Views and Overviews
Semiotics and Art History
Micke Bai and Norman Bryson
The basic tenet of semiotics, the theory of sign and sign-use, is antirealist. Human culture is made up of signs, each of which stands for something other than itself, and the people inhabiting culture busy themselves making sense of those signs. The core of semiotic theory is the definition of the factors involved in this permanent process of signmaking and interpreting and the development of conceptual tools that help us to grasp that process as it goes on in various arenas of cultural activity. Art is one such arena, and it seems obvious that semiotics has something to contribute to the study of art.'
From one point of view, it can be said that the semiotic perspective has long been present in art history: the work of Riegl and Panofsky can be shown to be congenial to the basic tenets of Peirce and Saussure,2 and key texts of Meyer Schapiro deal directly with issues in visual semiotics.3 But in the past two decades, semiotics has been engaged with a range of problems very different from those it began with, and the contemporary encounter between semiotics and art history involves new and distinct areas of debate: the polysemy of meaning; the problematics of authorship, context, and reception; the implications of the study of narrative for the study of images; the issue of sexual difference in relation to verbal and visual signs; and the claims to truth of interpretation. In all these areas, semiotics challenges the positivist view
1 We would like to thank Michael Ann Holly lor her very pertinent comments on this paper.
- See C. Hasenmueller, "Panofsky, Iconography, and Semiotics," journal of Aesthetics and Art Criticism, xxxvi, 1978,289-301; M. Iversen, "Style as Structure: Alois Riegl's Historiography," Art History, ll, 1979, 66-67; and MA. Holly, Panofsky and the Foundations oj Art History. Ithaca, 1984, 42-45. The semiotic nature of an apparently "natural" device like linear perspective is masterfully demonstrated in Hubert Damisch's seminal study, L'Originc dc la perspective, Paris, 1988.
1 See, e.g., M. Schapiro, "On Some Problems in the Semiotics of Visual Art: field and Vehicle in Image-Signs," Semiotka, I, 1969, 223-242.
'The clearest and most convincing overview of epistemological currents in the 19th and 2Uth centuries is Habermas's Erkenntnis und Intercssc of 1968 (Knowledge and Human Interests trans. J. Shapiro, London, 1972). Habermas's work has been challenged by psychoanalysts who believe that his idealized view of psychoanalytic practice as a constraint-free communication misunderstands their discipline. See, e.g., J. Rose, Sexuality in the Field of Vision, London, 1986. Habermas's oeuvre is also under pressure from the side of postmodern philosophy, most pertinently by J.-F. Lyotard, in, e.g., The Postmodern Condition, New York, 1980. These challenges do not, however, address Habermas's argument against positive knowledge, but his hope for a rational society. If anything, the authors are more skeptical than Habermas.
of knowledge, and it is this challenge that undoubtedly presents the most difficulties to the traditional practices of art history as a discipline.
Because of the theoretical skepticism of semiotics, the relationship between contemporary semiotics and art history is bound to be a delicate one. The debate between the critical rationalists and the members of the Frankfurt school, earlier on in this century, may have convinced most scholars of the need for a healthy dose of doubt in their claims to truth; nevertheless, much "applied science"—in other words, scholarship that, like art history, exists as a specialized discipline—seems to be reluctant to give up the hope of reaching positive knowledge. Whereas epistemology and the philosophy of science have developed sophisticated views of knowledge and truth in which there is little if any room for unambiguous "facts," causality, and proof, and in which interpretation has an acknowledged central position, art history seems hard pressed to renounce its positivistic basis, as if it feared to lose its scholarly status altogether in the bargain.1
Although art history as a whole cannot but be affected by the skepticism that has radically changed the discipline of history itself in the wake of the "linguistic turn," two fields within art history are particularly tenacious in their positivistic pursuit: the authentication of oeuvres— for example, those of Rembrandt, Van Gogh, and Hals, to name just a few recently and hotly debated cases—and social history.5 As for the former, the number of decisions that have an interpretive rather than a positive basis— mainly issues of style—have surprised the researchers themselves, and it is no wonder, therefore, that their conclusions remain open to debate." In section 2 ("Senders") we will pursue this question further. But, one might object, this interpretive status concerns cases where positive knowledge of the circumstances of the
" For the "linguistic" or, rather, rhetorical turn in history, see H. While, Metahistorv: The Historical Imagination in Ninetcodh-Century Europe, Baltimore, 1973, and especially, for a brief and convincing account of the fundamental rhetorical and semiotic nature of historiography, idem, "Interpretation in History," in Tropics of Discourse, Baltimore, 1978. The most detailed and incisive analysis of the rhetoric of historiography remains S. Bann's remarkable The Clothing of Clio, Cambridge and New York, 1984.
"See, e.g., the Rembrandt Research Project, in J. Bruvn, B. Haak, S.H. Levie el al, A Corpus of Rembrandt Paintings, The Hague, Boston, London, 1982,1987, 1989, review by L.J. Slatkes in the Art Bulletin, I \xi, 1989, 139-144.
SEMIOTKS AND ART HISI'ORY       1 75
making of an artwork is lacking, not because such knowledge is by definition unattainable. Attempts to approach the images of an age through an examination of the social and historical conditions out of which they emerged, in the endeavor of social history, are not affected by that lack.
The problem, here, lies in the term "context" itself. Precisely because it has the root "text" while its prefix distinguishes it from the latter, "context" seems comfortably out of reach of the pervasive need for interpretation that affects all texts. Yet this is an illusion. As Jonathan Culler has argued,
But the notion of context frequently oversimplifies rather than enriches the discussion, since the opposition between an act and its context seems to presume that the context is given and determines the meaning of the act. We know, of course, that things are not so simple: context is not given but produced; what belongs to a context is determined by interpretive strategies; contexts are just as much in need of elucidation as events; and the meaning of a context is determined by events. Yet whenever we use the term context we slip back into the simple model it proposes.7
Context, in other words, is a text itself, and it thus consists of signs that require interpretation. What we take to be positive knowledge is the product of interpretive choices. The art historian is always present in the construction she or he produces.8
In order to endorse the consequences of this insight, Culler proposes to speak not of context but of "framing": "Since the phenomena criticism deals with are signs, forms with socially constituted meanings, one might try to think not of context but of the framing of signs: how are signs constituted (framed) by various discursive practices, institutional arrangements, systems of value, semiotic mechanisms?'"'
This proposal does not mean to abandon the examination of "context" altogether, but to do justice to the interpretive status of the insights thus gained. Not only is this more truthful; it also advances the search for social history itself. For by examining the social factors that frame the signs, it is possible to analyze simultaneously the practices of the past and our own interaction with them, an interaction that is otherwise in danger of passing unnoticed. What art historians are bound to examine, whether they like it or not, is the work as effect and affect, not only as a neatly remote product of an age
Culler, xiv.
B Similar arguments within the social history of art, explicitly articulating art history with semiotics, have been put forward in a number of places by Keith Moxey. See "Interpreting Pieler Aertseii: The Problem of Hidden Symbolism," Nederlamis Kuitsthistorisch iaarboek, 1989, 42ff; "Pieter Bruegel and Popular Culture," The Compk'tc Prints of Pirter Bruegel the Elder, ed. D. Freedberg, Tokyo, 1989,42ff; "Semiotics and the Social History of Art," Acts of the 27th International Congress of the History of Art, Strasbourg, 1990 (in press).
" Culler, xiv.
long gone. The problem of context, central in modern art history, will be examined further from a semiotic perspective in section 1 here, and the particular problem of the reception of images, and of the original viewer, will come up in section 3 ("Receivers"), and again in section 8 ("History and the Status of Meaning").
In this article, we intend to conduct two inquiries simultaneously. On the one hand, we will examine how semiotics challenges some fundamental tenets and practices of art history. Although this is intrinsic to the article as a whole, it will receive greater emphasis in the first three sections. On the other hand and perhaps more important for many, we will demonstrate how semiotics can further the analyses that art historians pursue (this point will be centra] to sections 6 and 7). The parallel presentation of a critique and a useful set of tools conveys our view that art history is in need of, but also can afford, impulses from other directions. Since semiotics is fundamentally a transdisciplinary theory, it helps to avoid the bias of privileging language that so often accompanies attempts to make disciplines interact. In other words, rather than a linguistic turn, we will propose a semiotic turn for art history. Moreover, as the following sections will demonstrate, semiotics has been developed within many different fields, some of which are more relevant to art history than others. Our selection of topics is based on the expected fruitfulness for art history of particular developments, rather than on an attempt to be comprehensive, which would be futile and unpersuasive. This article does not present a survey of semiotic theory for an audience of art historians. For such an endeavor we refer the reader to Fernande Saint-Martin's recent study.10 Some of the specialized semioti-cians (e.g., Greimas, Sebeok) might see an intolerable distortion in our presentation. However, some of the theorists discussed here, like Derrida or Goodman, might not identify themselves as semioticians, nor might some of the art historians whose work we will put forward as examples of semiotic questioning of art and art history. In order to make this presentation more directly and widely useful, we have opted to treat semiotics as a perspective, raising a set of questions around and within the methodological concerns of art history itself.
The first three sections deal centrally with the semiotic critique of "context" as a term in art-historical discussion. From questions of context we move to the origins and history of semiotics, the ways in which these tools and critical perspectives have grown out of initial theoretical projects. The limits of space force us to consider just two early figures: Charles Sanders Peirce, the American philosopher (section 4), and the Swiss linguist Ferdinand de Saussure (section 5). In section 6 we present a semiotic view of psychoanalysis, demonstrating a variety of ways that psychoanalysis is bound up with semiotics and can be useful for art history, and then going on to discuss the most relevant concept, central in art history, that of the gaze. Psychoanalysis connects semiotics with an aware-
'" F. Saint-Martin, Semiotics of Visual Language, Bloornington, 1990.
176
the aut bulletin JUNE i'Wl volume lxxut NUMBER 2
ness of gender differentiation as pervasively relevant, indeed, as a crucial basis for the heterogeneous and polysemous nature of looking. Feminist cultural analysis has been quick to see the relevance of semiotic tools for its own goals. We wish to acknowledge that efficacy and we would have liked to demonstrate the inevitable "feminist turn" in semiotic theory itself by presenting the intersections between feminism's theorizing of gender, semiotics, and art history. But lack of space combined with the risk of overlap with an earlier survey article on feminism and art history published in this journal forced us, regretfully, to relegate feminism to the margins." Following the presentation of a psychoanalytically oriented semiotics, we go on to show the interpretive and descriptive, but also critical, value of a semiotically based narrative theory or narratology for the study of images— images that frequently have a narrative side that is not necessarily literary in background (section 7, "Narratology"). Instead of rehearsing the view of history painting as basically illustrative of old stories, a view that privileges language over visual representation, we demonstrate the specifically visual ways of story-telling that semiotics enables one to consider. Section 8 offers a few reflections on the status of meaning in relation to the historical considerations so important for art history.
One further question concerns the relation between the disciplines. Interdisciplinary research poses specific problems of methodology, which have to do with the status of the objects and the applicability of concepts designed to account for objects with a different status. Thus a concept mainly discussed in literary theory—for example, metaphor—is relevant to the analysis of visual art, and refusing to use it amounts to an unwarranted decision to take all images as literal expressions. But such use requires a thinking-through of the status of signs and meaning in visual art—for example, of the delimitation of discrete signs in a medium that is supposed to be given over to density.12 Rather than borrowing the concept of metaphor from literary theory, then, an art historian will take it out of its unwarranted confinement within that specific discipline and first examine the extent to which metaphor, as a phenomenon of transfer of meaning from one sign onto another, should be generalized. This is the case here, but not all concepts from literature lend themselves to such generalization. Rhythm and rhyme, for example, although often used apropos visual images, are more medium-specific and their use for images is therefore more obviously metaphorical.
Semiotics offers a theory and a set of analytic tools that are not bound to a particular object domain. Thus it liberates the analyst from the problem that transferring
"See the important article by T.G. Peterson and P. Mathews, "The Feminist Critique oř Art History," Art Bulldin, LXIX, 1987, 326ft'.
12 For the distinction between discrete and dense sign-systems, see N. Goodman, Languages of Art: An Approach to a Theory of Symboh, Indianapolis, 1976. This theory is much indebted to Wittgenstein. See A. Thiher, Words in Reflection: Modern Language Tiieory and Postmodern I id ion, Chicago, 1984.
concepts from one discipline into another entails. Recent attempts to connect verbal and visual arts, for example, tend to suffer from unreflected transfers, or they painstakingly translate the concepts of the one discipline into the other, inevitably importing a hierarchy between them. Semiotics, by virtue of its supradisciplinary status, can be brought to bear on objects pertaining to any sign-system. That semiotics has been primarily developed in conjunction with literary texts is perhaps largely a historical accident, whose consequences, while not unimportant, can be bracketed.13 As a supradisciplinary theory, semiotics lends itself to interdisciplinary analyses, for example, of word and image relations, which seek to avoid both the erection of hierarchies and the eclectic transferring of concepts.14 But the use of semiotics is not limited to interdisciplinarity. Its multidisciplinary reach— as journals like Semiotics demonstrate, it can be used in a variety of disciplines—has made semiotics an appropriate tool for monodisciplinary analysis as well. Considering images as signs, semiotics sheds a particular light on them, focusing on the production of meaning in society, but it is by no means necessary to semiotic analysis to exceed the domain of visual images.
1. Context
One area in which the semiotic perspective may be of particular service to art history is in the discussion of "context"1'—as in the phrase "art in context." Since semiotics, following the structuralist phase of its evolution, has examined the conceptual relations between "text" and "context" in detail, in order to ascertain the fundamental dynamics of socially operated signs, it is a held in which analysis of "context" as an idea may be particularly acute. Many aspects of that discussion have a direct bearing on "context" as a key term in art-historical discourse and method.1"
When a particular work of art is placed "in context," it is usually the case that a body of material is assembled and juxtaposed with the work in question in the hope
11 The intimate connection between semiotics and linguistics is a problem in Saussurean semiotics, which developed out of linguistics rather than the other way around, and not so much in Peircean semiotics, which came out of logic. See sections 4 and 5 here.
14 Examples of analyses of word and image interaction or comparison can be found in W. Steiner, The Colors of Rhetoric: Problem? in the Relation between Modern Literature and Art, Chicago, 1982, and Pictures of Romance: Form against Context in Painting and Literature, Chicago, 1988. See also the special issues of Poetics Today, x, 1 and 2, 1989, edited by Steiner. Also A. Kibedi Varga, "Stories Told by Pictures," in Style, xxll, 1980, 194-208, and "Criteria for Describing Word & Image Relations," in Poetics Today, x, 1989, 3Iff. For a critical examination of the hierarchies implied in many of these attempts, see W.J.T, Mitchell, konology: Image, Text, Ideology, Chicago, 1985, and M. Bal, "On Reading and Looking," in Sejniotica, lxxvi, 1989,283-320.
"The quotation marks around "context" ("text," "artwork," etc.) are meant to designate that at this place in our essay the word appears as an object of methodological reflection.
'"The points in this section and in section 3 ("Receivers") are worked out in more detail in N. Bryson, "Art in Context," in Studies in Historical Change, ed. R. Cohen, Charlottesville, Va., forthcoming.
SEMIOTICS AND ART HISTORV
177
that such contextual material will reveal the determinants that make the work of art what it is. Perhaps the first observation on this procedure, from a semiotic point of view, is a cautionary one: that it cannot be taken for granted that the evidence that makes up "context" is going to be any simpler or more legible than the visual text upon which such evidence is to operate. Our observation is directed in the first place against any assumption of opposition, or asymmetry, between "context" and "text," against the notion that here lies the work of art (the text), and over there is the context, ready to act upon the text to order its uncertainties, to transfer to the text its own certainties and determination. For it cannot be assumed that "context" has the status of a given or of a simple or natural ground upon which to base interpretation. The idea of "context," posited as platform or foundation, invites us to step back from the uncertainties of text. But once this step is taken, it is by no means clear why it may not be taken again; that is, "context" implies from its first moment a potential regression "without brakes."
Semiotics, at a particular moment in its evolution, was obliged to confront this problem head-on, and how it did so has in important ways shaped the history of its own development. We will discuss later the different conceptions of semiosis in Saussure and in the work of post-Saussureans such as Derrida and Lacan. Suffice it to say, for now, that in its "structuralist" era semiotics frequently operated on the assumption that the meanings of signs were determined by sets of internal oppositions and differences mapped out within a static system. In order to discover the meanings of the words in ä particular language, for example, the interpreter turned to the global set of rules (the langue) simultaneously governing the language as whole, outside and away from actual utterances (parole). The crucial move was to invoke and isolate the synchronic system, putting its diachronic aspects to one side. What was sought, in a word, was structure. The critique launched against this theoretical immobility of sign systems pointed out that a fundamental component of sign-systems had been deleted from the structuralist approach, namely the system's aspects of ongoing semiosis, of dynamism. The changeover from theorizing semiosis as the product of static and immobile systems, to thinking of semiosis as unfolding in time is indeed one of the points at which structuralist semiotics gave way to post-structuralism. Derrida, in particular, insisted that the meaning of any particular sign could not be located in a signified fixed by the internal operations of a synchronic system; rather, meaning arose exactly from the movement from one sign or signifier to the next, in a perpetuum mobile where there could be found neither a starting point for semiosis, nor a concluding moment in which semiosis terminated and the meanings of signs fully "arrived."17
J. Derrida, Dissemination, trans, with inrro. and additional notes by B. Johnson, Chicago, 1982. For a discussion of Derrida's theory of signification, see S. Melville, Philosophy beside Itself: On Deconstruct ion and
From this perspective, "context" appears to have strong resemblances to the Saussurean signified, at least in those forms of contextual analysis that posit context as the firm ground upon which to anchor commentaries on works of art. Against such a notion, post-structuralist semiotics argues that "context" is in fact unable to arrest the fundamental mobility of semiosis for the reason that it harbors exactly the same principle of interminability within itself. Culler provides a readily understood example of such nonterminability in his discussion of evidence in the courtroom.1* The context in a legal dispute is not a given of the case, but something that lawyers make, and thereby make their case; and the nature of evidence is such that there is always more of it, subject only to the external limits of the lawyers' own stamina, the court's patience, and the client's means. Art historians, too, confront this problem on a daily basis. Suppose that, in attempting to describe the contextual determinants that made a particular work of art the way it is, the art historian proposes a certain number of factors that together constitute its context. Yet it is always conceivable that this number could be added to, that the context can be augmented. Certainly there will be a cut-off point, determined by such factors as the reader's patience, the conventions followed by the community of art-historical interpreters, the constraints of publishing budgets, the cost of paper, etc. But these constraints will operate from an essentially external position with regard to the enumeration of contextual aspects. Each new factor that is added will, it may be hoped, help to bolster the description of context, making it more rounded and complete. But what is also revealed by such supplementation is exactly the uncurtailability of the list, the impossibility of its closure. "Context" can always be extended; it is subject to the same process of mobility that is at work in the semiosis of the text or artwork that "context" is supposed to delimit and control.
To avoid misunderstanding, one should remark that while the consideration that contexts may be indefinitely extended makes it impossible to establish "context" in the form of a totality—a compendium of all the circumstances that constitute a "given" context—semiotics does not in fact follow what may appear to be a consequence of this, that the concept of determination should somehow be given up. On the contrary, it is only the goal of totalizing contexts that is being questioned here, together with the accompanying tendency toward making
Modernism, Minneapolis, 1986. Umberto Eco, an important semiotician who draws upon Peirce but is also well versed in the Saussurean iradition, warns against a confusion between theoretical polysemy and actual interpretation, where limits are obviously in place. See his Rule o| the Render. Explorations in the Semiotics of Texts, 1979; Semiotics and the Philosophy of Language. 1984; and, more directly confronting deconstruc-tion, The Limits of Interpretation, 1990 (all three Bloomington). But the point is that these limits are socially and politically motivated, putting a practical stop to a theoretical polysemy. Thus the very thesis of polysemy provides clearer insight into the limits of interpretation and their motivations.
Culler, 139-152.
178        THE ART BULLETIN fUNE 1991 VOl.UMF I.XXIII NUMBER 2
a necessarily partial and incomplete formulation of context stand for the totality of contexts, by synecdoche. Certainly the aim of identifying the total context has at times featured prominently in linguistics (among other places). Austin's remark concerning speech act theory is a case in point: "The total speech act in the total speech situation is the only actual phenomenon which, in the last resort, we are engaged in elucidating."1" Semiotics' objection to such an enterprise focuses primarily on the idea of mastering a totality that is implicit here, together with the notion that such a totality is "actual," that is, that it can be known as a present experience. However, this by no means entails an abandoning of "context" and "determination" as working concepts of analysis. Rather, semiotics would argue that two principles must operate here simultaneously: "No meaning can be determined out of context, but no context permits saturation."2" Though the two principles may not sit easily together or interact in a classical or topologically familiar fashion, context as determinant is very much to the fore in semiotic analyses, and particularly those that are posfstruc-turalist.
