From the Critique of the Iconicity Critique to Pictorality Contents

Lecture 3: From the Critique of the Iconicity Critique to Pictorality

In order to show why Eco, Goodman, and others were wrong in their classical critique of iconicity, we will pursue a close reading of Peirce, but we will interpret his text in accordance with more recent findings in cognitive and perceptual psychology. We will arrive at the conclusion that there are two very different kinds of iconicity, which we will call primary and secondary iconicity. Even so, pictorial iconicity has its peculiarities, which we will also try to elucidate. We will also consider to what extent the linguistic model may still be helpful, and in which respects it is misleading.

Contents 3.1. Reading Peirce ecologically: From iconicity to iconic signs 5 The night of all iconicities 7 Iconicity in its tripartite structure 10 The ground as abstraction and relevance 14 On different types of types 17 Summary 21

3.2. Iconicity regained. The logic of the Lifeworld 22 Regressive iconicity and the sign function 23 Beyond the symmetry argument 24 Primary and secondary iconic signs 27 The criterion of independence 33 Summary 35

3.3. In the looking-glace, somewhat less darkly. Eco’s three critiques 35 This nose is not a nose. The case for conventionality 36 Saccharine by any other name… A plea for proportionality 41 Mirror, mirror, on the wall…What do you signify? 45 Summary 53

3.4 On the way to Pictorality: The case for resemanticisation 53 The conventional residue of iconic signs 54 The picture is no image. The Annigoni case 59 The process of resemanticisation 67 The Husserlean triad: picture thing, picture object, and picture subject 71 Pictorial consciousness and the scale of typicality 74 Summary 80

3.5 Perspectives, aspects, and “impossible pictures” 82 Wittgenstein’s rabbit as a Saussurean “form” 84 Possible depictions of impossible worlds 89 The Duck-rabbit and other impossibilities 94 From perceptual to pictorial perspective 102 The limits of the perspective metaphor in language 110 Perspective in the extended sense in language and pictures 112

Summary 117

References 117

So far, it has been argued that there is a place (although hardly prepared beforehand), as well as some precedents, for such a speciality as pictorial semiotics (first lecture), and that pictures are signs, which is not true of all meanings (second lecture). Notably, signs were shown to be meanings which are differentiated from the point of view of the subject, and which connect something that is directly present and non-thematic with something else that is only indirectly present and thematic. This is a definition, which, by combining criteria formulated by Piaget and Husserl, goes beyond both Saussure and Peirce. It remains to be shown, however, that pictures are iconic signs, but also that they are a peculiar kind of iconic signs. Unlike most others domains of study nowadays integrated into semiotics, the semiotics of pictures has hardly any precedents outside of semiotics proper: if we take pictorial semiotics to be involved with the study of pictorial signs per se, of some general property, more peculiar to pictures than iconicity, which may be termed pictorality, or picturehood, and if we suppose it to apply empirical methods to this study, then it certainly is a novel endeavour, far more so than linguistics, more, in fact, than literary semiotics, film semiotics and even the semiotics of architecture, all of which have known some more or less elaborate theoretical approaches before the coming of semiotics. This is so because the only other domain ever devoted to pictures, art history, always has been, and mostly continues to be, fascinated by the singularity of the individual work of art, which explains (without necessary justifying) the fact that it tends to shun all kinds of conceptual analysis. Indeed, pictorial semiotics is something of a newcomer within semiotics itself, for although Peirce gave us one of our most important theoretical tools for understanding the picture sign, i.e. the concept of iconicity, he himself hardly took any interest in pictures, and even the sub-category which he introduced to take care of them, the image, on closer scrutiny does not appear to adequately characterise pictorial iconicity. As for the other founding-father of semiotics, Saussure, he repeatedly insisted on linearity being one of the peculiar properties of verbal language, and at least once he opposed this feature to the multi-dimensionality found in pictures (cf. Saussure 1974:39); but though he was right in pointing to the differences in the ways in which signs are organised in verbal and pictorial “texts”, multi-dimensionality is not specific to pictures, and, on one

interpretation, it may even be found nor Saussure, but the phenomenolo- in verbal language. gy of Edmund Husserl, as developed The question concerning the by, among others, Aron Gurwitsch, specificity of the picture sign, as com- Alfred Schütz, and Maurice-Mer- pared to other signs, and as related leau-Ponty. In particular, this means to its sub-types, in contrast, has been that I will take the Lifeworld, also a central issue of general semiotics known as the world-taken-for-grant- since it rebirth in the 1960ies. It does ed, the common sense world, or, in involve the Peircean notion of iconic- the adaptation of the psychologist ity, less as it has been safeguarded by James Gibson, as ecological phys- the true Peirceans, than as it emerges ics, as the foundation for all possible from half a century of criticism, by meaning. It also imposes particular philosophers such as Bierman and requirements on us to explicate basic Goodman, as well as semioticians notions such as those of sign, iconic- such as Eco and Lindekens; and ity, and pictorality. In many ways, then rehabilitated, by, among others, ecological semiotics is closer to, and Groupe µ, and the present author; and more compatible with, the basic ten- it also concerns the Saussurean idea ets of Peirceans semiotics than with of the way meanings may be organ- the Saussurean brand, in particular ised, again as it was put to confused, as the latter was transposed outside and confusing, uses by Eco and oth- language by French structuralism. ers, and then completely reconceived In this framework, many of the con- in the light of the findings of percep- cepts of Saussure-inspired semiotics tual psychology, in particular by the retain their import, but only on a sec- present author. ondary level. However, one basic notion of Saussurean linguistics, hardly 3.1. Reading Peirce taken into account by French struc- ecologically: From turalism (but certainly by the Prague school), the notion of pertinence or iconicity to iconic signs relevance, will turn out to be a fun- The point of departure of the present damental ingredient in our interpre- approach to pictorial semiotics, tation of iconicity. It is in fact rel- variously known as “the Swedish evance, together with the notion of school”, “the ecological school”, the Lifeworld, which will turn out to and “the phenomenological school” (cf. Saint-Martin 1994; Carani 1999; For those in the know (others Nordström 2000), is neither Peirce may ignore this remark), I would like to add that I am not interested in the An argument could however be transcendental and constitutive aspects made for Lessing being the real pioneer of phenomenology, but only in what of pictorial semiotics. Cf. Wellbery 1984; Husserl himself calls “phenomenological Sonesson 1988:105ff. amd Lecture 4. psychology”. save the notion of iconicity from the will then see how the idea of a visual criticism directed at it from outside syntax was exaggerated into the idea orthodox Peircean quarters. of a double pictorial articulation, Ecological semiotics is, un- and, at the other extreme, was dis- like Husserlean phenomenology, but solved into the conception of density similarly to Gibson’s study of eco- and repleteness; and we will intro- logical physics, an empirical science. duce resemanticization as a peculiar If, along with Gibson (1978:228), feature of pictorial, as against other we describe pictorial semiotics as kinds of visual and iconic, meaning. “the science of depiction”, where The questions of what distinguishes the latter is considered to be a pecu- different picture categories, such as liar mode of conveying information, photographs and drawings, adver- its purview will involve, at the very tisements and caricatures, or picture least, a demonstration of the semiotic post cards and posters, will be taken character of pictures, a study of the up in a later lecture. peculiarities which differentiate pic- In the following, I will first de- torial meanings from other kinds of lineate a particular interpretation of signification (particularly from other Peirce’s division of signs, and then visual meanings, and/or other mean- proceed to elucidate the ways in ings based on iconicity, or intrinsic which the criticism levelled against motivation), and an assessment of iconicity, as well as against the com- the ways (from some or other point mon sense notion of picture, which of view) in which the several species it includes, can be eluded. This argu- of pictorial meaning may differ with- ment will critically involve a very ex- out ceasing to inhere in the category plicit notion of sign, which is absent of picture. In the following, we will from Peirce’s work, but is suggested have a look, first, at the notion of ico- by Husserlean phenomenology in nicity, not only to show that it exists, combination with the psychology and is involved in pictorial meaning, of Piaget; and it will also require us but that there are at least two kinds to introduce a concept of relevance, of iconicity, and that pictures must which is foreshadowed, rather am- be attributed to the first sub-category. biguously, in the musings about the The paradoxical nature of the kind of “ground” which Peirce sometimes iconicity ascribed to picture, prima- entertained, but is more clearly sug- ry iconicity, forces us to reconsider gested by the Saussurean notion of the picture sign as such, taking our pertinence. I will then indicate one inspiration in a very direct way from further way in which Peirce’s con- Husserlean writings. Fortunately, the ception of iconicity has to be amend- groundwork for this analysis was ed in order to account for the kind of laid already in the second lecture. We iconicity (which I will call primary

iconicity) instantiated in, among ogy, which tend to show that these other things, picture signs, and I will conclusions are unfeasible. More go on to approach some even more importantly, however, I have also peculiar traits of pictorality, which is suggested that the arguments against were we will make contact with the iconicity were mistaken, mainly be- Saussurean tradition in semiotics, as cause they construed language and well as, more intimately, with Hus- pictures, as well as the world of our serl’s seminal work on pictorial con- experience, i.e. the Lifeworld, in a sciousness. fashion which is incompatible with During the second half of the our empirical knowledge, i.e. with last century, the claim that there can that which we have good reasons to be no iconic signs came from two believe to be true about the world. rather different quarters. Philoso- Rather than making iconic signs phers like Bierman and Goodman, semiotically uninteresting, contrary only the first of which explicitly re- to what is suggested by these cri- fers to Peirce, started out from logi- tiques of iconicity, it is only by the cal considerations, together with a recognition of the reality of iconic set of proto-ethnological anecdotes, motivation that iconicity is opened according to which so-called primi- up as a domain for semiotics. This tive tribes were incapable of inter- is so at least for three reasons: first preting pictures; out-right semioti- of all, since an iconic sign is indeed cians such as Eco and Lindekens, similar to what it represents, it may on the other hand, wanted to show be used to manipulate and transform that pictures conformed to the ideal in numerous ways the ideas we hold of the perfect sign, as announced by about its referent, which is what gives Saussure, by being as arbitrary or rise to visual rhetoric (cf. Sonesson conventional as the sign studied by 1990; 1994b; 1996a, b, c; 1997a). In the “most advanced” of the semiotic the second place, a little investiga- sciences, general linguistics. Since tion will show us that iconicity may then, the question has largely gone inhere in signs in several different out of fashion, but the results of those ways, the main two of which we will disquisitions have, rather undeserv- call primary and secondary iconic edly, been taken for granted by later signs, but the sub-species of which researchers. In my own work on ico- may be very numerous (cf. Sonesson nicity, which dates from the period 1993a; 1998a, b, in press a). Finally, of low tide in the debate (Sonesson and perhaps most importantly, iconic 1989a, 1992a, b, c, 1993a, b 1994a, signs presuppose an uneasy equilib- b, 1995a, 1996d, 1997b, 1998b, rium between being the same and 2000a; in press a), I have quoted evi- being different which may explain dence from psychology and ethnol- that, to all appearance, pictorality

(and perhaps other iconic represen- of computer programming, or in cog- tations) is a late-comer in the proc- nitive psychology, where “iconic” ess by means of which human beings and “verbal codes” are opposed to distinguish themselves from other each other (e.g. Kolers 1977). Con- animals, and which may account for trary to the latter usage, iconic signs, the unique feats of the human race, in the sense of semiotics, appear just as much as verbal language (cf. in any sense modality, e.g. in audi- Donald 1999; Sonesson 2003a; in tion, notably in verbal language (not press d, e). Nevertheless, it will be only onomatopoetic words, but also useful to start out by revisiting the in the form of such regularities and old iconicity debate, not, however, symmetries which Jakobson 1965a,b before having returned to the origi- terms “the poetry of grammar”) and nal definition of the concept made by music (cf. Osmond-Smith 1972). Charles Sanders Peirce. Nor does iconicity in the Peircean sense have anything spe- The night of all iconicities cifically to do with what in art his- The iconicity debate has not been tory, following Panofsky, is called immune to some simple confusions. “iconography”, which is potentially It should be clear that, in semiotics, relevant to all pictures, but precisely the term “icon” is not normally to be in their non-iconic sense, as guided taken in its most common religious by an (explicit) reading program. In and art historical acception, to refer a parallel fashion, not all visual signs to a pictorial representation of per- are iconic in the semiotic sense; in- sons or events derived from the sa- deed, many icons found in computer cred history of Christianity, which is programs, as well as a great amount normally used as an aid to devotion. of visual signs appearing in public In fact, icons in the religious sense space, are actually aniconic visual are not particularly good instances signs. Finally, the sense in which of icons in the semiotical sense, for iconicity is discussed here is quite they are, as Uspenskij (1976) has distinct from that appearing in the shown, subject to several conven- More will be said about tions determining the kind of per- iconography and iconology in Lectures spective which may be employed, 4 and 5, when discussing the parallel to and the kind of things and persons Barthes often noted (which is really not so which may be represented in differ- straight-forward). ent parts of the picture. It seems to Many semioticians also are guilty of such a confusion of these two quite be less clear that the term is not to different senses attributed to the term be used to refer to all things visible, ‘iconic’: thus still in Eco 1999:100, in or to everything whose elements are spite of admitting his error in 1998: 10; graphically disposed, as in the jargon 1999:340. And all through Vaillant & Castaing in press. expression “cultural icon” – where nicity simply become irrelevant. it seems to refer to just about any- Some other readings of Peirce, thing (no matter the sign character which I am going to make, are more or the sense modality) which occu- controversial. Peirce is often taken pies a central position in a (popular to say that, given the class of all sub)culture. existing signs, we can make a sub- Sometimes, criticism levelled division into three sub-classes, con- against Peirce’s conception of ico- taining icons, indices, and symbols. nicity rest on obvious misinterpre- But it is of course easy to show that tations: for instance, iconicity is not many signs may have iconical, in- limited to the rendering of the ap- dexical, and symbolic features at the pearances of the ordinary perceptual same time. This seems to mean that, Lifeworld, as many theories would at least as applied to signs, iconicity, have it (as in many passages by Eco, indexicality, and symbolicity do not as well as in the definitions used by separate things, such as signs, but the Greimas school and Groupe µ), relationships between things, such but includes much more abstract re- as parts of signs. While Peirce nev- lations of similarity. In other cases, er seems to pronounce himself on the criticism (e.g. that of Bierman this issue, he has said that the per- and Goodman) turns out to be valid fect sign should include iconical and on one, but not another, possible, in- indexical as well as symbolic traits. terpretation of what Peirce’s says: for We may not care whether there are instance, it is often not clear wheth- such things as perfect signs, but this er Peirce wants to say that there are affirmation clearly implies (whether three kinds of relationships, iconic- Peirce was aware of it or not) that ity, indexicality, and symbolicity, iconicity, indexicality and symbolic- which each one alone are sufficient ity are relationships rather than ob- to transform something into a sign; or jects. This is why I will conceive of whether he is rather suggesting that, them in terms of “grounds” (which I among those things we call signs, for will take to mean foundations). some other reason (the existence of It should also be realised that a sign relation), some also have the there is an important structural (but property of iconicity or indexicality. of course not binary) argument in If we accept the second alternative, Peirce’s whole conception. He clear- for which some indirect arguments ly tells us there are three kinds of may be given, much criticism of ico- signs (or rather, sign relationships), at least from the point of view that I exclude “symbolicity” under the he has chosen (i.e. what motivates supposition of it being identical to the sign relation (but we will later see that this may the relationship between expression not be all that it is). and content). As far as I understand, this implies (again, whether Peirce tion regrettable, not only because I was aware of it or not) that we are am not at all convinced that the no- not free to add further kinds of signs tion of iconicity is incoherent (and, using quite different criteria (as for indeed, Goodman 1968 claimed the instance Sebeok and Helbo have same thing about the common sense done); nor to interpret some of these notion of picture), if we do not try to sign types in such a narrow way (e.g. understand it taking the picture as our indices as motivated by causality, prime example, but also because it or icons as rendering perceptual ap- means giving up what I believe is the pearances, the first of which is very positive part of the Peircean, as well common, and the second of which as Saussurean, heritage in semiotics, characterises the Greimas school and the search for a system: it means dis- Groupe µ) that the domain of possi- cussing picture signs without relat- ble signs relations is not exhausted. ing them to other iconical signs (or Two ways stand open to us: either whatever you want to call them), nor we accept that there can only be to signs in general. In short, it means three kinds of relationships between giving up semiotics as a comprehen- expression and content in a sign, and sive enterprise (or a “totalizing” one, then this will severely restrict our as the postmodernists were wont to options for choosing among the nu- say, a term which we in our post- merous definitions of the three sign postmodern age should be able to re- relationship given by Peirce. Or else claim in a positive sense). The very we have to demonstrate that, from point of semiotics, I take it, is to con- this same point of view, further sign tinuously relate the kind of signs we relationships may be established. are investigating to all other kinds of Here, I have decided to follow the signs. Semiotics first and foremost first path. is a comparative science. In will be noted that Klinken- berg (in press) and Blanke (in press), Iconicity in its tripartite both take a stand on this latter issue, structure but by refusing the very conundrum: In Peircean parlance, to put it sim- referring to the bewildering variety ply (but we will later see that this is which the term “iconicity” covers in all too simple a manner of putting the Peircean tradition, if we are to it), an icon is a sign in which the believe Eco’s first iconicity critique, “thing” serving as expression in they decide to use the term simply It must however be said, as I noted to mean pictorality. I find this posi- above, that, contrary to other domains of semiotics, pictorial semiotics is justified In fact, some predecessors of as such, since there is no comparable Peirce, such as Degérando, thought that to discipline otherwise interested in pictures be possible. at a general level. one respect or another is similar to, tive” meaning out of the abstract base or shares properties with, another structure, Greimas & Courtés (1979: “thing”, which serves as its content. 148, 177) similarly identify iconicity In fact, according to Peirce, there are with perceptual appearance. In fact, two further requirements: not only however, not only is iconicity not should the relation connecting the particularly concerned with “optical two “things” exist independently of illusion” or “realistic rendering”, but the sign relation, just as is the case it does not necessarily involve per- with the index, but, in addition, the ceptual predicates: many of Peirce’s properties of the two “things” should examples have to do with mathemat- inhere in them independently. ical formulae, and even the fact of Contrary to what is suggested being American, as in the Franklin by Groupe µ: s (1992) quotation from and Rumford example, is not really Dubois’ dictionary, iconicity, in the perceptual, even though some of its Peircean sense, is thus not limited to a manifestations may be (cf. Sonesson resemblance with the external world 1989a: 204ff). It is also common to (“avec la réalité extérieure”). When confound iconicity and picturehood, conceiving iconicity as engendering when in actual fact, if we rely on a “referential illusion” and as forming Peirce’s definition, pictures consti- a stage in the generation of “figura- tute only one variety of iconicity and are not even supposed to form the It should be noted that I will be best instances of it. On the contrary, avoiding peculiarly Peircean terms in the as we shall see, something additional following, as long as no harm is done by in necessary to account for the picto- that procedure: I will use ‘expression’ for what Peirce calls ‘representamen’ and rality of pictures. ‘content’ for his ‘object’: more precisely, If iconicity is part of a (trinary) I will roughly identify ‘immediate structure, then it cannot be discussed object’ with ‘content’ and ‘dynamical object’ with ‘referent’, though it might outside the framework of Peirce’s have been better to say that the former division of signs into icons, indices, is what is picked out of the latter by the and symbols. Within philosophy, ground. For the purpose of this article, I will completely ignore the ‘interpretant’, The Peircean use the term which is clearly also a part of meaning, “symbol” is of course problematic, since though not in the simple way suggested it contrasts with another sense, more by Ogden’s and Richard’s all too familiar common in the European tradition, and triangle. In many of my earlier works, I found for instance in the work of Saussure have argued for a relationship between the and Piaget, where it is a peculiar kind ground and the interpretant, and Johansen of icon, or, as in the work of Cassirer, (1993: 90ff) even claims the latter was a synonym for “sign”. However, it historically substituted for the former, but appears that “conventional sign” is not an I now think the relationship cannot be that adequate term for what Peirce means by straightforward, for reasons which will “symbol”, which may involve “law-like” partially appear below. relationships of other kinds (perhaps those 10 many divisions of signs have pre- something which is accepted by a ceded the one proposed by Peirce, community of sign uses, it could at ending up with two, or four, or more least, for a short time span, become categories. In some ways, these di- a sign to a single observer. But even visions may be more justified than this is contrary to the very conditions the Peircean one. However, there are described by Peirce: he specifically two reasons for taking our point of refers to the case in which the sign departure in Peirce: first, it is within loses its sign character, when it is not these frames that most of the discus- seen as a sign but is confused with sion has been conducted; and sec- reality itself (which could actually ondly, when we look beyond those happen when looking at a picture elements which have usually been through a key hole with a single eye, addressed in the discussion, we will producing what Husserl dismisses as find that Peirce’s theory offers some a “Jahrmakteffekte”), when, as Pi- help for developing a more subtle aget would have said, there is no dif- approach to iconicity. ferentiation between expression and Many semioticians, in particu- content (cf. Sonesson 1989a,I.2.5.; lar those who deny the existence of 1992b). Indeed, at least sometimes, iconic signs, apparently believes pic- the pure icon is taken to be some- tures to be typical instances of this thing even less substantial: an im- category. There are several reasons to pression of reality, which does not think that this was not Peirce’s view. necessarily correspond to anything Pure icons, he states (1.157), only in the real world, for “it affords no appear in thinking, if ever. Accord- assurance that there is any such thing ing to Peirce’s conception, a paint- in nature” (4.447). Thus, it seems ing is in fact largely conventional, to be very close to the “phaneron”, or “symbolic”. Indeed, it is only for the unit of Peircean phenomenology a fleeting instant, “when we lose the (itself close to the Husserlean “noe- consciousness that it is not the thing, ma”), which is anything appearing the distinction of the real and the to the mind, irrespective of its reality copy”, that a painting may appear to status (cf. Johansen 1993:94ff). In be a pure icon (3.362; cf. Sonesson this sense, the Peircean icon is some- 1989a;III.1.). It seems, then, that a what similar to that of cognitive psy- pure icon is not a sign, in the sense chology, for it involves “sensible ob- that the latter term is commonly un- jects” (4.447), not signs in any pre- derstood (although Peirce will some- cise sense: however, it comprises all times state the contrary). At first, it sense modalities. may seem that although the icon is In most cases, when reference not a socially instituted sign, i.e. not is made to icons in semiotics what is which are observed to obtain). actually meant is what Peirce termed 11 hypo-icons, that is, signs which in- might be a colour sample used when volve iconicity but also, to a great picking out the paint to employ in extent, indexical and/or “symbolic” repainting the kitchen wall. Indeed, (that is, conventional, or perhaps not only is it true that any picture in- more generally, rule-like) properties. volves the representation of numer- There are supposed to be three kinds ous relationships (though certainly of hypo-icons: images, in which case not all) obtaining in the perceptual the similarity between expression world reproduced, as Stjernfeldt (in and content is one of “simple quali- press) points out. More importantly, ties”; diagrams, where the similarity it is only at this level that there is is one of “analogous relations in their a real similarity between the pic- parts”; and metaphors, in which the ture and the world, as James Gibson relations of similarity are brought (1982) has shown. to an even further degree of media- In order to make sense of the tion. Diagrams in the sense of ordi- theory of iconicity, we have to intro- nary language are also diagrams in duce a distinction between iconicity the Peircean sense, e.g. the popula- per se, the iconical ground, and the tion curve that rises to the extent that iconical sign, which is partly, but the population does so. The Peircean certainly not unambiguously, sup- concept is however much broader, ported by Peirce’s writings. This can as is the notion of metaphor, which only be done by starting out from the would, for instance, also include the concept of sign (characterised in lec- thermometer. Contrary to the way in ture two), which is certainly nowhere which icons have been conceived in made explicit in Peirce’s work, or, the later semiotic tradition, diagrams, for that matter, in that of Saussure rather than pictures, are at the core of – and then develop it by means of Peircean iconicity; at least, they are the Peircean notion of “ground” of most interest to Peirce himself. Indeed, mathematical formulae and The ground as abstraction deductive schemes, which are based and relevance on conventional signs, are those To go from the concept of iconicity most often discussed in his work. to the iconic sign, we have to ponder Moreover, no matter how we choose the meaning of a notion, sporadi- to understand the simplicity of “sim- cally, but often significantly, used by ple qualities”, the Peircean category Peirce, i.e. the notion of ground. As of images will not include ordinary applied to signs, I will here suppose, pictures (which would actually ap- iconicity is one of the three rela- pear to be metaphors of metaphors), tionships in which a representamen although Peirce sometimes seems to (expression) may stand to its object say so: if anything, a Peircean image (content or referent) and which can

12 things or states

”representation” a ”expression” ”appeal” b b

Sign a a

sender receiver

Fig. 1. a) Bühler’s Point of view ≈ Interpretant Organon model (with “abstractive relevance” and “abstractive supplementation” ; b) our model Ground of the Peircean Content ≈ sign (the ground immediate as “abstractive object relevance” and

Expression ≈ Referent ≈ Representamen dynamical object be taken as the “ground” for their (2:228) describes it as something forming a sign: more precisely, it is which “stands for that object not the first kind of these relationships, in all respects, but in reference to a termed Firstness, “the idea of that sort of idea, which I have sometimes which is such as it is regardless of called the ground of the representa- anything else” (5.66), as it applies to men”.10 the relation in question. In one of his 10 On the ground as Firstness well-known definitions of the sign, and, paradoxically, as abstraction and a term which he here, as so often, comparison, and its relation to the uses to mean the sign-vehicle, Peirce “correlate”, cf. the discussion in the second lecture. 13 According to one of Peirce’s the ground would then be a poten- commentators, Greenlee (1975:64), tial sign. Indeed, if we take seriously the ground is that aspect of the refer- Peirce’s claim that the concept of ent that is referred to by the expres- “ground” is indispensable, “because sion, for instance, the direction of the we cannot comprehend an agreement wind, which is the only property of of two things, except as an agree- the referential object “the wind” of ment in some respect.” (I.551), then which the weathercock informs us. it must be taken to operate a modifi- Although Greenlee does not say so, cation on both the things involved. this would seem to make the ground The operation in question, I into that which separates the “imme- submit, must be abstraction or, as I diate object” (that part of the content would prefer to say, typification. In which is directly given through the one passage, Peirce himself identi- sign) from the “dynamical object” fies “ground” with “abstraction” ex- (roughly, the referent, i.e. mean- emplifying it with the blackness of ing connected to the content but not two black things (1.293).11 It there- given in the sign but present in other fore seems that the term ground could past or future signs). On the other stand for those properties of the two hand, Savan (1976:10) considers things entering into the sign function the ground to consist of the features by means of which they get connect- picked out from the thing serving as ed. i.e. both some properties of the expression, which, to extend Green- thing serving as expression and some lee’s example, would include those properties of the thing serving as properties of the weathercock per- content. In case of the weathercock, mitting it to react to the wind, not, for instance, which serves to indicate for instance, its having the character- the direction of the wind, the content istic shape of a cock made out of iron ground merely consists in this di- and placed on a church steeple. If we rection, to the exclusion of all other have to choose between Greenlee’s properties of the wind, and its expres- and Savan’s interpretations, all quo- sion ground is only those properties tations from Peirce which have some which makes it turn in the direction bearing on the issue would seem to of the wind, not, for instance, the fact favour the latter. And yet, it seems to me that, in order to make sense of 11 I would not like to conceal the the notion of iconic signs, we must fact that there are many other passages in admit that both Greenlee and Savan Peirce’s work (many of which are given by Eco 1998: 44ff; 1999:59ff) which seem are right: the ground involves both to state rather clearly that the ground is expression and content (cf. Fig. 1b). Firstness, which means that it cannot be a Rather than being simply a “poten- relation, nor any kind of abstraction, as I tial sign-vehicle” (Bruss 1978:87), understand it, that is, no typification. For discussion, see Lecture 2. 14 of its being made of iron and resem- of being relative to, the conception bling a cock (the latter is a property of reaction with, something else” by means of which it enters an iconic (6.32). All this serves to underline ground, different from the indexical the parallel with the principle of ground making it signify the wind). If relevance, or pertinence, which is so, the ground is really a principle of at the basis of structural linguistics, relevance, or, as a Saussurean would and much of semiotics inspired by say, the “form” connecting expres- it (Hjelmslev and Prieto, notably). sion and content: that which must But we could take this idea further, necessarily be present in the expres- adding to the notion of ground a sion for it to be related to a particu- more explicitly constructive aspect. lar content rather than another, and To many structuralists (the Prague vice-versa. This phenomena in well- school notably), relevance is a dou- known from linguistics, where often ble movement, which both serves conventional rules serve to pick out to downplay non-essential elements some properties of the physical con- and to add others which were antici- tinuum, differently in different lan- pated but not perceived: thus, it de- guages, which have the property of pends on the twin principles of “ab- separating meanings, i.e. of isolat- stractive relevance” and “appercep- ing features of the expression on the tive supplementation” embodied in basis of the content, and vice-verse. Bühler’s Organon model (cf. Fig 1a The difference is, of course, that in and Sonesson 1989a,II.4.2.), as well the iconic ground, the relation that as in the Piagetean dialectic between determines one object from the point accommodation and assimilation (cf. of view of the other, is basically non- Sonesson 1988,I.3.1). conventional (cf. Sonesson 1989a: III.1). On different types of types If the ground is a form of ab- In some ways, our model (first for- straction, as Peirce explicitly says, mulated in Sonesson 1989a) is simi- then it is a procedure for engen- lar to that suggested by Groupe µ dering types, at least in the general (1992:135ff) when they claim that sense of ignoring some properties of the iconic sign is constituted out of things and emphasising others, for three elements: the signifier, the ref- the purpose of placing them into the erent, and the type. The status of the same class of things. And if it serves “type” as a third element of the sign to relate two things (“two black is, as I have pointed out elsewhere things” for example, or “the agree- (cf. Sonesson 1996a; in press a), ment of two things” in general), it is rather curious. Indeed, the signifier a relation, and it is thus of the order (and the corresponding signified, ab- of Secondness, i.e. “the conception sent here, in favour of the referent) is

15 already a type, in the Saussurean con- confusing that only the typification ception. One might suggest that the on the side of the content is termed referent and the type of the µ model type, while the same procedure on should be related to each other in the the expression side gets hidden with- same way as Peirce’s “immediate” in a strange mixture of Saussurean and “dynamical object”, that is, as a and behaviourist terminology: for, part to the corresponding whole, or although the signifier, to Saussure, as Husserl’s “noema” and “object”, was certainly a type, this property is where the first is the standpoint tak- not topical in the term as given.13 en on the second. If so, however, the Börries Blanke (1998; in press) terminology is confusing.12 has formulated a similar critique of As I noted above, the ground Groupe µ;s first model, suggesting seems to account for the division that the double process of abstrac- between the immediate and the dy- tion, which (as he points out) I have namical object on the side of con- diagnosed, should be added to the tent; but I then proceeded to argue model. I must confess that I have that there must be a similar division long had trouble seeing the point of on the side of expression. Indeed, if this amalgamation, since by adding the ground is tantamount to abstrac- double abstraction to Groupe µ: s tion, as Peirce says, and if abstrac- model of iconicity we would sim- tion is the generation of types, then ply end up with my model. This is we should readily accept the distinc- no doubt failing to take into account tion, suggested by Groupe µ, be- the greater degree of explicitness of tween the referent and the type; but the µ-model, as well as the dynami- we should add to it the parallel dis- cal character suggested by the proc- tinction between the signifier and its esses of stabilisation and conformity type. It may seems that this is exact- between the type and the referent, ly what Klinkenberg (1996: 291ff; in and of recognition and conformity, press) suggests, when transforming between the signifier and the type the erstwhile triangle into a square comprising a stimulus, as well as a 13 Indeed, in relation to the examples, it appears to make more sense to relate signifier, a referent, and a type, in the type to the kind of “transformation” particular since the signifier is now connecting the expression (“signifier”) defined, in contradistinction to the and the content (“referent”); however, stimulus, as “un ensemble modélisé this is not congruent with the definition. It does, however, correspond to the de stimuli visuels correspondant à un following passage (where also the stimulus type stable” (1996: 293). Yet it seems has its type): “Le stimulus, comme le référent, sont tous deux des actualisations 12 Strangely, Klinkenberg (1996: du type. Mais entre eux, ils entertiennent 299f) points out that the type is not simply des relations que l’on nommera ci-après identical to the signified! transformations” (Klinkenberg 1996: 293). 16 Fig.2. Structure of the “iconic sign”, according to Klinkenberg 1996: 291

(in the first version), or the stimulus It should not be forgotten that, and the signifier (in the more recent to both Klinkenberg and Blanke, version; Fig. 2.).14 On closer consid- iconicity means pictorality. For us, eration, however, I think that there however, this very peculiar case will would also be something still lack- have to await later discussion. There- ing when adding double abstraction fore, let us now consider some cases, (as Klinkenberg 1996; in press and which are clearly iconic, without be- Blanke 1998; in press do) to the µ- ing pictorial. The weathercock con- model: i.e. the relational character of tains, as we have noted, apart form the two instances of abstraction, in the indexical ground connecting it to other words, the fact that the expres- the direction of the wind, an iconical sion is subjected to abstraction with ground, which has as its other rela- respect to the content which it repre- tum the perceptual impression pro- sents, and vice-versa.15 duced by a cock, which is thus not only an iconic, but also a pictorial, 14 It is not clear to me, however, ground. For the moment, however, why it should be necessary to distinguish we should consider such instances as stabilisation and recognition on the two axes. We will treat them as equivalent. the balance as a representation of justice, or the signs used in Blissymbol- 15 In a way, this is better suggested by the original µ-model of iconicity, in ics to indicate the properties of being which the type presided over the apex of above or below. There is a similar- the triangle, and thus over the process of transformation between (what I here we shall see later, when turning to call) expression and content. However, pictorality, there may really be several the arrows do not seem to justify this different processes of abstraction and/or interpretation, nor does the text. As transformation involved. 17 ity between the balance and justice also mental) domain. Thus, once (although we would of course not again, of all the properties possessed notice it if this had not been pointed by the material mark, only one is sin- out to us before, and indeed if there gled out by its association to the cor- were not a convention for attending responding content. to this similarity): just as the balance Given these preliminaries, is used to weigh material things, jus- it might be said that an indexical tice has the task of weighing different ground, or an indexicality, involves arguments, claims, and other men- two “things” that are apt to enter, in tal objects. Whereas the expression the parts of expression and content consists of a perceptual impression, (“representamen” and “object” in which is at least virtually dynamic, Peircean parlance), into a semiotic the content is an abstract property, relation forming an indexical sign, which cannot be perceived in any due to a set of properties which are direct way, but can only be derived intrinsic to the relationship between from a long sequence of verbal acts them, such as is the case independ- and other stretches of behaviour. It is ently of the sign relation. Indexical- the postulated relationship between a ity, which is a ground, and therefore balance and justice as such which op- a relation, is thus basically differ- erates an abstraction in both objects ent from iconicity, which consists involved, picking out the property of a set of two classes of properties of equilibrium or equity. The case of ascribed to two different “things”, Blissymbolics (Fig. 3.) is somewhat which are taken to possess the prop- different: here both expression and erties in question independently, not content seem to be clearly material, only of the sign relation, but of each but what connects them is neverthe- other, although, when considered less an abstract property. The sign from a particular point of view, these for “above” is a line inscribed above two sets of properties will appear to what is the level at which the signs be identical or similar to each other. of Bliss are customarily inscribed; it This is the sense in which indexicali- represents “aboveness” in any other ty is Secondness, and iconicity First- material (and no doubt indirectly ness. As for the Peircean symbol, or

Fig. 3. The signs of Blissymbolics meaning “man”, “above”, “woman” and “below”. The first and the third sign is a kind of depiction, however schematic, but the second and the fourth are clearly abstract.

