Sign and Symbol in Picasso

Abstract

Writers on the semiology of Cubism have often cited Daniel-Henry Kahnweiler’s 1946–48 descriptions of Cubism as a form of writing. They seem, however, to have overlooked Pablo Picasso’s 1945–48 statements about art as a sign language. The first section of this essay argues that Kahnweiler was in fact inspired by Picasso’s statements. The second section retraces the origins of semiology in nineteenth-century philology, its revival by Claude Levi-Strauss, his influence on critical theory, the rise of a semiological interpretation of Cubism, and the problems with this interpretation. The third section links Picasso’s 1945–48 statements about art as a sign language to his contemporary visual work; specifically, to his illustrations for Pierre Reverdy’s book of poems Le Chant des morts. The idea of art as a sign language is traced to Picasso’s 1924 drawings of “star charts” or “constellations”. However, Picasso’s 1945–48 designs using a similar vocabulary are analyzed as signifiers without signifieds—that is, as symbols, rather than signs.

Certainly Picasso did not imitate Grebo art but its lesson encouraged him to completely overturn the plastic arts of the West, and, together with Braque (and, soon after, Gris), to renounce all imitation. These painters turned away from imitation, having discovered the true character of painting and sculpture, which are forms of writing. The products of these arts are signs, emblems, and not a mirror, more or less distorted, of the external world. This example, once understood, immediately liberated the plastic arts from the servitude of illusionistic styles. The Grebo masks provide a pure example of the “sign”. The human face “seen”, or rather “read”, does not coincide with the details of the sign. Furthermore, seen in isolation, these details would no longer convey any meaning. Notably, the volume of the “seen” face is not inscribed anywhere in the “actual” mask, which reproduces only the contour of this face. This volume appears somewhere in front of the actual mask. The epidermis of the seen face exists only in the consciousness of the viewer, who “imagines” it, creating the volume of the face in front of the plane surface of the mask, at the end of the cylindrical eyes, which thus become eyes seen as hollows.

—Daniel Henry Kahnweiler (1948), Les Sculptures de Picasso1

1. Introduction

Yve-Alain Bois’s paper “The Semiology of Cubism”, presented at the symposium accompanying the MoMA exhibition Picasso and Braque: Pioneering Cubism (Rubin 1989), remains a fundamental example of the use of semiology in art history. In it, Bois draws several lessons from Kahnweiler’s discussion of the Grebo mask owned by Picasso. First, that Cubist signs are “arbitrary”. Second, that they “are constituted by differences or oppositions”. Third, that their meaning is variable, depending on the context in which they appear (Bois 1992, pp. 173–74).2 Bois’s emphasis on arbitrariness, difference, and context reflects his decision to interpret the Cubism of late 1912 via the critical framework of Ferdinand de Saussure’s structural linguistics.

Elsewhere in “The Semiology of Cubism”, Bois expands his critical framework to include the semiotics of Charles Sanders Peirce and the concept of ostranenie (“making strange”) introduced by the Russian Formalist Victor Shklovsky (Bois 1992, pp. 177–78). At the same 1989 symposium, Rosalind Krauss presented a paper on “The Motivation of the Sign”, adding Roman Jakobson and Mikhail Bakhtin to the list of linguistic theorists whose ideas might assist in the understanding of Cubism (Krauss 1992, pp. 273–75).

Published in 1992, these dazzling essays by Krauss and Bois, jam-packed with brilliant readings of individual works, made semiology into the canonical interpretation of Cubism, displacing earlier formalist readings. It has retained its canonical status to the present day.

Curiously, neither Bois nor Krauss noted that Picasso himself spoke on several occasions about the role of the sign in art.

For instance, in a 1945 conversation with the Hungarian photographer Brassaï, Picasso stated that “An artist should observe nature but never confuse it with painting. It is only translatable into painting by signs”. (Brassaï 1966, p. 162) In the same year, he told art critic André Warnod that the seeming realism of French classical painting was merely a convention:

There’s no proof that this image of nature is any more true than the images created at other epochs. Really, it is just a matter of signs. It has been agreed that this sign represents a tree, this other one a house, a man, a woman—just as, in language, the word ‘man’ evokes the image of a man in our minds; the word ‘house,’ a house and so on in every language, although it’s a different word in every language. It’s a convention that’s been agreed on; we communicate by using these customary signs.

(Warnod 1945; in Bernadac and Michael 1998, p. 53)

These 1945 statements by Picasso raise an obvious question: to what extent are Kahnweiler’s later discussions of the role of the sign in art paraphrases of ideas that the artist expressed in their frequent conversations?

Kahnweiler did not describe Cubism as a kind of “writing” or use the word “sign” in his 1916 essay “The Road to Cubism”.3 He utilized these terms for the first time in his 1946 monograph on Juan Gris, which includes a lengthy digression on the emergence of Cubism in the work of Picasso and Braque. Here, Kahnweiler states that a “painting is a representation of thought by means of graphic signs”, adding that “not only painting, but also sculpture, is writing”. He also draws a distinction between the “graphic emblem”, which refers to “pictorial signs with a single meaning”, and the “symbol”, which refers to “signs with a dual meaning” (Kahnweiler 1946, pp. 64, 66, 71).4

Just as Picasso said in 1945 that “classical” French painting was just one kind of sign language, Kahnweiler argues in 1946 that “the most varied forms of record…have each in turn represented perfect ‘likenesses’ for contemporary spectators”. As examples, he cites the diverse styles of “the frescoes of Pompeii, Byzantine seventh-century mosaics, [and] the works of Rembrandt”. However, he adds, Napoleon’s creation of the École des Beaux Arts put a halt to the evolution of academic art. Henceforth, only the classical style was considered valid, with the result that “the graphic emblems petrified”. As a result, the “average spectator” could no longer “read the writing of his time”. In Kahnweiler’s opinion, this breakdown in reading competence is the reason for the resistance experienced by avant-garde styles from Impressionism to Cubism (Kahnweiler 1946, p. 79).