As semioticians have tried to work through the complexities of the text/context distinction, they have developed a further caveat, concerning the stroke or bar (/) between the terms "text" and "context." That mark of separation presupposes that one can, in fact, separate the two, that they are truly independent terms. Yet there are many situations within art-historical discourse that, if we consider them in detail, may make it difficult to be sure that such independence can easily be assumed. The relation between "context" and "text" (or "artwork") that these terms often take for granted is that history stands prior to artifact; that context generates, produces, gives rise to text, in the same way that a cause gives rise to an effect. But it is sometimes the case that the sequence (from context to text) is actually inferred from its end-point, leading to the kind of metalepsis that Nietzsche called "chronological reversal."21 "Suppose one feels a pain. This causes one to look for a cause and spying, perhaps, a pin, one links and reverses the perceptual or phenomenal order, pain ... pin, to produce a causal sequence, pin .. . pain."" In this case, the pin as cause is located after the effect it has on us has been produced.
'■J.L. Austin, How To Do Thing* with Words, Cambridge, 1975, 148 (emphasis in the original). See also J. Searle, "Reiterating the Differences," Glyph, I, 1977,198-208; and Derrida, passim. -']. Derrida, "Living On: Border Lines," in Dt'coiidructimi mid Criticism, ed. H. Bloom et ai„ New York, 1979,81.
11 "The fragment of the outside world of which we become conscious comes after the effect that has been produced on us and is projected a IKistcrinri as its 'cause.' In the phenomenalism of the 'inner world' we invert the chronology of cause and effect. The basic fact of 'inner experience' is that the cause gets imagined after the effect has occurred." F. Nietzsche, Werke, ed. K. Schlechta, Munich, 1986, lit, 804; cited by I. Culler, On Deconstruction: Tlieory and Criticism after Structuralism, London, 1983, 86.
- Culler (as in n.21), 86.
Does one find comparable instances of such metalepsis or "chronological reversal" in art-historical analysis?
The answer may well be yes. Imagine a contemporary account of, say, mid-Victorian painting, one that aims to reconstruct the context for the paintings in terms of social and cultural history. The works themselves depict such social sites as racetracks, pubs, railway stations and train compartments, street scenes where well-to-do ladies pass by workmen digging the road, interiors in which domestic melodramas are played out, the stock exchange, the veterans' hospital, the church, the asylum. It would not be thought unusual for the art historian to work from the paintings out toward the history of these sites and milieux, in order to discover their historical specificity and determination, their detailed archival texture. Just this sort of inquiry is what, perhaps, the word "context" asks for; such reconstruction would be fitting and, one might say, indicated by the nature of the visual materials to hand.
But there are senses in which the procedure is still strange, despite its aura of familiarity. A primary difficulty is that those features of mid-Victorian Britain missing from mid-Victorian painting are rarely featured as part of the context that accounts for the works of art. A social history that set out, unassisted by pictures, to discover the social and historical conditions of mid-Victorian Britain might well attend to quite other milieux, different social sites, and indeed many other kinds of historical objects that do not readily lend themselves to pictorial representation. A harder social analysis might treat the pictures incidentally, in passing, as one sort of evidence among many. If one is going to study social history, why privilege works of art in such a way that the findings of historiography must be bound to the mise-en-scene of painting?
There are a number of observations that might be made at this point: for example, concerning the relations between art history and social history as disciplines both intertwined and impelled by different kinds of momentum, or concerning the role played by synecdoche in the rhetoric of art-historical discourses."3 The point that concerns us here, though, is that in the example chosen, the pictures have in some sense predicted the form of the historian's portrayal, that the work of art history is "anticipated by the structure of the objects it labors to illuminate."2' If that is so, then the "context" in which the work of art is placed is in fact being generated out of the work itself, by means of a rhetorical operation, a reversal, a metalepsis, that nonetheless purports to regard the work as having been produced by its context and not as
L1On synecdoche in historiography, see White, 1978 (as in n. 5) 51-80; on synecdoche as it functions within the rhetoric of art history, see Roskill, 3-35. See also D. Carrier's pertinent study of the rhetoric of art history and art criticism, Artwriting, Amherst, 1987,
i4 MA. Holly, "Past Looking," Critical Inquiry, xvt, 1990, 373. Holly's essay examines the general problem of "chronological reversal" in relation to the historiography of liurckhardt.
producing it. Moreover, in a further rhetorical maneuver, the work of art is now able to act as evidence that the context that is produced for it is the right one; the reversal can be made to produce a "verification effect" (the contextual account must be true: the paintings prove it).
In cases of this kind, elements of visual text migrate from text to context and back, but recognition of such circulation is prevented by the primary cut of text-stroke-context. The operation of the stroke consists in the creation of what, for semiotics, is a fantasmatic cleavage between text and context, followed by an equally uncanny drawing together of the two sides that had been separated. The stroke dividing "text" from "context" is the fundamental move here, which semiotic analysis would criticize as a rhetorical operation.25 From one point of view, as Derrida has argued, this cut is precisely the operation that establishes the aesthetic as a specific order of discourse. From another point of view, the cut (text/ context) is what creates a discourse of art-historical explanation; it is because the blade can so cleanly separate the two edges, of text and context, that one seems to be dealing with an order of explanation at all, with explanation on one side and explanandum on the other. To set up this separation of text from context, then, is a fundamental rhetorical move of self-construction in art history.
Semiotic inquiry has a further reservation about procedures of this kind; since it is concerned with the functioning of signs, it is particularly sensitive to the fact that in our example (a contextual account of mid-Victorian painting) the status of the paintings as works of the sign has in fact largely been effaced. This need not happen with all contextualizing accounts—and our example is, of course, only an imaginary case. What the example depends on is the idea of a number of contextual factors converging on the work (or works) of art. The factors proposed may be many; they may belong to all sorts of domains; but they all finally arrive at the artwork, conceived as singular and as the terminus of all the various causal lines or chains. The question to be answered was, "what factors made the work of art what it is?" And in order to answer such a question, it is appropriate and inevitable that some narrative of convergence will be produced. The question casts itself in just this convergent form: n number of factors, all leading toward and into their final point of destination, the work of art in question.
What semiotics would query here is the idea, the shape, of convergence. Certainly the model is appropriate if the object of the inquiry is assumed to be singular, complete in itself, autotelic. All the clues point toward
° The stroke is what Derrida critically describes as "the sans of the pure cut," a culling of the field that will be so sharp as to leave no traces of its own incision; a conceptual blade so acule that when the two sides of the cleavage are brought together the edges will perfectly rejoin; |. Derrida, The Truth in Painting, trans. G. Bennington and I. McLeod, Chicago, 1987,83-118.
SEMIOTICS AND AKI HISTORY 179
the one outcome, as in a work of detection. But the problem that is overlooked here is that insofar as works of art are works of the sign, their structure is not in fact singular, but iterative.'" Singular events occur at only one point in space and time: the guest at the country house party was murdered in the library; the Magna Carta was written in 1215; the painting was autographed and framed. But signs are by definition repeatable. They enter into a plurality of contexts; works of art are constituted by different viewers in different ways at different times and places. The production of signs entails a fundamental split between the enunciation and the enunciated: not only between the person, the subject of enunciation, and what is enunciated; but between the circumstances of enunciation and what is enunciated, which can never coincide."7 Once launched into the world, the work of art is subject to all of the vicissitudes of reception; as a work involving the sign, it encounters from the beginning the ineradicable fact of semiotic play. The idea of convergence, of causal chains moving toward the work of art, should, in the perspective of semiotics, be supplemented by another shape: that of lines of signification opening out from the work of art, in the permanent diffraction of reception.
It may be that scholars in certain other disciplines are more at ease than art historians with the possibility of a work of art that constitutively changes with different conditions of reception, as different viewers and generations of viewers bring to bear upon the artwork the discourses, visual and verbal, that construct their specta-torship. Admittedly, the openness of such a text or work of art can and has been appropriated and used in the name of a number of ideological exercises: the rehabilitation of the concept of the canon in literary criticism is one (the open text turning out to coincide with the shelf of masterworks, the rest remaining ephemeral and merely lisiblc); the cult of the reader as hedonistic consumer is another (a consumer who never reflects on the preconditions of consumption). But obviously the plurality attributed so selectively to the "classic" text (whether visual or verbal) is not excessive because it is a masterpiece. Rather the opposite: the openness of the classic is the result of that fundamental lack it shares with all texts, master-works or not. It is the consequence of the fact that the text or artwork cannot exist outside the circumstances in which the reader reads the text or the viewer views the image, and that the work of art cannot fix in advance the outcome of any of its encounters with contextual plurality. The idea of "context" as that which will, in a legislative sense, determine the contours of the work in question is therefore different from the idea of "context" that semiotics proposes: what the latter points to is, on
16 On "iteration," see Derrida, "Signature Event Context," in Limited Inc., 1-23.
"See P. de Man, Blindness and Insight, 2nd ed., ed. W. Godzich, Minneapolis, 1983.
'180      THE AK1 BULLE'llN IUNE W'Jl VOLUME LXXII1 NUMBER 2
the one hand, the unarrestable mobility of the signifies and on the other, the construction of the work of art within always specific contexts of viewing.
When "context" is located in a clearly demarcated moment in the past, it becomes possible to overlook "context" as the contextuality of the present, the current functioning of art-historical discourses, Such an outcome is something that semiotics is particularly concerned to question. It hardly needs remarking that the referent of "context" is (at least) dual: the context of the production of works of art and the context of their commentary. Semiotics, despite frequent misunderstandings of precisely this point (and especially of semiotic "play"), is averse neither to the idea of history nor to the idea of historical determination. It argues that meanings are always determined in specific sites in a historical and material world. Even though factors of determination necessarily elude the logic of totality, "determination" is recognized and indeed insisted upon. Similarly, in recommending that the present context be included within the analysis of "context," semiotics does not work to avoid the concept of historicity; rather, its reservations concern forms of historiography that would present themselves in an exclusively aoristic or constative mode, eliding the determinations of historiography as a performative discourse active in the present. The same historiographic scruple that requires us to draw a distinguishing line between "us" and the historical "them"—in order to see how they are different from us—should, in the semiotic view, by the same token urge us to see how "we" are different from "them" and to use "context" not as a legislative idea but as a means that helps "us" to locate ourselves instead of bracketing out our own positionali-ties from the accounts we make.
2. Senders
"Context," then, turns out to be something very different from a given of art-historical analysis. But no less problematic is the status of the concept of "artist"—painter, photographer, sculptor, and so forth. (To avoid some of the connotational baggage that comes with the label "artist," we use here the more neutral word "author.")28 It might seem at first that the idea of the author of a work of art is, again, a natural term in the order of explanation, and one that is now much more substantial and tangible that "context." As the idea of context is probed and tested, various disturbing vistas open up—-regressions,
mises-en-abyme,^ multiple or folded temporalities—but "author" seems much more stable. We may not be able in the end to point to a context, since in so many ways the context-idea involves lability and shifting grounds; but the author of a work of art is surely someone we can indeed point to, a living (or once living), flesh-and-blood personage with a palpable presence in the world, as solid and undeniable as any individual bearing a proper name, as reliably there as you or me.
Yet, as Foucault points out,3" the relation between an individual and his or her proper name is quite different from the relation that obtains between a proper name and the function of authorship. The name of an individual (as they say in Britain, J. Bloggs)'1 is a designation, not a description; it is arbitrary in the sense that it does not assign any particular characteristics to its bearer. But the name of an author (a painter, a sculptor, a photographer, etc.) oscillates between designation and description: when we speak of Homer, we do not designate a particular individual; we refer to the author of the Iliad or the Odyssey, of the body of texts performed by the rhapsodes at the Panathenaic Festivals, or we intend a whole range of qualities, "Homeric" qualities that can be applied to any number of cases (epics, epithets, heroes, types of diction, of poetic rhythm—the list is open-ended). "J. Bloggs" is in the world, but an "author" is in the works, in a body of artifacts and in the complex operations performed on them. Like "context," "authorship" is an elaborate work of framing, something we elaborately produce rather than something we simply find.
Some of the processes of this enframement can be seen at work in the strategies of attribution." Perhaps the first procedure in attribution is to secure clear evidence of the material traces of the author in the work, metonymic contiguities that move in a series from the author in the world, the flesh-and-blood J. Bloggs, into the artifact in question. The traces may be directly autographic— evidence of a particular hand at work in the artifact's shaping. Or they may be more indirect—perhaps documents pertaining to the work, or the physical traces of a milieu (as when an artifact is assigned to the category "Athenian, circa 700 B.C."). At this level, the most "scientific" stage of attribution, all sorts of technologies may provide assistance: X-rays, spectroscopic analysis, cryptography. What is assumed is that the category of authorship will be decided on the basis of material
:" Though the term "author" has some advantages over the term "artist" in this discussion, "author" has its own baggage of connotations. In some kinds of literary criticism, "author" is no less hagio-graphic than is "artist" in some kinds of art history; but we hope that the change of context here, from literary criticism to art history, will enable this range of meanings to he discarded. "Author" has the further disadvantage that, as a term brought into art-historical discussion from literary theory, it carries with it a connotation of "linguistic imperialism"—a name for the verbal artist being used for the visual artist. We are aware of this coloration, and we wish to state expressly that in our discussion the term "author" is meant to designate a function, or set of functions, not particularized by medium.
z" For this influential concept, see L. Dallenbach, it' Recit speculaire. Essni si(r la mise en abi/me, Paris, 1977 (The Mirror in the Text, trans. J. Whiteley and E. Hughes, Chicago, 1989).
111 M. Foucault, "What Is an Author?," in Language, Counter-Memory, Practice: Selected Essuys and Interviews, ed. D.F. Bouchard, Ithaca, 1977, 113-138.
!| In fact, what they say in Britain is likelier to be "Joe Bloggs"; for us, though, Bloggs can be a woman.
"Our description of attribution is not, of course, meant to be exhaustive.
SEMIOTICS AND Alii HISTORY 181
evidence, and what "author" names here is the work's physical origin. The techniques employed are essentially the same as those that would be used by a detective31 to establish whether J. Bloggs is guilty or innocent {whether the artwork is authentic or fake); and to this extent there is nothing as yet peculiar to art-historical discourse about the construction of authorship: the techniques are part of a general science of forensics. But attribution in art history involves further operations that lead away from science and technology into subtler, and more ideologically motivated, considerations concerning quality and stylistic standardization. Before, the "author" referred to a physical agent in the world, but now it refers to the putative creative subject. In the drastic changeover from scientific procedures built on measurement and experimental knowledge to the highly subjective and volatile appraisals of quality and stylistic uniformity, one already sees how multifarious are the principles that "authorship" brings into play. Not only are the principles diverse, which would make "authorship" an aggregated or multi-layered concept, but they are also contradictory—though the essentially unificatory drive of the concept of authorship as a whole will work to mask this, and to conceal the joins between conflicting elements from view.
If a certain measure of arbitrariness is already evident in the principles of quality and of stylistic standardization, a further and quite different range of the arbitrary is found in the procedures for "setting limits" to what counts within authorship. J. Bloggs, under the forensic principle, is the origin of all the physical traces that point to Bloggs's presence in the world, every one of them, however minute; forensics can consider all possible evidence, even the most unpromising. But "authorship" is an exclusionary concept. On one side, it works to circumscribe the artistic corpus, and on the other it works to circumscribe the archive. If the author were the physical agent J. Bloggs, we should have to count among Bloggs's authorized works every doodle, every jotted diagram, that Bloggs left in the world. Similarly, in defining the archive for Bloggs, we would have to admit into it the traces of every circumstance that Bloggs encountered in his life. As a concept, "authorship" turns out after all to entail the same regressions and mises-en-abyme involved in "context." And as it operates in practice, "authorship" manages these receding vistas through many variations on the theme of nonadmission.
Excluded from "authorship" are whole genres, and the decisions regarding such genres are historically variable to a degree. In our own time, graphic art occupies a mysteriously fluctuating zone between authorship (many graphics in magazines bear signatures) and anonymity (many others do not). Photography is similarly divided, with sometimes an expectation of authorship (for example, when photographs appear in museums, where
" On the relation between detectives and art historians (and psychoanalysts), see C. Ginzburg, "Morelli, I-'reud and Sherlock Holmes: Clues and Scientific Method," History Workshop, IX, 1980,5-36.
authorship operations are essential), and sometimes not (many photographs in daily newspapers). Among the forces that patrol these borders are those deriving from the economic matrix, since "authorship" in the modern sense has historically developed pari passu with the institution of property. Here the concept becomes a legal and monetary operation, closely bound up with the history of copyright law. And the forces must also include the protocols of writing and the rules governing what is to count as a correct mode of narration. For instance, a catalogue raisonne would be breaking those rules if it wandered into the realm of an author's doodles and napkin sketches, just as a biography of the author would be breaking them if it widened the aperture of relevance to the proportions of a Tristram Shandy. That such deviant narratives are rarely encountered is proof of the efficiency of the "authorship" operation, which is designed to prevent such aberrations. By a rule of correct narration or "emplotment," only those aspects of an author's innumerable wanderings through the world that may be harmonized with the corpus of works will count as relevant, and only a certain number of an author's traces will count as elements of the authorized corpus. The exclusionary moves are mutually supportive, and "correct" narration will set up further conventions, which vary from period to period, from Vasari to the present,'4 concerning exactly how much latitude may be permitted in describing the perimeters.
Authorship, then, is no more a natural ground of explanation than is context. To paraphrase Jonathan Culler, authorship is not given but produced; what counts as authorship is determined by interpretive strategies;3" and in the disparities among the plural forces that determine authorship are seen lines of fissure that put in question the very unity that the concept seeks, contradictions that the concept must (and does) work hard to overcome. Consider the following:
(B) physical agency (A) property "author"      (C) creative subject
(D) narration
Interdependent, these are various pressures that take different forms in different sites: in museums and auction houses, for example, (A) and (B) assume more centrality, and are subject to more exacting differentiation, than in departments of art history, where (C) and (D) may be more pressing than questions of monetary value or of forensics. In art history, modes of narration are of capital importance. And according to the view of many writers, from Barthes to Preziosi, the whole purpose of art-historical narration is to merge the authorized corpus
M On "emplotment," see White, 1978 (as in n. 5), 66-67; and Roskill, 7-10.
See Culler, xiv.
182       THE AKT BULLETIN lUNE WW VOLUME LXXIJI NUMBER 2
and its producer into a single entity, the totalized narrative of the-man-and-his-work, in which the rhetorical figure authur=corpus governs the narration down to its finest details.
What these writers find unacceptable is that such narratives are saturated with a romantic mythology of the full creative subject. Barthes writes: "The author is never more than the instance writing, just as / is nothing other than the instance saying I.. .. We know now that a text is not a line of words releasing a single 'theological' meaning (the 'message' of the Author-God) but a multidimensional space in which a variety of writings, none of them original, blend and clash."3" Preziosi writes:
The disciplinary apparatus works to validate a metaphysical recuperation of Being and a unity of intention or Voice. At base, this is a theophanic regime, manufactured in the same workshops that once crafted paradigms of the world as Artifact of a divine Artificer, all of whose Works reveal ... a set of traces oriented upon a(n immaterial) center. In an equivalent fashion, all the works of the artist canonized in this regime reveal traces of (that is, are signifiers with respect to) a homogeneous Selfhood that are proper(ty) to him.*'
The concept of "author" brings together a series of related unities that, though assumed as given, are precisely the products and goals of its discursive operations. First is the unity of the Work. Second is the unity of the Life. Third, out of the myriad of accidents and contingent circumstances, and the plurality of roles and subject positions that an individual occupies, the discourse of authorship constructs a coherent and unitary Subject. Fourth is the doubly reinforced unity that comes from the superimposition of Work upon Life upon Subject in the narrative genre of the lifc-and-work; for in that genre, everything the Subject experiences or makes will be found to signify his or her subjecthood. The mythology of this Subject is not only theophanic, it is also—as Griselda Pollock and others have shown—sexist: In a male-dominated art history "Women were not historically significant artists . .. because they did not have the innate nugget of genius (the phallus) which is the natural property of men."38
There can be little doubt that the discursive operations of authorship have been appropriated by ideologies with a heavy investment in the kind of Subject described here. In art history, and particularly through the formula of the monograph, the narrative genre of the man-and-his-work has exercised a hold over writing that is perhaps unparalleled in the humanities. To the extent that this
Vl R. Barlhes, "The Death of the Author," in Image—Music—Text, ed. and trans. S. Heath, New York, 1977,145-146.
17 Preziosi, 31.
!,G. Pollock, Vision and Difference: Femininity, Feminism and Histories of Art, London, 1988, 2.
has been the case, the author-function has enjoyed a hegemonic influence within the discipline, naturalizing a whole series of ideological constructs (among them, genius, genius as masculine, the subject as unitary, masculinity as unitary, the artwork as expressive, the authentic work as most valuable). But however much one may recognize the forcefulness of the critique of the author/Subject, it may now be just as critical to realize the strategic limitations operating upon it.
