible, as the image is, but which, in that invisibility, is the very condition of possibility of the visible. After all, the mark set against a background is the very basis for the very activity of the painterly. But this setting-against-a-background is not itself an image; it is what must occur for there to be an image. As such, it is necessarily withdrawn from even the « most austerely antinarrative, nonobjective paintings ».


Now this setting-off-against-a-background is differential. The logic it relies on is purely differential: this is not that, and indeed this stands out as being this only because it is not that, the background. Only because the image is not the background can we see it as image. This is very like Gestaltist logic (Lacan invokes Gestalt psychology fairly often in the Seminars, especially in II).

Krauss focuses her argument on « The Agency of the Letter in the Unconscious or Reason since Freud ». This piece is from 1957, when that linguistic turn is well under way; it is contemporary with Seminars IV on Object Relations and V on The Formations of the Unconscious. She draws a lot on Nancy and Lacoue-Labarthe’s The Title of the Letter: in particular on their argument that in adapting Saussure’s diagram of the sign, Lacan turns it into an algorithm, which it patently is not in Saussure.

Saussure’s diagram:

Lacan’s version:



where S (upper-case) is the signifier, and s (lower-case, italic) is the signified.

This is more than just an inversion. Saussure is indeed drawing a diagram, to show how the sign is structured internally in order to be a sign: it has two parts, which Saussure names, and a relationship between them. Lacan turns this into an algorithm, which is to say a systematic mathematical procedure for calculating a result. A = π r2 is the algorithm for calculating the area of a circle. If you know the radius, plug it into the algorithm, and what emerges is the area. For Lacan, the figure he offers is the algorithm for the Symbolic, the system of signifiers: this is how the Symbolic is generated.

Thus the Saussurean picture of the sign becomes a sort of formula – that’s crucial in Lacan’s preference throughout most of his career for expressing things in this quasi-mathematical way. Instead of the two interrelated parts of the sign we see in Saussure’s diagram, Lacan now inverts the whole thing to give the signifier pride of place, cuts the signified off from it with a bar, and insists that the sign has no unity at all. Instead a unity (the ellipse around the two parts of the sign in Saussure’s picture), we have an endless differentiation: the signifier is capable of generating meaning not because it has recourse to any signified whatsoever, but because it is now in relation with an indefinite and potentially endless set of other signifiers within the system.

In « Agency », Lacan takes Saussure’s model of the tree to task as misleading. Words don’t take on meaning because we can attach some extralinguistic signified to them; they work because we can connect them with other words. Think of the word « of »: commonplace, we all use it, it causes no problems (to native speakers: it’s a lot trickier if English is a second language). Dictionary definitions don’t give you a signified meaning outside of language, some sort of unlanguaged thought for which « of » stands (This is an « of »); they describe relationships « of » sets up among words around it. « Of » is a word whose function is to set up certain relationships between words before it and the words after it.

This is why Lacan puts the signified under a bar. We never reach a signified meaning which would be just pure concept, without reference to signifiers or to language. On the contrary, it’s the horizontal motion along the chain of signifiers, the relationships among signifiers, which give us meaning. (What do we do when we come across a word we don’t understand? We talk about it, use other words. Sooner or later, we find that we do know quite well what that word means.)

Möbius band

This is also why the signifying chain has the shape of a Möbius band for Lacan. An ordinary sheet of paper, like the Saussurean sign, has two sides, and you can only get from one to the other by crossing over an edge. A Möbius band, though, is what you get when you take a strip of paper and give it a 180o twist before joining it to the other. The odd property of a Möbius band is that it has only one side and one edge. Go far enough along the band and you end up back where you started, on the other « side ». Go far enough along the chain of signifiers (translation: talk about it), and you find that this is quite enough to produce, as an effect, a signified meaning.

As simple as it might seem, the S/s algorithm is enormously ambitious, as it’s from this that Lacan will develop all the properties of the unconscious. The radicality of Lacan’s approach is to show how so many things which are apparently properties of the psyche are actually properties of the signifier. A few years later, in the 1960 piece from Écrits, « The subversion of the subject and the dialectic of desire in the Freudian unconscious », he will develop an extraordinary series of 4 diagrams which begins with nothing more than the chain of signifiers, and then show in a series of successive layerings how, from nothing more than the properties of this chain, one can derive things as different as imaginary and symbolic identifications, desire, the structure of the superego, castration and jouissance, fantasy, drive, demand and desire, and a theory of why desire must always in its very nature remain unsatisfiable.