Two years later, in Les Sculptures de Picasso, Kahnweiler returned to the topic of signs in Picasso’s work. As Bois observes in “The Semiology of Cubism”, it emerges as part of Kahnweiler’s discussion of the Grebo mask owned by Picasso, cited as the epigraph to this essay. This was not the first time that Kahnweiler had discussed the Grebo mask. Thirty years earlier, in “The Road to Cubism”, first published in 1916 and revised in 1920, he noted that Braque and Picasso’s use of superimposed planes in their later Cubist work had been anticipated in dance masks made by the African artists of the Ivory Coast, explaining that:

These are constructed as follows: a completely flat plane forms the lower part of the face; to this is joined the high forehead, which is sometimes equally flat, sometimes bent slightly backward. While the nose is added as a simple strip of wood, two cylinders protrude about eight centimeters to form the eyes, and one slightly shorter hexahedron forms the mouth.

To this “tight formal scheme of plastic primeval force”, the African artists add a few “real details” such as the real details found in Cubist paintings of 1910–12. “The result in the mind of the spectator…is a human face”, floating in front of the geometric forms of the mask (Kahnweiler 1916; in Chipp 1968, p. 259).

In this early text, Kahnweiler saw the viewer’s perception of the mask as an example of the same cognitive process at work in the reading of Cubist paintings. From 1910 onward, Cubist pictures consisted of “a scheme of forms” resembling “an arrangement of planes, cylinders, quadrangles, etc.”. However, beginning in 1911, Braque and Picasso supplemented these abstract forms with what Kahnweiler called “real details”: an eye, a nose, the wooden scroll of a violin, etc. The “real detail” serves as a stimulus to memories of recognizable bodies and objects. “Combining the ‘real’ stimulus and the scheme of forms”, he wrote, “these images [Cubist paintings] construct the finished object in the [viewer’s] mind”. Kahnweiler’s early theory of Cubism was an amalgam of Lockean empiricism and nineteenth-century theories of perception (Karmel 2003, pp. 2–6).

Returning to the Grebo mask in his 1948 Les Sculptures de Picasso, Kahnweiler repeats the first part of his earlier description almost verbatim:

It should now be said that Picasso owned an African sculpture which completely escaped from servitude to the block, a servitude from which the Cubists wanted to liberate themselves. It was a Grebo mask. These Grebo masks represent the lower part of the face as a plane surface above which the tall forehead extends in a slight curve. From the place surface there project two cylinders, ten centimeters long and around five centimeters in diameter–the eyes–a triangular wedge–the nose–and a rectangular block–the mouth, these last two elements projecting less strongly.

(Kahnweiler 1948, p. 4)

What has changed between 1916 and 1948 is the interpretation placed on the mask. In 1948, the earlier empiricist reading was replaced by a new theory based on “signs”, “emblems”, and “forms of writing”. This new theory is laid out in the next paragraph of Les Sculptures de Picasso, cited at the beginning of this essay. In it, Kahnweiler takes Picasso’s 1945 remarks about signs and applies them to Cubism and the Grebo mask.

The fact that the semiological interpretation of Cubism is ex post facto does not make it invalid. It suggests, however, that it requires closer examination. I proposed a critique of it in my 2003 book Picasso and the Invention of Cubism. In the next section of this essay, I will give a summary of that critique.

Picasso’s 1945 exposition of a semiological theory of art also raises some very different questions. What prompted him to think about art and signs in 1945? Was the idea related, not to Cubism, but to his postwar work? And, if so, what does his postwar work tell us about semiology? I will address these questions in the third section of this essay.

2. Cubism and Semiology

The contemporary discipline of linguistics grew out of and superseded the nineteenth-century study of philology. Ferdinand de Saussure’s Course in General Linguistics, published posthumously in 1915 (de Saussure 1966), played a catalytic role in this transformation. However, Saussure did not arrive at his ideas ex nihilo; rather, they were products of the lively philological debates of the late nineteenth century (Karmel 2003, pp. 101, 112–13).

Philologists studied the evolution of language over time, decomposing modern words into “roots”, basic units of sound and meaning that could be traced back to ancient Sanskrit. These Sanskrit roots combined and recombined over centuries, giving birth to ancient languages such as Greek and Latin, and later to modern languages such as French, German, and English. In the course of this long evolution, the roots changed their pronunciation, but they did so according to regular patterns, so that the syllables of a modern word could be traced back to their remote origins.

F. Max Müller, an influential scholar of the historical school, argued that the meanings of modern words also derived from their Sanskrit roots. As he wrote in 1887:

The eight hundred roots which supply our dictionary can be reduced to about one hundred and twenty concepts. These one hundred and twenty concepts are really the rivers that feed the whole ocean of thought and speech. There is no thought that passes through our mind, or that has passed through the mind of the greatest poets and prophets of old, that cannot directly or indirectly be derived from one of these fundamental concepts.

(Müller 1888, pp. 32–33)

As a result, the worldview of European thought, however seemingly modern, reflected the social structure and occupations of ancient Aryan society.

Where historical philologists such as Müller believed that the meanings of roots endured throughout the process of combination and recombination, other scholars emphasized the ways that words’ meanings changed over time. Proposing that this evolution was guided by laws of “specialization” and “differentiation”, Michel Bréal (the inventor of the term “semantics”) argued that “the actual and present value of a word exercises such a power over the mind that it deprives us of all feeling for the etymological signification” (Bréal 1964, p. 176). Another French philologist, Arsène Darmesteter, observed that the names of objects typically derived from particular, even contingent, qualities. For instance, the Latin and French words for river, fluvius and fleuve, refer to the fact that a river flows; the English word “river” derives from Latin ripa, the bank of earth bordering a river. None of these names refers to water, the substance of which a river is composed (Darmesteter 1889, pp. 40–41). The leading American philologist William Dwight Whitney declared flatly that “every word handed down in every human language is an arbitrary and conventional sign” (Whitney 1979, p. 19).5

Saussure spent the 1880s teaching in Paris, where Bréal gave him his first teaching job, and his linguistic theories were shaped by the debates of the era. In his Course on General Linguistics, he states unequivocally that “Whitney quite justly insisted upon the arbitrary nature of signs; and by so doing, he placed linguistics on its true axis” (de Saussure 1966, p. 71). Similarly, Bréal’s “law of differentiation” seems to have inspired Saussure’s famous statement that “In language there are only differences” (de Saussure 1966, p. 120).