In those art-historical discourses that remain colored by romantic ideas of artistic creativity, the "death of the author" thesis may come as a shock, but in others it is old news indeed. Perhaps a rough and provisional distinction between "humanist" and "modernist" may be useful here. In humanist art-historical discourses, the author-function remains essentially sacramental. Doubtless it would be wrong to underestimate the persistence of this view, or to imagine that somehow progress had swept it away." Yet many modernist discourses in art history have defined themselves exactly by shedding this baggage of mythified authorship. Getting away from romantic suppositions concerning the creative, plenary subject has been the very badge of their modernity. Their goal has been not to perpetuate Michelangelism but to be rid of it; not to produce monographs of the author-genius, but to describe the limiting conditions that make the myth of genius impossible; to move, for example, from creators to patrons; from patrons to the competing interests within patronage; from popes to committees, and from committees to the raw documents that the archive supplies. Where the humanist discourses spoke of a full, creative subject, modernist discourses modernized art history precisely by emptying that subject, by evacuating its plenitude. In many ranges of its modern practice, the charge that art history is a theophanic regime accordingly falls rather flat; there the analogue might be not Lust for Life, but perhaps the nouvcau romaii.'11'
In modernist discourses, the author-function fulfils quite other purposes. It is not necessarily the central focus of analysis; it may even be incidental. Imagine a project of the following kind. Let us say that the art historian wishes to investigate the possible role played in a particular painting by an idea first stated away from art,
'"We cannot, for example, consign it to the 19th century- Rosalind Krauss's essay, "The Originality of the Avant-Garde: A Postmodern Repetition," in Art after Modernism: Rethinking Representation, ed. Li. Wallis, New York, 1984, has shown how central the discourse of authorial sell-origination has been to modernism. She writes: "The self as origin is the way an absolute distinction can be made between a present experienced lie novo and a fradifion-laden past. The claims of the avant-garde are precisely these claims to originality" (p. 18).
'"' In Preziosi's Rethinking Art History, the film Lust for Life of 1956, based on the life (or Ihe life and work) of Van Gogh, is presented as a distorting mirror in which art history may (or may not) recognize its own features. For detailed discussion of Preziosi's book, see the review by W. Davis in the Art Bulletin, lxxu, 1990, 156ft, and for a more-polemical one, M. Marmor in Art Documentation, 1990,233ff.
SEMIO I1CS AND ART HISTORY
183
in a philosophical treatise on perception." A humanist procedure here might be to posit a third term, such as the unity of the "period" or of the "culture," that would guarantee in advance that the philosophical treatise and the work of art are mutually related. But here the art historian is skeptical of such holistic moves and seeks more concrete relations. A certain ascesis is evident. There must, for example, be proof of actual contact between the treatise and the artist, whether directly (the artist read the treatise) or via "go-betweens" who relayed the ideas of the treatise to the artist. And it must be established that this particular idea, and not another, is present in the work of art; the idea must satisfy a condition of representability, which it will do if—to further specify the example—the philosophical idea concerns perception. As the analysis proceeds, visual properties of the work—its way of transcribing perception—are brought out; proof of contact is forthcoming; the place of go-betweens, relays between the treatise and the pictures, is clarified. Nowhere has there been any mention of genius; in fact there has been little mention of the artist beyond what has been necessary to the causal narration linking treatise, go-betweens, artist, and picture.
In a case like this, the "author" is obviously not part of the hagiography of the man-and-his-work; she or he satisfies quite different narrative needs. The governing figure is no longer that of synecdoche—the part standing for the whole—but that of metonymy,42 the linking together of contiguous events into a narrative continuum. The "author" is not an origin, but just one link in the chain. It is a link that cannot be dispensed with—the narration requires an agent to operate the chain that runs, treatise—go-between—author—work. But the agent required can be a "man without qualities." The only quality required is provable participation in the series of causation. The "author" is essentially transparent, like a window through which we look to see the causal factors that helped produce the work, like a funnel, essentially empty; like the neck of a funnel, penultimate. What the author-function enables is the closure of the chain; the
J' See M. Baxandall's extraordinarily scrupulous account of Chardin's relation to Lockean ideas of perception, in Patterns of Mention, 74-104.
42 Jakobson defined metonymy on the syntagma tic axis of language, the "horizontal" movement that links one word to the next in a sentence (or one link to Ihe next in a causal chain); he opposed metonymy to metaphor, a relation involving the "vertical" substitution of one element for another. A rough analogy for metonymy might be melody (as opposed to harmony) in music. This is the sense of metonymv in the present discussion. "Metonymy" in our discussion is used in Jakobson's sense, as a series of contiguous (syntagmatic) relations. See R. Jakobson, "Two Aspects of Language and Two Types of Aphasic Disturbances," in Selected Writings of Roman jakobson, The Hague, 1972, li, 239-259; P. Pettit, The Concept of Structuralists A Critical Analysis, London, 1975, 7-10; and R. Barthes. Elements of Semiology, trans. A. Lavers and C. Smith, New York, 1968, 58-61. For a clear comparison between various concepts of metaphor and metonymy, see K. Silverman, The Subject of Semiotics, Oxford, 1983. The rhetorical implications of metonymy for historiography are explored by White in "Interpretation in History," 1978 (as in n. 5), 51-80; the implications of metonymy in art-historical narration are explored in Roskil!, 31-34.
function is defined not by theophanies, in the archaic way, but by the "shape of convergence" that constitutes metonymic accounts. The idea of contextual factors that all eventually converge and terminate in the artwork is what produces the "author" here—an usher gathering in the various causal strands or chains, before the work. The author is needed not to open the work but to close it.
If the critique of the plenary subject is heard by many art historians as not referring to them, but as it were to their grandfathers, this may be because in certain ranges of disciplinary practice, art history's modes of narration have shifted away from the trope of synecdoche (of totality, of the great unities of period, style, and culture, art history's early dreams). What has replaced synecdoche is the trope of metonymy, of contiguous elements, a trope whose presence is the mark of modernization within art-historical discourses. The issue here is not to deny the viability of metonymy-based narration (or of any other kind of narration), only to see some of the consequences of the trope as it is used. The particular problem for metonymic narration stems from the logic of the figure itself, for metonymy is endless. Contiguities go back in time forever, one chain joining another, ramifying outward like the branches of a vine, a vine the size of the universe. And contiguities also travel forward in time, the artwork becoming contiguous with many contexts and discourses, touching them, traveling on with them, along plural paths that branch their different ways (how many ways we cannot know, for only the trope of totality will tell us that). To be narratable, metonymy requires two moments: inauguration and closure. Inauguration is the privilege, and also the essential function, of the narrator, the art historian: to open the discourse, to broach the subject, to start the metonymic chains. Closure is the privilege, as well as the essential function, of the "author"; through this figure the chains will be brought together in a movement of convergence and penultimacy, before "the end" finally appears on the screen.
The "death of the author" thesis in art history does not, then, apply evenly throughout art-historical practices. In regions where the "humanist" discourses of the full creator-subject have sway, the thesis may still have considerable moment as a critique of prevailing assumptions. In "modernist" discourses, where the "author" as creator-subject has already been evacuated, and where it exists more as a stripped-down narrative function, the thesis may be much less persuasive. One should remember that the space of a disciplinary field is not homogeneous; the same arguments mean different things in different places. Here the gendered organization of discourse becomes an important factor, for the "death of the author" thesis may not mean the same thing for men and for women.4'
When Barthes or Foucault first presented the "death of the author" thesis, it spelled release for many male critics
"See K. Silverman, The Acoustic Mirror: The Female Voice in Psychoanalyse and Cinema, Bloomington, 198S, 187-234.
184
THE ART BULLETIN JUNE [991 VOLUME LXX1II NUMBER 2
from the paternal powers of the master creators, the totemic fathers who, ruling the canon, also ruled criticism; for now the male critic could propose a different kind of activity, one that no longer assumed the heavy, guiding hand of those paternal ancestors.44 But for feminist critics, the same thesis could appear in another light. The demise of canonical authority might not be experienced as the same process by those who had never stood much chance of being included in the canon in the first place. The decade that saw the rise of the "death of the author" thesis, the 1970s, was the same decade in which the "first generation" of feminist art historians began demanding a place for women artists in the canon.45 It might seem that as soon as authorship for female artists was called for, the rules of the game changed so that "authorship" could appear as an archaic-concept, and the demand could be construed, by male critics, as "regressive." In a nonhomogeneous field, "progress" is not always to be measured by the same coordinates. To the male advocates of the "death of the author" thesis, the appropriations of Sherrie Levine, for example, might be taken as a critique of the Subject, where "Subject" has no connotations of gender; but to a feminist, Levine's refusal of authorship could be taken as "a refusal of the role of creator as 'father' of his work, of the paternal rights assigned to the author by law."4"
Similarly, the "death of the author" evident in "modernist" art history could be taken as "advanced" by those art historians who were working against a "humanist" background of the hagiography of the Author. But the procedures involved in "modernist" work can be taken in another sense if one's situation is different. For instance, it is part of the "scientific" quality of "modernist" causal narratives that the position of the analyst is not included in the narrative account. The narrator of the metonymic chains stands apart from those chains, "behind a glass wall";'17 the narrator occupies a point of knowledge outside the field, a point that is supposedly without qualities or investments.411 The narrator's nonimplication in the narrated sequence is guaran-
44 "The author is reputed the father and owner of his work: literary science therefore teaches respect for the manuscript and the author's declared intentions, while society asserts the legality of the relation of author to work (the 'droit d'miteur or 'copyright,' in fact of recent date since it was only really legalized at the time of the French Revolution). As for the Text, it reads without the inscription of the Father"; Barthes, "From Work to Text," in Image—Music—Text (as in n. 36), 160-161.
J'T.G. Peterson and P. Mathews propose the distinction between "first generation" and "second generation" feminist art historians in their article cited in n. 11 above.
"' R. Barthes, "From Work to Text," 160. Craig Owens applies Barthes's remark to the work of Sherrie Levine in "The Discourse of Others: Feminists and Postmodernism," in The Anti-Aesthetic: Essays on Postmodern Culture, ed. H. Foster, Port Townsend, Wash., 1983, 73.
*7 Preziosi, 35.
1,1 "The position of the analyst, in this panoptic regime, is a tacit space (that may be filled by similarly equipped or invested persons)—an ideally neutral Cartesian zero point ... divested of its own history, sociality, and conditions of investment and establishment. .."; Preziosi, 39.
teed by the fact that the causal chain ends pen ultimately with the "author" and finally with the artwork, in a clearly demarcated moment in the past. The metonymic narrative is third-person and aoristic, an overview from a place quite separate from artworks, authors, or causal chains. Against a background of "humanist" art-historical discourse, such separation might count as progress; but from another point of view, this apartness and this "objectivity" might be taken another way, as nonrecogni-tion and as disavowal of the creative authorial function of the art historian.
The moment of narrative closure, when all the metonymic chains draw to a convergent close, can also be read as a denial of the actual continuation of the contiguities in which the narrator stands. The movement of contiguities in fact passes on from the artwork into the art historian's own situation; the work of art is also contiguous with her or him. But the modernist discourses, which foreclose metonymic movement by getting the chain to end with the artwork, can work to deny this, making "contiguity then" eclipse and elide "contiguity now." The draining of hagiographic qualities from the "author" in the past can also be said to justify a corresponding emptying out of the qualities of positionality, motivation, and investment present in the author of the art-historical narrative, making the positionality of "modernist" art-historical discourse in general that much harder to think through.
The problem of the "author" is not, then, so different from the problem of "context." Semiotics assumes that not only artworks but the accounts we fashion for them are works of the sign; it especially attends to the governing tropes of art-historical rhetoric (where "rhetoric" docs not, of course, imply "ornament" or "embellishment," but names the fundamental conceptual shapes of art-historical accounts). From the viewpoint of semiotics, the modernist no less than the humanist discourses are constmcted in such a way as to prevent realization that when we confront works of art, we enter the field of the sign and of semiosis, of potentially infinite regressions and expansions; and that we deal with this situation by delimiting it from the place where we stand "now." In this process of concealing where we stand, the concept of "author" plays a crucial role, if as a result of its operation "author then" comes to mask—and to mask the masking of—"author now."
3. Receivers
Semiotics is centrally concerned with reception. As Barthes puts it, semiotic investigation "will not teach us what meaning must be definitively attributed to a work; it will not provide or even discover a meaning but will describe the logic according to which meanings are engendered."4" Semiotic analysis of visual art does not set out in the first place to produce interpretations of works of art, but rather to investigate how works of art are intelligible to those who view them, the processes by which viewers make sense of what they see. Standing
*' R. Barthes, Critique et verity Paris, 196b, 63.
SEMIOTICS AND ART HISTORY
185
somewhat to one side of the work of interpretation, semiotics has as its object to describe the conventions and conceptual operations that shape what viewers do— whether those viewers are art historians, art critics, or the crowd of spectators attending an exhibition. Modern art history has developed a number of ways to describe the role of the viewer and the "beholder's share." How do these accounts appear to the semiotics of reception?
It may be useful to begin by drawing a distinction between "ideal" and "empirical" spectators, since the distinction has been highly influential in directing art-historical discussions of reception. Empirical spectators are the actual, living, and breathing viewers of the sort we see in Rowlandson's cartoon Viewing at the Royal Academy (Fig. 1), walking through the exhibition space alone, in couples, or in groups, looking at the pictures around them and discussing what they see. The ideal spectator is a more abstract figure; broadly speaking, the term refers to the various roles ascribed to viewers by the paintings they see, the set of positions or functions proposed and assumed by each of the images on display. Within modern art history, the ideal spectator has been a continuing focus of interest, from Riegl's Dutch Group Portraiture,50 through the reception studies of Wolfgang Kemp/1 to the work that more than any other has brought the question of the spectator's modes of involvement with paintings to the forefront of art-historical discussion in English, Michael Fried's Absorption ami
Theatricality.^ Bringing empirical spectatorship into sharper focus has been the goal of a far more materialist analysis, which begins by investigating the actual traces left by actual encounters between viewers and works of art. As Nikos Hadjinicolaou puts it:
We must put forth another conception that sees the work of art as a relationship ... between an object and all the ways it has been perceived through history down to the present day; ways of perceiving that have untiringly transformed the work in a thousand and one ways. The work of art we have before us is the history of its consumption which has been determined 'each time' by the aesthetic ideologies of each present, these being in turn conditioned by the ideologies of contemporary social groups.55
From the point of view of the semiotics of reception, this project and its approach accord closely with the understanding of the concreteness, materiality, and sociality of sign-events among semioticians: "Communication and the forces of communication may not be divorced from the material base.""4 Nevertheless, the phrasing of the formulation here presents a number of difficulties. The clauses, "the history of consumption which has been determined ..." and "conditioned by social groups," involve a complex problematic, which we may be able to grasp if we attend to the word "groups."
511 A. Riegl, Z)as holländische Gruppenporträt (1902), ed. K.M. Swoboda, Vienna, 1931. M. Podro provides an illuminating analysis of Riegl's theory of spectatorship in The Critical Historians of Art, New Haven and London, 1982,81-95.
* W. Kemp, Der Anteil des Betrachters: Rczeptionsästheiische Studien zur Malerei des 19. Jahrhunderts, Munich, 1983. See also Kemp's article, "Death at Work: A Case Study on Constitutive Blanks in Nineteenth-Century Painting," Representations, \, Spring 1985, 102-123; and A. Neumeyer, Der Blick aus dem Bilde, Berlin, 1964.
5: M. Fried, Absorption and Theatricality. Painting and Beholder in the Age of Diderot, Berkeley, 1980.
53 N. Hadjinicolaou, "Art History and the History of the Appreciation of Works of Art," in Proceedings of the Caucus for Marxism and Art, no. 3—1, Jan. 1978,12f.
M V.N. Voloshinov, Marxism and the Philosophy of Language, New York, 1973, 21; see also R. Hodge and G. Kress, Social Semiotics, Ithaca, 1988, 37-78.
186       THE ART RUM FTIN JUNE IW VOLUME LXX11I NUMBER I
Let us suppose that analysis of reception discloses particular social groups, whose visual responses to particular works of art vary in semiotic terms, and that different groups (however these are defined) possess different codes for viewing even the same work. But the idea of the possession of codes of viewing cannot be taken for granted: if one is really going to address reception, it must be recognized that possession of codes of viewing is a process, not a given, and that members of groups acquire their familiarity with codes of viewing, and their ability to operate those codes, to varying degrees. Access to the codes is uneven: codes have to be learned and their distribution varies (and changes) within a group, even in those cases in which a group defines itself through its ability to manipulate visual codes in distinct ways. That is, even when attention to the conditions of reception discloses a particular group, which operates codes of viewing in a unique way, analysis of reception must still distinguish between degrees of access to those codes. If it does not do so, it is substituting an ideal case (hill possession of cultural skills, expertise, naturalness) for what is in fact an uneven process. The danger here is that the term "group" may function as an unacknowledged, and undetectable, synecdoche: in fact, members of the group have different levels of access to the group's codes and varying degrees of competence and expertise, but the condition of expertise is geiieralized to all of them.
Let us think, for example, of Rowlandson's cartoon of viewers at the Royal Academy. They are a motley crew: a gouty clergyman admires a bust; fashionable ladies point with their fans; some spectators bend over to quiz a nude, a man stands with his hands in his pockets looking lost. Or consider the crowds in contemporary engravings of the French Salon of the 1780s: men, women, and— surprisingly—children, gesticulating and craning to see, admiring or criticizing what is around them, talking, arguing, moving on. How much of this commotion has entered the historical record? Not much. What has entered it are highly specialized responses to exhibitions of this kind, literary productions that promoted or defended particular artists and schools and participated in various running debates about questions of taste. Such traces of criticism do not speak of viewing in a straightforward manner. Constrained by often highly restrictive literary conventions, which left little room to register the viewing response of their authors in any detail, they become expressive only when "enhanced" by reading between their lines and by carefully sounding their turns of phrase, their ellipses, to determine what by implication they may be giving voice to. If we meet with such obstacles within the public and recorded responses, and if we wish to develop further our concept of reception, what of all the other microscopic acts of viewing, each local and infinitesimal, which in their unseen trajectories failed to give rise to a discursive configuration that could survive? If such difficulties surround even the molar groups that reception analysis is able to identify, what of
all those other practices of looking, those swarms of viewers who left no trace of their ways and moments of seeing? There are many other viewers beside the ones who compose a treatise, publish pamphlets, or pen their memoirs. They are only a fraction, the smallest percentage, of actual reception. And how should we view this immense reserve?
Semiotics might answer this question as follows: we should remember that the reserve of unheard viewers is there, even when they cannot be retrieved; notice the absences in the record as much as what survives; and shift the terms of analysis from the actually documented viewers to the way the latter's discourses produce their own exclusivity. Take, for example, the absence of women. In the cartoon of the Royal Academy and in the engravings of the Salon, women are a visible and active presence. Yet, because women as a group lacked access to the machinery of the aesthetic treatise and of official taste, the archival traces of their spectatorship are slight. Furthermore, identification of "reception" with the archival traces left by the small fraction of male viewers who wrote their treatises, pamphlets, and memoirs is a synecdoche that can only reenact those exclusions that the archive has already performed. Indeed, it can even naturalize them further. A critical analysis of the codes in those documents, however, can avoid such repetitive-ncss. Such an analysis requires an extension of the archival project to a more comprehensive history of reception, one that would, ideally, uncover the hidden traces of other codes of viewing and add them to those we currently notice. As a canon has its exclusions, so has an archive: we need to look away from the obvious traces and the official records of reception, in order make the archive admit those whom it has set aside.
Semiotic analysis draws attention to the plurality and unpredictability at work in contexts of reception, in the forms of looking that have produced the discursive configurations evident in the archive. Surrounding those forms are other, submerged series of procedures that have addressed other needs, procedures whose traces can be discerned from the forcefulness of the attempts to repress them. Such series of procedures include codes of viewing that represent residual practices edged out by the rise of those later codes; the procedures also include codes that are hardly yet formed, emergent ways of seeing whose coherence has not yet been established and whose energies have not yet taken root. These codes are still tentative and faltering; they are configurations that still have to find each other and lock together to form a configuration that may emerge in the historical record.55
Within the fully configured codes of seeing, the ones that have made it into the arena of public taste and debate, are those existing in "debased" versions of the official protocols of viewing. On one side, there is the
55 See R. Williams's discussion of "residual" and "emergenl" cultures in "Base and Superstructure in Marxist Cultural Theory," Sell) Left teitw, Lxxxn, 1973,3-16.
SEMIOTICS AND ART HISTORY
187
practice of viewing works of art that seeks, but has not yet attained, confidence within a dominant visual discourse, that stammers and is not yet there, that does not fully grasp which responses do and which do not fall within the orbit of the sanctioned, and that yet seeks instruction and admission. On the other side are decaying versions of dominant practices of viewing, those that drift from the official model, through lapses of memory, disaffiliation, random variation, memory lapses, and so forth. As to the practices of resistance to dominant regimes of viewing, we know these are legion, and perhaps—comparatively—-they are better understood: they range from polite parody to outright defacement, from the clandestine inversion of existing rules of viewing to the invention of wholly new sets of rules, from subtle violations of propriety to blank refusal to play the game.56 In a separate category are complete idiolects of viewing, private languages of memory and habit that reorder the dominant codes into secret configurations of desire and identity, codes that may or may not be revealed to another human being, codes whose nature may or may not be consciously recognized.''7
Alongside the official records of reception, one must posit another world of looking, even before it can be specified in order to make it legible; against the "monotheism" of synecdoche, and its molar constructions, analysis has to assume the persistence of a "polytheism" of hidden and dispersed practices of looking at works of art, which while never giving rise to the consolidated forms of the review, the essay, the treatise, nevertheless constituted "reception" and "context" as historical realities.* One cannot know in advance what might enter into reconstructions of silent practice. What counts as much as what might eventually fill the space of the "reserve" is that such a space be conceptually created. It is not enough to escape the enclosure of synecdoche if the archive is simply extended toward as yet silent groups, for as each of these is identified and brought in, the logic of the "representative sample" is not yet challenged, only confirmed. In the same way that the extension of canons to include, for example, the work of women artists or artists of color nevertheless stays within the bounds of canonicity, so the discovery of previously unrecognized modes of visual practice stays within the logic of synecdoche unless the trope is deliberately confronted.
