The classification of modes of production outlined in Table 39 takes into account four parameters:
(i) the physical labor needed to produce expressions (ranging from the simple recognition of a pre-existent object or event as a sign to the invention of previously non-existent and un-coded expressions);
(ii) the type/token-ratio, whether facilis or difficilis (see 3.4.9.);
(iii) the type of continuum to be shaped; this continuum can be either homomaterial or heteromaterial; a continuum being homomaterial when the expression is shaped within and by the same material stuff as that with which the possible referent of the sign-function could be made (in cases where signs are used to mention things). All other cases imply a heteromaterial continuum which is arbitrarily selected (except in a few cases in which the matter of the expression is imposed by the direct action of the referent; for instance imprints are impressed upon a given material by the imprinter);
(iv) the mode and complexity of articulation, ranging from systems in which there are precise combinational units that are duly coded or overcoded to those in which there are texts whose possible compositional units have not yet been further analyzed.
A Typology 01 Modesof Sign Production
The table records the way in which expressions are physically produced and not the way in which they are semiotically correlated to their content; the latter is implied by two decisions that must be made either before or after the production of the expression.
For instance, in the case of recognition of symptoms, there is undoubtedly a pre-established motivation due to a preceding experience which has demonstrated that there is a constant physical relationship between a given agent and a given result; it has therefore been decided, by convention, that these resultant objects must be correlated with the notion of that agent under any circumstances, even when one cannot be sure that an existing agent has really produced the result. In the case of words (which may be classed among ‘systematically combinable units’) the correlation is posited after the production of the physical unit and is in any case independent of its form (this assumption being valid even if by unverifiable historical chance the origin of words had some sort of imitative motivation).
For this reason such non-homogeneous objects as a symptom and a word are posited in the same row; every object listed there can be produced according to its pre-existing expression-type (ratio facilis) and this happens irrespective of the reasons for which these objects were selected as the expression of a given content. All of them could be produced by a suitably instructed machine which only ‘knows’ expressions, while another machine could assign to each expression a given content, provided it was instructed to correlate functives (in other words, two expressions can be differently motivated but can function in equally conventional fashion).
On the other hand, all objects ruled by a ratio difficilis are so motivated by the semantic format of their content (see 3.4.9.) that it is irrelevant whether they have been correlated with it on the basis of previous experience (as in the case of footprints, where the semantic analysis of the content has already been performed) or whether the content is the result of the experience of ‘inventing’ the expression (as in the case of paintings). Therefore the motivated way in which they have been chosen (see the further analysis of imprints and projections below) does not affect their mode of production according to a ratio difficilis; they are correlated to certain aspects of their sememes – thereby becoming expressions whose features are also content-features, and thus projected semantic markers (20).
In this sense a machine instructed to produce these objects should be considered to have also received semantic instructions. One might say that since it is instructed to produce expressions, it is being fed with schematic semantic representations (21).
The items recorded in the row corresponding to the parameter ‘type/ token-ratio’ may look like ‘signs’, since to some degree they recall preexisting sign typologies. But they are not; they are short-hand formulas that should be re-translated so as ‘to produce imprints’, ‘to impose a vectorial movement’ or ‘to replicate combinable units’ and so on.
‘imprints’ or ‘examples’ must, at most, be understood as physical objects which, because of certain of their characteristics (not only the way in which they are made, but also the way in which they are singled out) become open to a significant correlation, i.e. ready to be invested with dignity of functive. In other words they are potential expression features or bundles of features. According to the system into which they are inserted, they may or may not be able to convey by themselves a portion of content. So that although they can also act as signs, they will not necessarily do so. It must be clear that the whole of Table 39 speaks of physical procedures and entities that are ordered to the sign-function but that could subsist even if there were no code to correlate them to a content. On the other hand, they are produced in order to signify and the way in which they are produced renders them able to signify in a given way.
A ready-made expression like /cherry brandy/ is the result of two procedures depending on a double type/token-ratio; it is constructed from two combinational units ordered by a vectorial succession; likewise a pointing finger is both a vector and a combinational unit, while a road arrow is both a stylization and a vector. Therefore items like ‘vectors’ or ‘projections’ are not types of signs and cannot be equated with typological categories such as ‘indices’ or ‘icons’. For instance both ‘projections’ and ‘imprints’ could appear to be icons but the former would imply an arbitrarily selected expression-continuum and the latter a motivatedly established one, while both of them (equally governed by a ratio difficilis) would be motivated by a content-type (though imprints are ‘recognized’, while projections are ‘invented’).
Imprints and vectors look like indices, but are in fact dependent on two different type/token-ratios. Moreover, certain categories (e.g. ‘Active samples’) come under two headings: they are the result of a double labor, since something must be replicated in order to be shown (ostension).
All these problems will be dealt with further in the following paragraphs. I have only anticipated some examples in order to stress the fact that one must not look at Table 39 in order to find types of signs. This table only lists types of productive activity that can give rise, by reciprocal and complex interrelations, to different sign-functions, whether they are coded units or coding texts.
Recognition occurs when a given object or event, produced by nature or human action (intentionally or unintentionally), and existing in a world of facts as a fact among facts, comes to be viewed by an addressee as the expression of a given content, either through a pre-existing and coded correlation or through the positing of a possible correlation by its addressee.
In order to be considered as the functive of a sign-function the object or event must be considered as if it had been produced by ostension, replica or invention and correlated by a given kind of type/token-ratio. Thus the act of recognition may re-constitute the object or event as an imprint, a symptom or a clue. To interpret these objects or events means to correlate them to a possible physical causality functioning as their content, it having being conventionally established that the physical cause acts as an unconscious producer of signs. As we will see, the inferred cause, proposed by means of abduction, is pure content. The object can be a fake or can be erroneously interpreted as an imprint, a symptom or a clue, when in fact it is the chance product of other physical agents: in such a case the ‘recognized’ object expresses a content although the referent does not exist.
In the recognition of imprints, the expression is ready-made. The content is the class of all possible imprinters. The type/token-ratio is difficilis. The form of the expression is motivated by the form of the supposed content and has the same visual and tactile markers as the corresponsing sememe, even though the marks of the sememe can be ‘represented’ by the imprint in various ways. For example the size of the imprinter determines (or motivates) the size of the imprint, but there is a similitude rule establishing that the size of the latter is always larger than the size of the former (even if infinitesimally). The weight of the imprinter determines the depth of the imprint, but this process is governed by a proportional rule (that is, an analogy in the strict sense outlined in 3.5.4.). With fingerprints, size is not a pertinent parameter since they can be correlated to their content even if enormously magnified.
These observations may help to clarify in what sense one could say that an imprint represents both a metaphorical and a metonymical operation. In fact imprints appear to be ‘similar’ to the imprinting agent and substitute for or represent it; and they can be taken as a proof of past ‘contiguity’ with the agent.
This explanation may work (and may indeed be used to distinguish imprints from clues and symptoms, see below) provided one accepts that the similarity of imprints to their possible cause is not immediately detectable, since certain transformational operations must be understood to have taken place, and that past contiguity with the referent is the result of a labor of presupposition performed when the sign is viewed as focusing on the world and interpreted as a mention (see Table 31).
All this means that, first of all, one must learn to recognize imprints (or to fake them). Imprints are usually coded; a hunter learns how to recognize the imprint of a hare without mistaking it for a rabbit’s. Insofar as they are coded, they rely on oppositional expressive systems; roughly speaking, one is dealing with oppositions such as ‘hare vs. rabbit’, though in fact these oppositions should be the product of a more finely analyzable system of pertinent spatial features. Semiotics has not yet done sufficient work on such expressive systems, but one of its provisional boundaries would be that of ‘imprints’ being not signs but rather objects to be inserted into a sign-function; in fact the trace of an animal, viewed as a sign-function, does not only imply spatial or tactile parameters (size or weight) but also vectorial cues (see 3.6.5.).
A trace is also interpreted in terms of its direction. This direction is another productive cue that can be falsified; one can shoe a horse backward so as to give the impression that the horse was going in the opposite direction. When interpreted as an imprint and as a vector, a given trace is correlated not so much with a coded unit (cat, horse, hare, SS soldier and so on) as with a discourse (a horse passed by three days ago going in that direction). Therefore the expression is no longer a sign but rather a text (22).
The correlational dynamics of imprints could be better explained when speaking of projections (3.6.9.) and in fact they are recognized as if they were projected on purpose. As projections, imprints can also be complex texts, in the sense that they can be imprints of very complex events; and in this way they cannot longer be considered as coded units.
But in the present section I am considering coded imprints, corresponding to a coded content; in this sense they are pertinent macro-units in some way analyzable into pertinent elementary features.
Anyway imprints are doubly motivated: once by the form of their content, and once by the presupposed relationship to their cause; therefore an imprint is a heteromaterial object (a cat’s paw mark in the mud is not materially the same as its possible cause) but its matter is strictly motivated by its cause.
Imprints (like any other recognition procedure) are conventionally coded, but the code is not established by an arbitrary social decision but is instead motivated by previous experiences; the correlation between a given form and a given content has been mediated by a series of mentions, inferences based upon uncoded circumstances, meta-semiotic statements (23). Since the experience of an event was constantly associated with a given imprinted form, the correlation, first proposed as the result of an inference, was then posited.
In the recognition of symptoms, the expression is ready-made. The content is the class of all possible causes (organic alterations). The type/token-ratio is facilis, for red spots do not have the same semantic markers as measles, nor does smoke have the same as fire. Nevertheless within the sememic representation of their content there is, among the markers, both the description and the representation of the symptoms.
This explains the way in which a symptom is correlated to the notion of its cause; the notion of the symptom constitutes part of the sememe of the cause and it is thus possible to establish a metonymical correlation between the functives (by a pars toto procedure). The process is notion-to-notion (or unit-to-unit) and the effective presence of the referent is not required. There can be smoke even if there is no fire at all, which means that symptoms can be falsified without losing their significant power. The ratio being facilis, it would be incorrect to speak of a certain ‘iconicity’ of symptoms; they have nothing to do with their content (or referents) in terms of similarity. When symptoms are not previously coded, their interpretation is a matter of complex inference and leads to the possibility of code-making.
Symptoms can be used for mentioning (smoke means «there is fire», red spots on the face mean «this child has measles»). In this case the mentioning procedure works as follows: by a coded and proved causality (contiguity) of the type ‘effect to cause’, an effective presence of the causing whole is deduced.
In the recognition of clues, one isolates certain objects (or any other kinds of trace which are not imprints) left by someone on the spot where he did something, so that by their actual presence the past presence of the agent can be inferred. It is evident that, when used for mentioning, clues work in exactly the opposite way from symptoms; by a coded and proved contiguity (of the type ‘owned to owner’) a possible presence of the causing agent is abduced. In order that the abduction be performed, the object must be conventionally recognized as belonging to (or being owned by) a precise class of agents. Thus if at the scene of a murder I find a dental plate I may presume that, if not the murderer, at any rate someone who has no more natural teeth has been there. If on the floor of a political party’s office, recently broken into, I find the badge of the rival organization, I may presume that the burglars were the ‘bad guys’ (obviously clues can also be falsified, and in cases like this they usually are).
As a matter of fact clues are seldom coded, and their interpretation is frequently a matter of complex inference rather than of sign-function recognition, which makes criminal novels more interesting than the detection of pneumonia.
One could say that imprints and clues, even though coded, are ‘proper names’, for they refer back to a given agent. The objection does not affect the fact that they refer, in any case, to a content, for there is nothing to stop the class to which the expression refers from being a one-member class (see 2.9.2.).
But in fact very seldom can imprints and clues be interpreted as the traces of an individual agent (indeed maybe never). When looking at the footprint on the island, Robinson Crusoe was not able to think about an individual. He detected «human being». When discovering Friday he was undoubtedly able to express the index-sensitive proposition «this is the man who probably left the footprint». But even if he had previously known that there was one and only one man on the island he would not, when looking at the footprint, have been able to refer it to a precise individual; the primary denotation of the expression would have been «human being» and the rest would have had to be a matter of inference. It is very difficult to imagine an imprint that mentions a referent without the mediation of a content (24). The only case would be that in which one sees a given individual in the act of producing a footprint; but in this case the footprint would not be ‘recognized’ as a sign, for it would not be ‘instead of’ something else, but ‘along with’ it (see the case of mirrors in 3.5.5.) (25). The same happens with clues. Even if I know that only one particular man, among the murdered person’s circle of friends, has a dental plate, I cannot regard the object left at the scene of the crime as a sign referring back to a «person x». The object simply means «person without teeth», and the rest is once again a matter of inference.
On the contrary many clues are overcoded objects. Suppose that I find a pipe in the same place. What makes me sure that a man was there? A social rule establishing that gentlemen smoke pipes and ladies don’t (the opposite would happen if I found a bottle of Chanel No. 5).
Ostension occurs when a given object or event produced by nature or human action (intentionally or unintentionally and existing in a world of facts as a fact among facts) is ‘picked up’ by someone and shown as the expression of the class of which it is a member.
Ostension represents the most elementary act of active signification and it is the one used in the first instance by two people who do not share the same language; sometimes the object is connected to a pointer, at others it is regularly picked up and shown; in both cases the object is disregarded as a token and becomes, instead of the immediate possible referent of a mention, the expression of a more general content.
Many things have been said about signification by ostension (see for instance Wittgenstein, 1945:29-30) and a purely ostensive language has been invented by Swift. Let me only remark here that in ostension there is always an implicit or explicit stipulation of pertinence. For example, if I show a packet of brand X cigarettes to a friend who is going shopping, this ostension can mean two different things: either «please buy some cigarettes» or «please buy this brand of cigarettes». Maybe in this latter case I would have to add certain indexical devices, such as tapping with the finger on the part of the packet which bears the name of the brand, and so on. Likewise, in other circumstances only a previous stipulation of pertinence makes clear whether, when showing a packet of cigarettes, I mean «packet of cigarettes» or simply «cigarettes».
At other times ostension may suggest an entire discourse, as when I show my shoes to someone not in order to say «shoes», but rather «my shoes are dirty» or «please shine my shoes». In these latter cases the object is not only taken as a sign but also as a referent and the indication constitutes an act of mentioning. As a matter of fact it is as if I were saying «shoes (ostension) + these (mention) + shoes (referent)».
This theory solves the problem of ‘intrinsically coded acts or object’ (see 3.5.8.) without implying that a part or all of the referent will constitute a part of the definition of the sign-function; the shoes are first of all viewed as an expression which is made with same stuff as its possible referent. Therefore ostensive signs (depending on choice) are homomaterial.
In principle ostensive production should be considered as governed by a ratio difficilis, for the shape of the expression is determined by the shape prescribed by the sememic composition of the content; yet in fact they constitute expressions whose form is already established by a sort of repertoire, and therefore they should be considered as governed by ratio facilis. For this reason I have classified them half way between both ratios. In practice they are already produced (as functional objects) and the problem of their type/token-ratio vanishes; but theoretically speaking (and considering them as if they had to be produced) they constitute a particular category of sign-functions in which both the ratios coincide.
Another characteristic of expressions produced by ostension is that they can be taken in two ways: as the conventional expression of a cultural unit (a cigarette means «cigarettes») or as the intensional description of the properties recorded by the corresponding sememe. So I can show a cigarette in order to describe the properties of a cigarette (it is a cylindrical body, several inches long, white etc.). This is the only case in which doubles can be used as signs.
So, as with doubles, in ostension the type/token-ratio becomes a token/token-ratio. Theoretically speaking, type and token coincide, which explains why ratio facilis and ratio difficilis also coincide.
All these observations might well lead to the conclusion that – in expressions produced by ostension – to distinguish expression from referent is a rather Byzantine exercise, which is by no means the case. Suppose that a crowd of men, each of whom has received a piece of bread, hold up these pieces (different in shape and size) shouting /more!/. The differences between referents disappear and only major pertinent features are virtually retained; the crowd is saying «we want more bread» (irrespective of its shape and maybe of its exact quantity). Inasmuch as it is shown, the bread works as a sign, and it is ‘made’ more elementary than it really is, becoming conventionally and virtually deprived of many of its physically detectable properties.
When an object is selected as a whole to express its class, this constitutes a choice of example. The mechanism governing the choice and the signifying correlation are based on a synecdoche of the kind ‘member for its class’. When only part of an object is selected to express the entire object (and thereby its class) this constitutes a choice of sample. The mechanism governing the choice and the signifying correlation is based on a synecdoche of the kind ‘part for the whole (of a member of a class)’. Instances of this case are those in which a tailor shows a small portion of a fabric in order to refer to the entire cut or indeed directly to the jacket (or shirt) made with this fabric, or a musical quotation referring to a whole work (/play me ‘ta-ta-ta-taaa’/ may mean «play me Beethoven’s Fifth»). On the other hand, an instance of ‘metonymical’ sample may be given by a lancet’s meaning «surgeon».
As Goodman (1968) remarks in an interesting discussion on samples, a sample can be taken as the sample of “samples”. Goodman also remarks that a polysyllabic word can be taken as the example of the general class of polysyllabic words. Since it is the case of a double, chosen or produced in order to exemplify not the physical properties of that token but the semantic properties of a metalinguistic sememe (see note 25), a preceding or presupposed discourse is always needed in order to stipulate the pertinence level. Without this previous convention, the ostension of the word /polysyllabic/ would be taken as the description of the properties of the expression /polysyllabic/ and not of every polysyllabic word, such as for instance /monosyllabic/.
Looking at Table 39, one may note that there is one sort of sample that is listed both under the heading of ostension and under that of replica. These are fictive samples, i.e., the sign-functions that Ekman and Friesen (1969) have called ‘intrinsically coded acts’ (see 3.5.8.).
If I pretend to hit someone with a fist, the meaning of the whole act is «I punch you». One could say that this was a regular ostension for I have chosen a token gesture in order to represent its class. As a matter of fact I have not so much ‘picked up’ an existing gesture as ‘re-made’ it, and in remaking it I have disregarded certain properties of the gesture (for instance, I do not really punch my interlocutor, and I therefore stop the trajectory of the gesture a little before its fullfilment). Thus I have replicated part of a gesture as a sample of the entire gesture. Thus so-called intrinsically coded acts are at once both ostension and replicas. Mimicry belongs to this category, as do ‘full’ onomatopoeias (that is, ‘realistic’ reproductions of a given sound by a human voice or other instrument, as opposed to onomatopoeic stylizations, such as /thunder/) (26).
Fictive samples are also homomaterial, because the replica is performed using the same stuff as that of the partially reproduced model. Therefore to call these full onomatopoeias ‘iconic’, in the same way that one calls the image of an object iconic, is to categorize them imprecisely, since images must be classified among projections (see 3.6.7.) where the expression-continuum is different from the stuff of the possible referent and the correspondence is fixed by transformational rules. A fictive sample does not need transformational rules since it is a homomaterial replica (a partial double) and as such has the advantage of being governed by a ratio facilis, while images are governed by a ratio difficilis. That a so-called intrinsically coded act is a matter of convention can be demonstrated by the fact that, in order to work as a sign-function, it required a previous stipulation (27).
This mode of production governs the most usual elements of expression, so that, when defining the notion of sign, one takes into account only replicable objects intentionally produced in order to signify. Thus the best known kinds of replicas are phonemes and morphemes; expression-units constructed according to a ratio facilis, using a continuum completely alien to their possible referents, and arbitrarily correlated to one or more content-units.
But this unit-to-unit correlation is not typical of replicas alone. Recognition and ostension likewise isolate units and symptoms, imprints, clues, examples, samples and fictive samples are coded by a unit-to-unit correlation. There are conventions by which a certain trace means «hare», a certain medical symptom means a given sickness, and an object taken as an example means a precise category. It is true that a footprint can ‘say’ more than «man», as we have seen, and that a packet of cigarettes may also mean «buy me some cigarettes», but then it is equally true that the word /cigarette/ may, under certain circumstances, stand for an entire discourse. This means that all the sign-functions depending on replica, ostension and recognition articulate given units in order to produce more complex texts.
Granted that this is so, one may go on to list under the heading of replicas not only verbal devices, but also ideograms, emblems (like flags), alphabetic letters, various coded kinesic features (for instance gestures meaning «come here», «yes», «no» and so on), musical notes, various traffic signals («stop», «walk», «no turn»), elementary graphic features, symbols in formal logic and mathematics, proxemic features and so on.
It is true that a word can be analyzed into more elementary, non-significant units (phonemes) and phonemes into more elementary, non-articulatory features, while an ideogram or an emblem must be taken as an unanalyzable unit. But this only means that replicable expressions work on different pertinence levels and may be subject to two, one, or no articulation.
During the sixties, semiotics was dominated by a dangerous verbocentric dogmatism whereby the dignity of ‘language’ was only conferred on systems ruled by a double articulation. A typical example of this fallacy is Lévi-Strauss’s discussion on the ‘linguistic’ properties of paintings, tonal music and post-Webernian music.
In verbal language there exist elements of first articulation, endowed with meaning (morphemes), which combine to form broader syntagmatic strings; these elements can subsequently be analyzed into elements of second articulation (phonemes). There is no doubt that meaning in language arises through the interplay of these two types of elements; but this does not mean that every semiotic process must come about in the same way. Instead Lévi-Strauss maintains that language cannot exist unless these conditions are fulfilled.
In his Entretiens (1961) with a radio interviewer, he had already developed a theory of visual works of art which outlined this viewpoint and he developed it more fully in the “Ouverture” to The Raw and the Cooked. In the former case he referred to a theory of art as iconic sign which he had elaborated in The Savage Mind, where he spoke of art as “a reduced model” of reality. Art is considered as the capture of nature by culture; it raises the brute object to the level of sign, and reveals in it a previously latent structure. But art signifies by means of a certain relationship between its sign and the object which inspired it; thus if it were not an arbitrary and a conventional phenomenon of a linguistic order it would no longer have the character of sign. If in art an appreciable relationship between signs and objects subsists, this is certainly due to the fact that, in one way or another, it presents the same types of articulation as verbal language.
Like verbal language, painting is supposed to articulate units which are endowed with meaning and which can be considered equivalent to morphemes (and here Lévi-Strauss clearly refers to identifiable images, and therefore to iconic signs); these units can be analyzed into minor articulatory elements (forms and colors) which only have oppositional value and are devoid of any autonomous meaning. According to Lévi-Strauss the ‘non-figurative’ schools forgo the primary level “and claim that the secondary level is sufficient”. They fall into the same trap as atonal music, they lose all ability to communicate and slip into “the heresy of the century”, the claim of “wanting to build a sign system on a single level of articulation”.
Lévi-Strauss’s text, which elaborates perceptive observations on the problems of tonal music (in which he recognizes, for example, elements of second articulation) is in point of fact based on a series of unfortunately dogmatic assumptions, namely that: 1) there is no language without double articulation; 2) double articulation is not mobile, the levels cannot be substituted or interchanged, their structure is based on deep natural structures of the human mind. But if one instead examines the functioning of various sign systems, one realizes that: (a) there are systems with various types of articulation or none at all; (b) there are systems whose level of articulation is changeable.
