Theory of Semiotics

Table of contents :
Foreword

Note on graphic conventions

0. Introduction—Toward a Logic of Culture

0.1. Design for a semiotic theory
0.2. 'Semiotics': field or discipline?
0.3. Communication and/or signification
0.4. Political boundaries: the field
0.5. Natural boundaries: two definitions of semiotics
0.6. Natural boundaries: inference and signification
0.7. Natural boundaries; the lower threshold
0.8. Natural boundaries: the upper threshold
0.9. Epistemological boundaries

1. Signification and Communication

1.1. An elementary communicational model
1.2. Systems and codes
1.3. The s-code as structure
1.4. Information, communication, signification

2. Theory of Codes

2.1. The sign-function
2.2. Expression and content
2.3. Denotation and connotation
2.4. Message and text
2.5 Content and referent
2.6. Meaning as cultural unit
2.7. The interpretant
2.8. The semantic system
2.9. The semantic markers and the sememe
2.10. The KF model
2.11. A revised semantic model
2.12. The model "Q"
2.13. The format of the semantic space
2.14. Overcoding and undercoding
2.15. The interplay of codes and the message as an open form

3. Theory of Sign Production

3.1. A general survey
3.2. Semiotic and factual statements
3.3. Mentioning
3.4 The prolem of a typology of signs
3.5. Critique of iconism
3.6. A typology of modes of production
3.7. The aesthetic text as invention
3.8. The rhetorical labor
3.9. Ideological code switching

4. The Subject of Semiotics

References

Index of authors

Index of subjects

Citation preview

A THEORY OF SEMIOTICS UMBERTO ECO

INDIANA UNIVERSITY PRESS Bloomington

London

Published by anansement with Bompiani, Milan Copyright© 1976 by Indiana University Prus All rights reserved No pan or this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopyin1 and recording, or by any information storaae and retrieval system, without permission in writin& from the publisher. The Association or American University Presses' Resolu• lion on Permissions constitutes the only exception to this prohibition. Published in Canada by Fitzhenry & Whiteside Limited, Don Mills, Ontario Manufactured in the United States of America Library of Conar- Catalo1ln1 in Publication Data Eco, Umberto. A theory of semiotics. (Advances in semiotics) Includes index. I. Semiollcs. I. TIiie. II. Series. P99.E3 74-22833 301.2' I ISBN 0•253-359SS-4 I 2 3 4 5 80 79 78 77 76

CONTENTS

vii

Foreword Note on graphic convention, 0. ln1roduc1ion-Tow1rd I Loaic of Cullure 0.1. 0.2. 0.3. 0.4. 0.5. 0.6. 0.7. 0.8.

0.9.

Design for a semiolic lheory 'Semiolics': field or discipline? Communicalion and/or significalion Polilical boundaries: lhc field Na1ural boundaries: two definitions of semiotics Nalural boundaries: inference and signification Na1ural boundaries: lhe lower lhreshold Nalural boundaries: the upper lhreshold Epistemoloaical boundaries

I. Si1nific1tion and Communication 1.1. 1.2. 1.3. 1.4.

3 3 7 8 9 14 16 19 21 28 32

An elementary communic1tional model Systems and codes The s-code u suucture Information, communication, sipification v

32 36 38 40

CONTENTS

vi 2. Theory or Codes 2.1. 2.2. 2.3. 2.4. 2.S. 2.6. 2.7. 2.8. 2.9. 2.10. 2.11. 2.12. 2.13. 2.14. f-1S.