Among other tilings, this confrontation takes place where the study of reception confronts the dimension of
*On "local resistance" to dominant regimes, see M. de Certeau, The Practice of Everyday Life, Berkeley, 1984.
57On private "idiolects" of reception, see R, Barthes's discussion of the punctilio in Camera Lucitia: Reflections on Photography, trans. R. Howard, New York, 1981. V. Burgin provides an interesting discussion of Barthes's text in "Diderot, Barthes, Vertigo," in Formations of Fantasy, ed. V. Burgin etal, London, 1986, 85-108.
a On "monotheism" and "polytheism," see M. de Certeau, Heterologies: Discourse on the Oilier, Minneapolis, 1986,89; and The Practice of Everyday life (as in n. 56), 48.
scoliotic play. Until it does, in fact, reach that point, "reception" risks (in the critique of semiotics) being made over into a primarily repressive and legislative idea, like so much of the context-idea in which it plays a conspicuous role. Semiotics argues that it is only in concrete material circumstances that signs operate; but it also raises a number of questions concerning the tracing of these operations in reception-analysis. Reception of works of art is certainly an aspect of the general category of consumption; but the nature of consumption cannot be discerned by looking into the mirror of production.39 Production may appear to be a highly determinate process, with a distinct sequence of stages and a clearly demarcated outcome, the product, although that, too, is a simplification. But reception is a work with signs that opens onto the "polytheism" of hidden and dispersed practices that make up semiotic play. Even in the chamber and season of its first appearance (the Royal Academy, the Salon), the work of art enters networks of semiotic transformation as volatile and as tangled as the glances of a crowd in any given minute of its life. Nothing can stop the movement of signifier to signifier in a visual text as it is actually being viewed by empirical spectators. To do so one would have to be able to overcome something that cannot be eradicated from texts—whether visual, verbal, written, or the "general text"—that at the moment when a text is made, enunciation and enunciated cleave at the most fundamental level, and the visual text sets out on its numberless trajectories of seeing, none of which will exhaust the mobility of signification.
Is the argument of a visual semiotics, that attempts to retrace or reconstruct reception to be abandoned as a goal? No—what one seeks is to add a question and to shift the claim. For visual semiotics, the better question to ask would be: from where, from what position, is the reconstruction being made? Art-historical discussions of reception seem to move between two poles: the plural, dispersed, often submerged "polytheism" of actual, empirical reception; and the delimitations of a discourse on viewing that produces out of this plurality a cast of viewers whose responses are said to follow the most determinable contours. Out of the welter of concrete reception is distilled a character, "the viewer" whose attributes vary from one narration to another. But however this figure is defined, the viewer is essentially a character, a personification, in stories of viewing written in the first place according to the disciplinary norms of the narratives they work within.
A plain example will make this clear. How is it that female spectatorship has not been, from the beginning, a primary object of art-historical inquiry into visual reception? Can it be said that it was because the archive was lacking in evident traces of women viewers, that because incidentally the representative sample typically con-
50 See J. Baudrillard, The Mirror of Production, trans. M. Poster, St. Louis, 1981.
188        IHE AKT BULLETIN JU.VE 199] VOLUME LXXIII NTJMBEB 2
cerned only male viewing, that the latter was extrapolated from the archive and presented as the self-evident context of reception? But obviously this is an insufficient explanation. It suggests more innocence than is plausible: it is as if the fault lay in the archive, not in ourselves; had the traces been those of some other privileged group, we would have privileged them instead. It is only by seeing from within the present context that institutional forces within art history have worked generally to silence the whole question of the roles played by gender and sexuality in the field of vision, that art historians are able now to begin to see the ellipses and silences within the archive. The present context of reception has latterly enabled art historians to realize different modes of reception and to recognize that it is art history that establishes them. This should, rationally, lead to investigation of the institutional forces that identify reception with some groups rather than others, and to examination of the narrative pressures that impound the welter of empirical viewing into stories of clear delineation. Until this occurs, art historians are probably at the mercy of the narratives that seem, within disciplines, naturally to explain things—stories of "reception then" serving to mask "reception now," in the same disciplinary ecology or episteme that makes "context then" and "author then" eclipse and elide "context and author now."
Thus far we have been discussing a single set of issues concerning context, author, and reception, and we have attempted to outline the semiotic critique of these terms. But in order to understand how semiotics has reached a position from which to form that critique, one needs to know something of semiotics' own development. Restrictions of space oblige us to keep our discussion of the history of semiotics to a minimum; here we shall single out just two traditions within semiotics, the first stemming from Peirce, and the second from Saussure.""
4. Charles Sanders Peirce
The semiotic philosophy of Charles Sanders Peirce (1834— 1914) is a complex logical system, much of which is relevant only to specialists.61 A few of his concepts,
""Fur a full discussion of the contribution of Peirce and Saussure to visual semiotics, see M. Iversen's article, "Saussure v. Peirce: Models for a Semiotics of Visual Art," in Vie New Art History, ed. A.L. Rees and F. Borzello, Atlantic Highlands, 1988,82-94.
"' Indiana University Press continues to bring out a new critical edition of Peirce's complete works, Writings of Charles 5. Peirce: A Chronological Edition, 1982-. Two selections from the older edition {Collected Papers, ed. C. Hartshorne, P. Weiss, and A. W. Burks, Cambridge, 1935-66) contain important texts: Charles 5. Peirce: Selected Writings, ed. with intro. and notes P. Wiener, New York, 1966, and Philosophical Writings of Peirce, ed. with intro. J. Buchler, New York, 1955. The short fragment selected by R. lnnis ("Logic as Semiotic: The Theory of Signs") in his useful anthology of various semiotic approaches, Semiotics: An Anthology, Bloomington, 1985, is particularly representative. The Monist devoted an issue to Peirce, Lxv, 2, 1982. In the Peircean tradition stands the volume edited by T.A. Sebeok, A Perfusion of Signs, Bloomington, 1983. U. Eco's Theory of Semiotics, Bloomington, 1976, is perhaps the best-known revisionist Peircean theory. The literature on Peirce has
however, have had an important bearing on developments in semiotic criticism. Peirce's theory is characterized by a trichotomistic structure that accounts for the ways signs function. Peirce's view is primarily dynamic; it describes the process of signification, which is called "semiosis." This process involves the production and the interpretation of signs, both equally fundamental. Hence, this theory provides a logical basis for a reader- or reception-oriented theory of art.
The process of semiosis works through three positions: a perceptible or virtually perceptible item—the sign or representamen—that stands in for something else; the mental image, called the interpretant, that the recipient forms of the object; and the thing for which the sign stands—the object or referent. When one sees a painting, say a still-life of a fruit bowl, the image is, among other things, a sign or representamen of something else. The viewer shapes in her or his mind an image of that something with which she or he associates this image. That mental image, emphatically not the person shaping it, is the interpretant. This interpretant points to an object. The object is different for each viewer: it can be real fruit for one, other still-life paintings for another, a huge amount of money for a third, "seventeenth-century Dutch" for a fourth, and so on. The object for which the painting stands is therefore fundamentally subjective and reception-determined.
Peirce's famous and often misquoted definition of the sign runs as follows:
A sign, or representamen, is something which stands to somebody for something in some respect or capacity. It addresses somebody, that is, creates in the mind of that person an equivalent sign, or perhaps a more developed sign. That sign which it creates I call the interpretant of the first sign. The sign stands for something, its object. It stands for that object, not in all respects, but in reference to a sort of idea, which I have sometimes called the ground of the representamen.62
The structure of address of the sign has been taken up by speech-act theory; the "more developed sign" points at the complex acts of interpretation, e.g., in scholarly work; the "ground" can be seen as the basis on which the
grown so enormous that it makes no sense to begin enumerating publications here. See A Comprehensive Bibliography and Index of the Published Works of Charley Sanders Peirce, ed. Texas Research Institute for Studies in Pragmaticism, Greenwich, 1977, and C.J.W. Kloesel, "Bibliography of Charles Peirce 1976 through 1980," lite Monist, lxv, 2, 1982, 246-276. The Institute for Semiotic Studies at Indiana University, directed by T.A. Sebeok, is a helpful source of current information. An important article that is particularly relevant for art historians with an anthropological interest is M. Singer, "Signs of the Self: An Exploration in Semiotic Anthropology," American Anthropologist, lxxxii, 1980, 485-507.
"C.S. Peirce, "Logic as Semiotic: The Theory of Signs" in lnnis (as in n. 61), 5.
SEMIOTICS AND ART HISTORY 189
interpretation takes place and comes closest to the more common concept of code.
As for the process of interpretation that this definition implies, it is obvious that the interpretant is constantly shifting; no viewer will stop at the first association. As soon as the mental image takes shape, it becomes a new sign, which will yield a new interpretant, and we are in the middle of the process of infinite semiosis. None of the aspects of this process can be isolated from the others, which is the reason why this theory is incompatible with any dichotomistic theory of the sign, such as Saussure's pair signifier/signified. Peirce insists that the thing that becomes a sign only does so when it begins to evoke its interpretant: "A Sign is a Representamen with a mental interpretant." This view is consistent with standard aesthetic theories, e.g., the German phenomenological school of Rezeptionscisthetik** which stated that a work of art only becomes a work of art in its concretization by a recipient. The Prague school of semiotics adopted Mukarovski's distinction between artifact, the mere object, and aesthetic object, the work when processed as work of art by readers or viewers, which is another aesthetic account similar to Peirce's more general account of semiosis.'"*
The interpretant is a crucial concept in Peircean semiotics. As we said earlier, it should not be conflated with the person doing the interpretation. A good example of the representation of an interpretant is Vermeer's The Artist in His Studio (Vienna, Kunsthistorisches Museum). Because Vermeer cannot have seen this scene while painting it, we must conclude that he imagined it, as an outsider would see it, for example. "The scene was nowhere but in his head."6" Hence, the scene of the painting artist presented from the back is in the first place and emphatically a sign or rather, a sign of its own sign-ness.
Peirce elaborates this basic trichotomy of the semiotic process in at times exasperating detail. A primary division of the field of semiotic inquiry is based on the relations between the elements of semiosis. The relation
™ A representative collection in this vein is W. Kemp's Dcr Betrachter isl im Bihi, Munich, 1988.
M In spite of the different epistemological backgrounds, there is a strong similarity in this respect between Peirce's and Lacan's concepts of the sign. Both concepts entail the notion that the fixation of the meaning of a sign is endlessly deferred, as Derrida also holds. The major difference seems to be that Lacan insists on the social formation of the practice of interpretation (his concept of the symbolic order). But as de Lauretis rightly argues, Peirce's interpretant, although presented as a mental image and therefore carrying the burden of mentalism, can be redefined as radically social in origin. For the Peircean ground without which no interpretant can occur, is, precisely, a common ground. See T. de Lauretis, "Semiotics and Experience" in Alice Doesn't: Feminism, Semiotics, Cinema, Bloomington, 1983, 158-186. De Lauretis presents an integration of Peirce's and Kristeva's semiotic theories, the former mediated through Umberto Eco's rethinking of Peirce in A Theory of Semiotics, Bloomington, 1976.
"5B. Rotman, Signifying Nothing: The Semiotics of Zero, New York, 1988, 35.
between the sign and the ground lead to grammar, whose most commonly studied aspect is syntax. The relation between sign and object leads to questions of meaning or semantics. The relation between sign and interpretant can be linked to questions of rhetoric as part of pragmatics by virtue of the idea that one thought brings forth another. This division into three fields of inquiry is more common in linguistics than in art criticism, but certainly deserves consideration. Pragmatics would be the dimension where the affective efficacy of a work is examined; semantics includes any hypothesis about the meaning of a work, e.g., iconography; syntactics studies the relations between elements of the image to codes or ways of meaning-production.
Although many of Peirce's elaborate typologies of signs derived from this basic theory have not been commonly taken up by art critics, the most famous of these, icon, index, symbol, deserves further investigation. We quote Peirce's own definitions, because this typology is frequently misunderstood.
An icon is a sign which would possess the character which renders it significant, even though its object had no existence; such as a lead-pencil streak as representing a geometric line. An index is a sign which would, at once, lose the character which makes it a sign if its object were removed, but would not lose that character if there were no interpretant. Such, for example, is a piece of mould with a bullet-hole in it as a sign of a shot; for without the shot there would have been no hole; but there is a hole there, whether anybody has the sense to attribute it to a shot or not. A symbol is a sign which would lose the character which renders it a sign if there were no interpretant. Such is any utterance of speech which signifies what it does only by virtue of its being understood to have that signification.'*
First of all, any identification of icon and the entire domain of the visual is wrong.67 As Peirce clearly states, the iconic is a quality of the sign in relation to its object; it is best seen as a sign capable of evoking nonexistent objects because it proposes to imagine an object similar to the sign itself. Iconicity is in the first place a mode of reading, based on a hypothetical similarity between sign and object. Thus, when we see a portrait by Frans Hals,
ts "Logic as Semiotic" (as in n. 61), 9-10.
"7 This is the error of Louis Marin's influential paper, "Toward a Theory of Reading in the Visual Arts: Poussin's The Arcadian Shepherds," most recently reprinted in Calligram, ed. N. Bryson, Cambridge and New-York, 1988, f>3ff. See also Marin's "The Iconic Text and the Theory of Enunciation: Luca Signorelli at Loreto (Circa 1479-1484)," in New Literary History, xiv, 1983, 553-596. For Marin, the "iconic text" is the visual text. Thus the sign-status of the icon is obscured, whereas the importance of the other two signs in visual art, the index and the symbol, is underestimated. And an ontological distinction between verbal texts as symbolic and visual texts as merely "natural" inevitably returns.
190       THE ART BULLETIN lUN't I'M VOLUME LXXEI NtrMBT.R 2
we imagine a person looking like the image, and we don't doubt the existence, in the time of Hals, of such a person; we don't demand substantiation of that existence by other sources. Similarly, we think we know the face of a self-portraitist, say, Rembrandt, even though other painters have presented a face of Rembrandt quite different from his self-portrait, just because we adopt the iconic way of reading when we look at Rembrandt self-portraits.1*
But the example of portraits might wrongly suggest that the icon is predicated upon the degree of "realism" of the image. An abstract element like a triangular composition can become an iconic sign whenever we take it as a ground to interpret the image in relation to it, dividing the represented space into three interrelated areas (Leo Steinberg, for example, makes this division in his paper on Las Meniims).M Instead of visuality in general, or realism for that matter, the decision to suppose that the image refers to something on the basis of likeness is the iconic act, and a sense of specularity is its result. A romantic sound of violins accompanying a romantic love scene in a film is as iconic as the graphic representation of Apollinaire's poem about rain in the shape of rain.
The concept of index has been brought to bear on visual art primarily by Rosalind Krauss.'" Peirce's description of the index emphasizes its symmetrical opposition to the icon: while the icon does not need the object to exist, the index functions precisely on the ground of that existence. His example suggests that real, existential contiguity between indexical sign and object (or meaning) is indispensible. But that existence need not be confined to "reality"; the indexical sign and its meaning can entertain such a contiguous relationship within the image itself. The many recent publications on the gaze and the look that take the represented look of the figures in the painting as their starting point, for example, implicitly state that there is an indexical relationship between the look and what is looked at. The represented voyeur looking at the nude body of a woman is an effective figure precisely because he stands for a real,
""This example is mentioned by S. Alpers in Rembrandt's Enterprise: The Studio and the Market, Chicago, 1988. For a study of Rembrandt's self-portraits, see H.P. Chapman, Rembrandt's Self-Portraits, Princeton, 1990, reviewed in this journal by D.R. Smith (LXXlt, 1990, 661-664). A semiotic perspective on his self-portraits, related to psychoanalysis, is proposed by Hal in Reading "Rembrandt." m "Velazquez' Las Meninas," October, xk, 1981,45-54.
"""Notes on the Index," October, m, 1977, 68-81, and rv, 58-67. These papers have had a decisive influence on the art history of modernism. Krauss gives many examples, one of which we cannot resist quoting: "To actually place one's tongue in one's cheek is to lose the capacity for speech altogether. And it is this rupture between image and speech, or more specifically, language, thai Uuchamp's art both contemplates and instances" (p. 206). The play here of indexicality is double: Duchamp indexes the loss of speech by representing the tongue in the cheek as concretely hampering speech, and Krauss reads this representation as an instance, a concrete case, of the loss of speech, which turns the representation as a whole into an index of speechlessness.
objectifying contiguity between look and object defined by looking as a real act. The index functions here in conjunction with the icon: the figure directing his eyes somewhere is taken to stand for a similar figure, a man looking at a woman. In the same way, the open mouths, iconically suggesting screams, of the popes in Francis Bacon's early portraits after Velazquez, function iconically because they also function indexically; the contiguity between screaming and the pain that induces it enhances the effectivity of the works.71
The most obvious use of the concept of index is the pointer. Pointing elements in an image are the most convincing case against the notion that the image is still and can be "read" in a momentary, punctual act. Pointers make us aware of the way our eyes move about the surface in different directions, some of which are suggested by indexical signs. When a figure points a finger in a certain direction, our look will follow the figure's directions.
One category of indexical signs gives the illusion of expressivity. Those are all the signs that refer to the maker of the image, ranging from the recognizable "hand" of the artist, the will to be expressive as in expressionist painting, to the signature. But the index functions in very different cases, and the most plausible one is not necessarily the most reliable one. The signature of the artist, for example, is an index of the person of the maker, even if it is a false signature; that is precisely why it is a sign, a stand-in for an absent other. From the perspective of the sender, a false signature is an icon (of the real signature) parading as an index. This possibility of falsification of the most materially grounded of the three signs is fundamental. As Eco has written with his characteristic philosophical insight parading as a joke, the sign can be defined as everything that can be used in order to lie.73 For in a Peircean view, the sign stands for something else, and that something is logically absent from the scene.
The signature has been the occasion for fascinating philosophical reflections in the work of Derrida. It is the most typical case of the Derridian "trace," the indexical sign that refers by contiguity, not simply to the past (the maker of the image) but, more importantly, to the future, the reading of it.71 The act of writing, which for Derrida is much more than graphic writing alone, is precisely the production of traces, and painting, drawing, and, most emphatically, etching are also acts of writing in this sense.74 Therefore, the notion of the index suggests that we do not only account for images in terms of their provenance and making, but also of their functioning in
71 See E. van Alphen, "Mystery-Portraits: The Detective Novel as a Theory for Bacon's Barly Paintings," in Representations of Death, ed. SB. Goldwin and H. flronfen, Ithaca, in press.
72 A Theory of Semiotics (as in n. 64), 10.
n J. Derrida, Of Grammatology, trans. G.C. Spivak, Baltimore, 1976.
"For a collection of literary analyses in this Derridian vein, see P. Kamuf, Signature Pieces: On the institution of Authorship, Baltimore, 1988.
SEMIOTICS AND ART HISTORY 19]
relation to the viewer: their structure of address. In a discussion with John Searle, Derrida insisted on the impossibility of determining exactly which speech-acts are performed, as Searle would like to do in his logic and typology of speech-acts. Instead, Derrida emphasises the constant interaction between the reader or viewer who tries, but in vain, to fix meaning and point down the act as a one-time, predictable performance.75
Finally, the symbolic sign in Peirce's theory must not be confused with the many different and often vague colloquial meanings of the word "symbol." As the definition in the quoted passage makes clear, it is dependent even more strongly than the two other terms on the act of interpretation that brings it to life, because without that interpretation it simply does not exist—as a sign, that is. What is not specified in this passage is the ground on which interpretation of a symbol comes about, which is the conventional rule of correlation between sign and object or meaning. The interpretant formed by the reader is possible because the latter knows what things usually mean in the culture in which the sign functions. The idiom of a particular language is conventional in the same sense as the idiom of iconographic traditions. "Translation" from one language into another, from an image into words that explain it, from one image into another image, all work by virtue of the knowledge of such idioms. These conventional signs are always also involved when we read iconically and indexically. We only come up with an iconic interpretation of a portrait because we know the convention of portrayal.
Symbolic interpretation, which always underlies other kinds of interpretations, is the most convincing evidence of the cultural specificity of pictorial traditions and styles. Even when we think the image is "realistic," we are in fact imbued with the convention that suggests that certain kinds of pictorial signs stand for "reality" more clearly than others. With the help of iconicity, which suggests that the image must have an object in the real world, and indexicality, which makes us aware of the real contiguities between elements of the image, and between those elements and ourselves, symbolicity by virtue of its very arbitariness is the most deceptive code. In everyday life, we tend not to question what is conventional; we don't even notice it. Much of art-historical work aims at denaturalizing these conventions, and to bring forth insight into the historical changeability of conventions. This kind of work can be seen as analysis of the symbolic as a code. Iconography is in this sense a semiotic approach.
