The new discipline was interested not so much in the good formation of a sentence (a study it delegated to linguistics) or in the relation between sentences and facts (which was unfortunately left in obscurity) as in enunciative strategies for making something «appear to be true.» And therefore interest was focused not on what happens when someone says It’s raining today and it really is (or isn’t) but on the mechanism according to which, by talking, one can induce someone to believe that it is raining today, and the sociocultural impact of that disposition to believe.
Consequently, when someone was confronted with an advertisement showing a glass of ice-cold beer, the problem was not so much to explain whether and why the image corresponded to the object (and we shall see later that the problem would not go away) as to explain what universe of cultural assumptions was brought into play by that image and how the image aimed at reiterating or modifying that universe.8
One invitation to consider the phenomenon of iconism should have come from the encounter with Peirce—and it should be said that most of the demand for a rereading of Peirce as a semiologist came precisely from within the semiostructuralist paradigm.9 But with regard to Peirce’s work there is no doubt that more attention has been paid to the aspect of unlimited semiosis, the growth of interpretations within the cultural Community (certainly a fundamental and indispensable aspect), than to the more properly cognitive moment of the impact with the Dynamical Object.
These were the reasons for the polemic against so-called naïve iconism, which is based on an intuitive notion of likeness. The polemic was directed not so much at Peirce as at those who had simply confused iconism (as a perceptual moment) with hypoicons. If by icon one meant an «iconic sign» (and therefore, in Peirce’s view, a hypoicon, whose «symbolic» or broadly conventional content he never denied), then saying that it possessed the properties of the object represented looked like a way of placing signs in a direct (and naïve) relation with the objects to which they referred, thus losing sight of the cultural mediation to which they were subjected (in short, treating phenomena of Thirdness as Firstness). I think (and the reader is referred to 2.8) I have made amends for those past simplifications, but it is also necessary to understand the reasons why people reacted the way they did at the time.
The virtually incontrovertible presupposition that hypoicons referred to their object by natural and immediate likeness, without the mediation of a content, was a way of reintroducing into visual semiotics that direct line between sign and referent that, with quasi surgical brutality, had been expunged from the semiotics of verbal language.10
It was a matter not of denying the existence of signs motivated in some way by something (and in fact I devoted to this problem the whole section on ratio difficilis in A Theory of Semiotics) but of making careful distinctions among motivation, naturalness, analogy, noncoding, «weak» coding, and unsayability. This attempt took various paths, some of which proved to be dead ends, but others led somewhere.
6.4. Dead Ends
As an example of an absolute dead end I would cite the attempt to examine not only hypoicons but also semiotic systems such as architecture through linguistic categories—for example, minimal distinctive units, double articulation, paradigm and syntagm, et cetera. This attempt could not have led very far, but historic reasons applied in this case too. Consider the debate with Pasolini (1967a), when he maintained that the cinema is based on a «language of reality,» an innate language of human action, in which the elementary signs of cinematographic language are said to be the real objects reproduced on the screen.
Although Pasolini was later to moderate the radicalism of those early statements in an essay that ought to be reflected upon anew today from a Peircean standpoint (1967b), his reaction was due to the fact that «hard-line» semiologists were interested in demythologizing—as they used to say then—all productions of realistic illusion and in revealing all that was artifice, montage, and pretense in the cinema. 11 And that is why we had come to individuate at all costs the presence of analyzable «linguistic» entities in films too, and I cite my own pages (1968: B4, 1.5–1.9)12 on triple articulation in the cinema, pages unfortunately still translated and republished in various anthologies but not worth rereading, unless for documentary purposes.
By way of an example of a path that certainly led somewhere but not in the direction intended, I would mention the attempt to reduce the analogical to the digital, i.e., to demonstrate that those hypoiconic signs that appeared to be visually analogous to their objects could also be broken down into digitalized units and were therefore translatable into (and producible by means of) algorithms. I am proud to have posed a problem that might have seemed an irrelevant technicality in the sixties but—in the light of computational theories of the image—is of the maximum importance today. But at the time the observation had rhetorical value only, because it suggested that the aura of «unsayability» surrounding hypoicons could be reduced. From a semiotic point of view this did not solve anything, because to assert the digital translatability of the image to the expression plane does not eliminate the question of how an effect of likeness comes about on a cognitive level.
6.5 Likeness and Similarity
The other path was to prove more productive. Since the notion of likeness seemed vague and in any case circular (that which looks-like is iconic, and that which is iconic looks-like), it had to be dissolved in a network of procedures to produce similarity.13 What the rules of similarity were was revealed to us by projective geometries, the Peircean theory of graphs, and the elementary concept of proportion itself. But this did not eliminate the problem of perceptual iconism, and of how an element of primary iconism—»likeness» in the sense of Peircean Likeness, the very basis of perceptual constancy—can survive even in the perception of hypoicons (based on criteria of similarity).
Taking their cue from Palmer (1978), May and Stjernfelt (1996: 195) propose the example as shown in figure 6.1:
Figure 6.1
Imagine a world represented only by the objects a—d (it is not necessary to establish whether this is a real universe or a possible world inhabited by abstract entities). Consider Ai, A2, and A3 as three different «iconic» representations of this world (incidentally, these would have every right to be considered as three interpretations of the world, just like those discussed in 1.8). Each of these three representations adopts a single criterion for establishing similarity: by expressing the property of being «higher than,» the criterion/ (applied to Ai) pertinentizes only the relations in terms of height between the four figures in the world, and this is why d is represented in d’ by a vertical line, abstracting from the undoubted property of breadth or horizontality that d shows in relation to the other three figures.
The criterion (in A2) again pertinentizes relations of height, but by representing the property of being «shorter than» (thus creating a visible relation of inverse symmetry between Ai and A2). The criterion h (in A3) is more complex: it pertinentizes the extent of area, but expresses the property «bigger than» through the mapping mechanisms employed in A2. In other words, the bigger an object, the shorter the vertical line that represents it. The three representations are certainly motivated by the nature of the objects (the length of the lines cannot be chosen arbitrarily), and therefore they certainly establish a hypoiconic relation between the representation and the represented. But this relation, which is defined as homomorphic and retains in the representation some structural properties of the thing represented, is not isomorphic, insofar as the representation does not have the same form as the thing represented.
This is a good example of similarity, motivated yet established according to rules. A certain «likeness» between each representation and the thing represented is maintained even when the rules of similarity are changed. By the way, this procedure corresponds to what in A Theory of Semiotics I defined as ratio difficilis (points of a virtual space of the content are projected onto the expression) and corresponds to what the post-Hjelmslevian tradition (and, most of all, the school of Greimas) has called the semisymbolic: where we have systems characterized not by the conformity between the expression plane and the content plane (as in a picture of a chessboard at a certain point in the game, or in a portrait) but through a correlation of two relevant categories of different planes (Greimas and Courtis 1979).
In more comprehensible terms, in Jakobson (1970) the motor gestures for yes and no are not motivated by an object (which?) that they «would resemble,» but they correlate, according to a nonarbitrary relation, a motor-spatial configuration (the movement of the head) with a categorial pair (affirmation and negation)—and even when in some cultures they seem conventionally different from ours, they nevertheless have a relation of motivation with the content they express.
Nevertheless, an understanding of these three representations is based on a perception of the difference in the length of the lines (not to mention the different format of the rectangles): now, this property of being longer or shorter is not established by the rule of similarity but is its requirement on the basis of the natural iconism of perception.
When I perceive a ball as such, I react to a circular structure. I