The heart of the Cours de linguistique générale is the section on “phonology”, where Saussure shows how the phonemes of any given language can be defined as a selection among different options for the production of sound. Every vocal sound requires an exhalation of breath. What fundamentally distinguishes one sound from another is the way that the breath it is modified by the mouth, larynx, and nose. The lips may be closed or open; the tip of the tongue may be placed against the teeth or the palate; the base of the tongue may be pressed against the top or side of the mouth. This fundamental sound, determined by the point of closure, may or may not be accompanied by the “voicing” of laryngeal vibration, by the “fricative” noise of air passing through a narrow space, or by the distinctive hum of sound routed through the nose. Every language uses a subset of these possible sound combinations. (Some languages add other forms of articulation, such as pitch). Within this combinatory logic, Saussure notes, the absence of a feature may be as significant as its presence. English speakers, for instance, distinguish between “d” and “t”. Both phonemes are produced by a sharp exhalation with the tip of the tongue pressed against the top of the mouth, just behind the teeth, but the former is voiced and the latter is unvoiced. In other languages, the distinction between “d” and “t” is inaudible; the two are considered the same phoneme (de Saussure 1966, pp. 38–64).

Saussure sums up his analysis by saying that “Every language forms its words on the basis of a system of sonorous elements…Phonemes are characterized not, as one might think, by their own positive quality but simply by the fact that they are distinct. Phonemes are above all else opposing, relative and negative entities” (de Saussure 1966, p. 119). The word “negative”, here, is justified by the existence of phonemes such as “t”, defined in part by its absence of voicing.

The section on phonology is the foundation of Saussure’s theory of language, and the part for which there is empirical evidence.6 However, the most influential part of Saussure’s theory is the following section on the nature of the linguistic sign. Here, he advances his famous division of the sign into two parts: signifier and signified, the first a word or “sound-image”, the second a concept. (He does not explain how the concept is linked to an object or objects in the real world.) Following Whitney, Saussure argues that the signifier—and therefore the sign as a whole—is “arbitrary” (de Saussure 1966, pp. 65–68). He slides momentarily back into the familiar nineteenth-century debate about how “signs”—that is, words—evolve, only to conclude that what is important is to study how language works at a given moment rather than how it changes over time. His argument here recapitulates the ideas of Michel Bréal, but Saussure introduces two useful terms—synchronic and diachronic—to distinguish the two approaches (de Saussure 1966, pp. 71–100).

Saussure’s presentation of “synchronic linguistics” is closely modeled on his phonology. Just as phonemes are “opposing, relative and negative entities”, so too “In language there are only differences. Even more important: a difference generally implies positive terms between which the difference is set up; but in language there are only differences without positive terms” (de Saussure 1966, p. 120)7. More concisely: “Language is characterized as a system based entirely on the opposition of its concrete units” (de Saussure 1966, p. 107).

In these oft-repeated statements, Saussure makes a leap from phonology to semantics, proposing that they function in the same way because they are both elements of “language”. However, he does not offer any evidence that the system of signs or “sememes” is organized around oppositions. He simply takes it for granted.

Saussure’s semiology inspired Russian Formalists such as Roman Jakobson and Victor Shklovsky, but they jettisoned his conceptual framework of arbitrariness and opposition in favor of ideas such as metaphor, metonymy, and ostranenie. In France, the influential linguist Joseph Vendryes (Vendryes 1921) continued the intellectual tradition of Whitney, Bréal, and Darmesteter, bypassing Saussure. As Christopher Green has argued, Vendryes may have influenced the criticism of Cubism in the 1920s; Kahnweiler cites him by name in his 1946 book on Gris. (Green 1992, pp. 83–87, cited in Karmel 2003, pp. 119–20).

Saussure’s Cours de linguistique générale was reprinted in France in 1931, but it seems not to have attracted attention outside the specialized field of linguistics until after World War II. The situation changed dramatically in 1955, when the anthropologist Claude Lévi-Strauss (1963) published an essay on “The Structural Study of Myth”, followed by a book on Structural Anthropology. Lévi-Strauss proposed that the significance of myth did not lie in its explicit narrative but in the repetitive “mythemes” of which it was composed. Like phonemes, mythemes were constituted by a series of differences. The true meaning of the myth was to be found, he argued, in the value system inherent in these differences.

To illustrate his new approach, Lévi-Strauss proposed an analysis of the story of Oedipus, together with several related myths, breaking them down into narrative units including:

“Oedipus marries his mother, Jocasta”

“Antigone buries her brother, Polynices, despite prohibition”

“Oedipus kills his father, Laios”

“Eteocles kills his brother, Polynices”

“Cadmos kills the dragon”

“Oedipus kills the Sphinx”

“Laios (Oedipus’ father) = left-sided (?)”

“Oedipus = swollen-foot (?)”

Having isolated these fragments, Lévi-Strauss experimented with different ways of arranging them, finally arriving at a grid with four columns. He placed Oedipus’ marriage to his mother and Antigone’s burial of her brother in the first column, which he characterized as containing examples of the “overrating of blood relations”. Oedipus’ murder of his father and Eteocles’ murder of his brother were placed in a second column demonstrating “underrating of blood relations”. A third column contained Cadmos’ killing of the dragon and Oedipus’ killing of the Sphinx, which both “refer to monsters being slain”. The names Laios and Oedipus, in the fourth column, both refer to “difficulties in walking straight and standing upright”. The items in the first and second columns demonstrated the over- or underestimation of family relations, two related themes. Lévi-Strauss linked the third and fourth columns by arguing that the killing of the monsters constituted a “denial of the autochthonous origin of man”, while the names indicating difficulty in walking signified “the persistence of the autochthonous origin of man”. The oppositions within the pairs of columns demonstrated how mythemes are structured around differences. Hence, Levi-Strauss’ insistence that this was a “structural” analysis comparable to Saussure’s analysis of language (Lévi-Strauss 1955; reprinted in Lévi-Strauss 1963, pp. 213–15).