18 Firstness Secondness Thirdness Principle Iconicity — — Indexicality = Ground Iconic ground — indexical ground Symbolicity = Iconic sign Indexical sign symbolic ground Sign (icon) (index) = symbolic sign (symbol)

Fig. 4. The relations between principles, grounds, and signs, in the present interpretation of Peirce. generic sign, it is literally groundless, solution, it seems to me, is to admit as least until it becomes a sign: there that, contrary to indexicality, iconic- is nothing in the thing serving as ity is not in itself a ground. Perhaps, expression, nor the thing serving as to use some of Peirce’s own exam- content which explains the sign rela- ples, the blackness of a blackbird, or tion. The principle of relevance ob- the fact of Franklin being American, taining between the two parts of the can be considered iconicities; when signs is produced merely by the sign we compare two black things or relation, which is why it is Thirdness Franklin and Rumford from the point (Cf. Fig. 4.).16 of view of their being Americans, If iconicity is Firstness, but the we establish an iconic ground; but ground is a relation, then the only only when one of the black things is taken to stand for the other, or when 16 This, no doubt, is what Peirce must Rumford is made to represent Frank- have meant – but I have my doubts about lin, do they become iconic signs (or this idea. Some iconical signs seem to rely on pre-existing conventions and/or hypo-icons). Just as indexicality is habits, which exist in the Lifeworld prior conceivable, but is not a sign, until it to any signs. Thus, for instance, the sign enters the sign relation, iconicity has meaning “woman” in the gesture system some kind of being, but does not ex- of the North American Indians consists in describing the braids on both sides of ist until a comparison takes place. In the head. This sign is of course iconical this sense, if indexicality is a poten- to the extent that it imitates the braids, tial sign, iconicity is only a potential and indexical, because it relates to the ground. head of the sign producer; but it only means “woman” because there is a pre- Projecting the distinctions existing convention in the society of North made in accordance with the preced- American Indians for women to wear their ing interpretation of Peirce onto the hair braided. Of course, the reason for this discrepancy is not doubt the vague µ-tological square (Fig. 2.), we end character of the Peircean sign concept. up with a model that has at least the 19 advantage of distinguishing the ex- give rise, with the iconical and index- pression type, the content type, and ical grounds constituting some kind the transformation type (Fig. 5.). It is of intermediate level. As we shall see difficult to know to what extent this later on in this course, the establish- model, as proposed here, still con- ment of such a separation of iconic- forms to the intentions of Groupe µ. ity and indexicality, the grounds that From our point of view, however, it they form and their corresponding helps clarifying the issues involved. signs, has the added advantage of lib- It still does not say very much about erating indexicality and iconicity for the specificity of the iconic, let alone other tasks than that of constituting the pictorial, sign. This specificity signs. For the moment, however, this resides in the nature of the transfor- discussion has permitted us to clarify mations and the different kinds of the structure of the sign, separating conformity. the various instances of typicality from the corresponding tokens. Summary So far, I have tried to show that the 3.2. Iconicity regained. concept of iconicity may still be ren- The logic of the dered useful in modern semiotics, conceived as an empirical science, Lifeworld involved with pictures and other The two most important arguments signs as we encounter them in our against the possibility of iconic everyday world. In part, this section signs were given by Arthur Bier- has consisted in a reading of Peirce’s man (1963), in the form of a direct writings, which, as all readings, but critique of Peirce. They were later rather more explicitly, is made from repeated, independently, as it seems, a particular angle of vision, that of in a much more well-known version, ecological semiotics. In order to and without any direct address to render possible this reading, we have Peirce, by Nelson Goodman (1968; relied on a concept of sign which is 1970). These arguments have later very much more specific than that been called the arguments of regres- found in Peirce’s work, and we have sion and of symmetry, respectively had recourse to the notion of ground, (Sebeok 1976:128). Both arguments with rather scant evidence, in the can be evaded, I believe, the first sim- sense of the structural principle of ply by accepting the specific concept relevance and the Husserlean process of sign which I introduced in the sec- of typification. As a result, we have ond lecture, the second, more labo- established a distinction between the riously, by taking the Lifeworld, or properties of iconicity and indexical- ecological physics, as being the pre- ity per se, and the signs to which they supposed background of all ordinary

20 Fig.5. Structure of the (pictorial) sign, as conceived here sign process. To take care of the argu- classified into a number of very ment of symmetry, therefore, we will general categories, such as “thing”, not only have to introduce the idea “animal”, “human being”, etc., and of a Lifeworld hierarchy of proto- therefore everything in the universe typicality, but we will have to distin- can refer to, and be referred to, eve- guish between two kinds of iconical rything else. Thus, if iconicity is at signs, one of which seems to change the origin of signs, all things in the the nature of iconicity in ways which world will be signs. Peirce himself Peirce could never have foreseen. would probably not be in the least Then we will see that both primary impressed by these consequences: he and secondary iconical signs, though certainly seems to think that Rumford for very different reasons, do not ful- can be an icon of Franklin since they fil Peirce’s requirement that iconic- share the property of being Ameri- ity should be independent of the sign cans, which is, if not a metaphysical function. property, at least a very general one. Nor would Giordano Bruno, or other Regressive iconicity and thinkers of the Middle Ages and the the sign function Renaissance be shocked (cf. Yates According to argument of regres- 1966; Gombrich 1972; etc.); neither sion, all things in the world can be would the Naturphilosophen of Ger-

21 man Romanticism, or Baudelaire and The import of the argument of other believers in the theory of “cor- regression really depends on how we respondences”, including the surre- interpret Peirce’s theory. If he meant alists, and even latter-day New Age to suggest that there are three prop- mystics. Hence, iconical signs of this erties, iconicity, indexicality, and kind are not only conceivable; they symbolicity, which, by themselves have been conceived throughout the and without any further requirement, greater part of human history. trigger the recognition of something Yet, it is true that these are not as a sign, then the argument of re- typical signs, let alone typical iconi- gression will create trouble for his cal signs; and pictures are most cer- conception. On the other hand, if he tainly not of this kind. The undesir- merely wanted to suggest that some- able consequences anticipated by thing that was already recognised as Bierman can be avoided, in his own being a sign could be discovered to view, if we introduce the provision be an iconical sign, rather than an in- that no icon should contain universal dexical or symbolic one, by means of characteristics as part of its meaning. tracing it back to the iconic ground, Thus, it seems, Bierman is satisfied, then the argument of regression will but it is obvious that such a provision have no bearing on it. No matter must be problematical, in particular what Peirce would have thought, the because it is not clear how we shall separation of the sign function from establish the limit beyond which the iconical ground, explicitly intro- characteristics become too general to duced in our reading, makes this ar- be included in the meaning of iconi- gument completely irrelevant. cal signs. It would also seem that, even apart from the “correspondanc- Beyond the symmetry es”, there are cases in which iconical argument signs have very general features in According to another argument, first their contents, at least if metaphors voiced by Bierman, which has later and symbols in the European sense, been termed the symmetry argument, are considered to be iconical. It is ar- iconicity cannot motivate a sign, for guable, that some pictograms stand while similarity is symmetrical and for relatively universal features, in reflexive, the sign is not. Pigments particular if based on synaesthesia. on paper, or carvings in a rock, could Indeed, both the balance as a sign stand for a man, but not the reverse; for justice, and the signs of Blissym- nor will they, in their picture func- bolics considered above, would be tion, stand for themselves. This argu- ruled out by such a provision. Such ment is based on an identification of as it stands, Bierman’s provision is the commonsensical notion of simi- scarcely acceptable. larity with the equivalence relation

22 of logic. No doubt, the equivalence inent becomes the references point, relation, as defined in logic, is sym- and prototypicality is only one of the metric and reflexive, and thus - can factors making an item eligible for not define any type of sign, since the the position, others being frequency, sign, by definition, must be -asym intensity, celebrity, information, and metric and irreflexive. As far as I so on. can see, there is no way to evade this If we generalise this finding to argument within a purely Peircean the case of signs, there is every rea- framework (except of course the son to suppose that a three-dimen- same argument as against the argu- sional object, rather than some lines ment of regression, if this is an genu- on a surface, would count as a natu- inely Peircean argument); ecological ral standard of comparison. While semiotics, however, has an answer to this relationship between three-di- offer. mensional and two-dimensional ob- The error consists in the iden- jects may well be a universal, it is tification of similarity with the easier to show the principle at work equivalence relations as defined by in cases that vary cross-culturally. logic. To make such identification is Among numerous apocryphal stories to suppose man to live in the world of tribes failing to recognise pictures of the natural sciences, when in fact as such, there is one verified case in he inhabits a particular sociocultural which the group (the Me’ studied by Lifeworld. Similarity, as experienced Deregowski) had never seen paper, in this Lifeworld, is often asymmet- and was therefore led to focus on ric and irreflexive. That this is true the material per se. When pictures of ordinary comparisons in verbal where instead printed on cloth, the language and in metaphorical visual Me’ immediately recognised their displays has now been experimental- sign function and perceived the pic- ly demonstrated (notably by Rosch tures as such. To these people paper, 1975; & Simpson & Miller 1976; being an unknown material, acquired Tversky 1977; & Gati 1978; cf. also such a prominence that it was im- Sonesson 1989a,220ff, 327ff). Here, possible for them to see it as a vehi- I will just consider one example: In cle for something else; on the other a task involving comparisons be- hand, it is precisely because paper tween countries, Tversky (1977: 333 is so trivial a material to us that we ff.) found that the statement “North have no trouble construing instances Korea is similar to Red China” was of it as pictorial signifiers (cf. Sones- chosen in preference to its inversion son 1989a: 251ff). In a similar vein, in 66 out of 69 instances; it was also it is natural that a Mexican woman, located higher on a scale. On the coming for the first time to Sweden, whole, that item which is most prom- should wonder at the presence of

23 higher things Fig 6. The basic model of the Lifeworld hierarchy things at hand

Most Most real well- ultra I-here known ultra things Most -now things ego- like Most understandable

lower things “Barbies” everywhere: in her experi- 1997a, b, 2000a). Husserl, Gibson, ence, real-world blond women were and Greimas all called for a science far less prominent than the doll made of “the natural world”, because they in their image. realised that nature as we experience It thus becomes necessary to it is not identical to the one known to posit a kind of taken-for-granted hi- physics but is culturally constructed. erarchy of prominence among the Like Husserl’s Lifeworld and Gib- things of the Lifeworld (see Fig. 6.). son’s ecological physics, but unlike For something to be a sign of some- Greimas’ natural world, semiotic thing else, it must be ranked relative- ecology will suppose this particular ly low on the scale of prototypicality level to be a privileged version of the applying to the “things” of the Life- world, “the world taken for granted”, world. Such a scale would be simi- in Schütz’s phrase, from the stand- lar to the basic metaphor underlying point of which other worlds, such as ordinary language, which Lakoff & those of the natural sciences, may be Turner (1989:160ff) call “The great invented and observed (cf. Sones- chain of being”. Indeed, these regu- son 1989a,I.1.4, I.2.1.). This world larities of the Lifeworld, together is characterised by a particular spa- with the similar laws of environmen- tial and temporal structure, by types, tal physics, formulated by James and, by regularities, or as Husserl’s Gibson, stand at the origin of an even says, “the typical way in which things broader domain of study, which I tend to behave”. The latter are the have called the semiotic ecology (cf. kind of laws of “ecological physics”, Sonesson 1993a, 1994a, b,1996a, c, in Gibson’s sense, which are “defied 24 by magic”, and which also forms the unlike that of the Cheshire cat, the foundations for Peircean abduction. human smile cannot exist independ- In my earlier publications, I ently. have referred the hierarchy of prom- Contrary to the argument of inence of Lifeworld things, in two regression, the symmetry argument different, but complementary, ways. may thus be warded off, without in- On the one hand, objects, such as the troducing a supplementary sign func- human body itself, in particular the tion and without amending the defi- face, but also common objects like nition of the iconic ground. On the chairs, must be so central to the hu- other hand, it supposes a complete man sphere that they will be recog- reconstruction of semiotic theory, nised with only scant evidence, even in which ecological physics, also though the invariants embodied in a known as the Lifeworld, is taken as particular picture are found in other the point of departure of all possible objects as well (cf. Fig. 9c.). In this meaning construction. case, the objects at the highest levels of the scale stand the best chance Primary and secondary of being selected. On the other hand, iconic signs I have argued that only objects low The alternative analysis in terms of down on the scale will be recognised conventionality suggested by Good- as susceptible of embodying a sign man, Eco, and others is conceived to function, without being particularly take care of the case of pictures, but designated as such, which in our paradoxically, it seems that it would culture is true of a sheet of paper or really be needed, not for pictures a canvas. A human being, a shape but for some other iconical signs which is easily recognised as such which rely on identity (and some with very scant indications, is per- others, such as “droodles”, e.g., in haps also that object which is most the extreme case, a line perceived as difficult to see as a mere signifier of “a dirty French card seen from the something else if he is not explicitly side”). Goodman’s, Greenlee’s and so designated, as in the theatre or in Eco’s contention that the referent of a ceremony. On the other hand, the each picture is appointed individu- human face, which is probably that ally, are incompatible with what psy- object which is most easily identified chology tells us about the child’s ca- of all, serves at the same time as sup- pacity for interpreting pictures when port for conveying other signs, the first confronted with them at 19 expressions of feelings and attitudes; months of age (as demonstrated in but then again, it is not the face but a famous experiment by Hochberg). its movements which are signifiers On the other hand, we do have to of these other signs. It is just that, learn that, in certain situations, and

25 according to particular conventions, knowing either that something is a objects which are normally used for sign or what it as sign of: of itself what they are become signs of them- as an individual object, of a particu- selves, of some of their properties, lar category (among several possible or of the class of which they form ones) of which it is a member, or of part: a car at a car exhibition, a stone one or another of its properties. A axe in the museum showcase or a tin car, which is not a sign on the street, cane in a shop window, an emperor’s becomes a sign at a car exhibition, as impersonator when the emperor is does Man Ray’s iron in a museum. away, and a urinal (if it happens to We have to know the showcase con- be Duchamp’s “Fountain”) at an art vention to understand that the tin can exhibition. There is never any doubt in the shop-window stands for many about their pure iconicity, or about other objects of the same category; their capacity for entering into an we need to be familiar with the art iconic ground — but a convention is exhibition convention to realise that needed to tell us they are signs. each object merely signifies itself; When used to stand for them- and we are able to understand that selves, objects are clearly iconi- the tailor’s swatch is a sign of its pat- cal: they are signs consisting of an tern and colour, but not of its shape, expression that stands for a content only if we have learnt the conven- because of properties which each of tion associated with the swatch (cf. them possess intrinsically. And yet, Sonesson 1989a,II.2.2. and 1994a, b, without having access to a set of 1998b). conventions and/or an array of stock When Man Ray makes a picture situations, we have no possibility of of a billiard table, we need no con-

Fig. 7. Man Ray’s billiard table

26 vention to recognise what it depicts (Fig. 7). However, if Sherrie Lev- ine’s (real, three-dimensional) billiard table is to represent Man Ray’s picture, there must be a label invert- ing the hierarchy of prominence of the Lifeworld (Fig.8). This shows that among the properties determining the probability of an object functioning as the expression of an iconic sign is to be found three-dimensionality rather than the opposite. Since the inception of modernism, and particularly in the phase known as post- modernism, the sign function of pictures has been at the centre of inter- Fig. 8. Sherrie Levine’s billiard tables est: it is thus not surprising that art- ists, such as Levine, should employ with the way iconicity is assigned to themselves to inverse the normal signs. A primary iconic sign is a sign Lifeworld hierarchy, which makes in the case of which the perception two-dimensional objects stand for of a similarity between an expres- three-dimensional ones, rather than sion E and a content C is at least a the reverse. But similar things also partial reason for E being taken to be happens in the world of everyday the expression of a sign the content life: the Mexican woman who found of which is C. That is, iconicity is Sweden to be full of “Barbies” made really the motivation (the ground), the same inversion, because, building or rather, one of the motivations, for on her particular Lifeworld experi- positing the sign function. A second- ence, she took real, animate, persons ary iconic sign, on the other hand, as being, at least for the duration of a is a sign in the case of which our speech act, representations of assem- knowledge that E is the expression bly-line fabricated objects, made of of a sign the content of which is C, inanimate matter, i.e. of dolls. in some particular system of inter- The relative part played by ico- pretation, is at least a partial reason nicity and conventionality in a sign for perceiving the similarity of E and may be used to distinguish primary C. Here, then, it is the sign relation and secondary iconicity. In fact, to that partially motivates the relation- be more precise, we should distin- ship of iconicity.17 guish primary and secondary iconic 17 Eco (1998: 27ff; 1999: 382ff) signs, since we are really involved apparently changes the names of my two 27 In a sense, what I here call sec- determine iconicity. In the case of ondary iconical signs are not very identity sign, the problem does not good examples of iconicity, as the consist in discovering the shared latter is characterised by Peirce, for properties – but it seeing the one item the definition clearly implies that, is a sign for another, rather than just in at least one sense, the iconicity of two members of the same category. the signs is not independent of their In other cases, the sign function must sign character: on the contrary, it is precede the perception of iconic- a precondition.18 Perhaps this does ity because there is too little resem- not have to be taken as an argument blance, as in the manual signs of the against Peirce’s definition: iconic- North American Indians, which, ac- ity per se may well be independent cording to Mallery (1881:94f), seem of the sign function, even though its reasonable when we are informed presence in signs may sometimes be about their meaning. In Arnheim’s conditioned by the sign function. terms (1969:92f), a “droodle” is dif- Pictures are of course primary, ferent from a picture in requiring iconical, signs, in this sense, and a key, as Carraci’s mason behind a they may well be the only kind there wall (cf. Fig. 9b), or in “Olive drop- is. However, identity signs do not ping into martini glass or Close-up of constitute the only case in which girl in scanty bathing suit” (cf. Fig. the sign function has to precede and 9a). While both scenes are possible to discover in the drawing, both are iconicities into mode Alpha and Beta, respectively, using the same definitions clearly underdetermined by it. There and similar examples; and although he are two ways in which we can try to fails to refer to me in this passage, the avoid such an ambiguity. One is to fill article were I first made the distinction in the details, in particular the details is in his bibliography (Sonesson 1993a, though given as 1994). The originality of that are characteristically different in Eco’s proposal, however, is the suggestion an olive and a navel, in the air and a (which is never spelt out) that the property pair of thighs, etc. At some point the that I have found to distinguish two kinds droodle will then turn into a genuine of iconicity could also be found in other grounds. Lopes (1995), on the other hand, picture. The other possibility, which seems to have happened on a very similar, is the only one considered by Burks if not identical, distinction, independently and Bierman, is to introduce an ex- of Eco and me. For some discussion, see plicit convention, such as Carraci’s Sonesson, in press b. Vaillant & Castaing, 19 in press, interpret Eco’s distinction in key. quite a different way, which seems to me 19 It is undoubtedly because he quite unwarranted. is mainly involved with droodles or 18 As we shall se later, there is quite logotypes close to being droodles that another sense in which primary iconical Vaillant (1997: 45ff) fails to see the signs do not conform to the criterion of relevance of the prototype hierarchy. independence. This also explains the same argument 28 a b Fig. 9. Two droodles and a picture which can be read as a droodle: a) Olive dropping into Martini glass or Close-up of girl in scanty bathing suit (inspired from Arnheim as adapted in Sonesson 1992). b) Carraci’s key (Mason behind wall); c) face or jar (inspired by Hermerén 1983:101);

According to Hermerén (1983: may be favoured because of proper- 101), it is only because of “the limi- ties of the expression itself, and is tations of human imagination” that not overridden by convention. we see Fig. 9c. as a human face, for In a curious little essay, David- it can equally well be perceived as “a sen & Davidsen (2000: 82) take me jar from above, with some pebbles to task, because, in their view, the and broken matches on the bottom, concept of a hierarchy of prominence and a stick placed across the open- only apparently solves the problem of ing”. Thus, it should be ambiguous accounting for the natural asymmetry in Bierman’s sense. It all depends on of the iconical sign: while it “might what is here meant by the limits of be taken to explain why an image of human imagination: Gestalt princi- a man is the iconic representation of ples, the face as a privileged percep- this man /…/, this does not contrib- tual object (cf. E. Gibson 1969:347 ute much more than to systematising ff.), and so on, all conspire to make relativism.” I don’t know what more one of the readings determinate. can be done. The sign cannot be any- While it is possible to find the ele- thing in itself. The point is precisely ments Hermerén suggests should be that the sign is a sign in relation to a there in the picture, it is impossible given lifeworld — in relation to the to see the interpretation as a whole general structures of the lifeworld, without being disturbed by the other as in the case of three-dimensional- reading. Thus, it seems that when an ity, or in relation to a particular so- expression has similarities to differ- cio-cultural lifeworld, as in the case ent contents or referents, one of these of the meaning of paper to the Me’. by Bordon & Vaillant (2002: 59), who Ecological semiotics is interested in however proceed to offer a more pertinent accounting for “the limitations of hu- counter-example to the prototype man imagination”, to use Hermerén’s hierarchy suggested here, which will be term – not to do away with them as considered below. 29 an obstacle to a deeper truth.20 conceived it above, the prototype hi- While all this serves to clarify erarchy would not apply to objects the nature of the pictorial sign, as a as such, but to some of their prop- particular kind of iconicity (though erties. Not the ice statue of a motor we will see that pictorality is pecu- cycle, but perhaps ice as a material, liar in other ways), it leaves a large might be argued to be “less promi- residue: secondary iconic signs can nent” (perhaps in the sense of more hardly be said to be determined more homogeneous, more ubiquitous in than in a negative way. Consider a the history of mankind, etc.) than counter-example to my prototype motorcycle parts. In the Me’ story, it hierarchy offered by Bordon & Vail- is paper, not particular things made lant (2002: 59): an ice statue of a mo- out of paper, which is ranked too tor cycle is less familiar to Parisians high on the scale. than a real motor cycle, and yet when But the example is also quite the former is exhibited in front of the different from those I have discussed town hall, there is no doubt to any- above, which either involved a two- one that the ice statue is the signifier, dimensional object representing a and the motor cycle the signified. three-dimensional one (as in the case The authors are guilty of sev- of pictures), or a single three-dimen- eral errors of interpretation, and yet sional object being the sign of the their example is interesting. The class of which it is a part, of some of prototype hierarchy is based on the its properties, and so on. It is, how- notion of prominence characterised ever, similar to cases I have taken by the cognitive psychologists Ro- up elsewhere (notably Sonesson sch and Tversky as corresponding to 1989a,III.6.5; in press b): the tailor’s prototypicality, frequency, intensity, dummy and artificial food made out celebrity, information, and so on. In of plastic or wax, as seen in Japanese this sense, if “familiarity” may, on restaurants. The outcome of that dis- some occasions, be the opposite of cussion was that there were certain prominence, there are many other properties that were intrinsically possibilities. But this also means that more prominent in the human world, the concept of prominence is rather such as, apart from three-dimension- unclear: at least, it seems to be too ality, animateness and movement. open-ended. In any case, as I have The last of these features may of course be the factor that makes the 20 Later on in the essay, Davidsen & Davidsen (2000) appear to criticise me for real motorcycle more prominent than finding the same sign iconical, indexical, the one made out of ice. But a more and symbolic, but that is only a problem, general argument could in fact be as mentioned at the beginning of this made from the example of artificial article, if we take these terms to apply to objects, rather than to relations. food. There is a reason why the wax 30 food is taken to represent the real should not be taken to be an all or food, rather than the opposite. Food none affair: just as a sign may con- is defined by the functional property tain iconical, indexical and symbolic (or the “affordance” as Gibson would properties at the same time, it may say) of being edible, and that is ex- very well mix primary and second- actly the property which wax food ary iconicity. lacks. The motorcycle, similarly, is defined by the property of being a The criterion of vehicle, which is an expectation that independence the ice statue can hardly fulfil. If this Apart from the reasons mentioned at shows that the ice motorcycle is no the beginning of this essay, there is real counter-example, it also demon- another sense in which pictures are strates the complexity of secondary far from being central instances of iconicity.21 The prototype hierarchy icons. As was noted above, the fact should not be expected to form some that an object serving as the expres- rigid structure fixed once and for sion of an icon and another object all.22 serving as its content possess, in Another, more important les- some respects, the same properties son of this discussion, however, is should not be a result of one of them that primary and secondary iconicity having an influence on the other. In the case of an icon (contrary to the 21 It might also be relevant, as has case of an index), “it simply hap- been suggested to me, that the motorcycle is as more enduring type of artefact than pens that its qualities resemble those the ice sculpture (Cf. Lecture 2). However, of that object, and excite analogous this argument could then be used to claim sensations in the mind for which it is that the bronze statue of Caesar is more a likeness” (2.299). Since both Fran- basic than Caesar himself. klin and Rumford are Americans, 22 This also shows that the “Barbie” Peirce claims, one of them may serve example given above is much more complex than what I hinted at: from the as a sign of the other; but the fact that point of view of animateness, and the like, Franklin is an American is quite uneven the Mexican woman does do really related to Rumford’s being one. But consider the doll to be more prominent there is at least one sense in which this than the human beings who are blond (or so I would hope). It is when attending is not true of pictures, not only in the to hair, skin colour, and the like, that case of a photograph (which Peirce she ranks the Barbie doll higher on the often pronounces to be an index), but scale than those alien human beings seen also in the case of a painting: in each in Sweden. This is the stuff of which metaphors are made. Cf. Sonesson 1989a, case, the “thing” serving as the ex- 2003c; 2005. Interpreted in another pression is expressly constructed in way, this anecdote might be the point of order to resemble the “thing” serv- departure of the distinction that is basic to ing as the content, although a direct cultural semiotics. Cf. Sonesson 2000b. 31 physical connection only exists in the at least in some cases of them (the first instance. Leonardo painted the exception being perhaps droodles). canvas known as Mona Lisa in order But we have already pointed out that to create a resemblance to the wife there is a sense in which primary of Francesco del Giocondo, and, al- iconical signs, but not secondary though the resemblance is of a much iconical signs, seem to realise such more abstract kind, the same is true independence. So what kind of in- of Picasso painting Gertrude Stein dependence are we talking about? or Kahnweiler. And it is as true of a It seems that Peirce’s criterion is in- synthetic computer picture showing sufficiently specified. Tentatively, I a lamp correctly illuminated from would like to suggest the following above right as of a photograph with distinction: secondary iconical signs the same subject. are made up of entities that have an In this sense, Peirce’s claim existential independence (they exist that the properties of expression and independently of the sign function content pertain to them independ- and the iconical ground) but no sem- ently seems more relevant to identity iosic independence (they are consti- signs than to pictures. When Fran- tuted as iconical grounds only as a klin is made to represent Rumford, consequence of being recognised as or, to take a more clearly iconical signs). As for primary iconical signs, (and visual) example, when a blond they have no existential independ- girl plays the part of Marilyn Mon- ence (the expression does not exist, roe, it is obvious that each of the as a likeness or as anything else, in- ‘things’ involved has an independ- dependently of its relation to the con- ent existence and a presence in the tent), but they do possess a semiosis world independent of the other; but independence (the iconical ground is in case of the painting, the “thing” constituted prior to the recognition which serves as expression is actu- of the sign function). ally created in order to resemble the If it is possible to distinguish “thing” which serves as the content; existential and causal independence, it comes into being precisely in order then there may perhaps be some fur- to accomplish this function. Thus, it ther kind of independence that is real- seems that the iconical ground of pri- ly characteristic of all iconical signs. mary iconical signs, or at least one Or perhaps — and I think this is the particular case of them (which may safest bet for the moment — iconici- be the only one), pictures, is not in- ty per se is really independent, while dependent of the fact that it connects iconic grounds and iconic signs may the relata of a sign function, whereas possess only causal or existential in- there is such an independence in the dependence, but never both. case of secondary iconical signs, or

32 Summary We have already touched on one of these ways: one property, which In this section, some pieces of criti- Peirce insists on as a defining crite- cism have been addressed against rion of iconicity, independence from the Peircean notion of iconicity, in the sign function, will not translate order to show that it cannot stand on onto pictures, at least if independ- its own, if it is not complemented by ence is taken in the sense of existen- a few essential new elements. First tial autonomy. But the case of picto- of all, to ward off the argument of re- rality is even more convoluted, as we gression, we need to use a more spe- will see shortly. cific concept of sign than that offered by Peirce and Saussure (as this was introduced in the second lecture). In 3.3. In the looking- the case of the symmetry argument, glace, somewhat less our only way of escape turned out to darkly. Eco’s three require a radical remake of semiotic critiques theory, basing it on the world-taken- Contrary to Bierman, Goodman, and for-granted, as developed by Husser- others, Umberto Eco remains untrou- lean phenomenology and Gibsonean bled by similarity being neither a suf- psychology, and further elaborated ficient nor a necessary criterion for in my own work, notably with the something being a (pictorial) sign. introduction of the concept of a hi- The different versions of Eco’s cri- erarchy of prototypicality. We ended tique of iconicity are too numerous up with a division of iconical signs ever to be fully discussed, but we can into two very different kinds, the pri- distinguish three essential periods: mary ones, where the perception of at the first stage, Eco (1968; 1970; similarity is at least part of the rea- 1972) is basically concerned to show son for postulating the sign charac- that iconical signs (the basic example ter, and the secondary ones, in which being pictures) are similar to linguis- case knowledge of the sign function tic signs in being conventional and is a prerequisite for discovering the analysable into features; at the sec- likeness. It should be clear that second stage (1976; 1978; 1984a, b), he ondary iconical signs violate one of abandons the idea of feature analysis Peirce’s basic requirements for ico- but wants to dislocate the required nicity: the independence of iconicity similarity sidewards, into some kind from the sign function. Fortunately, of proportionality. In the final stage, it may seem, pictures are primary however (1998; 1999), he seems to iconical signs: but there are ways in give up almost everything he has so which pictures do not seem to meas- far believed in, and, while retaining ure up to Peircean iconicity either. a tiny part for convention, basically

33 goes to the other extreme, making all icons into mirrors affording a direct view onto reality.

