The Contiguity Illusion

Décio Pignatari

When I start with the warning, as so many semioticians now do, that I am using language in its broadest sense, this seems to indicate that verbal imperialism is starting to crumble, at last. . . .

It was David Hume who first distinguished the two basic forms of association: association by contiguity and association by similarity or resemblance. Charles Sanders Peirce defined them (7.391, 7.392) and threw some darts at certain psychologists of his time who maintained that the existence of two forms of association was contrary to science. For them, association by resemblance was merely a subclass of the only valid principle of association—association by contiguity.

Why is contiguity endowed with such a privilege? The reason lies in a kind of logical illusion—the contiguity illusion—which we may also observe in the works of many contemporary linguists and semiologists. This illusion, in all appearances, was born directly from Western linguistic systems and has earned a droit de cité in the written notation of these systems, that is, the alphabetic code (in this connection, the fact that the unit letter was isolated many centuries before the unit phoneme cannot be dismissed). These linguistic systems favor assoeiation by contiguity. What we formerly called classical logic—i.e., Aristotelian and linear—is the logic embodied in the Greek idiom: it is a contiguity logic. The alphabetic code—a most powerful logical machine —is a highly abstract discrete source of signs with digital and metonymic characteristics. Words are formed by combinatory permutations, that is, syntagmatically, and are linked together following the pattern of predication (especially when the verb to be is employed; “this is that”), the logical pattern par excellence. Predication units, in their turn, are articulated by linkage elements called conjunctions —first-class connections when hierarchy is implied (hypotaxis) , and second-class connections when ‘non-hierarchy’ is implied (parataxis). From predication to sentences or propositions to concepts to Peircean arguments, we have the complete chain that has led ultimately to Western science and technology (at least if we believe in the verbalized histories of science and technology). Ergo, when we ‘talk logic’, or ‘talk science’, we mean that inferences have been drawn through contiguity. But is there no possibility of another kind of logic—that is, a logic by similarity?

The Western mind is contiguity-biased through language (the verbal code), which is itself based on contiguity. If we look at current European semiology, what we see is not so very different from the state of things as described by Peirce. What we see is ‘semantics’ nicknamed ‘semiology’, and Roland Barthes stating that semiology is a branch of linguistics rather than linguistics being a part of semiology. In fact, what we call logocentrism is another name for association by contiguity. When the word is taken as the central code, we are led to believe that all signs only acquire ‘meaning’ when translated into “Wordish,” the verbal code. The rational mind, it follows, is the contiguity mind. When dealing with analogy, ‘scientific’ minds become very cautious: analogy is a dangerous path to follow—it is almost . . . non-scientific.

In Peirce’s words :

Uncontrolled inference from contiguity, or experiential connection, is the most rudimentary of all reasoning. The lower animals so reason. A dog, when he hears his master’s voice, runs expecting to see him; and if he does not find him, will manifest surprise, or, at any rate, perplexity.

Inference from resemblance perhaps implies a higher degree of selfconsciousness than any of the brute possess. It involves somewhat steady attention to qualities, as such; and this must rest on a capacity, at least, for language, if not on language, itself. Primitive man is very industrious in this sort of reasoning. [7.445-46]

I doubt very much whether the Instinctive Mind could ever develop into a Rational Mind. I should expect the reverse process sooner. The Rational Mind is the Progressive Mind, and as such, by its very capacity for growth, seems more infantile than the Instinctive Mind. [7.380]

Firstness, secondness, thirdness. From sign to object to Interpretant; from icon to index to symbol. Firstness is form, the quality of a feeling ( feeling = “the immediate element of experience generalized to its utmost” [Peirce, 7.364]). Similarity thus comes first:

In this chaos of feelings, bits of similitude had appeared, being swallowed up again. Had reappeared by chance. A slight tendency to generalization had here and there lighted up and been quenched. Had reappeared, had strengthened itself. Like had begun to produce like. Then even pairs of unlike feelings had begun to have similars, and then these had begun to generalize. And thus relations by contiguity, that is, connections other than similarities, had sprung up. All this went on in ways I cannot now detail till the feelings were so bound together that a passable approximation to a real time was established. [Peirce, 8.318, “Letter to Christine Ladd-Franklin, On Cosmology”]