The sign-function Expression and content Denotation and connotation Message and text Content and referent Meaning as cultural unit The interpretant The semantic system The semantic markers and· the sememe The KF model A revised semantic model The model .. Q" The format of the semantic space Overcoding and undercoding The interplay of c.odes and the message as an open form

3. Theory of Sian Production 3.·1. A g�neral survey 3.2. Semiotic and factual statements 3.3. Mentioning 3.4. The problem of a typology of signs 3.5. Critique or iconism 3.6. A typology of modes of production 3.7. The aesthetic text as invention 3.8. The rhetorical labor 3.9. Ideological code switching

48 48 SO S4 57 SB 66 68 73 84 96 I OS 121 12S 129 139 ISI ISi IS8 161

172

191 217 261 276 289

4. The Subject of Semiotics

314

Rtftrtnce1

319

Index of outhor,

347

Index of 1ubjtct1

3Sl

FOREWORD

A preliminary and tentative version of this text (dealing with a semiotics of visual and architectural signs) was written and published in 1967 as Appunti per una semiologia de/le comunicazion/ visive. A more theoreli· cally oriented version - offering an overall view of semiotics and containing a long epistemological discussion on structuralism -was published in 1968 as la sr"'rrura assente. I worked for two years on the French, Gennan, Spanish and Swedish translations (only the Yugoslavian, Polish and Brazilian ones appeared with sufficient speed lo reproduce the original Italian edition without any addition) re-arranging and enlarging the book - and correcting many parts of It lo take Into account reviews of the Orsi Italian edition. The result was a book half way between· la sr"'tn,ra assellfe and something else. This 'something else' appeared in Italian as a collection of essays. u forme delco111e11uto, 1971. As for the English version, after two unsatisfactory attempts 11 lranslalion and many unsuccessful revisions, I decided (in 1973) to give up and lo re-write the book directly in English - with the help of David Osmond-Smith, who has put more work into adapting my semiotic pidgin than he would have done if translating a new book, though he should not be held responsible for the results of this symbiotic adventure. To re-write in vii

viii

FOREWORD

another language means to re-thi11k: and the result or this truly semiotic experience {which would have strongly interested Benjamin ue Whorl) is that this book no lonaer has anythina to do with La stru1111ra asse111e - so that I have now retranslated it into Italian as a brand-new wo,k (Trattato di

semiorica gc11erale).

Apart from the different {but by no means irrelevant) organization or the material, four new elements characterize the present text as a partial critique of my own precedin3 researches: (i) an attempt to introduce into the semiotic framework a theory of referents; (ii) an auempt to relate pragmatics to semantics; (iii) a critique of the notion of 'sign' and of the classical typologies or signs; (iv) a different approach to the notion or icon­ ism - whose critique, developed in my preceding works, I still maintain, but without substituting for the naive assumption that icons are non-coded analogical devices, the equally naive one that icons are arbitrary and ruuy analyzable devices. The replacement or a typology or signs by a typology or modes or sign production has helped me, I hope, to dissolve the umbrella· notion of iconism into a more complex network of semiotic operations. In doing so, the book has acquired a sort or 'chiasmalic' structure. In its first part, devoted to a theory of codes, I have tried to propose a restricted and unified set of categories able to explain verbal and non·verbal devices and to extend the notion or sign-function to various types of significant units, so-called signs, strings or signs, texts and macro-texts - the whole attempt being governed by the principle of Ockham's razor, no11 su111 multiplica11da entia praeter 11ecessitatem - which wpuld seem to be a rather scientific procedure. In the second part, devoted lo a tl,eory of sign productio11, I relt obliged to proceed in an inverse direction: the categories under consideration {such as symbol, icon and index) were unable to explain a lot or different phenomena that I believed to rall within the domain or semiotics. I was therefore forced to adopt an anti-Ockhamislic principle: emia su,11 m1dtipli­ t:t111da propter necessitatem. I believe that, under given circumstances, this procedure is also a scientinc one. I would not have arrived at the results outlined in this book without the help or many rriends, without the discussions that have appeared in the first six issues or the review VS-Quuderni di s111d/ semiutiL"i, and without confrontations with my students at Aorcnce, Bologna, New York University, Northwestern University, La Plata and many other places around the WOfld. Since the list of rererences allows me to pay my debts, I shall limit myself to warmly thanking my rriends Ugo Volll and Paolo Fabbri, who have helped me

Fo,�word

ix

throughout the various stages of the research - mainly by merciless criticism - and whose ideas I have freely used in various circumstances. Milan, 1967-1974.