Obviously one may suppose that there probably does exist a profound articulatory matrix which governs every sign-system and all its possible articulatory transformations, but this matrix must not be identified with one of its surface manifestations. This is precisely what Lévi-Strauss does when, for instance, he attributes a privileged status to the tonal system in music, forgetting that tonal system was born at a given historical moment and that the Western ear has grown accustomed to it. Lévi-Strauss rejects the atonal system (as well as the whole of non-figurative painting) for not being governed by a detectable double articulation, thus proposing the tonal system in music and figurative procedures in painting as basic and natural metalanguages, exclusively entitled to define (or reject) every other musical or visual ‘language’.
To confuse the laws of tonal music with the laws of music tout court is rather like believing that if one has a pack of French playing cards (52 plus one or two jokers), the only possible combinations among them are those established by bridge. Whereas on the contrary, bridge is a sub-system which makes possible an endless number of different games, but which could be replaced, still using the same cards, by poker, another sub-system which restructures the articulatory elements constituted by individual cards, enabling them to assume different combinational values and to form other significant arrangements (pair, three of a kind, flush, etc.). Clearly a given game (be it poker, rummy or bridge) isolates only some possible combinations among those permitted by the cards, but it would be a mistake to believe that any one of these combinations is the basic one.
It is true that the 52 (or 54) cards provide a choice which operates within the continuum of possible positional values – as do the notes of the tempered scale – but clearly various sub-systems can be constructed within this system; equally, there are card games which choose different numbers of cards – the 40 cards of the Neapolitan pack, the 32 cards of German skat. The real system which presides over card games is a combinational matrix which can be studied by games theory; and it would be useful if musical science were to study the combinational matrices which permit the existence of diverse systems of attraction; but Lévi-Strauss identifies cards with bridge, confuses an event with the structure which makes multiple events possible.
Playing cards bring us face to face with a problem which is very important for our investigation. Does the system of playing cards have two articulations? If poker vocabulary is made possible by the attribution of meanings to a particular articulation of several cards (three aces of different suits, equal to «three of a kind»; four aces equal to «four of a kind») we should consider the combinations of cards as significant strings of first articulation while the cards which form the combinations are elements of second articulation.
Nevertheless the cards are not distinguished merely by the position they assume in the system, but by a twofold position. They are opposed as different values within a hierarchic sequence of the same suit (ace, two, three . . . ten, jack, queen, king) and they are opposed as hierarchic values belonging to four sequences of different suits. Therefore two tens combine to form a «pair»; a ten, a jack, a queen, a king and an ace combine to form a «sequence»; but only the cards of the same suit can combine to form a «suit» or a «royal flush».
Therefore some values are pertinent features as far as certain significant combinations are concerned, and others are so as far as certain others are concerned. But is the single card the ultimate term of any possible state of articulation, thus resisting further analysis? If the seven of hearts constitutes a positional value in respect to the six (of any suit) and in respect to the seven of clubs, what is the single heart if not the element of an ulterior and more analytic articulation?
The first possible answer is that the player (who ‘speaks’ the language of the cards) is not in fact called upon to articulate the unit of suit, because he finds it already articulated in values (ace, two . . . nine, ten); but this point of view, though it may appear logical to the poker player, is already questionable to a player of other games (like the Italian ‘scopa’) in which the points (the units) are added up, and in which therefore the pertinent unit is that of suit (even if the additions have preformed addendae).
All these considerations force one to recognize that it is wrong to believe: 1) that every sign system act is based on a ‘language’ similar to the verbal one; 2) that every ‘language’ should have two fixed articulations. One should on the contrary assume that: (i) semiotic systems do not necessarily have two articulations; (ii) the articulations are not necessarily fixed.
Let us here list a series of different articulatory possibilities, following the proposal set out by Prieto (1966). It will be seen that there exist systems with two articulations, systems with only the first articulation, systems with only the second articulation and systems without articulation. Let us recall that (i) the elements of second articulation (called figurae by Hjelmslev) are purely differential units which do not represent a portion of the meaning conveyed by the elements of first articulation; (ii) the elements of first articulation, commonly called ‘signs’, are strings composed by elements of second articulation and convey a meaning of which the elements of second articulation are not a portion; (iii) there are signs whose content is not a content-unit but an entire proposition; this phenomenon does not occur in verbal language but it does occur in many other semiotic systems; granted that they have the same function as verbal sentences, we shall call these non-verbal sentences ‘super-signs’. In many semiotic systems these super-signs must be considered as strictly coded expression-units susceptible of further combination in order to produce more complex texts (Prieto, following Buyssens, calls these super-signs /sèmes/, but I prefer to avoid such a term, which may be confused with the term /seme/ or /sème/ employed in compositional analysis and standing for «semantic marker», therefore possessing a quite different meaning).
A typical example of super-sign is an ‘iconic’ statement such as a man’s photograph which not only means «person x» but «so and so, smiling, wearing glasses, etc.» (which could be a mere description) or «so and so is walking», which clearly corresponds to a verbal sentence.
Thus following Prieto’s suggestion, let us try to list various types of semiotic systems with various types of articulation:
A. systems without articulation: provide for super-signs which cannot be further analyzed in compositional elements:
1) systems with a single super-sign (for example the blind man’s white cane; its presence indicates «I am blind»; whereas its absence does not necessarily mean the opposite, as might be the case, however, for systems with zero sign-vehicle);
2) systems with zero sign-vehicle (the admiral’s flag on a ship; its presence indicates «admiral on board» and its absence «admiral off board»; the directional signals of an automobile, whose absence means «I am proceeding straight ahead»);
3) traffic lights (each unit indicates an operation to carry out; the units cannot be articulated among themselves to form a text, nor can they be further analyzed into underlying articulatory units);
4) bus lines labelled by single numbers or letters of the alphabet.
B. Codes with second articulation only: the units are super-signs. These cannot be analyzed into signs but only into figurae (which do not represent portions of the content of the main units):
1) bus lines with two numbers: for example line /63/ indicates that it «runs from place X to place Y»; the unit can be segmented in the figurae /6/ and /3/, which do not have any meaning;
2) naval ‘arm’ signals: various figurae are allowed for, represented by various inclinations of the right and left arm; two figurae combine to form a letter of the alphabet; this letter is not usually a sign because it is without meaning. It acquires the latter only if it is considered as an articulatory element of verbal language and is articulated according to its laws; however, it can acquire a conventional value within the naval code, indicating for instance «we need a doctor» and must then be considered as a super-sign.
C. Codes with first articulation only: the main units can be analyzed into signs but not thereafter into figurae:
1) the numeration of hotel rooms: the unit /20/ usually indicates «first room, second floor»; it can be subdivided into the sign /2/, which means «second floor» and into the sign /0/, which means «first room»;
2) street signals with units analyzable into signs: a white circle with a red border which contains the black outline of a bicycle means «cyclists not allowed» and can be broken down into the expression //red border//, which means «not allowed» and the image of the bicycle, which means «cyclists».
D. Codes with two articulations: super-signs can be analyzed into signs and figurae:
1) verbal languages: phonemes are articulated into morphemes and these in turn into broader syntagms;
2) telephone numbers with six digits: some can be broken down into groups of two digits, each of which indicates (according to position) a section of the city, a street, an individual; whereas each sign of two digits can be broken down into two figurae which have no meaning.
E. Codes with mobile articulation: in some codes there can be both signs and figurae but not always with the same function; the signs can become figurae or vice versa, the figurae super-signs, other phenomena can assume the value of figurae, etc.:
1) tonal music: the notes of the scale are figurae which are articulated into signs (partially significant configurations) such as intervals and chords; these are further articulated into musical syntagms. A given melodic succession is recognizable no matter what instrument (and therefore what timbre) it is played on; but if one changes the timbre for every note of the melody in a conspicuous fashion, one no longer hears the melody but merely a succession of timbres; and so the note is no longer a pertinent feature and becomes a free variant while the timbre becomes pertinent. In other circumstances the timbre, instead of being a figura, can become a sign bearing cultural connotations (such as a rustic bagpipe-pastoral) (cf. Schaeffer, 1966);
2) playing cards: here we have elements of second articulation (the units of the suits, such as hearts or clubs) which combine to form signs endowed with meaning in relation to the game (the seven of hearts, the ace of spades). These may combine into ‘card-sentences’ such as «full» or «royal flush». Within these limits a card game would be a code relying upon an expression-system with two articulations; but it must be noted that there exist in this system (a) some signs without second articulation, e.g. ‘iconological’ super-signs such as “King” or “Queen”; (b) iconological super-signs which cannot be combined into sentences together with other signs, such as the joker or, in certain games the Jack of Spades. Moreover the figurae can, in turn, be distinguished by both shape and color, and can be selected according to various pertinent criteria from game to game; thus in a game in which hearts are of greater value than spades, the figurae are no longer without meaning, but can be understood as signs. And so on: within the card system it is possible to introduce the most varied conventions of play (even those of fortune-telling) through which the hierarchy of articulations can change.
F. Codes with three articulations: according to Prieto it is difficult to imagine such a type of code for, in order to have a third articulation unit, one needs a sort of hyper-unit (the etymology is the same of ‘hyperspace’) composed of ‘signs’ of the more analytical articulation so that its analytical components are not parts of the content that the hyper-unit conveys (in the same way in which figurae are analytical components of signs but the former are not conveying a part of the meaning of the latter). It seems to me that the only instance of third articulation can be found in cinematographic language. Suppose (even if it is not that simple) that in a cinematographic frame there are visual non-significant light phenomena (figurae) whose combination produces visual significant phenomena (let us call them ‘images’ or ‘icons’ or ‘super-signs’). And suppose that this mutual relationship relies on a double articulation mechanism.
But in passing from the frame to the shot, characters perform gestures and images give rise, through a temporal movement, to kinesic signs that can be broken into discrete kinesic figurae, which are not portions of their content (in the sense that small units of movement, deprived of any meaning, can make up diverse meaningful gestures). In everyday life it is rather difficult to isolate such discrete moments of a gestural continuum: but this does not hold true for the camera.
Let me stress the fact that kinesic figurae are indeed significant from the point of view of an ‘iconic’ language (i.e. they are significant when considered as photographs) but are not significant at all from the point of view of a kinesic language! Suppose that I subdivide two typical head gestures (the sign for /yes/ and the sign for /no/) into a large number of frames: I would find a large number of diverse positions which I would not be able to identify as components of one particular gesture. The position //head tilted toward the right// might be either the figura of the sign //yes// coupled with the pointer //indication of the person on the right// or the figura of a sign //no// coupled with //lowered head// (a gesture that may convey various connotations). Thus the camera offers kinesic figurae devoid of content, which can be isolated within the spatial limits of the frame (see Eco, 1968, B.3.I.).
All these alternatives are suggested simply to indicate how difficult it is to fix, in the abstract, the level of articulation of some systems The important thing is avoid trying to identify a fixed number of articulations in fixed interrelationship. According to the point of view from which it is considered, an element of first articulation can become an element of second articulation and vice versa.
After establishing that systems have various types of articulation and that therefore there is no reason to bow to the linguistic model, we must also remember that a system is often articulated by setting up as pertinent features those elements which are the syntagms of a more analytic system; or that, on the contrary, a system considers as syntagms (the ultimate limit of its combinational possibilities) those elements which are the pertinent features of a more synthetic system. A similar possibility was observed in the example of sailors’ arm signals.
Language considers phonemes to be its ultimate articulatory elements, but the code of naval flags involves figurae that, in relation to phonemes, are more analytic (position of the right arm and position of the left arm), these combining to provide syntagmatic configurations (ultimate in relation to that code) which correspond, practically speaking (even though they transcribe letters of the alphabet and not phonemes) to the figurae of the verbal language.
However, a system of narrative functions contemplates large-scale syntagmatic chains (or the kind /hero leaves his house and meets an enemy/) which, for the purposes of the narrative system, are pertinent features, while for the purposes of the linguistic one they are syntagms. Thus a code decides on what level of complexity it will single out its own pertinent features, entrusting the eventual internal (analytic) codification of these features to another code. If one takes the narrative unit /hero leaves home and meets an enemy/ the narrative code isolates it as a complex content-unit and does not concern itself about the language in which it can be expressed and the stylistic and rhetorical devices which contribute to its construction.
All these are examples of successive overcoding. Usually in overcoding the minimal combinational units are the maximal combined chains of a preceding basic code. But sometimes there also is overcoding when the minimal combinable units or the minimal analyzable clusters of a given code are submitted to a further analytical pertinentization.
See for instance the various experiments in which a scanner is used to decompose and analyze an image into distinctive features, convey them to the computer by means of binary signals, and reproduce them in output through a plotter that draws very complex rasters capable of defining any type of image (their complexity is merely a matter of the complexity of the technical apparatus, but in theory it is by no means impossible to reproduce by means of a very refined raster Leonardo’s Mona Lisa, once it has been programmed in input by means of a very complex sequence of binary signals).
For example, Huff (1967) has produced and analyzed a number of images showing how they could be composed of: (a) elementary units formed by four dots of two sizes, allowing five combinational possibilities; (b) an infinite array of dot sizes, allowing continuous gradations; (c) elementary units formed by a grouping of three dots with two variations in size, so that their combinational possibilities support four types of elements (three small, none large; two small, one large; one small, two large; none small, three large); (d) arrays of dots of two sizes; (e) etc. In every case a question is raised: are we still confronted with a series of analogical sizes? Or are we faced with a series of discrete units such as phonemes, which are distinguished from each other by a series of distinctive features? In this case the distinctive features of the minimal graphic units, described by Huff, are: color, density, form, position of the elements, not to speak of the configurations of the lattice.
In any case Huff himself poses the problem of a binary reduction of the graphic code: “Perhaps (the designer) will even explore the minimal situation by working with elements of only two sizes, ergo, a binary system. In so doing he does meet a most formidable problem: for, in order to maintain a continuous surface, he must solve between two textural gradients in a manner other than the photomechanical process does. Perhaps these operative economies, practised by students of hand-produced rasters, constitute a finesse of little consequence for the computer graphic technique which hypothetically has the capability to formulate the light and shade characteristics of any conceivable surface, thereby matching the photomechanical process. It does seem, however, that the gradation of one size elements in tones of brightness rather than the gradation of one-color elements in varying sizes, though resourcefully adventurous, is ill directed effort – somehow contrary to the fundamental simplicity of digital or binary computers”.
Clearly Huff’s discussion concerns the practical possibilities of graphic realization and not the theoretical possibilities of an absolute binary reduction of the code. In this last sense the examples given by Moles (1968) seem more decisive. He shows for example lattices composed of a single right-angle triangle placed in the upper or lower corner of a square compartment, so as to be able to function in the opposition ‘empty place vs. full place’.
In any case discussion of the binary possibilities of rasters in photomechanical reproduction (which is governed by criteria of practicality) is outweighed by discussion of the possibility of a realizing any ‘iconic’ image by giving digital instructions to a computer which then transmits them to an analogical plotter (28).
Obviously the computer digitally commands a plotter which restores the image by ‘analogical’ means (Soulis and Ellis, 1967:150-151). Cralle and Michael (1967:157) further explain that “When we wish to plot something, we also have to say where to plot it. The addressing scheme normally chosen is obtained by imagining a two-dimensional Cartesian coordinate system, superimposed on the screen of the CRT. In both the horizontal (x) and the vertical (y) directions we can assign integers for each point to which the electron beam may be digitally deflected”. Experiments of ‘iconic’ reproduction by means of computers, such as those carried out in the Bell Telephone laboratories by Knowlton and Harmon, by the Japanese Computer Technique Group, show that the digital programming of ‘iconic’ signs can by now achieve in future high degrees of sophistication and that a greater sophistication and complexity is merely a question of time and economic means. Unfortunately this digital reduction concerns the possibility of replicating the expression using another continuum by a procedure which is not the one used by the artist. It does not concern the articulatory nature of the original expressive functive.
Computer experience tells us that it is possible (in principle) to analyze the original signal in figurae, but not that the signal was actually articulated by combination of pre-existent discrete entities. And in fact such entities are hardly identifiable, since the original signal was composed through a ‘continuous’ disposition of a ‘dense’ stuff. Thus replicability through computers or other mechanisms does not directly concern the code governing the replicated sign. It is rather a matter of technical codes governing the transmission of information (a signal-to-signal process), to be considered within the framework of communication engineering (see 1.4.4.). One could suspect that such procedures are rather connected with the production of doubles (see 3.4.7.) or partial replicas (see 3.4.8.). And this is so when a computer transfers an original ‘linear’ drawing into a plotted copy. But things go differently when an ‘absolutely dense’ oil painting is ‘translated’ into a ‘quasi-dense’ raster; in such a case it is very difficult to decide if one is dealing with a partial replica, an ‘icon’ or a pseudo-double.
Let us speak of transformation from expression to expression that offers a satisfactory laboratory model of the procedures required in cases of projection by ratio difficilis (see the models of projections in 3.6.7.).
These examples also demonstrate that, even in cases of non-replicable super-signs, there is the possibility of rendering them replicable using mechanical procedures that institute a ‘grammar’ there where was only a ‘text’. In this sense these experiments provide us with certain challenging theoretical suggestions about the nature of inventions.
Every assumption about the analogical nature of ‘iconic’ signs was always based upon (or aiming to support) the notion of the ineffability and the ‘unspeakability’ of those devices that signify through being mysteriously related to the objects. To demonstrate that at least the signals ordered to those sign-functions are open to analytical decomposition does not solve the problem but does eliminate a sort of magic. One could therefore say that the digital approach constitutes a sort of psychological support for the student who wants to further understand the mystery of iconism. When deciphering a secret message one must first be sure that it is indeed a message and therefore that there is an underlying code, to be ‘abduced’ from it; in the same way the knowledge that iconic signals also are digitally analyzable can help to promote a further enquiry as to their semiotic nature.
To return to the problem of replicas, one can replicate:
(i) features of a given system that must be combined with features of the same system in order to compose a recognizable functive;
(ii) features from a weakly structured repertoire, recognizable on the basis of perceptual mechanisms and correlated to their content by a large-scale overcoding, that must not necessarily be combined with other features;
(iii) features of a given system that must be added to a bundle or to a string of features from one or more other systems in order to compose a recognizable functive.
Features of type (i) have been considered in the preceding paragraph. Verbal language, for instance, combines elements of second articulation to construct elements of first articulation and therefore phrases. Features of type (ii) are stylizations; features of type (iii) are vectors.
I mean by stylizations certain apparently ‘iconic’ expressions that are in fact the result of a convention establishing that they will not be recognized because of their similarity to a content-model but because of their similarity to an expression-type which is not strictly compulsory and permits many free variants.
A typical example of this sort of replica is the King or the Queen in a pack of cards. We do not ‘iconically’ recognize a «man» and then a «King»; we immediately grasp the denotation «King» provided that certain pertinent elements are respected. It is also on this basis that ‘iconograms’ are coded, i.e. recognizable categories in painting such as the Virgin Mary, Saint Lucy, Victory, Athena, the Devil. In these cases the immediate denotation is a matter of ‘invention’ (they are productions governed by a ratio difficilis that establishes certain similarities with a male or female body and so on) while their full signification (this «man» is «Jupiter») is due to the presence of overcoded replicable features (stylizations).
So a painted image of the Devil is a super-sign which will be further analyzed when speaking of ‘inventions’. But, among other procedures, the replica of large-scale overcoded properties contributes to the structuring of such a sign-function. Insofar as it is an iconogram, the image of the Devil is a replica of a previously coded type, irrespective of a lot of free variants.
In fact when looking at the King of Spades or an image of the Virgin Mary we do not really have to grasp the representative meaning of the image, we do not interrogate the expression in order to guess, through a sort of backward projection at the format of the content-type. We immediately recognize this large-scale configuration as if it were an elementary feature. Some general properties having been respected, the expression is recognized as being conventionally linked to a certain content; the content can also be conceptually grasped without having recourse to its spatial and figural markers. The iconogram is a label.
In this sense even vaster configurations can be taken as stylizations; even if a more analytical glance will show them to be composed by more subtle operations. But if this analysis is not performed, they are received as if governed by a ratio facilis, even if they display the same markers as the corresponding sememe (ratio difficilis).
Let us list some of these large-scale stylizations, each category constituting a repertoire of conventional expression, therefore a sub-code:
(i) heraldic features such as the unicorn that supports the arms of the British royal family;
(ii) schematic onomatopoeias, such as /to sigh/ or /to bark/ (these could be analyzed as degenerate full onomatopoeias, and therefore as fictive samples (see 3.6.3.) but in fact they are currently accepted as arbitrary expressions;
(iii) coded macro-ambiental features, such as, in architecture, a house, a temple, a square, a street;
(iv) complex objects and their customary images (like the cars portrayed in advertising);
(v) musical types (a march, ‘thrilling’ music);
(vi) literary or artistic genres (Western, slapstick comedy);
(vii) all the elements of the so-called recognition codes (see 3.5.) by which a leopard is characterized by spots and a tiger by stripes (granted that an elementary ‘feline’ outline has been recognized on the basis of certain similarity procedures);
(viii) iconograms, as studied by iconology: the Nativity, the Last Judgment, the Four Horsemen of the Apocalypse and so on.
(ix) pre-established evaluative and appreciative connotations conveyed by iconological means, and mainly used in Kitsch-art: a Greek temple conventionally connoting «classical beauty», a given feminine image historically connoting «grace» or «sex»;
(x) other characterizations, such as one perfume immediately connoting «seduction» or «lust» while another connotes «cleanliness», the incense smell connoting «church», and so on.
Beyond a certain limit it is very difficult to distinguish a stylization from an invention, and frequently the decision is taken not by the sender but by the addressee, who in effect performs a labor of stylization on a given expression. Everyone has experienced how a given musical composition that has for many years been enjoyed as a complex text, with all its features subjected to intensive scrutiny, is at a certain point (as one’s taste becomes accustomed to the musical object in question) simply received as an unanalyzed form that means approximately «Fifth Symphony» or quite simply «Romanticism» or «Music».
Thus stylizations are catachreses of previous inventions, super-signs that could and should convey a complex discourse (being a text) and indeed almost take on the function of proper names. Their replica, however imprecise it may be, is taken as a sufficient token, and as such faithful to its expression-type. They are the proof that a ratio difficilis may, by force of continuous exposure to communication and successive conventions, become a ratio facilis.
Stylization may also combine with other devices to make up a discourse; for instance, by putting together certain stylizations with combinational visual replicas, a road signal could ‘say’: «this road is closed to trucks, cars must run at no more than 30 miles and U-turning is forbidden; please make no noise since there is a hospital in the vicinity».
Let us now examine those features that are not combinable with features of the same system but which exclusively collaborate with features of other systems so as to make up an expression. I have called them vectors.
The classic example is the one (already given in 2.11.4., 2.11.5. and 3.5.7.) of the pointing finger: dimensional features realized by a part of a human body, such as «linearity» and «apicality» are the same as those realized by a graphic arrow; in this sense the pointing finger should be considered as an expression produced by an aggregation of combinational units, like a verbal expression, and so should the arrow. But the finger moves toward something; there is a feature of direction (which naturally characterizes a lot of other kinesic features, though here these features of movement can be articulated with other features of the same type). This directional feature orientates the attention of the addressee according to parameters such as left’, ‘right’ or ‘up’ and ‘down’ and so on. But these are not simple spatial parameters of the type left vs. right’, to be used as combinational units in other kinesic configurations; they should instead be viewed as left-to-right vs. right-to-left’.
