Return to SCE Homepage
Return to Session Program

Autism and Representation
Case Western Reserve Univ
Cleveland, Ohio
28-30 October 2005



Michael Turnheim
Vienna, Austria

Do not cite without permission of author

Killing Me Softly (Autism, Time, and Representation)

There are certainly many different ways to speak about autism. The aspect which I have chosen to speak of today is that of speech and representation. Autistic persons speak in a very special way and some of them don't speak at all. I will try to relate this phenomenon of autistic speech (which I will have to describe more closely) to a very general thesis (which I will have to justify): that there is a relationship between language and violence. This thesis stems from psychoanalytic theory and it also occupies an important place in Jacques Derrida's texts. I'll start with a short explanation of the reason why it is possible to affirm the existence of such a relationship. To avoid misunderstanding, let me say right now that the choice to discuss speech does not mean that I consider autism as a simple problem of language.

Speaking is not just the expression of ideas or feelings nor is it just the description of objects or situations.1 Speaking is not simply the exteriorisation of something which already exists inside the speaker and it is not simply the image of something which already exists outside in the world. Speaking creates something which did not exist before and for this reason it is, in the first instance, an act which engages the speaker. It has consequences for the one who speaks. In a very global way we can say that the speaker has to assume what he says, that he has to take responsibility for it. In this sense, the effects of speaking (or at least of a certain type of speaking) come from something which is outside the speaker and which is at the same time his or her own creation. Words, once they are pronounced, function as if they were detached from the speaker and act on him like an exterior event. We can note here that this detachment is exactly what Derrida wanted to stress when he spoke of a function of writing logically anterior to the spoken word. I will return to this question of writing later.

To understand the consequences of speaking, we can start with the function of names or with the use of personal pronouns like \x93I\x94 or \x93me\x94. Let us consider something very simple: Names and pronouns are words and as words they function in a language which is constituted of elements. Since the number of elements of a language is necessarily finite, words have to be repeated. An element which would never be repeated would not be a word and in this sense something which happens in or through language can never be a pure event \x96 it can never be absolutely unique.

What is important now is that such repetition must necessarily create a tension, a tension between the supposed singularity of what is designated by the word and the repetition of the word, repetition which is, by definition, incompatible with singularity. It is in this sense that Lacan could write that \x93the symbol manifests itself first of all as the murder of the thing.\x942 If the word somebody else uses is my own name or if I myself use the word \x93I\x94, the simple possibility of the repetition of such a word creates a tension between my supposed or desired singularity and the necessary destruction of such singularity through repetition. I would like to be unique, but by the simple existence of language I loose that supposed singularity - a singularity, by the way, which has never existed. It is through this kind of loss or tension that speaking is linked from the beginning to violence. Since my name or the word \x93I\x94 are, like every word, used more than one time, I must accept the destruction of my supposed singularity. There is thus a sort of radical impossibility of the proper name \x96 the two terms \x93proper\x94 and \x93name\x94 are, at a certain level at least, incompatible.

Now, it is easy to see that the situation is even worse than that. I do not only loose my singularity through language, but the identity which I can acquire through naming will as well never be stable. Even if this constitutes a general law, it is easier to show it for a special type of sentence. Let's take the case where I say to a woman: \x93You are my wife.\x94 The interesting thing in this case is that I have said more than I thought, that I have said something without saying it, namely: \x93I am your husband\x94. Even if it was me who pronounced the first sentence, in addressing to the other I receive a new sentence from outside. The sentence \x93You are my wife\x94 creates a new situation in which I am no longer a bachelor, but a husband. After having said \x93You are my wife\x94, I must accept or assume to be what I have become through my act of speaking.

This shows that there exists a sort of after-power linked to speaking, a deferred action (in the sense of Freud's Nachträglichkeit ) of speech. On the one hand, this deferred action of speaking destroys my supposed singularity. On the other hand, it destroys the identity which I thought I had acquired through anterior naming. This deferred action of speech leads me to take a further step: there is not only something violent in naming, as I said before, but this violence seems to be linked to time.