This nose is not a nose. The case for conventionality When Peirce says that the sign is similar to its object, and when Peirce and Morris alike claim that they have properties in common, Eco (1968:188; cf. 1976:327f) thinks this is to some degree trivially true, but Fig. 10. One of Annigoni’s false to the extent that it is interest- portraits of Elisabeth II, possibly ing. For, Eco goes on to ask, what can the version referred to by Eco it mean to say that Annigoni’s por- with just two; the surface of the real trait of Queen Elizabeth II (Fig. 10) nose is full of pores and other ir- has the same properties as the queen regularities, but that of the painting herself? The example is of course is smooth; and corresponding to the not chosen arbitrarily: Pietro An- nostrils of the queenly nose, there are nigoni is a painter of the 20th century no apertures in the canvas, but only purporting to work in the tradition of two black dots. Morris, Eco admits, the Italian Renaissance and mostly is well aware of such problems, and known because of his portraits of has therefore proposed that iconic- celebrities such as Elisabeth II, John ity is a question of degrees; but such F. Kennedy, and the Shah of Iran. It a definition, Eco contends, can be may seem, therefore, that the differ- stretched to include anything, and ence between the portrait and reality must lead to the destruction of the is minimal. Lindekens, later on, will concept of iconicity. For semiotics, choose photography to have a good Eco therefore concludes, it is not whipping boy with which to bring enough to say that the iconic sign home the cause of conventionality. resembles its referent in certain re- For what Eco wants to show is spects. that, even so, the distance between Oddly enough, it is precise- the picture and reality is consider- ly with this definition that Eco able. Perhaps we could agree, he (1968:191) seems to end up a few suggests, that the shape of the nose pages later: iconic codes are said to is the same. But the nose of the real reproduce certain selected conditions queen has three dimensions, and that on the perception of the correspond- of her portrait must remain satisfied ing object. This selection in turn de- 34 pends on the codes of recognition, More important is the idea that the and also on graphic conventions. In selection of similarities is also con- our culture, for instance, the zebra, ditioned by the possibilities of the contrasted with the horse, will be particular pictorial expression used. identified by its stripes; but a- soci But Eco does not pursue this idea ety only acquainted with zebras and further. It should anyhow be obvi- hyenas will need to focalise on some ous that nothing of this will answer other feature. In the same way, Eco the fundamental criticism which (1976:349) says about Gombrich’s Eco himself addressed to iconicity: (1963:1ff) hobbyhorse, that it is even if we select just the stripes, or iconic in the most abstract sense, be- the propeller, or the horseback, it cause it only reproduces the straight remains true that, on closer inspec- line formed by the horseback. Noth- tion, those of the pictures are basi- ing of this is particularly original, cally different from those of reality. however. In the most general sense, But even so, this is not necessarily Peirce was well aware of the conven- an argument against the presence of tional nature of iconic signs. As for similarity: rather its concerns the lo- the codes of recognition, Vierkant cus of similarity. (1912:352) had already noted their Sometimes Eco (1968:201, existence when he observed that 208;1976:359) appears to return to the primitives make their pictures, his original radicalism: iconical signs including only that which must be only seem to reproduce the proper- observed in reality, from the practi- ties of their objects. A simple contin- cal point of view of the hunter; but uous line, we are told, will form the of course, Vierkant thought Occi- contour of a horse, and yet, “the only dental man would not make use of property the line has”, i.e. that of be- such operations. A more impressive ing “a continuous black line”, is “the testimony of the workings of recog- only property that the real horse does nition codes is fig. 11a-b, which re- not have” (1968:192; cf. 1976:328).23 duces the difference between a bird Now, obviously, there are other and a man to a minimum (drawings properties that the real horse does of the Bakairi, according to Vierkant 23 In his third critique, Eco (1998:14f; 1912:344). Paraphrasing Aristotle, 1999: 349f) explicitly rejects what he says we could say that to this tribe man is here, but he sometimes seems to go to the a beakless biped. opposite extreme of thinking there are lines in nature, and sometimes he accuses But rather than specific picture Gibson of holding this theory (‘outlines conventions, all this may well be are already offered by the stimulating conventions of the Lifeworld, either field’), when in fact Gibson’s point is in some specific socio-cultural varie- that perception is based, not on simple properties, but on relations between ty of it, or in its general human form. relations. 35 similar effect can be obtained only by solarisation). Curiously, the zebra, which to Eco is a horse with stripes, also appears in Kennedy’s (1974b:231 ff.) psychological study of picture perception. Children were able to interpret lines to stand for uneven surfac- Fig. 11. Bird and man, according to es as in a landscape layout, uneven the Bakairi conception, as shown illumination as in shadows, uneven by Vierkant 1912:344 texture as in the hems and cuffs of not possess; but, more importantly, knitted garments, and uneven pig- the line has other properties, at dif- mentation as in the hide of a zebra. ferent intensional levels – and so, the Here then, contours, not coloured common property may appear at one surfaces, are used for the stripes. of these levels (cf. Sonesson 1989a, Kennedy concludes that any discon- III.1.4.). In fact, Eco even mentions tinuities that are sources of optical another property of the line a few structures may be rendered by lines. lines later, viz. that it separates the Thus, we may answer Eco that the inner space, the horse, from the out- limits of the horse’s body and the er space, the non–horse; but he adds black continuous line have in com- that neither is this property found in mon the property of being disconti- reality. Then, mitigating once again nuities in optical organisation. the radicalism of his critique, Eco If the iconical sign has any speculates that the horse may really property in common with something, look like that if seen in profile con- Eco (1968:201) says, then it is not trasted against the background of the with the object itself but with its per- sky. This is certainly admitting too ceptual model; for the iconical sign much: the light may be such that the constructs a model which is homolo- horse loses its aspect of possessing gous to the model of perceptual re- a volume, but then it will appear as lations which we construct when we a silhouette. And silhouettes have recognise an object or remember it; limits (“Grenze”), but no contours only the matter in which the model (“Kontur”), as Volkelt (1963:28ff) is realised differ. But it seems doubt- says: the latter, but not the former, ful that there has ever been anybody detach themselves equally from out- who has claimed that pictures are er and inner space. There is nothing similar to anything else than what comparable to that in reality; more we perceive and/or know about their precisely, nothing like that is to be objects, say, to the deeper nature or seen in reality (in a photograph, a essence of things, except perhaps a 36 Neo-Platonist or Heidegger contem- (!) of the picture are not compara- plating van Gogh’s boots. Goodman, ble to the phonemes, for they have it is true, tells us about the numer- no opposition value: the same dot ous ways “in which the world is”, in may at one time signify an eye, order to dismiss the similarity argu- and then something completely dif- ment; but, clearly, the obvious can- ferent (Ibid.; cf. 1976:355 ff.). Eco didate for similarity comparisons is thinks these features must either be our ordinary, perceptual, Lifeworld. infinite in number, or else that they Pictures, according to Eco, de- should correspond to the elements of pend for their meaning on a code, geometry – but he also gives a list but only a very weak one; i.e. only comprising figure/ground, light con- with difficulty can they be dissolved trasts, etc. These are then combined into their elements. Oddly enough, into signs, corresponding to objects both Eco’s argument for the cod- which may be recognised: a nose, an ing of pictures, and his argument for eye, a cloud – further combined into this code being weak, are mistaken. “iconical statements” like “this is a In principle, Eco (1968:212, 217) horse”, or perhaps “this is a standing says, any analogue sign may be dis- horse seen in profile” (p234f; cf. also solved into a digital sign. As an ex- Sonesson 1989a,III.4.3). ample, Eco (1968:215ff; 1976:323 What Eco says about pictures f) cites the photograph, which must could equally well be said about be dissolved into dots before being verbal language. There is no way of reproduced in the newspaper; and finding the elements of a language the telephotographic technique for without being acquainted with its transmitting photographs from one particular scheme of interpretation. place to another. Now, these cases It is a common experience that one are plainly irrelevant: only when the is unable to discover even the lim- dots are brought together again will its between the words when listening configurations appear, and the pic- to an unknown language. In verbal ture is seen as such. The possibility language we can only distinguish of transmitting photographs dot by optional features from distinctive dot is no more relevant than the pos- ones in relation to a given content; sibility of making any picture into a and the case is of course the same for jigsaw puzzle. pictures, only that these would seem Such a “digitalisation” is dif- to have different contents on differ- ficult in practice, Eco continues: ent intensional levels, so that a fea- it is hard to tell the elements of ar- ture which is optional on one level ticulation apart (1968:203) and to becomes distinctive on another. As distinguish optional features from we said in lecture one, when discuss- distinctive ones (p204). The signs ing “Las Meninas”, more traits are

37 necessity for determining the sign ties between relationships, and met- “the Spanish infanta” than for the aphors, which involve relationships sign “little girl”, although these traits between relationships. For if we take partially overlap. Physically identi- this categorisation seriously, ordinary cal sounds will be heard as differ- pictures are not images, but rather ent phonemes in different contexts, some curious case of diagrams or, just as Eco’s dot changes meaning rather, metaphors. Indeed, perceptual with context; and in fact, physically psychology has shown us that what is quite different sounds are heard as similar between the expression plane the same phoneme (cf. Malmberg of a picture and reality as depicted 1966). On Eco’s account, it would can only be found on the level of seem, verbal language would also be relations between relations between a “weak code”. relations (cf. Gibson 1982; Sones- In the first version of his cri- son 1989a). The Annigoni portrait is tique of iconicity, as we have con- a perfect illustration of this point.24 sidered it in this section, Eco makes The only candidate for an image in the important observation that there Peirce’s sense would seem to be a is a sense in which the picture and colour sample, of the kind you bring its referent may be seen, on closer home to verify whether a particular inspection, to have no properties shade of paint will go together with in common. Nonetheless, he seems the rest of the furnishing of your to end up presenting as his own the apartment: here the simple quality of very same theory of pictures resem- colour is supposed to be the same. Yet bling their object in a few selected a picture is of course different from properties that he rejects in Morris. a diagram in the ordinary language To say that a resemblance exists be- sense of the term, which is included tween the picture and the model of among the Peircean diagrams: per- perception of the object is hardly to haps we could say that the picture, add anything new. Both when argu- as well as the diagram and the meta- ing that pictures depend on a code, phor, are caused by the perception of and that this code is “weak”, Eco re- relations between relations of some lies on erroneous conceptions of ver- or other degree, but that pictures bal language, which are contradicted are experienced as statements about by modern linguistics. similarities of simple qualities, while

Eco would have been better 24 Another illustration could be the advised to use his insights in order story Eco (1998: 16f; 1999: 355f) tells us to criticise the Peircean division of about Diderot observing a painting by icons into three types: the images, Chardin and speaking first as if he was seeing the real objects, and then stepping which rely on simple qualities, the closer to note the layers of paint on the diagrams, which concern similari- canvas. 38 diagrams and metaphors are seen as expression for the rather limited kind statements about relationships. Thus, of conventionality propounded by the similarity, which serves as a con- Eco in his first critique of iconicity. dition upon the perception of the pic- However, the same premises, which ture signs, is not of the same order have previously (in Eco 1968) been as the similarity, which is part of the used to demonstrate that iconic signs meaning of the self-same sign. depend on weak codes, now serve to show that there can be no iconic Saccharine by any signs, nor any figurae of iconicity, other name… A plea for but only “iconic texts”, which can- proportionality not be further analysed (1976:354ff). Most elements of Eco’s earlier cri- These reasons continue to be invalid, tique of iconicity recur in the iconic- even though the thrust of the argu- ity chapter of A Theory of Semiot- ment has now been changed. ics (1976:325ff), but the main point This time, it seems that Eco now is different. First, Eco tells us seriously rejects Morris’s selection that there is no mutual implication model: for although some of the ear- between “digital”, “arbitrary”, and lier examples and formulations reap- “conventional”, nor between “ana- pear, they have lost their central po- logue”, “motivated” and “natural” sition in the argument. In their place, (1976:323f; cf. 1968:208ff; and Son- new elements come to the fore. For esson 1989a,III.1.4.). And then he ar- instance, common sense tells us, says gues that it is naive, not only to think Eco (1976:329f), that sugar and sac- that iconic signs have the same prop- charin are similar, but, in fact, their erties as their objects, or are similar chemical formulae share no prop- to them, analogous to them, or mo- erties, and in its visual appearance, tivated by them, but also to claim saccharin should rather remind us of that they are arbitrarily coded, and salt. It is only the effects produced that they can be analysed into per- on our papillae by sugar and saccha- tinent units forming more than one rin that resemble each other, and this articulation. On the other hand, Eco they do, because the distinction be- (1976: 327) still believes that iconic tween sweet and salty things is tak- signs are culturally, or convention- en to be fundamental in our culture. ally coded, without however being Thus, at the very moment that Eco arbitrary, nor discrete. In the case of rejects distinctive features, he argues the dimension conventional vs. natu- for the existence of constitutive op- ral, however, Eco clearly opts for the positions in absentia, which are, at first term: iconic signs are not natu- least in part, responsible for the im- ral, whatever that means. So far, we pression of similarity. This is no con- only seem to have a more explicit tradiction, however, for these oppo-

39 sitions must be abductive rather than tainly not in prototypical pictures. structural, i.e. they are not sufficient The only thing that is necessary to in themselves to interdefine the ele- know beforehand, we noted, is the ments, but depends on our observa- hierarchy of the general and particu- tions of the regularities appertaining lar Lifeworld we inhabit. to certain elements of the Lifeworld If “analogy” is not just another (cf. Sonesson 1989a,I.3.3.). Unfortu- term for similarity, it means propor- nately, Eco gives no pictorial exam- tionality, Eco (p 337 f) claims; but ples of this constitution process: but then, he thinks, it must be a rule, such examples are easily discovered which establishes a relation between in the Bakairi version of the differ- at least three (?) terms. This rule may ence between a bird and a man (Fig. state that, if 10 corresponds to 1, then 11), and in some pictograms for la- 20 corresponds to 2; or it may just as dies’ and gentlemen’s lavatories as well stipulate that as 3 corresponds well as the correspond signs of Blis- to 9, 6 shall correspond to 18; there- symbolics (Fig. 3.). Nevertheless, it fore, Eco concludes, no similarity seems intuitively clear, that the more is required between the first and the important such abductive opposi- second term but this is created by the tions are in the constitution of a sign, rule (cf. 1976:335, 346, etc.). Eco is the less iconic it is felt to be. And if of course right in thinking that the a number of such oppositions tend first two terms of a proportionality to form a structure, or something do not have to be similar in any way; vaguely structure-like, iconicity fur- although, in his first example they ther decreases.25 are (cf. Sonesson 1989a,III.1.2.). But In Eco’s view, however, all even in Eco’s second example, simi- similarity is based on precise rules, larity is presupposed: not, of course, which have to be learned, and which a similarity of the first two terms, but stipulate which aspects of the object of the relation between these terms are pertinent. Only when we are fa- and the relation between the second miliar with the rules, Eco believes, pair of the proportionality. In fact, will we be able to discover the moti- there are a number of relations be- vation of the signs. This is, of course, tween 3 and 9; therefore, given Eco’s what Mallery (1881) called “reason- three terms (i.e. 3:9::6:x, if that is able” similarity which is, as we have what he means), we could neither fix already argued above, only found in the fourth term, nor determine the droodles, identities and other kinds relation. But we could still predict of secondary iconical signs, and cer- the few possible ones: if the rule is to multiply the first term of each pair 25 We will return to the secondary by three, the second term of the sec- use of structural elements in pictures in Lecture 4. ond pair will be 18, as in Eco’s ex- 40 ample; if the rule says we should add ganisations may be preserved. First, 6 to the first term, the term searched it is possible that reality, i.e. our par- for is 12; and if the rule requires us ticular Lifeworld, is not continuous to multiply the first term of the pair or at least that some parts of it are with itself, the term to be mentioned not; indeed, we have argued that re- is 36. If we are presented with all four ality is categorised (cf. Sonesson terms, however, they will make up a 1989a,I.2.1.), that is, discontinuous. simple structure in praesentia, which In this case, organisation is preserved only serves to select one among the if the semiotic system uses the same possibilities given by a structure in categories as the Lifeworld and re- absentia (cf. Sonesson 1989a,I.3.4.). lates them to each other in the same Similarity is defined by the latter, not way; but if Lifeworld categories are by the former, that is, it is defined by abolished and/or the members are the structure of mathematics. redistributed among the categories, And yet, even if Eco’s math- the system modifies the organisation ematical parallel proves wrong, he (as in Matisse’s “Nu bleu”, analysed might be right in his claim about pic- in Sonesson 1989a). Although Eco tures, so now let us consider this the- claims “iconic texts” can be no fur- sis independently. Suppose we want ther analysed, his notion of analogy to know the length of some marks: plainly supposes both expression in a system describing a continuous and content (or referents) to be seg- world, there will be different expres- mentable and differentiated. Neither sions for the mark which is 3/4 cm, continuity nor the precise categories and for the mark which is 1 1/4 cm, need to be preserved, but the rela- but in what Eco would call a “digit- tions between the categories have al” system, they might both come out to be kept up; and this would seem identical (cf. Roupas 1977:69ff).26 to presuppose the separability of the The former example is a particular categories. case of a structure–preserving map- The only example considered ping (cf. Janlert 1985:184), and this by Eco (1976:33ff) is Peirce’s ex- raises the question which other or- istential graphs, where the relations between the propositions of a syllo- 26 Roupas is really discussing gism are rendered by concentric cir- Goodman’s distinction between ‘dense’ cles. For instance, a reasoning like and ‘finitely differentiated’ systems. Cf. “All men are subject to the passions Sonesson 1989a and 1995 and Lecture 4. Although he uses a figure taken from – all saints are men – all saints are Palmer, instead of my example from subject to the passions” is expressed Roupas, Eco (1998: 12f; 1999:344ff) as the inclusion of the circle of men now makes the same argument for in that of the passions, and the inclu- the necessity of postulating general categories. sion of the circle of saints in that of 41 men. Eco censures Peirce for claim- the relations correlated in iconicity ing this to be a completely analogi- are prior to their relata, i.e. that they cal, iconical sign, for that which is are relational properties. Something represented is not even spatial. In- of this kind seems to be suggested by stead, he thinks there is a convention the theories of perceptual psycholo- which establishes that space a is to gists such as Gibson, Kennedy, and be taken to be related to space b, just Hochberg (cf. Sonesson 1989a,III.). like the element a’ is related to the While still claiming pictures to be element b’ (p 335). To begin with, conventional, Eco now denies the such an operation clearly requires possibility of analysing them into both the spaces and the elements to features. He fails to realise that his be segmentable and differentiated. own examples supposes there to be In the second place, while there may a basic motivation in the relationship be conventional elements in such a between the picture and its referent, specialised representation as an exis- as well as some kind of segmenta- tential graph (better known as a Venn tion of both reality and its signs. In- diagram), this proportionality is es- terestingly, to many psychologists sentially based on an iconical repre- engaged in the study of perception, sentation of the topological property pictures are motivated and resolv- of inclusion, a very abstract property, able into features! whose representation is in no sense In relation to Eco’s view, the less iconical that that of visual ap- conception propounded by Groupe pearance. As so often, Eco’s critique µ (1992:135ff) separating the iconic of iconicity (and that of many oth- sign into three elements: the signifier, ers) is based on the misconception the referent, and the type, has at least that iconicity is somehow essentially the merit of positing a categorical visual. level which Eco fails to recognise. In any case, the only property It will be remembered that I pointed preserved here as such is inclusion. out above that the ground seems to In a typical picture, however, a great account for the division between the number of relations obtain between immediate and the dynamical ob- each two units, or even between eve- ject on the side of content; but I then ry two elements of the pattern. A con- proceeded to argue that there must vention specifying all these relation- be a similar division on the side of ships would have to be very complex expression. Indeed, if the ground is indeed, and would probably have to tantamount to abstraction, as Peirce be made separately for each picture. says, and if abstraction is the genera- As a general theory of iconicity, or tion of types, then we should readily even of pictures, this conception is accept the distinction, suggested by not feasible. But suppose instead that Groupe µ, between the referent and

42 the type; but we should add to it the sence of its referent); 2) It is causally parallel distinction between the sig- produced by its object; 3) It is not nifier and its type. independent of the medium or the channel by means of which it is con- Mirror, mirror, on the veyed; 4) It cannot be used for lying; wall…What do you 5) It does not establish a relationship signify? between tokens through the interme- In his most recent work on iconicity, diary of types; 6) It does not suggest Eco’s (1998; 1999) sometimes seems a content (or only a general one such to give in completely to his critics, as “human being”); 7) It cannot be such as the present author, Groupe µ, interpreted further (only the object to etc., and then taking the “motivated” which it refers can). I will deal with nature of icons to a further extreme27. all these arguments in the follow- At other times, however, Eco (1997; ing, though not exactly in the order 1998: 10; 1999:241f) reaffirms the in which they are presented above. conventionality of picturehood, now In fact, they cannot be discussed in taken to be compatible with a ba- the order given, since some of the sic iconicity (which is exactly what affirmations turn out to be intercon- Eco’s critics such as myself said). nected. The most remarkable part of Eco’s On Eco’s account, then, the mir- latest critique, however, is his ex- ror is pre-semiotic. It is, according tension of the mirror model to some to Eco, an absolute icon, in Peirce’s phenomena that most semioticians, sense, and it would thus have been a including the earlier Eco, should perfect iconical sign, if it had been have considered to be iconical signs, a sign. This is certainly saying too and, in a way, it seems, to all iconical much, since an absolute icon, in signs. Peirce’s view, can only exist for a According to a theory first pre- fleeting moment, even in thought.28 sented in Eco’s (1984) dictionary en- Eco goes on to say that the mirror try on the mirror, and enlarged upon is no index, because, unlike a letter in his recent writings (1997, 1998, containing personal pronouns such 1999), the mirror is no sign. In par- as “I”, which continues to refer to the ticular, Eco quotes seven reasons for writer, a mirror sent by post ceases denying the sign status of the mirror, to indicate the sender and will now which can be summarised as follows: point to the receiver. It “is not even a 1) Instead of standing for something it stands before something (the mir- 28 There are certainly some ror image is not present in the ab- serious issues looming behind these interpretations, which cannot be dealt 27 For some examples, see the notes with here. See, however, Sonesson, 2003b; to the earlier discussion of Eco’s critiques. in press a, and Lecture 2. 43 Firstness in the Peircean sense” (my tinuity in time and/or space (they do italics), Eco continues, because it is not think the mirror image is part of already a relation, and thus a Sec- themselves) but rather that concerned ondness. On this point, I can only with the different nature of the two agree with him, except for the word- correlates (the cat takes the image of ing: it would have been more prop- a cat to be another cat). er to say that the mirror is already Let us start with the first argu- more than Firstness. In fact, I have ment, according to which the sign, said the same thing, not only about but not the mirror, supposes the ab- the iconic sign, but about something sence of the referent. In the case of more general which it supposes, the many signs, the content (or rather the iconic ground: it is already a rela- referent) is present together with the tion. In any case, if it is a relation, expression. Many signs function in it is a least already Secondness, so the way they function only in pres- why should it not be causal, as Sec- ence of their referent: this is the case ondness is in strict Peircean theory? with those pictures of birds with the Or, if we take causality to be a suf- names of their species written below ficient but no necessary criterion on them which are attached to the bird Secondness (as I would prefer), then case in the zoo. Indeed it is the case it might still be causal (cf. Sonesson with much of our language use: for 2003b; in press a). although the female personal pro- With reference to our more pre- noun, for instance, figures extensive- cise concept of sign, I see really no ly in the absence of a possible ref- reason to deny the sign character of erent, it does not tell us very much; the mirror: something which is com- and talking about the gorilla in front paratively more direct and less the- of it adds more than only shades of matic, the mirror image, stands for meaning. something which is less direct and Of course, bird pictures, and more thematic, the object in front of much of verbal language, function the mirror; and the person or thing also in the absence of their referent, in front of the mirror is clearly dif- although they function differently. ferentiated from the image in the Other signs, however, are more radi- mirror. Of course, animals and small cally dependant on their referents. children may have difficulty making Indeed, weathercocks, pointing fin- this differentiation, but that is exact- gers, cast shadows, and a lot of other ly what happens in the case of signs, signs cannot mean what they mean, as Piaget has indicated. The kind of if not in the presence of the object differentiation that does not obtain they refer to. Indeed, as we shall see, for animals and children is appar- co-presence is a precondition at least ently not the one involving a discon- for one kind of indexical sign. The 44 sign character of these signs only en- this means that there is no contradic- dures as long as the object is in their tion between causal production and presence, and such was no doubt the sign function. Even if causality originally the case also with person- does not define the sign function, nor al pronouns such as “I”. The classi- even the peculiar kind of sign termed cal definition of the sign, which Eco index, it is not incompatible with it. here refers to, is wrong in requiring These facts no doubt explain the absence of the referent. Differen- why Eco feels the need to demon- tiation, which defines signs, must be strate, not that the mirror is not an distinguished from absence. icon, but that it is no index. But his We shall now have a look at the argument is connected to another second argument, which says that the feature, which he takes to be char- mirror image is causally produced acteristic of signs: that they im- by its object, which is not the case ply the possibility of lying (cf. Eco with the picture sign. Thus, causality 1976:339f; 1984:202ff). Employing is taken to exclude the sign charac- one of the more classical instances ter. This is curious, because one of of indexical signs, Eco (1984: 214) Peirce’s most currently quoted defi- claims that one may use certain kinds nitions of the index (which is a sign) of chemical substances to produce says that it depends on a causal re- smoke, thus giving the appearance of lation between expression and con- there being a fire somewhere in the tent. In fact, a lot of indices depend neighbourhood. On the other hand, on causality, from the knock on the he maintains, we cannot lie with a door (caused by the hand) to the cast mirror image, as opposed to making shadow, the death mask and – some- something, which is not a mirror im- thing that is definitely also a picture age pass as such. – the photograph. But what exactly does this However, if we choose to define mean? The human species has been indices in terms of causality, then – accustomed for a very long time following the “structural argument” to interpret smoke as a signifier of which is have formulated elsewhere fire; but if chemical substances be- (cf. Sonesson 2001c and above) – it come common causes for producing will be impossible to exhaust the do- smoke, we will have to revise this main of signs by means of only three judgement. This is so because the sign types: indeed, many examples sign, in this kind of case, is based of indices given by Peirce are cer- on our observations of common tainly not causal. ”Real connection” Lifeworld regularities. Or to take a (exemplified most notably by conti- more topical example: we have been guity and factorality) is therefore at accustomed for the last centuries to better definition of indexicality. Yet believe that pictures which have a

45 peculiar granular appearance, which the weathercock has to do with the we call photographs, have been pro- relative importance of the constant duced by a more or less indirect con- and variable element in the meaning, tact between the surface of the im- that is, with Eco’s “content”. age and the objects represented; but This thus brings us to the sixth computer pictures are already forc- argument, according to which the ing us to reconsider this interpreta- mirror does not suggest a content, or tion. It is not clear if this means that only a general one such as “human those computer images are meant being”. The difference between the to pass as photographs; or whether pronoun, the mirror, and the weath- they are still the same kind of im- ercock depends on how far the con- age, produced in another way. So, if stant elements of signification (Eco’s something which looks to us like a “content”) go in a sign. We know that mirror image turns out to be capable “I” refers to the speaker or writer us- of appearing without a person being ing a particular instance of the sign, in front of the mirror surface, should and there are usually other ways of we conclude that it is something else discovering who the speaker or writ- trying to pass as a mirror image, or er is, or at least that he is not identical that mirror images are not what they to ourselves. The constant element of used to be? the weathercock is the indication of Pronouns like “I” change their the direction of the wind in the here meaning each time they are used, and now. The constant element of the yet retain this meaning once they mirror is the rendering of something are written down (or, one might add, visible placed presently in front of it. when the speech is recorded on tape). The variable elements are too many The mirror, Eco contends, continues ever to be retrievable; but it may yet to change its meaning for ever. How- be maintained that they all share a ever, the weathercock, one of Peirce’s number of predicates, such as being favourite examples of an index, be- visible, present in the here and now, haves in all these respects more like and so on. the mirror than like the pronoun: if The opposition that Eco posits sent as a message from the season- between mirrors and signs is seem- al resort, it will indicate the direc- ingly the same as other thinkers (e.g. tion of the wind at the place where Gombrich) have always postulated the receiver lives, not that which the as a difference between pictorial and sender observed before putting the verbal signs. It is often expressed device into the parcel. This is not to as a difference between singularity say that the weathercock functions and generality. A picture, it is said, exactly as the mirror. The difference can only show an individual person, between the mirror, the pronoun and not “a guard in general”, but some 46 very particular guard with individ- tions are irrelevant. The fact that we ual features. As applied to pictures, may see an object, and know that it is these arguments are no doubt wrong. there, without it having been pointed It is possible to construct very about to us, does not make the point- stract or schematic pictures (chil- ing finger less of a sign, and indeed dren’s drawings or logograms, for an index. Nor does the weathercock instance), which only convey very cease being an index just because we general facts. Indeed they are about may be able to discover the direction “a woman in general”, etc. But even of the wind already from the impact a photograph with an abundance of it has on our body. individual detail will only signify to Curiously, Eco all the time talks me something like “a young woman as if mirrors were only used to look dressed in 1920ies apparel”, if I do at ourselves. In fact, mirrors are not not happen to know the person in only used for seeing oneself but for question. This also applies to mir- seeing others and other things. Some rors: while looking at myself in the mirror types are actually specialised mirror, I may suddenly see some for such purposes. The rear mirror configuration, which I interpret, as of a car is used for discovering other “a man appearing behind my back”. cars coming from behind. A dentist I do not have to recognise him as uses a mirror to investigate the status Frankenstein’s monster to be fright- of our teeth. Indeed, a woman may ened. In mirrors, as in pictures, sin- know very well that she has lips, and gularity is not, in the last instance, in still use a mirror to ascertain that the sign, but in the use to which we she is putting the lipstick on to her put the sign. best advantage. Even supposing that At this point, it will be conven- Eco’s argument would have some ient to attend to a kind of generalisa- relevance, these mirrors are not used tion of the second argument: accord- to show something that is known ing to Eco, the mirror image is not an beforehand, as the presence of cars, index for the person in front of the teeth, or lips, but to investigate spe- mirror, because we do not need it in cial properties of these objects. Thus, order to know this fact; only the lack they are not “symptoms”, if we take of an image when the Invisible Man this word in the ordinary language or an vampire passes in front of the sense of an indexical sign that is un- mirror could perhaps be admitted to intentionally emitted. be a symptom. Nor is a mark on the We can now go back to the third nose observed in a mirror an index, argument, which claims that the mir- Eco says, because it is no different ror is not independent of the medium from the mark we observe directly on or channel by which it is conveyed. It our hand. However, these observa- is not clear whether Eco here means