Peirce's semiotic theory is relevant for the study of art because it helps us think about aspects of the process of art in society, in history, in a way that is not bound up with the artist's intention. It can contribute to the explanation of why certain elements of an image are particularly seductive or deceptive, suggesting depiction
6 The clearest account of this debate can be found in the afterword to the recent paperback edition of Derrida's Untiled Inc.
of something real, while specularity, a return to the self away from the real, is in fact the basis of the seductiveness. The mix of iconicity, indexicality, and symbolicity that every image presents emphasizes the other side of the illusory unity that the image is so easily assumed to be: its fracturedness, the "difference within" that allows for a view of image-seeing that is dynamic and positioned in time.
5. Ferdinand de Saussure
If Peirce's view of the way signs function is primarily dynamic, that of Ferdinand de Saussure (1857-1913) involves far more stasis. In order to understand how the words in a language (lattgue) function, Saussure argues that we must set to one side both the aspects of historical change within particular languages over time and actual utterances or performances of language in local situations (parole). What is sought is the global set of rules governing a language; the state of a language as a whole at a given and, as it were, "frozen" moment. Isolating the total body of rules allows investigation of the internal relations among the signs of a language, which, Saussure proposes, are what determine the meanings of those signs: the meaning of a word derives from its "diacritical" differences from other words. "Red," for example, acquires its meaning from the ways in which it is not "green," not "blue," not "yellow," and so forth. Although, as a linguist, Saussure's primary task was to elucidate the operations of verbal language, he also put forward the view that his method could be generalized: "One can conceive of a science that studies the life of signs at the heart of social life. ... Linguistics is only one part of this general science."7* Later analysts have extended the Saussurean model toward what is indeed a great variety of cultural systems. Roland Barthes, to take only one figure from this tradition, proposed the systematic study of "any system of signs, whatever their substance and limits; images, gestures, musical sounds,
objects, and the complex associations of all of these____"n
In Elements of Semiology Barthes set out to explore the operations of signs in the fashion system, the food system, the automotive system, the furnishing system;71* in Mythologies his analyses ranged even more widely.7"
Images are obviously included in this extended project/0 and it may be useful to see how semiotics, in its structuralist phase, tackles a particular work of art. Our
7" F. de Saussure, Cours de linguistique generale, Parts, 1960,33.
R. Barthes, Elements of Semiology (as in n. 42), 9.
" On the semiotics of clothing, see also R. Barthes, The Fashion System, New York, 1983; on the semiotics of food, see C. I .evi-Slrauss, The Raw and the Cooked, trans. J. and D. Weightman, New York, 1969; on the semiotics of consumer goods, see J. Baudrillard, he Systeme des objcls, Paris, 1968.
R. Barthes, Mythologies, trans. A. Lavers, New York, 1988.
""See, for example, Barthes's investigations of the image in Image— Musk—Text; and idem. The Responsibility of Forms: Critical Essays on Music, Art, and Representation, trans. R. Howard, New York, 1985.
192       THE ART BULLETIN' |UNE Wl VOLUME LXX1I] NUMBER 2
example is Edmund Leach's analysis of the Sistine Chapel ceiling."' Leach concedes the multiplicity of factors that may have influenced the way Michelangelo carried out his task, but he argues that among these should be included the internal logic of the scenes and narratives depicted in the nine central panels. These can be divided, Leach suggests, into three groups. The first shows God in the Cosmos without Man: God Dividing Light from Darkness, God Creating the Sun, Moon, Stars ami Planets, and The Spirit of God Moving on the Waters. The second group shows the Garden of Eden story, in which God and Man are together in Paradise: The Creation of Adam, The Creation of Eve, and The Fall and Expulsion from Eden. The third group relates the story of Noah, where sinful Man is in This World and separated from God: The Sacrifice of Noah, The Flood, and The Sin of Ham. The primary opposition governing the disposition of the ceiling panels is between This World, where Man is impotent, mortal, sinful; and the Other World where God is omnipotent, immortal, sinless. Equally important, in Leach's view, is the Liminal space between the terms of the polarities:
In any binary pair of the type "A"/"not-A," the boundary layer which is neither the one nor the other but both at once, "both A and not-A," is especially "interesting" because it is "repressed." This is where the theory of structuralism ties in not only with psychoanalysis but also with anthropological theories about magic and primitive religion. The boundary, the interface layer which separates categories of time and space, is the zone of the sacred, the forbidden, that which is taboo; God when seen from one side of the fence, Sin when seen from the other.1*"
Leach's decision to concentrate on the central panels stems from his Levi-Straussian" understanding of the importance of the "mixed" space in between the poles of the opposition, for here one finds complex and transgres-sive figures: Eve, who is shown in states of both innocence and "corruption" (her face echoing both the sinless uncreated Eve seen in The Creation of Adam, and the serpent who brings about the Fall); Adam, both pristine first man and ruined, postlapsarian figure of sorrow; the tree and serpent of Eden, which coil both toward and away from Paradise, the ambiguity assuming visual form as the spiraling and torsion in the center of the panel of The Fall. Pursuing such sacred ambiguities in other scenes, Leach finds further evidence of complex, contradictory figures: Cain, the first murderer, but also the
81 E. Leach, "Michelangelo's Cenesis: A Structuralist Interpretation of the Central Panels of the Sistine Chapel Ceiling," Semwtica, lvi, 1985, 1-30. An earlier version of Leach's interpretation appeared in the Times Literary Supplement for 18 March 1977, 311ff.
"; Leach, "Michelangelo's Genesis," 20.
Bi See E. Leach, Lcvi-Strauss, London, 1970.
" Similar procedures are to be found in an interesting, if neglected, study of a painting by Paris liordone, J.-L. Schefer's Scenograpliie d'un tableau, Paris, 1969.
founder of cities and the culture-hero; Haman, a precursor of Judas but also of Christ, in the same way that the figure of Holofernes is a cross-reference to John the Baptist, and Judith to Herodias.
This brief sketch, which does not begin to do justice to the complexity of Leach's interpretation, is enough at least to indicate how Saussurean procedures became elaborated within structuralism. There are perhaps three essential moves. First there is the drawing of a boundary round a distinct, enclosed corpus (Leach excludes the large nude male figures, the Sybils, most of the Prophets, the Tree of Jesse, and the medallions). Then there is the segmentation of that corpus into signifying units (here derived from the primary oppositions of "This World" and "the Other World," as well as the "trickster" zone produced by their merger). Finally, the segmented units are related syntactically, through a logic of transformation and condensation (Eve metaphorically connected with Mary, Esther, and Judith; Noah with Christ and Adam). All these have their parallels in Saussure's own methods: the drawing of a boundary around the langue (setting parole to one side); the segmentation of the langue into units (morphemes, phonemes, and so on); and the mapping of the units in terms of oppositional syntax.84 But each of the moves is problematic, and in poststructur-alist semiotics within the Saussurean tradition (but moving ever more remotely away from Saussure), quite different steps are proposed.
Derrida's text The Truth in Painting is probably the most systematic answer to the kind of visual analysis proposed by structuralism. The first move, the drawing of a nonpermeable perimeter around the work of art, is the subject of Derrida's lengthy excursus on "framing" ("The Parergon").85 It may be essential to the structuralist project, as well as to many others (mcluding "formalism" in art history) to maintain the fiction that the work of art is characterized by its apartness, that it inhabits an area of autonomy and separation from "extrinsic" concerns. But for Derrida semiosis is a process that can never be placed within a logic of enclosure: "There is frame, but the frame does not exist"** (il y a du cadre, mais le cadre n'existe pas). Where Saussure theorizes the sign as a fixed and static entity, with each signifier stitched to its stable signified, Derrida argues for the dynamism of signs: that a sign is not (as in Saussure) the conjunction between a signifier and its single, univocal signified, but the movement from one signifier to another, the motion between them.87 As motion, visual signification is therefore incompatible with the ideas of boundary, threshold, frame; it is a
15 Derrida (as in n. 25), sect. 2. "The Parergon" was published separately in English translation in October. ix, 1979, 3-41. For a lucid commentary, see Culler (as in n. 21), 193-199.
""Derrida (as in n. 25), 81.
"7). Derrida, La verite en peinture, Paris, 1978, 93.
SEMIOTICS AND ART HISTORY 193
"passe-partout.""" Investigating the conceptual structure of the frame in aesthetic discourse, Derrida finds that it is both fundamental to that discourse, and at a profound level absent from it. Fundamental, because without the idea of frame, there can be no object of aesthetics:
We must know what we are talking about, what concerns the value of beauty intrinsically and what remains external to an immanent sense of beauty. This permanent demand—to distinguish between the internal or proper meaning and the circumstances of the object in question—organizes every philosophical discourse on art, the meaning of art.... It presupposes a discourse on the boundary between the inside and the outside of the art object, in this case a discourse on the frame.89
Yet even though it is the idea of the frame that calls the discourse of aesthetics into being—for without it, that discourse could not open itself, could not define what it is that it discusses—that discourse, according to Derrida, cannot adequately theorize the frame or describe its opening move. What it can, of course, discuss is the "outside" of the work of art, which comes into being as outside once the concept of frame is in place; and equally well it can discuss the "inside," what is proper to the work of art. But the frame itself is consigned to a kind of conceptual limbo, for the reason {Derrida argues) that the frame is the one thing in the discourse of aesthetics that escapes the categories of "inside" and "outside."
In fact, the frame is both at once, a hybrid, a categorical aberration—which might be manageable if the discourse within which the "frame" operates, and which it also establishes, could permit a mediating zone between its two extremes (perhaps in the manner of the "trickster" category discussed by Leach). But the discourse of aesthetics, exemplified for Derrida by Kant, cannot allow such a zone of aberration to be admitted, since that would be to call into question its own primary move, the division of the field into "inside" and "outside." Instead, the frame is conceptually disavowed and repressed, becoming an ornamental supplement, an unnecessary and optional accompaniment to the work of art. Nevertheless, this relegation of the frame to the place of a mere incidental in aesthetic discourse cannot conceal, in Derrida's analysis, that the latter's central area of interest, the "inside" of the work of art, depends for its very being on the conceptual operation of the frame; that is, on an operation that threatens the clean separation of "outside" and "inside" to its foundations, since the concept of the frame is the undoing of the relation of "inside" to "outside" on which all else is predicated. Derrida's argument aims to expose the persistent logic of enclosure that allows there to be found in painting the stasis of transcendental contemplation (Kant), stabilized reference (Meyer Scha-piro), or ontotheological presence (Heidegger). Against
* Derrida, "Passe-Partout," (as in n. 25), Sect. 1. sa Derrida (as in n. 25), 45.
such a logic, and by pressing hard on the contradictions and incoherences of its fundamental moves, The Truth in Painting shows visual semiosis to be a matter of disfram-ing: an unending dissemination that, nevertheless, as Derrida himself but also Eco and many others have repeatedly pointed out, always occupies specific social and historical sites.""
If enclosure is impossible, so is segmentation. When one fails, so does the other: it is only within the confines of an enclosed and tabular space that internal oppositions can be established and "geometrized," and from those oppositions that the "significant units" can be derived. By the same token, the relation of opposition (which in Leach's analysis lies behind all the units of signification he isolates, including those of the intermediary, "sacred" zone) cannot survive the dissolution of the aesthetic boundary: as soon as the idea of a delimiting frame is questioned and the possibility of dynamic semiosis is admitted, the relation of opposition must give way to that of nonoppositional difference. The image becomes what it is by being traversed by flows of signification that cut across the boundary, making the image part of a general circulation of signs and codes within the social formation as a whole.
While Derrida's revisionary work within the Saus-surean tradition may have resolved some of the difficulties of structuralist methodology, it might be objected that the Saussurean legacy is of limited use in furthering understanding of visual art, since all of its models are verbal ones.''1 Saussure's projected science of signs may have called for an expansion of inquiry beyond the domain of language, but in practice the term "signifier" is modeled on the linguistic case. If this is so, it is right to wonder to what extent the "expansion" proposed by a "general" science of signs may in fact be an attempt at appropriation, the absorption of the visual domain into the empire of linguistics. For obviously there are a great number of aspects of visual art and visual experience that cannot be "translated" into language at all. As Michael Baxandall has put it, "In fact, language is not very well equipped to offer a notation of a particular picture. It is a generalising tool... the repertory of concepts it offers for describing a plane surface bearing an array of subtly differentiated and ordered shapes and colours is rather crude and remote.'"'2 The vocabulary of our languages is
n Derrida insists on this qualification, most emphatically in the afterword to Limited Inc. It is precisely because of the theoretically overriding effect of dissemination that we are forced to interrogate which social and political pressures do check the actual dissemination.
Some readers of Derrida sometimes cite as an indication of the persistent "verbal imperialism" of Derrida's project a certain impoverishment in his dealings with the visual domain, perhaps related to the poverty of the works of art discussed in The Truth in Painting (setting to one side those relating to Van Gogh). One might agree—but there is no reason after all to believe that such impoverishment, if present, is necessarily the result of Derrida's method. Moreover, if the arguments in Truth in Painting are accepted, the secure autonomy of a "visual domain" cannot simply be taken for gTanted.
"2 Baxandall, 3.
194
TUP ART BUI LETtN rUNT. 1991 VOLUME LXX1II NUMBER 2
able to scan for only a fraction of the hues that a painting presents us; the lexicon for shape is not much better. If the word "signifier" is modeled on the verbal case, surely everything that makes up the visual as a specific domain will be lost from view?
While such apprehensions are well based, in principle semiotics is well equipped to respect the specificity of visual and verbal discourses: not only does its key term, semiosis, embrace both visual and verbal practices of the sign, but its attention to sign-types forces it carefully to distinguish between those that belong to the linguistic domain and those that do not. In our "Narratology" section, for example, the narratives described are specifically visual ones; and the distinctions made by Peirce among symbol, icon, and index supply a firm basis for fine-tuned description of the differences between verbal and visual modes. But Peirce is one thing, Saussure and his tradition another: the question remains, whether the semiotics of Saussure is adequate to the specificity of works of visual art.
Take, for instance, the problem of the "significant unit." In language, the minimal blocks are abundantly clear: we all know what is a letter, a word, a sentence; we can pick out the fundamental "atoms" of language without difficulty. But with an image, the nature of what is to count as a unit is far from obvious. We might try to say that below a certain threshold, perhaps roughly corresponding to phonetics in language, there are marks that contribute to, but which do not yet produce, signification—individual brushstrokes or lines, or dots or pixels; and that above that threshold these as yet nonseman-tic marks emerge as productive of meaning. But can we say that the marks below the threshold are "units"? Or above the threshold? Particularly in the Saussurean tradition, the positing of meaning-bearing units— signifiers—seems essential.91 But a painting is a continuous surface, with marks that blend together inextricably. If no minimal units for images can be found, then a visual semiotics, deriving horn Saussure, must be an impossible endeavor: we cannot establish where the "signifier" actually is.
The objection is understable, but it may be misplaced. The problem of mismatch between words and images can, in fact, lead us in rather a different direction, toward the question whether the individual word actually is language's prima materia. In Saussure's view, a certain "atomism" of language undoubtedly prevails, and he is confident that, if we want to exemplify what language is, the individual word can stand in for all linguistic operations. Yet it is not obvious why this synecdoche should be accepted. The problem and source of the atomistic view could be said to be the semiotic positivism that claims ontological status for signs. If the sign is a thing that is there, then signs must be numerable, hence, discrete and intrinsically static; and the quest for the significant unit is
"'The Saussurean tradition, we note, is famous for its proliferation of "ernes": mythemes, semes, monemes, etc.
on."4 That quest is a reflection of a philologically derived linguistics that posits meaning as occurring at the level of the word or the sentence, but does not consider the larger aggregates, the bonding together of words and sentences in social practice, as discourse.0^ At this level, signs are not discrete but "dense":"" individual signs become molar, consolidated, fundamentally inseparable. If a visual semiotics pitches itself at this level, which is that of discourse and interpretation rather than of taxonomy, the difference between verbal and visual discourse is no longer one of the status and delimitation of the signs that constitute them. To think of semiosis as process and as movement is to conceive the sign not as a thing but as an event, the issue being not to delimit and isolate the one sign from other signs, but to trace the possible emergence of the sign in a concrete situation, as an event in the world.
Whether such a dynamic conception of the sign can still be called Saussurean is an open question: from one point of view, there is a certain continuity with Saussure, and for Dcrrida especially it was in part as a result of critical engagement with the stasis and fixity of the Saussurean sign that his own account of the dynamic semiosis came into being;''7 from another point of view, the changeover may be a genuine mutation, a break with the Saussurean legacy. But the problem of the "significant unit" is only one of the places at which Saussure's conception of the sign makes it hard—without drastic modification of the system—to see how a visual semiotics might be developed from a Saussurean framework. Equally challenging is what Saussure (speaking of language) referred to as the "arbitrariness" of the sign.
For Saussure, the relation that obtains in language between a signifier (a word) and a signified (its corresponding concept)'" is a matter of conventional agreement: "Apparently nothing would prevent the association of any idea whatsoever with just any sequence of
"'The points briefly stated here, concerning "minimal units" in language and in images, are argued at greater length in Bat, introduction and chapter 1.
"5See C. MacCabe, "On Discourse," in Theoretical Essays: Film, Linguistics, Literature, Manchester, England, 1985, 2-112. For a full elaboration of the theory of discourse in semiotic terms, see Hodge and Kress (as in n. 54).
" With apology to Nelson Goodman for this reversal of his term "density" (in Languages of Art [as in n. 12]).
°7 Though there are certainly other, non-Saussurean or obliquely Saussurean traditions that argue for the dynamism and the social embeddedness of signs and discourse, notably that which stems from Voloshinov/Bakhtim see Voloshinov (as in n. 54); M. Bakhtin, Rabelais and His World, trans. H. Iswolsky, Cambridge, 1968; Bakhtin: Essays and Dialogues in His Work, ed. G.S. Morson, Chicago, 1986; Rethinking Bakhtin: Extensions and Challenges, ed. G.S. Morson and C. Emerson, Evanston, 1989. On Derrida's engagement with Saussure, see R.M. Strozier, Saussure, Derrida and the Metaphysics of Subjectivity, Berlin, 1988, 160-288.
1B "The linguistic sign unites, not a thing and a name, but a concept and a sound-image"; Course in General Linguistics, ed. C. Bally and A. Seehehaye, trans. W. Baskin, rev. ed., London, 1974,66.
SEMIOTICS AND ART HISTORY 195
sounds.'"* There is nothing about the idea of a tree, for example, that indicates that the sound "tree" can be made to correspond to it; but can the same be said of images? To take a difficult case, is the "realism" one might find in a Renaissance painting, when comparing it with, say, a Chinese literati painting, a matter of convention, or is it a result of the fact that the Renaissance painting more directly reflects the natural world?"10 For if the Renaissance painting involves representational "directness," an uncoded access to nature, to the extent that it does so, its realism stands outside conventions and codes; it is not a product of sign activity at all, and therefore is not available as an object of semiotic inquiry.
Let us take two figures who have argued that there are no limits to the conventionality of representations: Roland Barthes and Nelson Goodman. It is part of Barthes's project, in many of his works, to take what might be thought to be a "natural" phenomenon, and to show that it is fact a cultural construction, a product of history and not of nature; the idea of "nature" being, in Barthes's view, a myth or a mystification of the workings of culture and history."" In Languages of Art, Goodman writes against the common-sense view that lifelike representations are "realistic" because they copy real things in the world, arguing that the effect of lifelikeness is a product of a certain kind of denotation, and that in the case of pictorial representations it is the result of the particular form of notation that pictures use (the notation of pictures, as opposed to that of schematic diagrams, being "dense" and "replete")."12 With both Barthes and Goodman, realism is entirely conventionalized, and the vraisetlt-hlance of an image is described as a quality that can be accounted for without reference to a criterion of fidelity to a natural world.110
It is often said against such conventionalism that it flies in the face of the fact that human viewing is not only a matter of codes and conventions but of inborn perceptual capacities, irreducible to cultural factors. Certainly an extreme form of conventionalism is contradicted by the admission of a natural human capacity to process
" Saussure, Course (as in n. 98), 76. The parallel between Saussurean "arbitrariness" and modernism's break with mimetic representation is developed by Y.-A. Bois in "Kahnweiler's Lesson," Representations, xvm, 1987,33-68.
""' See E.H, Gombrich, Art and Illusion: A Study in the Psychology of Pictorial Representation, rev. ed., Princeton, 1984, 84-86. For Nelson Goodman, however, European painting is no more able to justify the claim to realism, as defined by Gombrich, than any other sort of painting; its system of perspective, for example, is only a series of conventions (Languages of Art [as in n. 12|, 10-19). See also Damisch (as in n. 2); and N. Bryson, Word and Image, Cambridge, 1981,1-28.
"" On the unmasking of social myths, see Barthes, "Myth Today," in Mythologies (as in n. 79), 109-59; and J, Culler, Roland Barthes, New York, 1983,33-41.
"I! On "density" and "repleteness," see Coodman (as in n. 12), 225-232.