It is important to add that you need theoretical thinking in order to be able to postulate the existence of such violence. In other words, it is a violence that is normally forgotten \x96 no normal person would spend his or her life complaining about the violence of language. One can even say that what we call normality depends in an essential way on the forgetting of the violence linked to speech. And, on the other hand, it might be possible to explain at least some of the so-called psychopathological phenomena as an effect of the failure of such forgetting.

If I try to continue to consider this link between speech, time and violence, I can say that \x96 at a certain level at least \x96 it would the best if I were \x93nothing\x94, if I were already dead. In order to be able to support the necessary violence of speaking you have to be able to forget your own \x93ego\x94, to be able to bear the destruction of what would be a fixed identification of your own person. This does not mean that you don't have to have an ego. It means only that it is advantageous, even necessary, to be able to drop the ego at certain crucial moments of your existence. This is especially true for \x93strong\x94 propositions (Lacan opposed \x93full speech\x94 to \x93empty speech\x94) where it becomes evident that you are not going to be the same person once you have pronounced them. The somewhat traditional, but nevertheless interesting, examples Lacan uses here are two sentences, one of which I have already referred to: \x93You are my wife\x94. The other is \x93You are my master\x94. In both cases the speaker becomes somebody new through speaking, through his own words \x96 in the first case he becomes a husband, in the second case a disciple. The further step which I have introduced is that, in order to become a new person, it is necessary to be able to forget, to be able to drop the old ego. In a certain sense, to become a husband, the bachelor someone has been up to now has to die. The new ego is like a graft which has to be accepted without too strong an (or with a minimal) auto-immune reaction. But there would also be cases, so called pathological cases where there is a very strong reaction, and sometimes even a reaction so strong that it produces a breaking down of a life situation.

All this must sound very abstract. To make it less abstract, I'll give you a clinical example. For the moment, I am not yet speaking about autism, I am speaking about psychosis and I do so because I think that it will be interesting for an understanding of autism to compare it with psychosis. There is a very famous case of psychosis, the case of Daniel Paul Schreber who wrote an autobiographic book, Memoirs of my Nervous Illness , about his own mental disease which was commented on by Freud and Lacan and later by many other psychoanalysts. Schreber was a German judge who had a very important professional career at the end of the 19 th century. During the first half of his life he seemed to be more or less normal. But at the moment when he was nominated President of the court of justice he fell into madness. Looked at from this angle it seems that it was his nomination, his new name or title, which produced the catastrophic destruction of his apparently normal pre-psychotic world. Or \x96 to say it in a more technical way \x96 the beginning of his mental disease seems to be linked to the collapse of a pre-psychotic identification which, even if it revealed itself finally as having been insufficient, had allowed him to live more or less normally up to this moment. Compared to what he was, compared to his former \x93being\x94, his nomination constitutes an anticipation of the future: you are going to be President. Even if this nomination is what would normally be considered a success, it is also \x96 if we accept what I said before \x96 something violent. Schreber's nomination implies a new responsibility and, to be able to bear this, the ego has to change. The ego does not simply have to get bigger, but rather the old ego has to disappear to make place for a new one. To be able to become , to become something or somebody new, you must first of all be able to be nothing. And if you are not able, not able in an absolute sense, your world breaks down, which means that you go mad. In other words: Schreber's problem was that he had nothing but an ego, that he did not have at the same time an idea of his own nothingness, of his own death. His pre-psychotic identifications are revealed as being insufficient for a certain life situation, a situation of change.

We begin to see a specific temporal structure. Language allows us to go beyond empirical existence. It anticipates something which does not yet exist and requires us to be able to assume it subsequently and, if it concerns our own existence, to become it. Schreber was not able to be what language required him to become. It is very interesting that we find in Schreber's Memoirs that this same time structure manifests also itself at the level of speech itself, at the level of the sentence. Schreber describes in his book how God puts parts of sentences into his mind, strange sentences like: \x93Now I will face the fact that I am an idiot.\x943 The fact that the first parts of these sentences appear to him as hallucinations indicates that there is something in them which he is not able to bear. They don't become his own; they stay in the other's speech. The important thing here is that \x96 as Lacan has stressed \x96 the part of the sentence which can not be borne is always the part which contains personal pronouns, what linguists call \x93shifters\x94.4 This means that we can observe the same structure acting at different levels \x96 at the microscopic level of interrupted sentences and at the macroscopic level of life itself which breaks down. All this indicates that Schreber and - if you allow me this rather rapid affirmation \x96 psychotics in general, do not support the violence linked to language. And for this same reason they make us aware of the existence of such a violence which, as I said, is normally forgotten.

