Possible Worlds and Premature Sciences

Roger Scruton

  • The Role of the Reader by Umberto Eco
    Indiana, 384 pp, £10.50, September 1980, ISBN 0 253 11139 0
  • The Semiotics of the Built Environment by Donald Preziosi
    Indiana, 192 pp, £9.00, September 1980, ISBN 0 253 17638 7

Semiotics, semiology, hermeneutics, structuralist criticism – so many labels, but how many things? If there are distinctions here, they seem to be largely hereditary. The term ‘semiotic’ comes from C.S. Peirce, ‘semiology’ from Saussure. ‘Structuralism’ has meant one thing in anthropology, another in linguistics; its application to literary theory comes partly through the work of Propp and the Russian formalists. ‘Hermeneutics’ once indicated the nice interpretation of Biblical texts; now it denotes the nice interpretation of everything. In all these things, however, the niceties seem to be the same: technicality at the expense of theory, analysis at the expense of content, intensity at the expense of depth – in short, ‘vain babblings and oppositions of science, falsely so called’ (I Timothy vi, 20). And yet, wherever literature is taught, students have to perceive it through the veil of this new scholasticism, their observations muddled by technicalities borrowed from a thousand premature sciences, distracted by ‘methods’ which regard Mickey Mouse and the Mona Lisa, Superman and King Lear, advertising jingles and the works of Schoenberg, as equally legitimate objects of inquiry. Is this movement a reaction against critical moralism, expressed with a hesitancy so great that only massive recourse to technicality can prevent it from knowledge of its impotence? Or is it the first step towards some new critical method, a method sufficiently general as to assign an interpretation to everything that could be regarded as a ‘sign’?

Those questions are crucial for the future of literary education in our universities. Any answer to them must begin by recognising the uncertainty in a movement which now advances behind the banner of scientific inquiry, now slips sideways into the smoke of literary rhetoric. This uncertainty arises from the attempt to combine three independent enterprises, the first modest, the second speculative, the third rooted in fallacy.

The first is the quest for ‘levels’ of meaning in literary and other works. A poem may allow several readings (literal, allegorical, metaphorical, and so on), which develop with its movement, and take their structure from the narrative which unites them. Dante, in the Convivio, describes four such ‘levels’ of meaning, and the tradition which he exemplifies has continued to the present day. In particular, it can be seen in the French explication de texte, whether dressed in its old, cold, surgical garments, or fitted out in the latest jargon of linguistics. Thé method of much ‘structuralist’ criticism is that of Dante, the method of stepwise analysis, designed to reveal layers of meaning concealed within the enclosure of a ‘text’. The difference lies in a lust on the part of the structuralist for meanings which the author would not have recognised.

Semiotic and structuralist criticism also have their roots in a widespread speculation according to which scientific inquiry does not exhaust the modes of human understanding, being unable to describe the world as we experience it. We stand to this world (the Lebenswelt) in a relation not of observation but of ‘belonging’. From Kantian metaphysics and 19th-century anthropology arose the idea that a peculiar mode of understanding is reserved for this ‘human world’, a mode of understanding (Verstehen) which would show to be fraught with meaning what science displays as ‘neutral’ and ‘meaning-free’. Meaning belongs to human acts and gestures; it also lies dormant in the world. Perhaps, then, there is a general method that will reveal the meanings of things. Such a method will be unconcerned with explanation and prediction, but it will be of universal significance nevertheless, relating us to artifacts, to art, to things and to one another in a way that restores the centrality of human thought and action. It may even re-create that necessary illusion which scientific explanation took away, the illusion of human freedom.

The third intellectual motive behind this kind of criticism contradicts such speculations. If it is combined with them, it is because the seekers after ‘human’ understanding have been reluctant to relinquish, in their pursuit of it, the authority and objectivity of science. Hence there has emerged the desire for a ‘general science of signs’, or semiology. Without the assumption that such a general science is possible, the repeated borrowings from the technicalities of linguistics in order to describe the many things that are fashionably labelled ‘signs’ is nothing but a kind of alchemy, conjuring the illusion of method in the absence of the fact of it.