Associations by contiguity in the verbal code can therefore be summarized: to summarize a thesis is to retain its essence, to summarize a poem is to lose its essence (Valéry, 1954:1244). A summary cannot be made of a form. But what about a novel, or a short story? It is not by chance that in many Western idioms the term argument serves the reign of logic and the reign of fiction as well, to indicate the narrative line, the plot, the summary of a story. The work of Propp, for instance, has not been solely concerned with narrative functions, but also with the fact that narrative is built upon the predication pattern, upon associations by contiguity connected with cause/effect associations. Fairy tales and narratives in general present the same structure that can be found in predication, as well as in the sentence and the phrase (Propp, 1966: 121).

In 1968, I conducted a seminar at the Faculdade de Filosofia, Ciencias e Letras de Marília (São Paulo, Brazil) that was designed to analyze the phenomenon of plot rarefaction in modern fictional prose. Two major conclusions were reached:

1. The narrative depends on the number and the hierarchic structure of characters (hypotactic structure) ; and linear-discursive development is implied;

2. The reduction of the number of characters causes, at the same time, narrative 'impoverishment' (plot rarefaction) and simultaneity of actions (Joyce's Ulysses, the so-called French nouveau roman, and Alain Resnais' l'année dernière à Marienbad, serve as illustrations ).

This is why French semiologists shun modern prose analysis and restrict their narrative analyses to plot analyses—the plot being understood as logical 'argument'. Once more, they fall victim to the contiguity illusion, which in the Western world is stretched to such a degree that contiguity is even smuggled into similarity. The most startling example is the socalled analogic dictionary, such as we have in Portugal and Brazil, in which terms and expressions are clustered around certain themes (i.e., ideas related by contiguity ). It is interesting to observe that the organizing principle behind the making of these dictionaries is the same as that for the making of metaphors. Common dictionary entries and, even more so, those in a dictionary of rhymes, are truly analogic: they are organized following formal analogy, that is, following form similarity, in the same manner as ideogram organization. Their structure is paramorphic—the same structure that underlies poetic organization.

(En passant, it is worth noting, I think, that the written verbal code presents similarity patterns of its own, often quite independently of the sounds that they are supposed to represent [such as bdpq or BDPR]; we face here an almost autonomous language, with particular semiotic problems in modern poetry and communication [e.g., the iconization of the written verbal code]. In Mallarmé's Un coup de dés, all the e's of a special typeface stand for spermatozoa; in Poe's Berenice, the thirtytwo letters of the sentence que toutes ses dents étaient des idées stand for thirty-two teeth [Pignatari, 1974:95-96, 124]. A letter that is twenty feet high is no longer just a letter. )

Prose is contiguity's natural kingdom, as it were: fictional prose is a sort of non-antagonic contradiction, and poetry, an antagonic contradiction. Wben semiologists try to analyze what they formerly called the narrative syntagm, fundamental doubts arise in their minds—and in that of the poet—for a narrative syntagm seems not to be a common syntagm at all, but rather a paradigm, an icon—or, at least, a paradigmatic syntagm. Semioticians analyze a plot as if it were a concept—but we are dealing here with a model. A model is a nonverbal concept: that is, the equivalent of a concept based on contiguity is a model based on similarity. Fictional prose, or a biography, is a model, an icon. Of what? Of life, the reader's life. One's life is seen and felt as an immediate icon, an immediate model (Peirce's phaneron , or "quality of feeling" ). This is why a novel, not to speak of a play or a movie, may claim our attention even when there are no special stylistic features observed in it. To read a novel is to compare models, to exchange models—not to compare and exchange 'ideas' or concepts.