47 to speak about the different materials do paintings existing in one single employed, or about the fact of trans- copy (if we do not admit the repro- ference being possible.29 Histori- ductions as tokens, which most art cally, mirrors have been made out of historians would vehemently deny). different “substances”, that is, differ- Nor do any momentary signs com- ent materials: once upon a time, they ply with this criterion, from point- were made from metal sheets, which ing fingers to weathercocks or cast explains that Saint Paul could talk of shadows. For though the finger may us seeing “obscurely, as in a mirror”. endure, as does the mirror, the par- In this sense, the argument is histori- ticular act of pointing, just as that cally wrong.30 On the other hand, if of mirroring, does not repeat itself, Eco means to say that a particular nor does it admit a change of “sub- instance of mirroring is not trans- stance”. ferable from one mirror to another, The notion of momentary signs then something equivalent is true of does not appear to exist for Eco, many signs. If so, this criterion is and yet it is an important one. The hardly possible to distinguish from problem seems to be that Eco thinks the fifth one, according to which something, which once is a sign, signs suppose types to be mediated must then always be one. Howev- by tokens. er, if we exclude all signs that are Therefore, we now proceed to only momentarily signs of some- the fifth argument, which tells us that thing, most of the examples given by the mirror does not establish a rela- Peirce and others will not be eligible tionship between tokens through the as signs. You do not have to cut off intermediary of types. We may cer- a finger and send it off by post for tainly agree that mirrors do not com- it to change completely its meaning; ply with this criterion — but neither even in its natural position, the content to which it points is continuous- 29 The first is the common ly changing. In fact, weathercocks, interpretation of Hjelmslev’s distinction between form and substance, but the latter pointing fingers, and pronouns, seem is closer to being the correct one. to have functioned (and functioned 30 Indeed, Gregory (1997: 47ff, 57) as signs) much like the mirror, be- who maintains that the first mirrors were fore different techniques for preserv- made of polished copper, silver, gold, ing tokens (as opposed to types) of or obsidian, and before that were bowls signs were invented, a process which filled with water, the direct descendants of which were wetted slates hung on the perhaps begun with writing and now wall, also suggests that S:t Paul’s looking has reached the state of computer glace may have been have been an memory. This is also the only reason instance of “obscura mirrors which were Eco quotes for not recognising my so poor optically that they stimulated the imagination more then the eye.” suggestion (from Sonesson 1989a) 48 that the mirror is a “hard icon” in In fact, there are no zero-degree Maldonado’s sense: the indexicality mirrors: as people who use mirrors and iconicity of the mirror is only professionally, from dentists to sales momentary. But this reason will not clerks at the dressmakers, will readily do, since it would force us to deny point out to us, all mirrors are adapt- the sign status of numerous other ed to particular uses. Actually all signs. mirrors lie, or, more precisely, they We will now turn to the fourth interpret: they are adapted to differ- argument, according to which the ent professional uses, the “channel” mirror cannot be used for lying. having a particular fraction in the There is a more immediate retort: case of the dentist, a particular tint there are indeed mirrors that practice for the dressmakers, etc. constantly the art of lying. As Vilches Eco’s final argument, the sev- (1983:21) points out, the very busi- enth one, says that there is no chain ness of the mirrors in the Fun House of interpretants resulting from the is to do that. Indeed, it could be add- mirror as in the case of the sign. The ed that they lie in a systematic way: mirror cannot be interpreted further there is always the same distance – only the object to which it refers between the referent and the picture can. But of course the mirror may be object, at least from a given position the starting-point for a chain of inter- in front of the mirror, so there is ac- pretations, just as any feature of the tually a content (and since it is an ab- common sense Lifeworld. That is stract predicate, we could say there what the dentist does, the woman ap- is a type), which mediates between plying her lipstick in front of the mir- the subject and the mirror image. In ror, the driver who sees a car coming fact, Eco (1984: 217ff) considers the up behind him, the person seeing the case of “distorting mirrors” but re- monster (which is not a vampire) in jects them as counter-proofs for very the mirror, etc. Eco would say this obscure reasons. But this is not all: if amounts to interpreting the object, distorting mirrors are possible, then but this would only be true if we had all mirrors are no doubt somewhat accepted his other arguments. If mir- distorting (as are all photographs; cf. rors are adapted to their particular Sonesson 1989b; 1999), although we uses, as we just saw, then it really is are too accustomed to them to realise a question of interpreting the object it. So the mirror image is also con- as it is given in the mirror, roughly veyed to us with the fidelity permit- similar to the interpretation of ob- ted by the particular channel. This all jects through the intermediary of a amount to saying that, like the pic- picture. ture, the mirror has its “ground”, its Contrary to Eco, I think there is principle of abstractive relevancies. every reason to consider the mirror

49 to be a sign, an index and an icon: disappearance of the latter. Very lit- indeed, because it combines the tle seems to be left for the conven- functions of index and icon, it is (as tionalist theory of pictures, which, in observed in Sonesson 1989a;III.3.6.) other passages, Eco seemed willing what Maldonado has called a hard to maintain, in spite of certain modi- icon, comparable to X-ray pictures, fications. We are back where the thermograms, hand impressions on started, before Eco’s first critique, at cave walls, “acoustic pictures” ob- Barthes’ “message without a code”. tained by means of ultrasound, sil- And once again, iconicity appears as houettes, the configuration left on a complete mystery! the ground by a man out walking in Hiroshima at the moment of the nu- Summary clear blast, and pictures made with In contrast to Bierman’s and Good- “invisible light” to discover persons man’s logical critique of iconicity, hiding in the woods. It is true, as Eco Eco assembles a number of argu- (1999: 369) points out against me, ment, which are more general, com- that these latter signs leave some- mon-sense, and “cultural”. Many of thing endurable which serves as the these arguments are intriguing, but expression plane: but if observed at they always seem to miss the mark. the exact moment when a shadow is In the first version of his critique, cast, or an image appear in a mirror, Eco suggests that pictures are con- these latter phenomena may give the ventional and divisible in to features. same scientific assurance of exist- But basically, he seems to confuse ence which Maldonado ascribed to the conventionality of the Lifeworld the hard icons. with that of the picture sign (as we From here on, Eco (1998: 22ff; will show in more detail in the next 1999: 371ff) goes on to suggest that section). The second version of the television image is similar to Eco’s critique involves one impor- mirrors, and not to ordinary pictures, tant change: pictures are still sup- with the provision that the reflection posed to be conventional, but they passes over an electronical chan- now are alleged not to be divisible nel.31 He then imagines that the film, into features. Yet Eco’s argues for a the photograph, and the hyperrealis- relation of proportionality between tic painting are “frozen” mirror im- the sign and the world, which clear- ages. The difference here is simply ly implies that both the sign and the that now expression is separate from world must be segmented into parts. the content and thus can survive the The third version of Eco’s theory of iconicity starts out from the mirror, 31 We will see in a later lecture that which is claimed not to be a sign, this is unfeasible, quite independently of and then generalises this conception the mirror being a sign or not. 50 to (almost) all pictures. However, of the process of resemanticisation, thanks to our more stringent defini- which is to found, in part, in Woll- tion of the sign as well as of iconic- heim’s notion of “seeing-in”, but, ity, we can show that the mirror is principally, in Husserl’s considera- indeed a sign, and an iconical sign at tions on “pictorial consciousness”. that – and in that sense, it is similar Returning to my earlier discussion of to pictures. In other ways, however, these authors (in Sonesson 1989a), I as we shall discuss in the next sec- will expand on my previous conclu- tion, pictures are peculiar. sions to suggest a modification of the µ-model of the picture sign. Finally, 3.4 On the way to I will suggest a way in which the Pictorality: The case for iconical and indexical grounds, liber- resemanticisation ated from their function to constitute signs, will offer the foundations for a We have seen so far that pictoral- pictorial rhetoric. We will start, how- ity is usually confused with iconic- ever, from another end: by taking up ity, which begs the question of to Eco’s arguments again, and adding what extent pictures are special. The to them, to pinpoint the conventional same arguments, which Bierman di- aspects of pictorial and other iconic rected against iconicity, were used grounds. by Goodman in the case of pictures; and Eco constantly confuses pictures The conventional residue and icons, even in his latest work. of iconic signs Klinkenberg and Blanke even ex- Although the general argument for plicitly declare that they are going to the conventional character of pictures use the term iconicity to mean pic- is no doubt mistaken, it remains true torality. Peirce would probably take (as Peirce was the first to recognise), pictures to be “images”, that is, icons that all real pictures are largely con- based on simple qualities, but as we ventional. Some of the conventional have seen, this will not do, if he is re- traits of pictures depend, as we have ferring to the ground that motivates suggested, on the general character picture signs (which must consist of of the human Lifeworld. Many of relations between relations), as op- the other conventionalities attributed posed to the impression they create. to pictures are really inherent in the After showing more in detail why particular socio-cultural Lifeworld. pictures cannot by Peircean “imag- This means that, whenever some pe- es”, I will suggest that pictures are culiarities of an individual or a thing, organised in a special way, which some traits of the woman or the ze- I will call resemanticisation. I will bra, are locally given importance, also delve deeper into the foundation they also make up the features given

51 primary importance in a picture. and Kroeber’s introduction, p xxiv). Pictures, being a kind of visual Thus, for instance, in Mallery’s thinking, are required to follow the (1880-81) dictionary of manual ges- phenomenological rule of all think- tures we discover a great number of ing, according to which an object can different signs bearing the meaning only be seized each time from a par- “woman” or “female”: imitations of ticular point of view, and not in its the breasts; of the female sex organ; entirety, which means that a choice of the undulating contours of the fe- has to be made among the proper male body; of small size; of long hair; parts, the perceptual parts, and the and of the peculiar hairdo of the In- attributes of the object. Moreover, dian woman, with braids to the sides. much thinking, also that which goes This is really the story of the blind on in pictures, is made in terms of men and the elephant all over again: prototypes, that is to say, constru- the elements are all similar, but the ing an object as an approximation to way they are selected and divided a more typical instance of the same up into segments must be separately class; and even abductions and sim- justified (cf. Sonesson 1989a: 43ff, ple structures often intervene in the 223ff). This story may be recounted constitution of pictorial signs. using pictures. In fact, even a sign grounded in Thus, the iconical sign, like resemblance must pick up some of any perception, is unable to grasp its the infinite number of properties of object in more than one or a small the object which it takes as its signi- number of its aspects at the same fied, and reject all the others, in order time. Thus, for instance, a thing to constitute its own signifier. Only must be perceived from a particular some of the properties of the content point of view, and must likewise by are pertinent, or relevant, within the so rendered in a picture, or by means domain defined by the sign func- of an iconic gesture; even Cubism is tion. This appears to be true, not unable to integrated more than a few only of pictures, but of all iconical perspectives at a time. This is true signs, which is why there can be no not only of the perceptual parts (per- pure iconicity. In his early study of spectives) of a thing, but also of its manual gestures, Garrick Mallery proper parts, and of its attributes: an concluded, as we noted above, that object appears as a division block of a many of these signs seems “reason- more extended perceptual world (the able”, because the similarity between face as part of the body, the body of the sign relata could be observed by the room, the room of the apartment, a person acquainted with the cul- the apartment of the city, and so on); ture, or once the sign had been ex- and one or other of its properties is plained to him (Cf. Mallery 1881:94f highlighted by the way it is present- 52 ed (as a human being, a horseman, a lar some perceptual parts will be fa- general, a husband, and so on). The voured over the others, at least if there manual sign for woman obviously is no particular reason for picking represents a woman from a particular out some specific part. This choice perceptual perspective (the braids or is determined in multiple fashions. the curves are seen from the front), First, manual gestures, like pictures, selecting some proper part to the ex- carry with them the conventional clusion of others (hairdo, sex, or more traits of the Lifeworld in which they global properties like size and nature are first used. Even when referring of border lines), and insisting on par- the white man’s woman, the user of ticular attributes (secondary or even the American Indian manual signs culturally defined sexual characteris- will make the sign depicting the two tics). Also petroglyphs will insist on braids typical of the Indian woman’s some particular perceptual angle (a hairdo. In the context of the prehis- human being from the front, a ship or toric Lifeworld, it would seem, the an animal in side view, a plough and inclusion of a penis, or of some kind a chariot from above), some proper of weapon, both serve equally well parts (arms, legs and trunk of a per- to designate the male sex of a human son, sometimes no head, but often figure (see Burenhult 1981; Janson et sexual organs), and some attributes al. 1989), even though one is a body (sometimes man versus woman, but part, and other a cultural trait; more- never child versus adult, etc.). Ana- over, they may appear even when no ti’s “prayer” only conveys some very sexual act is involved, in one case, general traits of bodily shape, arms, and no war scene or hunting party, in legs, trunk, sometimes a head; and it the other. In the same way, the raised picks out one conceivable position, arms of the “prayer”, do not neces- that of raised arms. sarily stand for praying at this par- 32This choice is often not made ticular moment, even supposing that in an entirely arbitrary fashion, but was a possible interpretation of the some proper parts rather than oth- arm position in question at the time. ers, some attributes, and in particu- Second, like all thinking, pictures and manual signs designate cat- 32 As is clear from the name, Anati egories of things by describing their (1976) takes this figure to be engaged prototypes, that is, the best instances in praying. This seems to me a risky of the category. Thus, although not all hypothesis. Perhaps there really are anthropological universals for praying, but women have a markedly curvaceous I think it is safer to describe this figure as body, the manual sign involving an picture of a man, but only a droodle of undulating movements may be em- praying. If so, it remains to show that the ployed to designate them. Similarly, key is universal. Cf. Sonesson 1994a and Lecture 2. the petroglyph showing a man urging 53 a mule, or driving a plough, with his 1976: 400f). What is prototypical to over-sized penis in erection (repro- one culture may not be so to another, duced in Anati 1976:128 and Janson however, for old Chinese pictures, et al. 1989:20, respectively), may not and the early stages of their writing, be making love, either to his plough, show that they preferred to consider or to his mule, but is simply shown as a nose from the front, not in profile, a prototypical man, first because his as we do (cf. Lindqvist 1989:33). sexual organ is emphasised, and sec- In petroglyphs the plough or the ondly, because it is shown in its pro- chariot is often shown form above, totypical state (which is not to say its the animals attached to it from the most common state). Indeed, as we side (and sometimes one of them have suggested elsewhere (Sones- upside-down), and the man driving son 1988; 1989a; 1990) many visual them from the front (see Janson et al. signs standing for large categories 1989:20). are better seen as idealtypes, that is, In order to economise their ex- signs the expression of which exag- pressive resources, the users of pic- gerates certain features to a point not tures and gestures, like all other signs found in real instances of the catego- may, thirdly, be content to use only ry.33 The penis, in many petroglyphs, those traits in which the designated is not only of a disproportionate size, object differs from other, similar ob- but its erection is outside of natural jects. In the case of manual gestures, contexts for such as state. Also Ana- for instance, women’s small size is ti’s “prayer” may well be a idealtype relevant, in opposition to the (rela- in this sense (Fig. 12a): it exagger- tive) tallness of men. According to ates the erect posture which differ- a well-known example given by entiates man from other animal, by Eco (1968:191), the zebra, which in adding the vertical extension of the our culture may be contrasted to the arms to the customary stature, and by horse by means of its stripes, would introducing the raised arms outside a have to be differentiated in another natural context for such an act. way in a culture being familiar with Most beings and objects clearly a single other animal species, the hy- possess a point of view from which ena, which also has stripes. A more they are most characteristically extreme example are the petroglyphs shown, and more easily identified: of the Bakairi tribe contrasting man it has been demonstrated that the with a bird (Figure 11a-b): only the side-view is prototypical to animals upper part of the figures differ, so and vehicles, but furniture, shirts, that it might be said, paraphrasing and trousers, as well as human be- Aristotle, that to this tribe man is a ings, from the front (see Rosch et al. beakless biped. Anati’s “prayer” ap- 33 Cf. Lecture 4. pears to pick out a more basic trait 54 Fig 12. a) One variety of Anati’s “prayer” (from Anati 1976:46, passim); b-c) prehistoric petroglyphs

a b)e) man;man c)f) woman woman distinguishing man from other ani- a general way, the sexual organs are mals, which is more firmly grounded more natural distinctive features be- in the general structure of the Life- tween the male and the female than world: the erect posture, exaggerated trousers and skirts; but, pictorially, by being extended to the arms. the selection of these traits, rather The signs used for man and than any of the other bodily differ- woman in Blissymbolics (and also ences between men and women, is sometimes as logograms to indicate quite arbitrary. men’s and women’s washing rooms; Taken together, all this means Figure 12b-c) may be read pictori- that, although a part of the expres- ally, but the traits included are not sion of a sign can be iconical for a chosen for the sake of a correct and part of the content, it may very well complete rendering, but to establish be included in the sign for conven- the distinction: the trousers’ legs of tional, and even, in a stronger sense, the male versus the woman’s skirt arbitrary reasons. Of course, in some (Note that, on this interpretation, the cases this inclusion is far from being pictorial equivalents of the woman’s arbitrary, but is then motivated by legs emerging below the skirt are not considerations quite foreign to the included; the rendering of the wom- pictorial rendering, perhaps by the an’s body is less complete than that necessities of symbolism. of the man). In some respects this is If the inclusion in a picture of similar to the opposition, often found certain, in themselves iconic, traits, is in petroglyphs, between the stick fig- not necessarily iconically, or at least ure with a the straight line appended, not pictorially, motivated, the exclu- which is easily interpreted as the sion of such traits does not have to man’s penis, and an identical stick be pictorially, or otherwise iconical- figure with a dot (or, strictly speak- ly, justified. Thus, for instance, Anati ing, a cup mark), which could be the (1976) repeatedly mentions the pres- woman’s vagina (cf. Anati 1976; Bro- ence of “asexual figures”, but in fact, by-Johansen 1967; Burenhult 1981; the failure to depict any sexual organ, Gudnitz 1972; Janson et al. 1989). In just as the absence of heads, in other

55 cases, may well lack all pictorial sig- in its totality, but only transposed in nificance; the proper parts in ques- time, as an undulating movement. tion may have been neutralised, sim- Rock carvings, as all other pictures ply because their inclusion was not executed on a surface, lack the third important in that context. Similarly, dimension of the real world: actu- Tilley (1991: 68ff, 102ff, 136, 142, ally, because of being carved in the 146) makes too much of what might rock, the petroglyphs, unlike most well be the same kind of neutralisa- pictures, have a prominent third ditions, when he argues that the elks, mension, but this cannot be used for lacking antlers, must be females, rendering the third dimension of the and thus embody a female princi- perceptual world. ple, whereas the stick-figures, which have phalluses or no sexual indica- The picture is no image. tion, must represent males. Sexuality The Annigoni case may simply not be relevant here. Another respect, in which Eco If Anati (1976:60) is right in (1968:188; cf. 1976:327f) does have suggesting that a couple consist- a point, but not the one he presumes, ing of one headless figure and one is when he asks what it can mean to normal one indicates the separation say that Annigoni’s portrait of Queen between body and soul in the same Elizabeth has the same properties as individual, then the headless figure is the queen herself. As he points out, really iconical, but not in the sense the nose of the real queen has three of pictorial, as we have describe the dimensions, and that of her portrait latter function above, but rather in a only has two; the surface of the real symbolical sense: no real headless nose is full of pores and other ir- person is depicted. regularities, but that of the painting Finally, it should be mentioned is smooth; and corresponding to the here that pictures and gestures are nostrils of the queenly nose, there necessarily conventional because are no apertures in the canvas, but they are themselves objects of differ- only two black dots. And we could ent categories from most of the ob- certainly go on comparing further ject that they represent. For instance, details with similar results. Thus, it since the expressive resources of can hardly be denied that a paint- manual signs are essentially the hands ing of the queen (as opposed, for in- and their movement, the (limited) stance, to an actor impersonating her vertical extension of the female body on the theatre, or even a statue) is can only be described as far as one of profoundly different from the queen its properties is concerned, its high- herself. But this does not show that est point; and the curvaceousness of the painting is conventional, at least the female body cannot be rendered not in any obvious sense. 56 It follows, instead that there is furnishing of your apartment: here something wrong with the Peircean the simple quality of colour is sup- division of icons into three types: the posed to be the same. Yet a picture images, which rely on simple quali- is of course different from a diagram ties, the diagrams, which concern in the ordinary language sense of similarities between relationships, the term, which is included among and metaphors, which involve rela- the Peircean diagrams: perhaps we tionships between relationships. Fol- could say that the picture, as well lowing these definitions, ordinary as the diagram and the metaphor, pictures are not images, but rather are caused by the perception of re- some curious case of diagrams or, lations between relations of some rather, metaphors. Indeed, percep- or other degree, but that pictures tual psychology has shown us that are experienced as statements about what is similar between the expres- similarities of simple qualities, while sion plane of a picture and reality diagrams and metaphors are seen as as depicted can only be found on statements about relationships. Thus, the level of relations between rela- the similarity, which serves as a con- tions between relations, and so on dition upon the perception of the pic- (cf. Gibson 1982; Sonesson 1989a). ture signs, is not of the same order The Annigoni portrait is a perfect il- as the similarity, which is part of the lustration of this point. Whatever is meaning of the self-same sign. similar between the portrait and the If, as I suggested above, there real queen must be searched on some is an implicit structural argument in higher level of relations between re- Peirce’s division of signs into icons, lations. And yet we do not have to indices, and symbols, then this should search for it. It is directly offered to also be true about the further division our perception. Indeed, the similar- of icons into images, diagrams, and ity is there, not in some real reality, metaphors. If it is implied that this but in the human Umwelt. The same division should be valid both for the goes for the lines, which Eco (fol- iconic grounds, and for the impres- lowing Gombrich) denied any real- sion created, then we have falsified ity. The lines are there, in the human this argument. It seems hardly sat- Umwelt. isfactory to say that the picture is a If the Peircean notion of im- metaphor at the level of grounds, but ages is not an empty category, then an image at the level of perceptual the only candidate for inclusion results, but for the time being this would seem to be a colour sample, is all we are able to do. More than of the kind you bring home to verify in the case of the primary triad, the whether a particular shade of paint present one really stands in need of will go together with the rest of the emendation. This may really be a

57 case in which we had better abandon of pictorial semiosis (cf. Sonesson trichotomania. 1989a,III.4.2-3). Saussure’s (1974:39) observa- Eco (1976) later retracted him- tion that, whereas language is oneself completely, arguing that there dimensional, painting depends on a could be no distinctive features in semiotic system deployed in multi- pictures. Thus, he ended up holding ple dimensions, has never acquired the same position as that expressed any followers, and rightly so, for it earlier and more clearly by Good- would be impossible to demonstrate man: that pictures were capable of no that these dimensions are relevant further analysis, i.e. that they were, as for the combinations (syntagms) and Goodman puts it, “dense and replete”. selections (paradigms) of some kind Density is a property of sign systems of pictorial sub-elements (cf. Sones- the possession of which implies that son 1988; 1991, 1992e; 1993c.). For no matter how close a division of the a long time, however, semioticians signs is made into smaller parts, it tried to demonstrate the existence will always be possible to proceed of some kind of minimal unit of pic- with the division, introducing a third torial meaning, sometimes termed unit between each earlier couple of iconeme, which was supposed to have items, and so on indefinitely. Den- no meaning of its own, but served to sity is semantic when it applies to discriminate the meanings of larger content units (to referents, in Good- wholes, just a phonemes do in rela- man’s nominalist terms), and syntac- tion to words or morphemes. Eco tic as far at it involves the varieties (1968: 234f; 197l: 77ff), who was an of expression (Goodman’s “marks”). early proponent of this conception, A dense system is replete when its even went so far, at one time, as to signs can be divided from many dif- suggest the existence of something ferent, perhaps an infinite number, similar to a double articulation in of viewpoints. Density and replete- pictures (adding a third articulation ness, in Goodman’s view, apply to in the cinema; 1968: 244), where pictures both as carrier of reference dots and lines made up the equiva- and as exemplifications, that is, in lent of phonemes, noses and eyes terms more familiar to semioticians, corresponded to words, and horses as “iconic” and “plastic” language and men were on the same level as respectively (cf. Groupe µ 1979). As sentences; but quite apart from the I have shown elsewhere (Sonesson fact that the latter two levels concern 1989a;III-2.3-5 and III.6.1.; 1995a), units of the referential world, which this would mean that pictures are may be rendered differently in picto- semiotic atoms, in the original sense, rial signs, such an analysis does in i.e. not susceptible of being divided no way account for the peculiarity in any non-arbitrary way into smaller

58 units. However, as I argued against all these relationships would have to both Eco and Goodman (in Sones- be very complex indeed, and would son 1989a), and as Eco (1998: 12f; have to be made separately for each 1999:344ff) now himself admits, the picture. But suppose instead that the interpretation of pictures supposes relations correlated by iconicity are the identification of general catego- prior to their relata, i.e. that they are ries, both in pictures and in the per- relational properties. Something of ceptual world. And there can only the kind seems to be suggested by be categories that may be correlated the theories of perceptual psycholo- to the extent that the world and the gists such as Gibson, Kennedy, and picture are susceptible of segmenta- Hochberg (cf. Sonesson 1989a,III.): tion.34 no “simple qualities”, but the rela- In his earlier work, however, tions between the relations coincide Eco claimed all similarity to be based in the picture and the perceptual on precise rules, which have to be world. While still claiming pictures learned, and which stipulate which as- to be conventional, Eco now denies pects of the object are pertinent. Only the possibility of analysing them into when we are familiar with the rules, features. He fails to realise that his Eco thought, would we be able to own examples supposes there to be discover the motivation of the signs. a basic motivation in the relationship As pointed out above, the only exam- between the picture and its referent, ple considered by Eco (1976:33ff) is as well as some kind of segmenta- Peirce’s existential graphs, where tion of both reality and its signs. On the relations between the proposi- the contrary, to many psychologists tions of a syllogism are rendered by engaged in the study of perception, concentric circles. Contrary to what pictures are motivated and resolv- Eco claims, we noted, there is one able into features. This is because property that is preserved here, viz. their features are relational. inclusion. In a typical picture, how- To Gibson, features are the ever, a great number of relations ob- same as “invariants, in the mathe- tain between each two units, or even matical sense, of the structure of an between every two elements of the optic array”; “formless and timeless pattern.35 A convention specifying invariants that specify the distinc- 34 Simple spatial features are tive features of the object” (Gibson not enough, of course, as I argued (in 1971:31); “relations between rela- Sonesson 1989a) against both Kennedy tions, for which there are no names and the usage to which Bouissac puts and no mathematical expressions”; Hubel & Wiesel’s spatial detectors, and as Eco (1998: 14f; 1999:350ff) now argues in Sonesson 1989a,III. would seem to take against Hubel & Wiesel. further the claims made by Stjernfeldt in 35 This argument, formulated already press. 59 “invariants of structure which are (1974a: 44) is aware of this problem: nameless and formless” (1978:228). what optics is still not able to do is This certainly seems curious, as Gib- to discover that which is common to son’s critics have not failed to no- all angles that are projections of hu- tice (cf. Goodman 1984:11 f; Janlert man beings. That there are such in- 1985:156ff): for while it is conceiv- variants is however, as far as I under- able, that a mathematical invariant stand, simply an article of faith.36 So cannot “be put into words”, it is dif- why could we not suppose instead ficult to understand how it can lack that these kinds of invariants, unlike mathematical expression. According those of depth and simple shapes to Kennedy (1974a: 44) solid angles perhaps, are qualitative rather than are too complex to be computed by quantitative, and have both a name the optics of the present time; so and a time of birth? mathematical expression is suppos- This would certainly bring us edly forthcoming in the future. Purdy closer to the Ganzheitspsycholo- and Sedgwick have demonstrated gie, though not so much the Berlin that pictures really do contain geo- school, which Gibson explicitly re- metrical information for depth (cf. jects, but rather the Leipzig school. Sedgwick 1980); and Perkins (1973) The examples given by Gibson has shown that, from some angles of (1978:230) could easily be under- vision, certain geometrical volumes stood in this qualitative way: straight can be derived from their orthogo- lines, curves, closure, intersection, nal projection on the picture plane. parallelism, coincidence, and other But that the information is present features are invariants which the child does not mean that it can be used by must discover and correlate with in- the perceptual system, as Hochberg variants of the environment, such as acutely remarks. More important for occluding edges, corners, pigment- the moment is the consideration of borders, etc. Features of this kind what kind of information it is that have been studied more thoroughly has to be picked up. by Kennedy and Hochberg. Although It may be remembered, that, ac- Gibson clearly identifies his invari- cording to Gibson (1978:228), what 36 The fact that animate action (and is directly seen are things like the the action of different animals, as well as cat on not “that peculiar, furry, mo- different movement pattern of the kind bile layout of surfaces”. Therefore, we call walking, running, etc.) can be perceived simply from the movement of it is not enough to show, that there a constellation of dots seen in the dark is geometrical information for depth, would seem to lend some plausibility and for simple geometrical shapes, to this claim – but even if these on the picture plane; we need to find constellations may be mathematically described, it does not follow that this is the formula of cat-ness. Kennedy how they are perceived. 60 Fig. 13. Features according to Kennedy (1974b: 231): 1) occluding bound with background air; 2) occluding bound with background surface; 3) occluding edge with background air; 4) occluding edge with background surface; 5) concave corner; 6) convex corner; 7) crack ants with features, Kennedy seems to tures can be defined. It could be add- feel that his features are different. ed, I think, that if there is similarity, Kennedy (1974a: 28ff) starts it exists between the content and the out radicalizing Goodman’s critique referent (between the picture object of similarity, in order to include more and the picture subject, in Husserl’s subtle relationships. Langer, who, in terms, as we will say below), not Kennedy’s view, is more precise than between either of these and the ex- Peirce and Wittgenstein about the pression (i. e. the picture thing). This distinction, says that similarity may problem is assumed to be resolved consist in “a proportion of parts”, by both Langer and Kennedy. “an arrangement of elements analo- The convention theory is also gous to the arrangement of salient rejected by Kennedy for obvious visual elements in the object” which, reasons. Gibson’s original point-pro- as Kennedy (p34) observes, is rejection theory will not do either: the ally two criteria, for a stretched rub- same configuration can be formed ber sheet changes its proportions but with white chalk on the blackboard, retains the adjacencies of its parts, and with blue ink on white paper, i.e. its topology. Two configurations and there will be no single iden- close to each other on the pictorial tical spot (p.42). Later, however, surface may, as Kennedy points out, Kennedy suggests that the projection represent a neighbouring house and theory can be made feasible, if it is a distant hill. Thus, it is certainly not reformulated for features instead of in terms of the distribution of chemi- spots. Indeed, if we are concerned cals on the pictorial surface that pic- with outline pictures, Gibson’s col-

61 our spots are clearly irrelevant for lysable into features — which is ex- there is no colour (p 106ff); so per- actly the opposite proposition to that haps, instead, Kennedy suggests, it is of Eco. So far, Kennedy’s suggestion the whole object which is projected certainly seems the more reason- from the world to the picture. Since able. it is possible to depict, and to recog- Like Kennedy, Hochberg nize the depiction of, unknown ob- (1972:69; 1978a: 190ff; 1980:51; jects, Kennedy argues, entire objects 1994; 1998) thinks picture percep- cannot be projected, so perhaps it is tion is essentially based on perspecti- parts of objects that combine to form val cues. Pigment on paper can stand new configurations, which are pro- for the edge of a surface, such as the jected. But different objects have dif- contour indicating the silhouette of a ferent parts, Kennedy continues his key; it may represent the corner where argument, so this would give us an three surfaces meet, as in the Necked infinite number of elements. In order cube; or else it can represent the line to arrive at “a small set of units – a where a round surface, for instance a small vocabulary”, Kennedy (p 108) doll’s body, passes out of sight, just suggests another analysis in terms of as the earth disappears from view at contours, shadowing, highlights, etc. the horizon. At least some of these Contours, for instance, have features features, like many of Kennedy’s, are like “concave corner, convex corner, already on the content side: the edge occluding edge, occluding bounds, and the “horizon”, for instance, look strands, cracks, etc.” (p 110 ff; cf. exactly alike, until they are seen as fig. 13). Features like these must be parts of the object they compose on higher-order variables, so it is not the pictorial surface. The case of the clear how they differ from Gibson’s different angles between the lines is invariants, at least from Gibson’s ex- different, for they can be interpreted amples quoted above. by a properly programmed computer In any case, Kenney’s fea- (cf. Winston’s programme discussed tures result in a new segmentation in Boden 1977). In addition to the of perceptual reality, a new “form” perspectival features, Hochberg given to the old “substance”, for (1972:73ff; 1980:76ff) also consid- the human body will no longer dis- ers distinctive or canonical feature, solve into head, trunk, legs, etc., but the nature of which can most easily into bounds, edges, and corners. Of be grasped from an example: given course, at some unconscious level, a facial oval, a toothbrush mous- perceptual reality is supposedly ana- tache is sufficient to indicate Chap- lysed in the same way. But this means lin, if the potential set is made up of that to Kennedy, and perhaps to Gib- movie stars: but in a wider context, son, pictures are motivated, and ana- a bowler hat, instead of the charac-