Viewed from this angle, psychosis is liberty of speech. Psychotics are speech virtuosos. Language allows them to advance into regions of speech where effects are produced for which they are not able to take responsibility. This it what distinguishes them from normal or neurotic persons. They are victims of a sort of time loop in which they get lost. There is a tension between a level where a technical mastery of language exists and another level \x96 psychoanalysts would say: the level of the unconscious \x96 where the deferred effect of what has been said should be, but is not borne or assumed.

If we try to say it in a more technical, psychoanalytical way, we come to the following formulation of the situation: What we call normality seems to be linked to the possibility to bear the momentary dissolution of the speaker's own representation, of his ego. Psychosis, on the contrary, or at least certain forms of psychosis, seem to be linked to an absolute fixity of the ego. There is an ego which allows the speaker to create symbolic entities which go beyond an existing situation, but the subject is not able to assume the deferred action of such a process. The deferred action of speech triggers the outbreak of psychosis, triggers the transition from the pre-psychotic to the psychotic state. And it is also responsible for speech phenomena like Schreber's interrupted sentences.

If we want now to compare autistic speech with normal and psychotic speech, it seems best to begin with those autistic persons who speak, who even speak a great deal, but in a very special way \x96 the so-called Asperger cases. Oliver Sacks for example writes that they show \x93verbosity, empty chatter, cliché-ridden and formulaic speech\x94.5 Others stress the existence of difficulties on the pragmatic level of speech. Autistic persons tend to misunderstand the pragmatic context. When they hear for example the question \x93Do you know where Mommy is?\x94 they simply reply \x93Yes!\x94 Autistic speech is also marked by a mechanical aspect, by a lack of intonation. It is also generally recognized that autistic persons have specific difficulties with the use of personal pronouns.6

My thesis about autistic speech is linked to what I have said up to now about psychosis. I have said before that psychosis is liberty of speech and that psychotics are, in a certain sense at least, speech virtuosos. When we hear autistic speech we have, on the contrary, the impression that the modification consists in an extreme limitation of speech productions. The very simple formula which I would like to propose to you to explain this contrast is the following: the temporal loop which is at the origin of the psychotic catastrophe and of certain specifically psychotic speech phenomena, is never produced in autism. The often-remarked poverty of autistic speech is in the first instance a way of speaking in which the deferred action of language - which in \x93normality\x94 as well as in psychosis touches in two very different ways the identity of the speaker himself - can never occur. In a certain way autistic persons are like normal persons \x96 they don't produce sentences which result in consequences they would not be able to bear. But of course there is a radical difference from the beginning, the fact that the sentences produced in autism are exactly those which don't have any subjective consequence at all . This could explain why autistic persons like Temple Grandin are attracted by science \x96 if one considers that science is by definition a \x93de-subjectivated\x94 discourse.

If we turn again to the comparison of autism and psychosis, there is another very important difference between the two. In psychosis there is always a \x93trigger\x94. There is a specific moment in life at which one can say that psychosis \x93begins\x94, sometimes very late in life, as it was the case for Schreber. In autism the situation is different. Even if in some cases of autism there seems to be a sort of regression after a very limited period of development, in most cases autism seems to be present from the beginning of life or at least extremely early. From the point of view which I propose here, this means that autism is marked by a specific temporal structure from the beginning. There is not, as is the case in psychosis, a pseudo-normal period supported by some primitive identification which later breaks down. There is no breakdown because anticipation as such does not seem to work or works only in an extremely limited way. In a very essential way autism does not go beyond some limit. So as it was already the case for Schreber, we can here again say that the sentence and life in general have the same temporal structure. Even if every sentence is developed in time, autistic speech seems to be a way of speaking which excludes the deferred action of meaning. Autistic persons never produce sentences which could transform or abolish some identification. And for this same reason they never enter into \x93existential\x94 situations which could produce a general breakdown as we can observe it in psychosis.