The addressee does not have to physically follow that direction (nor indeed does there have to be anything in the indicated direction for the pointers to be significant, see 2.11.5.). As a matter of fact there are two ‘directions’: one is actually and physically perceptible and is an expressive feature; the other is the ‘signified’ direction and is mere content. The directional feature is produced according to a ratio difficilis because the produced direction is the same as that of which one is ‘speaking’.
In order to understand vectors, one must also think of other kinds of directional feature, and one must free the term ‘direction’ from spatial connotations (this perhaps being better realized by the word ‘vector’ or ‘vectorialization’). One may thus regard as vectorial devices the increasing or decreasing of vocal pitch and dynamics in paralinguistic features: for instance when uttered, a ‘question intonation’ is a vectorialization; the nature of a musical melody is grasped not only because of the articulation of combinational units but because of their precise temporal succession. Thus even syntactic-phrase markers must be considered as vectors (29).
In /John beats Mary/ it is the direction of the phrase (a spatial direction in the written phrase and a temporal one in the uttered one) that makes the content understandable; by changing round the proper names the entire content is reversed. Again, a vectorialization is neither a sign nor a complete expression in itself (except taken as an expression signifying a pure vectorial correlation as in /a⊃b/), but rather a productive feature that, in conjunction with others, contributes to the composition of the expression (30). One could say that in some cases a vector by itself can give rise to a sign-function; suppose that I hum an upward pitch-curve; I can succeed in signifying «question» (or «I am questioning» or «what?») by imposing a direction on a sound-continuum without resorting to any other device. But this is a case of coded stylization.
Many vectors are governed by a very schematic ratio difficilis so easily recognizable that, as happens with stylizations, a sort of catachresizing process takes place and the ratio difficilis practically becomes a ratio facilis. The case of the interrogative humming cited above is a typical example of this process.
Half way between replica and invention there are two kinds of productive operation that are not usually considered as semiotically definable. The first one concerns the disposition of non-semiotic elements intended to elicit an immediate response in the receiver. A flash of light during a theatrical performance, an unbearable sound, a subliminal excitation, and so on, are to be listed among stimuli rather than signs, as was stressed in 3.5.5. But in the same paragraph we noted that, when the sender knows the possible effect of the displayed stimulus, one is obliged to consider his knowledge as a sort of semiotic competence, for to him a given stimulus corresponds to a given foreseeable reaction that he expressly aims to elicit. In other words, there is a sign-function by which the stimulus is the expression plane of a supposed effect functioning as its content plane.
Nevertheless the effect of a stimulus is never completely predictable, especially when inserted among other more specifically semiotic elements within a text as a pseudo-sign. Suppose that a speaker is elaborating a persuasive discourse according to the rules of judiciary rhetoric and trying to arouse in his addressees feelings of pity and compassion. He can utter his phrases in a throbbing voice, or with barely detectable vibrations that could suggest that he is tempted to cry. These supra-segmental features could obviously be either paralinguistic devices or mere symptoms indicating his emotional state; but they might also be stimuli he inserts into the discourse in order to provoke some degree of identification in his listeners and to pull them toward the same emotional state. He is using these devices as programmed stimulations but does not know exactly how they will be received, detected, interpreted. The speaker is thus half way between the execution of certain rules of stimulation and the displaying of new unconventionalized elements that might (or might not) become recognized as semiotic devices. Sometimes the speaker is not sure of the relation between a given stimulus and a given presupposed response, and he is more making than performing a tentative coding of programmed stimuli. Therefore these devices stand between replica and invention; and may or may not be semiotic devices, thus constituting a sort of ambiguous threshold. So that even though the expressive string of programmed stimuli can be analyzed into detectable units, the corresponding content remains a nebula-like conceptual or behavioral ‘discourse’. The expression, made of analyzable and replicable units (governed by a ratio facilis) may then generate a vague discourse on the content plane. Among such programmed stimuli one might list: (i) all the programmed synesthesiae in poetry, music, painting, etc.; (ii) all so-called ‘expressive’ signs, such as those theorized by artists like Kandinskij, i.e. visual configurations that are conventionally supposed to ‘convey’ a given feeling directly (force, grace, instability, movement and so on) and that have also been studied by the theorists of Einfühlung or empathy; insofar as these devices hold a motivated relationship with psychic forces or ‘reproduce’ physical experiences, they should be dealt with in the paragraph concerning projections (3.6.7.); insofar as they are displayed by a sender who knows their emphatic effect, they are programmed stimulation (and therefore precoded devices) of which, however, the result (on the content plane) is only partially foreseeable; (iii) all production of substitutive stimuli described in 3.5.8.; (iv) many projections, about which more will be said in 3.6.7.
Anyway one should carefully distinguish between this sort of programmed stimulus and the more explicitly coded devices used to express emotions, such as body movements, facial expressions, and so on, now so precisely recorded by the latest researches in kinesics (Ekman 1969) and in paralinguistics.
Another kind of spurious semiotic operation is pseudo-combination. The most typical example is an abstract painting or an atonal musical composition. Apparently a Mondrian painting or a Schoenberg composition is perfectly replicable and therefore appears to be composed by systematically combinable units. These units are not apparently endowed with meaning but they do follow combinational rules.
Nobody can deny that there is an expression system even though the content plane remains, as it were, open to all comers. These examples are thus more open signal textures than sign-functions; for this very reason they appear to invite the attribution of a content, thus issuing a sort of interpretive challenge to their addressee (Eco, 1962). Let us call them visual or musical propositional functions that can only ‘wait’ to be correlated to a content, each being susceptible of many different correlations.
Thus when hearing a post-Webernian sound cluster one detects the presence of replicable musical units combined in a certain fashion and sometimes one also knows the rule governing this kind of aggregation of material events.
However, the problem seems to change when one is dealing with abstract expressionist paintings, random music, John Cage’s happenings and so on. In these cases one can speak of textural clouds which lack any predictable rule. Can one then continue to speak of a pseudo-combinational operation? It is exactly this kind of artistic operation which prompted Lévi-Strauss (1964) to deny any linguistic nature to these phenomena, in view of their lack of discrete units or of oppositions based on an underlying system.
One could respond that in these cases the entire material texture, through its very absence of rules, opposed itself to the entire system of rules governing ‘linguistic’ art, thus creating a sort of macrosystem in which manifestations of pure noise are opposed to manifestations of informational order. This solution has the advantage of elegance and does in fact explain many of the intentions behind the work of ‘informel’ artists, but it is equivalent to maintaining that even in non-semiotic phenomena there is a semiotic purport insofar as they are displayed in order to make absent semiotic phenomena relevant. In this sense the creation of ‘art informel’ would be the same as silence in order to ‘express’ refusal to speak.
As a matter of fact there is another reason why many examples of this kind of art have at least the nature of pseudo-combinations. The clue is given by artists themselves when they tell us that they examine the very veining of material, the texture of the wood, canvas, iron or sounds and noises, trying to find in them relationships, forms, new visual or auditory paths. The artist discovers at the deeper level of the expression-continuum a new system of relations that the preceding segmentation of that continuum, giving rise to an expression form, had never made pertinent. These new pertinent features, along with their mode of organization, are so detectable and recognizable, that one becomes able to isolate the work of a given artist, and thus to distinguish, for instance, Fautrier from Pollock or Boulez from Berio.
In this case the establishing of pseudo-combinational units does not precede the making of the work itself; on the contrary, the growth of the work coincides with the birth of the systems. And, provided that these forms convey a content (which is sometimes identical with a metalinguistic account of the nature of the work and its ideological purport), an entire code is proposed as the work is established.
Let me stress that we are here dealing with three problems: (i) the segmentation performed below the level of the recognized expression form, that is, a further segmentation of the expression-continuum; this aspect will become very important in section 3.7. when speaking of the aesthetic text; (ii) the complexity of this segmentation at various levels, which sometimes makes it impossible to detect distinguishable units, thereby making it impossible to establish replicable expression types; when this happens pseudo-combinational units cannot be replicated (in post-Webernian music some sound-clusters can be replicated – indeed there is a score prescribing their way of performance – while others can only be ‘suggested’ by the composer and require an inventive participation on the part of the performer; a Dubuffet painting can hardly be replicated); (iii) the invention of new expression levels along with their possible segmentation and systematization; in such cases pseudo-combinational procedure turns into purely inventive procedure, thus bringing us to the last item in the present classification of modi faciendi signa.
In Table 39 pseudo-combinational units are nevertheless listed among the modes of production governed by a ratio facilis because, as long as they are replicable, they have to reproduce an expression type, though it seems doubtful that they represent a definite case of sign-function so much as one of an ‘open’ signal. But if their constitutive units are not detectable, they are not replicable, and they thus remain half way between sign production and the proposal of new possibilities for manipulating continua.
It is not by chance that programmed stimuli have on the contrary been listed in the same row as examples and samples, in a middle position between ratio facilis and ratio difficilis. Sometimes, as the empathy theorists assume, there is a sort of ‘motivated’ link between a certain line and a certain feeling, and thus cases of stimulation rely on procedures of projection or stylization.
We may define as invention a mode of production whereby the producer of the sign-function chooses a new material continuum not yet segmented for that purpose and proposes a new way of organizing (of giving form to) it in order to map within it the formal pertinent element of a content-type. Thus in invention we have a case of ratio difficilis realized within a heteromaterial expression; but since no previous convention exists to correlate the elements of the expression with the selected content, the sign producer must in some way posit this correlation so as to make it acceptable. In this sense inventions are radically different from recognition, choice, and replica.
Everybody recognizes an expression produced by recognition because a previous experience has linked a given expression-unit with a given content-unit. Everybody recognizes an expression produced by a choice made on the basis of a common mechanism of abstraction, such as the acknowledging of a given item as representative of the class to which it belongs. Everybody recognizes an expression produced by replica, because the replica replicates an expression-type which has already been conventionally correlated with a given content. In all these cases, whether the ratio is facilis or difficilis, everybody recognizes the correspondence between a token and its type because the type already exists as a cultural product. Whether the token expression reproduces a content type, as in the case of imprints, or an expression type, as in the case of phonemes and words, the procedure follows certain basic requirements.
If one views a type (whether of content or of expression) as a set of properties that have been singled out as pertinent, the token is obtained by mapping out the elements of the original set in terms of those of the token set. This procedure can be represented by Table 40, where the xs represent the pertinent properties of the type and the ys non-pertinent and variable elements (31).
In cases of ratio facilis mapping presents no problem; it simply involves the reproduction of a property using the same sort of material as that prescribed by the type. In the case of a phoneme the type may, for instance, prescribe ‘labial+voiced’ (thereby implying: by means of human phonation), thus establishing how to produce a [b].
The notion of mapping is somewhat more problematic in cases of ratio difficilis, because the type of a ratio difficilis is a content unit, a sememe, and its properties are semantic markers, and are not in principle linked to any particular expression continuum.
So what does one mean by mapping the pertinent properties of a glass of wine within another material so as to produce the recognizable wet imprint of a glass of wine upon a table? Formulating the question in this way might make for a puzzling answer, but this is because of one’s ‘referential’ bias. As a matter of fact the imprint of a glass of wine does not have to possess the properties of the object «glass of wine» but it does have to possess those of the cultural unit «imprint of a glass of wine». And in this case the semantic representation of the entity in question entails no more than four semantic markers, i.e. «circle», «red», «length of the inradius (or diameter)» and «wet». To map these markers within another material simply means to realize the geometrical and chemical interpretants of the sememes «circle», «red», «diameter X» and «wet». This done, the mapping process is complete, and the realization of a token of the content type a comparatively easy matter. In this sense one cannot maintain that the imprint of a hare’s paw is an iconic feature in the same way as is the image of a hare. In the former case the content type is culturally established, whereas in the latter one it is not (except in cases of stylization).
The only problem would appear to be: in what sense does a circle of a given diameter realized upon a table map the semantic markers «circle» and «diameter X»?
But on second thoughts, that question is not so different from asking in what sense a labial and voiced consonant maps the abstact phonological type ‘labial+voiced’ in sound. In the latter case the answer seems easy enough: there are certain sound parameters which permit the realization and recognition of the replica (as to how the realization of a parameter is recognizable, this sends us back to basic perceptive requirements that, as was noted in 3.4.7. and 3.4.8., are postulates rather than theorems for a semiotic theory).
Thus one need only repeat that (as was underlined in 3.4.2.) various expressions may be realized whether in accordance with spatial parameters or phonic parameters in order to justify listing the replica of a circle in the same theoretical row as the replica of a phoneme. The only difference is that the sound features governing the reproduction of a phoneme are not content markers, while the spatial features governing the reproduction (even if virtual, as in the recognition of imprints) of a geometrical figure are. This – as we have seen – is exactly the difference between ratio facilis and ratio difficilis.
Now if one considers Table 39, one notices that all the cases of ratio difficilis concern content types in which the most important semantic markers are toposensitive, i.e. figural or vectorial properties. This brings us back to the problem outlined in 2.7.2.: not every semantic marker can be verbalized. When semantic markers can be verbalized they have undoubtedly acquired a maximum of abstraction; previously culturalized and frequently expressed through verbal devices, they can even be arbitrarily correlated with other non-verbal devices (for example a geometrical form in a road signal meaning «stop»), and through the mediation of verbal habits they can easily be detected. In these cases it is true that, as Barthes and other theorists say, non-verbal semiotic systems rely on the verbal one. But there are markers that cannot be verbalized, at least not completely, so that they cannot be conveyed by a metalinguistic definition verbally expressed.
The spatial disposition of the imprint of a hare’s paw cannot be verbally meta-described. It is, however, hard to assert that is has no cultural ‘existence’, and the proof is not in the fact that it can be ‘thought’ (which would be an extra-semiotic and somewhat mentalistic argument) but in the fact that it can be interpreted in many ways. For instance one can conceive of an algorythm which, when fed into a plotting machine as input, would produce as its output a drawing of a hare’s paw. The fact that this drawing is more schematical than a real imprint is a further proof of the present thesis: the cultural notion of such an imprint (a sememe) is neither the same as its perceptual model nor as the corresponding object.
The process from perceptual model to semantic model and from semantic model to an expressive model governed by a ratio difficilis, may be represented as in Table 41.
In other words, given a perceptual model as a ‘dense’ representation of a given experience, assigning to the perceived object x the properties x1, x2, x3, x4 . . . xn, that perceptual model gives rise to a semantic model which preserves only, let us say, three of the properties of the dense representation. It is not said that all those selected semantic markers are necessarily verbalizable items; many of them may be toposensitive relationships.
At this point it would be possible to express this semantic model (a sememe) by means of an expressive device. If the markers of the sememe were non-toposensitive, the correlation content-expression could be an arbitrary one. Since, however, in this case some markers are toposensitive, the correlation is motivated, and must follow in principle the rules governing every type/token-ratio, i.e. rules of transformation.
Let us now add something about the double mapping outlined in Table 41. The first kind of mapping (from percept to sememe) does not need to be semiotically explained: it follows the rules governing every phenomenon of abstraction – both in conceptual and ‘Visual’ thinking – and is therefore a procedure depending on the mechanisms of human intelligence (which is not to say that even this procedure could not be seen as a semiotic one, but rather that the definition of this problem constitutes one of the ‘political’ boundaries of semiotics – see Introduction).
The second kind of mapping should be identical to that which governs the production of a triangle that is similar to another, given certain spatial parameters and conventions (such as that size is irrelevant, but sides must be proportional and angles ‘equal’). Let us call this procedure a transformation: “every biunivocal correspondence of points in space is a transformation. What concerns us is the existence of particular transformations that leave certain prominent properties of the geometrical entities to which they are applied unchanged” (32). This concept of transformation fits cases of token-to-token reproduction as well as those of type/token-ratio perfectly (this being one of the postulates of semiotics). But it also explains cases such as the production (even if virtual) of an imprint, which is why in 3.6.2. even imprints were said to be cases of transformation.
But in cases of type/token-ratio, mapping by similitude takes place between an expression type (and thus the model of an object) and an expression token (and thus another physical object). In the case of the imprint, on the other hand, we are considering similitudes established between a semantic model and its physical expression. We are once again concerned with the difference between ratio facilis and ratio difficilis.
At this point two problems arise:
(i) how to ‘map’ from a content-model into an expressive one, i.e. from a non-physical reality into a physical continuum;
(ii) how various kinds of mapping may be listed according to a degree of conventionality reached by the content-type and its toposensitive complexity.
If, in Table 39, imprints (even if accidentally replicated rather than recognized) were not classified as straightforward transformations under the heading of inventions, this was for a good reason. In the case of an imprint the content-model already exists. It has, in one way or another, been culturally established. When replicating an imprint one is mapping from something known. And there exist similitude rules establishing how to embody in a material continuum certain semantic toposensitive properties of a sememe (as in the case of the glass of wine). So the mapping procedure by ratio difficilis in Table 41 is not so different from that performed in cases of ratio facilis (Table 40). This mapping is undoubtedly motivated by the sememic representation of the supposed object but is at the same time ruled by mapping conventions. The main problem arises when trying to determine how it is possible to map onto an expression continuum the properties of something which (because of its cultural oddity or formal complexity) is not yet culturally known.
It must be stressed that one is not here concerned with the representation of a golden mountain, or a man with ten eyes and seven legs. It is very easy to infer the nature of unknown elements from the addition of known ones, just as language manages to express unheard-of events by articulating recognizable units. But the real puzzling problem is not so much how one may represent a man with ten eyes and seven legs as why one may visually represent (and recognize as represented) a given man with two eyes and two legs. How it is possible to represent a man standing and a lady sitting under a tree, a calm landscape with clouds and a corn-field behind them, a given light and a given mood – as happens in Gainsborough’s Mr. and Mrs. Andrews?
Since this complex content is not a unit but a discourse (and the painting is not a sign but a text), and since that content was not previously known by the addressee grasping for the first time from an expression for which no type previously existed, how is it possible to define this phenomenon semiotically? The only solution would seem to be that painting is not a semiotic phenomenon, because there is neither pre-established expression nor pre-established content, and thus no correlation between functives to permit signification; thus a painting should appear a ‘mysterious’ phenomenon which posits functives instead of being posited by them.
Nevertheless, if such a phenomenon seems to escape the correlational definition of ‘sign-function’, it certainly does not escape the basic definition of a sign as something which stands for something else: for Gainsborough’s painting is exactly this, something physically present which conveys something absent and, in certain cases, could be used in order to mention a state of the world.
With this example we have arrived at a critical point in the present classification of modes of sign-production. We now have to define a semiotic mode of production in which something is mapped from something else which was not defined and analyzed before the act of mapping took place. We are witnessing a case in which a significant convention is posited at the very moment in which both the functives of the correlation are invented.
But for the semiotician this latter definition has a rather familiar ring to it. It curiously resembles the problems (so vigorously rejected by at least three generations of linguists) surrounding the origins of language or the historical rise of semiotic conventions.
Now if such a problem can be rejected when it is proposed from an abstract and roughly archeological point of view, it cannot be escaped when approached from the viewpoint of a phenomenology of modes of sign-production. Let us therefore assume that the problem of these transformations listed as inventions and based on a ratio difficilis (depending on a toposensitive content model) raises the question of the activity of code-making (see Table 31 and paragraph 3.1.2.).
We may now revise the mapping process as proposed in Table 41 as follows (Table 42).
Here (i) relevant elements are picked up from an unshaped perceptual field and organized in order to build a percept; (ii) by means of abstractive procedures very similar to the rules that govern cases of stylization (see 3.6.5.) the percept is mapped onto a semantic representation, the latter being the cultural simplification of the former; (iii) this semantic representation is either arbitrarily associated with a set of expressive devices, as in the case of systematically combinational elements and other kinds of replicas, or (iv) mapped into a transformation according to conventional rules of similitude. These procedures explain every kind of sign production listed in Table 39, except inventions.
An invention can take place in two ways, one moderate and the other radical:
(a) Moderate inventions occur when one projects directly from a perceptual representation into an expression-continuum, thereby realizing an expression-form which dictates the rules producing the equivalent content-unit (Table 43).
This is the case, for instance, in Gainsborough’s painting, as indeed in all so-called ‘classical’ paintings. The same thing happens in the first reproduction (or recognition) of an imprint.
From the sender’s point of view, a perceptual structure is considered as a coded semantic model (even though nobody else would yet view it in this way), and its perceptual markers are mapped into an as yet unshaped continuum according to the more commonly accepted rules of similitude. The sender therefore proposes rules of correlation even though the functive-content does not as yet exist. But from the addressee’s point of view the result is simply an expressive structure.
Using the painting as an imprint, he makes his way backward inferring and extrapolating similitude rules, and finally re-constitutes the original percept. However, the process is not an easy one; sometimes addressees refuse to collaborate, and consequently the convention fails to establish itself. The addressee must be helped by the sender and consequently a painting can never afford to be entirely the fruit of an inventive transformation. It must offer various clues: stylizations, perhaps some pre-coded combinational units, a number of fictive samples and of programmed stimuli. Thus, by dint of a series of complex adjustments, the convention is established.
When this process is successful a new content-plane, lying between the percept (which is only remembered by the painter) and the physically testable expression is brought into being. This is not so much a unit as a discourse. What had been raw content-continuum perceptually organized by the painter in the first instance now gradually becomes a new cultural arrangement of the world. A sign-function emerges from the exploratory labor of code-making, and so establishes itself that the painting generates habits, acquired expectations, and mannerisms. Expressive visual units become sufficiently fixed to be available for further combinations. Stylizations come into being.
The painting now offers manipulable units that may be used for further sign-production. The semiosic spiral, enriched by new sign functions and interpretants, is now ready to start all over again.
(b) The case of radical inventions is rather different, in that the sender more or less bypasses the perceptual model, and delves directly into the as yet unshaped perceptual continuum, mapping his perception as he organizes it (Table 44).
In this case the transformation, the realized expression, is a shorthand device whereby the sender fixes the results of his perceptual labor. Only after carrying out this expressive labor can he arrive at a perceptual model and then subsequently a sememic representation. This process has been present at all the great innovative moments in the history of painting. Take the case of the Impressionists, whose addressees absolutely refused to ‘recognize’ the subjects represented and said that they ‘did not understand’, that the painting ‘did not mean anything’, that real life was not like that, etc. This refusal was due to the addressees’ lack not only of a semantic model to which the mapped items might be referred, but also of a percept to guess at, since they had never perceived in this way.
In such cases what takes place is a radical code-making, a violent proposal of new conventions. The sign-function does not as yet exist, and indeed sometimes fails to establish itself at all. The sender gambles on the possibility of semiosis, and loses. In one or two cases it is only centuries later that the gamble comes off and the convention is established. All these procedures will be further examined in the section devoted to the aesthetic text, thereby implying that code-making and invention are aesthetic activities.
Curiously enough, this assumption carries speculation about languages back to the position adopted by Giambattista Vico, who proposed that languages rise as poetic inventions and are only accepted by convention afterward. This is not to say that the conclusion of this chapter should be interpreted as a verification of idealistic theses that, when carried to extremes, result in the rejection of any semiotic science (or at least of any recognition of the social import of codes). In fact, no one ever really witnesses cases of total radical invention, nor indeed of total moderate invention, since texts are maze-like structures combining inventions, replicas, stylizations, ostentions and so on. Semiosis never rises ex novo and ex nihilo. No new culture can ever come into being except against the background of an old one.