The idea of a general science of signs is, I think, rooted in fallacy. What makes a science? There is a science of fish because fish are similarly constituted, obey similar laws, have a discoverable essence, over and above the evident facts which lead us so to label them. (Fish constitute a ‘natural kind’.) Buttons, by contrast, have no such essence, and no common identity besides the function which we already know. There can be no general science of the constitution of buttons: if there is a science of buttons, it is a science of their function. Now signs are clearly more like buttons than like fish, and a general science of signs will therefore be a science not of constitution but of function. But what is this function? Semiology refers us to language, road signs, facial expressions, food, clothes, photography, architecture, heraldry, basket-weaving, music. Are all these ‘signs’ in the same sense, or in any sense? The word ‘sign’ means many things, and points to many functions. Do we suppose that a cloud signifies rain in the way that Je m’ennuie signifies that ‘I am bored’? Of course not, since no cloud can have the function of a sentence. From a, scientific point of view, one suspects that there is not one thing here but thousands. What is common is only a small feature of the surface of each, usually as familiar to us as the function of buttons. If there is a common essence of ‘signs’ it is sure to be very shallow; semiology pretends that it is deep.

But there is more to be said. There are sciences whose subject-matter is contested and speculative, based on analogy and hypothesis, rather than on any intuitive knowledge of the existence of a natural or functional kind. Perhaps linguistics itself is such a science: at least, its subject-matter and its methods are still much in dispute. Semiology leans on linguistics, hoping to base itself on an analogy between language and other sorts of ‘sign’. The basis of the analogy is twofold. First, all human behaviour can be seen as expressive. It reveals thoughts, feelings and intentions, not all of which would be spontaneously acknowledged by the agent. Secondly, and more important, the modes of human expression can sometimes be thought to have a certain structure which they share with language. It is this which seems to bring them within the purview of Saussurian linguistics.

According to the Saussurian model, a sentence is a ‘system’ composed of ‘syntagms’. A syntagm can be defined as a set of terms which may replace one another without destroying the system – without rendering the sentence ‘unacceptable’ to speakers of the language. For example, in the sentence ‘John loves Mary,’ ‘loves’ may be replaced by ‘hates’ or ‘eats’, but not by ‘but’, ‘thinks that’ or ‘swims’. Now consider another example, discussed at length by Barthes in his Eléments de Sémiologie, the example of the menu. A man might order the following: oeufs bénédictines, followed by steak and chips, followed by rum baba. That is an ‘acceptable’, system: in our society, the same menu in reverse would not be ‘acceptable’. Furthermore, each dish belongs to a ‘syntagmatic unity’: it can be replaced by some dishes and not others. Steak and chips can be replaced by ham salad but not by a glass of Sauternes – for that would be unacceptable. (One can see at a glance that railway trains and classical architecture also exhibit this kind of ‘structure’.) What follows?

Consider Barthes’s actual interpretation. Steak and chips is supposed to ‘mean’ (according to an essay in Mythologies) ‘Frenchness’. Suppose that the ‘meaning’ of oeufs bénédictines is ‘Catholicism’, and that of rum baba ‘sensuality’: what now is the meaning of the whole system? Does it mean that French Catholicism is compatible with sensuality? Or that being French is more important than being Catholic? Or being sensual a fundamental part of both? There is no telling, since while the system has structure, it has no grammar. Which is to say that there is no way in which the meanings of the parts determine a meaning for the whole. Hence we should not speak of ‘syntax’: the idea of syntax is the idea of a potential meaning, so that syntactical rules must depend upon semantic interpretation. (Frege showed this truth to be fundamental to both logic and language.) It follows that the search for ‘syntagmatic’ and ‘paradigmatic’ structure plays no part in any science of non-linguistic ‘signs’. The two aspects of the analogy with language (expression and structure) cannot be conjoined in the manner that the analogy requires. No amount of technicality can suffice to conjoin them. So it is impossible to assume that the functions of linguistic and non-linguistic ‘signs’ will be similar, and the grounds for supposing that there might be a general science applicable to both of them collapse.

We must therefore abandon the scientific pretensions of semiology. And in fact we find that semiological criticism has relied on linguistics, not for the latter’s scientific claims, but for its technicalities alone. These technicalities nevertheless create an illusion of objectivity. They put the ‘semiological critic’ in a position of peculiar superiority. They give him a justification for avoiding the question of how meaning is objectively determined in language (since their application makes contact with no semantic inquiry); at the same time, semiology lays claim to a special objectivity of its own. This ‘objectivity’ consists in the fact that the semiologist will look for those features of a text that ‘reveal’ things (for example, an attachment to some ‘ideology’), while remaining indifferent to the professed intention of the writer. This is what is meant by the claim that semiology explores ‘levels’ of the text that may not have been apparent to the author. Because Dante was analysing his own poems, he could not raise the question whether the ‘levels’ that he discerned in them belonged to the writer’s intention, the reader’s perception, or both. But the problem for the semiologist is no different from the problem for Dante: what makes an interpretation legitimate? No ‘method’ of ‘decoding’, as the semiologist would express it, will suffice to answer that question. The semiologist commonly sets about the task of ‘demythologising’ the text before him, with the intention of remythologising it in accordance with his own rival ideology. But that can provide no criterion of legitimate interpretation. This can easily be seen by reading Barthes’s analysis of Balzac’s Sarrasine, but I shall take an easier example.

