If there is one time and place to which I can allocate the origins of the concerns of this book, it is the occurrence of a dialogue at Murdoch University between one of my students and myself over Saussure's definition of a signified as a concept. I chose an example from film, specifically Andy Warhol's eight-hour uninterrupted shot of the Empire State Building in New York. (I did not claim to have seen it: conceptual art is always more fun to hear about than to experience.) The signified, at least the denotative signified, of this shot was, I argued, of the concept of the Empire State Building. Had I seen the film I might never have said this, but even so my rationalization of this view fell dead upon my own ears: of course the denotative signified of this shot was the accursed building itself, although connotatively it may, as the title suggests, hold all sorts of meanings about American imperialism and hubris in constructing such a monument to 'manifest destiny'.

Accordingly, it was not long after that that I became interested in Peircean semiotics, rather than the logomorphic and logocentric semiology (the distinction is Eco's) of Saussure and such followers as Roland Barthes. Somewhat late I came to realize that the latter were using 'denotation' and 'connotation' in a way that ran counter to the philosophical usage, at least on the surface. The clearest statement I know of the latter is that of the Polish logician, Kasimierz Ajdukiewicz: [1]

Every meaningful sentence, whether true or false, states something. What is stated by a sentence is called in German 'Sachverhalt'. A literal translation of that term, e.g., 'state of things' or 'state of affairs', is seldom used in English. In the latter language what is stated by a sentence is usually called a 'proposition'. In this sense, a proposition is neither a linguistic expression, nor a psychological act of thinking, nor any 'ideal meaning', but something that belongs to the sphere of objects to which a given sentence refers.

To continue Ajdukiewicz's exposition,

every name [2] (i.e., such a name which may be used as a predicate in a sentence and consequently may be called a predicable name as distinguished from a proper name) denotes its extension or denotation, and connotates its intension or connotation. The extension or denotation of a predicable name is the class of all individual objects with reference to which that may be truthfully used as a predicate.... The intension or connotation of a name is a set of properties whose conjunction holds with reference to each of those, and only those, objects which are elements of the extension of that name. In other words, the connotation of a name is a set of properties which univocally determines the extension of that name.

A little earlier, the author writes:

The need for a definition of the proposition can also be seen in connection with the puzzling question, what is stated by a false sentence.... The difference between a meaningless sentence and a false one consists just in that the former states nothing while the latter states something - which, however is not a fact. The definition of proposition should provide an answer to the question, what is stated by a sentence, whether true or false. Such a definition would have to point to a kind of entities which in the case of true sentences may be called 'facts'.

The distinction between such propositions which are facts and such which are not is the same as the distinctions between true and false propositions.

Such propositions are conceived as 'objective'. I take exception to the word 'objective' because it begs the very serious epistemological question of what it is for a statement (not 'sentence' or 'proposition') to be unqualifiedly true, supplying already the 'hard' realist analysis of it. I prefer Richard Rorty's formulation of truth as 'characterizing the view which would be agreed upon as a result of argument undefeated by irrelevant considerations', which is roughly Peircean Consensualism. It is also what I call, going back once more to Peirce, fragmation. As Rorty points out, 'The two are largely coextensive in philosophical discourse (i.e., "objectivism" and "intersubjectivism")', and that to me smells of disaster for an unbiased look at 'discourse' itself. [3]

This may also be the place to exorcize the valorization of the work of Ferdinand de Saussure, whose 'theory of language relies implicitly on a rationalist theory of meaning and consciousness, since it rests on a notion of signs as representing ideas which precede any actual utterances and are, consequently, timeless and context free'. [4] The truth or falsehood of entities, then, are properties of objective entities and as such belong to the objective world. They may be understood, therefore, to be those mysterious objects which Frege called 'truth' and 'falsehood' and related to sentences as their denotations.

Similarly, as Ajdukiewicz points out, 'it is our intention to give such a definition of proposition which conceives it as an objective, and not a linguistic or psychological entity, connected with the sentence by a semantic, and not a syntactic or pragmatic relation, viz., the relation of stating'. [5] And later: 'We might speak not only of denotation and connotation of names, but also of denotation and connotation of other expressions. Further, one might try to generalize them so that the denotation of sentences would be, in conformity with Frege, their truth-value, and the connotation of sentences would be an entity which we might call the proposition'. [6]

While diametrically opposed to Frege on the nature of the furniture of the world, I thought it worthwhile, as well as delineating the philosophical use of 'denotation' and 'connotation', to introduce the notions of sentence, proposition and statement, although I agree with Foucault that while meaning attaches to sentences and truth or falsity to propositions, statements have as their correlates the essential discursive formations.