As was said in 2.1. and 2.4. there are no signs as such, and many so-called signs are texts; signs and texts being the result of a labor of correlation in which many variously intertwined modes of sign production take part. If ‘invention’ were a category within a typology of signs, then it would be possible to isolate absolute and radical inventions, which would constitute real examples of the birth of language, demonstrating the continuous recurrence of the ‘auroral’ moment through which, every day in everyone’s life, language comes into the world – just as Croce’s linguistics maintained, overestimating the creative power of the speaking subject.
But since invention is, on the contrary, one among various modes of sign production, collaborating with others to correlate functives and to establish various sign-functions, the idealistic fallacy is avoided.
Man is continuously making and re-making codes, but only insofar as other codes already exist. In the semiotic universe there are neither single protagonists nor charismatic prophets. Even prophets have to be socially accepted in order to be right; if not, they are wrong.
The products of semiotic invention, even if viewed as potential super-signs, are ‘fuzzy’ signs. They do not establish straightforward oppositions as much as possible gradations and they are more subject to undercoding than to coding. It would be wrong to assert that a painting is a complex of recognizable signs like a poem. But it would be equally wrong to maintain that a painting is not a semiotic phenomenon; it represents the moment in which a semiotic phenomenon comes into being, the proposal of a possible code by making use of remnants of previous ones. This being the case, it must be stressed that there are different kinds of transformations, some of them closer to the making of a double for the purposes of pure perception or use, others more akin to a semiotic procedure. Let us list, at any rate, three grades within this continuum.
First of all there are congruences or casts (33): a point in the physical space of the expression corresponds to each point in the space of a real object. One example of this is a death mask. But death masks can be ‘understood’ even if one does not know the model-object (and as a matter of fact they are frequently displayed in order to allow one to detect the physical properties of a person one has never known). Death masks are not absolute congruences (in the full geometrical sense of the term); they discard as irrelevant skin texture, color and many other properties; in fact they can also be reproduced on a smaller scale without losing their representative power. So they, too, must be governed by conventions of similitude. When looking at a death mask one ‘maps backward’. But at the end of the projection stands not an object, but a content-type: therefore they are sign-functions.
Furthermore, it is clear enough that death masks can be faked. So that, however you look at it, these heteromaterial casts must be signs. Only homomaterial congruences are not signs, and these are in fact absolute replicas or doubles! Secondly there are projections (34): points on the space of the expressive token correspond to selected points on the space of toposensitive perceptive or semantic models. Strong similitude rules are at work; one must in fact learn to recognize this kind of ‘image’. There are different styles of projection, and they are easily falsifiable.
Any naive interpreter of a projection ‘reads’ it as an imprint, that is, as the direct mapping from the actual aspects of a thing!
On the contrary, the projection is always the result of a mapping convention by means of which given traces on a surface are stimuli compelling one to map backward and to postulate a content-type where one only sees an expression-token. So it is always possible to project from nothing or from contents to which no referent corresponds (as in a classical painting representing mythological heroes). The existence of social conventions in projections (so that is is possible to map from a perceptual or a semantic model) make easy the reverse procedure, that is, to map from the projection to an unexisting and supposedly projected entity. What reinforces our criticism of naive iconism is that since it is possible to draw false iconic signs, iconism is a matter of a highly sophisticated semiotic convention (35). When considered as mentions, projections are frequently false; they try to assert that something exists, which actually looks like the expression item, when this is not the case at all; they can thus display images of Julius Caesar as well as of Mr. Pickwick, irrespective of the differing ontological status of the two. It is in cases of projections that the so-called ‘scales of iconism’ can be accepted as heuristically useful.
Thirdly, there are graphs or topological transformations (36), in which spatial points in the expression correspond to points of non-toposensitive relation; such is the case in Peirce’s existential graphs (see 3.5.3.): a spatial expression displays information about a correlation which is not spatial at all but which instead concerns, for instance, economic relationships, as when one uses the graph of Table 45
to express the following relationship: “every dependent worker belongs to the class of exploited and alienated proletarians”.
In any case this whole range of inventive representations, from congruences or casts to graphs, displays not signs but texts; when these texts first appear there is as yet no distinction between pertinent and irrelevant features. It is only in the course of decoding them that pertinent features emerge, and they begin to produce signs (and thus their own mannerisms). Because of the difficulty of isolating the content-type to which they refer (by a procedure of ratio difficilis), these texts are not easily replicable.
To successfully copy a painting is no mean feat, and to farce a Rembrandt may well be considered a para-artistic achievement, for it is very difficult to detect the pertinent properties on which the significant power of the expression relies, and only remarkably skilled counterfeiters have a capacity for isolating and reproducing them. When only one person in the world is able to falsify a mode of invention (i.e. not to copy a given painting, but to paint according to the same type of inventive procedure) the code proposed by that painting has not yet been accepted by a culture; when it becomes possible to paint à la manière de, then the invention (as a code-making proposal) has succeeded semiotically; a new convention exists. But it is clear that the present discussion is continuously shifting from the problem of code-making and of the acquisition of new conventions, to the problem of the aesthetic use of a language. Any discussion of invention inevitably opens up the problem of the ambiguous, self-focusing and idiolectal use of a code, and compels us to return, once more, to the discussion on aesthetic texts.
The typology of modes of production of the signal, outlined in this section, has definitely clarified the fact that what one usually calls ‘signs’ are the result of many intertwined modes of production. For instance, a perfume of incense, if smelled in a church, is only a case of recognition, that is, a symptom by which one recognizes that a liturgical ceremony is taking place. When produced, it is at the same time the replica of a stylization and a programmed stimulus. When used during a play in order to suggest a mystical situation, it is both a programmed stimulus and a fictive sample (the incense for the whole ceremony).
A smile can be a symptom or the replica of a stylization, and sometimes even a vectorialization. A musical melody, when quoted in order to recall the entire symphony from which it has been extrapolated, is a sample; but it can be the replica of a text composed by combinational units and, sometimes, even a complex of programmed stimuli mixed with pseudo-combinational units. And it is usually all these things together. A geographical map is the result of a previous transformation (half way between a projection and a graph) which has definitely become a stylization, and as such is the result of a replica. Clothes in general are replicable stylizations with intertwined pseudo-combinational units and programmed stimuli.
The problem becomes more difficult when one must define a painting. In any case, a painting is certainly not ‘a sign’: it is a complex text resulting from the network of many modes of production.
One might suppose that a portrait of a given man represents a perfect case of ‘proper name’ necessarily sending back to a physical referent (while the verbal proper names have been demonstrated in 2.9.2. to have a content). One could better say that such a portrait is neither a sign nor a complex super-sign but rather a mention (/this is a man and he possesses these properties . . . /). On the other hand it might be said that the same portrait is equivalent to a description. Goodman (1968:1.5.) remarks that there is a difference between the picture of a man (the portrait of Napoleon) and a man-picture (the portrait of Mr. Pickwick). In fact such a portrait embodies different types of activities, practically covering the entire range of types of semiotic labor outlined in Tables 31 and 39. It is a mention because, through programmed stimuli, it displays the surrogate of a percept and by means of some graphic devices attributes to it the markers of a possible corresponding sememe; it is an invention insofar as the perceptual model does not yet exist; it is a factual judgment (/there exists a man so and so/) and a description (/a man so and so/). Being still uncoded it is at the same time relying on a lot of already coded features, and the invention is made acceptable by the intervention of coded imprints, stylizations, samples, pseudo-combinational units, vectorializations, and so on. Therefore such a portrait is a complex text whose content ranges from a coded detectable unit («Mr. So and So») to an infinite discourse or a content-nebula. But as far as the portrait is accepted and recognized by a culture, it creates a ‘type’ (in the sense of a ‘literary type’, intended as the representation of some ‘universal’ properties: the Hero, the Gentleman, the Beautiful Lady, la belle dame sans merci, and so on). At this point it becomes the model for further stylizations. So what in a given historical period may be viewed as an inventive projection, in another period becomes a stylization.
The same happens with the so-called ‘architectural signs’. Even if many researches in semiotics of architecture have tried to isolate the existence of ‘architectural signs’ (37), it is absolutely clear that even the most elementary architectural configuration is always a text. Let us consider for instance a staircase. It is undoubtedly a semiotic device which signifies certain functions: but to compose such a device productive labor is requested to display the following features: (i) articulation of pseudo-combinational units; (ii) vectorializations (the staircase indicates a direction making recourse to toposensitive parameters); (iii) programmed stimulations (the staircase in a certain way obliges one to move one’s feet for climbing up); (iv) stylizations (the staircase corresponds to a precise typology); and so on.
It is not without sense to try to isolate precise expression-units in architecture, but it is indispensable to take into account the lot of productive features that these units bring into play.
All this reminds us that, the more a text become complex, the more complex is the relationship between expression and content. There may be simple expression units that convey content-nebulae (see for instance many cases of programmed stimulation); expression-clusters that convey a precise content unit (a triumphal arch can be a very elaborate architectural text and nevertheless convey a strictly conventional abstraction such as «victory»); precise grammatical expressions, composed of replicable combinational units, such as the phrase /I love you/, that in certain circumstances convey dramatically a content-nebula; and so on.
This must not allow one to neglect to isolate precise sign-functions when they are detectable, but serves to remind one that in the semiosic process we are usually facing undercoded or overcoded texts.
When more analytical units are not detectable, it is not a case of denying the existence of a semiotic correlation; the presence of the cultural convention is not only witnessed by the emergence of so-called elementary signs. It is first of all revealed by the detectable existence of modes of semiotic production (recognition, ostension, replica and invention) that the present section has outlined and whose presence demonstrates that – even when there are not precise unit-to-unit correlations – there is, however, a posited sign-function.
The aesthetic use of a language deserves attention on a number of different levels: (i) an aesthetic text involves a very peculiar labor, i.e. a particular manipulation of the expression (see 3.7.2.); (ii) this manipulation of the expression releases (and is released by) a reassessment of the content (see 3.7.3.); (iii) this double operation, producing an idiosyncratic and highly original instance of sign-function (see 3.7.4.), is to some degree reflected in precisely those codes on which the aesthetic sign-function is based, thus releasing a process of code changing (see 3.7.5.); (iv) the entire operation, even though focused on codes, frequently produces a new type of awareness about the world (see 3.7.6.); (v) insofar as the aesthetic labor aims to be detected and scrutinized repeatedly by the addressee, who thereby engages in a complex labor of interpretation, the aesthetic sender must also focus his attention on the addressees’ possible reactions, so that the aesthetic text represents a network of diverse communicational acts eliciting highly original responses (see 3.7.7.).
In all these senses the aesthetic text represents a sort of summary and laboratory model of all the aspects of sign-function: it can perform any or all productive functions (being composed of various types of judgment and acting as a meta-semiotic statement) and it can require any kind of productive labor.
At the same time Table 31 can be viewed as the simplified representation of what happens when an aesthetic text is produced and interpreted.
All of which shows clearly enough why the semiotician may be interested in examining the aesthetic experience. But there are other reasons for attempting a semiotic approach to aesthetic texts, since one may also hope to thereby clarify many problems that traditional philosophical aesthetics has left unsolved.
Typical of many philosophical aesthetic theories is that, rather than define the poetic message, they list the effects that any reader of poetry (or viewer of a visual work of art) may feel. What differentiates the response of philosophical aesthetics from that of the layman is the sophisticated architecture of rhetorical devices which, by means of an imaginative interplay of metaphors, translate a sum of truisms.
Some so-called aesthetic definitions of art can be translated by the statement “art is art” or “art is what produces an aesthetic effect”.
One example is Croce’s theory of the cosmic quality of art: the whole life of the cosmos breathes within the artistic representation, the individual pulsates with the life of the Whole, and the Whole is revealed in the life of the individual. “Every genuine artistic representation is in itself the universe In every word the poet writes and in every creature of his imagination there lies the whole of human destiny, all human hopes, illusions, griefs, joys, greatness and misery; the entire drama of Reality, which develops and grows up upon itself for ever, suffering and rejoicing . . .. ” (38). This definition of poetic effect seems to correspond to certain impressions that we have had in our aesthetic experience; but it is vague and unsatisfying because it says, in an elegant way, what we feel or have felt, but not why. So we must now see whether the semiotic approach gives a better explanation of this effect.
According to the well-known subdivision of the functions of language put forward by Jackobson, a message can possess either one or a combination of the following functions: a) referential; b) emotive; c) imperative; d) phatic; e) metalinguistic; f) poetic. The message assumes a poetic function (though in this context it is preferable to call it an ‘aesthetic’ one, granted that we are dealing with every kind of art) when it is ambiguous and self-focusing. Obviously all six functions can coexist in a single message; in the greater part of everyday language (as well as in aesthetic messages) they are constantly interrelating and overlapping, although one of the functions usually predominates, thereby characterizing the message. Semiotically speaking ambiguity must be defined as a mode of violating the rules of the code. There are totally ambiguous messages (such as /wbstddd grf mu/, which violates both phonetic and lexical rules), syntactically ambiguous messages (such as /John has a when/, which violates subcategorization rules) and semantically ambiguous messages (such as the well-known /green colorless ideas sleep furiously/) but not all these types of ambiguity necessarily produce an aesthetic effect (even though, when inserted in an appropriate context, they could).
Another form of ambiguity is the stylistic one. Coseriu (1952), distinguishing between system and norm, suggests that a langue may allow different performances, all considered as ‘grammatical’, yet some of them will acquire an appearance of ‘normality’, while others will be considered as stylistic (i.e.: upper-class, vulgar, literary) variations. Latin allows one to say /Petrus amat Paulum, Petrus Paulum amat, Paulum Petrus amat/, but the third expression looks less normal than the preceding two. In receiving the third one, the addressee immediately grasps a connotation of excessive elegance.
These norms depend on stylistic subcodes assigning an additional connotation both to isolated words and (more frequently) to ready-made sentences. Stylistic norms are thus an instance of overcoding (see 2.14.3.). When hearing /Paulum Petrus amat/ I am not really concerned with the fact that a man named Petrus loves a man named Paulus; I am interested in the ‘poetic’ (or perhaps ‘Kitsch’) nuances that the expression may suggest. Ready-made rhetorical sentences are also examples of overcoding.
Some stylistic approaches to criticism (Spitzer 1931) speak of the aesthetic as a deviation from the norm. This is not entirely satisfactory because not every deviation from the stylistic norm constitutes an aesthetic achievement: /Amat Paulum Petrus/ is semantically comprehensible and stylistically deviant but it sounds merely rather odd. Moreover, the theory does not make clear whether poetic deviation has to be viewed in relation to the everyday norm or to a poetically established one. In fact there can be deviations of both types.
However, ambiguity is a very important device because it functions as a sort of introduction to the aesthetic experience; when, instead of producing pure disorder, it focuses my attention and urges me to an interpretive effort (while at the same time suggesting how to set about decoding) it incites me toward the discovery of an unexpected flexibility in the language with which I am dealing.
A first step toward an aesthetic definition of ambiguity might be represented by the postulate according to which in aesthetic texts an ambiguity on the expression plane must involve a corresponding ambiguity on the content plane. /Paulum Petrus amat/ or /A mat Paulum Petrus/ undoubtedly deviate from certain expressive norms but do not affect the conveyed content, which remains unchanged. /Colorless green ideas sleep furiously/ is more akin to an aesthetic achievement because the shock received by the breaking of certain rules forces the hearer to reconsider the entire organization of the content.
A characteristic of aesthetic texts singled out by the Russian formalists is the ‘priëm ostrannenja’ (Šklovskij, 1917), the so-called “device of making it strange” (Erlich, 1954): in order to describe something which the addressee may have seen and recognized many times, the author unexpectedly uses words (or any other kind of sign) in a different way. One’s first reaction is a sense of bewilderment, of being almost unable to recognize the object. Somehow the change in expressive device also changes the content. Thus art “increases the difficulty and the duration of perception” and describes the object “as if one were seeing it for the first time” so that “the aim of the image is not to bring closer to our understanding the meaning it conveys but to create a particular perception of the object”. This explains the poetic use of archaisms, the difficulty and obscurity of artistic creations when presented for the first time to an audience as yet unprepared for them; or those rhythmic violations which art brings into play at the very moment when one expects obedience to the customary ‘golden rules’: “in art there is ‘order’ and yet there is not a single column of a Greek temple that follows this order exactly, and aesthetic rhythm consists of a prosaic rhythm that has been violated . . .. it is a question not of complex rhythm but of violation of that rhythm and of a violation such that it cannot be predicted; if violation becomes the rule, it loses the force that it had as an operational obstacle”. Thus Šklovskij (1917) anticipates by some thirty years the analogous conclusions of so-called ‘informational aesthetics’ (Moles, 1958;Bense, 1965; Meyer, 1967; Zareckij, 1963).
A violation of norms on both the expression and the content plane obliges one to reconsider their correlation, which can no longer be the same as that foreseen by the usual code. In this way the text becomes self-focusing: it directs the attention of the addressee primarily to its own shape. There are self-focusing messages in which the ambiguity ratio is at its most elementary level and yet these messages are more akin to an aesthetic achievement than are merely ambiguous ones. Jakobson’s study of a political slogan such as /I like Ike/ (1960) has become famous (39).
Ambiguity and a self-focusing quality are by no means entirely concentrated upon the planes of expression and content as considered up to now. In the aesthetic text both the labor of the sender and the attention of the addressee are focused on the lower levels of the expression plane.
So let us consider, by examining a few examples of aesthetic experience, some of the qualities of an aesthetic sign-vehicle. These qualities are obviously already familiar to any aesthetic enquiry, but they deserve a semiotic explanation. That someone likes Ike (meaning) is immediately understood whether the sender says /I like Ike/ or /It is Ike that I like/ (sign-vehicles); and in both cases certain formal rules of the expression system are observed. But these rules do not necessarily emerge in everyday linguistic usage; language is used, frequently without a complete consciousness of the underlying competence, in order to communicate contents. Nevertheless, as Jakobson has demonstrated, in the case of /I like Ike/ the addressee’s attention is focused on the phonic matter of the message. There is something in this sentence which goes beyond the usual correlation between expression and content; something that, as it were, falls very easily upon the tongue. That something seems to be so ungraspable that the first aesthetic reaction consists in asserting that art, above and beyond its own ‘linguistic’ form, also conveys a ‘je ne sais pas quoi’. In this way aesthetics becomes the philosophy of the unspeakable.
Suppose we are looking at an Italian Renaissance palace with an ashlar-work facade. If the palace were drawn or photographed one could understand the architect’s ‘idea’, the so-called, ‘form’ of the artifact and other drawings could provide one with the plan and the entire set of geometrical rules which directed the builder. But when the palace is directly viewed something else happens.
Not only does it take time to comprehend, imposing a shifting angle of vision, and thus introducing time as one of the indispensable components of the architectural experience. But also the material itself, with its unevenness and its tactile stimulation, adds something to our conceptual understanding. The architectural system has given rise to a certain number of units, i.e. stones of a certain size; each is clearly a pertinent element of a segmented masonry-continuum. But what about the inner texture of the individual stones, since undoubtedly much of the appeal of ashlar work derives from this factor? Modern aesthetics would say that aesthetic enjoyment brings into play even the microstructures of the material from which is made (40). Which is true enough, except that a semiotic definition of these microstructures must go on to say that they represent the pertinent elements of a further segmentation of the material in question, thus suggesting the possibility of a more basic form of the expression. Aesthetics is not only concerned with hypersystems such as the various connotations that the work of art conveys above and beyond its immediate communicative appearance; it is also concerned with a whole series of hypostructures.
Let us return to the comparison between the semiotic levels of a code and those of a sign-function, as outlined in 2.2 (Table 6). The theory of codes considers an expression level in which, on the basis of an as yet unshaped continuum, a syntactic system gives rise to a structured set of signal-units; the code assigns the units of this plane to units of a content plane in which an as yet unshaped continuum has been structured into a set of cultural units by a semantic system.
When on the contrary one considers a sign-function in itself one has to take into account a sign-vehicle conveying a given meaning. The sign-vehicle is realized by moulding a particular channel; in other words, the stuff of which the sign-vehicle is made is the continuum from which the expression form has cut out its expression units; these units, if not inserted into a sign-function, are mere signals. A signal is a material fact and can consequently be studied and qualified by information theory. The signal is the token aspect of a unit of the expression-substance. Let us call this physical aspect of the signal the matter of the sign-vehicle.
The examples of aesthetic enjoyment examined above prove that in the aesthetic sign-vehicle matter plays an important part, and does so because it has been rendered semiotically interesting. In other words not only can the sign-vehicle (as an expression-unit) be detected as a pertinent element of the expressions system; even the material consistency of the sign-vehicle becomes a field for further segmentation. Using the everyday rules of a language I can utter a word in many ways, changing the pronunciation, stressing certain syllables differently, or altering intonation patterns; yet the word remains the same. But in aesthetic discourse every free variation introduced in ‘uttering’ the sign-vehicle has a ‘formal’ value. This means that even those features that usually pertain to the continuum and that a semiotic approach does not need to consider (instead leaving them to some physical or physiological discipline) here become semiotically relevant. In the aesthetic text the matter of the sign-vehicle becomes an aspect of the expression-form.
A red flag on a highway or at a political meeting can be based on various differently manipulated matters in order to be grasped as an expression: but the quality of cloth and the shade of red are in no way relevant. What is important is that the addressee detects /red flag/. Yet a red flag inserted in a pictorial work of art depends, among other things, upon its chromatic quality, in order to be appreciated (and to convey its signification).
In order to produce the conventional sign /cross/ one need only to cross two sticks. In order to produce a cross for the treasury of a medieval king, it was necessary to use gold and precious stones; each gem contributed to the aesthetic effect of the work because of its size, its weight, its transparency, its brilliancy, etc. Gold and jewels were appreciated because they could be manipulated and insofar as they actually were manipulated in a certain way. But even when viewed as an occurrence of a specific material, they were worthy of particular attention: that material was already charged with cultural signification. So that, wrought with equal care, a cross out of iron and glass would not have had the same aesthetic relevance; gold and jewels were significant stuff before the craftsman began to work on them.
In practice there is an empirical limit beyond which this material consistency, even though segmented to its utmost, can only be viewed both by the artist and the addressee as a cluster of unpredictable hypoforms. Beyond this limit there may still be perceptive and emotional effects but there are no more significations. Once it has moved beyond this threshold the work of art seems to stimulate reactions but not to communicate contents. Which might seem to confirm the opinion of those who assert that in art there is something more than ‘language’, a sort of irreducible ‘aesthetic information’ radically different from ‘semantic information’ (Moles, 1958; or Brandi’s distinction between ‘semiosis’ and ‘astanza’ or ‘presence’, 1968) (41).
But if these microstructures are not considered formally, then it is easy enough to assert that in aesthetic experience there exists a ‘je ne sais pas quoi’ that escapes ‘rational’ consideration. I can recognize a phrase, an image, a melody but something remains that I cannot grasp by means of the commonly accepted semiotic categories. Thus the impression of ‘unspeakability’. Fortunately a lot of pre-and para-semiotic disciplines are nowadays able to tell us something more precise about these phenomena.