Now there are of course several questions linked to my thesis, essentially two. First: why does this happen? And second: what is the relationship of this temporal structure to the other phenomena of autism?

I will certainly not take the risk of pretending that I know the cause of autism. Almost everybody says now that it is biologically determined. I don't know. All I want to do is try to push the analytical explanation as far as possible without attempting to fix an ultimate cause. For such an analysis, the theory of Melanie Klein's disciple Donald Meltzer seems to me quite convincing.7 What he postulates is that a very early mechanism of normal and psychotic development, the so called projective identification, does not work in autism. What Melanie Klein means by the expression projective identification is that there exists from the beginning some inner \x93bad object\x94, in other words an inner heterogeneity, an inner corpus alienum which, in the imaginary world of the child, is normally expulsed or projected. The effect of this is that the bad object is now outside and that it creates persecutory anxiety. In other words: the price to pay for cleanness inside is the existence of badness and persecution outside. Even if there will be later, through the effect of the so-called depressive phase, a sort of normalization, for Melanie Klein everybody passes through a paranoid phase during which he is the object of the persecution of the other. The first \x93theory of mind\x94 would be a sort of primitive paranoia.8 Now the important thing here is that even if this transitory paranoia creates something painful, at least the subject identifies with something. It becomes something in his relationship to the other. This means that the production of a first identity or of an ego depends on persecution. For Meltzer the problem with autism is that this whole process does not work. In other words \x96 the problem is that autistic persons never go through this primordial paranoia. Since they have never created an enemy, they have never been persecuted and, for this same reason, they don't have an ego.

What one can say now is the following. Psychosis \x96 or at least a certain typical form of psychosis - is marked by the fact that, due to projective identification, an ego exists, and the problem is that there is nothing but an ego. The primordial persecution is produced but the development goes no further. What is called megalomania would be one of the possible expressions of such a predominance of the ego. The existence of the ego in psychosis permits the anticipation, the time-loop, the formation of the sentence and especially of certain \x93existential\x94 sentences which, by deferred action, would require the destruction of this same ego. If this destruction can not be accepted, anticipation produces the typical psychotic breakdown. If we follow the Klein-Meltzer theory, we can say that in autism there is no ego at all and, for the same reason, there is no anticipation, no breakdown and no interruption of sentences.

In a very general and schematic way we can thus distinguish three possibilities: 1. Normality and neurosis: existence of an ego allowing anticipation plus ability to assume the consequences of such anticipation 2. Psychosis: ego allowing anticipation without such an ability to assume 3. Autism: no ego, no anticipation, no necessity to assume whatsoever.

Maybe you will have the impression that I have chosen a sort of linguistic approach to autism. But I do not think that this is the case. I have tried to speak about the conditions of different forms of speech rather than to speak simply about speech. In trying to trace back as far as possible these conditions of speech, I have referred to Melanie Klein's theory. I have said that the absence of original paranoia in autism is due to the fact that some inner corpus alienum has never been expulsed. This means that the autistic structure is characterized by the persistence of an inner surplus or excess. In a very general way we can say that this surplus of excitation, which never gets outside and consequently never gets linked to language, can explain many autistic symptoms beyond the typical speech phenomena. It can explain a specific irritability of autistic persons, the fact that they seem to be exposed without any defense to situations or stimuli which normally would be borne without any problem. In other words \x96 the level on which I try to analyze autistic phenomena is the level of some primordial partition between a marked and a non-marked space, between language or representation and something else. The autistic speech alteration would already be a consequence of some particularity at the level of this original partition between language and something else.