And as a way of bringing opposite factions together, I add to this introductory chapter a thought from Barthes: that denotation is the last of connotations. Or in the words of Christian Metz: 'In summary, a way of filming is always the place - the matter of expressions, not in the code - where the signifier of connotation and the signifier of denotation overlap: the first, because there is one or another way, the second, because one films the object in all kinds of ways'. [7]

My own view, as it will be developed, is already implied in the bivalence of statements both of intension and of extension. Following Louis Hjelmslev, my project is a bifurcation of both signifier and signified into their formal and substantive aspects, as in the figure below:

	Expression plane	Content plane
Form	Structures of technique or 'style'	Thematic structure
Substance	Materials of the medium	Diegesis-producing content

This is a diagram, let us say, of the components of a fiction film, but such a quadripartite model will recur in this dissertation, always taking expression, or the signifier, or enunciation, or narrating, on the left-hand side and the signified, or enunciated, or narrated, on the right. I do not of course regard all these concepts as interchangeable, but at base they do come down to the distinction between expression plane and content plane, which replaces forever, one hopes, misleading oppositions of form and content as correlates.

The contention, then, is that Saussure's semiology excluded the extensional aspect of the signified and in so doing removed semiology from the actual world of utterances and readings, works and texts. And it is at this point that I stress the latter distinction, because while the text is an intersubjective possible referent of discourse, what I am calling the work is not; the latter is closer to Ingarden's notion of the 'aesthetic object', which was rigorously to be distinguished from the work of art, whose virtual structure I take to be that of the text outside any reading(s) of it. [8] The major problem of dispensing with the referent as Saussure does is to give an account of the intentionality (not just intension) of utterances/texts in the absence of their goals. Moreover, 'intension' (conceptual) and 'extension' (referential) are logical correlates like 'good' and 'bad', since if we do not have at least the concept of each we cannot explain either one, much less define it. To put this another way, the two are dialectically related, so that a change in one logically entails a change in the other.

This is true of both language and discourse, which usually but not necessarily makes use of language. And I shall draw four distinctions between them in the body of this text. But if an act is an act of discourse, then, as Christian Metz points out, we tacitly imply a 'speaker', the author of the discursive act. Incidentally, I shall throughout use 'discursive act' where Jrgen Habermas would use 'communicative act' and ordinary language philosophers in the tradition of J.L. Austin, 'speech acts'. Habermas sees that not all communications involve speech, and so broadens the concept; I do not see the intention or the fact of communication in meaningful acts, and so will broaden it still further.

I further see the use of a concept of intradiscourse, as when one admits something previously refused to oneself, to go with both inter- and extradiscourse. As Metz says, in discourse one always infers a speaker, which is one of its differences from language, and in consequence do not even view a film as a perceptual experience that is 'just there'; this is 'speech', so there must be a speaker. [9] This is to say that we view even the most realistic of films as artefacts, even if we do not infer a real artificer: it is merely a fact of cognition, but notably of abductive, not inductive or deductive derivation. [10] Thus we are 'spoken by' our discursive systems; I hope to show that reciprocally we also 'speak' them.

This view of reference, I believe, does not fall into the half delineated by Eco, of 'referential fallacy', because at no point is extension to be compared with intension; but we need both for intelligible discourse, since to understand a concept entails in part knowing what counts as an instance of it. Note at this point also the notorious and dubious use of 'null denotation', to which I shall be returning. For now, we turn to Ajdukiewicz: [11] 'there are names which are composed of other names, but not on the conjunction principle. E.g., the name "circular or red" is based not on conjunction but on alternative. The connotation of the name "circular or red" may not be considered to be the set of properties of circularity and redness. Should one think so, one would ascribe to the name "circular or red" the same connotation as to the name "circular and red". Since these two names differ in their extensions, that is, in their denotations, this would be in contradiction with the principle that the connotation of a name univocally determines its extension'. [12]

Despite all this, some theorists, like Geoffrey Leech in his Semantics, confuse 'denotation' with 'conceptual meaning'. [13] Stuart Hall clouds the issue further by saying 'the term "denotation" is widely equated with the literal meaning of a sign'. But is the 'literal' its referents? -

because this literal meaning is almost universally recognized, especially when visual discourse is being employed, 'denotation' has often been confused with a literal transcription of 'reality' in language - and thus with a 'natural sign', one produced without the intervention of a code.

There is also the possibility for the Rationalist unwilling to clutter up his system with items from the real world, that, while 'extension' must be admitted as a correlate of 'intensions', nonetheless it is empty, for there can be classes without members, or whose members are not mentioned. This view is attractive to me because it narrows still further the distinction between factual and fictional discourse. Perhaps we do ostensively refer in certain instances, but I am suggesting that this is not the first task of denotation, which is to participate in the creation of possible worlds. 'Connotation, on the other hand, is employed simply to refer to less fixed and therefore more conventionalized and changeable, associative measurings, which clearly vary from instance to instance and therefore must depend on the intervention of codes'. [14] One wonders at the usages of 'literal' here, but that is not the main issue: what Hall has inadvertently done is to provide an account of linguistic/discursive indetermining. He goes on: 'We do not use the distinction - denotation/connotation in this way. From our point of view the distinction is an analytic one only'. It is rather like 'literal'/'metaphorical' in that respect, as we shall see. Moreover:

It is useful in analysis, to be able to apply a rough rule of thumb which distinguishes those aspects of a sign which appear to be taken in any language community at any point in time, as its 'literal' meaning (denotation) from the more associative meanings for the sign which it is possible to generate (connotation). But analytic distinctions must not be confused with distinctions in the real world. In actual discourse, most signs will combine both the denotative and the connotative aspects.... It may, then, be asked why we retain the distinction at all. It is largely a matter of analytic value.