'"•"See Barthes, "The Reality Effect,' in T. Todorov, ed., French Literary Theon) Today; A Reader, trans. R. Carter, Cambridge, 1982, 11-17. For an incisive critique of Goodman's account of the "reality effect," see R. Wollheim, "Nelson Goodman's Languages of Art," in On Art and the Mind: Essays and Lectures, London, 1973,290-314.
imagery, but one may not be convinced that all forms of conventionalism are threatened by this to the same degree. Discussing the drive (Trieb), Freud proposes (in the Three Essays on Sexuality) a model in which a biological or "hard-wired" program (for instance, the infant's need to suckle) becomes the support for forms of symbol-ization that deviate from the drive and transform it; these forms are said to be "propped" on the instinct (entsteht in Ahlehnung an), and also to swerve away and to produce effects in human subjectivity that are irreducible to instinct.1"1 Freud here is far from being a pan-conventionalist, and his model in fact depends on the instance of a biological program; but he also suggests something highly relevant to the discussion of conventionalism in art, that the model he proposes "represents the model of every drive," including the scopic drive."b To the extent that the recognition of images depends on a prior, hard-wired program, it stands outside semiosis; but recognition of such limits to semiosis does not exclude the possibility that the most complex forms of symboliza-tion, for example of the kind we associate with visual art, may be propped superveningly upon the nonsemiotic base.
6. Psychoanalysis as a Semiotic Theory
For many scholars, the most useful contribution of semiotics to art history concerns semiotic's focus on the socially constructed nature of signs. Here semiotics reinforces tendencies within the social history of art that view representations not as the reflection of a reality found elsewhere but rather as an active and shaping force in society. But semiotics is a complex and multiple field, and next we will be looking at two further aspects of semiotics that, though apparently less directly relevant to the project of the social history of art, may, we would argue, helpfully complicate the latter: psychoanalysis and narratology.
Psychoanalysis is a mode of reading the unconscious and its relationship to expression, and as such it is a semiotic theory. Using this theory for the study of visual art assumes that art bears traces of the unconscious. Moreover, many of the key concepts of psychoanalytic theory have a specific visual status (the imaginary, the gaze), or refer to visual experiences (castration anxiety, the mirror stage), to sign-making (condensation, displacement), or to concepts we tend to visualize (the breast, the phallus). When psychoanalytic theory is used to read visual art or literature, however, the transfer of a method for curing subjects to a method of reading a work poses the problem of the nature of the relationship between the theory and the work of art. Psychoanalysis is a "talking cure" in which the patient does the talking, the interpreting; in psycho-criticism the work cannot talk, so
Freud, Three Essays on Sexuality, vn in The Standard Edition of the Complete Works of Sigmund Freud, ed. J. Strachey, London, 1953-66. See also the commentary of J. Laplanche, in Life and Death in Psychoanalysis, trans. J. Mehlman, Baltimore, 1976, 8-24.
11,5 Laplanche (as in n. 104), 8,11.
1%
THE ART BULLETIN JUNE 1991 VOI UMF LXXIII NUMBER J
who is the patient? If psychoanalysis tends to take on the status of a master code that can be "applied" to art, one can also argue that the critic is the patient who does the talking (s/he is the only one who talks), while the work of art is the analyst who orients the analytic work (the analyst is typically silent, but strongly structuring of the analytic work). One can even argue that the discourse of psychoanalytic theory is the patient whose unconscious is uncovered by the slips it produces when confronted by the visual work (many recent analyses of Freud's texts follow this lead). Without prejudging the nature of the relationship, then, we assume here that the relationship between the work and psychoanalysis is an interaction between two discourses, conducted among three subjects: the psychoanalytic theorist, the work, and the critic.
With this in mind, we can distinguish a few basic methodological models that have been and can be followed. The most common one in classical psycho-criticism is the analogical model, based on an assumed analogy between the processes and products of the practices of psychoanalysis and visual art. In the early days of psycho-criticism, the analogy was based, broadly speaking, on one of two comparisons.
The first studies works of art insofar as they display the psychoanalytic theory's narrative of the development of the subject. The work's story and the theory's story are analogous. Cases of Oedipal desire, incest taboo and its transgression, and castration anxiety were collected and interpretations of classical masterpieces as Oedipally motivated appeared: Freud's interpretation of a childhood memory of Leonardo da Vinci is perhaps the most famous case, along with Jones's Oedipal interpretation of Hamlet.M' Our position as outlined means that we hardly have a place for this kind of diagnostic reading that would focus on the symptoms of psychic pathology or disease. But the questions Freud raised apropos of Leonardo, and also of Michelangelo's Moses, can also be seen primarily in terms of a reader-oriented commentary. Freud's and Jones's interpretations begin with a Morelli-like detail—the little finger in the Moses, the question why Hamlet does not kill his uncle right away—that becomes an obsession of the critic. Thus, the relationship between critic and work, which Freud and Jones took to be that of analyst to patient, can be reversed. The process that made them think they were the analysts applying their master code to the work can be called transference, a projection on the work-analyst of roles the critic-patient needs to work through.1"7
™ For an overview of all the "models" that can be distilled from Freud's writings on art, see F. Orlando, "Freud and Literature" in Poetics, XIII, 1984, 361 ff. P. Fuller, Art and Psychoanalyse, London, 1980, criticizes Freud's art criticism astutely. The tensions inherent in the case of Oedipus himself have been analyzed playfully by H. Verhoeff, "Does Oedipus Have His Complex?", Style, xvm. 1984,261ff.
1WJ. Gallop, "Lacan and Literature: A Case for Transference," Poetics, Xlll, 1984, 301ff. Gallop continues the line proposed by S. Felman in "To Open the Question," Yale French Studies, l.v/l.vi, 1977, x. Felman's recent book, Jacques Lacan and the Adventure of Insight: Psychoanalysis in
The second comparison the analogical model works with is between the work and psychoanalytic theory itself. When Freud alleged Sophocles' Oedipus Rex as the "source" for his theory named after this work, he was in fact doing this: the work became more than the illustration of the theory; it became its analogon, structurally identical to it. The work is a staged theory—but as the founding example shows, the theory is also a staged work. This is why Freud's writing is so strongly literary— and visual, constantly leaning on visual metaphors. The maximal gain of such a procedure is that it sheds a particular light on a work, which can be interpreted in many other ways. The use of either of the two comparisons does not protect the work against arbitrary interpretations. In fact, the theory and its interpretive schemas have been taken as a whole, as a story in themselves, which is superposed on the work. Such doubling presents the inconvenience of leading invariably up to allegory. And allegory is a flight away from the signifier toward an elusive, logocentric meaning outside. The work can be appropriated for ideologically dubious uses, and no check on this appropriation is possible.1"8
The analogical model is not entirely pointless, however. Analogic flunking is one of the basic modes of dealing with the world as it is accessible in discursive practice. The procedure presented here is one instance of that mode. It helps to gain access to discursive objects that do not by themselves yield their meanings, that is, it offers precisely the analogies we need to make meaning. Hence those interpretations have didactic, semiotic, and communicative value; objects rather than method, they offer relevant material for the study of the place and function of psychoanalytic concepts in culture.
Another variant of the analogical model, the medical model, reverses the relationship between theory and work almost entirely. Psychoanalysis is now not the informant but the informed discipline. Art becomes a document and illustration. The major problem with this approach, in addition to the dubious relevance for the study of art, is the difference in status between the unconscious and the highly and consciously elaborated work of art. The problem is not that the text has no unconscious, or that the status of visual features is questionable in relation to symptomatics; it is that interpretation of this kind is psychoanalytic rather than visual. Such interpretations often offer coherent and complete accounts of the work, but the form of these accounts is entirely medical. As to the work itself, the evidence tends to consist of loose ends and disparate
Contemporary Culture, Cambridge, 1987, is of crucial importance for those interested in the connections between semiotics, art, and psychoanalysis.
™ On allegory, the key text remains W. Benjamin, The Origin of German Drama, trans. J. Osborne, London, 1977. J. Fineman, "The Structure of Allegorical Desire" in October, xii, 1980, 47-66, is also relevant. On allegory in contemporary visual art, see C. Owens, "The Allegorical Impulse: Toward a Theory of Postmodernism," in Art after Modernism, ed. B. Wallis, New York and Boston, 1984,203-235.
SEMIOTICS AND ART HISTORY 197
details whose relative importance for the work as a whole goes unargued.
Distinct from these various forms of the analogical model is the specification model. Here, psychoanalysis is used as a searchlight theory, allowing specific features to be illuminated, sometimes explained but primarily read, by means of psychoanalytic concepts.10" The goal of such interpretations is not to confirm the psychoanalytic content of the material, but to make explicit in what ways the presumed subject exposes itself as existing through various psychoanalytically theorized problems. For instance, the famous and over-quoted woodcut by Durer, Draftsman Drawing a Recumbent Woman of 1525, invariably alleged as a statement on the technique of perspective, can be argued to represent a fear of women: the standing ruler, not without phallic overtones, also protects the draftsman from the proximity of the woman who is, in addition, doubly screened off, by the screen that works like the bars of a prison window, and by the veil that covers her body, otherwise exposing her genitalia to the stern gaze of the man. The woodcut does not illustrate the psychoanalytic concept itself—say, castration anxiety—-but is a representation of a unique instance of it. The defensiveness implied in the representation is transferred to the scientific endeavor that is its overt theme: perspective. As it happens, perspective is itself a device for distancing. The defensiveness that colors the image so insistently—grid, stand, grim look—may remain unconscious to the viewer, but it does make the work even more attractive as a scientific demonstration of perspective, its overt purpose. The critical response to the print demonstrates the reassuring quality of its distancing: whereas it is very frequently cited, the danger it neutralizes is hardly ever addressed. The psychoanalytically informed interpretation supports the technical one, but it also underscores its importance and explains the urgency that is betrayed by its insistent quotation. Thus the psychoanalytic concept informs the analysis but does not reduce the work to what it is not; it is not allegorical. This feature distinguishes the specification from the analogy: analogy summarizes while specification expands the realm of meaning.
The point of psychoanalysis is then neither the diagnosis of a psyche, nor its contribution to the interpretation itself, but the possibility it offers to gain access to issues of visual art. In the example by Durer, the "immediate" meaning of the work—the statement on perspective— does not disappear but gains more depth, becomes culturally framed, and shows the complicity of scientific development and gender relations.'1"
""This is, for example, how E.H. Spitz discusses the relevance of psychoanalysis for art in Art and Psyche: A Sludv in Psychoanalysis and Aestlietics, New Haven, 1985.
On the complicity between science and gender politics, see E.F. Keller, Reflections on Gender and Science, New Haven, 1985. S. Alpers, "Art History and Its Exclusions," in N. Broude and M.D. Garrard, Feminism and Art History: Questioning the Litany, New York, 1982,
What we will designate here as the hermeneutk model is different from the preceding models in that it does not use the content of psychoanalysis to inform the work but, instead, draws upon psychoanalytic assumptions and axioms such as its theorization of repression, its views on semiosis, and its theory of the subject, and it uses these as descriptive concepts. This approach draws less upon the developmental theory arid more upon Freud's and La-can's semiotic intuitions. The procedure is not so much interested in traces of the Oedipal drama or the pre-Oedipal conflations, but rather in traces of the unconscious and the forms these take, disturbing coherence: forms of censorship such as condensation and displacement, and contradictions, incoherences, and their status in relation to the coherent, "conscious" propositions the image offers. This procedure can yield relevant views on the work in relation to the more common interpretations of its overt semiotic system.
Condensation occurs when one sign—any detail of the work—refers to different meanings, whether mutually unrelated or even inconsistent. Thus two stories are represented at once, one of them often inavowable. Margaret D. Carroll demonstrates, for example, that Rubens's Rape of the Daughters of Leucippus of 1615-18 represents a pernicious collusion of two different events: a violent rape and a pleasurable heterosexual encounter."1 This collusion is, of course, culturally coded in the discourses surrounding rape, which contribute to the perpetuation of the idea that women enjoy rape. Thus it connects to an ideological condensation while its status as a condensation in the psychoanalytic sense needs further examination. And this meaning is included in this overt and pernicious condensation: the exposure of the women in the most graceful poses, while the bodies of the two abductors, although "coming down on" the women both literally and in street parlance, are hidden and confused with the bodies of the horses. While the latter confusion may be read as an indication of the idea of male sexuality as bestial, hence, irrepressible, the hiding of the male bodies is symptomatic of an insecurity about their own grace. Thus, reading the work as an unconscious condensation, its sexual violence receives a new light: as compensation for the lack of confidence in the seductive power of the two males. As in the case of Diner's woodcut, then, the psychoanalytic hermeneutic helps to provide the more obvious interpretation with an explanatory dimension that gives it more depth, more nuance, and more ideological relevance.
While condensation and its twin concept in Freudian rhetoric, displacement, both concern semantic complexity and the plurality of meaning, displacement is often helpful in revealing a hidden "other side" of an overt meaning. In Picasso's response of 1957 to Velazquez's Las
183-199, is one of the rare commentaries on the Dürer woodcut that pays attention to the gender aspect.
111 M.D. Carroll, "The Erotics of Absolutism: Rubens and the Mystification of Sexual Violence," Representations, xxv, 1989,3ff.
198
TI IC ART BULLETIN IUNE WI VOLUME LXXUI NUMBER 2
2 Velazquez, Lös Menituis, 1636. Madrid, Prado
Meninas, for example (Figs. 2-3), the central positions of the proud painter, on the one hand, and the reflected king and queen, on the other, have been displaced onto two other figures: the man in the background and the little princess. Already spatially central in the Velazquez, in Picasso's painting the princess is now also central via the emphasis given her through color, light, and space. The yellow of this figure, which sets her off against the rest of the work, rhymes with the yellow surrounding the man in the open doorway; the square form of her dress repeats the rectangular space of the open door; and her arms are in the same position as his. This displacement of emphasis onto the man and the princess can be seen as pointing toward an aspect of the scene already present in the Velazquez but screened off, there, by the crowded stage: the vulnerability of the little girl. Leading Picasso's work away from the class issue and the dignifi-cation of the art of painting involved in the Velazquez, this displacement of attention reveals the less avowable sexual concerns that the older work displays only sublim-inally, if at all. So far, Picasso's work proposes a reinter-pretation of Velazquez's statement on class, shifting power relations in the direction of sexuality. There is nothing unconscious in this yet; rather, sexual explicit-ness was a fashionable topic in Picasso's time. But the explicit sexual reorientation escapes full coherence. Although the other figures can be discerned, their cubist treatment, usually emphasizing presence, here hides them, reducing them to mere eyes, and the only other striking figure is a red shape at the bottom right with a distinctly phallic head. With that figure as an index,
3 Picasso, las Meninas, 1957. Barcelona, Museo Picasso de Barcelona
metonymically pointing toward the issue of sexuality, the man in the doorway suddenly becomes threatening, and the girl, frontally exposed to the viewer, is helplessly surrendered to the violence of vision. The other figures now uncannily repeat that violence, Like phantoms in dark corners, rather than protecting the girl; but this meaning displaces the threat from sex to visuality, contaminating the painting as well as our concept of sex with guilt. Reading the Velazquez back from the perspective of the Picasso, we realize that there, too, the presence of the other figures merely exposed the girl more emphatically, enhancing her loneliness; and from the overt drama of court life, the threatening aspect of vision is displaced on sexuality.112
This psychoanalytic hermeneutic can be brought to bear on art-critical discourse as well. Thus Sander Gil-man's use of visual illustrations in his article "Black Bodies, White Bodies," juxtaposing images of monstrosity in women of color to representations of a clitoris with a phallic form, displays an ambivalence toward his own sex that his discursive comments also suggest. In T. J. Clark's discussion of Manet's Olympia (Louvre, Paris) the contrast between the two women obscures the similarity in their figuration; Gilman, taking up Clark's analysis, sees Olympia as skinny and the black woman as plump, setting up a contrast that is questionable in its implica-
"; The literature on las Meninas is abundant and well known. Key texts are Foucault's introduction to The Order of Things, trans. A. Sheridan, New York, 1973, which proposes a semiotic reading. Precisely the semiotic nature of this reading—its readerliness—has been misunderstood. John Searle attempts to explain Foucault's reading through speech act theory in "Las Meninas and the Paradoxes of Pictorial Representation," Critical Inquiry, VI, 1980,477^488, whereas for Foucault the painting exemplified his own theoretical view of classicism. Joel Snyder and Ted Cohen misunderstand both Foucault and Searle and counter their readings with scientistic calculations on the "real" rather than the semiotic point of view, in "Reflections on Las Meninas: Paradox Lost," Critical Inquiry, VII, 1980,429.
SEMIOTICS AND ART HISTORY
199
lions. Both men are blind to the similarity between the women and to the nature of the bouquet of flowers, which we read as a metaphor for "flourishing" genitalia."3
The examples quoted here are all related to a problematic of vision, and that is, of course, no accident. The most central concept that links psychoanalysis and visual art is that of the gaze. That centrality could, however, also be a problem, as we will suggest at the end of this section.
The psychoanalytic description of vision has been a particular focus within the work of Jacques Lacan. Two texts in particular, in which vision is described, have become highly influential in film studies, though their implications for the study of painting, sculpture, and photography are in principle no less powerful than for the cinematic image. Lacan's Four Fundamental Concepts of Psychoanalysis lays down a general theory of the role of signification in shaping visual subjectivity,"4 while the essay "The Mirror Stage as Formative of the Function of the T " deals more specifically with the formative stages of visual experience.""
In The Four Fundamental Concepts Lacan extends into the sphere of vision an argument central to his project in psychoanalysis, that human subjectivity is profoundly and constitutively shaped by the institution of symbolism or sign-activity in its largest sense. To the sum of cultural processes in which symbolism occurs, Lacan gives the name, the symbolic (or the symbolic order), a term meant to designate the entire domain brought into being by the social circulation of signifiers. In other Lacanian texts, more concerned with the role of linguistic structures in determining subjectivity, it may seem that the symbolic order is identical to language; one may come away with the impression that the human being's entry into the cultural field of semiosis occurs at the moment when speech is acquired, and that "the signifier" whose operations Lacan proposes as determining the structures of subjectivity is simply another name for "the word." But in The Four Fundamental Concepts Lacan assumes that the signifier can be visual as well as verbal, and that just as the signifier in the domain of language produces a speaking subject, so in the domain of vision it produces a "seeing subject," that is, a subject whose mode of seeing is the product of the signifier as it operates upon vision.
One aim of Lacan's discussion is to dislodge the Cartesian notion that the subject stands at the center of vision, in a position of mastery over its visual field; this conception has no more validity, in Lacan's view, than
TJ. Clark, The Painting of Modern Life: Paris in the Art of Manet and His Fotloiuers, London, 1985; S. Gilman, "White Bodies, Black Bodies: Toward an Iconography of Female Sexuality in Late Nineteenth-Century Art, Medicine, and Literature," Crilieal Inquiry, XTl, 1985, 2t)4ff.
mLe Seminaire de Jacques Lacan, Livre XI, Les Quatre Concepts fondamen-taux Ac la psychanatyse, Paris, 1973; trans. A. Sheridan, The Four Fundamental Concepts of Psycho-Aitalysis, New York, 1977.
"5 In Ecrits: A Selection, ed. and trans. A. Sheridan, New York, 1977.
the corresponding notion in the theorization of language, that the subject possesses mastery over speech. Learning to speak involves an insertion into preestab-lished systems of verbal discourse that lay down in advance the paths or networks that the speaker's words are obliged to follow. In the same way, once the subject learns to "see" it is obliged to orchestrate its personal visual experience with the socially agreed descriptions of the world around it; thereafter, deviation from this social construction of visuality can be named and dealt with, variously, as hallucination, misrecognition, or "visual disturbance." What is seen is formed by paths or networks that exist before the subject and continue to operate in the social formation long after the individual's demise. The visual field therefore has the character of a "ready-made." In the "Tyche and Automaton" section of The Four Fundamental Concepts Lacan describes the mechanisms of repetition that give visuality its coherence. Once installed in the symbolic order, the subject must refer all new visual "data" to the chains of signifiers that now cut across and organize visual experience. To these chains or tracks, preestablished in the social formation and then internalized by the subject, Lacan gives the name the gaze (le regard).
To expose the prefabricated character of the visual discourses into which seeing subjects are inserted is, however, only one aim of Lacan's discussion. The subject's entry into the various visual discourses that make up the gaze might, after all, be a simple and pacific process, no more a matter of conflict or anxiety than the programming of a computer, or the acquisition of any human skill. What makes both the entry into the gaze, and the subject's subsequent occupancy of the field that the gaze forms, into a complex and conflict-ridden process, is the subject's systematic misrecognition of the externality of the visual discourses through which it organizes sight: for unlike a computer programmed to manipulate chains of signifiers, the human being is, according to Lacan, also structured in such a way as to produce a continuous sense of itself as autonomous headquarters of signification, an "I" that looks out at the world from a central vantage-point, experiencing the visual field as a horizon always composed around itself.
The register of psychic processes in which this sense of centralized identity is produced Lacan calls the imagi-naty, a term that may, however, lead to considerable confusion unless glossed further. In both French and English the terms imaginairc and imaginary carry connotations of illusion or fictionality, as though the ego were unreal or in some way a mirage. Lacan's point, though, is not that the Imaginary is a realm of illusion, as against the "real" and objective structures that make up the gaze or the symbolic order in general. Lacan reveals his debt to post-Kantian thought in his assumption that the human subject has no direct access to reality; all transactions with the kind of objective world postulated, for example, by the physical sciences, are mediated, he argues, on the one hand by the work of the symbolic order (producing
201)        I'HL ART BULLETIN |UNE Wl VOLUME LXXI11 NUMBER 2
signification), and on the other by the work of the imaginary (producing identity): the real, in Lacanian theory, is a logically empty category. It is not that the sense of self generated within the imaginary is false; Lacan's emphasis is rather on its instability, and on the vicissitudes of the imaginary as it attempts to generate a sense of centralized ego in the face of a symbolic order that does not of itself produce, or even require, the notion of center for its operations to proceed.