Levi-Strauss’ analysis of myth offered a model that could seemingly be adapted to every area of the humanities. Within classical studies, it inspired a structuralist approach developed by Jean-Pierre Vernant, Marcel Detienne, and Pierre Vidal-Naquet. Further afield, the literary critic Roland Barthes, originally a champion of the “writing degree zero” of Alain Robbe-Grillet, applied it to the study of modern literature, summarizing the ideas of Saussure and his followers in a 1964 essay on the “Elements of Semiology”, followed by an ambitious “Introduction to the Structural Analysis of Narratives” (1966). In 1967, the philosopher Jacques Derrida, hitherto known as the author of a slim volume on Husserl, published a massive tome, On Grammatology, proposing a fundamental rethinking of Western metaphysics on the basis of Saussure’s linguistics. The psychoanalyst Jacques Lacan, previously associated with the Surrealists, began using Saussure to explicate Freud as early as the 1950s, and his approach became widely influential after the publication of his selected essays in 1966 and of his study The Four Fundamental Concepts of Psychoanalysis in 1973.8

Within art history, Pierre Dufour proposed the first semiological analysis of Cubism in 1969. A 1973 colloquium on Cubism at the Université de Saint-Étienne featured semiological interpretations by Jean Laude and Françoise Will-Levaillant. Pierre Daix invoked semiology in his introduction to the catalogue raisonné of Picasso’s Cubist work that he and Joan Rosselet published in 1979. In contrast, in the United States, MoMA’s massive Picasso retrospective of 1980–81 inspired numerous biographical interpretations of Picasso’s work. Influential critic Rosalind Krauss attacked this biographical approach in a pair of polemical articles; as a corrective, she proposed a semiological interpretation of Cubism, inspired by Daix but attempting to apply Saussure’s concepts with greater rigor. Krauss’s articles laid the foundation for Yve-Alain Bois’s (1990) essay “Kahnweiler’s Lesson” and for his (1992) paper on “The Semiology of Cubism”, as well as for her own paper on “The Motivation of the Sign” (Krauss 1992).9

Earlier formalist interpretations of Cubism were in effect diachronic, interpreting the style as a waystation on the road to pure abstraction. In contrast, the semiological approach is synchronic, reading Cubism as a self-sufficient language composed of geometric “signifiers” that are neither realistic nor abstract. They refer indirectly to the real world, much like elements in a written language. Conventional pictures of a wineglass, table, or violin resemble the objects they depict. In contrast, the words “wineglass”, “table”, or “violin” possess meaning only as part of a larger language; they signify without resembling. The semiological interpretation argues that Cubist signs, too, signify without resembling.

Unfortunately, the problems with Saussure’s semantic analysis, discussed above, are repeated and amplified in its later iterations. In Lévi-Strauss’ seminal analysis of the Oedipus myth, for instance, the Saussurean model functions as a kind of Procrustean bed, where the individual “mythemes” are stretched or trimmed to fit into the columnar grid.

Antigone’s burial of her brother Polynices epitomizes the tension between moral responsibility to family and moral responsibility to the state. To describe it as an example of the “overrating of blood relations” is to miss the point of Sophocles’ tragedy: that there can be an irreconcilable conflict between private and public morality. Nor is Oedipus’ marriage to his mother Jocasta an example of the “overrating of blood relations”. It is a transgression of a universal human taboo. Only by mischaracterizing these two events can Lévi-Strauss place them together in the first column of his grid.

Similarly, it is deeply misleading to characterize the events in the second column as examples of the “underrating of blood relations”. Oedipus does not know that Laios is his father when he kills him; indeed, the death is a result of Oedipus’ pious effort to avoid fulfilling the prophecy that he will kill his father and marry his mother. Similarly, Eteocles does not “underrate” his blood relation to Polynices: the two brothers are locked in a fatal conflict because, as brothers, they have an equal claim to the throne of Thebes.

The third column includes examples of “monsters being slain”. This seems straightforward. It is paired with the fourth column, which contains, not events, but etymological analyses of names. It is not news that “Oedipus” means “swollen foot”. (His feet are swollen because they were pierced and fastened together when he was a child.) However, the fact that his father’s name, “Laios”, is similar to the ancient Greek word for “left” seems to be an original observation.

At first glance, the items in columns three and four seem completely unrelated: it is not obvious what either of them have to do with “autochthonous origin of man”. On reflection, Cadmos and the dragon evoke this theme because, after killing the dragon, Cadmos extracts its teeth and sows them in the ground like seeds; they then spring up as men. Born from the earth rather than from human mothers, they are indeed autochthonous. However, Lévi-Strauss somehow reads this episode as a “denial of the autochthonous origin of men”. For no apparent reason, he lumps the death of the Sphinx into the same category.

Reminding the reader that Oedipus is lame and suggesting that his father (“Laios” = “Lefty”) also had problems walking, Lévi-Strauss comments that “in mythology it is a universal characteristic of men born from the Earth that at the moment they emerge from the depth they cannot walk or they walk clumsily”. This is not true in Greek mythology: the men born from the dragon’s teeth, in the legend of Cadmos, begin immediately not only to walk but also to fight. To support his interpretation, Lévi-Strauss instead invokes examples from the mythology of the Pueblo and the Kwakiutl, falling back into the Jungian or Frazerian idea that all mythologies, from all times and places, are actually one. Right or wrong, Levi-Strauss’ interpretation of “Oedipus” and “Laios” as symbols of autochthonous origins makes it clear that he needs to misread the legend of Cadmos as a denial of autochthony in order to create a Saussurean opposition between columns three and four of his model. Actually, if Lévi-Strauss’ interpretations are correct, the items in both columns all represent the same thing (Lévi-Strauss 1963, pp. 215–16).