In fact, as I have said, the distinction is indispensable for the perception of discourse even if x, the denotation on one occasion, appears as y, the connotation, in a different context. The level is that of everyday discourse and is predicated upon intentionality, even if, at a 'higher' level neither of the denotation/connotation pair is taken to be more important than the other. This emerges with particular clarity if we understand connotation to be intension and denotation to be extension, or the contract may be drawn as concept and referent. The latter, however, glosses over an important fact of connotation: it refers. But does it refer to a class or to the members of that class? To embrace the latter of those options seems to me to be to fall into what Eco calls the Referential (or Extensional) Fallacy, while to acknowledge as Saussure does only intension, might with justice be called the Intentional Fallacy, and one that totally ignores the pragmatics of language, or its status as discourse. Assuredly, there is no intentional ghost in Saussure's machine, but then it is not going anywhere either.

So we are left with intensional and extensional classes, and it is one of the polemical aims of this thesis to show that no extensional class is empty, which is the same as denying Nelson Goodman's notion of 'null denotations' when the denoted is fictional. I shall be adverting also to Foucault and Peirce on this question later. At present, all I want to say is that this places a premium on certain aspects of existence that may be less exclusive and more commonplace than, say, a rugged Empiricist, fresh from the cricket pitch, would allow. As we shall see, substance is not a quality by itself (for Hjelmslev), but rather the workings of forms on matter, and it may be that in some ways an imagined matter is preferable to the quotation.

A little reflection on that much-used but near useless term 'literal' might have disabused Hall of the notion that connotations are not logically necessary to denotation, being as it were the other face of the linguistic coin. But it is true to say, with Hall though not in his sense, that they are 'useful tools for distinguishing, in particular contents, between the presence/absence of ideology in language [and also] the different levels at which ideologies and discourses intersect'. [15]

It is true, as I have admitted a propos of indeterminacy that there is an inner, private aspect to connotation, and it is perhaps that which gives discourse its thrust and interest. Perhaps that is why intersubjective and actual, private and public, empirical data are necessary if we are to think through someone else's innermost workings. This is language at work, and it requires a full-blooded pragmatic approach involving others' intentions as well as intensions if we are ever to understand it.

And, as I have said, discourse always implies 'discourses'. It is in the search for those persons that, as a welcome by-product, we sometimes communicate, even though we cannot know when. But there is the irreducible trace that all discourse contains a message about how it is to be taken. To avoid pedantic overkill, I will not preface each instance of 'communication' with 'meta-', but that is only because it is universal: all discourse is meta-discourse, consciously or not. [16]

That is, it is ontically preferable on the edge of Ockham's Razor, to generate our criteria of numerical identity from possible world semantics rather than a semi-medieval concept of substance. On the other hand, it is useful to a theory of fictional intentionality to seem to be dealing with real objects, because as we shall see the line between documented fact and fiction can be so fine as to vanish when it comes to defending one's extensions as actual, like Burke and Hare, rather than fictional, like Sherlock Holmes. I choose Holmes deliberately because of the cult that has grouped around the stories concerning him, or rather around Holmes himself, since no devotee is permitted to doubt the onetime actual existence of their hero. And I will go so far as to say that Sherlock Holmes is more real than the vast majority of his contemporaries (that is, the people who were alive at the time of Holmes' mythical - in every sense - existence). And it is the Real not the actual that impels people to act, since what is actual but fails to become real may as well not have happened except for the actual and real events it may have been a necessary condition of.

This might be summed up in Foucaultian terms if we allow that sentences concern meaning, propositions truth and statements 'discursive formations'. I am taking the last to be the Real and the propositional its actual. At least in theory this is the case, but Foucault also insists that every sentence implies a statement and therefore the Real, as no denotative class can be empty because it will be filled by the Real, whatever the status of the actual. This is to say also that speech is always discourse and therefore self-fulfilling or self-actualising, and will become empirically so in the behaviour of those who believe it. Thus does the discursive practice become, through its 'positivity', the Real and even at last the literally actual.