Arithmetic can be derived from logic in various ways. Frege, Russell and Zermelo each attempted such a derivation, and their proofs have a common intellectual structure. But where Frege speaks of ‘concepts’, Russell refers to ‘classes’ and Zermelo to ‘sets’. Applying semiology, we could say that in each case the choice of vocabulary ‘reveals’ (or better ‘codifies’) certain ‘meanings’. For Frege arithmetic is the objectification of the inner life; for Russell it is the expression of class consciousness; for Zermelo it is the description of a collectivity. The bourgeois individualist, the aristocrat and the egalitarian each reveals himself in the language that he chooses. Such an account could be given all the rhetorical and ‘structuralist’ support that Barthes gives to his analysis of Sarrasine. But we know that it is irrelevant to an understanding of the texts. How do we know that?

A question like this bewilders the semiologist, since his technicalities can generate no solution to it. In the last analysis, his answer must be the same as Dante’s: an interpretation is a legitimate ‘reading’ if it develops through the text, acquiring its strength and elaboration from the movement of the literary surface. But perhaps there is more to be said. And it is characteristic of ‘semiotics’ – the new discipline which sometimes forswears, but at other times times endorses, the scientific pretensions of semiology – to attempt to say something more than the obvious in answer to our question. Does semiotics, then, represent some new departure, and does it escape the strictures that we have laid upon the general ‘science’ of signs? The principal exponent of semiotics is Umberto Eco, who holds the only existing chair in this possibly non-existing subject, and whose writings – which take their initial inspiration from the general classification of signs put forward by C.S. Peirce – have proved influential in Italy, France and America. Eco himself has shown a tireless energy in collecting and applying the results of modern semantics, putting forward (in La Struttura Assente) a ‘semiotics’ of architectural forms which has been studied, if not for its content, at least for the enterprise that it represents. He has looked for his theories of the non-linguistic ‘sign’, not only in Peirce and his followers, but also in Chomsky’s transformational grammar, in the theory of speech acts, and now in the ‘possible world’ semantics of the modal logician. And while this may seem to be just another case of Tel Quel dilettantism Eco’s enterprise is distinguished by spasmodic attempts at clarity, and by a relative freedom from that café-table Marxism which recently transferred its favourite point of observation from Les Deux Magots to La Coupole, acquiring in the process a profound respect for the commodities of the consumer culture, and a keen desire to participate in their fetishisation. For such reasons – and because he writes also in English – Eco continues to attract attention, and his awareness that there are important intellectual questions that semiology leaves unanswered has led to a transfer of faith to the newer ‘semiotics’. It seems to express a greater plasticity of outlook, and, being able to regard Julia Kristeva and Saul Kripke as equally relevant to its intellectual enterprise, it draws attention from every corner of the academic world.

Let us, however, ignore the higher reaches of semiotic speculation, and attend for a moment to that mundane application through which it first proved influential: the application to architecture. We find the old semiological fallacies enduring uninterrupted. This can be seen at once from reading Donald Preziosi’s The Semiotics of the Built Environment, which puts forward a theory of the significance of architectural forms. This theory, where intelligible, is palpably erroneous, consisting in a renewed iteration of the fallacy that sequential organisation is a kind of ‘syntax’. Like most practitioners of semiotics (and in architectural theory they are increasingly many), Preziosi begins by assuming what he ought to be proving: namely, that architecture is sufficiently like language for the methods of analysis appropriate to the one to be usefully transferred to the other. And he masks this assumption by using the word ‘code’ to refer to both. By what pre-established harmony of folly has this word ‘code’ come to be brandished in so many places and in the name of so many disciplines, as though it were sufficient to call a thing a ‘code’ to achieve some special claim to be able to decipher it? ‘Like verbal language,’ Preziosi writes, ‘the built environment – what will here be called the architectonic code – is a panhuman phenomenon.’ The italicised words are held by Preziosi to have a technical meaning; they are assembled, with others, in a glossary, which explains the crucial term ‘code’ thus: ‘the system (qv) of ordered relationships among significant formations’. For further elucidation one turns to the entry for system, which says ‘system: see code.’ From which one can deduce what significance these terms have for the author. As one can see already, ‘code’ is going to be used whenever there is sequential pattern: the idea of semantic interpretation has been brought in by a sleight of hand.