Lévi-Strauss concludes that “the Oedipus myth…has to do with the inability, for a culture which holds the belief that mankind is autochthonous…to find a satisfactory transition between this theory and the knowledge that human beings are actually born from the union of man and woman” (Lévi-Strauss 1963, p. 216). Indeed, Lévi-Strauss sees myth in general as a device for reconciling a culture’s contradictory beliefs. This may well be true. But to make this argument, he feels compelled to squeeze his evidence into a Saussurean apparatus of columns and thematic oppositions. Saussure’s phonology is a genuinely scientific system, with clear laws subject to empirical verification. His semantic theory is a daring hypothesis. Lévi-Strauss’ structuralism is a mere pseudoscientific construct.

The same pseudoscientism afflicts the semiological interpretation of Cubism. A genuine Saussurean interpretation of Cubism would have to demonstrate that its graphic language was ordered around a series of oppositions, and that its signifiers were constituted, like phonemes, by different combinations of these opposing terms. Indeed, Daix, Krauss, and Bois point to some conspicuous formal oppositions found in Cubist pictures (above all, the papiers collés of fall 1912): specifically, oppositions between straight and curved lines10 and between “transparent” (blank) and “opaque” (colored or textured) elements. However, arranging these two oppositions into a grid yields a language with only four visual phonemes or signifiers. This is obviously inadequate to account for the richness and complexity of Cubist signification.

Perhaps the language of Cubism includes other oppositions not explicitly described by semiological critics, making it possible to define a longer series of visual phonemes? At the 1989 symposium, I asked Bois whether this might be the case. He replied that “trying to find the phoneme in the Cubist painting would be atrociously stupid” (Rubin et al. 1992, pp. 175, 177). If so, the Saussurean model does not seem appropriate.

Writing about this topic in 2003, I proposed an alternative semiology of Cubism, based on the ideas of Darmesteter rather than Saussure. As noted above, Darmesteter observed that, depending on the language in question, a current of water may be described as a “fleuve” (something that flows) or as a “river” (something with borders). The underlying material goes unmentioned, while the name evokes a particular quality of that material. Together, they form a composite signifier consisted of qualifier and qualified: one specific, the other general.

Cubist signs, from 1910 onward, are also generally composite in nature. There is a geometric form, typically a rectangular plane, which indicates the presence of something without specifying what that something is. On top of this, there is a “realistic” marker, typically simplified and cartoon-like, which imbues the geometric substrate with a particular identity. A nose or an eye transforms a plane into a face; a breast, into a woman’s body; the lapel of a jacket, into a man’s body; an f-shaped sound hole, into a violin. The marker serves as a qualifier, the geometric plane is the thing that is qualified. Together, they form a composite signifier indicating a recognizable object (the signified). (Karmel 2003, p. 122)

I thought, in 2003, that I was the first to propose the terms “qualifier” and “qualified”. Some years later, however, reading Douglas Cooper’s 1962 monograph about Nicolas des Staël, I came across the following passage:

De Staël was always concerned in his pictures with space, movement, light and those basic formal and tonal relationships which constitute the essence of a scene as perceived by the eye. But he knew how to transpose these into pictorial equivalents whose form and tonality immediately convey the elements of a natural scene without the artist being obliged to ‘qualify’ them with eye-wearying descriptive detail.

(Cooper 1962, pp. 6–7)

The leading Cubism scholar of his generation, Cooper was intimately familiar with its figurative language. As he implies, Cubist artists such as Picasso and Braque create abstract forms that are “equivalents” to reality but that nonetheless require additional, legible, details in order to “qualify” as recognizable figures or objects. Before his untimely death, De Staël was widely regarded as Braque’s artistic heir. In this passage, Cooper argues that De Staël goes beyond Cubism by making his abstract “equivalents” legible without the addition of figurative qualifiers. De Staël’s abstraction is a semiological system—but not a Saussurean one.

3. Sign and Symbol in Picasso’s Work of 1945–48

Picasso himself does not mention Cubism in his 1945 statements about art as a sign language. What, then, impelled him to address this topic? It seems likely that the stimulus was a 1945 commission from the publisher Tériade to illustrate Pierre Reverdy’s book Le Chant des morts (“The Song of the Dead”), a collection of poems responding to the nightmare of the war years. Picasso did not complete the commission until 1948, but he made several trial illustrations in November 1945.

Although the title of Arabesque (Figure 1) suggests a symmetrical, allover composition, the actual drawing is asymmetrical: a circle flanked at left by a curved “barbell” and ornamented at right by a pair of short lines extending in a tiny “V”.11 These were forms that Picasso had previously employed—or would shortly employ—in his figurative work. The barbell is a version of the dot-and-line motif found in a series of 1924 drawings discussed below. The circle with a “V” is clearly related to the schematic image of a faun’s head found in Picasso’s 1946 mural The Joy of Life (Musée Picasso, Antibes) and in many related studies. If Arabesque were a horizontal image, it might be read as a horned head atop broad shoulders. However, Picasso has rotated the configuration clockwise by 90° so that, like an arabesque, it functions as an abstract composition even if it contains figurative elements.12