Krauss cites Nancy and Lacoue-Labarthe here, to say that what this does is destroy the classical view of the sign, as Saussure presents it. Quite right: but rather than see that as an unfaithfulness to Saussure or sloppy reading, it’s important to see that as precisely Lacan’s intention. Lacan dismantles the sign, if we take sign in its classical sense as something that represents something else (i.e. for a subject). He is not interested in the sign, but in the signifier, as algorithm, the thing from which everything else follows. The signified is redundant: it’s something a system of signifiers will produce, as an effect. (We should repeat that this is not a matter of Lacan’s arguing that language doesn’t represent things. Of course it does. No one has any problems with that. It’s a matter of saying that that function of representation is an effect, not a cause. Language doesn’t work because it represents ideas or the real. On the contrary, it is able to represent only because it works as a system of signifiers. Representation might be an important effect of language, but it’s not the basic mechanism behind it. What Lacan is offering is an attempt at a rigorously anti-nominalist theory of meaning.)

So, Lacan does indeed abandon and dismantle the very idea of a sign, as something which represents something else (sor a subject), and instead offers the more cryptic definition of a signifier as that which represents a subject for another signifer. This is precisely where he breaks with Saussure, where the Saussurean image of the sign becomes an algorithm, something whose working is itself a part of the symbolic. It is not, as Krauss suggests, the point where Lacan loses the way. It’s the deliberate and quite radical point from which he departs, the thing which sets him on his way.

We need also keep in mind something else that follows from this, as it will do a lot to dismantle Krauss’s argument in turn. If the Symbolic has no unity, but is produced entirely in and as an effect of difference, then the Symbolic, the Logos, can never be simply « impervious », as Krauss characterises it. The Symbolic must necessarily be furrowed, at every instant, with the lack which both makes it possible and at the same time guarantees that its empire will always remain incomplete. Logos is a sort of project: it never ends, never finally closes in on itself as completion. To see it as complete is, quite precisely, to see it in terms not of the Symbolic, but of an image, a presence to be grasped in a single gestalt. A much more complex and interesting question than that of escaping the supposed imperiousness of the Logos is how we are to conceive of Logos as this necessarily ever-incomplete background, as forever divided against itself. This is where what Lacan calls sexuation comes into the picture. Sexuation is not just sexual difference either as a biological fact or as a cultural production, but is also that very division of the Symbolic, the Logos, against itself. There is no single position from which one can totalise the Symbolic; there are always at least two radically incompatible and irreducible positions available within it. Far from being opposed to each other, as Krauss suggests, narrative and the image alike may function as ways of forestalling the anxiety which is a consequence of that radical division within the Symbolic.

Earlier, I invoked Gestaltist distinction between image and background, and suggested that for the image to work as an image it must necessarily partake of the logic of the symbolic. For us even to be able to perceive it as image rather than as all but unimaginable blooming buzzing confusion, we must be able to perceive it against what it is not. That is, the very thing that allows there to be an image in the first place is the basic relation which characterises the Symbolic: two terms, X and not-X, each of them taking on its meaning only through its juxtaposition with the other. The Symbolic is an irreducible condition of the image. As soon as we have an image, we also have the Symbolic.

The relation of the Imaginary and the Symbolic is thus far more complex than Krauss makes out. Rather than see the image as the nostalgic and liberatory outside of an imperious Logos, for Lacan the Imaginary and the Symbolic arelimits to each other.

Schema L

For Lacan, psychonanalysis must address the subject rather than the ego. That is, it has to take into account the unconscious and the symbolic, not only the imaginary and fantasised aspects which make up the ego. The ego is, surprising as it may seem, an object: the fantasy of the self objectified, frozen, no longer subjected to the vicissitudes of an unconscious. It is a fantasy a greal deal of positivist psychology buys into – hence Lacan’s famous insistence that psychoanalysis is not a psychology.