62 teristic fringe, is needed in order to perspectival cues, then, would be distinguish Chaplin from Hitler. The secondary to some more qualitative insistence on “potential sets” gives types of holistic properties; and this the argument a distinctly structural- would explain, not only that non- ist ring; but neither the fringe, the perspectival pictures are readily in- moustache, nor the hat are “purely terpretable, but also the possibility of negative terms”, and so must some- “impossible pictures”. how be recognized for what they are More recent proposal, which do in themselves. If we can explain how not distinguish between the percep- this is possible, we have also perhaps tion of pictures, and that of perceptual the general explanation for picture reality, such as those of David Marr perception. (discussed, for instance, by Bermú- Indeed, the modification of Gib- dez 2005) or Irving Biedermann (cf. son’s picture theory was provoked Peissig, Young, Wasserman, & Bied- by the Ryan & Schwarts experiment, ermann 2000) do not show any fear which showed that cartoon-type of an “infinite number of elements” drawings were more rapidly rec- which, according to Kennedy, must ognized than more literally correct result from different objects having pictures, including photographs (cf. different parts. Marr, for instance, E. Gibson 1969:100 ff; Hochberg discusses the single example of the 1978a: 193 ff; Perkins 1975; Perkins human body, conceived as a series of & Hagen 1980); and yet the rejection interlocked parts inscribed into three- of the point-projection theory could, dimensional shapes. Bidermann’s as Kennedy (1974a: 42) observes, be theory more explicitly take into ac- motivated from the common sense count the difficulty of recognizing observation that white chalk on a objects when seen from different blackboard can form the same picture points of view, and so may include as blue ink on white paper. But what the pictorial view as an instance; at if cartoon features, or “canonical no- least, the perceptual abilities of pi- tation”, are crucial to all pictures? geons are studied using pictures. What if there are some “invariants” “Geons” are such components of ob- for moustache-hood, bowler hat- jects which can be recognized from ness, and fringe-ness, which are not different perspectives and which in- perspectival in nature, i. e. which do clude the relation to other component not refer to spatial layout? Consider of the same objects. again Kennedy’s landscape (cf. fig. While no agreement seems pos- 15): how much would we make of it, sible at present of the nature of the if the house had not been a typical parts into which objects and pictures house, just as the hills, and the quite are analysed, it is clear that, if any- non-perspectival cloud and sea? The thing, pictures, just like perceptual

63 reality, must somehow by analysable of this mountain ridge. But not indef- into features, for some identification initely: only up to a point set by the to take place, and these features must principle of pertinence embodied in be, at least in part, motivated by sim- the pictorial medium. ilarity, if a mapping from what is for Consider Hünig’s (1974: 5ff) all practical purposes an “infinite” argument regarding the impossibility number of features is to take place of pictorial features. Fig. 15a looks from reality to picture and vice-ver- like a fish, but fig. 15b is easily seen sa. to be a bird; since the only difference between them is the elements found The process of in fig. 15c, one might suggest that resemanticisation they are indirectly signifying units, In spite of the existence of pictorial like phonemes. Against this proposi- features, Goodman’s observations on tion, Hünig adduces the fact that fig. density are not entirely off the mark. 15d, which does not contain the ele- Indeed, once we have determined ments in fig. 15c, can also be seen as whether a particular scribble realises a bird, as well as the further facts, that the category of a tree or of a wom- the very same elements are parts of a an’s profile, the drawing will tell us a star in fig. 15e, and form the arms of lot about the particular conformation a man standing on his hands, in fig. of the crown, or the nose, the hair- 15f, which, without these elements, cut, and so on (Cf. Fig. 14a-b). Simi- would instead be similar to a tree, as larly, once we realise that a “black in fig. 15g. Unfortunately, this argu- wiggly line”, as in Goodman’s ex- ment is invalid. To begin with, if both ample, does indeed represent Mount fig. 15b and fig. 15d mean “bird”, Fuji, rather than a population graph, they will simply be synonymous ex- its minute details and variations will pressions, and this fact has no bear- inform us about the particular shape ing whatsoever on the existence of

c Fig. 14. Gregory’s tree: a) seen a tree ; b) seen as smoking woman’s profile; c) quadrangular face

64 Fig. 15. Iconical figures and configurations (from Hünig 1974 :5f and Sonesson 1989a : 297).

features, for synonyms cannot be to be. expected to have the same set.37 In It has been noted by philoso- the second place, the other cases phers, from Husserl (1980) to Woll- cited actually confirm the thesis they heim (1980), that we seem to “see” were designed to oppose: for if the the content of the pictorial sign di- elements in fig. 15c are able to enter rectly “into” its expression. This is into so many configurations having true in a quite concrete sense. For quite different meanings, they really instance, although no real faces are appear to be units lacking meaning quadrangular, we have no trouble in themselves, exactly as phonemes identifying Figure 12c as a face; and, are. The real point, however, is that more to the point, we can even indi- once the elements in fig. 15c are cate the precise place of the expres- added to fig. 15a, the fish is not only sion plane where the ears are lack- changed into a bird, but the elements ing. This certainly has something added themselves take on the aspect to do with that peculiar property of of bird’s legs; and when the same el- iconic signs, observed by Peirce, and ements are introduced into fig. 15g, called exhibitive import by Greenlee, the tree is not only transformed into which makes it possible for icons to a man standing on his hands, but the convey more information than goes elements themselves now appear to into their construction (cf. Sonesson be the arms of the man. The differ- 1989a,III.3.6. and III.5.1.) ence is not where Hünig expected it The specificity of pictorial 37 In fact, these pictures should meaning can be illustrated by Ma- really be intensionally specified, at gritte’s familiar picture, “Le viol” least as ‘standing bird’ and ‘flying bird’, respectively; then verbal language appears (Fig. 16) which may be seen either to be more motivated in this case than as a face or as a woman’s trunk; it pictorial semiosis, for in the latter there is is precisely because of this double, nothing standing proxy for the invariant contradictory appresentation that it birdness. 65 property which is peculiar to pictorial meaning: the parts which are meaningless in isolation become carrier of particular portions of the overall meaning, once they are integrated into the whole. Like the phonemes /m/, /æ/, and /n/, forming the word /mæn/, the strokes and dots making up the picture of a man are in themselves meaningless even when considered in their particular spatial location; however after having been put together, the phonemes continue to be deprived of meaning as such, whereas the strokes and the dots begin to take on the aspects of different proper parts and attributes of Fig. 16. “le Viol”, by René Magritte the man they contribute to form. Put is instructive. Beginning with the simply, the different parts and prop- smallest elements, no particular erties of the man are not distributed meaning is suggested. But at least when putting the two half-circles containing two smaller circles side- by-side, we seem to be seeing two breasts. This interpretation is at its most determinate at the penultimate configurational level; but, at the highest one, when the hair is added, another interpretation, that of a face, gains the upper hand. Once we reach this level, some details which were present beforehand lend their support to it: the holes in the small circles, and their relative dimension, makes them look much more like the pupils of eyes than as nipples; indeed, the proportional location of the inner details are more nearly those of a face Fig.17. The double analysis of “Le than of a trunk. Viol” and its consequences for Now this points to the second resemanticisation

66 among the phonemes /m/, /æ/, and / tion, is really only the familiar device n/, as they are among the strokes and known as the hermeneutic circle, by dots forming the corresponding pic- means of which some elements are ture. This does not mean, of course, determined by the whole of which that every detail of the picture may they form a part.40 The comparison be given a particular reading in terms is interesting, although I don’t think of details of the object represented; it is entirely correct. To begin with, there is a lower, as well as a higher, if resemanticisation is identical with level where this projection of mean- the hermeneutic circle, it is at least ing from whole to parts cease to op- operating at another level: while the erate.38 This process, by which mean- hermeneutic circle, as we normally ings accrues to pictorial features, understand it, works on the foun- may be termed resemanticisation. It dations laid down by the ultimate will be noted, then, that pictures do constituents of language, that is, not have double articulation, as was phonemes and words, resemantici- once argued by Eco and Lindekens, sation starts from scratch. The case nor do they lack elements without is similar to what I have in another their own signification, as has been context claimed about the opera- widely argued since; their case is dif- tions of rhetoric: whereas the trans- ferent again.39 And this is precisely formational rules of verbal rhetoric the property that is not found in mir- work on and modify the result first rors, nor in identity signs: it is pecu- brought about by the combinatory liar to pictures. rules of grammar, pictures first arise It has been suggested by Vail- from rules of transformation applied lant & Castaing (in press) that what to perceptual reality. In this sense, I have here, as well as elsewhere, they are immediately rhetorical (cf. termed the process of resemanticisa- Sonesson 1997a, 2004a, in press c). But there is a second difference, 38 Vaillant (1997: 101ff), who adopts which may point to something even my notion of resemanticisation, rejects more peculiar to pictures: whereas the idea of upper and lower iconical thresholds (Sonesson 1989a: 317f), but the hermeneutic circle distributes he interprets them wrongly to imply a the meaning of the whole back to the level “en-deça duquel l’unité de plan de parts, it does so only on the level of l’expression ne veut plus rien dire”, etc., which would make resemanticization 40 As they note, there is also a identical to double articulation of sorts. parallel to Gestalt psychology, but All that is meant is that, beyond these that is recognised already in my levels, the projection of meaning back description, since I distinguish different from the whole to the part is no longer a configurational levels of the picture, one-to-one mapping. Cf. lecture 4. at which it can be interpreted as being 39 This argument is given more fully complete, for instance, in ‘Le Viol’, as a in Sonesson 1989a,III.4. face or as a trunk. 67 content. But resemanticisation oper- expression and content are present to ates more immediately on the level us in different ways within the pic- of expression.41 This is because re- ture. It is, however, an excellent idea semanticisation, as I have described on the part of Blanke to bring Hus- it, depends of what Wollheim calls serl’s work to bear on the µ-model “seeing-in”, and what Husserl has of iconicity. To show this, I have to described as being “perceptually im- return, however briefly, to my own agined”, which is a phenomenon that earlier discussion of Wollheim and, I believe to be unique to pictures. in particular, Husserl (in Sonesson 1989a,III.3.5-6.). The Husserlean triad: Wollheim’s contribution con- picture thing, picture sists in pointing out that, in Wittgen- object, and picture subject stein’s (originally Jastrow’s) famous To determine the nature of pictoral- duck/rabbit picture, the phenomenon ity (which he calls iconicity), Börries of “seeing-in” something in the sur- Blanke (in press) also has recourse face, whether it is a duck or a rabbit, to the theories of Wollheim and Hus- is distinct from the phenomenon of serl, using them to argue that picto- “seeing-as”, that is, of seeing some- rality is essentially a case of double thing as being either a duck or a rab- categorisation: something is both a bit. We see the rabbit or the duck al- tangle of lines on paper and a rabbit. ternately, but we see the surface and Both as an interpretation of Wollheim what it depicts at the same time (for and Husserl, and as an elucidation of which Wollheim refers to the work the nature of pictorality, this charac- of Pirenne, which can however be terisation seem to me rather mislead- substantiated by the later experi- ing. At least if we follow Husserl, ments of Hagen, Elliott and Perkins, we really would have to recognise quoted in Sonesson 1989a,III.3.5.). a triple categorisation (into picture The first is not restricted to - picto thing, picture object and picture sub- rial consciousness: even in the real ject). More importantly, however, I perceptual world, we may mistake a do not think we are justified in con- shrub for a ruffian, if perhaps not a ceiving categorisation to be the es- rabbit for a duck. sential operation involved in picto- Husserl’s much earlier work rial consciousness, because, just as delves much deeper into the issue. in any other sign, as we have seen, Two similar things assume the char- 41 Of course, one might prefer to acter of a picture only when pictorial take the hermeneutic circle in a more consciousness is attached to them, general, abstract sense, as all kinds of Husserl (1980:17, 16, 138f) contends determination of the parts from the whole, in which case resemanticisation is a (and, in addition, the similarity must special case of the hermeneutic circle. be “anschaulich”; p 135). Pictorial 68 consciousness puts three instances still valid, because even high-qual- into relation: the picture thing (origi- ity colour photographs, as well as nally the “physical picture”), the paintings, are unable to render the picture object, and the picture sub- full scale of colours present in the ject (“Bildding”, “Bildobjekt” and real world of perception. According “Bildsujet”, respectively). When the to Husserl (p 18), however, there is picture is said to be lopsided, this also a different kind of difference concerns the picture thing; but when between the picture object and the we complain about the failure of the picture subject, for while that Berlin photograph to resemble the person castle which we see is here, where photographed, it is the picture ob- the picture is, the Berlin castle itself, ject that is incriminated. However, as a thing, remains in Berlin. it might seem less clear what consti- Non-coincident traits stand out tutes the difference between the pic- on the background of the coincident, ture object and the picture subject. depicting ones, Husserl observes; In the photograph of a child, a and, while the extent to which pic- figure can be seen which is in some ture object and picture subject differ respects similar to the child, but dif- varies from one picture type to an- fers from it in size, colour, etc. The other, and for each particular case, miniature child in a greyish violet is there must be some difference, for of course not the child that is “intend- without a difference and an aware- ed”, i. e. conceived (“vorgestellt”). ness of it, there can be no pictorial The real child, the picture subject, is consciousness (p 30ff, 20, 82,138), red-cheeked, has blond hair, and so merely “Täschungen à la Panop- on, but the picture object can only tikon, Panorama, etc.”, i.e. “Jahr- show up “photographic colours”. The markteffekte” (p 41). Unlike the pic- first, then, which is what is “seen-in”, ture subject, both the picture thing in Wollheim’s sense, is the picture and the picture object are “appear- object.42 The second is the picture ances” (“Erscheinungen”), i.e. they subject. It should be noted immedi- are directly perceived (p 27 f, 489 f). ately that, although “photographic Our seeing of the picture object is of colours” do not mean the same thing the same kind as ordinary perception to us as to Husserl, the distinction is (p 133 f), and yet it is somehow “ab- normal” (p 490). For picture thing 42 Husserl uses the very phrase and picture object can by no means ’seeing in’ all through his extensive research manuscripts (e. g. pp 24 f, 28, 30 be identified: the latter is no part of f, 32, 34 f, 167, etc.). Unfortunately, each the physical picture, as the pigments time he becomes more precise, it appears and the lines are; and while the pic- that he is thinking about the relation ture thing is flat, the picture object is between the picture object and the picture subject. three-dimensional (p19f, p82f, 138f, 69 143).43 us so, we would seem to have an ap- Interestingly, it is for the rela- presentation (cf. Husserl 1939:174ff; tion between the picture object and Luckmann 1980), in other words, the picture subject that Husserl re- what we have called a semiotic func- quires similarity (p 138f and pas- tion above. It is possible to thematize sim), i.e. for two instances which are the picture thing, as when we note roughly equivalent to Peirce’s “im- that the picture is lopsided, but “trotz mediate” and “dynamical objects”. meinender Zuwendung zum Bild- Except once, when he says that a re- ding bleibt die erregte Erscheinung lief is comparatively more similar to des Repräsentierenden Bildes mitbe- its picture object (p 487 ff.), Husserl merkt” (p 137; cf. p 488). Likewise, never discusses the similarity of the it appears to be possible to thematize picture object and the picture thing. the picture object, at least in order to Nor does he consider the similarity note the “extensity” and the “inten- of the picture thing and the picture sity” of its pictorality. subject, which is the closest we come Husserl’s term for the equiva- to Peirce’s relation between the “rep- lent of Wollheim’s seeing-as is com- resentamen” and the “object”. It is in petition (“Wettstreit”), as when we the relation between the picture object are not sure whether we see a hu- and the picture subject that pictorali- man being or a wax-doll (p 277, 401, ty may be more or less extensive, and etc.). This is opposed to the condi- more or less intensive, i.e. concern a tion of contradiction (“Widerstreit”), greater or lesser number of proper- in which case one of the percepts is ties, and realise them to a greater or perceived without “belief”, in other lesser degree (“Extensität” and “In- words, is “cancelled out”, as when tensität der Bildlichkeit”; p56f).44 we see the doll and the human being The picture thing and the picture it stands for. According to Husserl, object are directly perceived; but the however, pictures are different again: picture subject, which is what is in- they do not simply constitute “illu- tended (“gemeint”; pp 23 ff., 30, passions”. The picture object is no doubt sim), is only indirectly given; there- a fictum, but not an illusionary one fore, although Husserl does not tell for, unlike the latter, it is discordant already in itself, so that no attention 43 For reasons discussed in Sonesson 1989a,III.3.6., I would not give such to the environment is needed in order importance to the impression of three- to cancel it out (p 490). Also, while dimensionality as Husserl does, but I will the illusionary fictum appears in the not enter this discussion here. ordinary world, the picture presents 44 That third part of the iconicity itself in a reserved space (p 480). question elected for consideration by Indeed, the paper surrounding the Husserl is thus not the same as that chosen by Peirce (Cf. Sonesson 1989a, III.1.1.). drawing can be observed, as can the 70 frame, the wall where it is placed, the From the picture thing via the picture room, etc., but there is no ordinary object to the picture subject (and, as perceptual apprehension (“Wahrneh- I will later argue, to the extrapicto- mungsauffassung”) for that part of rial referent), there is (if the aesthetic the paper where the drawing is. Of function does not apply) increasing course, the paper apprehension is all thematization, and, roughly speak- the time “co-conscious”, for it is part ing, there is also decreasing direct- of the continuous field of vision, but ness. There is decreasing directness it has lost its contents to the picture only if the most large-scale consist- object apprehension. There is a con- ent reading is required, for otherwise tradiction, but the picture object pre- the picture object is the most directly vails (p 45f). And yet, the picture ob- given, and directness decreases in ject is unreal, for it contradicts that both directions from there. which is present in the here and now: Thus we see that the picture in the consistent wall perception, a corresponds to at least three differ- piece is “covered” by the picture (p ent categories: the picture thing, the 482).45 picture object, and the picture sub- In conclusion, it seems that ject. Indeed, I will go on to suggest while the picture object has a kind of that the latter must be distinguished existential priority, because it is most from the picture referent. Moreover, immediately seen, the picture thing it becomes clear that these catego- has another claim to the same pri- risations do not occur at the same ority, because it is compatible with level of perception: while the picture the most extended range of the envi- thing is an ordinary percept, the pic- ronment. This should not impede us ture object is merely “perceptually from seeing the picture as a sign, in imagined”, and the picture subject is the sense of the semiotic function, al- not perceived at all. They are not re- though of a rather peculiar variety.46 lated to each other as categories, but as levels of thematisation and direct- 45 Many more complications envisaged by Husserl are discussed in ness within the sign. Sonesson 1989a,III.3.5-6. Pictorial consciousness 46 Husserl himself claims there is a sign relation even between the picture and the scale of typicality thing and the picture object, but he does Although typification is an impor- so for erroneous reasons: he thinks the tant ingredient of the Lifeworld, as sign function (here clearly identified with conventionality) to be required for us to so often described by Husserl, he know which side of the picture is up and never seem to take it into account is which is down. However, as a simple his discussion of pictorial conscious- experiment using the comic strip “The ness. Actually, he does, but only im- Upsidedowns” shows, this information is part of the very picture object. Cf. Sonesson 1989a,III.3.6. and Lecture 2. 71 plicitly so. The difference between But this makes nonsense of the idea, types and occurrences is in fact con- suggested by Husserl himself, to tained in the distinction between pic- compare the picture object and picture thing, picture object, and picture ture subject, as to “extensivity” and subject, to which must be added the the “intensivity” of their respective picture referent. It will be remem- properties. bered that, to Klinkenberg (1996; in It seems to me that the picture press), in the second version of the subject is made to accomplish a dou- µ-model, referent stands to type as ble task, which it cannot really sus- stimulus stands to signifier, but the tain, that of content type and refer- fact that the former item of each cou- ent. It would of course be an error to ple is a typification is blurred by the identify the triad picture thing, pic- terminology. In Husserl’s case, typi- ture object, and picture subject, with cality is hidden by being found be- expression, content and referent. The tween the different instances. picture object is perceived, which As we noted above, Husserl the content of, for example a verbal maintains that there must always be sign, is not; and there is a real sense a difference, however small, between in which the picture object is present picture object and picture subject, in here and now, together with the pic- terms of the “extensivity” and the ture thing, which the verbal content “intensivity” of their respective prop- can hardly be said to be. Nor is it erties. If so, it should be sufficient to feasible to assume that the picture attenuate the “intensive” and “exten- subject is identical to the referent, in sive” differences between them, in the sense of a concrete object of the order to have them approach gradu- world, or even in the sense of being ally, and then in the end coincide, at a type standing for a number of such least as a thought experiment. But instances. Many pictures may not this could never happen, not even in have referents, in any of the latter thought, because the picture object senses, but they clearly have picture is here, where the picture thing is, subjects: such is the case not only of but the picture subject is somewhere the notorious unicorn, but of all the else, in the place assigned to it in the creatures emerging out of Escher’s Lifeworld (cf. pp 18, 79): indeed, as and Reutersvärd’s pictures.47 we have heard, the Berlin castle, no The whole point of the “impos- matter where the picture is moved, sible pictures” is that they point be- will remain in Berlin. Moreover, the yond themselves to something which picture object is perceived, but the picture subject is only something 47 More will be said about Escher about which information is con- and Reutersvärd and about “impossible pictures” in the final section of this veyed (see Sonesson 1989a,III.3.6.). lecture. 72 cannot exist, their equivalents in the few; cf. Husserl 1980:38: the picture three-dimensional world (cf. Son- object renders the picture subject in esson 1989a,III.3.4.): indeed, their one of “seiner Erscheinungen aus picture things are quite possible, as der Synthesis”). Furthermore, the are in this sense their picture objects picture object differs in its attributes (consider the importance of “recog- both “intensively” and “extensively”, nizability” to Escher). This is, I sub- as Husserl puts it, from the picture mit, the most interesting interpreta- subject; or at least this is the usual tion of the notion of picture subject: case. It is also possible for a picture as the potential real-world equivalent object to show just a proper part of of that which is “seen in” the pic- a picture subject and, at least in one ture thing, that is, of the picture ob- sense, only complete objects can ex- ject. Husserl (p 490) could be taken ist in the Lifeworld. Thus, we could to suggest just this, when he claims take the picture subject to be the pic- that what is seen in the picture is ture object as it must be completed in corrected for its deviations from the order to be something, which could idea we have of the corresponding possibly be encountered in the Life- type, which imposes constraints on world. But this is only possible be- the possibilities of perception: being cause, at least as types, at a general made of plaster contradicts our idea level, we know the Lifeworld objects of a human being, so we withdraw it beforehand. We can easily complete from the picture object. our Lifeworld prototypes of a wom- Thus, if the picture subject is the an, a sitting person, an almost naked projection onto the common, three- person, and so on, to arrive at the pic- dimensional Lifeworld of the picture ture subject which is Titian’s Sacred object, the reverse is also, and pri- Love (though not at the iconographic marily, true: at least on some inten- level); but the picture subjects corre- sional level, Titian’s Sacred Love (to sponding to the “impossible objects” pick one of Husserl’s favourite ex- can only be anticipated as some very amples) and the Escher-Reutersvärd general kind of Lifeworld objects, as kind of objects also reflect potential “box-like”, or something of the kind things of the common Lifeworld, in (and outside the purview of real box relation to which they are felt to be possibilities). incomplete. For instance, at the level The reminder of this discussion of perceptual parts, the picture ob- (from Sonesson 1989a,III.3.5-6) ject is normally based on just a sin- should help us realise what else has gle noema (or a few noemata in the to be put into the model of iconicity case of Cubism), and no real world (which is actually a model of picto- object, and so no picture subject, rality) envisaged by Groupe µ. Many can have just one noema (or only a years ago, I suggested there are re-

73 ally three iconicity questions (cf. tations. So, apart from what he had Sonesson 1989a,III.1.1.): the rela- been able to “see into” the surface of tion between expression and content, the pictures and the relief (and what the relation between content and ref- is already there in the statue), that is, erent, and the relation between ex- apart from the picture object, which pression (or perhaps rather the sign emerges directly, but already trans- as a whole) and the referent. In the formed, from the different picture case of the picture, as we have seen, things encountered, he must have the issue is further compounded by had recourse to a number of typifi- the existence of a fourth instance: cations, such as concrete object, liv- the picture object. It is the possible ing being, horse-like creatures, and iconicity of the relation between the so on, which helps in constituting the picture object and the picture subject picture subject.48 As we know, this that is discussed by Husserl. But all did not impede him from finding, at similarity breeds a certain dissimilar- the end of his search, the referent we ity. Husserl, as we saw, talks about now know must have been a rhinoc- the extensivity and the intensivity eros. Others, of course, have come of the common properties. Thus, if up with the empty set. there are three (or four) relations of It follows that the extrapictorial iconicity, there is the same number of places in which a transformation 48 As I have often pointed out above may take place. (and in Sonesson 1992a, 1994a), one of Consider the story of Marco the many ways in which predominantly iconical signs like pictures and gestures Polo’s discovery of the unicorn (re- are necessarily conventional derives from told by Eco 1999: 57ff). Or rather, let the fact that they are themselves objects us consider the preconditions of that of different categories from most of the story (not mentioned by Eco). How object which they represent. For instance, since the expressive resources of manual did Marco Polo get his idea about the signs are essentially the hands and their looks of a unicorn? He might have movement, the (limited) vertical extension read about it, but most probably he of the female body can only be described had also seen pictures, which would as far as one of its properties is concerned, its highest point; and the curvaceousness allow him to have a more complete of the female body cannot be rendered idea of its appearance. He might have in its totality, but only transposed in seen paintings, tapestries, book illu- time, as an undulating movement. Rock minations, but perhaps also reliefs carvings, as all other pictures executed on a surface, lack the third dimension or even statues. He had certainly not of the real world: actually, because of been made to think that unicorns are being carved in the rock, the petroglyphs, beings of paint, woven tissue, ink, or unlike most pictures, have a prominent stone, nor that they are flat as they third dimension, but this cannot be used for rendering the third dimension of the appear in the first series of represen- perceptual world. 74 referent, if there is one, is something for pictorial rhetoric. The first sys- quite distinct from the picture sub- tematic approach to pictorial rheto- ject (unlike the case of mirrors; cf. ric was realised by Groupe µ (1979; Sonesson 1989a,III.3.5 and 2003b). 1992). For reasons that would be too Thus, it seems, we must think of the long to review here (but cf. Sonesson picture thing as being on the expres- 1996a, b; 2001a; 2004a; in press c), I sion side of the sign, while the pic- find their model rather unsatisfactory ture object occupies some kind of in the end. But I think another model intermediate position; the picture may be established which relies on subject is similar to the content; and several dimensions, some of which the referent, which strictly speaking are iconicity and indexicality. falls outside the sign, is still the refer- In the first section of this essay, ent. The picture object must be seen we liberated the iconic and indexical as a reflection within the sign, of the grounds from their absorption into picture subject: and thus, the similar- their respective sign functions. This ity of the picture object to the pic- is what we need here, because the ture subject, in absolute terms, mat- elements which are connected iconi- ters less than the place of the picture cally or indexically are not parts of subject in the hierarchy which allots signs, but elements depicted, on one their relative centrality to the objects hand, and elements which are ex- of the Lifeworld. This is what de- pected to be depicted, on the other. termines the possibility for pictures There are two indexical dimen- to say new things about a very old sions, one involving contiguity, and world. the other factorality (the relation of At the beginning of this lec- parts to whole): both operate by an ture, I suggested that the recognition increasing degree of divergence from of the iconical character of pictorial expected integration. For instance, signs was only the beginning of a on the level of factorality, we expect real semiotics of pictures. In fact, if to see pupils in Haddock’s eyes, but the primary property of the picture instead there are bottles; and, on the sign is to suggest a similarity to real- level of contiguity, we anticipate that world perceptual appearances, then ice cubes and an aperitif bottle should the very point of their sign function be placed in an ice-pail, but instead is to subtly modify our view of these they are located inside Coliseum. On real-world appearances. Similarity the iconical dimension, there may be make us expect more similarity, and more similarity than expected (e.g. connection make us anticipate inte- the roof and the street which have gration. But, in pictures, these expec- identical forms in Magritte’s “Les tations are generally deceived. This promenades d’Euclide”), or more is what constitutes the foundations commonly, more dissimilarity than

75 expected, going to the extreme of much more thorough investigation, contrariety (the male adult mouth called “pictorial consciousness”. on van Lamsweerde’s little girl) and Even Husserl’s account is seriously even contradiction (Magritte’s pipe incomplete, and the notions of pic- contradicted by its title). The third di- ture thing, picture object, and picture mension may perhaps also be called subject do not only have to be made the dimension of symbolicity, for it more precise, but the latter must also concerns signs containing other signs be distinguished from the picture (e.g. pictures of pictures). Thus, we referent. When confronted with the see that the ground, in the sense in µ-model of iconicity, Husserl’s anal- which we have taken the term, is not ysis shows us however that there are only useful for constituting signs: it several different relations of iconic- also may explain operations within ity, and, for the same reasons, sev- signs.49 eral distinct loci of transformations. Finally, we have seen that iconical Summary and indexical grounds, and in a way There is no tradition in semiotics for also the sign function, may be used determining the specificity of -pic within the picture sign in other ways tures, within the domain of iconical than just to constitute the sign. signs. The category which Peirce Pictorial semiotics, I have ar- calls “images” may have been in- gued, must determine the specificity vented precisely to do this job, but of pictures, within the general do- we have seen that it cannot serve main of iconical signs, not just offer such a purpose: if understood in models for the analysis of particular terms of grounds, pictures are cer- pictures. Although my own point of tainly not made up of “simple prop- departure is Husserlean phenomenol- erties”. Against structuralist semiot- ogy, reviewed by newer approach- ics, which tried to import the notion es within perceptual and cognitive of double articulation from linguis- psychology, I have found it useful tics to, among other things, pictorial to start out from the work of Peirce semiotics, I have shown that pictures and his critics, in order to develop function according to a very differ- my own, more complex, conception ent, quite peculiar process, which I of iconicity. The latter model in- have called resemanticisation. This volves a distinction between iconic- is a property whose specificity to pic- ity per se, the iconical ground, and tures must itself be explained relying the iconical signs, and it supposes a on what Wollheim calls “seeing-in”, much more explicit concept of sign and which Husserl, in an earlier, than those found in the Peircean and Saussurean traditions alike. I have 49 Cf. Sonesson 1996b, 1997a, 2001a, 2004a; in press d. also suggested that there are at least 76 two very different kinds of iconical resources (apart from those, like cin- signs, primary and secondary ones, ema, television, and video, which which inverse the relations of condi- are based on picturelike media), is tioning between the ground and the capable of showing (and, indeed, in sign function, and that the primary a sense, is constraint to show) true kind can only exist because of more perceptual perspective. One of the general assumptions embodied in things which most clearly separates the Lifeworld or ecological physics. the the picture object and the pic- In determining the specificity of the ture subject is that the former offers picture sign, we were led to reject its a particular perspective on the latter. identification with the Peircean con- Unlike what happens in the noemat- cept of “image”, as well as the inde- ic matrix, however, the (perceptual) pendence (in one sense) of iconicity part is not an inextricable moment of from the sign function. Unlike both the perceived whole. The part stands the first and second articulations in out as a separate object. It becomes verbal language, I argued, pictures an object on its own right, given gain their meaning from a process I in its proper noemata. However, it called resemanticisation: the projec- would be misleading to equate per- tion back from the whole to the parts spective, as here understood, with of a globally constituted meaning, at linear perspective, as discussed in art the level of expression as well as that history. Indeed, the fact that perspec- of content. But resemanticisation it- tive is discussed in relation to many self is only an effect of something other kinds of semiotic resources, more fundamental: the specific func- including language, suggest that is tioning of the distinct instances mak- has a much broader mening, also in ing up the picture sign: the picture pictures. The problem, nevertheless, thing, the picture object, the picture is that the latter may be too broad to subject, and the picture referent. Be- make any sense. tween them, there are different rela- Perspective, as applied to per- tions of iconicity — and thus differ- ception, compounds several different ent possibilities of transformation. things: most generally, that which is seen in perspective is given in one of 3.5 Perspectives, its variegated appearances (which aspects, and means that the object is question is “impossible pictures” this thing of which we have experience but many other things in addi- A correlative of the peculiarities of tion); the object is given from one of pictorality which we have studied the many position which make up the in the last section is that pictures, space sourrounding it (what we will alone among all species of semiotic later call positional perspective); and

77 Fig. 18. Different kinds of ambiguous figures: a) Wittgenstein’s duck/rabbit: b) the Necker cube; c) Black cross on white background, or the reverse; d) “the devil’s turning fork”; d) cat which is also a coffee pot. it is given to a particular person who, when applied to other types of me- in the most simple case, happens to diation than perception and pictures, occupy the space in question (later perspective may often involve, not called personal perspective). The position as the place from which the ascription of perspective to posi- object is experienced, but as defined tions and persons may be found also in relation to some kind of landmark in semiotic resources not so close to (for instance “behind the mountain”, perception as pictures. However, in where the mountain is the landmark, pictures the source of that which is the object is on one side of the moun- given is always (visual) perception, tain, and the person having the expe- but when other semiotic resources are rience on the other side). Perspective involved, it may very well by knowl- in this sense seems to be impossible edge (in pictures, as in perception, in pictures. In the second place, other we really know much more than we semiotic resources such a language see, notably, we know about the sides may mix different perspectives (in which are turned in other direction. the sense of position and person, and The classical distinction between the involving perception or knowledge), “conceptual” pictures of children and but this is hardly possible in pictures, “primitives” and the perceptual ones apart from some rather marginal cas- invented by the Renaissance is large- es. ly misleading). On the other hand,