The circumstances of the commission for the illustrations of Le Chant des morts deserve closer examination. Picasso and Reverdy had known each other since at least 1916, when they helped organize a banquet in honor of Guillaume Apollinaire, badly wounded during his military service. In 1917, Reverdy founded Nord-Sud, a short-lived but influential journal of poetry and art criticism whose title evoked the new metro line running between Montmartre, the home of the prewar avant-garde, and Montparnasse, the preferred neighborhood of the rising generation of artists and poets. Picasso’s work did not appear in Nord-Sud, whereas Reverdy reproduced Cubist compositions by Georges Braque and Fernand Léger, and published Braque’s influential “Pensées et réflexions sur la peinture” (Braque 1917).13 However, Reverdy and Picasso grew closer in the 1920s. Picasso provided three etchings for Reverdy’s 1922 volume of poems Cravates de Chanvre, and in 1924, Reverdy wrote the preface to Pablo Picasso (Reverdy 1924), a pictorial survey of the artist’s evolution from the Blue Period to his new classical style.14 During these years, both men frequented Parisian high society; Picasso, thanks to his engagement with the Ballets Russes; Reverdy, to his intimate friendship with fashion designer Coco Chanel. In 1926, however, both effectively withdrew from this social world; Picasso shifting his allegiance to the Surrealists, Reverdy going into spiritual retreat in the countryside near the Abbaye de Solesmes. During World War II, Picasso remained in Paris, a beacon of the French avant-garde under Nazi occupation, while Reverdy became an active member of the Resistance. After the liberation of Paris, Reverdy wrote Le Chant des morts, a series of laments for the spiritual damage of the war, couched in the Symbolist imagery of Emile Verhaeren and Maurice Maeterlinck.

The profound originality of Picasso’s illustrations for Le Chant des morts becomes more apparent when they are placed in the context of his earlier drawings and prints for books by friends or for reprints of classic texts. Often these illustrations were portraits of the author, as in his frontispieces for Guillaume Apollinaire’s Alcools (1913) and for Reverdy’s Cravates de chanvre. At other times he depicted figures or scenes related to the text, as in his etchings for Max Jacob’s Saint Matorel (1911) and for Ovid’s Metamorphoses (1931). Picasso’s most ambitious book project, prior to Le Chant des morts, was an illustrated edition of Honoré de Balzac’s story Le Chef-d’oeuvre inconnu published by Ambroise Vollard in 1931. The complex illustrative program of this book has been little studied, probably because the original edition is rare and precious, and the only reprint edition is a small-format edition with the text translated into German (de Balzac 1987).

Balzac’s 1831 story concerns an imaginary seventeenth-century painter named Frenhofer, who labors in secrecy for ten years on a painting of a female nude. When he shows the finished canvas to two colleagues, what they see is “confused masses of color and a multitude of fantastical lines”. The only recognizable element of the picture is “a bare foot emerging from the chaos of color”. Their disappointment drives Frenhofer to madness: he burns his paintings and dies that night (de Balzac 1831).15

A parable of the fate of the avant-garde artist, Balzac’s story appealed strongly to artists and writers of the Impressionist generation. Paul Cézanne is frequently quoted as having exclaimed, “Frenhofer, c’est moi!”16. For Picasso, Balzac’s reference to “a multitude of fantastical lines” must have seemed to anticipate the drawings of nudes and figures lost in a web of interlacing lines that appear in his sketchbooks of 1925–26 and in two major canvases of 1926, Painter and Model (Musée Picasso, Paris) and The Milliner’s Workshop (Centre Pompidou, Paris).17

Picasso decided on an unusually complex program of illustration for the book of Balzac’s story. He created a dozen new etchings, several of them showing an artist seated at an easel, facing a seated model, others of them apparently unrelated to the story. As a rule, scholars discussing the book reproduce one or more of these etchings. However, the text is also decorated with 63 woodcuts reproducing interlace drawings of figures, heads, and still lifes from Picasso’s 1925–26 sketchbooks. In this context, the interlace drawings seem like attempts to recreate the style of Frenhofer’s nude, although there is no evidence that Picasso had read Balzac’s story by 1925. The woodcut reproductions function as vignettes, placed seemingly at random throughout the story. Before Balzac’s text, Picasso added a purely visual section “by way of an introduction”, consisting of sixteen pages of woodcut reproductions of a series of distinctive dot-and-line drawings (Figure 2).18

These dot-and-line drawings, copied from a 1924 sketchbook, are of critical importance for Picasso’s illustrations to Le Chant des morts. Lattices of lines and curves with black dots at their intersections and termini, they seem at first glance to be abstractions, but on closer inspection reveal musical instruments or seated figures. The apparently random scatter of the dots, transformed by the lines into meaningful configurations, makes them resemble star charts representing the constellations, and they inspired similar constellation images by numerous later artists. Picasso’s star charts exemplify a Surrealist faith in creative interpretation. Strictly speaking, constellations do not exist. They are created by collective acts of imagination: picking out groups of stars, imagining lines drawn between them, and interpreting those lines as schematic pictures. In effect, the constellations demonstrate the mind’s power to free itself from appearances and to impose its dreams and desires on the world. Picasso’s star chart drawings seem to show this process at work, as if he had begun with an undifferentiated field of dots, then discovered guitars and figures within them and marked their contours with lines.19

The visual language of Picasso’s star charts does not fit neatly into the usual semiological classifications. The musical instruments in Three Guitars are drawn with rounded bodies and long, straight (or gently curved) necks. One can hardly say that these contours are “arbitrary signs”. They are more like what Charles Sanders Peirce calls “icons”—that is, signs based on resemblance.20 What makes this representation harder to see is the repetition of the schematic contours, nested within one another, and the addition of the dots. Together, these transformations draw the viewer’s attention away from the basic schema of each instrument, making it seem like a random patch of dots and lines. The original, legible schema is “motivated”, as Saussure would say, by its resemblance to an actual musical instrument. As Picasso multiplies its elements, the design comes to seem increasingly arbitrary. But this is not the foundational arbitrariness of the spoken or written signifier: it is a simulated arbitrariness, a post facto withdrawal of meaning.

Two decades later, Picasso recalled the formal vocabulary of the constellation drawings in the “barbell” at the left of Arabesque (Figure 1). Here, the motif of a curved line terminating in dots has been isolated, enlarged, and rendered in a freer style, using brush and lithographic wash instead of pen and ink. The hard-edged dots and lines become gestural blobs and streaks. The new meaning of this transformed motif can only be understood within the overall program of Picasso’s decorations for Le Chant des morts, which was dictated by the book’s unconventional printing.