For instance during the last century many techniques for measuring microstructures have been developed. From Birkhoff’s formula to the various proposals put forward by Bense, and certain techniques arising from the application of information theory (Moles and others), the distribution and order of textural items is becoming increasingly open to quantitative measurement: electronic computers, along with their scanning and plotting devices, are able to analyze lines, points and spatial intervals in their structural interrelationships, while sound-recordings and oscillographs are revealing the spectral formants of sounds, there defining structures where frequencies, durations and stresses were previously conceived of as the ultimate terms of musical science. Tonal nuances, intensity of colors, consistency and rarefaction of materials, tactile sensations, synaesthetic associations, all so-called ‘emotive’ performances, such as supra-segmental features and ‘musical’ gestures, vocal inflexions, portamenti and vibrato in singing, plus many other features that until a few decades ago were considered as uncoded, are now being investigated by semiotic disciplines that deal with the so-called lower levels of communication (Stankiewicz, 1964). Supra-segmental features and free variants that linguistics was not able to recognize as proper objects for study are now being tackled by paralinguistics – as well as, in other semiotic codes, whistling and drum languages or gestures, mimicry, facial expressions.
Recent Soviet studies on the levels in poetry (Toporov, 1965; Kolmogorov, 1962) remind us that, as Hjelmslev (1928) said, it is dangerous to establish a theoretical distinction between grammatical and extra-grammatical elements, or between the intellectual and the emotional use of a language. Both so-called extra-grammatical elements and emotional effects obey rules which have not yet been isolated. Trubeckoj (1939:IV.4.) described as ‘emphatic’ and therefore as ‘expressive’ (although conventional) certain phonological features that have since been coded into oppositional systems, or at any rate into gradated sequences.
It is no chance that, having started out from the material consistency of the aesthetic sign-vehicle, we have arrived at contemporary disciplines which do not study aesthetic phenomena at all (or at least not in particular) but are rather branches of a theory of codes. The reason for this is that there is a strong relation between the further segmentation of the token matter of a given sign-vehicle and the further segmentation of the expression plane of an entire semiotic system. In other words, the aesthetic experience, by revealing that within its basic matter there is a further space in which sub-forms and sub-systems can be isolated, suggests that the codes on which the aesthetic sign relies can likewise be systematically submitted to such further segmentation. The pertinentization of the token matter of the token sign-vehicle demands the pertinentization of those aspects of the expression-continuum that have up to now been considered as ‘hyposemiotic stuff’.
The aesthetic experience thus advocates the ‘semiotic civil rights’ of the ‘segregated’ continuum. A work of art performs a semiotic redemption of its basic matter (thus succeeding in a task that the Plotinian God never managed to accomplish, in spite of his emanational power).
After having experienced the pertinentization of matter achieved by the aesthetic sign-vehicle, one is forced to reconsider the expression system as a whole, in order to see whether it, or any of the sign productions permitted by it, can be subjected to the same mise en forme. Thus the diagram a la Hjelmslev outlined in 2.2.3. must be rewritten as in Table 46.
Hjelmslev (1943:52) asserted that “purport remains, each time, substance for a new form”, only qualifying his statement with the remark that this further segmentation of the continuum is a matter for other approaches than the linguistic one. We have now seen that this further segmentation is neither extraneous to the linguistic approach nor to the various semiotic ones.
Table 46
As long as semiotics continues to develop, the continuum will be further segmented and therefore better understood : the aesthetic experience provides a special opportunity for increasing this understanding.
As the first of its results, this further ‘culturalization’ of matter produces a further conventionalization of sign production (which is in some ways another sort of overcoding). One immediate consequence for aesthetics and art criticism is that this kind of new knowledge removes many phenomena from the realm of individual ‘creativity’ and ‘inspiration’ and restores them to that of social convention (42). But a study of this kind also becomes indispensable for the reverse process; since it is only when all that can be coded has been coded that actual innovation and real insight into the expressive possibilities of a given communicative medium can occur.
The study of all the systems that enrich the expression-continuum of each code (which may already be known, but are never exhaustively exploited as far as the flexibility of the expression plane is concerned) represents one of the main tasks undertaken by the discipline that the Prague School called poetics (43)
Thus an increased degree of organization within an expression-continuum will inevitably involve a parallel increase within the content-continuum. Looking at a work of art, the addressee is in fact forced to question the text under the pressure of a twofold impression: on one hand he ‘guesses’ that there is a surplus of expression that he cannot completely analyze (though maybe he could). On the other hand he vaguely senses a surplus of content. This second feeling is clearly aroused by the surplus of expression but it occurs even when this surplus of expression is not consciously grasped.
Suppose we are reading the well known verse by Gertrude Stein: A rose is a rose is a rose is a rose. At first glance nothing would seem to be more ‘normal’ than this sentence. From the point of view of the expression all the elementary rules of the English code are fully respected. As for the content, it seems to offer the most elementary kind of information, the tautology for truism. In order to convey a tautological content the expression seems to rely upon an excess of redundancy (the same redundancy being realized on the content plane; tautology is in fact mere semantic redundancy).
Nevertheless the message gives the impression of saying something that is semantically rich and therefore highly ambiguous. The feeling of ambiguity is suggested, first of all, by the excess of expressive redundancy, which violates a stylistical norm. Rather as, when white is perceived, the physicist recognizes an excessive simultaneous overlapping of colors, so this stubborn repetition of a banal statement makes one suspect that each time the same expressive items return they mean something different. Neither botany nor logic has ever accustomed one to accept as normal such an uninformative statement, which constitutes a sort of deviation from definitional norms. These two excesses of redundancy (on expression and content planes respectively) produce an increase of informational possibilities: the message has in effect become a source of further and unpredictable information, so that it is now semantically ambiguous.
From this point on, the addressee is entitled to suppose that /rose/, in every one of its occurrences, might be connected with different connotative subcodes, e.g. the allegorical, the iconological, the iconic. The work is thus ‘open’ to multiple interpretations (44). The contextual interaction brings to life more and more meanings and, as soon as they come to light, they seem fraught with yet other possible semantic choices. It is indeed difficult to avoid the conclusion that a work of art communicates too much and therefore does not communicate at all, simply existing as a magic spell that is radically impermeable to all semiotic approach.
However, this ‘magic spell’ is not as radically impermeable as might seem to be the case. First of all, it is open to a semiotic commutation test; if one changes one contextual element, all the others lose their primitive function and are usually unable to acquire another; they remain unbalanced, as on a chessboard where a bishop has been replaced by a third castle. If there is such contextual solidarity, then there must be asystematic rule.
This means that a work of art has the same structural characteristics as does alangue. So that it cannot be a mere ‘presence’; there must be an underlying system of mutual correlations, and thus a semiotic design which cunningly gives the impression of non-semiosis.
The aesthetic text is like a multiple match played by different teams at a time, each of whom follows (or breaks) the rules of their own game. Is it possible that in such a situation the way in which the baseball players deviate from their norm has something to do with the way in which the soccer players deviate from their own? This is rather the impression given by a work of art, so that a foul committed by a baseball player reveals itself not only as a witty solution that the rules of baseball must henceforth admit, but also as a device that should put into a different strategical perspective the ‘hands!’ committed by a distant soccer player.
Thus art seems to be a way of interconnecting messages in order to produce a text in which: (a) many messages, on different levels and planes of the discourse, are ambiguously organized; (b) these ambiguities are not realized at random but follow a precise design; (c) both the normal and the ambiguous devices within a given message exert a contextual pressure on both the normal and ambiguous devices within all the others; (d) the way in which the norms of a given system are offended by one message is the same as that in which the norms of other systems are offended by the various messages that they permit.
At every level (for every message) the solutions are articulated according to a homologous system of solutions, and every deviation springs from a general deviational matrix. Therefore, in a work of art a super-system of homologous structural relationships is established rather as if all levels were definable on the basis of a single structural model which determined all of them. But every system ruled by that deviational matrix is not only homologous to the others so ruled. Where this is the case, a work of art might be an admirable complex of interconnecting structures, but it would not necessarily have any particular semiotic status. However, on the basis of this structural arrangement of mutual homologies, the work of art seems to acquire a new status as a super sign-function.
Insofar as the aesthetic text has a self-focusing quality, so that its structural arrangement becomes one of the contents that it conveys (and maybe even the most important one), the way in which the rules are rearranged on one level will represent the way in which they are rearranged on another. Furthermore, it is the ambiguous arrangement of one level that provokes a reassessment on another: in /a rose is a rose is a rose is a rose/ the puzzling redundancy of the lexical level stands for a semantic complication on the definitional one. Thus the deviational matrix not only represents a structural rearrangement: it entails a rearrangement of the codes themselves. It thus represents the proposal of a new coding possibility.
This new code is apparently spoken by only one speaker, and understood by a very restricted audience; it is a semiotic enclave which society cannot recognize as a social rule acceptable by everyone. Such a type of private code is usually called an ‘idiolect’. The rule governing all deviations at work at every level of a work of art, the unique diagram which makes all deviations mutually functional, is the aesthetic idiolect. Insofar as it can be applied by the same author to many of his own works (although with slight variations), the idiolect becomes a general one governing the entire corpus of an author’s work, i.e. his personal style. Insofar as it is accepted by an artistic community and produces imitations, mannerisms, stylistic habits, etc., it becomes a movement-idiolect, or a period-idiolect, studied by criticism or the history of ideas as the main artistic feature of a given historical group of period. Insofar as it produces new norms accepted by an entire society, the artistic idiolect may act as a meta-semiotic judgment changing common codes(45).
The work-idiolect, the corpus- idiolect, the movement-idiolect and the period-idiolect form a hierarchy of increasingly abstract models each of which constitutes the individual performance of an underlying competence, granted that not only do competences allow performances but that performances also establish new forms of competence. The aesthetic idiolect produces over-coding rules; for example nowadays it is impossible to perform certain lexical cross-breedings without recalling Joyce’s pun-technique. If someone, whether consciously or unconsciously, follows the rule for making mots-valise a la Joyce he is not speaking ‘ungrammatical’ English so much as ‘Finneganian’. What he is really saying is far less important than the underlying statement: “lam joycing”.
To detect an aesthetic idiolect is no easy matter, and can in fact only be accomplished when the idiolect in question is highly standardized. The more the work submits to ‘commercial’ influence, the more connected to previous idiolectal experiences its underlying idiolect will seem; the more the work is immediately recognizable as ‘true art’ (i.e. Kitsch art, Midcult, philistine Beauty, art pompier, pocket -musee imaginaire), the more the idiolectal model will recur unvaried and clearly recognizable on all levels.
But even when the critic has isolated the idiolect of a work, this does not mean that he is in possession of a formula that could engender similar works. If considered as a work-idiolect the formula could only permit the production of another work that was absolutely identical to the first. If considered as a corpus or period-idiolect, the structural model is no more than a general schema to be embodied in a new substance. The difference between that schema and a given work is the same as that between a code and its possible messages. But another reason for the ^reproducibility of works of art is that however carefully the idiolect is isolated, it will never take into full account the form of the work’s lower levels. As a matter of fact although each further act of criticism will bring to light more precise idiolectal definitions of the work, the greater the stature of the latter, the more the critical process will constitute continuous and unfinished approximations. In other cases the formula can produce satisfactory new works; but here the more exactly the imitator understands the idiolect, the more eager he will be to emphasize the model that he has isolated; thus pastiches are more acceptable than mere imitations. When successful, a pastiche (such as Proust’s re-making of Balzac or Flaubert) represents a witty piece of criticism; it clarifies the characteristics of its models, ironically stressing some of their nodal or peripheral devices. The ruling presence of the aesthetic idiolect can be either detected by a critical analysis or confusedly ‘felt’ by an intelligent though non-technical reading.
The addressee ‘senses’ the surpluses of both expression and content, along with their correlating rule. This rule must exist, but to recognize it requires a complex process of abduction: hypotheses, confrontations, rejected and accepted correlations, judgments of appurtenance and extraneity. This process produces three of the results mentioned in 3.7.1.: existing codes are focused and submitted to change or partial revision; the relation between accepted content-systems and states of the world is frequently challenged; and a new type of ‘conversational’ interaction is established between the sender and his addressee.
Let us examine these three results in the following paragraphs.
The semiotic notion of an aesthetic idiolect explains the vague impression of ‘cosmicity’ that the addressee feels when contemplating a work of art. Insofar as every one of its levels is semiosically interconnected, the aesthetic text continuously transforms its denotations into new connotations; none of its items stop at their first interpretant, contents are never received for their own sake but rather as the sign-vehicle for something else. If the idiolect were rendered metalinguistically explicit, the reading of the work would be nothing more than a correct decoding.
Peirce recognized that a moment of hypothetical tension arouses a feeling similar to that engendered by a piece of music. One can thus understand why and how the interpretive effort demanded by a work of art releases this kind of strong and complex feeling that aestheticians have named in various ways (pleasure, enjoyment, excitement, fulfillment and so on), always believing that it was a form of ‘intuition’. There is some degree of philosophical laziness in merely labelling as ‘intuition’ every experience that demands an excessively subtle analysis in order to be described. But common artistic experience also teaches us that art not only elicits feelings but also produces further knowledge. The moment that the game of intertwined interpretations gets under way, the text compels one to reconsider the usual codes and their possibilities. Every text threatens the codes but at the same time gives them strength; it reveals unsuspected possibilities in them, and thus changes the attitude of the user toward them.
Through the close dialectical interrelationship maintained between message and code, whereby each nourishes the other, the addressee becomes aware of new semiosic possibilities and is thereby compelled to rethink the whole language, the entire inheritance of what has been said, can be said, and could or should be said. By increasing one’s knowledge of codes, the aesthetic message changes one’s view of their history and thereby trains semiosis. While doing this, the aesthetic experience challenges the accepted organization of the content and suggests that the semantic system could be differently ordered, had the existing organization been sufficiently frequently and persuasively challenged by some aspect of the text.
But to change semantic systems means to change the way in which culture ‘sees’ the world. Thus a text of the aesthetic type which was so frequently supposed to be absolutely extraneous to any truth conditions (and to exist at a level on which disbelief is totally ‘suspended’) arouses the suspicion that the correspondence between the present organization of the content and ‘actual’ states of the world is neither the best nor the ultimate. The world could be defined and organized (and therefore perceived and known) through other semantic (that is: conceptual) models.
This epistemological principle might seem to be mere metaphorical license. It is certainly a common enough experience to ‘feel’ (while reading a poem, watching a play, looking at a painting, etc.) that maybe ‘things’ are not quite as they usually seem. However, to simply suggest that a work of art ‘tells the Truth’ would be of little semiotic value and would not greatly differ from certain poetic statements (like “Beauty is truth, truth beauty”), which, when assumed as philosophical principles, are really astoundingly silly.
So that, in order to be verified, this semiotic principle requires a thoroughgoing analysis of the semantic shiftings engendered by an aesthetic text (see for instance, Eco, 1973 ƒ: although not aiming at an exhaustive demonstration, that essay at least tries to furnish general directions for a further and more concrete examination of the problem). If aesthetic texts can modify our concrete approach to states of the world then they are of great importance to that branch of a theory of sign production that is concerned with the labor of connecting signs with the states of the world.
Finally, aesthetic texts possess one quality that makes them a peculiar example of sign production labor intended to establish pragmatic relations between communicators, through a complex network of presuppositional acts. Inasmuch as the idiolect constitutes a sort of final (though never completely achieved) definition of the work, to read an artistic product means at once: (i) to induce, that is to infer a general rule from individual cases; (ii) to abduce, that is to test both old and new codes by way of a hypothesis; (iii) to deduce, that is to check whether what has been grasped on one level can determine artistic events on another, and so on. Thus all the modes of inference are at work. Like a large labyrinthine garden, a work of art permits one to take many different routes, whose number is increased by the criss-cross of its paths.
First of all the comprehension of an aesthetic text is based on a dialectic between acceptance and repudiation of the sender’s codes – on the one hand – and introduction and rejection of personal codes on the other. If the more usual form of abduction consists in proposing tentative codes in order to disambiguate an uncoded situation, then aesthetic abduction consists in proposing certain tentative codes in order to make the author’s message understandable. The addressee does not know what the sender’s rule was; he tries to extrapolate it from the disconnected data of his aesthetic experience. He may believe that he is correctly interpreting what the author meant, or he may decide to test new interpretive possibilities upon the text the author has set out before him. But in so doing, he never wants to completely betray the author’s intentions. So that in the interpretive reading a dialectic between fidelity and inventive freedom is established. On the one hand the addressee seeks to draw excitement from the ambiguity of the message and to fill out an ambiguous text with suitable codes; on the other, he is induced by contextual relationships to see the message exactly as it was intended, in an act of fidelity to the author and to the historical environment in which the message was emitted.
In this dialectic between fidelity and initiative two kinds of knowledge are generated: (a) a combinational knowledge about the entire range of possibilities available within the given codes; (b) a historical knowledge about the circumstances and the codes (indeed all the norms) of a given artistic period. Thus the semiotic definition of the work of art explains why, (i) in the course of aesthetic communication an experience takes place which can neither be reduced to a definite formula nor foreseen in all of its possible outcomes; (ii) yet at the same time this ‘open’ experience is made possible by something which should have (and indeed has) a structure at all levels. Thus the semiotic definition of an aesthetic text gives the structured model for an unstructured process of communicative interplay.
A responsible collaboration is demanded of the addressee. He must intervene to fill up semantic gaps, to reduce or to further complicate the multiple readings proposed, to choose his own preferred paths of interpretation, to consider several of them at once (even if they are mutually incompatible), to re-read the same text many times, each time testing out different and contradictory presuppositions.
Thus the aesthetic text becomes a multiple source of unpredictable ‘speech acts’ whose real author remains undetermined, sometimes being the sender of the message, at others the addressee who collaborates in its development (46).
A theory of sign production must take into account the labor performed in order to overcode and to switch codes. As was said in 3.1.1., this activity is commonly registered under the heading of rhetoric. In this section I shall try to show: (i) in what sense traditional rhetorical categories can be inserted into a semiotic framework; (ii) in what sense some of the problems connected with overcoding and code-switching go beyond the usual rhetorical framework and ask for either a new and semiotically oriented rhetoric or for new and autonomous branches of semiotics; (iii) in what sense many of the discussions about Ideology’ and ‘ideological discourse’ come within the scope of a semiotically oriented rhetoric and how the entire problem of ideology can be studied from a semiotic point of view (Genette, 1966; Todorov, 1967; Groupe µ, 1970; Barthes, 1970). In order to achieve these aims, let us try to summarize and schematize the objects of classical rhetoric, adding certain items that ancient rhetoric did not consider, but that modern rhetoric, whether semiotically oriented or not, either does or should.
Table 47
Classical rhetoricians viewed their discipline as the art of persuasion. Persuasion was not necessarily an underhand device but rather a socially oriented form of reasoning that did not deal with ‘first principles’ (such as those of formal logic, i.e. identity, non-contradiction and the excluded middle principle) and could not therefore use apodictic syllogisms. Thus rhetoric, like dialectics, was only dealing with probable premises (and therefore ones that were open to discussion); while dialectics aimed to derive an acceptable conclusion from these premises on reasonably logical grounds, rhetoric overtly dealt with enthymemes, i.e. syllogisms that also moved from probable premises, but to emotionally and pragmatically influence the listener. In recent times the so-called ‘new rhetoric’ (Perelman, 1958) has definitely reduced apodictic discourses to axiomatical systems alone, and has listed all other types of discourse (from philosophy to politics or theology) under the rhetorical heading. Thus almost all human reasoning about facts, decisions, opinions, beliefs and values is no longer considered to be based on the authority of Absolute Reason but instead intertwined with emotional elements, historical evaluations, and pragmatic motivations. In this sense the new rhetoric considers the persuasive discourse not as a subtle fraudulent procedure but as a technique of ‘reasonable’ human interaction, controlled by doubt and explicitly subject to many extra-logical conditions.
If rhetoric is considered in this way, it represents one of the more complex manifestations of sign production, involving the choice of given probable premises, the disposition of rhetorical syllogisms (or other forms of many-valued logic) and the necessary ‘clothing’ of expressions with rhetorical figures. This activity has its own rules (Perelman has listed many of them) and in the last analysis it constitutes the object of a semiotics of conversational interaction. The main requirement of this activity is that the rules be respected; and one of the most important of these rules is an explicit recognition of the one-sidedness of the premises and an acceptance of the principle that, under different circumstances, the issues might also differ.
But there is also ‘aberrant’ performance of the same type of sign production which results in an ‘ideological’ discourse (and under that heading I would list all forms of fraudulent propaganda and mass persuasion, as well as many so-called ‘philosophical’ statements); I mean by ideological discourse a mode of argument that, while using probable premises and considering only a partial section of a given semantic field, pretends to develop a ‘true’ argument, thus covering up the contradictory nature of the Global Semantic System and presenting its own point of view as the only possible conclusion (whether this attitude is deliberately and cynically adopted by a sender in order to deceive a naive addressee, or whether the sender is simply the victim of his own one-sidedness). The problem of the ideological discourse, which may help to throw a new light on the structure of the Global Semantic System, will be examined in 3.9. It too concerns a triple manipulation at all rhetorical levels (inventio, dispositio and elocutio).
In order to better understand these points something has to be said about what is commonly and restrictively believed to be ‘rhetoric’ in its entirety, that is, the various techniques of elocutio. In order to force the listener to pay attention to the premises and arguments one must stimulate his attention; it is here that rhetorical figures (or the various figures of thought, figures of speech and tropes) come in, these being the embellishments by means of which the discourse acquires an unusual and novel appearance, thus offering an unexpectedly high rate of information.
Unfortunately, in the last two centuries rhetoric has suffered from a rather bad reputation because of the two ways in which the notion of ‘figure of speech’ can be viewed. According to the ancient theorists a figure of speech was a schema of unexpectedness that provided the rules for replacing a word (along with corresponding concept), by means of other words and concepts. In this sense they are generative rules of overcoding: and this is how figures of speech will be considered in the following paragraph.
But rhetorical usage, throughout the centuries, has generated a lot of established rhetorical expressions. Starting as the theory of a particular type of manipulation of language, rhetoric has become, step by step, a store of pre-established instances of manipulation. Thus rhetoric frequently meant a repertoire of ready-made sentences offered as models of ‘good writing’ or ‘good speaking’. This repertoire included either pre-tested stylistic devices with an overcoded connotation of “artistry” (one outlet for these ready-made syntagms is Kitsch – see Eco, 1964), a mode of sign production which cajoles its audience by the use of formulas which have already been tried out and have acquired a certain prestige, or pre-established connotations with a fixed emotional value (figures such as /fatherland/, /free world/, the image of the mother and child connoting ‘pure’ feelings, etc.), and so on.
In this sense rhetoric is the result of a millenary overcoding that has in some cases produced catachreses, that is, figures of speech so strictly coded that the entity for which they stood has definitely lost its proper sign-vehicle, as in the case of the /table’s legs/.