It was Paul Valéry, it seems, who first called attention to the necessity of an Analogic —not mere analogy. To him, da Vinci was a nonverbal philosopher:

À peine notre pensée tend à s'approfondir, c'est à dire, à s'approcher de son objet, essayant d'opérer sur les choses mêmes (pour autant que son acte se fait choses), et non plus sur les signes quelconques qui excitent les idées superficielles des choses, à peine vivons-nous cette pensée, nous la sentons se séparer de tout langage conventionnel . . . penser profondément . . . est penser le plus loin possible de l'automatisme verbal. [1959:1263]

Philosophie systems are no more than written or scriptural systems, even when philosophy tries to assure itself against the "danger de paraître poursuivre un but purement verbal" (Valéry, 1959:1268). What Valéry says about the graph may be said of Peirce's interprétant —it is almost a definition:

Le graphique est capable du continu dont la parole est incapable; il l'emporte sur elle en evidence et précision. C'est elle, sans doute, qui lui commande d'exister, qui lui donne un sens, qui l'interprète; mais ce n'est plus par elle que l'acte de possession mentale est consommé. [Ad marginem , he notes, "et de plus une analogique ," and proceeds.] On voit se constituer peu à peu une sorte d'idéographie des relations figurées entre qualités et quantités, langage qui a pour grammaire un ensemble de conventions préliminaires (échelles, axes, réseaux, etc.); pour 10gique, la dépendence des figures ou des portions de figures, leurs propriétés de situation, etc. [Valéry, 1959:1266-67]

The paradox of Peirce's interprétant is that it is a meaning process, at the level of law and generalization and, at the same time, something like a moveable verbal and iconic supersign that is guiding the whole process, a dynamic model of signic relationships—an icon. As the meaning of a sign is another sign (cf. the dictionary), as one sign saturates into another of the same nature, so one code continually saturates into another, the verbal into the iconic and vice versa. Ultimately, thirdness saturates into firstness (cf. Valéry's graph). For Peirce, artists and seientists alike are creators of icons, and, in this light, we can understand Einsteins view:

I seldom think in words at all ... I work out the idea first . . . and then, much later, I try to explain it in words. [Metheny, 1968:15]

And this also helps to shake Benveniste's apparently matter-of-course idea that:

Une chose au moins est sûre: aucune sémiologie du son, de la couleur, de l'image ne se formulera en sons, en couleurs, en images. Toute sémiologie dun système non-lingüistique doit emprunter le truchement de la langue, ne peut donc exister que par et dans la sémiologie de la langue. [1969:130]

From this we may deduce that metalanguage (the interprétant) is always and necessarily verbal in nature. This is not so. Any translation from code into code implies a metalingual operation (metasignic seems to be a more precise term ). Any object—a work of art, for instance— will always resist (i.e., will always be different from) its description or analysis. The smile of Mona Lisa is metasignic in relation to critic and critics; semiotic is not "the study of relations between code and message," as Umberto Eco states (1968:94).

Semiotic is the study of the relations among sign systems; semiotic is always intersemiotic. Moreover, to what langue is Benveniste referring? The system of spoken utterances or that of written words? Or both? Here we have two different sign systems: a vocal, continuous, analogical, iconic sign is translated into a visual, discrete, digital, arbitrary sign. This is a semiotic fact, not a linguistic one. Phonological units may be translated as well—and even better—into iconic codes through phonogram plates or frequency apparatuses, such as a video-phonogram. Sign systems exist and are created for the sake of knowledge and information; each must have its own untranslatable 'bits' of information. Meaning results from the confrontation of sign systems—and difference is always implied in the process. Consequently, we are free to raise the following questions: Can a "sémiologie de la langue" exist without the written code? When we talk about 'red', are we not dealing first of all with the word red rather than with the visual phenomenon of color? Is not Benveniste's statement a 'normal' sample of verbal automatism and contiguity illusion? Is it possible for human beings to build a house with words? Can a bird do this? Was da Vinci an iconic thinker?