Reverdy completed the manuscript of Le Chant des morts in January 1945 and sent it to Tériade (Stratis Eleftheriades), the publisher of the luxury art journal Verve. Enchanted with the poet’s rounded, casual, but legible script, Tériade decided not to have the poems set in type but instead to reproduce Reverdy’s handwriting. In his original manuscript, Reverdy presumably inscribed each page of text on the recto of a standard vertical sheet (or on the recto and verso). An exact reproduction would therefore have consisted of a series of vertical sheets. For the publication, however, Tériade decided to place two handwritten pages side-by-side on each printed sheet, simulating a page spread in a conventional book. Reverdy also wrote out a title page, a half-title, and table of contents identifying each poem by the first few words of its first line.21

The text begins with a long poem extending over fifteen pages, followed by a series of shorter texts. Reverdy quietly signals the beginning of a new poem by beginning the text a few lines down, leaving a small space at the top of the page. At the end of a poem, there is typically a larger empty space at the bottom of a page. In later, printed editions (such as Reverdy 2016), some of the poems acquire new titles, printed at the top of the page. In the original, handwritten edition, there are no titles or printed marks to indicate the beginning of a poem. At first glance the text seems to consist of a single book-length text.

The task of illustration thus presented an unusual challenge for Picasso. Because the layout of lines on each page was already fixed, he could not create spaces for vignettes by breaking up a long column of printed text, as he had done in Le chef d’oeuvre inconnu. Conversely, he seems to have recognized that the homogenous layout of the text—and the absence of titles—left him with the responsibility of marking the beginnings and endings of the individual poems.

Picasso’s partner, the painter Françoise Gilot, suggested a solution to these challenges. Recalling medieval manuscripts she had studied in a library in Montpellier, with “large ornamental initial letters in red, beautifully refined in their abstract form”, she described them to Picasso and “suggested he pattern his illustrations after them”. After making a handful of studies, including Arabesque (Figure 1), Picasso set the project aside. It was revived in 1948, when a rare-book dealer offered him a manuscript in which “the body of the text was written in the usual black Gothic script but the initials were very large ornamental letters painted in red and abstract in form”, showing “a definite Arabic influence”. Picasso bought the manuscript and soon thereafter returned to work on the illustrations for Le Chant des morts (Gilot and Lake 1964, p. 193).

The particular manuscript purchased by Picasso has not, as far as I know, been published. However, a review of comparable manuscripts suggests that he was influenced not only by the design of individual initials but also by the use of red to mark important names and dates, and by the way that ornamental bands and figurative marginalia often provide a framework enclosing the written text.

In the Rutland Psalter, for instance, each page of text is bordered on the left by a hard-edged “bar” ornamented with alternating rinceaux and capital letters (Figure 3). Elongated dragons extend from the top or bottom of the band, framing the text along that edge, while a variety of animals and human figures help fill the space at bottom (and sometimes also at top). The thick lettering forms a solid block. Where the words do not extend all the way to the right margin, the empty space is filled in by a “line-ender”, a colored band filled by a repeating decorative motif.

Similarly, in a typical page spread from the first, long poem in Le Chant des morts (“Il a la tête pleine d’or”), both text blocks are bordered at left by vertical red streaks, with a shorter red streak extending part way across the top of the text (Figure 4). The two streaks are joined by a red blob, recalling the barbell of the 1948 Arabesque. Where a short line of text leaves a space at right, Picasso fills it with a red blob and/or horizontal streak.

Later in the book, Picasso uses the barbell motif in multiple capacities, sometimes simply as a lateral border (Figure 5) and sometimes curving over the top of a text block to indicate the beginning of a new poem (Figure 6). The barbell here serves as a framing device; it would be a mistake to attribute a figurative meaning to it.

The designs that fill the empty spaces at the ends of individual poems also often incorporate the barbell motif. At the end of the short poem “Les fenêtres nues de l’exil” (“The bare windows of exile”), the space is filled by a horizontal barbell curving into the lower right corner of the page (Figure 7). At left, an additional half-barbell descends straight to the bottom; at right, a second half-barbell curves upward into the text area, emphasizing the closing line, “Vous comprenez que ce n’est rien” (“you understand it’s nothing”).

At the end of another short poem, “Je ne pense plus qu’à la nuit” (“The only thing I think of is the night”), a long barbell descends down the left margin of the text, then curves to the right to cradle the poem’s last words, “immobile au point mort/la ruche de lumière” (“immobile at the point of stillness, the beehive of light”). Another curved barbell connects the lower corners of the page. The two are joined by a pair of short vertical lines (Figure 8).

The last page of the long opening poem, “Il a la tête pleine d’or” (“He has a head full of gold”) displays only a few lines:

Miracles du sommeil	Miracles of sleep
Les mains liées dans les ornières	Hands linked in the furrows
Les pieds au ciel	Feet in the sky

Picasso emphasizes the left border with a long straight barbell (Figure 9), a more definitive version of the bordering streaks in the interior of the poem (Figure 4). The empty space below the text is occupied by a large circle with small “horns” at right. Two short lines connect the circle to the barbell, like those connecting the barbells at the end of “Je ne pense plus qu’à la nuit” (Figure 8).

Like the vignettes that break up the text in Le Chef d’oeuvre inconnu, these configurations at the ends of the poems are poised on the border between figuration and abstraction. In the earlier illustrations, the interlaced images of heads and figures related, somewhat distantly, to the theme of the text. What, if anything, do the configurations in Le Chant des morts represent, and what do they have to do with the text?

The composition at the end of “Il a la tête pleine d’or” (Figure 9) is unmistakably a version of Picasso’s Arabesque of 1945 (Figure 1). As discussed earlier, elements of this composition are linked to the Arcadian theme of a horned faun. However, Picasso has once again rotated the composition so that the circle no longer reads as a horned head.22 In any case, such Arcadian imagery would be completely foreign to Reverdy’s nightmare vision of life in wartime.