Pressing on, despite this unpromising start, one learns that ‘a built environment is an ongoing, dynamically unfolding array of signs, existing’ (one is happy to note) ‘spatially and temporally.’ In other words, the architectonic code consists of signs. What is a sign? The answer is that ‘an architectonic sign is a combination of a formation with a meaning.’ So Preziosi is committed to semantics after all. And perhaps that is no bad thing: for do we not speak of the meaning of architectural forms? Do we not regard buildings as in some sense speaking to us, even, as it were, observing us, with something on the tip of their (our) tongue? Perhaps a ‘semiotics of the built environment’ will solve the mystery which such speculations generate. But how should it proceed? Preziosi begins by isolating the architectural vocabulary, which he identifies with the ‘geometrical properties’ of formations. This suggestion is already highly implausible. Imagine two buildings with identical geometrical properties, in an identical arrangement (or ‘syntax’), the one built in limestone, the other cast in steel. Is it not absurd to say that they must have the same ‘meaning’ for us?

Let us suppose, however, that we have isolated the architect’s ‘vocabulary’. No progress can now be made without the assumption that meaning is determined by convention. If it is not, then the reference to vocabulary, and later to syntax, is spurious. Now it is fairly clear to anyone who has thought about these matters that, while what is sometimes called meaning in architecture (but which might equally be called aesthetic character) is influenced by convention, it is not merely the result of convention. An architect cannot guarantee the ‘meaning’ of his work by obedience to rule. It is hardly surprising, then, if Preziosi’s attempts to interpret his architectural ‘syntax’ are hesitant and vague. After extracting the ‘syntactical’ rules which generate the forms of ancient Minoan houses, he writes as follows: ‘If we take an “architectonic code” as meaning, among other things, sets of elemental units, relationships among those units, and rules governing relationships …’ What he is saying, clearly, is that architecture ‘means’ its own ‘syntax’. In other words, we have failed to advance from syntax to semantics. We have syntax without semantics, and therefore neither. None of Preziosi’s discussion advances beyond this point, which is hardly surprising, since there is no advance to be made, and however much he may seek to draw our attention to the ‘ongoing semiotic bricolage of daily life’, his observations concerning its principal manifestation remain at a level of vagueness that no linguistic alchemy can conceal.

Umberto Eco’s book of essays promises something more substantial. The collection reports and explores specimens of ‘unlimited semiosis’ (the creeping tendency to accumulate meaning). These specimens are the ‘texts’ of the semiotician, and the word ‘text’ applies to everything that might be read as a sign, and so be infected by this disease. Imitating the Barthes of Mythologies (but with rather less wit), Eco explores the narrative devices of Ian Fleming and the emotional significance of the Superman comics. He also pays attention to modern music, to the once popular melodramas of Eugène Sue, and in the piece which enunciates his latest theoretical advances, to a literary joke by Allais, culled from Breton’s Anthologie de l’Humour Noir.

Like Preziosi, Eco introduces, and relies upon, the unexplained notion of ‘code’; and since he recognises that a literary work has a writer and a reader, he regards textual codes as the property of both. The title of his collection derives from his supposition that the reader may be in part the creator of what he reads, according to the extent to which the text is, as he puts it, ‘open’ or ‘closed’. It is rather difficult to understand the distinction between the open and the closed text. It is clearly meant to provide some substitute for the (ideologically unacceptable?) distinction between high and popular culture. But Eco’s account is odd and sometimes contradictory. The closed text (such as the Superman comic-strip) aims at what a more old-fashioned critic might have called ‘stock responses’. Paradoxically, this also makes it ‘open’ – open, that is, to any possible ‘aberrant de-coding’. Such texts purportedly generate no standard of what is ‘right’ or ‘wrong’ in reading them, since they elicit no creative participation from the reader. The truly open text, by contrast, because it makes demands, aims also to discipline its reader into the adoption of those ‘codes’ (or habits of interpretation) which would be appropriate to it. The open text is addressed to an ideal reader (whether or not suffering from an ideal insomnia), and is therefore ‘closed’ to aberrant interpretations. Ian Fleming is a writer of closed texts, James Joyce a writer of open texts. We have here an important critical distinction, corresponding to the more familiar distinction between mechanical contrivance and creative art. Unfortunately, semiosis soon sets in, and the distinction rots away. The text becomes ‘blown up’, and then ‘narcotised’, a purveyor of ‘topics’, ‘isotopics’, ‘semantic markers’. Words like ‘extensional’, ‘intensional’ and ‘macroproposition’ enter without explanation. As well as phonemes and sememes, we eventually acquire the inscrutable ‘styleme’. None of these technicalities is either explained or seriously exploited. When it is said, in the conclusion to the introduction, that ‘ruled by a constitutive mechanism of unlimited semiosis, the semantic space can be reduced only through the co-operative activity performed by the reader in actualising a given text,’ all that is meant is what we already knew: that since a work of art bears many meanings, the reader must choose between them.