At this point some examples may help to clarify the discussion; we shall move from a case in mathematics, to language, to the image. When I perform a mathematical calculation, the symbols I manipulate make no reference to my situation; they are impersonal in the sense that the numbers are no "nearer" to me before or after the calculation; though the numbers have not spontaneously rearranged themselves, and 1 am without doubt the operator who works with them, they bear no traces of my personal activity or existence. To enter into the realm of mathematics is, accordingly, to risk being effaced as "i."1"1 The signifiers in language seem at first sight more personal; for instance, language hospitably provides the personal pronoun "I," through which one may enter into verbal systems as one cannot with mathematical ones. But, again, the personal pronoun "I" is, in fact, impersonal: anyone may use it, so that the "I" of language actually refers no more to me, in my specific situation and identity, than it does to anyone else—even a computer or an auto-teller may use the signifier "l.""7 And similarly with the gaze, though I may experience the visual field around me as a horizon whose center I always occupy, so that all of the vision appears to unfold inside me ("somewhere behind the eyes"), the signifiers of sight can no more be my unique possession than the symbols of mathematics can, or the "personal" pronoun "1." And insofar as the social coherence of the visible world requires me to submit my visual experience to the operations of the visual and the verbal signifier, it obliterates me as the center of my lived horizon at the very same moment when 1 seem, to myself, to occupy its heartland."8 Holbein's The French Ambassadors is the work chosen by Lacan to demonstrate this simultaneous process of possession and dispossession in the field of vision. The ambassadors are masters of symbolism, in possession of all the codes of knowledge, of science and art, fashioned in their social milieu: but their visual field is cut across by something they cannot master, the skull
that casts itself sideways across their space, through anamorphosis. For Lacan, the subject's entry into the networks of signification involves a force of decentering so profound that the metaphors for the process are necessarily those of death; the skull is emblematic of the fate of sight as the latter is subjected to the work of the symbolic order.
Evidently Holbein's painting is not the only work of art in which this Lacanian dialectic between the symbolic and the imaginary orders might be discovered: the structures that Lacan describes are in principle capable of manifesting themselves in a wide variety of instances (raising a methodological problem that will be discussed below). For example, in her Lacanian reading of the work of Francis Bacon, Brenda Marshall argues that Bacon's work draws the viewer into complicity with a structure of perversion in which Oedipal law is both asserted and denied. In her essay on Edward Hopper, she finds that although Hopper's images show what is in some respects a world of resolute mundanity and ordinariness, they are nevertheless shot through with the affect of death."" This thematics of loss opens up a rich field of possible interpretation, and in film studies the Lacanian emphasis on what might be called the "mortification" of sight has influenced a number of analyses of cinema's capacity to engulf the viewer in its imaginary world (so that everything unfolds before the masterly witness of camera and spectator). Such analyses note cinema's ability to interrupt that centering of sight around the viewing subject in order to expose the workings of the cinematic apparatus and its symbolic codes, producing a loss, separation, and nostalgia.1211
Yet there is an obvious problem of method here—one that has nothing to do with the accuracy or inaccuracy of Lacan's account of visuality. What Lacan offers is a theory of visual subjectivity in general, not in the first place a theoiy of visual art, and in fact his discussion of The Ambassadors follows what was earlier described in this section as the analogical model, where the work of art serves primarily as an allegory of the terms used in psychoanalytic theory. One might well raise the objection that the painting predictably ends up in a subservient and colonized role, as the mere illustration or staging of the theory. And one might further object that such allegorical instrumentalism might be served just as well, not by Holbein's Ambassadors, but by any picture whatsoever; or, for that matter, by any visual phenomenon or
""See He Seminar of lacaues Lacan, ed. J.-A. Miller, u, Cambridge and New York, 1988, 294-308.
117 On the relations between the first-person pronoun and its operator, see E. Benveniste, Problem* in General Linguistics, trans. M.E. Meek, Coral Gables, 1971, 195-246.
'"The topic of the decentering of the vision in Lacan (and Sartre) is examined in N. Bryson, "Vision in the Expanded Field," in Vision and Visuality, ed. H. Foster, New York, 1988,87ff.
""B. Marshall, "The Lacanian Body: Studies of the Unconscious in Fainting and Aesthetics," Ph.D. diss., University of Melbourne, 1990.
Marshall's study is perhaps the most systematic attempt so far to bring the Lacanian understanding of the unconscious, and of vision, to bear upon painting; her analyses include works by Frida Kahlo and John Singer Sargent, besides those of Bacon and 1 lopper.
'■"See C. Melz, The Imaginary Signifier: Psychoanalysis and the Cinema, trans. C. Britton, A. Williams, B. Brewster, and A. Guzzetti, Blooming-ton, 1982; J.-P. Oudart, "Cinema and Suture," Screen, xvm, 1977-78, 4, 35-47; K. Silverman, "Suture," in The Subject of Semiotics (as in n. 42), 194-246; idem, The Acouslk Mirror, 1-41.
SEMIOTICS AND ART HISTORY 201
scene that could be made to allegorize the gaze (there is nothing in the theory that requires its illustration to be a representation, a picturing).
It may seem that such objections severely qualify the usefulness of Lacan's terms in discussing works of art. The difficulty is not that of finding other examples besides The Ambassadors that might accord with the terms of Lacan's description of vision; the problem is just the opposite, that—if Lacan's general view is accepted—all pictures (or photographs, or sculptures) would do just that. The principles of curtailment and of counterexample seem absent from the argument. This does not mean, however, that there is no way that Lacan's discussion can be made to intersect with individual works of art in all their complex particularity. Following the specification model earlier described, it remains possible to use Lacan's terms (the symbolic, the imaginary, the real, the gaze) not in order to allegorize them through art, but to make explicit those features in a given work that these terms, and perhaps only these terms, describe.
The specification model has the further advantage of being able to provide principles of resistance and counterexample. That is, it may well prove to be the case in practice that Freud's or Lacan's terms will undergo revision in the light of particular works. A clear example here is Leo Bersani's analysis of Assyrian sculpture, which depends on a radically revisionist reading of Freud's texts on repetition and sublimation.121 A similar instance of major revision occurs in film studies, where certain theorists have been obliged to insert into Lacan's account a political elaboration of the qualification that the gaze does not structure the field of vision in the same way for both male and female subjects: since the most powerful visual discourses have been created, historically, by men and for men, the gaze should be theorized, they argue, as a male construction; it cannot be separated, as it is in Lacan, from questions of sexual domination.122 The application of Lacan's general theory of visual subjectivity according to the specification model has, in fact, resulted in a number of compelling interpretations of individual works. An excellent example here would be Rosalind Krauss's analysis of early Surrealist works by Max Ernst.121 What a collage (more precisely, an overpaint-ing) such as The Master's Bedroom explores, Krauss argues, is exactly the "ready-made" nature of the visual field; and Surrealism's interest in automatism may in certain cases need to be thought through, Krauss implies, not as
1:1 L. Bersam, The Freudian Body: Psychoanalysis and Art, New York, 1986, 67-79; and L. Bersani and U. Dutoit, The Forms of Violence: Narrative in Ancient Assyria and Modern Culture, New York, 1985.
133 See, in particular, L. Mulvey, "Visual Pleasure and Narrative Cinema," and "Afterthoughts on 'Visual Pleasure and Narrative Cinema' inspired by Duel in the Sun," both in Feminist Film Theory, ed. C. Penley, New York, 1988. Mulve/s position is reviewed from a Lacanian point of view by K. Silverman in "Fassbinder and Lacan: A Reconsideration of Gaze, Look and Image," Camera Qhscura, XIX, 1989, esp. 59-60. In this article Silverman offers a seminal revisionist analysis of Lacan's theory.
111 R. Krauss, "The Master's Bedroom," Representations, xxvm, 1990, 55ff.
the pursuit of spontaneity and randomness but in the opposite direction, as dramatizations of the prefabricated and machinic nature of visual subjectivity (it may well be that Lacan himself encountered the idea of the visual "automaton" in the Surrealist circles to which he belonged in Paris in the twenties and thirties). The Lacanian model of the gaze is able to pick out, searchlight fashion, a host of signifying details in Ernst's image and to integrate them into a powerful interpretation of the work as a whole. And if it is objected that of course Lacan's "Surrealist" theory fits Surrealist art, since both come—so to speak—from the same stable, one can point to many other Lacanian-inspired readings in which the question of "circularity" does not arise: Kaja Silverman's work on Fassbinder,124 Jacqueline Rose's analysis of Leonardo, '", certain investigations of David and Ingres/2" and Michael Fried's account of Thomas Eakins.127
Those recent developments in semiotic theory that connect with psychoanalysis have strongly shaped semiotics' engagement with issues pertaining to gender as well as other aspects of the subject.12" Indeed, there are three key issues in semiotic theory that have been grounded in differentiation, the relationship between individual subject and social pressures, and power relations: the issue of ideology, that of subjectivity, and that of the relation between the latter and interpretation. All three represent aspects of the key concept of Peirce's semiotic theory, the interpretant, which anchors the production of meaning both in the subject's hic-et-nunc agency and in the social field where ideology shapes the imaginable priorities.129 In order to understand how the
m Silverman (as in n. 122), 54ff.
'* J. Rose, Sexuality in the Field of Vision, London, 1986,225ff.
134 N. Bryson, Tradition and Desire: From David to Delacroix, Cambridge and New York, 1983.
157 M. Fried, Realism, Writing, Disfiguration: On Thomas Eakinsand Stephen Crane, Chicago, 1987.
"Two relatively recent surveys of feminist approaches to art history have made available to the general art-historical audience the increasing number of works in this vein. T.G. Peterson and P. Mathews (as in n. 11), 326ff, discuss feminist art history in terms of generations, beginning with a first generation identified primarily with Linda Nochlin's early work, which is not always treated fairly. We also have problems with the family metaphor implied in the use of generation as a leading principle. Lisa Tickner's "Feminism, Art History, and Sexual Difference," Genders, m, 1988, 92, divides the field according to three views of sexual difference; as an experiential category, a category in discourse, and in psychoanalysis. We highlv recommend these two articles, both of which provide an extensive bibliography of studies of the intersection of art history and feminism. Instead of arguing the relevance of semiotics for feminism, we would like to focus on the relevance of feminism and other theories of gender and of ideology for semiotics' theorizing the subject, and intend to do so elsewhere. See M. Bal, On Meaning Making, Sonoma, 1992 (in press).
12'On gender and class, see, e.g., N.B. Kampen, "Status and Gender in Roman Art: The Case of the Saleswoman," in Broude and Garrard (as in n. 110); and her Image and Status: Working Women in Ostia, Basel, 1981. On the fcmmc fataie see B. Dijkstra, Idols of Perversity: Fantasies of Feminine Evil in Fin-de-Siecle Culture, New York, 1986, and S. Casteras, The Substance of the Shadow: Images of Victorian Womanhood, New Haven and London, 1982.
202
inn. art nuu.rnN iu.nc 119] voiUMt lxxi.i numot.r 2
ideological effect of art can be produced by ideological representation, it is necessary to account for the formation of subjectivity in culture.110 The ultimate rationale for the examination of gendered semiosis is itself anchored in semiotic conceptions of the place of the subject in culture. Semiotics rests on assumptions concerning the subject and a subjectivity grounded in difference, with gender difference as a central one.131 As will become clear in the following section, an image does not represent a single subject-position, but rather an organized plurality of them, thus suggesting what Bergcr called "ways of seeing" to the viewer. This relationship between "I" as speaking subject and the spoken "I," and the ambiguous relationship of the "you" to these two "I"s and the subsequent formation of subjectivity in representation, has been particularly well articulated in film theory.132
In a seminal essay that demonstrates the inherent bond between semiotics and subject theory, Teresa de Lauretis addresses the difficult problem of the relationship between the social experience that constitutes "women" as a group of subjects (not identical but identically "hailed," and thereby ideologically produced), and the semiotic practice of meaning production. Unlike many attempts to theorize the subject in meaning production, which fail because they stay on either side of the divide between sociological and psychoanalytic approaches, De Lauretis takes the difficult but rewarding avenue of Feirce's concept of the interpretant.m Leaning on Peirce on the one hand, and on Lacan on the other, she describes subjectivity as an effect of meaning, "a semantic value produced through culturally shared codes"'" and experience is the process wherein that
'"'On discursive practices and their relations to ideology, see M. Foucault, The Arclieology of Knowledge, trans. A.M. Sheridan Smith, New York, 1475. On the relation between ideological manipulation and the formation of the subject, which will be discussed shortly, see idem, "The Subject and Power," in H.I.. Dreyfus and P. Rabinow, Michel Foucaul): Beyond Structuralism and Hcrmcneutics, Chicago, 1983.
111 See in particular C. Pollock, Vision am! Difference. In her earlier work, Pollock discussed the artist in more experiential terms; see G. Pollock and P. Parker, Old Mistresses: Women, Art and Ideology, New York, 1981, and G. Pollock, "Women, Art and Ideology: Questions for Feminist Art Historians," Woman's Art fournal, tv, 1983, 39^17.
n:Mosl central are K. Silverman's The Subject of Semiotics (as in n. 42); and, for specific analyses that focus on the intimate connections between the voice and the image. The Acoustic Mirror.
'""Semiotics and Experience," in Alice Doesn't (as in n. 64). This article is an example of the benefit that semiotic and feminist theory can derive from the integration of both through a psychoanalytically informed subject theory. De Lauretis defines experience as "a process by which, for all social beings, subjectivity is constructed. Through that process one places oneself or is placed in social reality, and so perceives and comprehends as subjective (referring to, even originating in, oneself) those relations—material, economic, and interpersonal—which are in fact social and, in a larger perspective, historical" (p. 159). De Lauretis elaborates her view as a response to arguments stated by U. Eco in A Theory of Semiotics (as in n. 64) and in The Role of the Render (as in n. 17); and with J. Kristcva, Desire in Language: A Semiotic Approach to Literature and Art, ed. L.S. Roudiez, New York, 1980. The best critique of Kristeva's theory is Silverman's in The Acoustic Mirror, chap. 4, which is an indispensible addition to De Lauretis's necessarily brief treatment of it.
°* "Semiotics and Experience," 167.
happens. Obviously, looking at representations, or making them, is part and parcel of that experience. In a Peircean view, the production and reception of signs are basically a similar activity with a similar result: both receiver (interpreter, reader, or viewer) and producer form interpretants. These are not arbitrary, individual, or idiosyncratic; interpretants are new meanings resulting from the signs on the basis of one's habit. And habits, precisely, are formed in social life. "The individual's habit as a semiotic production is both the result and the condition of the social production of meaning."135 Thus, not only is experience a legitimate basis of interpretation, it is the only possible one. If those who have hitherto believed that they can reach universal validity are willing to see the general validity of this view and, therefore, to abdicate their illusionary claim in favor of semiotic specificity, much indispensible scholarly as well as political progress can be gained.
A dynamic view of the sign, including the signs that constitute visual art and the discourses about it, can help to denaturalize the exclusions that have resulted from those particular framings, as well as, conversely, to use framings to counter these exclusions without falling back into positivistic claims to truth. One such framing that permits a more open and equitable interpretation is narratology, which is the subject of the next section.
7. Narratology
Accounts of narrative in visual art tend to focus on the question of how images are able to narrate stories.136 Although such accounts have great usefulness, the underlying presupposition seems to be that images are a priori handicapped in this competition; narrating is primarily a matter of discourse, not of visuality. Hence, attempts to overcome the limits of the visual, brave as they arc, will have to be considered with indulgence. But from a semiotic perspective, various theories of narrative have been developed that can be brought to bear on visual art, without presupposing that narrative is somehow a foreign mode in visual art. Perhaps the best known example of such a theory is the one implied in Barthes's famous book S/Z.'17 Barthes develops an interpretation of Balzac's short story Sarrasine throLtgh an analysis of five codes
"'-Ibid., 179.
""Recent examples are A. Kibedi Varga's essay (as in n. 14), and his Discours, recit, image, Liege and Brussels, 1989; and W. Steiner, Pictures of Romance, Chicago, 1988.
WS1Z, Paris, 1970, trans. R. Miller, New York and London, 1975. Barthes's initial, structuralist theory of narrative can be found in Elements of Semiology (as in n. 42). Barthes's essays on visual images that are not bound up with his narrative theory are mostly in Image—Music— Text (as in n. 36). On photography he published Camera Lucida (as in n. 57). His Empire of Signs (Geneva, 1970), trans. R. Howard, New York, 1982, offers a semiotic interpretation of Japanese culture, which is implicitly a semiotic of ethnocentrism. Two of Barthes's older texts deal with visual material, albeit in a linguo-centric way: Mythologies (as in n. 79), remaining essays trans. R. Howard as Vie Eiffel Tower and Other Mythologies, New York, 1979; and Systeine de In mode, Paris, 1967, trans. M. Ward and R. Howard as The Fashion System, New York, 1983. A most accessible introduction to Barthes's work is J. Culler (as in n. 101).
SEMIOTICS AND ART HISTORY 203
that the reader allegedly activates when reading this story. The essay is attractive because it is reader-oriented while defining the act of reading as cultural constrained.
The proairetic code for Barthes is a "series of models of action that help readers place details in plot sequences: because we have stereotyped models of 'falling in love,' or 'kidnapping,' or 'undertaking a perilous mission,' we can tentatively place and organize the details we encounter as we read."138 In a way this is a narrative version of an iconographic code.1"1" The hermeneutic code presupposes an enigma and induces us into seeking out details that can contribute to its solution. Although this code may seem less relevant for visual art, we claim that there is a hermeneutic code at work for the viewer, precisely when an image's subject is hard to make out. The semic code inserts cultural stereotypes, "background information" that the viewer brings in to make sense of figures in the image in terms of class, gender, ethnicity, age, and the like. With the help of the symbolic code, the viewer brings in symbolic interpretation to read certain elements, e.g., "love," "hostility," "loneliness," or, for that matter, "theatricality," "vanitas," or "self-referentiality." Finally, the referential code brings in cultural knowledge, such as the identity of the sitter for a portrait, the program of an artistic movement, or the social status of the figures represented. Together, these (and other) codes produce a "narrative," a satisfying interpretation of the image in which every detail receives a place. This narrative is emphatically produced by the reader to deal with the image; it produces the story through the processing of a strange image into a familiar mind-set.14"
The intertwining of codes produced by prior discourses of a culture makes Barthes's approach congenial with Bakhtin's theory of narrative.14' Barthes starts from the receiver, the reader (or viewer) of the work. Bakhtin's concept of polyphony, the intertwining of different voices in the novel, resulting in heteroglossia or the cacophony of incongruous strands of cultural discourses, takes the same issue up from the other side, the side of the sender or writer, in our case the maker of the image. For our purposes, the major gain of this view of semiotics is the awareness that the image is not unified. Indeed, classical art criticism and history has tended, just like literary studies, to seek the interpretation that accounts
m Culler, Barthes, 84.
"'In Travels in Hyper reality Eco analyzes Casablanca as a "cuitfilm" because of the predominance and intertextual play of the stories projected by what Barthes calls the proairetic code. A comparison between iconography in visual art and iconography in literature is undertaken in Bal, chap. 5.
imlx\ this vein an instructive analysis of Francis Bacon's work is published by E. van Alphen, "The Narrative of Perception and the Perception ol Narrative," Poetics Today, in press.
M. Bakhtin, The Dialogic Imagination, ed. M. Holquist, trans. C. Emerson and M. Holquist, Austin, 1981. K. Hirschkop and D. Shepard, eds., Bakhtin and Cultural Theon/, Manchester and New York, 1989, offers a wonderful collection of critical essays on the usefulness of Bakhtin in a perspective of cultural studies. A good introduction is T. Todorov, Bakhtin: The Dialogic Principle, trans, by W. Godzich, Minneapolis, 1984.
for each and every detail of the work within the same framework. Thus details that don't fit are ignored or set aside as unimportant or as "mistakes," evidence of a foreign hand, of studio practice, a process that ultimately discounts the work as nonautographic rather than contributing to a heteroglossic view of the work. Bakhtin helps us to accept that even when the image is made by one artist, the inherent hetero-discursive nature of the culture of which this artist is a product necessarily brings in elements of alterity, if only to be repressed to the margins. Thus an image that overtly represents the intervention of women in a fight between men such as David's Sabine Women (Paris, Louvre) cannot help inserting indexes of homosocial interest.14" Velazquez's Las Meninas has been argued to present a visual heteroglossia in that it is at the same time narrative and descriptive, displaying the act of painting within the description of courtly life as still life.143 This mix of modes is fundamentally different from the famous self-referentiality (Foucault), paradox (Searle), or unity (Snyder and Cohen) proclaimed by other critics, who ultimately gave the painting a unity it so stubbornly refuses to yield.144 Paradoxically, Alpers's refusal to appropriate the entire painting for the narrative mode enabled the critic to do more justice to the painting as narrative than the other, simple narrative readings.