The other not simply linguistic aspect I would like to stress is a specific modification of time in autism. Normal speech presupposes a certain unfolding of time which does not exist from the beginning. Speech is a sequential phenomenon - one word comes after the other. As you might know, for Jacques Derrida the priority which occidental thinking accords to the spoken word, to language already unfolded in time and to sequential phenomena in general, belongs to what he called logocentrism and phonocentrism. It belongs to a tradition \x96 our own tradition - in which writing is considered to be a secondary picture of temporally unfolded speech, of speech already unfolded in time. But it is evident that such a conception of writing is linked to a whole tradition and especially to alphabetical writing presented as a secondary mapping of events which appear already in a sequential order; and it is also evident that this is not the only way of writing. Pictograms, like in Chinese writing, are not sequential mappings of sounds and this allows us to think that writing in its most general way is not simply sequential, that \x96 from a certain point of view \x96 it is situated before what we usually call time, before time as a sequence of events. What Derrida calls trace is in a certain sense pre-temporal writing and it produces violence though the simple nomination of something or somebody, through the entry of the unique in a system. This means that there must be a sort of original, unthinkable knot between writing and violence (Derrida speaks of \x93archi-writing\x94 and \x93archi-violence\x94), and all the rest comes only afterwards \x96 sequential time, normal speech, etc. So I think \x96 and I am not the only one \x96 that autism is first of all a time problem.9 What I have said before about the specific difficulties autistic persons have with deferred action in speech can be interpreted, in a larger way, as an effect on speech of a fundamental problem with time in autism. Time has to be opened and this opening is violent. There has to be a sort of deal \x96 I accept the violence of the unfolding of time and I get something else. This deal does not work in autism, it is refused.

To avoid dogmatic statements, let me just add that there is of course no complete absence of time in autism. There is an extreme poverty of time which we can observe on many levels \x96 not only in speech but also in drawing. If you draw something, you first have a sort of project, a draft which will orient your successive actions. Since in autism there is a difficulty with time, the drawings sometimes become very strange. Rather than the realization of the original project we see a progressive destruction of what seemed to be the initial intention. There are isolated meaningful actions which do not seem to be linked to what was before and to what will come after. On the level of speech there are strategies which help to be able to survive in a social world constituted by a certain unfolding of time, strategies which autistic persons like Donna Williams describe in an interesting way \x96 for example speaking by imitating whole sentences spoken by another person (\x93echolalia\x94), which means that the unfolding of time is borrowed from the other, ready-made.10 And the last thing which I want to stress is that, of course, this poverty of time does not mean that autistic persons simply live in present time. Present time never exists. It is always already lost through representation \x96 even before you speak, even before you have been born, the other has already spoken about you which means that you have already lost both supposed singularity as well as full present time. The problem with autism is that it never gets far beyond that loss, that autistic persons are permanently confronted to this primordial, normally forgotten loss of present time and of singularity. This means that the autistic world is not simply based, as one could think, on the avoidance of violence in general. It seems rather linked to the fact that the subject never gets beyond a first violence and that the immediacy of the pain caused by original loss never allows autistic persons to enter in the very complex structure of representation, time and violence which I have tried to present to you.


1. As you know this has been stressed in an interesting, if - in a certain sense - also problematic, way by Austin in his work on the performative aspect of language.

2. LACAN, \x93Function and field of speech an language\x94, in Ecrits/A selection , New York , Norton, p. 104

3. Daniel Paul SCHREBER, Denkwürdigkeiten eines Nervenkranken, Leipzig, Oswald Mutze, 1903, p. 217 sq.

4. LACAN, \x84On the possible treatment of psychosis\x94, in Ecrits/A selection , op.cit., p. 186

5. Oliver SACKS, \x93An Anthropologist on Mars\x94, in An Anthropologist on Mars , London , Picador, 1995, p. 234

6. Simon BARON-COHEN, Autism / The facts , Oxford , Oxford University Press, 2003, p. 47 sq.

7. Donald MELTZER, Explorations in Autism , Old Ballechin Strtah Tay , Clunie Press, 1975

8. Cf. Michael TURNHEIM, \x84Autistische Geistesblindheit\x93, in Das Scheitern der Oberfläche , Berlin and Zürich, diaphanes, 2005, p. 13-40

9.Cf. Peter FUCHS, \x84Kommunikation und Autismus\x93, in Die Umschrift , Frankfurt am Main Suhrkamp, 1995

10. Donna WILLIAMS, Nobody nowhere , New York , Avon Books, 1992





Return to Top of Page

Return to SCE Homepage
Return to Session Program