The ego, then, is firmly on the side of the Imaginary. It sees itself as formed in a series of mirror-relations with others, small-o others, other egos. These mirror relations are identifications with objective properties: I see in this other certain qualities I like, certain qualities I don’t like; certain things remind me of myself, or embody what I want to be, or don’t want to be, and so on. The subject, though, is on the side of the Symbolic, and of the unconscious, because it involves that entire dimension of the world which forms me but of which I am never directly or fully aware. Here, my identifications are not with small-o others, but with the big Other, the Symbolic system within which, even though I cannot see it as a totality, I am nevertheless positioned and by which I always stand to be judged. Symbolic identifications are always endless questions, unanswerable in principle: what should I do to be loved, or a good person? What’s the right thing to do here? What’s really wanted of me?

An adequate way of diagramming that would be like this:

It’s inadequate because it simply juxtaposes the Symbolic and Imaginary to each other – as if they were, as they seem to be for Krauss – just a choice we could make, depending on whether we’re dealing with language or an image.

I suggested that the real relationship between the two is that they are each other’s limit. To see how that works, we once more apply a 180o twist. Imagine that diagram drawn on a sheet of paper. Now keep the bottom edge fixed, the ego and Other on it, and give the top edge a full 180o twist, so that other and subject change places. What you get is this, which Lacan calls the Schema L:

The Symbolic and the Imaginary relations cross one another, chiasmatic and occluding. On the top, we have the ego relations, which tend towards a coherence (an imaginary one). Beneath that, we have the symbolic relations that subtend and support that very perception of coherence. As the diagram suggests, those relations are largely unconscious: they are not entirely visible, because they are occluded by those apparent ego unities where one edge of the twisted sheet dives under the other.

Again, what we have here is a very similar Gestaltist logic: the ego, as object, as image, appears against a ground of the unconscious. Because symbolic relations are the support of the image, to focus on them is to make the image itself vanish. That imaginary unit of the ego – or in Freudian terms, its essential narcissism – is something which is subtended by and left incomplete by the processes of the symbolic. One name for that incompleteness would be symbolic castration. (This is a term fraught with misreadings. It isn’t theft of a pleasure or a wholeness, something which is rightly ours but of which we have been robbed by our entry into the symbolic. Instead, it’s the cost of our access to pleasure, and to subjectivity itself: without it, there would be no pleasures, no subjectivity at all. The cost is that all desires and all pleasures appear incomplete and thus endless. We find ourselves measuring them up against a purely imaginary pleasure which has no actual existence and which we never find in the world. Castration is the name psychoanalysis has for the realisation that desire is in fact this endlessness: what desire desires is desire, more desire; the object one actually attains is never quite it. And what’s more, rather than being a fate to bemoan, that’s actually how we all want it. Hell would be what happens when all your desires do get fulfilled, completely, once and for all…)

So, the Symbolic is a limit to the Imaginary: it’s where the apparent unity given us in the image reaches its limit. The unity of the image is always illusory. The child sees itself as a whole in the mirror but it does not see a whole; what it sees is incomplete, parts of a body, and what makes what it sees visible as an image is that very incompleteness. There are many parts of the body the child does not see in the mirror, including aspects which are perfectly visible to others. The image is an image only because it does not show all; because what it does show takes on its intelligibility as image only if it is understood that there is an unseen aspect as well. The paradoxical lesson of Lacan’s famous article on the mirror phase (Écrits) is not the obvious one of the primacy of the Imaginary; it’s that the Symbolic is necessarily there beforehand…

But we can also reverse things. A Gestalt may invert itself, so that the vase becomes two faces and vice versa. There’s a peculiar sort of projective geometry at work here (in projective geometry, all theorems run in pairs: wherever a statement is true, you can generate a complementary true statement by exchanging the words point and line). If the Symbolic is a limit to the Imaginary, the Imaginary also functions as a limit to the Symbolic – and as Lacan will develop in his later work, both of them have another limit, one which will have asymmetric effects on both: this is the Real, as shown in the well-known schema of the Borromean rings.

                                            Venn diagram

And just as the interruption of the Imaginary by the Symbolic can be thought of as exactly what Freud calls castration, so too we can find a clear Freudian name for this Gestaltic inversion, whereby the Symbolic is interrupted by the Imaginary. In the second Seminar (1953-4), The Ego in Freud’s Theory and the Technique of Psychoanalysis, Lacan will say that this imaginary function of the ego is nothing less than resistance. (Why does the analysund resist? Because the ego does not like where the symbolic chain of associations is leading.)