78 Wittgenstein’s rabbit as a strued by different scholarly speci- Saussurean “form” alities, is compounded by the host of near-synonymies which immediately When literary historians, linguists, offers itself in any languages: instead film theorists, art historians, and so of perspective, we could almost as on, discuss perspective, they are con- well talk about “view”, “point of cerned with different things, yet with view”, “viewpoint”, “standpoint”, phenomena which may not entirely “outlook”, “approach”, “angle (of lack overlapping features. Naturally, vision)”, “attitude”, “aspect”, etc. we should expect linguists and liter- Indeed, a partial overlap is even to ary historian to be referring, at least be found with some more “techni- some of the time, to the same thing, cal terms” such as “seeing-as”, “see- since both are involved with language ing-in” (as the terms are just by Her- — but, as we shall see, not even this merén and Wollheim), “intention”, is obvious. There have been attempts “propositional attitude” and so on (as to compare perspective in film and commonly used in philosophy and literature (cf. Currie 1995), but they some parts of linguistics), “form” do not seem to take must heed of the as opposed to “substance” (as used terminological differences. The most is semiotics following Saussure and, famous case of a systematic attempt in particular, Hjelmslev), and indeed to compare (or rather equate) literary “aspect” (as used by Wittgenstein). and art historical concepts of per- We shall have occasion to return to spectives is found in Boris Uspensk- some of these terms in the course of ijs classic book A poetics of compo- this section. sition (1973 in English).50 It seems to me that, from Uspenskij onwards, if It could be argued that semi- not before, there has been too much a otics is the science of perspective tendency to find similarities between — or at least of point of view. We the arts – and the semiotical systems have Saussure’s (1968: 26) word for which lie at their foundation – rather it : he talks about the difficulty of than attending to their differences. “linguistics and all the other semi- ological sciences” stemming form The difficulties in discussing the fact that they are not involved perspective, as it appears in different with something material, but only semiotical systems, and as it is con- with “the point of view taken on 50 It was my dissatisfaction with something material”. This is paral- this book that first prompted the present observations (although on the way it lel to a more famous saying, which happened to crossbreed with some other appears in the Cours de linguistique themes of mine), but, in the end, I have générale, put together by Saussure’s found it more convenient to state my pupils, according to which “it is the conception in a positive way, rather than criticising Uspenskij. point of view which creates the ob- 79 ject”. As Luis Prieto (1975a: 114; said about the Lifeworld, subjec- 1975b: 225f) has convincingly dem- tive-relative. The subject in ques- onstrated, this is not, as so often have tion does not necessarily have to be been thought, a positivistic profes- a solitary individual but could just as sion de foi : rather, it is another way well be a group subject or even an of expressing the famous figure in anthropologically (or perhaps zoo- Saussure’s Cours of the double cut, logically) universal subject. What isolating some portion out of the two Saussure seems to mean, in the end, amorphous masses of sounds and is that something which appears to meaning. But it adds something : the be identical (something “material”) relation to a subject. is presented as being different from Today most linguists and psy- the point of view of different sub- chologists would probably not want jects, where the canonical case is to talk about any amorphous mass, taken to be the speaker of a particu- neither on the side of the signifier lar language — or, as suggested in nor on that of the signified : at least the unpublished remarks, the user of in the case of verbal language, the any other semiotic system. most fundamental categories through The most obvious example of which we perceive the world appear this thesis is of course the differ- to by part of the universal make-up ent ways in which sounds are used of the human species, in the form of by different languages, some posit- “prototype categories” preferentially ing a difference where other do not. grasped at the “basic level” (cf. Ro- The case is more easily illustrated by sch 1975; et alia 1976). And even if the different ways in which writing these categories are found in percep- systems interprets the visual figure tion and cognition before they ap- which looks as follows: “H”. In the pear in language, it is probable that Latin script, it is of course the letter they will appear also in those semi- “H” (though in some language, it is otic systems which make use of the really not pronounced at all, and in visual modality, such as, in the cases some cases it merely represents aspi- considered here, the cinema and stat- ration), but in the Russian alphabet, ic pictures. it corresponds to the sound which we This very general concept of render by the letter “n”, whereas in perspective (or, literally, point of the Greek alphabet, it now represent view) in no way pertains to any the sound /i/ (as in the world “be”), particular (spatial) position, incar- while the Ancient Greeks used it to nation, opinion, or mode of expres- designate the sound /e/ (as in the sion ; it amounts to the suggestions world “egg”). In Saussure terminol- that we are involved with something ogy, it may be said that the shape “H” (“a meaning”?) which is, as Husserl remains the same substance, but is 80 given varying forms by these differ- as”). The first phenomenon could ent writing systems. In other terms, be instantiated also by a completely different principles of relevance are unambiguous picture. Wollheim also applied to the same object. rightly points out that “seeing-as” is As a first approximation, it may not restricted to pictures — it also seem that this is similar to Wittgen- applies to humidity spots, clouds, stein’s (1971: 227ff) famous picture bushes, etc. Indeed, I have argued (taken over from the psychologist that “seeing-in” is a variety of the Joseph Jastrow, otherwise known for semiotic function, the sign function, having written, together with Perice, as it appears in pictures (cf. Sones- the first American article of experi- son 1989a: 263ff and above). mental psychology 1884) which may Well before Wittgenstein, Her- be seen as either a duck or a rabbit: merén, and Wollheim, in some man- physically, it might be suggested, uscripts written between 1898 and what is present on the page of Witt- 1925, Husserl (1980) discussed a genstein’s book is the same ink blot, phenomenon apparently identical but as the perceiving subjects adopts to “seeing-in” describing it as be- different stands on the blot, a duck ing “perzeptiv imaginiert”. In our or a rabbit is seen (Fig. 18a. roughly time, the psychologist James Gib- corresponds to Wittgenstein’s ver- son (1980) used the term “indirect sion). Again, the ink blot could be perception”. Husserl went on to dis- considered to be a single substance tinguish the picture thing (the mate- which is given different forms. Witt- rial object which may be suspended genstein, of course, talked about dif- obliquely), the picture object (that ferent “aspects”. which you “see into” a photograph, Wittgenstein’s description gave in spite of the “photographic col- rise to a discussion among some aes- ours” present in the material thing, thetically interested philosophers. etc.) and the picture subject (the real Thus, both Hermerén and Wollheim thing we think we see but which is at one time thought what Wittgenstein actually somewhere else, with cor- calls aspect could more properly be rect colours, etc.). When Husserl here described as a kind of “seeing-as”. talked about “photographic colours”, As we have seen, Wollheim (1980) he was no doubt thinking about black later argued against this position, and white photography. But it is still pointing out that there were really true of colour photography – and of two different phenomena involved any other kind of picture — that it here : seeing the blot as something conveys a reduced colour scale as other than a blot (“seeing-in”) and compared to perceptual reality (cf. seeing it as either a rabbit or a duck Hochberg 1979: 25: cf. Sonesson or even some third thing (“seeing- 1989a: 270ff).

81 Both “seeing-in” and “see- sake, but merely as a “stand-in” for ing-as” could be considered to be something else: our attention (or “in- some kind of “perspectives”, in the tention”, as Husserl would have said) most general sense : they are differ- “goes through” the expression but ent ways of talking about something does not stop there: it is focused on which, from another point of view, the content. This is also true of Witt- is identical. It should be noted that, genstein’s “aspect” to the extent that in cases like these, “seeing-as“ is it can be identified with “seeing-in”, dependant, for its possibility, on the that is, when we take it to involve the “seeing-in“ : pictures may contain ink blot as opposed to the rabbit and information which is minimal to the the duck. point of making it possible to reduce Neither Saussure’s “point of the real-world difference between a view” (that is, the “form”, which rabbit and a duck, etc. It would cer- is instantiated in pictures as “see- tainly be very difficult to confuse a ing-as”) nor Wittgenstein’s “as- real rabbit and a real duck. Nor are pect” (here reduced to “seeing-as” combinations of a cat and a coffee as in the choice between the rabbit pot feasible in the actual perceptual and the duck) are really central in- world (cf. Fig. 18e.) Thus, the simi- stances of what we ordinarily mean larity is at the level of expression — by the term “perspective”, at least in the picture object, not the picture when applied to the world of vision. subject. Also in the case of clouds, It might be useful to consider some bushes, or damp spots the sign func- of the other examples mentioned tion is a prerequisite for “seeing-as” by Wittgenstein: the Necker cube, — at least if we take these equiva- where either end may be perceived lencies to be simple musings, as as being closer to the observer (Fig. they clearly were to Leonardo da 18b), actually seems to be more akin Vinci, and not errors of judgement, to the central idea of perspective. which may occur in the dark, or dur- Then there is the shift between fig- ing some special mental states. The ure and ground, which Wittgenstein box which the child pretends to be exemplifies with the black cross on a a house, which is a further example white background, which could also of Wittgenstein’s, is a case of “sym- be seen as the reverse (Fig. 18c), bolic play”, as thus of the semiotic but which may be more familiar as function. the two profiles facing each other Clearly, it is only in a very which can also be seen as a vase. loose way that the semiotic function These instances may be compared to may be identified with perspective. It so-called “impossible figures” (for would then simply mean that some example, “the devil’s turning fork”; object is not perceived for its own Fig. 18d), which to ordinary vision

82 Fig. 19. The classical revolving stair of Escher seems to form wholes, but which do jects, but only as lines on a surface. not go together if we attend to the Thus, the alternatives as well as the details. As is well-known, Escher impossibilities are only given thanks (Fig. 19) and Reutersvärd (Fig. 20) to the sign function. have used such figures in their paint- As far as I understand, Wittgen- ings; but also some of Klee’s works, stein did not mean to argue that one in which for instance a mother and a “aspect” necessarily precludes the child share a contour line (Fig. 21), perception of the other; but this is are “impossible” in a similar (though exactly the use to which Gombrich less geometrical) way. In fact, none (1960) puts Wittgenstein’s example. of these figures are impossible as It may be true that we cannot see the figures, that is, as constellations of rabbit and the duck at the same time, lines. They simply do not correspond nor, perhaps, the two orientations of to any conceivable (and perceivable) the Necker cube; but in other cases, real-world object. Nor is the Necker as in the cat which is a coffee pot cube or the cross ambiguous as ob- 83 (Fig. 18e), as well as in Arcimboldo’s the other hand, we do only need to paintings, we have no trouble see- suppose that there are objects in the ing the contradictory aspects. Here world, and that there are different again, as I noted earlier, the doubling ways of having access to them. We of the aspect is only possible once do not have to say anything about the sign function is given. which these objects are, nor describe It seems obvious that both Saus- the modes of access them. For more sure’s notion of “form” and Witt- concreteness, however, I prefer to genstein’s conception of “aspect” rely on folk ontology, also known involves something which is con- as the “science of the Lifeworld”, ceived as being different from what “ecological physics”, the “natural it “really” is (that is, immediately, in world”, or “naïve physics” (cf. Son- “direct perception”), in such as way esson 1992a, b; 1996a, 1997, 2001a, that there are different (though some- b, c).51 times concurrent) alternatives for what this other thing is. In contrast, a Possible depictions of perspective does in no way represent impossible worlds any “indirect perception” (unless The interesting thing about so-called combined with “form”): it offers one “impossible pictures” – which are re- of several possible modes of access ally very possible pictures, although to a single identical object. In fact, they seem to represent objects which as I observed above, the Necker cube could never exist in the three–di- may actually be construed as repre- mensional world – is that their very senting alternative perspectives, but absurdity highlights the way pictures only if we suppose that it is the same relate, by means of dissimilarity as cube that we are seeing from differ- well as similarity, to the perceptual ent angles. This possibility does not world. I think some considerations exist for the cross, let alone the duck- on these “impossible pictures” will rabbit. show us why perspective, in the In order to talk about things strict sense (which is more than lin- being the same or different, and ear perspective) is only possible in something being merely a different pictures. mode of access to an identical ob- Unlike pictures as such, impos- ject, it seems we must make assump- sible pictures have long be discussed tions about how the world is made by perceptual psychology, being up, that is, in the famous Quienean treated as one among several kinds phrase, about what there is. Acutally, of “visual illusions” used to argue for us to establish a difference be- for the constructivist view of percep- tween form/aspect on the one hand, 51 Cf. Lecture 2. More will be said and perspective in the strict sense on about this in Lecture 4. 84 tion. As Gibson has often remarked, pictures are analysed in order to draw conclusion about the three-dimensional world of perception (and this is still true today, in the work of Hoff- man 1998). The task of the perceptual scheme, according to Hochberg, is essentially to fill in the details which peripheral seeing fails to capture. For while the information picked up from the environment, when it falls on the fovea, is indeed abundant, the Fig. 20. A Reutersvärd picture on gaps between the fixations have to be a Swedish stamp spanned with constructions built up should at once realize that such pic- on fragmentary cues. Painters have tures cannot be interpreted three-di- used this fact, Hochberg (1979:25ff; mensionally (cf. Gregory 1966); but 1994) suggests: Rembrandt devel- in fact their three-dimensional inter- oped one style for the fovea, and an- pretation is only rejected, Hochberg other for the periphery; and impres- claims, when the contradiction is ap- sionism devised a way of painting, parent in one momentary glance, for the result of which simulates a scene otherwise the gaps will be coherently when viewed peripherally, while reconstructed using the most plausi- merely indicating “painterliness” to ble assumptions. the fovea (p 36). We cannot pursue First, we turn to the famous these complex issues here, but will Penrose triangle, which was the ba- remain content to consider the case sis of many of Escher’s constructions of outline pictures. In Hochberg’s view, depicted objects, as well as real ones, differ most from each other at their edges: thus, when perceiving an object or a picture, we are likely to fix some of the edges, or the lines standing proxy for the latter, and then fill in the rest from assumptions. This is what Hochberg (1972:59, 1978a:154; 1980:63) tries to establish in his discussion of “impossible pictures. His point is that if we really had access to complete information from the environment, as Gibson suggests, we Fig. 21 “Mother and child” by Klee

85 X

a b

c d e f g

h i j

k l m Fig. 22. “Impossible pictures” and some all too possible ones (a-g inspired in Hochberg 1978a: 154; h-k from Perkins 1981: 152ff; l-m from Sonesson 1989a).

(cf. Escher 1967; Ernst 1978); it was patterns. But although the right and actually invented by Reutersvärd in left sides of Fig. 22b are inconsist- 1934, 24 years before the Penroses’ ent, because the line marked x must article (cf. Bresti 1985). Neither this change its function as the corner of figure, nor the two to follow should a dihedral angle somewhere between look solid according to the Minimum the two sides, this contradiction is principle, as defined by Hochberg, not obvious, and does not interfere because they are neither simpler nor with the figure’s apparent tridimen- more consistent when considered as sionality. The reason for this, Hoch- whole 3D objects than they are as flat berg (1980:63) claims, is that the

86 two sides cannot be encompassed by contorted object; there is even a sug- foveal vision in one single glance. gestion of a spinning movement. The But this is possible in Fig. 22c, and Penrose triangle can also be read in consequently, the figure looks flat this way, but hardly Hochberg’s rec- and inconsistent, Hochberg tells us. tangle (fig. 22b): however, at least That 22c is not flatter merely - be on a second viewing, it may give the cause of its proportions is shown by impression of being slightly bent, 22d, which is just as compact as the rather than inconsistent. Nor can I former, but appears tridimensional. In agree with Hochberg about the most other cases, too, it is what is grasped natural interpretation of the second in the momentary glance that deter- set of pictures. These are similar to mines the interpretation. Thus, when the wellknown Necker cube, which the gaze is directed at point 1 in fig. admits of two readings, which only 22e, we see a perfectly consistent differ as to orientation (e. g. fig. 22j). object, viz. a partly transparent box But in the classical Necker cube, oriented as in 22g; but with the gaze both readings are equally consist- turned to point 2 in fig. 22e, -per ent, and the aptitude for reversal is spective reversals will be perceived quite resistant also to the addition to occur, and the orientation of the of further cues: even if anterior and figure in 22f will alternate with that posterior lines are drawn in differ- of 22g, even though only the latter ent colours, reversal can take place, is consistent with the local depth cue and a tridimensional wire construc- at point 1. These two sets of pictures tion is treated in the same way (see are taken by Hochberg to show that Kennedy 1974a: 138 ff). only those lines which fall near the In Hochberg’s view fig. 22e, is fovea act as cues in the representa- also open to two readings, one of tion of the depth and the corners of which is inconsistent, but fig. 22f an object, and and that the remaining and 22g only admit of one reading part of the percept, including orienta- each. However, it seems to me that tion, is hypothetically reconstructed also fig. 22f and 22g can be seen in on this basis. either of two orientations, where the Hochberg’s argument appears second is inconsistent; and while it to rest on his own perceptual experi- takes some effort to find the second ence in relation to the figures, and it reading of fig. 22f, fig. 22e appears seems possible not to share that ex- to shift from one orientation to the perience. I, for one, find it quite im- other quite spontaneously. The case possible to see fig. 22c as flat; nor do of fig. 22e is even more curious: I I think that it is properly speaking in- find it hard to see the reading which consistent. Instead, it appears to me is consistent, for I immediately per- to represent a strangely twisted and ceive the righthand side of the figure

87 to be the farther side of the box. Even This is very much like Gregory’s when fixing point 1, I often continue hypothesis-testing, as Perkins & to see the cube in this orientation, Cooper (1980:122) also remark ; and and although I sometimes manage to it has a somewhat more conceptual see the consistent reading, where the ring about it that Hochberg’s pro- right-hand side of the figure is the posal. Yet local cues cannot be the side side of the box which is closer whole story: Perkins & Cooper have to the viewer, the figure will soon re- already demonstrated that the pres- vert to the first, more stable percept. ence of symmetry is essential to the It would of course be interesting to way a picture is interpreted, and now know whether most people would they observe that symmetry suppos- agree with me or with Hochberg, on es cues to have been picked up from the interpretation of these pictures, all over the figure, and then to have or if they would suggest some third been integrated with each other. Un- reading; but even if my perceptions fortunately, Perkins & Cooper do not are completely idiosyncratic, they develop this line of thought. Howev- throw some doubt on Hochberg’s er, it was already demonstrated in the argument, for the physiological con- studies of the Granzheitspsychologie straints which it posits must apply to (Sander, Volkelt, etc) that holistic, each and every case.52 non-configurational properties have Perkins’s (1981:152ff) argu- primacy in all perception; and I have ment is similar (also cf. Perkins & suggested, as the most favourable Cooper 1980:122ff): fig.22h and 22i though somewhat implausible inter- have different orientations; fig. 22j pretation of both Eco and Gibson, can be seen both ways; and fig. 22k that what is identical from the pic- has no coherent interpretation, for the ture to reality is some kind of holistic righthand part is like that in fig. 22i, properties.53 and the left-hand part looks exactly There is no denying the im- like that of fig. 22h. The reason for portance of local cues, however. At this, Perkins (1981:154) submits, is some very elementary level fig. 22h- that the perceiver begins with some k, as well as fig. 22e-g, are probably local cue, and then attempts to ex- seen simply as “angularity” and “clo- tend the interpretation suggested by sure” or, somewhat more concretely, this cue to the entire figure until the as instances of “box-ness”. It is im- procedure runs into trouble, permit- possible, I think, to see fig. 22k as a ting a new cue to be taken up, and flat pattern. But then local cues are again extended as far as possible. essential. In the case of Gregory’s tree (fig. 13), it will be remembered, 52 In fact, in an informal experiment, a small detail is sufficient to change I found confirmation for all my readings, except for the ambiguity of fig. 21f-g. 53 Cf. Lecture 4. 88 completely the interpretation of the that Hochberg’s figures may be per- whole figure. But when it comes to ceived in other ways than those he the Necker cube and its variants, takes for granted, contradicts such there is not even that much of a dif- an identification. If, at least to some ference: here, as in Rubin’s vase persons, fig. 22c does not appear to which is also two facial profiles, the be flat, but three-dimensional and mouse which is an old man’s face, contorted, and even may appear to and the duck which is a rabbit, all be executing a spinning movement, the features of expression of the two then the presence of contradictory pictorial signs are identical, but the local cues in the space encompassed thematic hierarchies are different, by a single momentary glance is not because the thematic centre has been enough to destroy the three-dimen- differently located in the two cases. It sional percept. If the fixation of point should be possible to attend to local 1 in fig. 22e does not automatically cues, without failing to capture ho- change the orientation of the box listic properties, if, as in the noema, perceived, the fovea cannot be the there is a horizontal consciousness sole responsible for the election of accompanying that of the thematic the perceptual theme. Further, if fig. centre. As it is originally presented 22e and 22g are easily seen in such by Hochberg, one would expect pe- orientations as suppose inconsist- ripheral vision to be sufficient for ent percepts, consistency cannot be picking up some information, if only the fundamental criterion, and if the of a sketchy kind; but Hochberg later reading of fig. 22e giving rise to the only takes account of the fovea. It is most stable percept is the inconsist- possible, however, that peripheral ent one, there must be cues which seeing conveys the kind of informa- override consistency: perhaps, in tion gathered from tachistoscopic ex- this case, holistic properties of “box- posure by the subjects of Sander and ness”, which induce the most proto- and other Ganzheitspsychologen: typical orientation of the box to the holistic properties. If so, at the same viewer. Again if, in spite of the dis- time as the thematic centre gives us tance between the contradictory lo- the local cues (if we are able to pick cal cues, fig. 22b appears to be bent them up) the fringes offer holistic in- rather than being an ordinary frame, formation. Even the pictorial expres- the contradictory cues must some- sion would thus take on the mean- how have been sampled, but only ingful organization of the noema. when the over-all interpretation has But this noematic organization been established. cannot simply be identified with the A much more general point retinal distinction between fovea and needs to be made: if pictorial inter- periphery. The fact, referred above, pretations were simply contingent

89 on foveal fixations and their limit- between “genuinely impossible fig- ed scope, we, as well as Hochberg, ures”, which represent solid, stere- would have had to modify our in- ometrical bodies, and figures which terpretations many times during our are “not genuinely impossible”, extended scrutiny of the figures; in because they are not completely fact, however, there are a few pos- closed, or delimited, and fail to sep- sible, and some preferred, interpre- arate the “figure” clearly from the tations, which remain identical even “ground”. Bresti (p 17f) also quotes as we learn they are inconsistent. It Reutersvärd himself as saying that is conceivable, of course, that the re- the genuinely impossible figures are sult of the first fixation is stored in “more impossible”, because the ob- memory, and continues to influence jects which they render seem to be perception from there; but there may perfectly tangible, and the drawings be other sources also for the selec- can be coloured, and shadows added, tion of the perceptual theme (like the just as in the case of pictures repre- cultural expectations determining if senting ordinary objects. The distinc- something is seen as a window or tion is not quite clear, particularly as a tin can in Deregowski’s 1976:22f Bresti (p25) also presents a drawing picture). The important fact, how- by Reutersvärd which is “completely ever, is that we are not observering closed, although it is not a genuine- the real world, with its continuously ly impossible figure”, which seems changing appearances, in which what a contradiction in terms. However, is theme over and over again be- Gregory (1966:72 f) also suggests comes thematic field and/or fringes, that, while the Penrose triangle and and vice-versa, but a static artefact the Penrose staircase (used in Es- whose structures are fixed forever.54 cher’s “Ascending and descending”; The noematic adumbration in which cf. Escher 1967:76; Ernst 1978:90 the picture is given changes, but not ff) correspond to objects that can- the object given in the picture. Even not exist, the “three-stick clevis”, the fixation on a detail, which is con- also called “the devil’s turning fork tradictory, therefore does not change (fig. 18d), is a rather different kind the overall interpretation. of impossible object, which cannot even be seen — and the latter figure The Duck-rabbit and other is clearly among those which Bresti impossibilities considers as “not genuinely impos- In his discussion of Reuters- sible”. From the intuitive determi- värd’s “impossible figures”, Bresti nation, which is the only one at our (1985:24f) introduces a distinction disposal so far, it appears that Hoch- berg’s examples are all “genuinely 54 Cf. Lecture 2 about organism- impossible figures”. independent artefacts. 90 Bresti is of course wrong in sides, and vice-versa; and occlud- suggesting that Reutersvärd has, in ing bounds enclosing surfaces can- any sense, rendered fourth-dimen- not fit in with occluding bounds en- sional space; but it is true, as Kulpa closing air space; but this is exactly (1982:14) says, that only those ob- what happens in the devil’s turning jects that we tend to see the pictures fork. However, this explanation does as representing are impossible, while not seem to account for “genuinely the same drawings may also corre- impossible figures”: in most of the spond to other, quite possible ones work of Escher and Reutersvärd, – but, unfortunately, the latter must and in Hochberg’s examples, there then be two-dimensional patterns. are no “cracks” becoming “wires”, Indeed, Deregowski (1973:171ff) and so on, but “wires” in orientations found that only those who were 3D- which exclude each other. As a prac- perceivers, according to his construc- tical test of “genuine impossibility”, tion test, had trouble copying “the we may, following Reutersvärd, try devil’s turning fork” (cf. Kennedy to colour our figures: in the devil’s 1974a: 146ff). The works of Escher turning fork (fig. 18d), we must give and Reutersvärd illustrate how the up the attempt at some point between “rationalization of sight”, in Ivins’ the ends of the lines; but in Hoch- (1938: 1946) terms, just as all other berg’s figures, colouring the “wires”, formalizations, may, once they have if anything accentuates the contorted been accomplished, be detached three-dimensionality of the objects from their foundations, and applied perceived (fig. 22 l-m). to alien, and sometimes rather “irra- There is also some truth in Gre- tional” purposes: that is, the means gory’s curious observation that some which permit the rendering of the impossible figures are not only non- real world also allow its unreal (and existent, but cannot even be seen. impossible) extensions. The whole For if we try to follow one of the in- point of the “impossible figures” is ner lines of the devil’s turning fork, that they appear to be three-dimen- or Kulpa’s monobar, from one end sional, but we know they cannot be. to the other, there is some interme- Impossible objects, Kennedy diary point at which the inconsist- (1974a: 146 ff; 1974b: 235 ff) ency becomes flagrant; but in Hoch- claims, are made up of the same eco- berg’s figures, as in Penrose’s tribar, logically attested features as possible and in Escher’s stairs, there does ones, though in ecologically impos- not seem to exist any such single sible combinations: in the real world point at which three-dimensionality of perception, “cracks”, which have breaks down. For, strictly speaking, surfaces on both sides, cannot turn the devil’s turning fork can be seen, into “wires”, which have air on both and is seen: viz. as a fork. It is only

91 when we try to capture the inner de- figure from Harper’s Weekly, which tails that the fork becomes diabolic, took it over from Fliegende Blätter or virtually unseeable. Indeed, it (a German humour magazine more shifts from one interpretation to the famously used by Freud in his writ- other, both equally inconsistent, or ings about jokes; cf. 23a-b). Inter- dissolve in a general blur, instead of estingly, however, unlike Jastrow’s appearing real but contorted. In this and Wittgenstein’s duck/rabbit, the respect, the Reutersvärd picture so version in Harper’s Weekly has the contradictorily described by Bresti duck’s bill tilted slightly upwards, (1985:25), is really not a genuinely and the Fliegende Blätter version impossible picture, although it is en- even more so, which, as John Kihl- closed by a continuous contour: it is strom (2006a) observes, makes the genuinely unseeable. reversal form one interpretation to The essential fact, however, is another seem more dramatic. Kihl- that the devil’s turning fork is seen as strom (2006b-c) also points to the a fork, and that Hochberg’s figures existence of another similar figure, are seen as frames, although there alternating between the semblance exists no consistent way in which the of a dog and a chef (the first version local cues may add up to this impres- of which comes from an old Gibson sion. Independently of the perspecti- article), and he discovers for himself val cues, there seems to be a primary the possibility of interpreting another message about fork-hood, frame- pattern as either a whale or a kanga- ness, and what have you. But how, roo (Fig. 24a-b). then, is this message conveyed? Phenomena such as the duck/ The best way to view “impos- rabbit are not, as Kihlstrom points sible pictures” is perhaps as figures out, visual illusions properly speak- which are intermediate between two ing. Rather, they form “a class of other figures both of which -corre reversible, ambiguous, or bistable spond to possible objects (as some of figures commonly discussed by per- those discussed by Hochberg). Seen ception psychologists” (Kihlstrom in this way, the impossible pictures 2006c). The difference, as Kihlstrom are not so far from ambiguous figures, (2006a) sees it, is that the former de- such as the duck-rabbit: in both cas- pend on “unconscious inferences in es, it suffices to change a small detail perception” (and, thus, I take it, on to obtain a figure which is possible universal human tendencies), while or, as the case may be, non-ambigu- the latter “illustrate the role of expec- ous. This brings us back to Wittgen- tations, world-knowledge, and the stein’s duck-rabbit, which Wittgen- direction of attention”. Thus, Kilh- stein himself attributes to Jastrow. strom goes on to say, children tested In fact, Jastrow himself adapted the on Easter Sunday are more likely to 92 Fig. 23a-b. The duck/rabbit in Jastrow’s version (a) and the original in Fliegende Blätter (b) see the duck/rabbit as a rabbit – if, involve what Deregowski called the it could be added, these children are epitomic aspect of pictures: the iden- part of American culture or of the tification of the depicted object.55 Americanized world after some time The dog/chef picture and the after 1970. To my mind, this does not whale/kangaroo picture, as well as touch on the basic difference: visual Wittgenstein’s version of the duck- illusions are effects of the whole rabbit, are all basically what we have on the parts, which make metrical- earlier, following Arnheim, called ly identical lines seem different in “droodles”, that it, not pictures prop- size, and the like. Ambiguous figures erly speaking, but visual displays have to do with the identification of according to secondary iconicity. In the marks on the surface as particular this sense, they are like damp spots, objects of the Lifeworld. They are similar to impossible pictures, in that 55 See Lecture 2. They involve they their ambiguity, just as the impossi- epitomic aspect, but do not necessarily bility of the latter, resides in their re- exclude they eidolic aspect, as we shall see, in the case, for instance, of the Necker lation to the content level. They thus cube.