Perhaps the dialectically interactive nature of intension and extension which ensures that if one is invoked, so is the other - contrary to some received ideas of a reconstructed Saussure - is best argued for by example. Andrei Tarkovsky's astounding film A Mirror (1975) may be the first truly subjective film, since without knowledge private to Tarkovsky one cannot relate to the extension of the film in any of the established Hollywood-illusionist or avant-garde structuralist ways, because he has simply either re-staged or made up, but at any rate formalized, events from his own past in such a way that not only are the figures or literalisms that occur in the film placed on the paradigmatic axis of discourse, but a recognizable context is likewise often absent. Thus the viewer finds the film extensionally opaque, although he/she may be reminded of some personal event: it is the job of A Mirror to interact morphogenetically with whomever is reflected in it and its discourse is an open subject requiring a mirror system for the creation of meaning, but not of film. I have argued elsewhere that the subject of films is involved via identification, thus recreating the mirror phase of Lacan, which is doubly imaginary in that one misidentifies oneself and manufactures desired feedback.

With A Mirror this procedure is out of the question, since we are deprived of both narrative and content and must create out of the whole cloth of one's consciousness a relation that it is not sharable with other members of the audience; it (the film) is a private experience because of the absence of the usual marks of attempted communication, and one either takes it or leaves it, although symbol-hunters may be well advised to stay away: the film is 'incomprehensible'. In their terms the Soviet bureaucracy predictably reported it as formalist, as too subjective and too expressive for the pedagogic purposes of Socialist Realism).

A Mirror in this respect is like music without Tchaikovsky's cannons (or canons). My point, however, is not merely to communicate the excitement of a new kind of cinema. A Mirror providentially appeared just as film theory's encounter with Lacan gave it a further metaphor for film: not meaning either within the frame or outside it like looking through a window, but emerging 'between the screen and the spectator', as Godard put it. This has far-reaching consequences for the notions involved in the splitting of the subject. What I want to point out is the consequence for the film's intensions, which are made the more abstract and non-pragmatic (or rather pragmatic but open) by their pure empty form, simply because the extensions are impenetrable.

If Point Blank is a film of the Imaginary, A Mirror's double imaginary structure renders it to the viewer like a language one cannot read - though this is overstating the case; not all of it is subjective, though it is all subjectively viewed - and it is this which makes this seemingly imaginary film symbolic, even though we must assign our own meanings. In effacing himself within the text, Tarkovsky in A Mirror is the voice of the boy who overcomes his stuttering at the beginning and speaks; and what he speaks is the film, an impressionistic and above all subjective collage in which the viewer must create (as in a mirror) his/her meanings.

More mundanely, there is another objection that might be levelled at the dialectical (or correlative) relationship of intension to extension. It arises out of Russell's famous distinction between knowledge by description, where the objects of our beliefs are described to us, and knowledge by acquaintance, where we have a personal, if not necessarily private, experience of the said object. And these modes of knowledge bear striking resemblances to, respectively, knowledge by intension and knowledge by extension. Yet one may occur without entailing the other: therefore they are not logically dependent upon, or dialectically related to, each other.

But let us examine what we may take as a paradigm case of extension without intension, the case of describing colours to a congenitally blind person. We may teach him/her non-visual means of identifying the colours of objects so that he/she can discriminate coloured objects, i.e., have the extensional meaning of them. (And it has been suggested that talk of the temperature of colours, 'hot' reds and 'cool' greens, may actually have a thermal base, so that we cease to be talking metaphorically when we speak of them; or perhaps what we thought to be metaphorical was 'in fact' literal, though not intended to be: and vice versa. The problem of the propositional force of metaphors will become important later in this thesis.) But we cannot make the blind person sense the colours visually. Must we then say that the subject has grasped the extensional meaning of colour-words without requiring the intensional? I think not, since sensations are not the intensions of words, even in the case of 'simples' like 'red', because they always appear in a context and are never simple. What the blind person has learned is how to discriminate between say red and green objects by some other means (which would probably, but not necessarily, be by means of another sense-organ). She or he has, to paraphrase Jay Rosenberg, the use of an inner representation which bears the same relationship to the object as the colour-word stands to it. [17] Or: colour-words do not denote sensations, though they may connote them.

More strongly, I will reiterate the view that connotations presuppose denotations, and equally that denotations entail connotations (of some sort) if they have any meaning at all in use as discourse. In terms of language rather than discourse, a semanticist, like Leech, may imagine that he has a distilled, connotationless meaning, but let him/her use it. A fortiori, there can be no pure intensions either, at least in discourse, because to imagine/think/say anything nominative entails some instantiation of it, no matter how imaginary it may be. In short, we must acknowledge that we must always and as a matter of logic, discourse on the trees when we discourse on the forest, and vice-versa.