These and other results of rhetorical overcoding cannot be the object of a theory of sign production and should rather be that of a theory of codes that deals with overcoded ready-made expressions. When used in sign production, they have a merely ornamental role and when employed to cover up a cruder content (as in the case of an expression like /peace with honor/ used instead of «peace but not immediately») they fall under the heading of elocutio as used in ideological discourses (see 3.9.).
But rhetorical figures are not merely ‘embellishments’. When originally and creatively used they do in fact change the way in which the content is taken into consideration. A semiotic explanation of rhetorical figures can be attempted by developing the theory of the sememe (as outlined in 2.11.) along with that of Model Q (see 2.12.).
In this paragraph I shall limit myself to a further consideration of two typical rhetorical figures, metaphor and metonymy. According to Jakobson (1956) these depend respectively on the axe of paradigm and syntagm;they thus represent two different procedures: one of substitution by similarity and the other of substitution by contiguity.
If one considers two sememes some of whose ‘readings’ have semantic markers in common, one can easily understand what is meant by ‘similarity’. Granted that both the sememe «dog» and the sememe «friar» possess the same connotative marker of «fidelity» (to their master) and «defense» (dogs defend their masters and friars defend the principles of the religion) it was easy during the twelfth century to invent for an order of mendicant friars (the Dominicans) the metaphor “dogs of God” (domini canes) (47). In this way the notion of ‘similarity’ no longer involves a suspected resemblance based on the thing itself (even though it often helps to make people believe so); a ‘similarity’ between semantic markers is simply a semic identity. On the other hand metonymy often seems to be a simple matter of overcoding; substitution by syntagmatic contiguity is based on the fact that, given a ready-made syntagm, established habits will permit one of its elements to be substituted for another. Thus given the accepted semiotic judgment /the President of the United States officially lives in the White House/ it is easy to use /the White House/ as a metonymy for /the President of the United States/.
However, a further consideration of such mechanisms shows that the fact of living in the White House is conventionally accepted as a semantic property of the cultural unit «President of the U.S.» – granted that a semantic system is more like an encyclopedia than a dictionary (see 2.11.3.).
Thus, in order that there be a conventionally accepted contiguity between two items of a ready-made syntagm it is necessary that such a syntagm be a semiotic statement. Since a semiotic statement attributes to a sememe some of its coded markers, metonymy also relies on the sememic spectrum of a given cultural item. Instead of being a case of semic identity it is a case of semic interdependence.
This semic interdependence can be of at least two types: (i) a marker standing for the sememe to which it belongs (/the sails of Columbus/ for «the ships of Columbus»); (ii) a sememe standing for one of its markers (/Harry is a regular fish/ for «Harry swims very well»).
However, the notion of semic interdependence does not take into account the difference between synecdoche and metonymy posited by classical tradition: the former being ‘a substitution within the framework of the conceptual content’, and the latter a substitution ‘with other aspects of reality with which a given thing is customarily connected’ (see for instance Lausberg, 1949).
It is true that such a distinction is based on a confusion between intensional and extensional approaches and does not take note of the nature of the sememe as encyclopedia. In this latter perspective one cannot accept that the relationship between ‘grapes and bunch’ is a synecdoche while the one between ‘grapes and Bacchus’ is a metonymy, since even the fact that grapes (and wine) are in some way connected to Bacchus should be registered by the semantic representation of «grapes».
But it cannot be denied that the dichotomy we have proposed is poorer than the traditional classification. Distinctions such as pars pro toto, totum pro parte, genus pro specie, species pro genere etc. (concerning synecdoche) and causa pro effecto, effectus pro causa, a possessore quod possidetur, inventas ab inventore, ab eo quodcontinet quod continetur, etc. (concerning metonymy) seem to be rather important from a semantic point of view (48).
One could object that the addressee usually ‘understands’ a metonymy or a synecdoche disregarding these distinctions and only grasping general relationships of interdependence. But it is also true that these distinctions directly concern the ‘good’ organization of the sememe and the problem (discussed in 2.11.1.) of semiotic entailment or “meaning inclusion”. As a matter of fact, if a sememe were a non-hierarchical aggregation of disconnected markers, one should say that the sememe «male» has the denotative marker «human» and the sememe «human» can have a connotative marker of «male». But the system of semiotic inclusions asks for a precise hierarchization; therefore every marker denotes (by semiotic entailment) the class in which it is included and connotes the members of which it is the class (see also Greimas’ opposition between ‘axes and semes’ in 2.9.5.). Thus a sememe denotes the genus of which it is a species by hyperonymy, and connotes the species of which it is a genus by hyponymy («scarlet» denotes «red» and «red» connotes «scarlet»). This explains all the rhetorical distinctions linked to the phenomenon of synecdoche. As for metonymy, a satisfactory solution can be reached by inserting within the semantic representation n-places predicates according to a typology of roles or ‘cases’ (cf. 2.11.1.). In this way one can record relations such as causa pro effecto and vice versa, a possessore quod possidetur, ab eo quod continet quod continetur etc.
Let us examine the Aeneid, 10,140:
vulnera dirigere et calamos armare veneno
where / vulnera dirigere/ means «to inflict a blow (in order to cause wounds)» and stands for/ dirigere tela/, /dirigere ictus/, /dirigere plagas/ or /vulnerare/. Let us suppose that it stands for /dirigere tela/ (with /dirigere ictus/ the result would be the same). A tentative representation of / telum/, excluding many other possible selections and referring to a standard Latin, appears as follows (where R is the Result of the action exercised):
Then/ vulnera dirigeref is a metonymy of the type (i) – marker for sememe – and represents a case of substitution of the instrumental cause by the effect. If the same expression stands for /vulnerare/ the rhetoric mechanism would not change, except that it would be a little more complex:
In fact «vulnus» instead of «vulnerare» would be a substitution of the efficient cause by the effect, but there is also a partial substitution of the connotation of «direction» for the directional act of wounding: a very risky synecdoche indeed (that can work only when supported by the ‘stronger’ metonymy).
Let us now suppose that, in order to indicate a friend of mine who is a bachelor, I say /that unlucky seal!/. Provided that my audience has read Katz, Fodor and Postal (and in the circle that I move in everybody has), the substitution is easily understood. The problem that arises here is whether this rhetorical figure is a metaphor or a metonymy. Since /bachelor/ as human male being and /bachelor/ as seal are both readings of the same sememe, one should speak of replacing a sememe with one of its markers and therefore of a type (i) metonymy. Nevertheless it is clear that the substitution was based on the ‘identity’ of the marker «unmated» (which is more general than «never-married» and which in any case springs from «never-married» because of the redundancy rules). At this point only two solutions are possible: either the two «bachelors» are simply different but synonymous sememes, or one must speak of metaphor even when dealing with a semic identity between two different readings of the same sememe. In any case, insofar as metaphor is no longer considered as a similarity between things, and metonymy is no longer considered as a contiguity between things, the categories of both semic identity and semic interdependence are a matter of infra- or inter-sememic connections. In this perspective even semic identity is permitted by the underlying texture of the semantic system, so that a sort of structural contiguity supports and governs all these rhetorical interactions. Metaphors and metonymies are made possible by the existence of a semantic global universe whose format is that of Model Q.(49).
At this point it becomes necessary to establish by what rule ‘good’ metaphors or metonymies are distinguished from ‘bad’ or ‘trivial’ ones. A tentative solution might be as follows: a good metaphor occurs when the ‘identical’ markers are comparatively peripheral and particularly characteristic of the two sememes in question. To call a group of warriors /men!/ is clearly an embryonic metonymy because all soldiers are men, but «man» is a marker shared by many other sememes and consequently this kind of substitution does not particularly characterize a warrior. But if one says /I have two thousand swords at my disposal/ in order to convey that one has two thousand warriors at one’s command, then we have a more successful metonymy since warriors are the only type of men to have swords.
A better example of successful metonymy is offered by the interdependence established by the Romans between «gladiator» and «ready to die» or «death-seekers» (Ave Caesar, morituri te salutant!). In this latter case not only does the metonymy seem more ‘inventive’ but it increases one’s awareness of the semantic entity «gladiator».
Suppose now that one substitutes «warrior» by «gladiator», and «gladiator» by « moriturus». Not only are warriors seen in a less customary light but they are also characterized by a peripheral marker that is shared by other sememes that might up to this point have been considered far removed from the one under examination. For instance it now becomes possible to associate a warrior metaphorically with a «scapegoat» (as a « moriturus» by definition), so that an army of warriors may be defined as /the scapegoats of the King’s ambitions/. Insofar as «scapegoat» has a marker of «innocence» the way to a more complex network of substitutions is open; the warriors can become /two thousand innocent swords/. And so on. At the extreme point of this substitutional shifting the way in which warriors are usually viewed has enormously changed; the connotations of «fierceness», «courage», «pride», and «victory» do not disappear, but merge with antonymous connotations such as «fear», «sorrow», «shame» and «defeat».
The rhetorical tracing of underlying connections in the semantic fields has revealed fertile contradictions. Since it has to take place between branches of the sememes, and since any node within these branches is the patriarch of a new sememe (see Model Q), rhetorical substitution, by establishing further connections, runs the whole gamut of the Global Semantic Field, revealing its ‘topological’ structure. In this activity contextual and circumstantial selections are frequently switched and overlapped, and short circuits of all sorts create sudden and unpredictable connections. When this process is rapid and unexpected and joins up very distant points, it appears as a ‘jump’ and the addressee, though confusedly sensing its legitimacy, does not detect the series of steps within the underlying semantic chain that join the apparently disconnected points together. As a result he believes that the rhetorical invention was the product of an intuitive perception, a sort of ‘illumination’, or a sudden revelation, whereas in fact the sender has simply caught a glimpse of the paths that the semantic organization entitled him to cross. What was for him a rapid but distinct look at the possibilities of the system becomes for the addressee something vague and indistinct. The latter attributes to the former a superior intuitive capacity, whereas the former knows that he had a more immediate and articulated view of the underlying structure of the semantic system. Both have, however, discovered a new way of connecting semantic units, so that the rhetorical process (which can, in some cases, equal the aesthetic one) thus proves itself to be a form of knowledge, or at least a way of upsetting acquired knowledge (50).
Suppose, reformulating Table 15 in 2.9.6. in order to get an ad hoc example, that there is an axis containing two semantic units (u1 and u2), that are usually considered mutually incompatible, because their first respective denotative markers are units derived from an oppositional axis (α2 vs. α1), but that, through α1, have a connotation γ1 in common (Table 48).
Let us now suppose that, through a series of rhetorical substitutions, a sememe can be named (and therefore rendered rhetorically equivalent: 
either (i) by one of its markers (a case of metonymical substitution, represented by mtn followed where necessary by the marker via which the connection is made), or (ii) by another sememe with which it shares a given marker (a case of metaphorical substitution represented by mtf, followed by the marker upon which the substitution relies), as shown in Table 49.
Table 48
Table 49
Provided that the rules not of formal logic but of rhetoric are in play, then ul (because of its equivalence to u2) acquires both markers α1 and α2, which were previously seen to be antonymically incompatible (Table 50).
Table 50
Sometimes the incompatibility thus challenged reveals itself in some form of ‘wit’ (a baroque device, as in oxymorons: /a strong weakness/). Sometimes the oppositional axis is really upset and the addressee must ask himself whether it needs to be reorganized. At other times the incompatibility remains unaccepted by the codes, despite which the rhetorical figure continues to operate, thus creating a feeling of unbalance and allowing logicians to assert that natural languages have no logic.
It is in fact frequently the case that sign production procedures in natural language are without logic but do involve rhetoric, this latter being nothing more than a fuzzy logic. When only words and elementary figures of thought are involved, these incompatibilities are commonly accepted as legitimate rhetorical games.
It would seem therefore that only elocutio is involved, while both the premises (the concern of inventio) and the arguments (the concern of dispositio) remain unchallenged.
But suppose, for instance, that the semantic tree of «bachelor» proposed in the KF Model could be rewritten as in Table 51, in accordance with the Revised Model:
Table 51
It is thus possible to metaphorically substitute the bachelor-seal for the bachelor-man because both possess the very specific marker «unmated» and to elaborate the rhetorical definition /that unlucky seal/ for a friend who has never married. Through a further metonymical series of substitutions it would then be possible to substitute «animal» for «human» and to call him a /poor beast/. Everybody would accept this series of substitutions as a pleasantly ironical joke without denying its legitimacy on the grounds of a two-valued logic or of an excessively rigorous semantic theory.
Nevertheless the joke may conceal a ‘poetic truth’ which is simply a form of inferred knowledge: what if being a bachelor really were a sad condition, in spite of the many peripheral connotations which suggest that bachelors are free, happy, ready for all sorts of erotic adventures? These suspicions entail a revision of many customary premises; a jocular figure of speech may thus release a discussion about whole areas of accepted values.
However, the discussion of accepted values becomes explicit when it assumes an enthymematic aspect. When enthymematic reasoning explicitly assumes that the premises from which it starts are probable (i.e. matters of opinion), the rules of the game are observed and straightforwardly persuasive intercourse results. But very little is needed to turn a straightforwardly persuasive argument into an ‘ideological’ one. The threshold between these two types of reasoning will be demonstrated by a concrete example.
In 1969, and for many years before, dietetic foods were largely advertised in the American market. Since sugar was supposed to produce fatness, and fatness was linked with several illnesses, including heart attacks, dietetic foods eliminated sugar and replaced it with cyclamates. In November 1969 a medical research program discovered that cyclamates could produce cancer. Thus all dietetic foods that advertised the presence of cyclamates among their main ingredients had to be removed from the market. Because this decision was causing an economic crisis among many industrial corporations, new packaged dietetic foods were sold advertising the absence of cyclamates, and further stressing their elimination by adding the label: “with sugar added”.
At first sight the solution might sound rather paradoxical, for it is clearly idiotic to advertise a dietetic food by stating that it contains sugar – a substance widely recognized as a fattening element. Nevertheless this new approach to advertising was accepted by the consumers.
In order to explain this phenomenon one might well assume that until November 1969 American society accepted some sort of implicit coding which established the series of mutually exclusive oppositions and connotatively established implications recorded in Table 52.
Table 52
A sort of hyperconnotation marked the second column as positive (+) and the first as negative (-). On the grounds of this coded series of correspondences and opposition, a series of semiotic statements could maintain that sugar produced fatness (and therefore heart attacks) while cyclamates produced slimness (and therefore guaranteed a longer life).
The factual statement associating cyclamates with cancer acted (through the authority of the scientists that pronounced it) as a metasemiotic statement which gave rise to new semiotic statements associating cyclamates with cancer and death. Thus within the space of a few days the social competence accepted a new series of correspondences and oppositions (see Table 53).
The fact that sugar was fattening fell into the background. Many newspaper interviews showed that people were of the firm belief that it was better to get fat than to get cancer. And it was easy to accept the idea that a more arduous slimming process achieved through eating lightly sugared products was preferable to a predictable neoplasm.
One must remember that the sememic spectrum of «sugar» did not change: it continued to be coded as fattening and therefore (by a normal redundancy rule connecting obesity with circulatory diseases) as somewhat dangerous to health. All that happened was that the sememic representation of «cyclamate», although not losing the marker «slimming», acquired that of «cancer». Thus in order to explain the reversal of the oppositional series, one must postulate the existence of a rhetorical premise which was in fact accepted by everybody and recorded in the interviews: “Better fat than dead”. Insofar as this premise was a typical rhetorical ‘endoxon’, a matter of common opinion, the rules of enthymematic reasoning permitted its use for persuasive discourse. By doing so, in November 1969 one would have been performing an acceptable persuasive argument.
Suppose that my doctor had said: “Well, clearly you’ll have to lose weight, but cyclamates are far too dangerous; you’d do better to give up those dietetic foods for a while. You know what I mean: better fat than dead”. The argument would have been acceptable, since the doctor was not claiming to demonstrate an absolute truth, but merely trying to persuade me about a choice between two sets of values.
What made the advertiser’s argument into a typical example of propaganda and ‘ideological’ discourse was the fact that the positive status acquired by sugar when compared to cyclamates (the axis placing them in opposition being roughly ‘ways of dying’) was applied to an argument concerning dietetic foods (the axis in question being: ‘ways of slimming’). Thus sugar appeared to have a positive effect on slimming, when in fact it had nothing of the sort. It thus was surreptitiously given a «slimming» marker to which it had never had any right, socially recognized or otherwise.
This ‘ideological’ operation was performed by code-switching and thereby displacing an emotional connotation. A recognized but forgotten (or concealed) premise (“better fat than dead”) charged sugar with a positive marker (but only according to the uncoded contextual selection «vs. cyclamates»). The ad-men then retained this marker as if it could be conventionally associated with the sememe «sugar» in every context.
This example demonstrates two things: (i) the way in which code-shifting is brought about; (ii) the fact that semantic sub-systems acquire a given status in accordance with a given contextual selection and that this status does not or ought not to remain the same when viewed from a different contextual point of view. The discussion about cyclamates was an example of ‘ideological’ code switching, because it pretended that the structure of a given semantic sub-system remained the same under any circumstances.
In any case the problem of ideological code-switching deserves a more thorough examination. In order to do so let us summarize all the preceding points and then establish a more abstract laboratory model.
In 2.14.1. when dealing with the interpretation of the expression /he follows Marx/ I said that it involved a degree of ideological connotation (is following Marx good or bad?) that determined the interpretation but did not depend on any previous coding. In this sense the ideological background on which the interpreter relied in order to disambiguate the sentence was reached through a complex inference, involving a series of presuppositions about the sender or the object of the sentence. Detection of the speaker’s world vision depends on a process of interpretation rather than on previous codes. Thus ideology would appear to be an extra-semiotic residue which is able to determine semiotic events, acting as a catalyst in many abductive processes, but which escapes cultural coding (see also Table 30).
But what has to be presupposed (since it is not assured by any previously established code) is that the sender subscribes to a given ideology, whereas the ideology itself, the object of the presupposition, is an organized world-vision which must be subjected to a semiotic analysis.
A semantic system or sub-system is one possible way of giving form to the world. As such it constitutes a partial interpretation of the world and can theoretically be revised every time new messages which semantically restructure the code introduce new positional values. A message which states that /Martians eat babies/ not only charges the sememe «Martians» with a connotation of «cannibalism» but carries a whole chain of connotations resulting from the global axiological attribution of «negativity». Clearly a series of messages which explain that Martians do eat babies, but babies of a different species, just as we eat ‘baby’ animals, could change the global axiological connotation. But such a revision of the code implies a series of meta-semiotic statements which question the connotative subcodes – this being the critical function of science.
But in general any addressee will turn to his own cultural inheritance, his own partial world vision, in order to choose the subcodes that he wishes to apply to the message. To define this partial world vision, this prospective segmentation of reality entails a Marxist notion of ideology as ‘false conscience’. Naturally, from the Marxist point of view this false conscience is born as a theoretical disguise (with pretensions to scientific objectivity) for concrete social relationships and given material conditions of life.
Ideology is therefore a message which starts with a factual description, and then tries to justify it theoretically, gradually being accepted by society through a process of overcoding. For a semiotics of codes there is no need to establish how the message comes into existence nor for what political or economic reasons; instead, it is concerned to establish in what sense this new coding can be called ‘ideological’.
Let us imagine a container divided into two parts, Alpha and Beta, by a partition in which there is a small hole. At both sides gas molecules move at different speeds. To guard the hole there is what the kinetic theory of gases calls ‘Maxwell’s demon’. The demon is an intelligent being who confounds the second principle of thermodynamics by allowing slower molecules to pass from Beta to Alpha, while only letting the faster ones through from Alpha to Beta. He thus causes an increase of gas temperature in Beta. We might also imagine that our demon (who is more intelligent than Maxwell’s) assigns the same speed to all fast molecules. Knowing both the number of molecules and the standard velocity, we should be able to ascertain both pressure and heat with the same unit of measure.
Let us image that the demon, for every n molecules passing into Beta, emits a signal: each signal-unit communicates only the number of molecules judged pertinent for our purpose (for instance a given calculation of the pressure and heat tolerable in a given situation). So that it is the purpose that determines the criteria of pertinence. If the demon – as an emitter – has a very simple code such as “yes vs. no” one needs no more than an electric signal to indicate the unit of measure. Repetitions of the signal indicate the sum of the units of measure. Let us suppose that /Z/ denotes «minimum» (heat and pressure) and /ZZZZ/ denotes «maximum».
If the receiver of the message is a machine, it registers «minimum» or «maximum» values and reacts according to instructions received. The signal, in this case, is not a ‘sign’, nor does the machine ‘understand’ its ‘meaning’. If on the contrary the receiver is a human being, his reaction transforms the signal into a sign that is the correlation between an expression and a content. But at the same time the human addressee will add certain connotative markers to the denotative ones.
For instance, the expression /ZZZZ/ when referring to the ‘calculus for heat’ connotes certain positive values that cannot be taken into account when the ‘calculus for pressure’ is considered. Moreover, if a given quantity of heat is required in order to make the room more comfortable, it will connote certain markers that obviously change when the heat is needed as a form of energy for producing something. The same goes for the expression /Z/ («minimum») so that one can establish a sememic representation for both expressions, according to different circumstantial selections (Table 54).
Both sememes, if they are to be established, require that culture subdivide the semantic space into a series of oppositional sub-systems of which only a limited number are taken into account by the various readings of the sememe.
Thus every circumstantial selection isolates oppositions that are sometimes semantically identical, sometimes not entirely homogeneous, and at others straightforwardly contradictory (Table 55).
Table 55
(1) |
(2) |
(3) |
PRESSURE |
HEATING |
PRODUCING |
min vs. max |
min vs. max |
min vs. max |
low vs. high |
low vs. high |
low vs. high |
lack vs. excess |
lack vs. plenty |
lack vs. plenty |
security vs. danger |
discomfort vs. comfort |
-energy vs. +energy |
good vs. bad |
bad vs. good |
bad vs. good |
If we represent the composition of a given sememe as the successive branching into different positions of diverse semantic axes (see 2.9.6.) then the representation of the sememe «maximum» should have at least two incompatible readings represented respectively by a continuous and a dotted line (Table 56).
I shall define as an ‘ideological’ inventio a series of semiotic statements based on a previous bias (either explicit or otherwise), i.e. the choice of a given circumstantial selection that attributes a certain property to a sememe, while concealing or ignoring other contradictory properties that are equally predicable to that sememe, granted the non-linear and contradictory format of its semantic space. Thus semiotic statements represented by either continuous or dotted lines will be viewed as ideological ones.
A non-ideological statement would be a meta-semiotic one that showed the contradictory nature of its semantic space. This kind of meta-semiotic statement is represented in Table 57.
Table 56
Table 57
I shall define as an ‘ideological’ dispositio an argument which, while explicitly choosing one possible circumstantial selection as its main premise, does not make clear that there exists a contradictory premise or an apparently complementary premise which leads to contradictory conclusions – thus concealing the contradictory nature of its semantic space.
I shall also define as an ‘ideological’ dispositio an argument which although undertaking the comparison of two different premises, chooses ones that do not possess mutually contradictory markers, thus consciously or unconsciously concealing those that could upset the ‘linearity’ of the argument.