One might, crudely, read the intersecting barbells at the end of “Les fenêtres nues de l’exil” (Figure 7) as a schematic image of a dog with a raised head and a tail.23 But this cheerful, domestic, image would be antithetical to the closing lines of the poem:

Notes pures de l’agonie	Pure notes of agony
L’amour la folie le délire	Love madness delirium
Vous comprenez que c’est rien	You understand it’s nothing

Similarly, if the linked barbells at the end of “Je ne pense plus qu’à la nuit” are read as a creature with four legs, one extended, the image corresponds to nothing in the text. Picasso, a famously acute reader of poetry, would not have inserted such incongruous images into Reverdy’s text.

Picasso’s vignettes for Le Chant des morts resemble the ideographic characters of Chinese, assembled from originally figurative marks that generate new meanings when combined. But this resemblance is deceptive. Unlike his dot-and-line drawings of 1924, Picasso’s blob-and-streak designs of 1945–48 are not disguised versions of figurative images.

There is a paradox here: in precisely the period (1945–1948) when Picasso and Kahnweiler articulate a theory of art as a sign language (“It has been agreed that this sign represents a tree, this other one a house, a man, a woman”), Picasso invents a sign language that signifies nothing. More precisely, it is a language assembled from potentially figurative elements: the streaks suggest limbs, the blobs suggest appendages (such as hands or feet) or points of articulation (such as shoulders or hips). In effect, Picasso’s strategy is to foreclose representation while keeping open the latent possibility of signification.

Kahnweiler’s discussion of “ideographic writing”, in his 1946 book on Juan Gris, offers a clue to Picasso’s thinking. As noted earlier, he distinguishes between “graphic emblems”, which are “pictorial signs with a single meaning”, and “symbols”, which are “signs with a dual meaning”. The uneasy relationship between “sign” and “symbol” constitutes a long-standing problem within the theory of language. Umberto Eco observes that writers disagree as to which is the more inclusive term. Peirce, for instance, treats the “icon”, the “index”, and the “symbol” as different kinds of “signs”. For Peirce a “symbol” is an arbitrary term whose meaning is determined only by convention; it corresponds, in other words, to what Saussure means by “signifier”. Surveying the literature on this topic, Eco concludes that the “sign” generally possesses a fixed meaning within a clearly defined larger system. Conversely, a “symbol” suggests many possible meanings; it functions outside any single interpretive framework. Eco traces this idea to Goethe and the Romantics, for whom “symbols are polysemous, indefinitely interpretable; they realize the coincidence of the contraries; they express the unexpressible”, etc. Discussing the symbol in Gérard de Nerval, Eco concludes that “the content of the symbol is a nebula of possible interpretations…The symbol says that there is something that it could say, but this something cannot be definitely spelled out once and for all” (Eco 1984, chp. 4, “Symbol”, pp. 130–63, esp. pp. 136, 141–42, 159–61). This Romantic idea provided the point of departure for the Symbolist poetry of Mallarmé and his followers, whose key qualities have been described as: “deliberate ambiguity; hermeticism; [and] the feeling for the symbol as a catalyst (something which, while remaining itself unchanged, generates a reaction in the psyche)” (Lucie-Smith 1972, pp. 54–55). With Mallarmé, we come full circle, for his poetry exerted a significant influence on Picasso and the invention of Cubism.24

The formal language of Picasso’s illustrations for Le Chant des morts is different from that of Cubism, but both styles have a similar goal: to serve as a catalyst, inducing a mental state in which the viewer can enjoy an exceptional degree of pictorial ambiguity. In this altered state of visual perception, the process of decoding a picture becomes an end in itself. It does not matter whether or not the process arrives at a conclusion. The viewer may never succeed in translating the picture into a recognizable scene or figure. But neither does the picture abandon the idea of figuration, becoming an “abstraction”—that is, an arrangement of forms satisfying in itself, without reference to the real world. The viewer remains haunted by the sense that the picture is, in some undefinable sense, an image of the real world.

Later in 1948, Picasso deployed the sign language he had invented for Le Chant des morts in an important composition, The Kitchen, which he realized in two closely related canvases. Eight feet wide and almost six feet tall, these works are at the same scale as Jackson Pollock’s Number 1A, 1948 (MoMA, New York), and their “allover” arrangements of lines, curves, and dots seem almost equally abstract. In her memoir Life with Picasso, Françoise Gilot described the genesis of the composition:

La Cuisine…was based on the kitchen in the Rue des Grands-Augustins, where we sometimes ate our evening meal. The kitchen was painted white, and in addition to the usual equipment there were the birdcages. Aside from the birds the only touches of color were three Spanish plates on the wall. So, essentially, the kitchen was an empty white cube, with only the birds and the three Spanish plates to stand out from the whiteness. One night Pablo said, ‘I’m going to make a canvas out of that––that is, out of nothing.’ And that is exactly what he did. He put into it all the lines of force that build the space, and a few concentric circles that look like targets—the Spanish plates. In the background, vaguely, are the owl and the turtledoves.

(Gilot and Lake 1964, pp. 219–20)

In the version of the painting in the Musée Picasso, Paris, the decorated plates and the birds are recognizable, as are the leaves of a plant. However, the rest of the canvas is taken up with a complex arrangement of branching lines, with dots at their intersections and either dots or short lines (like serifs) at their termini. Presumably the lines are related in some way to the actual furnishings of a kitchen: stove, sink, and so forth. Indeed, in this version, light and dark patches in the background suggest the presence of a table at center, windows at left, and a door at right. In the version owned by The Museum of Modern Art, New York, the identifying features of the birds, plates, and leaves have been suppressed, and the background is a mottled, dirty white. As a result, the picture seems completely abstract.25

It is here, in his illustrations for Le Chant des morts and in the New York version of The Kitchen, that Picasso arrives at a pictorial equivalent to the arbitrary sign described by Saussure. Discarding the figurative image, or abstracting it to the point of invisibility, he severs the link between signifier and signified: the drawn or painted mark now functions exclusively in relation to the other marks around it. The result is that the sign ceases to be a sign, becoming, instead, a symbol.