Such indeterminacies as remain after we have exhaustively tested our intensions extensionally belong to intentionality, but are in themselves less than 'bits', i.e., they are possible differences which make no difference. Yet I would not take Ockham's Razor to them, because of their potential bodying forth in actions and/or discourse. And, to advert to Smith and Wilson's Modern Linguistics, it is unlikely that two speakers ever use precisely identical 'grammars'. [18]

It may be helpful at this point to create a visual model for the splitting of expression and content planes:

Signifier/Enunciation	------------->	Signified/Enounced
Form of expression	------------->	Form of content
	is entailed by
Substance of expression	------------->	Substance of content

In fact I regard this splitting of signifier and signified and all other binary models as deficient for discourse, because this always involves an interpretant, but as a model of language compared with signifier-signified the enunciation-enounced dichotomy serves its purpose. As long, that is, as it is remembered that the missing intentionalities and interpretants tend to float down on either side of the great disjunctive divide. In this respect witness Stephen Heath's awkward apportioning of them in his Introduction to Barthes' Image-Music-Text - not to mention another blunder I will be adverting to shortly - along these lines:

Every énoncé is a piece of parole; consideration of énonciation invokes not only the social and psychological (i.e. non-linguistic) context of énoncés, but also features of langue itself, of the ways in which it structures the possibilities of énonciation (symbol-indexes such as personal pronouns, tenses, anaphores are the most obvious of these linguistic features of énonciation). [19]

Patently, Heath's conclusions are invalidly drawn since they depend upon an equivocation between language and discourse, each being brought in to shore up the other's shortcomings. To put this point aphoristically: language is a good object for analysis; discourse is a good subject of analysis.

What, surely, énonciation and énoncé are displacing are signifier and signified, not langue and parole. As a measure of the hegomony of Screen writers at that time, no-one to my knowledge has come forward to contest the validity of the distinction, or its appropriateness.

These terms, according to Wilden, permit the argument that we have allowed meaning to be absorbed into signification, possibly as a result of erecting an epistemology on the basis of phonetic oppositions, and continually associate rationality with 'either/or' disjunctive meaning replacing the 'both/and' of analogic thought, leaving meaning in my first sense behind; but this is the subject matter of Chapter Three.

It is at this point that I want to borrow the concepts of intensions and extensions from Carnap (not that he would approve of what I am attempting to do with their assistance) and equate them respectively with intension/connotation and extension/denotation. This is not so revolutionary, since Alexander's primer on philosophy, The Language and Logic of Philosophy, runs them together as a natural course of action. [20]

Difference of usage denotes, however, that there are several other crucial differences between intension/extension and signification/meaning. Most relevant for the purposes of this chapter is the fact that significations and meaning are not conceivable in isolation from one another while intensive and extensive are mutually independent, if still influential. For philosophers and linguists of a binary bent, such a tidy schema as the former must be a godsend: rationalist thinkers and the tradition of Saussure simply excusing meaning as being outside their practice, and too 'intuitive' to be bothered with anyway. Meantime such a 'follower' of Peirce as Charles Morris had produced a behaviourist semiotic which rendered the interpretant redundant, its troublesome inquantifiability simply removed to make the world care for behaviourists.

Both solutions, however, founder on the principle that to know the meaning of a word and not just to know how to behave aptly, we need to perceive imaginatively the difference between a world in which this concept has extensions from one which has not, which I will call the Principle of Reciprocity (of propositional meaning and truth-conditions in a speech as against a statement). Therefore I propose to replace the concepts of signification and meaning with the dialectically linked ones of intension and extension.

Having invoked Peirce and his 'pragmaticist' theory of truth, it is fitting to return to the notion of ternary, not binary, models. Janos S. Petofi takes his cue from Carnap, if not his idealism, and constructs this model:

This is derived from his Text-Structure-World-Structure theorizing, with which I concur at least on the point already argued by Russell and Foucault (in their vastly differing ways) that 'the representation of the intension of a word is always a text'. [21] As can be seen the ternary model rules out behaviourism as the 'Sinn' has more than token value. In practice we can assume that the 'Sinn' is always so filled because there is no such thing as a null connotation: it is self-contradictory to suppose it.

However, authors such as Nelson Goodman in Languages of Art have argued that there may be null denotation, which is to say extensional classes devoid of numbers. He says this because he does not want to say that fictional statements are meaningless, or indistinguishable from one another because they all have no referents. [22] David Novitz wants to deny denotation to such 'utterances' as paintings altogether, saying that they do not denote but resemble their objects and so 'mean' them. [23] This debate seems to me reminiscent of a sort of cocktail-party fracas in which an artist claims his works are lines and shapes, while a guest who has commissioned the artist to paint his elderly, wealthy aunt insists on their representational links: it is simply two ways of seeing; but Novitz is wrong to think he has refuted Goodman. Of course pictures are more like other pictures than they are like anything else, and one does not have to drag in non-representational art to prove it. And they do denote. But I will quarrel with Goodman that they ever have null denotation, since the existence of the sitter is irrelevant to the appreciation of the painting (except in special, usually lubricious, circumstances). In fact, I want to use some of Goodman's own concepts against him.