But the difficulties for the ideal reader of Eco are compounded by a far from ideal author, who goes on to confuse ‘openness’ with indeterminacy, as exemplified by improvisatory music, and then with ‘suggestiveness’, so that the paradigm switches first to the auditory confusion of Stockhausen and then to the fastidious impressionism of Mallarmé and Debussy. What had begun by seeming to be a genuine theory ends as a concatenation of independent distinctions, none of them adequately characterised.

In his final essay, Eco brings forward a truly formidable body of technicalities, in the attempt to arrive at a satisfactory ‘reading’ of a short story by Allais. This story is a joke, directed against the reader, who is encouraged to fill in its lacunae. In the course of doing so, he involves himself (the implication is, through no fault of the author’s) in a startling self-contradiction. A story like this is, of course, a gift to the semiotician. Professor Eco uses it to introduce the latest (and most fashionable) developments in modal logic – in particular, the theory of possible worlds. The result scarcely makes sense, and it appears to me that the author may well be aware of this.

The semantics of modal logic arose in the following way. Certain words, such as ‘possibly’ and ‘necessarily’, which seem to be indispensable to ordinary as well as to scientific discourse, present problems in the theory of meaning. In particular, they generate ‘intensional’ contexts, contexts like ‘It is possible that …’ which, when completed by some sentence ‘p’, yield sentences which may be true or false independently of the truth or falsehood of p. Ordinary logic is ‘extensional’: the truth or falsehood of any complex sentence depends on the truth or falsehood of its parts. And modern logic has developed almost entirely on the assumption of extensionality: without that assumption logic seems hardly to proceed. How then can ordinary logic either represent or elucidate sentences about the necessary and the possible? It is clear that these sentences have a logic. Following a suggestion of Leibniz, the modern logician translates ‘It is necessarily p’ as ‘p is true in all possible worlds.’ ‘Necessarily’ then becomes a ‘quantifier’, and the logical properties of quantifiers can be described in extensional terms. This theory is capable of systematic development.

It might seem natural to suggest that a work of fiction is nothing but the (partial) description of a possible world. Perhaps, then, the theory of possible worlds can be used to elucidate our understanding of fiction, and thereby show us how jokers like Allais can tie us in impossible knots. This, I suppose, is the intuition that underlies Eco’s attempt to introduce the technicalities of this branch of ‘model theory’ (as it is called) into his discussion of Allais’s story. Unfortunately, there are two serious drawbacks. The first is that Eco does not seem to understand the theory from which he persistently borrows. Even his grasp of extensional logic is tenuous, as is shown by the fact that when he chooses to express himself in logical notation rather than in English, the result is usually ill-formed (examples on pages 213 and 236), When he comes to write of model theory itself, the result is strictly unintelligible.

Secondly, it is doubtful whether model theory can do anything to elucidate our understanding of fiction. The difficulty here was pointed out over two millennia ago. In the Poetics Aristotle drew attention to the fact that impossibilities are frequent in fiction. But, he argued, there is no reason to complain about a ‘probable impossibility’, which is always preferable to an ‘improbable possibility’. (Consider the many difficulties that surround the idea of incarnation, so brilliantly dramatised by Wagner in the Ring. It is impossible that Wotan should be identical with Walse, but, were he to be so identical, it is probable that he would have fathered Siegmund and Sieglinde.) In other words, a fictional world may not be a possible world. What then happens to the semantics of fiction? A determined model theorist could think of ways around the problem. But neither problem nor solution is hinted at by Eco. Moreover, it is clear that he would not wish to get entangled in such inquiries. All he seeks in the abstruse regions of model theory is the rhetoric of technicality, the means of generating so much smoke for so long that the reader will begin to blame his own lack of perception, rather than the author’s lack of illumination, for the fact that he has ceased to see. Therein lies the disease of semiosis. What is true of Eco is, so far as I can see, true of all the practitioners of his art. Perhaps it needs another Ben Jonson to reveal the complex motives behind all this. But we can be sure that, while there was a humane beginning to this madness, its ‘method’ leaves humanity behind.