24ca-b: The chef/dog and the whale/kangaroo (from Kilhstrom 2006b-

93 clouds and the Rorschach test, with frame alignment does not simply the caveat that the latter not normally apply here to the fact that front part “bistable”, but shift between unlim- of the kangaroo is to the left on the ited number of possibilities. Interest- drawing, while that of the whale is ingly, Jastrow’s duck/rabbit and the to the right. This is, I take it, part of original version in Fliegende Blätter the reconstrual of component parts. are not droodles, but real pictures. Instead, reference-frame concerns It is thus even more remarkable that the way in which the depicted object they allow for different interpreta- is taken to be oriented in space, that tions. is, in the third dimension, depth. The Kilhstrom (2006b) suggest there whole point of the Necker cube is that are three different kinds of reversible either side of it may be seen as being or ambiguous figures, the principles closer to the observer. In order to see of which may combine in also com- the figure as whale, we have to see bine in single figures. the lower, right sight as being closer to us; if we opt to see it as a kanga- Figures that are subject to reference-frame realignments, such as the Necker Cube. Fig- roo, we must perceive the higher, left ures that are subject to figure-ground revers- side as the closest. Here, as Hochberg als, such as Rubin’s Vase-Faces. Figures that are subject to reconstruals of the whole or of (1980:51) notes in another case, that component parts. of a doll, the lines represent a round In the particular case of the surface at the point of passing out of whale/kangaroo, Kihlstrom observes, sight, as the earth disappears form there is a reference-frame realign- view at the horizon, not the corners ment, in which the front of the whale where the edges of the object itself becomes the back of the kangaroo. At meet, as in the Necker cube, or the the same time, there are also recon- edge of the surface, as in case of the struals of component parts of the fig- silhouette of a key. The whale, or the ure, in which the whale’s flipper be- kangaroo, is so as to speak transfixed comes the kangaroo’s foot, while the at a particular position in relation to whale’s tail becomes nose and ears. the viewer. There is no figure-ground reversal. As far as I understand, the 56As the example of the Necker cube duck/rabbit is basically a reconstrual suggests, the notion of reference- of component parts, in which the bill of the duck is transmuted into 56 Without reference-frame reversal, the ears of the rabbit, and so on. At the whale’s flipper becomes the airplane’s wing; the kangaroo’s leg becomes the least in the Jastrow and Fliegende branch on which the bird sits; or with Blätter versions, however, there also whole reconstruals of the object, the appears to be a slight change of ref- whale can also be seen as a dolphin, and erence frame: the nose of the rabbit the kangaroo can also be seen as a fox (Kilhstrom 2006b). seems closer to the observer than the 94 ears, and the bill of the duck appears the outer limits of the figure becom- to be closer than the back of its head. ing parts of its internal demarcation. There are non figure-ground revers- There is therefore also a small change als. But what exactly are we to un- of reference frame, for the difference derstand by the reconstrual of com- in distance from the observer of the ponent parts? Does it involve simply upper and lower lines is different substituting a pair of rabbit ears for in the bathing trunks and the glass a duckbill interpretation, and may it interpretation. This means that the also occasion a different division of whole does not even end at the same the whole into parts, that is, a seg- boundaries in the two interpretations. mentation into bigger or smaller As for Gregory’s tree (Fig. 14.), we units? Hoffman (1998: 89, 95), who have seen that the component parts also discusses the duck/rabbit figure, of the two interpretations are quite claims that the duck and the rabbit different, and of very different ex- must have the same parts, according tensions. to the minima rule, which states that Physical situations themselves, shapes should be divided ”at negative as Hochberg (1980:48f) observes, minima, along lines of curvature, of do not present any lines to the eye. the principle curvatures”. Therefore, Kennedy (1974a: 150ff; 1980; ”what is an ear of the rabbit, for in- 1994; & Fox 1977) had blind people stance, is an upper bill of the duck.” identify objects rendered by raised But this does not seem to be true: points; since they managed fairly that which is an essential part of the well, Kennedy concluded that what rabbit, the mouth opening, is a fairly both lines and raised points represent irrelevant inward bend of the lines is not light, but “the spatial juxtapo- making up the duck. Also, while the sition of surfaces” (1980:297). Else- bill is one component of the duck, where, Kennedy (1974 b215) tells us the rabbit has two ears, and as part of that a line, which is usually a deposit the body they must not necessarily of pigment, is “an in homogenization be taken to end at the same point as on a surface between two bounda- the bill. More differences of this kind ries (called contours) enclosing the can surely be observed in the Jastrow width”. It seems reasonable to con- and Fliegende Blätter versions. clude that raised points, as well as Arnheim’s droodle (Fig. 9a) is lines, are such inhomogenizations on basically a reconstrual of component a surface. Of course, raised points, parts, the olive becoming a navel and unlike lines, are three-dimensional; the Martini glass becoming bathing but since the third dimension is the trunks, but there is also a kind of fig- condition of possibility for their be- ure-ground reversal, or rather, an ex- ing tactually accessible, it can prob- tension of the figure, that which was ably be reckoned as non-pertinent. In

95 both cases, then, inhomogenizations Kennedy claims, to take Rubin’s ex- on surfaces stand for inhomogeni- periments to show that both contours zations in the space through which cannot be given interpretation at we move. It is not clear if any kind once: in fig. 25, this is only true if we of inhomogenization will do. Nor is see one half of the picture as a head it clear if such inhomogenizations profile; but not, for instance, if we may stand for anything else than in- consider it to be a clam on its edge homogenizations in space. Lack of with its two sides clasped tight. homogeneity is indeed a very ab- But there is no reason to stop at stract property for lines and objects this point; for while it is true, semi- to share. otically speaking, that either just one There is another property of lines boundary of a line, or both bounda- that in any case is not shared by the ries, may turn out to be pertinent un- edges of objects (nor by the elements der a particular interpretation, there of photographs, cut-outs, silhouettes, is also the further possibility that, etc.): that of having two boundaries besides the boundaries, the pigment that, so to speak, “face out” in differ- between them possesses a meaning ent directions. This is the difference in the picture (cf. Sonesson 1989a: (referred to in the discussion of Eco 265f). The interior line of Kennedy’s above), which Volkelt (1963: 28ff) drawing (fig. 25) could well be the noted between limits and contours, mouth of the clam, in which case the where the latter, but not the former, pigment stands for the mouth open- detach themselves equally from inner ing itself, while the boundaries, fac- and outer space. These boundaries ing on to the pigment and each other, are different, Kennedy (1974b: 218) represent the inner edges of the mov- observes, if the line is not straight: able body parts around the opening. for instance, the outer contour of a It should be noted, however, that ac- glove is like a mitten, but the inner cording to this same interpretation, one is like a hand with the fingers only the outer boundary of the oval slightly spread out; and the numeral line has a meaning in the drawing, 9 has one contour like 9, and another viz. as the outer edges of the clam’s which looks like . Depending on body (or, strictly speaking, its “hori- whether we give prominence to one zons” in Hochberg’s sense). But the contour or the other, different parts same drawing may also represent a of the drawing may emerge as “fig- wire construction, in which case both ure” and “ground”.57 But it is wrong, boundaries and pigment, of the oval as well as of the inner sinuous line, 57 And since Kennedy 1974a:154 correspond to particular parts of the thinks there is evidence of figure/ground effects also in ”haptic pictures”, he two borders; cf. 1980:294f and 1994 about apparently also thinks raised lines have ”ambiguities”. 96 represented object. In Arnheim’s droodle, the pigment does change meaning somewhat, because the middle vertical line, understood as the stand of the glass, makes up a body of its own, but understood as the limits between the legs, it is an empty space between two contours facing each other out- wards from left and right, respectively. More importantly, however, in the bathing suit interpretation, the main figure of the picture does not end with this middle line, branching out into a triangle standing on its head, as Fig. 25. A clam. Lines for it does on the glass interpretation. In- different uses stead, these lines are simply internal demarcation in a figure whose limits lines, sometimes one of its contours, cannot be seen (but which are sup- and sometimes both, which have to plied from our world knowledge).58 be taken a relevant for identifying The decision of what to take the figure. Curiously, these epitom- as figure and ground seems to be ic impossible figures, which would primary. It decides what is taken to seem to require as contradictory a be the theme and the fringes, as ap- interpretation as the corresponding plied to the picture. In most cases, of eidolic ones, have never been treated course, figure/ground reversals do as such. Consider, once again, Klee’s not suggest themselves, so the figure “Mother and child” (Fig. 21): part of is immediately given. The next ques- the line must be taken here, as the tion which involves frame alignment border both of the mother’s body and and component parts (which are at of the child’s body, each facing in a least partly dependant on each oth- different direction. Thus local cues er), would seem to that the operation seem to be contradictory, but the which decides, without the thematic whole does not offer any real prob- field, which part is the theme prop- lem of interpretation. Once again, er and which are simply parts of its the global interpretation gains the field. upper hand. It seems then, that in many cas- From perceptual to es, it is sometimes the pigment of the pictorial perspective 58 Cf. Lecture 4. In order to situate the similarities and 97 differences in which perspective, called the “sensate quality”; and the in the sense of a particular, subjec- phase which best permits the antici- tive-relative, division of an object, is pation of the complete temporal suc- rendered by means of such semiotic cession is called the “pregnant mo- resources as verbal language and pic- ment”. tures, it will be necessary to attend If we are to believe Lessing (and, briefly to the general differences be- in fact, many others who have writ- tween these types of semiotic media- ten about pictures since then, includ- tion.59 The most serious study of the ing Goodman 1968, who talks about difference between “literature” (that “density”), visual art is not only able is, most of the time, verbal language to describe the whole of space, but it in general) and “painting” (pictures cannot avoid doing so: pictures have and, to some extent, other visual to show “fully determinate entities”. modes of mediation) remains to this Taken literally, this must mean that very day Laokoon, first published in pictures are unable to pick up “sen- 1766 by the German writer Gottfried sate qualities”. Even if we limit this Ephraim Lessing.60 claim, as is no doubt intended, to It is the conviction of Lessing sensate qualities in the visual modal- that time cannot be adequately ren- ity, this is certainly not true: as I have dered in pictures, which is why visual shown elsewhere (in Sonesson 1989; art should ideally pick up one single 1994a), notably against Goodman, moment, and, in a parallel fashion, the “density” of pictures is only rela- language, which it not very conver- tive, and all kinds of abstraction are sant with space, should be content found in them.61 This applies to the to describe a unique attribute. Then, expression plane, in the case of more according to Lessing, an extension or less schematic pictures, such the to the whole will take place in the “impossible pictures” and the “am- imagination, spatially in language biguous figures” discussed above: but and temporally in pictures, that is, is also applies to the content plane of in the domain that the system cannot some pictures the expression plane adequately render. The property that of which is fully “dense”. Thus, for most easily allows such an extension all practical purposes, many pictures to the whole of the (spatial) object is are not about a particular person in 59 The following sections are adapted one or other disguise, but about more from Sonesson 2004b. 61 Simply put, “density” to Goodman 60 But, of course, even his ideas means that, no matter how fine the may be developed using more semiotics, analysis of something (e.g. a picture) as has been done by Wellbery (1984), into meaningful units, it will always be Bayer (1975; 1984), and the present author possible to posit another unit between (notably Sonesson 1988). More will be each two of those already given, and so on said about this in Lecture 4. indefinitely. Cf. Lecture 4. 98 or less abstract roles in relatively ge- ties is that, in the ordinary picture, neric situations. the space of representation is, at the To Lessing, in any case, the pic- same time, a representation of the ture is unable to abstract: Homer may space of ordinary human percep- show the gods drinking and discuss- tion, which impedes an organisation ing at the same time, but that is too by other systems. We have seen this much information to put into a single already in the case of the “duck/rab- picture. But as the example shows, bit” and similar figures, that the the- it is not the amount of information matic organisation serves the task of that is crucial, but the possibility identifying and presenting the ob- to organise it: verbal language has ject, that is, it is taken in charge, first fixed means for conveying relative by the epitomic function, and, in the importance, newness, focus, etc. The second place, if some resources are picture, however, in the prototypical left unused, by the eidolic function. sense of the term, may possess some In the history of art, these difficul- corresponding mechanisms which ties were at least partially overcome are not sufficiently known, but hard- by Cubism, Matisse, as well as some ly any systematic and content-neutral forms of collages and synthetic pic- means for organising such informa- tures, and it has been even more tion: that is, in Halliday’s (1967-68) radically modified by visual systems terms, there are no fixed devices for of information, logotypes, Blissym- separating that which is given from bolics, traffic signs, etc.. (cf. Sones- that which is new, and that which is son 1988; 1996b; in press a, b). Yet the theme (what we talk about) form it remains true that pictorial repre- the rheme (what is said about it). sentations lack systematic means for Indeed, although “background”, as rendering what Halliday has termed applied to language, is originally a “information structure”. visual metaphor, just as is “perspec- Although pictures do not render tive”, that which the picture places in the world in the form of “fully de- front is not always the most weighty terminate entities”, they have to di- element, with importance decreasing vide up the world in bigger chunks in according to increasing apparent dis- order to convey information about it tance; nor is necessarily the central than is the case with verbal language, figure the most semantically promi- and they lack any general means nent one.62 for imposing an internal structur- One of the principal difficul- ing on these chunks, apart from the one given in perception. In terms of 62 Self-styled semioticians like Kress more modern cognitive linguistics, & van Leeuwen (1996) claim the contrary, the same two points might be driven but it is difficult to grasp their justification for so doing. Cf. Lecture 4. home by saying that pictures cannot 99 pick one image scheme without also tungen”. Nothing similar exists, for having to choose several others, and instance, in language (and thus not in they are unable to organised these literature). Turning Lessing against schemes in order of relative impor- himself, we could say that there is no tance.63 The positive side of the same such “sensate property “ which, once observation, however, is that it is only being made explicit by language, in pictures (and to some extent in the could give us an impression of expe- cinema) that perceptual perspective riencing the whole of perceptual per- can be rendered as such, that is, as spective. By means of other semiotic it appears in the world of our direct resources, we may fix the relation perception. As far as I understand, of the positions of the observer and it is only here that it can be given that which is observed, whereas pic- the form of perceptual adumbra- torial perspective, like true perception that is has in reality. My second tual perspective, takes into account claim, however, to which I will turn all the phases connecting those two in a moment, is that there is another positions. Thus, the true perceptual sense in which pictorial perspectives, perspective of pictures could be con- in spite of being at the origin of the trasted with the mere positional per- metaphor, are further removed from spective of other semiotics means. perceptual perspective than others. Linguistic “perceptual” per- Pictorial perspective shows how spective is about relative position of something is perceived, not just that the viewer and the thing seen, not it is perceived: that is, it can, and in about the “how”, the nature of the as sense must, show all the details of link between the perceiver and the the perceptual relation connecting the perceived. We must grant, however, perceiver to the main object of per- that true perceptual perspective ap- ception. In Husserl’s terms, it shows pears to imply something about the the world in “perzeptuelle Abschat- relative positions of subject and object. In fact, there are some very 63 The term “image scheme” is used schematic pictures, exemplified by by such linguists as Lakoff, Langacker, traffic signs, which are more- simi Talmy, and many others, but I have nowhere seen any clear definition of lar to the kind of perspective we find what it means. It clearly implies that in language and literature. Even so, linguistic meaning is different from however, the picture shows, in ad- what logicians call propositions but is in dition to position, some aspects of some way more similar to pictures. The visual representations used, in particular, the “sensate qualities” of the object by Langacker and Talmy, suggests that positioned. This could be exempli- images schemes are some very abstract fied by the traffic signs for bus stop kinds of pictures corresponding to a (in Europe seen from the side, but in single or a very limit number of objects or events. Cf. Lecture 2. Mexico from the front), the car (from 100 Fig. 26. Different kinds of “positional perspectives” in traffic signs the front), the air plain (from above), gy, the perspective (which, as I sug- and that for a street with restricted gested above, is one way of dividing circulation, which mixes the per- objects into parts) becomes an object spectives (Cf. Fig. 26). Similar ex- in its own right.64 As perceptual psy- amples could of course by found in chologists such as Gibson, Pirenne, Egyptian frescoes and pictures from and others never tire of pointing out, many other alien cultures, in space it is only when we look at a picture and time. So the picture (at least the through a loophole, using one, im- schematic variety) may render some- mobile eye (“cyclopean vision”) that thing akin to positional perspective, there is any possibility of confus- but, again, it cannot do so by showing the picture with reality. In fact, ing mere position; but it is the only the surface is always perceived as a type of semiotic resource that allows surface. Although not part of any of for another possibility. these traditions, Perez Tornero (1982) There is, however, a second pe- gives a nice illustration of this two- culiarity of pictures, which makes layered perspective (Fig. 27). This them more different from real-world is the reason for calling perceptual perception than other semiotics, perspective, when manifested in pic- means: they are two-layered per- tures, for pictorial perspective. spectives. A picture is immediately a Strictly speaking, pictorial per- perspective on a perspective. This is spective does not imply any particu- not what happens in the perceptual lar positional perspective. In general, world : as Husserl argued and Gib- it cannot be supposed that depicted son after him, we “see through” the perspective is the perspective from perspective to the thing as such. This which the perceiver “sees” the mes- is not so in pictures. Even the best sage. This is certainly not true about linear perspective will not permit us ordinary lineal perspective, for al- to see “through” this surface to the though the “right” position may be thing depicted: instead, the perspec- 64 Actually, Husserl identifies tive is part of what is depicted. In “aesthetic perception” with terms of Husserlean phenomenolo- phenomenological reduction in his discussion of the “Bildbewusstsein”. 101 geometrically determined, it is nev- by the creator. er the one exclusively used by the Perhaps we should say, then, observer (as shown by Pirenne; cf. that pictorial perspective implies po- Sonesson 1989a:255ff). Anamorphic sitional perspective, but only an un- perspective may actually force the specified version of it. In other words, observer to adopt a particular posi- no particular subject or person occu- tion; but this is a very peculiar de- pying such a position is posited. If vice. In the second place, perceptual we take positional perspective to be perspective is not necessarily the the most abstract form of perspec- perspective from which the sender tive (since it only describes relative of the message created it – although positions), then it may be said that he has not doubt looked at the thing pictorial (and of course perceptual) depicted from the perspective de- perspective adds information about picted, among many others. The lat- that which is perceived, while per- ter point obviously does not apply to sonal perspective contributes infor- photography – only the camera has mation about the perceiver. In order to “look” from that perspective. The to illustrate these relationships (Fig. photographer may in fact be in front 28-30), I have adapted some sche- of the camera, as in Cindy Sherman’s matic figures from Langacker (2001, well-known photographs. Moreover, in press). However, the emphases in the case of computer-generated placed on different parts and the de- images, in particular those derived velopments of the subject pole are by means of an algorithm, the per- my own contributions. spective rendered is not even neces- Commenting on Lessing’s the- sarily one among those experienced ory, Bayer (1975; 1984) submits that

Fig. 27. The perspective of the picture and of the observer. From Perez Tornera 1982, as adapted in Sonesson 1989a.

102 the cinema should be able to synthe- We have seen that it is not the sise the advantages of language and relative position as such which is pictures: it has access to the whole important in pictorial perspective, of the visual world, and to temporal but the properties of the very process succession at the same time. This is of viewing. In a film, however, it is probably an all to simple conception precisely this relative position which of film as it is experienced, as may be seems to be important (somewhat like illustrated by the case of perspective. in language) : as-seen-from-above, The film picture is of course similar as-seen-from-the-right, etc. 67In to the static picture in being able to some contexts, films seems to em- render perspectival adumbrations. phasise the spatial perspective (the However, it is probable that these position), in others the personal per- adumbrations have less of a part to spective (the identity of person doing play in film perception, since they the seeing). For instance, a perspec- change all the time as the film -de tive from above does not necessarily velops. Thus, they are probably less involve positing somebody looking in focus — except in the experience down from above (God or the angels of the film analyst who may play the — except in some relatively recent film over and over again, look at it films by Wim Wenders). But if we in slow motion, and even investi- are first shown a person looking and gate each picture separately.65 On the then some view, we tend to think that other hand, unlike the static picture, what we are seeing is the perspective the film picture does permit us to of that person, in particular if the film “see through” the perspective to the then cut back to the same person, or thing given through the perspective, another one, looking. These are el- as in the ordinary perceptual world. ementary viewing habits, which are This is not surprising, since the film learnt very early by children (and picture, like the perceptual world, is emulated by some recent robots). It in perpetual movement. On the other is because of the temporal succession hand, because of the construction of – and the possibility of cutting it up the camera lens, it is still a case of in discontinuous chunks (the “mon- cyclopean vision.66 tage” of film theory) — that this is

65 In fact, with the generalisation kind of film. Some robots being built of the video library, not to mention the nowadays are able to integrate different DVD, this possibility becomes more camera views, just as the human eye does, easily available. Contrary to what film but so far, only the robot can “see” this. analysts like Metz, Bordwell, etc, always 67 I am of course referring to the seem to suppose, the cinema is no simulation of perceptual perspective longer necessarily the typical channel of here : there are other devices in films for circulation for film pictures. rendering some qualities of the viewing 66 It is possible to imagine a different process, most obviously blur. 103 Fig. 28. Perspective construction (inspired in Langacker 2001, in press). B = Observer; M: consciousness; VF = visual field; MF = maximal field of vision; OS = immediate field of vision; T = theme. This is essentailly a translation of Langacker’s scheme: what makes it into a positional perspective, however, is the fact that it is the relative position of M (or B generally) and T which is emphasized (thick line). possible in the cinema, but hardly in identified in static pictures: the gaze static pictures.68 of Matisse, of van Gogh, etc. On the Here, as if often the case, we are contrary, ascribed perspective here of course concerned with ascribed seems impossible. There have been perspective. What we see is ascribed arguments claiming that some donor to a person different both from the figure appearing in mediaeval paint- creator of the pictures and the observ- ings, and even some other marginal er. You must be at least a little of a person, is the one whose perspec- film connoisseur to identify the low- tive is shown, but this seem to me standing camera as the gaze of Ozu, a contrived parallel. These persons and the like. If anything, however, it are part of the (perspectival) picture, is the perspective of the creator, in even if they are painted on the border this peculiar sense, which may be of the pictorial space. What is lacking in the picture is the shift from the 68 Except of course in comic strips space of the perceiver to the thing and photo novellas. But these still feature perceived which we have in the per- static pictures, which are seen as objects ceptual world as well as in the film. in their own right. Perspective may We could of course have a picture therefore be more effective in the comics than in the cinema. where we “look over the shoulder” 104 Fig. 29. Perceptual perspective (same conventions as in Fig. 28). Here it is not the relative postion but its forms of transition (thick line) which are emphasized, and at least, in the static picture, the visual field rather than the immediate field of vision. of some person, as we have in the The limits of the cinema, but it is only in some wider perspective metaphor in context that such shots will identify language the perspective of what is seen at the The most elementary fact of linguis- other side of the subject’s body as tic perspective (in the limited per- being his perspective. ceptual sense) is that it is opposed There is of course another way of to non-perspectival devices : some embedding one person’s perspective words and phrases do not seem to within another, and that is by “quot- embody any (perceptual) perspective ing” it, which is obviously possible at all. This is of course impossible in static pictures and also, though in pictures as well as in film (even perhaps less naturally, in films. But though some abstract pictures may since this seems to be a case in which lack detailed perceptual adumbra- the perspectival metaphor, projected tions, as does traffic signs). So there onto language, is returned to picto- is an initial choice between perspec- rial media, it may be better to discuss tive and non-perspective, which does first the nature of perspective in lan- not appear within other semiotic re- guage and literature. sources, at least not within visual ones. An exception to this may well

105 Fig. 30. Personal perspective. Same conventions as in Fig. 228-29, with the following additions: S = sight; K = body; U = conception; E = emotion; I = information, knowledge; UM = forms of expression. be those kinds of abstract renditions realising perspective. of objects (e.g. the cube identified as In linguistics, perspective most closure plus angles) which are pro- immediately involves positional per- duced by small children and brain- spective : the difference between the damaged persons. But this case if speaker’s perspective and that of of course perspectival in the wider some person mentioned or implied non-perceptual sense of singling out (including the listener’s perspective, some properties – another of the three notably in some pronouns and verb modes in which the object may be di- forms). Such linguistic perspective, vided, which were discussed above. as codified in language, normally More to the point, traffic sign, Bliss, involves a relationship to the body logotypes, and many other schemat- of a person, which means that po- ic visual figures only have positional sitional perspective and personal perspective. But all these are at the perspective are hard to distinguish. limit of being pictures. Cubist paint- Thus, perspective presupposes em- ings, and “icons” in the religious bodiment, egocentric space, or, as sense of the term (as described by Us- linguists have earlier said, the origo penskij 1976a) contain several (more of the I-here-now. A rather curious or less) perceptual perspectives. But example is the sentence “The sun is choosing several perspectives is also right above the cabin”, pronounced a way of following the obligation of by someone standing on his head

106 (Langacker 2001: 37). In written lan- to be completely impossible in pic- guage (the most common material of tures (and the cinema) : there are of literature today, in spite of Lessing) course objects in the picture which the connection to the body of the can be described, linguistically, and speaker and hearer is of course much perhaps even, in a sense, perceived, looser. As we have seen, there are as being in front of and behind oth- only very indirect ways of embody- ers, but the picture is never struc- ing the listener’s or some third per- tured according to such a perspec- son’s perspective in static pictures, tive. There is no point of reference while there is much more leeway in apart from the implied observer of the cinema. However, two phenom- the pictured scene. In his book about ena to which we will now turn char- Russian icon paintings, Uspenskij acterise linguistic perspective but are (1976a, b) argued that these paint- hardly conceivable outside of verbal ings, in their central, religiously lad- language. en parts, were structured according A notion of perspective which to what he calls an “inner perspec- is linguistically very relevant in- tive”, that is, a view from the other volves some other object or person side of the painted scene ; but, even functioning as a kind of “landmark” so, this would not involve any inde- or, as I will say in the following, a pendent point of reference used by point of reference : something or the perceiver, but God as some kind somebody is in front of the house, of super-subject — it would simply behind the hill, etc. Perhaps it would mean that the “point of view of the be more correct to say that it is only picture” were even further from co- in language that something distinct inciding with the “point of view of from one’s own body may be used as the perceiver” than in other pictures. a point of reference. Someone who Another peculiarity of linguistic says that somebody else “came up perspective is when we identify with onto the stage” may himself be on the object, that is, we put our self in the stage, but it is also possible that its position: we speak, for instance, he uses the announcer as a point of of “the front of the house”. Linguists reference (Cf. Langacker 2001: 29ff). call this phenomenon “intrinsic The sentence “Dick is to the left of front” and define it as being the side Tom” may be true at the same time containing the main perceptual appa- as “Dick is to the right of Tom”, if ratus, which is in the direction of mo- Dick and Tom face you, and you in- tion and/or which is characteristical- voke either egocentric space or Tom ly oriented to the observer (Miller & as a point of reference (Cf. Miller & Johnson-Laird 1976: 400ff: Cf. Van- Johnson-Laird 1976: 275ff). deloise 1986). The first two criteria Perspectives, in this sense, seem involve some kind of identification

107 with the object, the third one rather Perspective in the with a second person, an Alter. Thus, extended sense in it supposes some kind of dialogical language and pictures projection, which is probably only The term “perspective” is readily possible in verbal language. Once used by linguists and philosophers again, descriptions such as these can in a much more extended sense: it of course be applied to pictures, both comprises all the different ways of as objects and as depictions, but they “dividing up the object” which we have no consequences for the organi- discussed earlier, thus, in addition to sation of the pictures. It could be said perceptual parts, proper parts and the- that intrinsic front is a property also matic hierarchies of properties. This only described by language but actu- often involves using perception as a ally forming part of our knowledge metaphor for cognition. Even though of the world. However, it seems to be he does not use the term “perspec- transferable to novel objects within tive” (but “construal”), Langacker language. (1991; 2001) clearly conceives the A more general conclusion difference between active and pas- might be drawn at this point. Only is sive forms of the verbal construction verbal language is it possible to use in this way. Analogously, differenc- something different from the subject es of tense and aspect (for instance itself as a point of reference. A more passé simple/pasado vs imperfait/ complete analysis would certainly re- imperfecto in French and Spanish) veal that what characterises language may be readily construed in this way. is the capacity to use an additional I will call this a conceptual perspec- point of reference, concurrently with tive. one’s own body. Something is “be- Literature may obviously em- hind the hill” in relation not only to ploy conceptual perspective, since the hill but also to the speaking sub- it is a resource contained in verbal ject. Or, more precisely, something language (also in less codified forms is “behind the hill” in relation to than those considered above). It ap- the relationship between the subject plies not only to the real subject of and the hill. The subject is primarily the situation of communication (or relevant as a body. Both referential at least to the creator) but also to a points and intrinsic fronts involve a number of ascribed subjects (the bodily positional perspective, or a “hero” but also the narrator). The bodily perspective, for short. No oth- only problem concerns the possibil- er semiotic resource than language ity of distinguishing this perspective seems to be able to make use of this from the other types. In static pictures kind of perspective. as well as in the cinema, the situa-

108 tion is somewhat different (although it only happens under very peculiar even there “pure” cases may be dif- circumstances. It is not a straightfor- ficult to find). Our idea of the style of ward way of using these semiotics a particular painter or photographer resources as might be said about lit- depends at least partly on conceptual erature and other kinds of language perspective: this is the case with van use. Gogh’s strokes (now even as fea- Elsewhere, I have distinguished tures in graphics software) as well as generic and specific picture depic- with Cartier-Bresson’s incomplete tions (Sonesson 1994). A picture scenes. However, even in this case is may “quote” another picture, but it difficult to conceive the possibility if the picture in question cannot be of imputing the perspective shown to identified from other sources, it is depicted or otherwise fictive persons. simply generic, and there is no way In the cinema, on the other hand, it we can know whether it is similar to can be done, because of the ongo- “direct speech”, “reported speech” ing pictorial flow. The difference in- or something intermediate. The case volving perceptual perspectives thus if different with specific picture de- seems to be reproduced. pictions. In Velázquez’s painting A more commonly discussed in- “Las Meninas”, several paintings by terface of perspective studies in lan- Rubens and Jordeans may be identi- guage and literature is the phenom- fied as hanging on the walls. In fact, enon often termed “erlebte Rede” or, it has even been argued that what we more generally, the possibility of re- see are the copies of these paintings producing (or only partly reproduc- made by Martínez del Mazo. Here, ing) the speech (or thought) of oth- we encounter the closest equivalent ers within your own speech. For the of “direct speech”. If anything, Pi- present purpose, we need to begin by casso’s version of “Las Meninas” is considering the general possibility comparable to “reported speech”, be- of reproducing “the other’s speech”, cause while we can recognise some in Bakhtin’s parlance, without con- elements of Velázquez’s painting, fusing it with our own, by means of they are all filtered through the style different semiotic resources (cf. Ba- of Picasso. Hamilton’s version of khtin 1981; 1984; 1986; Voloc&inov). Picasso’s “Las Meninas”, however, The question whether this operation may contain some elements of “er- uses direct speech, reported speech, lebte Rede”, with reference to both or something in between will have Picasso and Velázquez. This is at to wait. Obviously literature, as all least true in relation to Picasso, since other uses of language, is capable of Hamilton borrows the latter’s “way such reproduction. Even static pic- of speaking” (style, typical themes, tures and the cinema can do this, but etc.) even when it is not realised in

109 this particular work. other’s language” within one’s own In order to have something more speech: this is the way in which Ham- directly similar to “direct speech” in ilton “quotes” Picasso. This could be pictures, we would need to include a called a stylistic perspective. In the reproduction of some earlier picture case of both language and pictures, in our picture. Since paintings are it is not clear whether such a stylistic traditionally conceived to be unique, perspective can be – or even should the only way of doing this would be be – distinguished from conceptual to use a reproduction, as Duchamp perspective. did with Mona Lisa in “L.H.O.O.Q.”. However, as the term is used in This is of course more naturally done the study of language and literature, in the cinema, where there is not “erlebte Rede” also involves oth- supposed to be a unique copy of the er phenomena. Thus, for instance, work. Photographs, digital pictures, when Hellberg (1984) talks about and in fact all pictures whose copies “empathy markers”, it seems that it are given the same value as the origi- is not so much the other’s expres- nal (the “first copy”), that is, which sive means, but his or her emotions are not considered to be works of art, which are involved. Perhaps the case are of course equivalent to film here. is comparable to Jakobson’s (1963) Yet in all these cases quotation will “emotive function”, which is defined only function as such when we rec- to concern everything which in- ognise the work from other sources volves the “sender” of the message, (even when it is only the pictorial but which is then (in part) exempli- genre which is quoted, as in Cindy fied (as the choice of work also sug- Sherman’s “Film Stills”). To con- gests) by reference to the emotions struct “the language of the other”, of this “sender”. It could be argued quite independently within one’s that emotional perspective necessar- own language, when this other is not ily implies stylistic perspective, even a specific real person, as happens in though the opposite does not hold. a novel, seems to be impossible in But even this does not seem to me to the picture. be true. If I write, for instance, “He It is of course impossible to fully exclaimed that Damn! he would kill discuss here the relevance of some- me”, some much stronger word than thing as complex as “erlebte Rede” “damn” may really be used, but the to other semiotics resources than lan- emotional tone may still be that of guage, but I would like to add a few the other speaker. considerations. The analogies sug- Again, nothing similar seems to gested above suppose that “erlebte be possible in pictures. Like any hu- Rede” consists in incorporating parts man artefacts, pictures may of course of the expressive resources of “the express the emotions and sentiments 110 of their creators. And they may con- given expression (using for instance vey emotions to the observer in a special effects) followed or preceded way that is not completely arbitrary. by a scene in which this emotion is But a picture cannot express the sen- assigned to a depicted person. timents of the persons depicted. Of The meaning of perspective in course, it may depict the facial ex- literature is of course predetermined pressions and gestures of the persons by the resources offered by language. involved in the scene – but then it And yet it is perhaps no accident that simply conveys the emotions avail- literary scholars tend to conceive able to the observer of the perceptual perspective in a rather different way world. Nor can a picture render the from linguists. They are concerned emotions of the creator of a picture with the one who does the observing that it quotes. If we include values and/or who detains the knowledge, in emotional perspective, we might much less the one who has feelings argue that Egyptian frescoes express with respect to something and/or who the values of the Pharaoh, by depict- gives expression to such feelings. A ing him much bigger than everybody verbal text as a whole no doubt may else (something comparable to “sub- contain viewpoints that are not given jective maps”). But in fact, Egyptian any particular linguistic expression. frescoes no doubt expresses the val- Genette (1983: 48ff) describes focal- ue of the society at the time. Some- isation using a drawing with a head thing similar applies to the mediae- “observing “ (or otherwise having val donor who had himself painted access to) a bubble containing a head much smaller than the saints: this “observing” another head in a bub- value relation is really part of his so- ble, etc. This should not be confused ciety, and is expressed more clearly with Perez-Tornero’s picture of the in the contract with the painter. Even viewpoint of the picture – the latter the “inner perspective” attributed to does not contain any little head doing God in Orthodox icons does not re- the observing. It is a static reproduc- ally express God’s values or emotion standing for itself. Something tions. It simply embodies a way of similar to what Genette describes conceiving the world that would be can only be found in specific picture well known to the contemporary ob- depictions. server from many other sources. Against other literary scholars In the cinema, on the other hand, who have taken the perception meta- it seems quite possible to express phor rather literary, Genette insists emotional perspective attributed to that focalisation does not involve others. Because it is made up of a se- perception but only the amount of quence of pictures, a film may show knowledge shared — or, more ex- us some scene in which an emotion is actly, the restrictions imposed on