He distinguishes generally between allographic and auto-graphic art, saying, most unsatisfactorily, that 'a work of art [is] autographic if and only if the distinction between original and forgery of it is significant.... If a work of art is autographic, we may also call that art autographic'. [24] Thus music is allographic, because it is the score which determines a compliance class in any performance, while painting is autographic. But where does this leave the cinema? As William Luhr and Peter Lehman point out, a film's script is to film more like an artist's notes towards his painting than it is like the execution of a musical score. [25] Thus it is allographic in one way, autographic in another, and unfortunately does not confront the dilemma. [26] But after twenty years or so of debate over the attribution of authorship in a film, it is the director who has, not exceptionlessly but substantially, been recognized, including of course those directors, such as Alfred Hitchcock, who do not write their own screenplays.

Luhr and Lehman insist that there is no equivalent in film to the compliance-class of performances of music, and this would seem to leave the issue permanently unresolved. But perhaps a way out can be found through the work of Horst Ruthrof who, in his book The Reader's Construction of Narrative, re-theorises the usual division of signifiers and signifieds as a presentational process and a presented world which are both significant, i.e., distinct signifieds, constructed from one and the same chain of signifiers. It will come as a shock to learn that Ruthrof is a phenomenologist, but I have not before seen the signifier taken out of anonymous discourse and placed as a signified in the reader/viewer who must realize the text of an author with his/hers, the reader/viewer's own propensities and competences.

The result of applying this to film would be to make the director the provider of a compliance-class of perceptions in ourselves, the audience. It follows that the best, indeed only viable, method of film criticism is for the reviewer to be honest enough to review the work s/he has made of the filmed text, as should have been obvious in any case since, as Raymond Belaer points out, film is 'The Unattainable Text'. So we must, even with our own projection equipment forever be frustrated in catching a meaning on the wing and pinning it down, since the only way to do that would be to halt the projection, with the consequence that film loses what Eco calls its third articulation, which is action or motion, and become something else, a photograph.

This is not, I stress, an abandonment of critical principles to the subjective, and of course the text is viable for all to see (though not from the same angle); what I am suggesting is that we air our private 'works' (or what Roman Ingarden called 'aesthetic objects') in intersubjective dialogue. There is, of course, no 'objective', since this would imply perception without the input of the perceiver, which is impossible. In such a way we can be guided back to the text to see it as someone else has done, or at least make the attempt. Only new and fresh works can be the result.

It is essential at this point to state that I am not, in valorizing extensions, advocating a reference theory of truth, i.e., one in which real objects or their surrogates occur in denotation. On the contrary, I want to say that denotation is not designation or like anything else but contains 'a recursion on a concept like satisfaction or reference'. But these notions we must treat as theoretical.

The trouble with Luhr and Lehman's position, and by extension Goodman's, is that it is absolutist: either a film is art or it is not, and if it is it must have been directed by an auteur, whom the Luhrian critics who started it all conceived of as creating coherently consistent bodies of texts - or rather, a text, since they took up Jean Renoir's remark that 'in his lifetime a director makes only one film' as their rallying-cry, and it certainly applies to Renoir. Peter Wollen in Signs and Meaning in the Cinema dismissed the films of talented but non-auteur directors as 'unreadable palimpsests', a nice example of the Intentional Fallacy sweeping into film theory, only to be retracted four years later in the book's revised edition. [27]

I take the position that in fact all films are art, mostly, as in any other medium very bad art - but undeniably so. What else could they be? Or rather, all films can be art: the realization of the viewer, here abstracting instead of concretizing as he/she would in a novel, is still essential in the hermeneutic circle - or rather spiral, since each twist is self-correcting. As for whether they are allographic or autographic, I again refer to the reading which determines figures and ground, and consider films on a continuum between the two modes, which are not binary and mutually exclusive, but analogic and inclusive.

This would seem to relegate film to the class of allographic arts, but by enlisting Goodman and Luhr and Lehman, I hope to show that it is after all autographic, original or no, in the service of the auteur policy. Goodman, speaking of silent films, suggests that the script is neither the cinematic work nor a score for it but, though used in producing the film, is otherwise as loosely related to the work as is a verbal description of a painting to the painting itself. The corollary to this is that in a sound film only the spoken dialogue would be a notational system.

As Luhr and Lehmann phrase it, 'a film script is not the text of the film as a play script might be legitimately called for the play'. [28] The transforming artist in a film, then, is the director, which was a claim made by the 'auteur theorists' of the 1960's, though it is not the same as the prior claim that a single member of the filmmaking body can be identified as its auteur. The two claims are quite separable, but were confusingly argued for under one heading, viz., the auteur theory (or policy). For example, Richard Corliss' book Talking Pictures [29] is an auteur theory for the screenwriter. The truth is, of course, that it is all in how one views a film, although I believe the most seasoned filmgoer, like the experienced concertgoer, comes to sense, read or understand film in a way that is denied his/her no less vocal colleagues, and that it is this that brought acrimony into the debate. Among the former I include no single living American critic except Andrew Saris, who has himself become rather 'literary' of late. The battle is finally over with a slightly ludicrous victory for the director-auteur critics, a victory which has infiltrated even into Hollywood itself, so that the credit titles of films now read 'A Film by Walter Grassman' or some other hack. Not even the director-auteurists saw all directors as artists: the journeymen (even those of some distinction like William Wyler) were bundled off as mere metteurs-en-scene.