Suppose that there is someone who believes (or wants to make people believe) that maximum heat in the Alpha-Beta system can give both optimal heating and an optimal state of productivity. This person can set out his argument so as to show that such aims are mutually compatible and that they jointly produce a “desirable” situation (which we might call «welfare»). The argument can thus arrange the two sub-systems on which both circumstantial selections rely in the symmetrical fashion outlined in Table 58.
Table 58
This enthymematic model shows that there is no contradiction between the pursuit of optimal heating and that of optimal productivity.
The oppositions and the successive connotations of the square ‘abed’ (representing the premise “high heat makes for good heating”) are so complementary to these of the square ‘abef’ (representing the premise “high heat makes for good production”) that if one considers the lateral triangles ‘ace’ and ‘bdf one will see that «discomfort» can be considered as a metonymy for «-energy» and «comfort» as a metonymy for «+energy». According to the rhetorical rules outlined in 3.8.3. these substitutions are in fact permitted by the sememic representation outlined in Table 54. It is in fact obvious enough that loss of energy can cause a less comfortable heating situation (while comfort can in turn be caused by a good supply of energy): the substitution of effect for cause or vice versa providing an excellent example of metonymy.
The above example of dispositio is ‘ideological’ because it does not take into account the potential contradiction between, on the one hand, «production and pressure» and on the other «heating and pressure».
Let us set out in Table 59 the symmetrical correspondences between these two sub-systems, in order to demonstrate the contradiction which clearly arises.
Table 59
It will be seen that the lateral triangles present pairs of antonymous markers: «excess vs. plenty», «security vs. discomfort», «danger vs. comfort» are not mutual interpretants, nor can they be mutually substituted (except in cases of ironical oxymoron). The second level of this prismatic structure shows the incompatibilities that arise when the two points of view are compared; the base diagrammatically demonstrates the whole network of mutual incompatibility, each link producing an opposition ‘good vs. bad’. The same happens when setting out ‘pressure vs. production’ diagrammatically.
Table 60 needs no further commentary.
Table 60
When the connections showed by Tables 58, 59 and 60 are disregarded, two ideological discourses may potentially occur, one asserting that heating and producing arc primary values to be pursued at any cost for the sake of general happiness, while concealing that they are not compatible with general security, since they produce danger; the other asserting that the most important thing is absolute security for all members of the social group in question, while concealing the fact that, if completely attained, this excludes any increase in productivity and welfare.
I am not trying to argue that the production of a good heating system is a ‘bad’ goal; nor indeed that aiming at maximum safety is a ‘bad’ idea. On the contrary, any honestly persuasive discourse about the aims of a social group must take into account all of these goals; but it must at the same time acknowledge on what grounds (i.e. according to which premises) the values are preferred and to what extent they are mutually exclusive.
In fact a critical survey of the values in question will show that they are mutually exclusive only if taken as absolute (i.e. logically formalized). Whereas in fact they are all fuzzy concepts. A critical survey of the semantic composition of these concepts would show that they were open to gradation; there is a series of intermediate states between «-energy» and «+energy» and likewise between «security» and «danger» (danger being a very low state of security and vice versa). It would thus be possible to isolate a middle portion of the energy scale which coincided with that of the security scale (provided that their ‘gradients’ were inversely proportional).
Table 61
But in making such a calculation one has already trespassed upon the realm of ‘ideology’ by performing a critically persuasive discourse. This discourse can obviously be rejected by any interlocutor who has established a radical scale of priorities, such as “better rich than safe” (or “better safe than rich”) (51). A critical semiotic survey of ideological discourses does not eliminate the speaker’s pragmatic and material motivations and therefore does not change the world (or the material bases of life). It can only contribute to making them more explicit (52).
The ideological discourse endeavors to conceal these various options, and must therefore involve a rhetorical labor of code shifting and overcoding. Thus anyone who accepts the harmonious and symmetrical correspondence between production and heating (Table 59) may well forget that the semantic unit «maximum» on which he bases his view represents not only a maximum of heat and energy but also a maximum of pressure. The only connotations connected to this unit are those of «plenty», «comfort» and «energy», which rapidly become names for it (53); consequently, when someone maintains that maximum heat also means «danger», the statement is received as semantically anomalous, and believed to be referentially false. It is thus ‘ideologically’ interpreted as a malignant effort to disrupt the ‘law and order’ which governs one’s uncontradicted semantic universe (i.e. one’s culture, world vision, religion, ‘way of life’, etc.).
To recall that /maximum heat/ is not only an expression suggesting «wealth» and «comfort», but also a sign originally produced in order to mention a state of the world, and to realize that, physically, this state of the world was and is a growth in pressure – all this would mean putting on its feet a ‘philosophy’ that was used to standing on its head. But ideology is a partial and disconnected world vision; by disregarding the multiple interconnections of the semantic universe, it also conceals the pragmatic reasons for which certain signs (with all their various interpretations) were produced. This oblivion produces a false conscience (54). Thus a theory of codes (which looks so independent from the actual world, naming its states through signs), demonstrates its heuristic and practical power, for it reveals, by showing the hidden interconnections of a given cultural system, the ways in which the labor of sign production can respect or betray the complexity of such a cultural network, thereby adapting it to (or separating it from) the human labor of transforming states of the world (55).
This transformation cannot be performed without organizing such states of the world into semantic systems. In order to be transformed, the states of the world must be named and structurally arranged. As soon as they are named, that system of sign systems which is called ‘culture’ (which also organizes the way in which the material forces are thought of and discussed) may assume a degree of extra-referential independence that a theory of codes must respect and analyze in all its autonomy. Only in that way is it possible to outline a theory of sign production that (even when approaching the relationship between signs and things in terms of Truth and Falsehood) will profit from a purely semiotic approach.
When the Alpha-Beta system gives rise to an ideological imbalance, and semantic sub-systems stand ‘on their heads’, there are only two possible ways of stopping the process:(i) to make the Alpha-Beta container explode, so that the existence of pressure comes to light and destroys the oblivion of false conscience; this act, sometimes called ‘revolution’, represents another semiotic threshold that our discipline must recognize and cannot afford to trespass upon; when the Alpha-Beta container explodes, the whole system of semiotic units will go up with it, and will need to be re-built (even though there may be no semioticians around to record the new phenomenon).
(ii) to demonstrate (through a survey of the contradictory format of the semantic universe, getting back toward its sources as far as is possible by moving along the branches of the content systems and across the various code shiftings and concretions of different sign-functions) how much broader than most ideologies have recognized is the format of the semantic universe.
Granted that it is daring, but by no means absurd, to maintain that issues (i) and (ii) are mutually compatible, the semiotician may not have much to say on this matter, but he will have something to do. The labor of sign production releases social forces and itself represents a social force. It can produce both ideologies and criticism of ideologies. Thus semiotics (in its double guise as a theory of codes and a theory of sign production) is also a form of social criticism, and therefore one among the many forms of social practice.
1. In fact any of the judgments listed in (viii) can be translated as a non-‘locutory’ act. For instance semiotic judgments can produce speech acts like “Are all bachelors males?”, “If only all bachelors were males!”, “I assure you that all bachelors are males!”; “Is this object a pencil?”, “Look, a pencil!”, “What a horrible pencil!”; etc. Factual judgments are susceptible to the same type of translations: “Is moon really walked on by human beings?”, “Is that pencil really black?”, “I define this pencil as black”, “Gosh! Human beings on the moon!”, “Eco-freaks, beware! Man has arrived on the moon!”. Even if logically speaking all these speech acts can be reduced to assertive sentences, and even if transformationally speaking their underlying phrase markers can be reduced to that of a declarative sentence (Katz, 1972:201 ff; sed contra, Lakoff, 1971 a), they pose a lot of semiotic problems. I shall not be particularly dealing with them in this book but I am inclined to view the disciplines investigating such problems from language analysis to sociolinguistics, from researches in interactional behavior to ethnomethodology as constituting an essential chapter of general semiotics (Austin, 1962; Searle, 1969; Cohen, 1973; Gumperz and Hymes, 1972; Cicourel, 1973).
2. When a factual judgment is emitted, the most natural attitude is to test it. Testing factual judgments is the first duty of the scientist, the historian, the newsman as well as the prudent man. It would be wrong to say that semiotics is not concerned with this activity of testing; except in the case of testing mentions, factual judgments do not send immediately back to an actual perceptum and demand for more mediating operations, each of them implying the recourse to a new level of semiotic conventions (for example an historian must check a factual judgment by controlling written records, archeological witnesses, and so on). This kind of labor, recorded in Table 3 under the head of ‘focusing on world’ should be further approached by a general semiotics. Up to now this labor has been studied by logical semantics (as well as by the methodologies of the various sciences). That until now semiotics and logic seem to have marched quite independent of each other (in spite of the methodological and philosophical chance offered by Peirce) is due to the fact that (except in the case of Morris) semiotics was more linked with linguistics and cultural anthropology. But this threshold between logic and semiotics becomes more and more imprecise and it comes to be widely trespassed by the latest researches in generative semantics and transformational grammar.
3. The way in which both factual and semiotic judgments can change the code (a way that will be further analyzed by the various sections of the present chapter) could solve a widely discussed epistemological problem, i.e. the possible relationship between a structural and a dialectic logic. If semiotic systems are structures and if the first property of a structural arrangement is the mutual solidarity of its elements (therefore implying an homeostatic permanence of the structural whole), how then can structures transform themselves into other structures (i.e., how then can codes change)? There are many positions held by different authors in some way concerned with a dialectical interpretation of structural approach; Seve (1967) maintains that structures are only transitory configurations of the ‘material process’ and that a structural logic is only the science of the “internodal segments of the dialectical contradiction”; thus structural logic is only an ‘analytical reason’ which, even though useful and necessary, does not grasp in its totality the dialectical process. Godelier (1966) maintains that there are two types of contradictions, the one within the structure and the one between the structures; the former practically corresponds to the self-contradictoriness of each code as outlined in 2.12-13.; the latter depends on the appearance of new material phenomena and could be equated with the necessity for factual judgments outlined in the present chapter (see for instance the example of cyclamates given in 3.8.5.). This double aspect of code changing is also considered by Lotman (1970) in his typology of culture, and he too accepts the idea of continuously intertwined action of both principles (changing from outside and changing from inside). The cybernetical and mathematical background of this question is dealt with by Apostel (1960) and especially by Piaget (1968). I think that the dialectic between semiotic judgments and factual judgments, along with the dialectic between codes and sign production, can constitute the basis for a dialectical theory of structural evolution in semiotics.
4. After having elaborated his criticism of Russell’s dichotomy between meaning and denotation (and after having reduced it to the more suitable complementarity of signifying and mentioning) Strawson advances a conclusion which perfectly fits its philosophical purposes but does not help us in developing a semiotic theory. He says: “Neither Aristotelian nor Russellian rules give the exact logic for any expression of ordinary language, for ordinary language has not exact logic.” The purpose of a theory of codes was to see if ordinary languages have, if not an exact one, at least one logic. Maybe the problem is not of finding a logic, if logic is only the theory of a formalized language. The problem is to find a semiotic theory which is surely different from a formal logic, but which is nevertheless able to dissolve the shade of skepticism suggested by Strawson’s quotation, which may lead to the suspicion that natural languages cannot have a theory, which has to be refused if a semiotics is to subsist.
5. Once this is asserted, one may easily admit that there is a lot of difference between the semiotic function performed by a photograph and the one performed by an actual object. But in the present context identities are more important than differences. Another objection might be that the above theory of mention does not hold as far as individuals are concerned. What does it mean to say /this man is John/? According to what has been said in 2.9.2. à propos of denotation of proper names, it means that the semantic properties commonly correlated by a group-code to the lexeme /John/ coincide with the semantic properties that the same group-code might correlate to a given perceptum. It means that one must associate to the token perceptum the same notion that presumably associates to the lexeme /John/: a man who is the brother of Mary, the assistant manager of the local bank, the one that I have frequently described as the best of my friends and so on.
6. A difference exists between two ways of intending the use of /is/ and of the pointer in mentioning. If – indicating a penguin – I say /TTiis is a cat/ and I mean «this object has the property of being a cat», then I pronounced a semiotic index-sensitive judgment which simply represents a wrong use of the code; from the point of view of a theory of mentioning it leads to a false statement. If on the contrary I mean «the name of this animal is ‘cat’», then I have pronounced an exageratedly arbitrary meta-semiotic judgment that can be erroneously accepted only by a naive anthropologist who has chosen me as an untrustworthy informant.
7. Since the above analysis is performed according to the Revised Model proposed in 2.11. and since the above page represents an effort to approach the classical problems of philosophical semantics from the point of view of a semiotically oriented semantic theory (a ‘meiging’ that has been foreboded in note 1 of this chapter) it would have been more challenging to test the power of this approach on the sentence /the present king of France is a bachelor/. May I propose from now on the use of this sentence as a symbolic guarantee of a possible merging of different approaches to the same set of problems?
8. Anyway, even though accepting the idea that Proust can translate Elstir, it would be hardly believable that Mondrian can translate Spinoza’s Etica more geometrico demonstrata . . ..
9. See in Goodman (1968:99 ff.) an interesting discussion on artistic fakes and on the ‘autographic’ and ‘allographic’ arts; the former are not susceptible of notation and do not support free performance, the latter can be translated into conventional notation and the resulting score can be performed through variations of a certain extent (like music). The difference between autographic and allographic is linked to the distinction between dense and discrete signals (see later, 3.4.6.).
10. Given a scale of replicability, as far as one steps down from its top (n/n of fidelity) to the inferior degrees, after a certain ratio, one seems to have trespassed on a threshold; one passes from the universe of replicas to one of similarities (see 3.5.3.). As a matter of fact such a scale is not an homologous one because the very notion of property changes beyond the threshold; in cases of replicas it is a matter of real ‘same’ properties, in cases of similarities it is a case of transformed and somewhat projected properties (see 3.6.7.).
11. Let us assume that in every case of ratio facilis there is the possibility not only of customary replicas but even of absolute duplication; it is possible to print a double of a previous word, to print a double of a king of spades and obviously to produce a double of a road signal (every replica, in this last case, being at the same time a double). The signs ruled by a ratio facilis can be translated by some other notation (see note 9). I can translate phonemes by means of Morse dashes and points, musical sounds from the temperate scale into written notes, and so on.
12. One should thus reformulate Morris’s position by saying that these formators are just features and not signs, in the same way as a phoneme is not a sign but a combinational feature whose presence or absence changes the nature of the word. But this assumption, although it tightens up the analysis, does not change the problem. There are certain expressive features that seem to be motivated by the toposensitive markers of their content. (That is, features which directly convey a given portion of the content expressed by the expression of which they are a syntactic component).
13. On the manifold senses of the word ‘analogy’ during the whole history of philosophical, theological and mathematical thought, see La linea e il circolo by Enzo Melandri (1968). However fascinating this concept may have been and may be, I prefer to retain here only its most concretely operational sense. Even because, when speaking of iconic signs, analogy is frequently used as a synonym of unspeakability, ineffability, native resemblance: that is, simply in order not to recognize iconic devices as signs and to avoid any semiotic enquiry on them.
14. See Gibson, 1966, p. 227: “Physical optics makes a distinction between ‘real’ and ‘Virtual’ images. In optics, what I have called a screen image (the picture made by projecting shadows on a surface, the structuring of an array by artificial variation of illuminations) is called a ‘real’ image, and so it is. What I call an optic array (the structured stimulus for an eye, chambered or compound) when it comes from a mirror or a lens is said to produce a ‘virtual’ image. The apparent face in the mirror and the apparently near thing in the field of a telescope, are objects in effect, not in fact, but they are not pictures or sculptures or screen images”.
15. One could say that mirrored image is used as a sign in some cases, such as when I see in the mirror a person coming behind me or when I use two mirrors in order to check the cut of my hair on the nape of my neck. But these are mere cases of artificial extension of my sight, not so different from the fact of using a magnifying glass in order to see what usually I cannot.
16. The problem of ill-counterfeit doubles remains open half way between a real double, a tentative replica and an iconic representation of the original. What is a bad imitation of a wooden cube? the Xerox of a drawing? a photomechanical reproduction of a painting, perfect in any particular, except the texture of the canvas substituted by a sheet of appropriate paper? Some of these phenomena could be listed among imprints (see 3.6.2.), some others seem to escape a precise definition. I think that, even though they are produced as doubles, they may become signs when chosen as such; in this case their semiotic nature depends on the context in which they will be inserted, and on the kind of explanation that accompanies them.
17. To prove the conventionality of the transcriptional systems, Gombrich also refers to two photographs of the same corner of Wivenhoe Park, which clearly indicate how little Constable’s park has in common with that of the photograph; but he does not then go on to suggest that the photograph constitutes the parameter for judging the iconicity of the painting. No square inch in the photograph is identical with a mirror image; the black-and-white photograph only reproduces gradations of tone between a very narrow range of greys. Not one of these tones corresponds to what we call “reality”. The scale depends largely on the photographer’s choice in the darkroom; given two photographs (reproduced by Gombrich), the one printed within a narrow scale of greys produces an effect of misty light while the other, where stronger contrasts were used, gives a different effect.
18. If every complex iconic representation is a code in itself, then there is not one iconic code, but many, maybe thousands and thousands of them. Take as an hypothesis that the known languages are not the only languages in the verbal universe, but that there are a million other possible, and actually existing, languages as well; this does not spare us the task of studying semiotically how these languages are structured. It is the same in the iconic universe, The assumption that there exist as many ‘languages’ as messages is then a drastic assumption. It can be translated as follows: 1) there exist as many iconic languages as personal styles of an author; 2) there exist as many iconic languages as there are styles and manners typical of a school or period; 3) if the various iconic works of art treat the existing codes ambiguously, as also happens with verbal works of art (which is the “langue” of Joyce?), the iconic styles reserved for non-aesthetic uses follow instead more predictable systems of rules; therefore there are recognizable iconic codes in mass communication, photography, comic strips and movies.
19. To the extent that Peirce established part of his program of a typology of signs (only 10 types on the programmed 66) every sign appears as a bundle of different categories of signs. There is not an iconic sign as such, but at most an Iconic Sinsign which at the same time is a Rheme and a Qualisign, or an Iconic Rhematic Legisign (2.254). Nevertheless the classification was still possible for, according to Peirce, the different trichotomies characterized the signs from different points of view and signs were not only precise grammatical units but also phrases, entire texts, books. Thus the partial success of the Peircian endeavor (along with his almost complete failure) tells us that if one wants to draw a typology of signs one must, first of all, renounce the straight identification of a sign with a ‘grammatical’ unit, therefore extending the definition of sign to every kind of sign-function.
20. All this once again requires that the concept of semantic component be freed from a verbo-centric fallacy. The semantic representation of a given expression also contains non-verbalizable markers, such as directions, spatial dispositions, relations of order and so on. Thus the content of the word /dog/ must have among its markers images of dogs, while the content of the image of a dog must have among its markers also the concept of «dog» (along with its intensional description) and the corresponding word in all possible languages. As has been said in 2.11.3. such a semantic encyclopedia is more a regulative hypothesis than a matter of individual knowledge; it is the virtual social semantic representation of a given sememe, which must be postulated in order to permit a theory of codes, able to explain every concrete act of communication.
21. However, one ought not to believe that this makes the difference between an analogical and a digital machine, for even an analogical one can produce tokens according to a ratio facilis (see TV scanning and broadcasting).
22. When Robinson Crusoe discovered the footprint of Friday on the seashore, the footprint conventionally denoted «man» but also connoted «barefooted». Being imprinted on the sand with a feature of direction, the context //footprint+position+direction// also was a text meaning// «a man has passed here».
23. Since Robinson believed himself to be the only human being on the island, a labor of inference and presupposition had thus to be performed by him so as to arrive at the conclusion: «I am not the only man here» or «there is another man on the island». This could have involved some meta-semiotic statements about the definition of the island.
24. When a trace is not previously coded, one is obliged to think that every point in the trace must have corresponded to a point on the imprinting referent. One could say that the imprint in this case is really an index, in Peirce’s sense. As a matter of fact, in this case the imprint is not a sign but an act of mention. An act of mention has to be verified. But to verify a mention (see 3.3.5.) means that – given the expression /this object is six meters long/ – one has to decide if the referent, taken as an ostensive sign (3.6.3.) can be called an object and if its dimensional properties can be defined as /six meters long/. In other words, in order to check a mention, one must already know the properties of the mentioned object. Suppose now that an explorer discovers the tracks of an unknown animal. One usually believes that it is enough to reconstruct by the traces the imprinting objects by a sort of backward projection. But in order to abduce the animal as the cause of the imprint, one must already possess some general content-schema. It is only by interpreting the traces as those of one or more known animals that one can extrapolate the form of the paw of the unknown one. The explorer is not merely tracing a sort of direct line between the elements of the track and the elements of its cause; a series of content units works as mediating cues. In other words, the explorer is abducing an unknown code using excerpts of the existing ones. Only in science fiction is there the case of imprints of ‘I-don’t-know-what’; and in these cases it is absolutely impossible to go backward from the trace to the referent, which is called, faute de mieux, ‘The Thing’.
25. Suppose that one finds a huge footmark. The first naive inference could be: here has passed a giant. There is a content unit (duly coded) that has made the inference possible. But this unit is usually marked with the connotation «legendary». Thus one is interpreting a mention and discovering that this mention is mentioning something unexisting and therefore not mentionable. In such a case it is possible to assert that the mention is false; therefore the expression is a lie. If the lie had been expressed through words it would have been a mere lie; since it is expressed by a drawing, it is usually enjoyed as a joke. For it is easy to produce words, but it is less easy to produce images, especially if they are customarily produced by unintentional agents; moreover, the perfection of the image requires more skill than the perfection of a phrase. Thus the fake is amusing for two reasons: (1) it is an elementary case of artistic skill; (2) it falsifies something that was commonly believed to be non-falsifiable, that is, the product of a non-intentional agent. Men are supposed to lie, things aren’t; thus to make things lie seems to be a rather curious achievement. So we laugh.
26. When, in a Western movie, Indians emit a coyote’s cry, this full onomatopoeia plays a double role. Indian to Indian it is a purely arbitrary device, duly coded so as to transmit certain information; Indian to Whites it is a fictive sample produced in order to mean «coyote» (instead of «Indians»); therefore it is a lie.
27. Jocular gestures of menace, if not previously coded, may be taken as effective ones. If one looks at Marcel Marceau without knowing mimicry conventions one believes that he is a fool.