The contiguity illusion reigns everywhere among the works of semiologists. Let us illustrate this once more with some statements of Eugen Bär, from his "The Language of the Unconscious According to Lacan." Bär begins by assuming a 'scientific' posture, as always seems to happen in those soft sciences called the human sciences':

I shall make a distinction between what Lacan wants to convey and the way he says it. First, he purports to be scientific, but presents his theory in the form of a literary text, sacrificing clarity not for accuracy, but for aesthetic ambiguity. . . . For this reason, Ecrits should be read as a literary text and, for scientific purposes, be presented in a clearer form. Some such form is attempted in my article. [1971:243]

Here we see the traditional mistake—the dichotomy f orm I content, signifier/signified leading to the contiguity illusion that it is possible to translate form by translating 'content'. But the following lines suit our purposes better:

Lacan refers the reader to the works of R. Jakobson, who (1956:58) distinguishes two fundamental operations of speech: (1) to select certain linguistic units and to substitute one for another; and (2) to combine the selected linguistic units into units of higher complexity. The operation in (1) is based on the principle of similarity. For instance, substitutive reactions to the stimulus "hut" in a psychological test were the following: the tautology "hut"; the synonyms "cabin" and "hovel"; the antonym "palace" and the metaphors "den" and "burrow." The operation in (2) is based on contiguity. For instance, to the stimulus "hut," the combining operations result in metonymie appositions such as "burnt out," "is a poor little house," "thatch," "litter," or "poverty" (1973:24950). For Jakobson, and subsequently for Lacan, metaphor is an instance of the operation in (1); metonymy, an instance of the operation in (2).

First, we must ask:

why hut -----→ cabin, hovel, palace, etc.,

and not hut -----→ hat, hit, hot?

Secondly, Jakobson's discovery is based on the formal similarity of phonemes: would this not lead to paronomasia rather than to metaphor? Third, it is interesting to note that the same process seems to underline operations of predication, synonym, and metaphor—from the Western point of view, of course. Chinese and Japanese ideograms have no synonyms and no predication, only paronomastic metaphors , or better, paramorphic metaphors. It is not my intention here to discuss the consequences of the contiguity illusion on the study of Lacan's ideas by Bär. If, however, we compare the different renderings of the Freudian formula, Wo es war soll Ich werden,

we can see that Lacan is not only saying, but showing what and where unconscious language is, with all its reverberations and varying hues of certainty: Là où/ Là haut, c'était/ s'était, mon de/ voir. Vienna (vienne) is hinted at as a compliment to the master—so does poetry. One must approach it with the instruments of Analogic, not simply with those of Logic (from logos, word, verb—that is, contiguity sign, or symbol, in Peircean terms ). And when one says Analogic, icon is implied. (I agree, by the way, with Sebeok [1976:9-10] on the issue of the expression of nonverbal communication, in the sense that I see no use whatsoever in discussing it if it points to either body or gesture languages or to other iconic sign systems. I do not, however, quite agree with him when he talks about 'iconicity'; on the same ground, we would be entitled to talk about Verbalicity'. For the sake of semiotic, the distinction Verbal/ nonverbal' should be abolished once and for all; Verbal/iconic', or 'symbolic/iconic' should replace this dichotomy—with the verbal and the iconic being the two central sign systems.

Unconscious language is basically paralanguage, pre-verbal, iconic, paratactic, and paronomastic—a quasi-sign if we are talking about poetry (Pignatari, 1974:56-59). Hence Freud's special affection for parapraxes, as Bär puts it (1971:265) and as Freud himself has put it in so many instances. What is common to the languages of the arts, of children and primitive peoples, of schizophrenics and of the Unconscious, is the paratactic, iconic organization of signs. Even Lacan is too word'minded, but his poetic language is the only way to reach and try to apprehend the transverbal, or iconic. A psychoanalysis of Iconic Man (not only in works of art) should be possible. Perhaps it would also be a psychosynthesis of psychoanalysis.

Words, especially written words, are the most abstract instances of signs; poetry is the most concrete instance of words—and icons are the 'eastern' side of signs, as words are their 'western' counterpart. Recent discoveries in brain structure seem to point to the fact that this organ, too, is divided into a contiguity lobe (i.e., the left one) and a similarity lobe (the right one). It is high time that semiologists, critics, and pedagogues started making efforts to develop their right ones.