In the intervening period, structuralism had crossed the English Channel and struck at the hearts of Leavisite and other critical orthodoxies. So it is something of an understatement on the disingenuous part of Wollen to mention in passing that he had 'changed his views' since the first edition concerning signs and directors. The new movement came to be called auteur-structuralism and accordingly auteurs were bifurcated Solomon-style into, e.g., John Ford and 'John Ford', the first a person of little interest who made films and the latter valorized as an almost Platonic critical form. When all about were losing their heads, or immolating their human auteurs, it was left to Robin Wood, a man of uncommon common sense, to write of Hitchcock's description of Psycho as 'a lark' and not enjoyable to anyone with a sense of humour, that 'Hitchcock is a greater artist than he knows'. Capping this encomium, John Russell Taylor, author of the endorsed biography of Hitchcock, made much plainer when he said that Hitchcock was not a conscious intellectual-moral giant cramming his films with theological meanings - although they are there, all right - but a genius who simply 'thought in film'. (Michael Powell of Peeping Tom has said this of himself, a striking concordance given the Grand Guigniol trilogy of 1959-60 whose other panel is Georges Franju's Eyes Without A Face.)

This expression is not readily explicable since everyone considers him/herself an adequate film critic, particular those who never go to mere 'movies', and will at the drop of a frame offer 'educated' opinion on the products of the most complex medium yet devized. No-one would presume to do this with music, or admit that it must be learned, the hard way, by sitting through millions of miles of footage not simply listening to the dialogue, admiring the photography or enjoying the performances of favourite actors. Film is always discourse, specifically narration (excepting only the freeze-frame), and we do infer a speaker, as Metz says, and s/he is invariably out of reach, a semiotic blind spot but a hermeneutic subject for an epoché, and so we do attend to 'Alfred Hitchcock', though not necessarily banishing his orator, Alfred Hitchcock, in whose unconscious it all began. He was indeed a greater artist than he knew, as is apparent in his marathon interview with disciple Fran‡ois Truffaut, where Truffaut constantly reveals meaning Hitchcock had not consciously at least planned and the latter, like a basilisk, basks in the sun of this young intellectual filmmaker's worship, only occasionally twitching his tale and seemingly eager only to discuss the technical means by which he achieved certain effects.

This is not the fiasco Hitchcock's 'intellectual' detractors like to think, since all is germane to its Hitchcock-text. To adapt a rather unkind remark by, I believe, Taylor again, on John Huston and Michaelangelo Antonioni, it is perhaps better in film to be intelligent without being intellectual, rather than intellectual but unintelligent.

To return from filmic discourse to Foucault and the truth conditions for sentences, the logical form attributed to sentences, and the whole machinery of terms, predicates, connectives and quantifiers, none of this is open to direct confrontations with the evidence. [30] That is to say, 'denotation' refers to a class, not its members, except indirectly. This fact about denotations is one reason for my belief that as an actual referent is not logically necessary, there is a continuum, not a disjunction, of statements all of which have what Foucault calls a correlate, and that is all that is necessary - or desirable, if one recalls Swift's philosophers groaning under the weight of their conversation pieces.

The function of the external world, the 'out there' of philosophy, is to underwrite the concepts of numerical vs. qualitative identity, and even here it is threatened, as we shall see, by possible-world semantics. As Wittgenstein would say, there is no room for proof here because there is not the possibility of real doubt.

However there are circumstances, like Peirce's indexical relation or Foucault's discursive formations, in which the referent is internal to the model, but these are logically secondary to or parasitic upon the main structure of discourse in which not the referent but the class to which it belongs constitutes the extension. This reference has a relatively small but essential part in any theory of denotation which, as Mario Bunge points out, is not to be confused with designation, the latter being more like naming.

Therefore, as the burden of this opening chapter has it, we can say things about the actual, but always embedded in some modality and register, which becomes the Real in the best Marxist thought, and a confirmation of the symbolic and imaginary for the conservative. The gist of it is that conservation is based on value which many people hold to be self-evident. Boiled down this means that they have mistaken the cultist for the natural, a classic case of 'naturalization'. A potent searching of the schizophrenia of conservatives which allows them to suppose abominations while personally being quite decent people is found in the George V. Higgins' novel Dreamland, whose title is the only overt expression of the author's attitudes. As always, in Higgins, the novel is mostly dialogue, only this time it is in the first person, so that we share the travails of our 'moderate' hero. It is only at the end that one realizes that the events he has described are empirically not compossible, and that our complex author has resorted quite unthinkingly to a series of lies, including his own, frauds and swindles. So the story (as distinct from the discourse) makes no sense at all - except to our implicit author, who would probably retreat into catachresis if his inconsistencies were put convincingly to him. (He lives in the Real and false, while some of us try to live in the Real and true.) The trouble is, of course, that in reaching it, few conservatives will look beneath the easy, witty surface to the depth structure within, and so nothing changes.