28. The science of music offers us an excellent model for research. Music, insofar as it has a tonal ‘grammar’ and has elaborated a system of notation, analyzes the continuum of sounds into pertinent features (for pitch-tones and semitones; for rhythm – metronomical beats, crochets, semibreves, etc.). Any musical discourse can be accomplished by articulating these features. The immediate objection is that, even though the notation prescribes how ‘to speak’ musically on the basis of a digital code, still the single message (execution) is enriched with non-coded free variants. Thus a glissando, a trill, a rubato, the length of a pause are considered (both in common and in critical language) as ‘expressive’ devices. But, although it recognizes these expressive facts, the science of musical notation does everything possible to codify and make replicable even the variants. It codifies the tremolo, the glissato, adds notations such as “allegro ma non troppo”, etc. As will be noted, these codifications are not digital, but ‘analogical’, and proceed by degrees (more or less) summarily defined. But though they are not digital they can be digitalized; to demonstrate this, however, one goes not to the level of notation for an interpreter, but to that of technical codes of sound transcription and reproduction. Every minimal expressive variation in the grooves followed by a phonograph needle on a record corresponds to a precise feature. It will be said that these features cannot be considered as discrete either, since they proceed as a continuum of curves, of more or less accentuated oscillation. But when one considers the physical process by which we pass from this continuum of gradated curves, through a sequence of resultant electric signals and a sequence of acoustic vibrations, to the reception and retransmission cf the sound through the amplifier, one is concerned with digital phenomena. From the groove to the needle, before the input into the electronic blocks of the amplifier, sound is reprodueed through a continuous model; but from the input into the electronic blocks to the output of sound from the amplifier, the process becomes discrete. The communications engineer is always trying, even in computers, to transfer analogical models into digital codes, a process which is always possible. The discrete nature of the signal is confirmed by the existence of noise. Noise can exist in the channel precisely because the emission of electrons, which is what the signal is, is a discrete phenomenon. If the passage of information were continuous for the whole process, there would be no noise.
29. The whole of generative grammar is practically concerned with the problem of vectorialization in the ‘deep’ phrase marker. The notions of “government”, “command” and “embedding” are vectorial notions (concerning hierarchical dispositions, ‘up and down’ or ‘before and afterward’ relations). In this sense, vectorializations should be considered as contextual features, which however does not contradict the assumption made in 2.9.3., according to which the representation of the sememe involves even combinational rules. In principle, since a vectorial device is a position within a spatial or temporal arrangement, given the representation of items in the form of n-places predicates, one could assume that any representation of argument displays, as a contextual selection, the representation of its role within a given
But such a solution gives rise to a first problem: is the vectorial position to be represented the one realized by surface structure or the one realized by deep structure? Lakoff (1971 a) shows that the expression /Sam claimed that John had dated few girls/ is open to both readings /Sam claimed that the girls whom John had dated were few/ and /The girls whom Sam claimed that John had dated were few/. Vectorial devices in surface structure are misleading and only a correct ‘vectorialization’ of the quantifier in a deeper semantic structure can disambiguate the expression. Nevertheless the labor of producing signs (and in particular ‘code observing’) consists in mapping the deep semantic representation in the surface structure by several rules or “constraints”. These rules establish how vectorial devices may be realized and interpreted in surface structure. At this point the semantic representation (in terms of generative semantics) is an interpretant, or a compound of vectorial devices, which acts as a metalinguistic device able to explain the labor of choosing and producing signs. Thus vectorializations are present (i) in the metalinguistic representation of a semantic theory, (ii) in the semantic representation of the combinational possibilities of a given sememe, (iii) in the very labor of inference and presuppositions displayed in interpreting a context. Since all these problems are not sufficiently clear either for a transformational grammar or for an interpretive or for a generative semantics (and they are rather a matter of a discussion in progress) let us assume that, for the time being, the problem of vectorial devices in verbal languages is a very puzzling one and demands further research. This demonstrates that even the most peculiar problems of a linguistically oriented semantics cannot be solved without solving general semiotic problems concerning vectors at large.
30. In different languages the position of the lexeme within the surface structure does not represent the vectorial feature which permits the correct disambiguation of the sentence. For instance, in Italian one commonly says /Giovanni ama Maria/ (John loves Mary) – where the Agent comes before the Object – but it is also possible to say (even if with a shade of ‘poetry’) /ama Maria Giovanni/. In this second case the position does not permit us to recognize who is the Agent. Nevertheless, when speaking, the ‘positional’ feature is easily substituted by a paralinguistic one (as Lakoff shows, 1971 b): if I say /ama Maria, Giovanni!/ stressing on the second name, I have explicitly ‘vectorialized’ (and paralinguistic devices are substituted, in the written expression, by comma and exclamation mark). Sanders (1974) reminds us that in Spanish it is possible to say both /Manuel me presentö a ti/ (Manuel presented me to you) and /Manuel te me presento/. But in the second case we are entitled to speak of overcoding, since the expression can be considered as a ready-made syntagm, or an idiom.
31. For all the problems dealt with in this paragraph I am indebted to Krampen (1973) and Volli (1972). The whole discussion on iconic signs which has developed in VS magazine (1972, 1973) through the interventions of Farassino, Casetti, Bettetini, Volli, Verón, Metz, Osmond-Smith and others has been useful in order to clarify many points of the present discourse, which is presented as a critical summary of the above contributions. I am also indebted to Tomás Maidonado (personal communications), Thomas A. Sebeok (personal communications), Roman Jakobson (personal communications) and many others.
32. Volli (1972:25), referring to the conceptions of geometry expressed by Klein in his Erlangen Program, 1872, according to which “every geometry is the study of the properties that remain unchanged in relation to a determined transformation group”.
33. “Let us first of all consider congruences, which is to say the transformations which make each segment correspond to a segment of equal length. They leave the metrical properties of the geometrical entities unchanged, and therefore obviously do the same for the affinitive, projective and topological ones, etc. We are concerned with what we would intuitively call the ‘equality’ of geometrical figures. Two objects ‘with the same form’ even if made of different substances, within the limits of the approximation mentioned above, are evidently the simplest possible iconic signs of each other; their iconic relation can be rigorously defined in terms of congruence between the respective forms of expression. It may be noted that mirror reflections are also congruences” (Volli, 1972:25).
34. Let me range under the name of ‘projections’ all the phenomena as homotetiae and projective transformations: “Another interesting type of transformation is provided by the homotetiae, which in elementary geometry are called similarities and do not preserve all the metrical properties of the figures, while the affinitive, projective and topological properties remain unchanged. A plastic model of a building is the most typical example of an iconic sign based on homotetiae. We then have projective transformations, which are those that give rise, among other things, to correspondences of perspective, and in addition to the topological properties maintain unchanged all the projective properties, such as that of ‘being a straight line’, ‘being a second degree curve’, the two-way relationship between the four points of a straight line, etc. Photographs and most graphic reproductions constitute examples of iconic signs based on projective transformations of the form of expression of the object. But in this type of sign both the perceptual principles of approximation, and the fact that the form of expression of the iconic sign is the transformation of only one part of that of the object, play an important part” (Volli, 1972:25).
35. Even the projections that seem to work as ‘indexical’ imprints of a given set of objects are in themselves the result of a highly simplified mapping from a few pertinent traits of a perceptual model. See Gibson (1966:190 ff.) for the way the representation of a room is realized from a given station point. The most customary mode of projection consists in considering only edges and borders. While the scientific (even though abstract) representation of the scatter-reflection of lights from surfaces gives “a dense space-filling network of reverberating rays from everywhere, every normally recognizable image of the object in the room maps only the rays from the faces of the surfaces and objects . . . not those from the facets of things . . . . It can be observed that each face, as defined by edges or corners of the surfaces, corresponds to a perspective projection”; moreover, all is mapped from a monocular point of view. It is the addressee that, in interpreting the image, fills up the empty spaces and maps backward from the ‘iconic’ abstraction to the perceptual model. The representation of a perceptual field is based on projective conventions, and as such has to be learned. For this reason the representation of unknown objects becomes difficult and cannot be solved by means of projective conventions, so it needs both programmed stimuli and a combination of stylizations of various types.
36. “Finally we have topological transformations, which only conserve certain very elementary properties, such as the continuity of lines and the arrangement of potential systems. The most typical and important example of iconic signs that derive from topological transformations is that of diagrams. The diagram of the underground or of a railway system, or that of the structure of an electric or electronic apparatus only have certain fundamental properties of arrangement in common with the object to which they refer, yet they conserve a great capacity for explanation, clarity and information. Topological transformations can also take into account orientation, as in a map of a city that reproduces one-way streets” (Volli, 1972:25-26).
37. In Eco (1968) – English translation as Eco (1973 e) – I tried to define architectural signs as manufactured objects and circumscribed spaces that signify possible functions (going up and down, coming in or out, sheltering, gathering together, sleeping, eating, celebrating events, etc.) on the basis of previous codes. In this sense I distinguished between processes of stimulation (a step that I stumble over in the dark, forcing me to raise my legs) and processes of signification: a //staircase// consists of the articulation of morphological elements which express the function for ascending; the staircase can be used only if recognized as such, and can be recognized without being used (it can even signify its function without in fact allowing it, as in cases of trompe-l’oeil). Thus the content of an architectural expression is a class of possible functions. In Eco (1968) I had also distinguished between two types of signified function: primary functions, i.e. the first and immediate denotative content of the expression (such as going upstairs, standing at the window, living together, etc.), and secondary functions, i.e. the mediate connotative content (such as various ‘symbolic’ meanings, «mystic atmosphere», «deference», «triumph», and so on). In the article listed as Eco (1972 c) I have also tried to propose a model for the compositional analysis of architectural objects, such as for instance a column, taking into account many contextual and circumstantial selections, such as for instance the presence of the column among ruins, which adds to the object many ‘archeological’ and ‘historical’ connotations, duly recorded by the critical literature on architectural aesthetics. Both Eco (1968) and Eco (1972 c) were still linked to the notion of signs criticized in the present book. One should now read them by substituting for the notion of ‘architectural sign’ that of ‘architectural text’ in which many modes of sign production are simultaneously at work.
38. Breviario di estetica, Bari: Laterza, 1913, 9th ed. 1947, pp. 134-35.
39. “The political slogan ‘I like Ike’ [ay lay k ay k] succinctly structured, consists of three monosyllables and counts three diphthongs [ay], each of them symmetrically followed by one consonantal phoneme, [. . l . . k . . k]. The make-up of the three words presents a variation: no consonantal phonemes in the first word, two around the diphthong in the second, and one final consonant in the third. A similar dominant nucleus [ay] was noticed by Hymes in some of the sonnets of Keats. Both cola of the trisyllabic formula ‘I like/ Ike’ rhyme with each other, and the second of the two rhyming words is fully included in the first one (echo rhyme), [layk] – [ay k], a paronomastic image of a feeling which totally envelops its object. Both cola alliterate with each other, and the first of the two alliterating words is included in the second: [ay] – [ayk 1, a paronomastic image of the loving subject enveloped by the beloved object. The secondary, poetic function of this electional catch phrase reinforces its impressiveness and efficacy”. In fact this message acquires part of its ‘charm’ because it brings into play phonic elements, and thus a sort of ‘musical’ quality which is linked neither to syntactic devices nor to semantic ones. It obviously shows that if one changes the expression the content is also affected (one cannot obtain the same effect by saying /It is Ike that I like/ or /Ike is liked by me/). But it also shows that a particular pronunciation of the slogan (perhaps one with slang overtones), reinforces its ‘aesthetic’ effect.
40. By integrating and articulating differently a classification suggested by Bense (1965) we may say that in an aesthetic message there are many levels of information: (a) physical supports: in the verbal languages they are tones, inflexions, phonetic emissions; in visual languages they are colors, stuffs; in music they are timbres, pitches, dynamic stresses, temporal durations; they are expression-purport or elements of the material continuum which the signal is made of; (b) differential elements of the expression plane: phonemes, rhythms, metrical lengths, positional relations, geometric and topological forms, etc.; (c) syntactic rules: grammar rules, proportional relationships in architecture, perspectives in painting, musical scales and intervals; (d) denotative contents; (e) connotative contents; (f) overcoded ready-made strings: stylistic subcodes, rhetorical systems, iconographical models, and so on; (g) others. Bense, however, speaks of a global ‘aesthetic information’ which is not achieved on any of these levels in particular, but rather on the level that he calls ‘Mitrealität (which is signified by all the correlated levels). Bense’s ‘Mitrealität’ seems like the general contextual situation of improbability that the work exhibits, but the term (because of the Hegelian mold of the author) is slightly colored with idealistic connotations. Mitrealitdt seems then to denote a certain ‘essence’ – maybe the Beauty – which can be intuitively grasped but not semiotically singled out. On the contrary one of the tasks of a semiotic approach to aesthetics must be its capability to analyze and describe in systematic terms why the impression of Mitrealität arises. In the following paragraphs the notion of aesthetic idiolect will satisfy this need.
41. Nevertheless, even though one admits that the author is molding something that elicits responses without really ‘communicating’ contents, one has to assume that either the author or the art analyst knows that these clusters of unpredictable and undetected microstructures ought to produce a predictable response. There is at least a moment in which the relationship between clusters and responses is coded (or abduced) as a semiotic correlation. These are cases of programmed stimuli (see 3.6.6.). Imagine that a critic knows that a given artistic procedure (recognized as ‘the secret of that great master’) appears in many works of that master and inflexibly produces an ‘unspeakable’ feeling and a specific kind of enjoyment; this kind of knowledge would constitute the last (and at any rate the more satisfactory) form of ‘rationalization’ that semiotics can exert about a work of art.
42. The complexity of the inferior levels, which seems to escape any definition, has induced some aestheticians to consider these levels as extraneous to the poetic form. Galvano della Volpe (1960), who was fighting against the various idealistic theories of the ‘ineffability’ of art, in order to avoid everything which could escape a rationalistic approach to art, has excluded all the ‘musical’ values that cannot be coded. These values were relegated to the range of ‘hedonistic’ elements, stimulators of an extra-aesthetic enjoyment. Hence the repudiation of many phonetic phenomena that, according to many critics, constitute the principal appeal of poetry; hence the rhythm looked suspiciously at; hence the statement that the aesthetic value is what remains when a work is translated into another material support, where musical devices can change in impact but there remains a kind of logical and conceptual pattern capable of keeping the really significant relationships. It is curious that this effort toward a ‘rationalistic’ theory of art happens to leave to the ‘irrational’ elements much more than they deserve. To renounce extension of the rationalistic analysis to the lower levels of the expression form leads to equating ‘rational’ with ‘semantic’ – which is a very poor way to refute the idealistic approach to the ‘unspeakability’ of art. The correct trend, even from the point of view of Della Volpe, should be, on the contrary, the one outlined in the present chapter, that is, to overcode the expression substance.
43. “Poetics deals primarily with the question, What makes a verbal message a work of art? Because the main object of poetics is the differentia specifica of verbal art in relation to other arts and in relation to other kinds of verbal behavior, poetics is entitled to the leading place in literary studies. Poetics deals with the problem of verbal structure, just as the analysis of painting is concerned with pictorial structure. Since linguistics is the global science of verbal structure, poetics may be regarded as an integral part of linguistics . . . . Many poetic features belong not only to the science of language but to the whole theory of signs, that is, to general semiotics. This statement, however, is valid not only for verbal art but also for all varieties of language since language shares many properties with some other systems of signs or even with all of them (pansemiotic features)” (Jakobson, 1960:351).
44. See Eco, 1962, where the theory of the ‘open work’ is approached from a pre-semiotic point of view.
45. The aesthetic idiolectic has not to be considered a code ruling one and only one message (which, however, should not be viewed as a theoretical contradiction); it is rather a code ruling the various different messages which compose that complex network of messages called ‘aesthetic text’. Nevertheless it is (or it may in principle be) a code ruling one and only one text, a code destined to produce a unique discourse. This (and nothing else) explains the creative and individual character of a work of art. Ruled by the idiolectal aesthetic code and connecting various messages which are to be taken as radical instances of a rearranged underlying system, the work of art is a system of systems (see Jakobson and Tynianov, 1927; Wellek and Warren, 1942).
46. Roland Barthes (1963c.) once said that the work of art “est une forme que l’hstoire passe son temps à remplir”. I agree with this statement but I would prefer to re-translate it into the categories of the present semiotic approach: the work of art is a text that is adapted by its concrete addressees so as to fulfill many different communicative purposes in diverse historical or psychological circumstances, without ever completely disregarding the underlying rule that has constituted it.
47. The above example is perhaps a too perfect one, for there is also, along with the semantic substitution, a witty phonetic echoing: thus the metaphor is reinforced by the pun, that is, the substitution on the content plane is reinforced by the co-presence on the expression plane. On this aspect of the ‘metaphor+pun’ see Eco, 1973b, on Joyce’s technique; the pun is a figure of speech that, instead of substituting a ‘tenor’ by a ‘vehicle’ (according to Richard’s definition) makes the word corresponding to the vehicle embody the word corresponding to the tenor. In such cases the ‘necessity’ of the substitution is reinforced, such as happens with the rhyme according to Jakobson, where a parallelism on the expression plane co-involves a parallelism on the content plane. Thus, and the remark is particularly valid for medieval culture, one is further convinced that nomina sunt numina or that nomina sunt consequentia rerum.
48. These distinctions are also recorded by the medieval definitions of loci: “Quis, quid, ubi, quibus auxiliis, cur, quomodo, quando But these questions only partially concern the format of the sememe. Some of them concern contexts and circumstances, therefore a series of uncoded presuppositions. Let us then assume that there exist cases of empirical contiguity that cannot be recorded by a semantic theory. For instance, when dreaming, one frequently relies on connections based only on one’s personal and idiosyncratic experience. More open to coding appear some metaphors such as the substitution of penis by vertical objects (already discussed in semiotic terms by Morris, 1938). Thus it would be risky to call metonymies all the substitutions by ‘contiguity’ occurring during the dreams or the psychoanalytic verbal interaction – and interpreted as such by the analyst. Psychoanalysis seems to be rather a matter of interpretation of as yet uncoded texts, producing further overcoding, and the discourse of the patient has many points in common with an aesthetic text whose idiolect has to be discovered by the critic. Many of the ‘contiguities’ discovered in this text are contextual presuppositions, called “referentials” by Hiz (cf. 2.11.1.).
49. The functioning of metaphors and metonymies explains the mechanism of every other trope, that is, all the substitutions by immutatio. Periphrasis is the substitution of a lexeme with the whole (or a great part) of the markers of the corresponding sememe. Antonomasia is both a sublease of synecdoche (species pro individuo) and periphrasis. Since the sememic representation analytically (or semiotically) includes the negation of the antonym, litotes are a common case of substitution of the sememe by a marker, and ironies are the direct use of the antonym (but, being in fact figures of speech and of thought, they involve more complex contextual substitutions). Eptphases are types of synecdoche, and hyperboles are types of metaphor. A different discourse should be made for the figures of speech and of thought (which proceed by adjectio, detractio and transmutatio) and which rely on phonological, syntactical and inferential mechanisms. Groupe p. in Rhétorique générale carefully distinguish between figures of expression (metaplasms and metataxes) and figures of content (metasemenes and metalogisms); it must be stressed that the suggestions offered by the present book work only for a study of metasemenes.
50. Katz (1972:8.4.) proposes to add a rhetorical representation to the canonical theory of grammatical components, saying that “the theory of grammar requires a next subtheory, namely, a theory of rhetorical form, and, further, that grammars require a new component to express the rhetorical interpretation of superficial phrase markers”. Katz thus assumes that rhetorical manipulation may only alter surface structures without interfering with their semantic interpretation. It should be clear that the theory outlined in this book is exactly the opposite, even if one might admit that in exceptional cases some rhetorical manipulation may only determine very peripheral connotations. But also, in these cases, rhetoric is concerned with semantics and I think that it is impossible to consider the rhetorical component independent of the semantic one; they are rather to be viewed as two sides of the same semiotic problem.
51. For example: “Our society has to improve and to increase production; many sacrifices will be required of every member of the community in order to achieve our goal. Individuals do not count in respect to the collective welfare”. The same kind of coding is implicitly established by another statement of the type: “Productivity produces money, money produces welfare; obviously in this struggle for life – or in this free competition – somebody must be overcome; this is the price to pay for an affluent economy”. All these statements constitute implicit meta-semiotic judgments imposing some semantic rules for connecting values. On the other hand different premises may be stated: “Better poor than slaves of our own affluence”; “A society in which somebody dies in order to produce wealth for somebody else is an insane society”; “Burn grass not oil”; “Social security must be the first care of this government”. All of these statements (easily recognizable as belonging to different ideologies) are nevertheless as many examples of rhetorical persuasive efforts to make people think in another way; therefore they constitute meta-semiotic judgments aiming to assign new connotations to old semantic items.
52. There are no objective rules for transformation of an ideology into another one; the disconnection of the semantic space allows one to demonstrate that different biases produce different semantic organizations. There is no theory of the ideologies that would be able to test and to improve them. There is a semiotic technique of analysis that allows one to destroy an ideology by opposing to it another ideology, the latter showing the falsity of the former (and vice versa). The choice of the right or of the correct bias is not a semiotic matter. Semiotics helps us to analyze different ideological choices; it does not help us to choose.
53. At this point we can take up and develop a problem already aired by Barthes (1964 b), that of the relationship between rhetorical formulae and ideological positions. It is not likely that a Communist would indicate the necessity of the Third World’s struggle against the Western powers by the phrase /the defense of the free world/ even if he considered the autonomy of colonial people to be the only form of freedom for which it was worth fighting. The rhetorical formula /defense of the free world/ is henceforward strictly associated with political positions which are identified with the United States, their allies and their ideological vision. Naturally the same operation could be accomplished on a formula such as /brotherly help to the socialist allies/. Thus a certain way of using a language is identified with a certain way of viewing society.
54. The whole of the above discussion verifies the most current definitions of ideology. The sense given to this term by the French ‘idéologues’ of the eighteenth century is practically equivalent to our conception of semiotics as a ‘genetic’ criticism of ideologies. Ideology as conscious code-switching is what Engels called “a process that the so-called thinker accomplishes consciously but with false conscience. The true moving forces that determine him remain unknown (otherwise it will not be an ideological process,” (. Letter to Mehring). Ideology as unconscious code-switching is described by Jaspers as “the complex of thoughts and representations appearing as an Absolute Truth to the thinking subject for the interpretation of the world . . . producing a self-deception, a concealement, an escape (from reality) ”Die geistische Situation der Zeit). On the other hand the Marxist ‘positive’ sense of ‘ideology’ as an intellectual and political ‘weapon’ serving the social purpose of active modification of the world does not contradict the preceding negative definition; in this sense an ideology is taken without denying its one-sidedness and without concealing what it refuses; except that a previous system of premises has clarified what one wants to get and what one prefers (see note 50), on the basis of a given theory of society and of material needs.
55. The ideological labor can assume even more complex forms. Through metaphorical substitution it is possible to equate «energy=comfort» and to oppose them to «danger». It is possible to translate «danger» into «less security» and therefore demonstrate that more energy implies less security, which is a price that might be paid to have high comfort. It is possible to introduce surreptitious fuzzy concepts without establishing their exact gradation; suppose that the productivity theorist, having asserted as an acceptable premise that All Comfort for All is incompatible with All Security for All, proposes a pseudo-logical square of the type
All Comfort for All – All Security for All
Some Security for Some – Some Comfort for Some
in which appearently the first antonymous tuple opposes two contraries, the second one two converses, while a general comfort which implies some security seems to be a fair issue. It suffices to recognize «All» and «Some» as fuzzy operators that change their own semantic nature depending on their positions, and the pretended logical exactness of the square is challenged: does All Comfort for All mean «equally distributed to everybody» (socialism) or «potentially at the disposal of everybody» (free competition)? How much is the first «Some» quantifying «Security»? And for how many is the second one? The ‘game’ could continue indefinitely. Only when each term is sent back to its position within the codes and semantically analyzed, can the ideological labor can be unmasked and be taken back to a persuasive discourse based upon a logic of preference.