The contradiction of the 'scientific' attitude that is derived from the contiguity bias lies in the fact that its champions are inclined to consider puns, that is, paronomasia, almost as a nonscientific language. At the same time, they state that mathematics is the model for scientific thinking. Mathematical statements operate by actual similarity among signs: algebraic expressions are algebraic puns, iconic puns. Algebra is a seience of the eye, Gauss used to say. And a mathematician who is not also a poet will never be a creative mathematician, Poincaré would have added. And then Peirce would have rejoined:

a great distinguishing property of the icon is that by the direct observation of it other truths concerning its object can be discovered than those which suffice to determine its construction. Thus, by means of two photographs a map can be drawn etc. Given a conventional or other general sign of an object, to deduce any other truth than that which it explicitly signifies, it is necessary in all cases, to replace that sign by an icon. This capacity of revealing unexpected truth is precisely that wherein the utility of algebraical formulae consists, so that the iconic character is the prevailing one. [2.279]

The reasonings of mathematicians will be found to turn chiefly upon the use of likenesses, which are the very hinges of the gates of their science. The utility of likenesses to mathematicians consists in their suggesting in a very precise way, new aspects of supposed states of things. . .. [2.281]

For this reason, the icon is the sign of all possible lies as well as the sign of all discoveries—the heuristic sign, the open sign (4.531) par excellence:

They [the icons], one and all, partake of the most overt character of all lies and deceptions—their Overtness. Yet they have more to do with the living character of truth than have either Symbols or Indices.

I am too much indebted to Roman Jakobson to dare disagree with him. Nevertheless, I must dare—and may my errors and eventual nonerrors alike be as so many more homages to him. Peirce suggests that metaphors are linked to predication (7.590). Furthermore, he makes a distinction between a pure icon ("a possibility alone is an Icon purely by virtue of its quality; and its object can only be a Firstness") and iconic signs, or hypoicons , which are divided into three types: images, diagrams , and metaphors. In this connection, metaphors are thirdnesses among firstnesses: they refer mainly to symbols (i.e., words). In semiotic terms, we could define metaphor as a hypoicon by contiguity, meaning a similitude or parallelism between certain supposedly observed features of the referents of the signs ('semantic' similarity or likeness of the signified), as when we say "John is an eagle." But the case would be different if we were to say "Neagle is an eagle." Here it is as if an external similarity were internally embodied in signs : sound and image tend to mimic certain features supposedly existent in the referents. Paronomasia establishes syntactic similarity or likeness between signifier and signified. What we see here is the process of iconization of the verbal. Even if we take formal similarity as metaphorical (ultimately, resemblances among sounds and letters have little to do with resemblances among features of objects which are compared), even then the phenomenon is entirely different in quality, because it represents the passage from contiguity to similarity association, from symbol to icon— an icon of the process of similarity. Metaphors point to; paronomasia attempts to portray. The "de-wording" of a word does not occur by common metaphor alone: paronomasia (paramorphism) must be introduced. Moreover, paramorphism alone may assume the complete role, discarding metaphors entirely (hence poetic "music"). It follows, or so it seems to me, that on the verbal level, paronomasia (paramorphism on all levels) and not metaphor characterizes the paradigmatic axis.

Paronomasia disrupts discourse (hypotaxis), spatializes it (parataxis), and creates a non-linear, analogical-topological syntax. In a poem, horizontal paronomasia (alliteration, colliteration) creates melody; vertical paronomasia, harmony. Rhyme is the most common vertical paronomasia. Mallarmé's On coup de dés and the so-called concrete poems function on horizontal/vertical audio-visual paronomasias. Repetition of sounds is repetition of sounds in time, and this creates a spatial rhythmic grid—a diagram, a topological syntax. Rhythm is an icon. Sound-timing is rhythm, as is space-timing (e.g., in dance, cinema, or an assembly line) and space-spacing (in architecture or in a picture). Rhythm is thus a relational icon.

Summing up, sound similarity causes space similarity and correspondences—and here we have the basis for the iconic syntax that underlies poetry and such prose works as Tristram Shandy, Ulysses, and Finnegans Wake. Paronomasia is the bridge from the verbal to the iconic; for this same reason, it is said to be poetic', literary, and, chiefly, 'nonscientific'. Yes, most people believe that science is essentially verbal. They do not seem to realize that verbalized science is a translation of iconic science.