In Foucauldian terms, the novel will make a discursive formation in its final incoherence so that the conservative - or more specifically 'patrist', in G. Rattray Taylor's [31] sense - reader will either not notice it or must violently reject the picture of his/her own discursive formations. In the case of such a rejection, such a reader cannot possibly doubt the self-evident (because always present) 'truths' that govern his/her life without severe ego crisis. This is what Foucault calls the 'positivity of discourse'; it is not so much addition or deletion of facts, which the 'hard-headed' conservative prides him/herself on being so realistic about, but a question of their inevitability for all time.

Such a mind, working within strictly defined parameters efficiently enough, is threatened by counter-instance and tends to deny it as aberration - 'madness' - and, for instance, could not comprehend such a remark as that childhood was a sentimental nineteenth-century invention, or that there was no homosexuality as s/he understands it among the ancient Greeks. Which takes us to the core of Foucault's thesis on discourse, which is that it is highly intensional, the extensional pay-off being radically different accounts of actions which may have been physically identical but intentionally disparate. Thus Julius Caesar was not homosexual even though his troops, Suetonius relishes relating, sang him into battle as 'every woman's man, and every man's woman', and was copiously polymorphous in his sexuality. Of his initiation, Suetonius writes before dissolving into hypothetical dots that it was 'so this descendant of Venus lost his virginity in Bethynia' - to another man, one adds, as the conqueror of Gaul apparently went to extravagant lengths to be femininely attractive, drag and all. The point is that his warriors loved him for this, among other things, and Caesar would lie stretched upon his pallet to receive his men's vocal tribute.

It was a different sexual act from any that might be described by chroniclers today, who cannot escape the force of what Foucault calls the 'positivity' of discourse, since even describing extensionally the physical aspects of any action cannot escape connotation, the way of looking at actions which is embedded in discourse and which one must verbally disclaim before launching into the minutest 'literal' description, since, I say again, extension and intension live by looking in each other's washing. Anyone who still believes that, for example, medical texts are above (or below) such positivity should acquaint themselves with some of R.D. Laing's examples of 'objective' writing among psychiatrists in The Divided Self or if they are fortunate enough to secure one, a text on such pathologies as 'masturbatory insanity' in the early part of this century. One laughs now, but the regimen of cold showers, constant attendance by burly orderlies and even handcuffs were grim enough for the victims of 'objective medicine'. Nor were the psychoanalysts enlightened post-Freud about the horrors of sexual deviance: Wilhelm Stekel, perhaps one of Freud's less gifted acolytes, concludes an hilarious compendium of sexual polymorphism that would send William Burroughs back to the security of his needles, with an emotional 'Retrospect and Prospect' [32] which pleads the cause of these 'unfortunates', who seem in the double-bind context of psychoanalysis to include everyone but the good doctor himself, now revealed as a master of the Catch-22: whatever you're doing it is infantile and to deny that merely compounds the disease with hypocrisy.

But the untestability of psychoanalysis restores it to its true role of poetic metaphor, even if its discursive formation has often resulted in the incarceration of the rare sane by the many beserk: to be 'out of formation' is crazy, or at least one must cover it well, perhaps with a medical prophylactic.

I will not want to quarrel with Foucault's trifurcations of the concept of sign either: sentences produce meanings (when we use them; this is true of all semiotic activity); propositions yield truth or falsity; and discourse, epistemically prior to either, creates discursive formations which already predisposes certain meanings and 'truths'. We are engaged in discourse, through various channels and sign-systems, from the day we are born until the day we die. Like communication, meaning and significance are by-products of this essentially self-locked mode of expression and growth through feedback. It is sad to say that communication is a spin-off, but I believe it to be ineluctable: even supposing telepaths feel the same things at the same time, they still have to employ representational means, which entails (Eco) the ability to lie, or simply to be mistaken, or to use words with an undetectably small amount of intensional similarity, still we are alone.

To advert to Dasheill Hammett a moment, a person is (i) lying, (ii) mistaken, or (iii) lying to him/herself. In The Glass Key, Ned Beaumont flatly tells a woman who relates a dream to him that she is lying. And Sam Spade of The Maltese Falcon (book and film) finally accepts Brigid O'Shaunnessy's latest tale, not because he believes it, but because he is too tired to go on disbelieving her. All of Hammett's heroes achieve a precarious poise because they make their own rules, and even then have trouble wondering whether they are lying to themselves. Celebrated by the French existentialists, Hammett would have understood Foucault very well, while remaining dubious about the truth-value of propositions. Stephen Marcus, in a structuralist introduction to The Continental Op, a collection of Hammett's stories, [33] finds the story about a man who left his job and family only to be traced some years later with a near-identical job and family especially interesting because the man's name is Charles Pierce, and he does not believe that this is a chance event given Hammett's wide-ranging interests and his