111 knowledge as compared to the om- film as well as in any other pictorial niscient narrator. That which then media in exactly the same sense as characterises other kinds of narrators in reality, that is, indirectly through is whether they have access to any perception. consciousness or only to the outside The amount of information world, and whether they have ac- conveyed about other minds thus cess to a single consciousness only depends on the aptness and will- (which may or may not correspond ingness of the persons depicted for to the person described as “I”) or to sharing this information with us. In several ones. In such cases, we will some cinematic genres, such as “ac- talk about information or knowledge tion films”, information about other perspective. It will be noted that what minds is largely irrelevant; and there is here taken as the zero degree of the are certain directors (such as Bres- classification is omniscience, where- son) who intentionally try to elimi- as the point of departure of pictorial nate the kind of access we have to perspective is always the knowledge other minds in the real world. On the identical to that which is given to the other hand, the cinema (and, to some perception of one person placed in a extent, other pictorial media) has the particular position. possibility seldom present in real- Even though knowledge per- ity to convey some particular types spective does not have the same of contents of a person’s conscious- systematic importance outside of ness, by means of such devices as literature, it may still be relevant to close shots of faces and extended other uses of language (even though shots (which are comparable, in this the omniscient narrator then is re- respect, to static pictures). vealed as a rather utopic personage). As such, however, the knowl- In pictorial media, on the other hand, edge perspective only seems to be- knowledge perspective is hard to come relevant in the cinema (as well isolate, since knowledge here must as in a series of pictures, for instance of necessity be translated into seeing a comic strip; cf. Sonesson 1988; (even in the cinema, if we abstract 1992a) when a clear dissociation is from the verbal part of the mes- perceived between the knowledge of sage). But this does not in any way the creator (and/or the correspond- mean that we are barred from entering ascribed subject) as it is trans- ing the consciousness of all persons lated into something which may be involved (apart from the director), seen, and the knowledge of the spec- contrary to what is suggested when tator. This happens, for instance, in Hemingway’s style is described as Hitchcock’s “Stage fright” where we being derived from the cinema: in are shown what the narrator (which, fact, other minds are given to us in a at this point, is the suspected assas- 112 sin) wants us to know, which is not literature, pictures and the cinema, identical to what he does know, that respectively, offer the resources ca- is, what actually happened. But this pable of expressing these different can only be discovered retroactive- kinds of perspective. In particular, I ly, when we are shown the real se- have suggested that perceptual per- quence of events towards the end of spective can only be emulated in the picture. pictures. The presence of pictorial perspective, which serves to identify Summary the object perceived (epitomic func- In this section I have described a tion) and its placement in the spatial kind of relationship to the subject world (eidolic function), at the same which may aptly be called a perspec- times serves to limit the possibilities tive while distinguishing it from oth- of adopting other kinds of perspec- er similar phenomena (such as “as- tive in pictures. pect”): it is a case in which an object is conceived as being constant, while References the modes of access which a subject Anati, Emmanuel (1976). Evolution and may have to it varies. I have also style in Communion rock art. Capo di singled out some different kinds of Ponte, Edizioni del Centro. Arnheim, Rudolf (1969) Visual thinking. perspective: positional perspective, Berkeley & Los Angeles: University of which only focuses on the respec- California Press. tive spatial positions of the subject Bayer, Duo, (1975): Lessings and the object; perceptual perspec- Zeichenbegriffe und Zeichenprozesse im “Laokoon” und ihre Analyse nach tive, and its particular variety, picto- der modernen Semiotik. Stuttgart. Diss. rial perspective, which are concerned — (1984) Laokoon - Momente einer with the exact perceptual adumbra- semiotischen Ästhetik, in Das tion in which the subject has access Laokoon-Projekt, Gebauer, Gunter, ed., 58-102. Stuttgart: Metzlersche to the object; and personal perspec- Verlagsbuchhandlung. tive, which is involved with the way Bakhtin, M.N,. (1981) The dialogic in which the subject itself is modified imagination. Austin, University of in presence of the object. In addition, Texas. — (1984) Problems of Dostoevsky’s I have distinguished different com- Poetics. Minneapolis, University of ponents within the consciousness of Minnesota Press. the subject, such as his body, his feel- — (1986) Speech Genres and other Late ings, his conception, his thought and Essays. Austin, University of Texas. — (1990) Art and Answerability. Austin, his means of expression. Against the University of Texas. background of a distinction between — (1993) Toward a Philosophy of the Act. the resources of verbal and visual Austin, University of Texas. media, I have also tried to determine Bierman, Arthur K., (1963) That there are no iconic signs, in Philosophy and to which extent ordinary language, phenomenological research, XXI11:2,

113 December; 243-249. Deacon, T. (1997). The Symbolic Species: Blanke, Börries (1998) Modelle des The Co Evolution of Language and the ikonischen Zeichens, in Zeitschrift für Brain. NY: Norton. Semiotik 20: 3-4, 285-304. Deregowski, Jan, (1973) Illusion and – (in press) L’iconicité comme double culture, in Illusion in Nature and Art. catégorisation. To appear in VISIO: Gregory, R. L., & Gombrich, E. H., thematic issue: Iconicity revisited/ (eds.), 160-191. London Duckworth & Iconicité revisité, Sonesson, Göran (ed), Co, Ltd. Third impression 1980. Spring 2005 – (1976) On seeing a picture for the first Boden, Margaret, (1977) Artificial time, in Leonardo 9:1, 19-23. Intelligence and natural man. New Donald, M. (1991). Origins of the Modern York: Basic books, inc., 1977. Mind. Three Stages in the Evolution Bordon, Emanuelle, & Vaillant, Pascal of Culture and Cognition, Cambridge, (2002) Le status du signe iconiqwue Mass.: Harvard University Press. entre iconicité et intertextualité. In Eco, Umberto, (1968) La struttura assente. VISIO, 6:4, hiver 2001-2002, 57-74. Milan: Bompiani (quoted from the Bouissac, Paul, (in press) Eroticon: Swedish translation, Den frånvarande l’iconicité renversée To appear in strukturen, Lund, Cavefors 1971).. VISIO: thematic issue: Iconicity – (1971) Die Gliederung des filmischen revisited/Iconicité revisité, Sonesson, Codes. In Semiotik des Films, Knilli, F, Göran (ed), Spring 2005 (ed.), 70-93. München: Hanser Verlag. Bresti, Carlo, (1985) De omöjliga figurerna – (1976), A theory of semiotics. Bloomington: /foreword/ in Reutersvärd, Oscar, Indiana University Press. Omöjilga figurer i färg, 5-30. Lund: – (1984). Semiotics and the philosophy Doxa. of language. Bloomington: Indiana Bruss, Elisabeth, (1978) Peirce and University Press. Jakobson on the nature of the sign, in – (1997) Kant e l’ornitorinco. Milan: The sign — Semiotics around the world. Bompiani Bailey, R.W., Matejka, L., & Steiner, P., – (1998) Réflexions à propos du débat sur (eds.), Michigan Slavic Contributions: l’iconisme (1968-1998). In Visio, 3, 1, Ann Arbor; pp. 81-98. 9-32. Broby-Johansen, R. (1967). Oldnordiske – (1999) Kant and the platypus. New York: stenbilleder. København: Gyldendals. Harcourt Brance & Co. Burenhult, Göran (1981). Stenåldersbilder. Ernst, Bruno, (1978) Der Zauberspiegel Stockholm: Stureförlaget. des M. C. Escher. München: Moos Bermúdez, José Luis (2005) Philosophy (Dutch original). of psychology. New York & London: Escher, M. C., (1967) The graphic work ot Routledge. M. C. Escher. , New York: Ballentine Carani, Marie (1999) Voir, penser, décrire books (Dutch original 1960). et percevoir l’objet visuel. In Visio, 4; Genette, Gérard (1983) Nouveau discours 2, 9-30. du récit. Paris: Seuil. Currie, Gregory (1995) Image and Mind. Gibson, Elenor (1969) Principels of Cambridge: Cambridge University perceptual learning and development. Press. New York: Appleton-Century-Crafts. Davidsen, Erling, & Munkholm Davidsen, Gibson, James, (1966). The senses Helle (2000) Iconicity and Evidence. considered as perceptual systems. In Iconicity - A Fundamental Problem Boston: Houghton Mifflin Co. in Semiotics. Troels Deng Johansson, – (1971) The information avaiiable in Martin Skov & Berit Brogaard (eds.), pictures, in Leonardo 4:1, 27-35. 81-90. Aarhus: NSU Press – (1978). The ecological approach to visual 114 perception of pictures, Leonardo, 4:2: & New York. 227-235. Gudnitz, Fred (1962) Broncealderens – (1982), Reasons for realism. Reed, Monumentalkunst. København: Martins Edward, & Jones, Rebecca, (eds). Forlag. Hillsdale New Jersey: Lawrence – (1972) Bronzealderens billedkunst. Eearlbaum Ass. Holbæk: Historiskt Samfund for Gombrich, E. H., (1960): Art and Illusion. Holbæk Amt. London: Phaidon Press. Second edition Halliday, M.A.K., (1967-68): Notes on 1962. transitivity and theme in English, in – (1963) Meditations on a hobby horse. Journal of linguistics, III:1, 37-81; London: Phaidon. III:2, 199-244; IV:2, 179-215. – (1972) Symbolic images. Studies in the art Hellberg, Staffan (1984) Satsens subjekt of the Renaissance. Londo: Phaidon. och textens. I Nysvenska studier, 64, 29- – (1982) The image and the eye. Oxford: 81. Phaidon. Hermerén, Göran (1983) Aspects of Goodman, N. (1968). Languages of art. aesthetics. Lund: CWK Gleerups. London: Oxford University Press. Hochberg, Julian (1972) The representation – (1970) Seven strictures on similarity, of things and people, in Art, Perception, in Experience and Theory, Foster, L., and Reality, Gombrich, E. H., Hochberg, & Swanson, J.W., (eds.), Cambridge. J., & Black, M.(eds,) 47-94. Baltimore MA: University of Massachussets Press and London: John Hopkins University 1970; pp. 19-29. Press. _ (1984) Of mind and other matters. – (1978a) Perception. Second, revised Cambridge. Mass., Harvard University edition. Englewood Cliffs: Prentice- Press. Hall, Inc. Greimas, A.J. & Courtés, Joseph, (1979) – (1978b) Art and perception, in Handbook Sémiotique: Dictionnaire raisonné de of Perception, vol. X; Carterene, E., & la théorie du langage. Paris: Hachette. Friedman, M., (eds.), 225-258. N. Y. Greenlee, Douglas, (1973) Peirce’s Concept and London: Academic Press. of Sign. The Hague & Paris: Mouton. – (1979) Some of the things that paintings Gregory, Richard (1966) Visual illusions, are, in Perception and pictorial in New horizons in psychology 1. representation. Nodine, C., & Fischer, Foss, Brian, ed., Penguin, 68-96. D., eds., 17-41. New York: Praeger Harmondsworth 1966. Reprinted 1973. Publishers. – (1997) Mirrors in mind. Oxford: W.H. – (1980) Pictorial functions and perceptual Freeman-Spectrum. structures, in The perception of pictures, Groupe µ (Dubois, J., Edeline, Fr., Volume II: Dürer’s Devices. Hagen, Klinkenberg, J.M., Minguet, Ph, , etc.) Margaret, (ed.), 47-93. New York: (1976). La chafetière est sur la table, Academic Press. Communications et langages, 29, 36- – (1994) The construction of pictorial 49. meaning. In The semiotic web 1992/93: – (1979) Iconique et plastique: sur un Advances in visual semiotics, Sebeok, fondement de la rhétorique visuelle, Th., & Umiker-Sebeok, J., eds., 109- Revue d’ésthétique, 1-2, 173-192. 162. Mouton de Gruyter, Berlin & New – (1992). Traité du signe visuel, Pour une York. rhétorique de l’image. Paris: Seuil. – (1998) Gestalt theory and its legacy: – (1995) Iconicsm., in The semiotic web organization in eye and brain, in 1992/93: Advances in visual semiotics, attention and mental representation. In Sebeok, Th., & Umiker-Sebeok, J., Perception and cognition at century’s eds., 21.46. Mouton de Gruyter, Berlin end. Hochberg, J., ed., 253-307. San

115 Diego: Academic Press. Iconicity revisited/Iconicité revisité, Hoffman, Donald, (1998) Visual Sonesson, Göran (ed), Spring 2005 intelligence. how we create what we Kennedy, John M. (1974a). A psychology see. New York: Norton. of picture perception. San Francisco: Husserl, Edmund (1939). Erfahrung Jossey-Bas, Inc. und Urteil. Prag: Academia – (1974b) Icons and information, in Media Verlagsbuchhandlung. and symbols, Olsen, David (ed), 211- —(1980) Phantasie, Bildbewusstsein, 240. Chicago; National Socieity for the Erinnerung. Husserliana XXIII. The study of education. Hague: Nijhoff. – (1994) Picture perception, in The semiotic Hünig, Wolfgang (1974) Strukturen des web 1992/93: Advances in visual Comic strip. Hilsesheim & New York: semiotics, Sebeok, Th., & Umiker- Olms Verlag. Sebeok, J., eds., 185-216. Mouton de Ivins, William M., (1938) The Gruyter, Berlin & New York. rationalization ot sight. New York: – Fox, N., (1977) Pictures to see and pictures Metropolitan Museum of art. to touch, in The arts and cognition, – (1946) Art & Geometry, Harvard Perkins, David & Leondard, Barbara, university press. New York: Dover (eds.), 118-135. London & Baltimore: publications, Inc., 1964. Johns Hopkins University Press. Jakobson, Roman, (1963) Linquistique Kihlstrom, John F. (2006a) Joseph Jastrow et poétique. In Essais de linguistique and His Duck – Or Is It a Rabbit? In générale /I/. Paris: Minuit. http://socrates.berkeley.edu/~kihlstrm/ – (1965a) À la recherche de l’essence du JastrowDuck.htm langage. In Diogène 51, 1965: 22-38 — (2006b) A new reversible figure – and – (1965b) Poesie der Grammatik und an old one. In http://socrates.berkeley. Grammatik der Poesie. In Mathematik edu/~kihlstrm/SEP06.htm und Dichtung, Gunzenhäuser, R, & – (2006c) Provenance of the Chef-Dog Kreuzer, H, (eds.), 21-32. München: Reversible Figure. In http://socrates. Nymphenburger Verlagsbuchhandlung. berkeley.edu/~kihlstrm/ChefDog.htm – (1975) Coup d’œil sur le développment de la sémiotique. Bloomington: Indiana Klinkenberg, Jean-Marie (1996) Précis University Press. de sémiotique générale. Bruxelles: de Janlert, Jan-Erik, (1985) Studies in Boeck knowledge representation. Umeå: – (in press) Clés cognitivistes pour une Institute of information processing. solution au problème de l’iconisme. Janson, Sverker, Lundberg, Erik B., To appear in VISIO: thematic issue: & Bertilsson, Ulf, eds. (1989). Iconicity revisited/Iconicité revisité, Hällristningar och hällmålningar Sonesson, Göran (ed), Spring 2005 i Sverige. Third, revised edition. Kolers, Paul (1977) Reading pictures Helsingborg: Forum. and reading text, in The arts and Johansen, Jørgen Dines, (1993) Dialogic cognition, Perkins, David & Leondard, semiosis. Bloomington: Indiana Barbara, (eds.), pp. 136-164. London & University Press. Baltimore: Johns Hopkins University Jappy, Tony. (2000): Iconicity, Press. Hypoiconicity. In Quiroz, João, ed. The Kress, Gunther, & van Leeuwen, Theo Digital Encyclopaedia of Charles S. 1996: Reading images. The Grammar Peirce. http://www.digitalpeirce.org/ of Vusual Design. London & New York: jappy/hypjap.htm Routledge. – (in press) In Defense of Iconicity To Kulpa, Zenon, (1982) Oscar Reutersvärds appear in VISIO: thematic issue: upptäcktsresor i okända världar, in 116 Reutersvärd, Oscar, Omöjliga figurer i Klincksieck. färg, 4-15. Lund: Doxa, Miller, George A., & Johnson-Laird, Philip Lakoff, George & Turner, Mark (1989) N. (1976) Language and perception. More than cool reason. Chicago/ Cambridge, Mass. The Belknap Press. London: Univ. of Chicago Press. Nordström, Gert Z (2000) Semiotiken Langacker, Ronald W. (1991) Concept, idag – lingvistisk imperialism eller image, and symbol. Berlin & New York: fenomenologisk narcissism. In Bild i Mouton de Grynter. skolan, 1, 16-19. — (2001) Viewing and experiential Osmond-Smith, David, (1972) The iconic reporting in Cognitive Grammar. In process in musical communication, In Linguagem e Cogniçâo, Soares da Versus 3, 31-42. Silva, Augusto (org), 19-49. Braga: Peissig, Jessie, Young, Micahel, Associaçäo Portugeusa de Linguística. Wasserman, Edward, & Biedermann. – (in press) Viewing arrangements and Irving (2000) The pigeon’s perception experiental reporting. In Papers from of depth-rotated shapes, in Picture the International Conference on Culture perception in animals. Fagot, Joël, ed., and Space, Quebec City, October 2001. 37-70. Hove: Psychology Press. Quinn, Andrew (eds.). Pérez Tornero, J.M. (1982): La semiótica Lessing, G. E., (1766): Laokoon -- oder de la publicidad, Barcelona: Editorial über die Grenzen der Malerei und Mitre. Poesie. Berlin. Reprint: Stuttgart Perkins, D., (1973) Compensating for Philipp Reclam Jun. 1964. distortion in viewing pictures, obliquely, Lindqvist, Cecilia (1989). Tecknens rike. in Perception & psychophysics, 14:1, En berättelse om kineserna och deras 13-18. skrivtecken. Stockholm: Bonniers. – (1975) A definition of caricature Liungman, Carl. G., (1991) Dictionary of and Caricature and recognition, in symbols. Santa Barbara, California: Studies in the anthropology of visual ABC Clio communication 2:1, 1-24. Lopes, Dominic (1996) Understanding – & Cooper, R., (1980) How the eye Pictures. Oxford: Clarendon Press. makes up what the light leaves out, Luckman, Thomas (1980). Lebenswelt und in, The perception of pictures, Volume Geschichte. Paderborn: Schöning. II: Dürer’s Devices. Hagen, Margaret, Mallery, Garrick, (1880-81): Introduction (ed.),. 95-130. New York: Academic to the study of sign language among Press. the North American Indians and A – & Hagen, M., (1980) Convention, context, collection of gesture signs. In Aboriginal and caricature”, in The perception of sign languages of the Americas and pictures, Volume I: Alberti’s Window, Australia I., Umiker-Sebeok, J., & Hagen, Margaret, (ed.), 257-286. New Sebeok, Th., (eds.), 1-76; 77-406. New York: Academic Press. York & London: Plenum Press 1978. Piaget, Jean (1945). La formation du – (1881). Sign language among North symbole chez l’enfant. Neuchatel: American Indians. Reprinted: The Delachaux & Niestlé. Third edition Hague & Paris: Mouton 1972. 1967. Maldonado, Tomás, (1977) Vanguardia – (1967). La psychologie de l’intelligence. y racionalidad (Translation of Paris: Armand Colin. Avanguardia e razionallta. Einaudi – (1970). Epistémologie des sciences de Editore, Turin 1974). Barcelona: l’homme. Paris: Gallimard. Gustavo Gili. Picado, José Benjamim, (in press) The Metz, Christian (1991). L’énonciation Problem of Iconism: a semiotical impersonnelle, ou le site du film. Paris: dogma. To appear in VISIO: thematic

117 issue: Iconicity revisited/Iconicité Iconicity revisited/Iconicité revisité, revisité, Sonesson, Göran (ed), Spring Sonesson, Göran (ed), Spring 2005 2005 Sonesson, Göran, (1988) Methods and Peirce, Charles Sanders, (1931-58) models in pictorial semiotics. Report Collected Papers I-VIII. Hartshorne, C, of the Semiotics, Project, Lund Weiss, P, & Burks, A, (eds.). Cambridge, University. MA: Harvard University Press. – (1989a). Pictorial concepts. Inquiries into Prieto, Luis (1966). Messages et signaux. the semiotic heritage and its relevance Paris: PUF. for the analysis of the visual world. – (1975a). Pertinence et pratique. Essai de Lund: Aris/Lund University Press. sémiologie. Paris: Minuit. – (1989b). Semiotics of photography. On – (1975b). Essai de linguistique et tracing the index. Report from the sémiologie générale. Genève: Droz. Semiotics project. Lund: University. Rosch, Eleanor, (1975) Cognitive reference – (1990). Rudimentos de una retórica de la points, in Cognitive psychology, 7:4, caricatura. In Investigaciones semióticas 1975, pp 532-547. III. Actas del III simposio internacional —, Simpson, C., & Miller, R. S., (1976) de la Asociación española de semiótica, Structural bases of typicality effects, in Madrid 5-7 de diciembre de 1988; Journal of experimental psychology: Volumen II, 389-400. Madrid: UNED. human perception and performance, – (1991). Kroppens byggnad och bruk. 2:4, pp 491‑502. Några semiotiska modeller. In Att tala Roupas, T. G. (1977) Information and utan ord. Människan icke-verbala pictorial epresentation, in The arts and uttrycksformer, Göran Hermerén, (ed.), cognition. Perkins, D., & Leondar, B., 83-100. Stockholm: Almqvist & Wiksell eds. 48- 79. Baltimore & London: John International. Hopkins University Press. – (1992a) Bildbetydelser. Inledning till Saint-Martin, Fernande, (1994) Avant- bildsemiotiken som vetenskap. Lund: propos: Pour en finr avec la mutité de la Studentlitteratur. peinture. In Nouveaux actes sémiotiques, – (1992b). The semiotic function and 34-35, 1-4. the genesis of pictorial meaning. In Saussure, Ferdinand de (1968-74) Cours Center/Periphery in representations de linguistique générale I-II. Edition and institutions. Proceedings from critique par Rudolt Engler. Wiesbaden: the Conference of The International Harrossowitz. Semiotics Institute, Imatra, Finland, Savan, David, (1976) An introduction July 16-21, 1990., Tarasti, Eero (ed.), to C.S. Peirce’s semiotics. Toronto: 211-156. Imatra: Acta Semiotica Toronto Semiotic Circle. Fennica. Sebeok, Thomas A., (1976) Contributions – (1992c). Comment le sens vient aux to the doctrine of signs. Bloomington & images. Un autre discours de la méthode. Lisse: Indiana University Press/Peter de In De l’histoire de l’art à la sémiotique Ridder Press. visuelle. Carani, Marie(ed.), 29-84. Sedgwick, H. A., (1980) The geometry of Québéc: Les éditions du Septentrion/ spatial layout in pictorial representation, CÉLAT. in in The perception of pictures, Volume – (1992d). Le mythe de la triple articulation. I: Alberti’s Window, Hagen, Margaret, Modèles linguistiques et perceptifs (ed.), 33.90. New York: Academic dans la sémiotique des images. In Signs Press. of Humanity/L’homme et ses signes. Stjernfeldt, Fredrik (in press) Christ Proceedings of the Fourth congress levitating and the vanishing square. of the International Association for To appear in VISIO: thematic issue: Semiotic Studies, Barcelona/Perpignan, 118 Mars-April 1989, Balat, Michel, a la retórica de la cultura/Från Deledalle-Rhodes, Janice, & Deledalle, varseblivningens till kulturens retorik, Gérard, (eds.), Volume I, 149-156. in Heterogénesis 15, 1-12. Berlin: Mouton de Gruyter. – (1996d) Iconicité de l’image — Imaginaire – (1992e) Bodily semiotics and the de l’iconicité. De la ressemblance à extensions of man. In Center/Periphery la vraisemblance. Actes du Premier in representations and institutions. Congrès de l’Association internationale Proceedings from the Conference of de sémiologie de l’image, Blois, The International Semiotics Institute, novembre de 1990. In VISIO, 1, 1, pp. Imatra, Finland, July 16-21, 1990., 23-34. Tarasti, Eero (ed.), 185-210. Imatra: – (1997a) Approaches to the Lifeworld core Acta Semiotica Fennica. of visual rhetoric. Visio 1, 3, 49-76. – (1993a). Pictorial semiotics, Gestalt – (1997b) The ecological foundations of psychology, and the ecology of iconicity, in Semiotics Around the World: perception. Review of Saint-Martin, Synthesis in Diversity. Proceedings Fernande, La théorie de la Gestalt et of the Fifth International Congress of l’art visuel. In Semiotica 99: 3/4, 319- the IASS, Berkeley, June 12-18, 1994. 399. Rauch, Irmgard, & Carr, Gerald F., eds., – (1993b) Die Semiotik des Bildes, 739-742. Berlin & New York: Mouton Zeitschrift für Semiotik 15: 1-2, 131- de Gruyter. 164. – (1998a) /Entries/ in Encyclopaedia of – (1993c) The multiple bodies of man. Semiotics. Bouissac, Paul, ed., New Project for a semiotics of the body. in York & London: Oxford University Degrés, 74, d-d42. Press. – (1994a). Prolegomena to the semiotic – (1998b) That there are many kinds of analysis of prehistoric visual displays. pictorial signs, Visio, 3, 1, 33-54 In Semiotica, 100: 3/, 267-332. 47-64. - – (1999) Post-photography and beyond. – (1994b) Sémiotique visuelle et écologie From mechanical reproduction to digital sémiotique, in RSSI, 14, 1-2, printemps production. In VISIO 4, 1, 11-36. 1994, pp. 31-48. 47-64. – (2000a) Iconicity in the ecology – (1995a) On pictorality. The impact of the of semiosis, In In Iconicity - A perceptual model in the development of Fundamental Problem in Semiotics. visual semiotics, in The semiotic web Troels Deng Johansson, Martin Skov 1992/93: Advances in visual semiotics, & Berit Brogaard (eds.), 59-80. Aarhus: Sebeok, Th., & Umiker-Sebeok, J., eds., NSU Press. 67-108. Mouton de Gruyter, Berlin & – (2000b) Ego meets Alter: The meaning New York. of otherness in cultural semiotics”, in – (1995b) Indexicality as perceptual Semiotica, 128-3/4, 537-559. mediation, in Indexicality. Pankow. – (2001a) La retorica del mondo della Christiane (ed.), 127-143. SSKKII vita. In Modo dell’immagine. Bologna: Report 9604.Gothenburg: Gothenburg Esculapio, Pierluigi Basso (ed.), 93- University. 112. Actes of the 5th Congress of the – (1996a) An essay concerning images. International Association for Visual From rhetoric to semiotics by way of Semiotics, Siena, Italy, 1998). ecological physics. Review of Groupe – (2001b) De l’iconicité de l’image à µ, Traité du signe visuel. Semiotica, l’iconicité des gestes. In Oralité et 109-1/2, 41-140. gestualité. Actes du colloque ORAGE – (1996b) Le silence parlant des images, 2001, Cavé, Christian, Guaïtella, Protée 24:1, 37-46. Isabelle, Santi, Serge (eds), 47-58. – (1996c). De la retórica de la percepción Paris. L’Harmattan.

119 – (2001c) From semiosis to ecology. On the rhetorics, Urbino 2002, Badir, Sémir, / theory of iconicity and its consequences Klinkenberg, Jean-Marie (eds.) for an ontology of the lifeworld. In – – (in press d) The extensions of man VISIO, 6, 2-3, 85-110. revisited. From Primary to Tertiary – (2001d) The pencils of nature and culture: Embodiment, in Krois, John, ed., The New light on – and in – the Lifeworld. body and embodiment. London & New In Semiotica 136-1/4, 27-53. York: Rouledge. – (2003a) The Symbolic Species revisited. — (in press e) From the meaning of Considerations on the semiotic turn embodiment to the embodiment of in biology and cognitive science. meaning. In: Zimke, T., Zlatev, J. and R. Workings Papers of the SGB poject. In Frank (eds.) Body, Language and Mind. http://www.arthist.lu.se/kultsem/pro/ Vol 1. Embodiment. Berlin: Mouton. SGBWP3.pdf Tilley, Christopher, (1991). Material – (2003b) Why the mirror is a sign — and culture and text. The art of ambiguity. why the television picture is no miror. London & New York: Routledge Two episodes in the critique of the Tomasello, M. (1999). The Cultural Origins iconicity critique. In Special issue. of Human Cognition, Cambridge, MA: Iconicity. European journal for semiotic Harvard University Press. studies. 15-2/4, 2003, 217-232. Tversky, Amos, (1977) Features of — (2003c) Uber die Möglichkeit von similarity, in Psychological review, bildhaften Metaphern. In Zeitschrift 84:4, pp 327‑352. für Semiotik, 25: 1-2, 2003: Metaphern —, & Gati, l., (1978) Studies of similarity, in Bild und Film, Gestik, Theater und in Cognition and categorization, Rosch, Musik, Johansen, Jørgen Dines, & E. & Lloyd, B., eds., pp 79-98. Hillsdale: Posner, Roland, (Hrsg.), 25-38 Erlbaum Ass. – (2004a) Rhétorique du monde de la Uspenskij, Boris, (1972): Structural vie. In Ateliers de sémiotique visuelle, isomorphism of verbal and visual art, i Henault, Anne, & Beyaert, Anne (eds), Poetics 5, ss. 5-39 83-100. Paris: PUF. — (1973): A poetics of composition. The _ (2004b) Perspective from a semiotical structure of the artistic text and typology perspective. In Essays on fiction and of compositional form. Berkeley, Los perspective. Rossholm, G., ed, 255-292. Angeles, & London: University of Bern: Peter Lang. California Press. – (2005) De la métaphore à l’allégorie dans — (1976a) Semiotics of the Russian Icon. la sémiotique écologique. In Protée, Lisse: Peter de Ridder Press. 33:1, printemps 2005, 77-92 — (1976b): The language of ancient – (in press a) From Iconicity to Pictorality. painting, i Dispositio. Revista hispanica de A view from ecological semiotics. semiótica literaria 1, 3, ss. 219-246 To appear in VISIO: thematic issue: Vaillant, Pascal (1997) Interaction entre Iconicity revisited/Iconicité revisité, modalités sémiotiques. De l’icône à Sonesson, Göran (ed), Spring 2005 la langue. Paris: Université Paris-Sud – (in press b) Iconicity strikes back: the (Orsay) third generation – or why Eco still – & Castaing, Marie-Françoise (in press) is wrong. To be published in Actes of Iconicity as Recognizability. To appear the 6th congress of the International in VISIO: thematic issue: Iconicity Association for visual semiotics, Quebec revisited/Iconicité revisité, Sonesson, City, 2001 Göran (ed), Spring 2005 – (in press c) La rhétorique de la perception. Vandeloise, Claude (1986) L’espace en Recherche de méthode. To be published français. Paris: Seuil. in the Acts of the Urbino symposion on 120 Vierkant Alfred (1912) Das Zeichnen der Naturvölker. In Zeitschrift für angewandte Psychologie 6, 299-373. Vilchez, Lorenzo (1983) La lectura de la imagen. Barcelona: Paidós. Volkelt, Hans, (19673) Grundlagen der Psychologle. München: C. H. Beck’sche Verlagsbuchhandlung. Voloc&inov, V.N. (1973) Marxism and the Philosophy of Language. New York : Seminar. Wellbery, David E. (1984) Lessing’s Laocoon. Semiotics and aesthetics in the Age of Reason. Cambridge: Cambridge University Press. Wittgenstein, Ludwig (1971) Philosophische Untersuchungen. Frankfurt/M.: Suhrkamp Wollheim, Richard (1980) Art and its objects. Cambridge: Cambridge University Press. Yates, Frances (1966) The art of memory. London: Routledge & Kegan Paul.

121