Thus, translating the famous Jakobsonian 'poetic function' from linguistics into semiotic, we have:

Verbal/properly—Peircely symbolic language acquires a so-called poetic function when an iconic system is infra-intra-superimposed on it.

Corollary: . . . when an analogical syntax is superimposed on a logical one.

NOTE: In certain iconic systems, like painting, phenomena seem to occur in reverse. For instance, 'metaphoric' surrealism is rather literary (by contiguity); metonymie' cubism is rather paronomastic (paramorphic).

SOME EXHIBITS/EVIDENCES/ HYPOTHESES

(a) Paratactic construction tends to destroy linearity:

John looks around, jumps, says good morning, cleans his collar, spins, stands still, starts running, waves at someone.

Here we see that verbs tend to "commonounize," framing interchangeable shots of equal value. The Brazilian linguist Myriam Lemle, in trying to establish a Basic Portuguese of some sort, has observed that parataxis prevails in the speech construction of the lower social classes.

(b) Sound organization creates space organization.

In Byron's "And where he gazed a gloom pervaded space," apparent linearity of sound turns into a nonlinear, spatial panorama—a spacesound landscape. Moreover, we cannot dismiss its calli-typographical dimension: here, we can 'hearsee' the eyes in gloom, as in Dante's

In Byron’s verse, gloom stands also for “eye(s),” as commonly happens in poetry, where we find a kind of ad hoc meaning. Poetry creates a semiotic ad hoc dictionary, with semiotic ad hoc synonyms: words do not mean what they are expected to because they are no longer words. They acquire iconic features—and I insist on the idea that sound features are iconic features, as we can observe in music, singing, noise, and speech.

(c) Freud's parapraxes have the same iconic structure, as we can see, among so many instances, in the classic deciphering, by Aristandros, of Alexander of Macedon's dream: Satyr = Sa Tyros = thine is Tyros (Freud, 1950:11). Until recently, and despite Lacan's efforts, analysts, essayists, and semiologists have been speaking of the "association of ideas." Strictly speaking in semiotic terms, there is no such thing, but only association of forms : the meaning of a sign is another sign, and this meaning is exerted by the interprétant, which, in turn, is iconic in nature and is a superor metasign that continually establishes diagrams of meaning, as I have stated above and elsewhere (1968:93; 1974:32-35; and with Lucrécia D'Alessio Ferrara, 1974:3/24). Summing up : you cannot have a thought (thirdness) without its form (firstness ).

At this point, I believe it is useful to remember that it is utterly impossible to understand Peirce's semiotic without taking into serious account what he called his ideoscopy —his overall trichotomic system. Besides, like Marx, he came from Hegel (not without objections) : "My philosophy resuscitates Hegel, though in a strange costume" (1.42). And I find also that it is not completely void of sense to say that Peirce did for language what Marx did for history. For this reason, it is difficult to accept Jakobsons suggestion that Peirce, ultimately, follows the old classic and Saussurean logic concerning the dichotomy signifier/signified (1975:8). As for secondness (reaction, real fact, antithesis, index, etc.), much work still needs to be done in the years ahead.

Form (another name for icon) is firstness, and its chief mode of organization is by co-ordination (parataxis). This does not mean that we do not have an iconic hierarchy; the difference is that the iconic hierarchy is established analogically, not logically. For instance, in architecture, hierarchy may be established, among other things, by likenesses and differences in size, volume, and quantity of elements; in photography, cinema, and television, by occupied space and by the whole range of distances and positions, from close-ups to panoramic shots. An analysis of a movie plot which does not take into consideration these iconic meanings is only an analysis of a verbalization. The same may be said of other sign systems, such as graphic and industrial design, music—or dreams:

To be language-conscious is to be aware of iconic organization. To be truly language-conscious is to be free from the contiguity illusion.

In closing, I have roughly laid out the correspondence table represented in Fig. 1.