In spite of the importance of Barthes's and Bakhtin's insights into the various factors that collaborate or compete to produce narrative, the one factor that keeps slipping away is the traffic of meaning from source to destination and back. In traditional narrative theory, the concept of the narrator as the source of information or the utterer of the speech act of narrating has favored a model of a unified voice: one narrator determines what the reader is going to get.14" Replacing the author by the narrator, or the artist by an implied orchestrator of the
l4- See Bryson (as in n. 126).
141 S. Alpers, "Interpretation without Representation or, the Viewing of Las Meninas," Representations, I, 1983, 31-42.
144 See Foucault; Searle, 477-488; and Snyder and Cohen, 477, all as in n. 112.
145 The best known book in English is W.C. Booth, The Rhetoric of Fiction, Chicago, 1961, which introduces the concept of implied author as a fictional stand-in for the author and the orchestrator of the narrative as a whole. Booth's attempt to rid literary studies of the moralism that blames the author for any unacceptable ideological statement in the text was important in its time, but only half successful. The unified nature of his implied author only displaced the problem from "real" author to implied author, still not managing to account for heterodiscur-sive strands and ideological disjunctions.
Exceptions to this unified model not influenced by Barthes or Bakhtin are G. Prince's speech act theory of narrative, Narratology: The Form and Function of Narrative, Berlin, New York, Amsterdam, 1983; and A. Banfield's linguistic theory of narrative, Unspeakable Sentences, London, 1982. G. Genette, Narrative Discourse: An Essay in Method (Paris, 1973), trans. J.E. Lewin, Ithaca and London, 1980, begins to differentiate between narration and focalizatiou (a term that replaces the muddled traditional term of point ofview, which isfarless technical in narratology than in visual analysis) but fails to account for localization within semiotics and thereby falls back into the older privileging of the narrator.
204       THE ART BULLETIN JUNE 1991 VOLUME LXXTJI NUMBER 2
4 Artemisia Gentileschi, Susanna, 1610. Pommersfelden, Schloss Weesenstein
image, does not really help in understanding the various signs at work in an image.
Attempts to atomize the informational sources of narrative in view of a semiotic conception of texts may be more useful. One such attempt distinguishes between three narrative agents: the narrator or speaker, who is the source of the utterance, the focalizer or source of the vision presented in the utterance, and the actor or agent acting out the fabula (the sequence of events presented).14*' This model allows for integration of two important views: the idea that signs are organized, and the possibility of "the difference within."147 Precisely because the narrators of a text hold discursive power, they are also able to embed the vision of somebody else into that text, as in the phrase, "She saw that he noticed that the lipstick on his collar had not escaped her." Here one narrator conveys three views, nested like Russian
""The following is drawn from M. Bal, Narrate/log}/: Introduction to the Theory of Narrative, Toronto, 1985, further elaborated in Death and Dissymmetry: The Politics of Coherence in the Book of fudges, Chicago, 1988. This theory is used for visual analysis in Reading "Rembrandt."
147 This phrase is used by B. Johnson in her A World of Difference, Baltimore, 1987.
5 Rembrandt, Susanna Surprised by the Elders, 1647. Berlin-Dahlem, Gemaldcgallerie
dolls, and each based on signs positioned on different levels: in the fabula, two actors are confronted, one of whom, the woman, constructs on the basis of a sign of facial expression that the other, the man, has in turn constructed on the basis of the sign of her own facial expression, a third construction, the sign of lipstick on his collar, which he may up till then have been unaware of himself.
In this structure of embedding, one voice conveys in a single discourse a visual dialogue by images that seems typical of language. In fact, something like the literary concept of Free Indirect Discourse, where not just the vision but even aspects of the voice of another figure are embedded in the narrator's monologue, might seem impossible in visual images where this hierarchical ordering does not apply and where all elements of the configuration of subjects are present on the one surface. Yet this is a deceptive unification of the status of the various elements. Imagine the story of Susanna surprised by the Elders. In a traditional painting of the subject by, say, Rubens, the figure of the naked woman is presented to the viewer, exposed for sight and delight (Munich, Alte Pinakothek). The elders, the represented focalizers, provide a position from which the viewer can interpret the woman's body; in other words, they offer a point of identification. The woman, either looking away or looking at the viewer in compliance, does not counter the voyeuristic position offered. This is, then, a simple, one-strand narrative. In contrast, Artemisia Gentileschi's Susanna (Fig. 4) is claimed by Mary Garrard to suggest discomfort at the situation of voyeurism, implying a critique of it.148
148 See M.D. Garrard, "Artemisia and Susanna," in Broude and Garrard (as in n. 110), 147-171. See also Garrard's monumental Artemisia Gentileschi: The Image of the Female Hero in Italian Baroque Art, Princeton, 1988, reviewed in this journal by G. Pollock (Lxxu, 1990,499-505).
SEMIOTICS AND ART HISTORY 205
What are the narrative signs of this critique? To give one example, Garrard mentions the uncomfortable stone bench that Susanna is sitting on. As opposed to the lush foliage that, alluding to the garden of earthly delight, traditionally represents the cheerful focalization of the Elders, this stone bench concretizes the experience of the threatened woman. The structure of focalization in this painting is more complex than in the Rubens: the men exchange glances, "telling" each other visually their scheme; Susanna looks away, refusing to interact with her assailants; and the hard and unpleasant features of the Elders counter any tendency to identify with their viewing position: spying on Susanna. Instead, Susanna's distressed look, although directed neither at the viewer nor at the men, so as to preclude the idea of compliance, suggests that the unpleasant men we see are as she sees them: Susanna, here, focalizes her assailants, and the result of that focalization is signified to the viewer, before the latter reaches their focalization, which in its mutuality excludes the viewer. Thus, the single "discourse" of voyeurism, so prevailing in the Western tradition, contains not a narrative of voyeurism but a counter-narrative of anguish and vulnerability.
Between Rubens's and Gentileschi's Susannas we can read Rembrandt's Susanna in Berlin, where an Elder sits on a throne while the other busies himself with Susanna differently. On the one hand, the female nude is vulnerable and young; in iconography she recalls the Venus Pudica and hence evokes eroticism. She looks at the viewer, who can read her as helpless and calling for sympathy, or as an object of sadistic lust. But the Elder who with one hand undresses her, with the other hand signals the direction of his look, which literally overlooks Susanna's body. With the help of another iconographic reference, to Diirer's Melencolia I, we can construct a narrative that complicates and critiques the theme of voyeurism. This narrative refers not so much to the woman's focalization as to the man's failure at focalization. Once a viewer identifies with this man, he or she is struck by the failure of looking and the resulting paralyzing melancholy. The image presents three figures looking in different directions and in different modes, breaking up the unity that would cast this work in the too-encompassing category of "the nude."
What this view of narrative suggests, then, is that the act of looking at a narrative painting is a dynamic process. The viewer moves about the surface to anchor his or her look at a variety of positions. These positions are not just alternatives, as a pluralistic view would have it, but are interrelated and embedded.m Whichever position one chooses to endorse in the Rembrandt, the other two cast their shadow over it. If one looks with the
man in the upper right, comfortably sitting and watching the scene, one's eye travels with him to his colleague, and there as that embedded position is occupied, one fails to see. If one responds to Susanna's look and her experience of having been attacked, one then becomes aware that the man undressing her is also emphatically not looking at her. In Bakhtin's terms, the discourse of melancholy and the discourse of rape are in dialogue with each other and with the discourse of voyeurism. But that view does not account for the embedding of the former two themes in the overall concept of the latter. Therefore, we would rather contend that both melancholy and rape, as well as the opposition between these two, are embedded within the discourse of voyeurism, embodied by the man at the upper right.
The semiotic nature of this model emphasizes the sign-status of the elements involved in this reading. As we abandon the illusion of unity, the focused eye of the active Elder, engaged in an act of looking unrelated to the nude body, stands out as a sign of a whole range of meanings, which include ideas about melancholy as an index of artistry. And these meanings are close to losing any connection with the Susanna theme. In competition with this cluster of meanings stands the allusion to the Venus Pudica made by the gesture of Susanna's left hand. But this sign is in turn overruled by the slight shift backwards of her hand, turning the sign of statuesque display into that of a narrative agency pushing away the threatening elder.
This example also demonstrates that the atomization of narrative within a discursive order—the integration of pluralization with embeddedness—allows for an account of ideology that breaks away from monolithic readings without falling back into an "innocent" relativism. Thus those paintings by Anselm Kiefer that have disturbed some viewers because of their allusions to Nazism, while being hailed by others as a critique of fascism, can be seen as a debate with Nazism. Kiefer's dialogic approach refuses to silence, ignore, repress, and thereby conserve fascism today. By integrating another partner in this debate, the tradition of linear perspective, that emblem of realism and objectivity, Kiefer's works also signify the complicity of art with politics. The suggestion is that perspective and the scientific pursuit it stands for, collaborate with fascist tendencies, which an obliteration of the Nazi past facilitates. The resulting narrative presents a highly complex account of both fascism and painting, within which the various possible focalizers take their share in the production of meaning. The narrative of the paintings is constituted by the tensions between these focalizers.iS"
Narrative semiotics does not merely identify subjects
Bakhtin's radical view of heteroglossia, as well as Derrida's account of polysemy in the concept o( dissemination, is often construed as liberal pluralism. This is a mistaken interpretation of both theorists' positions. On Bakhtin's critical edge, see Hirschkop, Bakhtin (as in n. 141); on Derrida's disavowal of the pluralistic posilion, see Limited Inc.
On the question of Kiefer's position toward recent Cerman history, see A. Huyssen, "Anselm Kiefer: The Terror of History, the Temptation of Myth," October, xlviii, 1989,25-15.
206       THE ART BULLETIN lUNE l'Ml VOLUME LXXI1I NUMBER 2
within the image and their relations of embedding; it also allows one to specify the nature, place, and effectivity of each subject's agency. It provides a possibility for reading images against the grain of the alleged opposition between discourse and image by interpreting elements as signs of negation, as in the Elder's not-seeing; as signs of syntactical connection, as in the bench in the Gentileschi, shifting focalization from the Elders to the victim; as signs of causality, like the anguished look of Rembrandt's Susanna as caused by the physical attack upon her person. Most important, narrative semiotics provides insight into visual narrative, as distinct from analysis of visual allusions to verbal narratives.
It is crucial to keep in mind that narrative is not a one-sided structure. Address, the ways in which a viewer is invited to participate in the representation, is perhaps the most relevant aspect of a semiotics of subjectivity. According to the linguist Emile Benveniste, language inscribes the subject of discourse as the implicit "I" who speaks. Certain linguistic categories, such as "I" and its correlate, "you," but also other elements such as pronouns and adverbs like "here" or "yesterday," have no referential value but only mean in terms of the discourse itself. Thus the subject of discourse is defined by these deictic words, whether or not s/he is the same person as the speaker who conveys the speech, say, the narrator. The noncoincidence between the speaker and the subject of the discourse is precisely the condition of possibility of narrative.1''1 The equivalent of "I" in a painting is often the figure whose act of looking is represented, thereby suggesting participation in a story: "1" takes the shape of an "eye." But that story involves the onlooker as well. If Manet's OSympia scandalized its contemporary viewers, we now say it was because the woman participated too fully in her own display; rather than contenting herself with being the "third person" whose body was objectified in an impersonal narration for the sake of the onlooker, this woman looks actively at the viewer, so much so that her objectifkation is nullified, and the viewer, who is now no longer the "I" who can take possession of the woman, is offered the position of the "you" hailed by the woman and held accountable for his act of looking. For any "I" implies a "you" whom it addresses. The French philosopher Louis Althusser claimed that this address, this "hailing," constitutes the subject in ideology: it forms the subject as what the "ideological state apparatuses" wish the subject to identify with.1''2 However much autonomy a particular viewer may have (or assume to have) in front of a painting,
according to this theory subjectivity is always produced at least by the interaction between the "I" of the work and the "you" this "I" addresses.1''1
8. History and the Status of Meaning
The preceding sections have each presented questions of visual art from a semiotic angle, and in each case our readers may have continued to worry: but what about history? How do these approaches account for the historical status of the images, their origins and their original intentions, and the ways they were interpreted by their contemporary audience? As we announced in the introduction and emphasized thereafter, we are less sure than many colleagues about the possibility of reconstructing those origins and the relevance of the attempt to do so. In addition to our previous remarks on this topic, we wish to point out that semiotics involves three issues that complicate the historical search: intertextual-ity, polysemy, and the location of meaning. In this section we conclude our presentation by briefly outlining the problems these issues entail for the certainties and relevances of the historical search for origins.
The term inlcrtextuality was introduced by the Soviet philosopher of language Mikhail Bakhtin. It refers to the ready-made quality of Linguistic—and, one can add, visual—signs, that a writer or image-maker finds available in the earlier texts that a culture has produced. For art historians, this concept may seem to overlap with that of iconographic precedent. To a large extent, it does. Yet in crucial aspects it does not. Iconography seems nothing but the examination of just this reuse of earlier forms, patterns, and figures. However, three features, and crucial ones, distinguish intertextuality from iconography. In the first place, iconographic analysis tends to take the historical precedent as the source that virtually dictates to the later artist what forms can be used. By adopting forms from the work of a certain artist, the later artist declares his allegiance and debt to his prestigious predecessor.154 Michael Baxandall has already convincingly proposed to reverse the passivity implied in that perspective, and to consider the work of the later artist as an active intervention into the material handed down to her. This reversal, which also amounts to a deconstruc-tion of the relation between cause and effect, already challenges the idea of precedent as origin, and thereby makes the claim of historical reconstruction problematic.155
A second difficulty in the juxtaposition of intertextual-
151 E. BenvenLsle (as in n. 117).
,sl L. Althusser, Lenin ami Philosophy and Other Essays, trans. B. Brewster, New York, 1971, and For Marx, trans. B. Brewster, London, 1977.
1 ' The expression "reading as a man" is the equivalent of "reading as a woman," put into currency by J. Culler in his famous section of this title in On Deconstruction, 43ff. Culler distinguishes three moments in feminist criticism that may have their parallel in art history: identification with the roles of women, reevaluation of the traditionally female
roles, and deconstructing the oppositions on which these value-systems are based. In a mitigated manner, all three approaches can be seen in G. Saunders's The Nude: A New Perspective, London, 1989.
IMThis is how H. Perry Chapman explains Rembrandt's use of forms taken from Raphael and Titian in his self-portraits of the middle period. See Chapman (as in n. 68).
Baxandall, 58-62.
SEMIOTICS AND ART HISTORY
207
ity and iconography is the place of meaning. Icono-graphic analysis frequently avoids statements about the meaning of the borrowed motifs. To borrow a motif is not a priori also to borrow a meaning. The concept of intertextuality, in contrast, implies precisely that: the sign taken over, because it is a sign, comes with a meaning. Not that the later artist necessarily endorses that meaning; but she will inevitably have to deal with it: reject or reverse it, ironize it, or simply, often unawares, insert it in the new text. This is how Mary Garrard uses precedents in her basically iconographic analysis of Artemesia Gentileschi's Susanna.1* Thus, referring to Diirer's Melencolia I in the pose of the aggressive elder in his Susanna in Berlin, Rembrandt cannot help bringing in the quite unsettling meaning of that precedent, suggesting that illegitimate and abusive looking paralyses the transgressor.
A third difference resides in the textual character of intertextual allusion. By reusing forms taken from earlier works, an artist also takes along the text out of which the borrowed element is broken away, while also constructing a new text with the debris. Reusing a pose employed earlier in a self-portrait, Rembrandt inserts the discourse of self-portraiture into his Bellcma of 1633 (New York, Metropolitan Museum of Art).157 The new text, say, a mythography, is contaminated by the discourse of the precedent, and thereby fractured so to speak, ready at any time to fall apart again. The fragility of the objectifying, distancing device of mythography is displayed by this taint of "first-person" subjectivity. In Benveniste's terms, the historical narrative is infected by subjective discourse. Such a view has obvious consequences for the interpretation of this painting in terms of gender.
One can push this reflection of the implications of intertextuality further, in the direction of the kind of self-reflection advocated by Habermas.1'" For the art historian, like any viewer of images, cannot but bring to the pictures her own legacy of discursive precedents, and reading images entails the inevitable mixture of these signs with those perceived in the work. The allusion to Melencolia, for example, occurred to us for reasons that Habermas would wish us to explore further, and it is immediately obvious that they have to do with contemporary gender issues. This addition from the present is emphatically not to be taken as a flaw in our historical awareness, or a sign of failure to distance ourselves from our own time, but as an absolutely inevitable proof of the presence of the cultural position of the analyst within the analysis, which, from a semiotic point of view, is not surprising. To take that presence into account makes the
analysis, in fact, more rather than less historically responsible.
This leads to the second issue, that of polysemy. Since readers and viewers bring to the images their own cultural baggage, there can be no such thing as a fixed, predetermined, or unified meaning. Attempts to fix meaning provide, in fact, the most convincing evidence for this view. The field in which struggles over meanings are fought is a social arena where power is at stake. A good example of this mechanism is allegory, the interpretation of, say, a mythical story and all its representations as referring to something outside itself. On the one hand, allegory demonstrates the fundamental polysemous nature of signs. If stories can mean something entirely outside of themselves, then there is no constraint. This freedom is viewed positively by Paul de Man, for example.15" We wish to express some caution, though, which is warranted by the same cases that come up again and again: the allegorical interpretations of mythical stories of rape as "really" dealing with tyranny and the establishment of democracy. Intertextual analysis will bluntly refuse such abdication of the meaning imported by the sign: if rape means political tyranny, then the bodily, subjective experience of the woman raped cannot be divorced from the politics at stake. The myth of Lucretia, then, is allegorical, but with a vengence. The alios of allegory is, after all, not only "other," but also "within."
This problem we have with allegory is, in turn, allegorical for a larger problem implied by polysemy. For the dynamism of signs implied in this view might be mistaken as an abdication of the scholarly position altogether. Derrida's concept of dissemination is the most radical endorsement of the view that no interpretation can be privileged over any other.11'0 In spite of the attraction of this idea, especially as a corrective to the remnants of positivism still pervasive in the humanities, we wish to advocate some caution here again. For the play of interpretation is surely not entirely free, or else there would be no cause for chagrin about power relations and exclusions in academic practice. In agreement with Wittgenstein's notion of a language game, semiotics proposes to see signs as active, and requires them to be deployed according to rules, and public. A sign, then, is not a thing but, as we have said, an event that takes place in a historically and socially specific situation. Sign-events occur in specific circumstances and according to a finite number of culturally valid, conventional, yet not unalterable rules, which semiotics calls codes. The selection of those rules and their combination leads to specific interpretive behavior. That behavior is
* Garrard's work has been mentioned in n. 148.
157 The reference is put forward by G. Schwartz, Rembrandt: His Life, His Paintings, Harmondsworth, 1985.
IS" Habermas (as in n. 4). His concept of self-reflection has been demonstrated to lead easily to an idealistic "purity" of power relations. See Lyotard (as in n. 4).
• P- de Man, Allegories of Reading: Figura! Language in Rousseau, Nietzsche, Rilke, and Proust, New Haven, 1979. De Man develops his argument around the root alios, other, and thus allegory becomes itself an allegory for the acceptance of otherness within. In the wake of Walter Benjamin, a whole school of allegorists has written positively about allegory, as has Owens in the seminal article cited earlier (as in n. 108).
'""Derrida (as in n. 17).
I
208       THE ART BULLETIN IL'NE I1 VOLUME I.XXIII NUMBER 2
socially framed, and any semiotic view that is to be socially relevant will have to deal with this framing, precisely on the basis of the fundamental polysemy of signs and the subsequent possibility of dissemination. In the end, there is no way around considerations of power, inside and outside the academy.
Mieke Bal's books include Reading 'Rembrandt': Beyond the Word Image Opposition, Cambridge and New York, 1991, and Death and Dissymmetry: The Politics of Coherence in the Book of Judges, Chicago, 1988 [Department of Art and Art History, University of Rochester, Rochester, N.Y. 1.4627 and Instituut voor Algemene Literatuurwetenschap, Spuistraat 110, 1012VT Amsterdam, The Netherlands].
Norman Bryson's latest book is Looking at the Overlooked: Four Essays on Still Life Painting, Cambridge, Mass., 1990, and he has edited the anthology Calligram: Essays in New Art History from France, Cambridge, 1988 [Department of Fine Arts, Harvard University, Cambridge, Mass. 02138].
Frequently Cited Sources
Bal, M., Reading 'Rembrandt': Beyond the Word-Image Opposition, Cambridge and New York, 1991.
Barthes, R., Image—Music—Text. ed. and trans. S. Heath, New York, 1977.
Baxandall, M., Pattertis of intention: On the Historical Explanation of Pictures, New Haven and London, 1985.
Culler, J., Framing the Sign: Criticism and Its Institutions, Norman, Okla., and London, 1988.
Derrida, J., Limited inc., Evanston, 1988.
Preziosi, D., Rethinking Art History: Meditations on a Coy Science, New Haven and London, 1989.
Roskill, M., The Interpretation of Pictures, Amherst, 1989.
Silverman, K., The Acoustic Mirror: The Female Voice in Psychoanalysis and Cinema. Bloomington, 1988.
White, H., Tropics of Discourse, Baltimore, 1978.