NOTE ON GRAPHIC CONVENTIONS Sin&le slashes indicate something intended u an expression or a sign­ vehicle, while guillemets indicate something intended u content. Therefore /xxxx/ means, expresses or refers to «xxxu. When there is no question of phonolo&Y, verbal expressions will be written in their alphabetic form. How­ ever, since lhil book is concerned not only with verbal signs but also with objects, imases or behavior intended u signs, these phenomena must be ex­ pressed through verbal expressions: in order to distinguish, for instance, the object automobile from the word automobile, the former is written between double slashes and in italic. Therefore lau1omobilcl is the object conespond­ in& to the verbal expression /automobile/, and both refer to the content unit «automobile». Sin&le quotation marks serve to emphasize a certain word; double marks are used for quotations. llalic denotes terms used in a technical 1ense.

INTRODUC TION: TOWARD A LOGIC OF CULTURE

O. I. Design for a semiotic theory 0.1 .I. Aims of the research The aim of this book is to explore the theoretical pouibility and the social function of a unified approach to every phenomenon of signification utd/or communication. Such an approach should take the form of a general semiotic 1heory, able to explain every case of sign-function in terms of underlying systems of elements mutually correlated by one or more codes. A desillJI for a general semioticsJ a result lhe notion of 'meaning' as 'cultural unit' becomes applicable not only to the cotr,ortmotic terms but also to J)l11cotqurt11UJtic ones. Instead of putting into structural relationship names of intellectual

76

A THEORY OF SEM IOTICS

qualities, names of colors, or tenns of kinship, Apresjian (1962) Indicated fields which place pronouns in opposition (pronouns which designate animate things vs. pronouns which designate in11nlmott things : also for instance the place occupied by /you/ in English compared with the place occupied by /tu/ /lei/ /IIOi/ In Italian) or fields of verbs which designate different operations within the same sphere of operations (for example to advise, to assure, to convince, to inform etc., all of which belong to the sphere of transmission of information). This enables us to race the problem or the po11ible content or • 'functional monemes' (see for syncategorematic terms and or the so-called instance i-h, 1969). See on this subject 2.1 1.S., in which some suggestions for a componentill analysis of syncategorematic terms are proposed. • Space (as Naturally structural semantics hopes to establish the Semantic the Form of Content In Hjelmslev's sense) in its totality. But this aim, which can constitute a general hypothetical framework for research, comes up against two obHacles; one empirical and the other inherent in the semiotic process. The first obstacle is that until now such studies as have been undertaken only arrived at a structuring of very restricted subsystems, such u for example that or colors, or botanical classifications, or metereological terms, etc. The second obstacle is due to the racl that the lire of semantic fields is briefer than that of phonological systems where the structural models attempt to describe forms which remain unchanged for long periods or time within the history of a language. Since semantic fields give shape to the units or a given culture and establish portions of the world vision belonging to that culture, movements of acculturation and critical revisions of knowledge are enough to upset a semantic field. If Sau11ure's metaphor or the che11board is accurate, the movement or one piece will suffice to change all the relationships of the system. Therefore it is enough that as the culture • • develops, the term /Kunst/ be given areas of application which are much wider than usual, for the whole system of thirteenth-century relationships studied by Trier to be changed, thus depriving the term /l/11/ of its value. 2.8.3. The segmentation of semantic field• In what sense does a semantic field show the world vision belonging 10 • a culture? Let us go back to one or the classic examples of the theory of semantic fields and examine the way In which a European civiliution analyzes the color spectrum by assigning names (and therefore establishing cultural units) to various wave-lengths expre11ed In millimicrons.

n

11,eory of Cod,i

a, b. c, d. e. r. g.

red orange yellow green blue indigo violet

800-650 l11ll 640,590 mµ 580-550 l11ll 540-490 l11ll 480-460 rnµ 450-440 rnµ 430.390 rnµ

A preliminary and naive Interpretation might propose that the spectrum, divided into wave-lengths, constitutes the referent, the object or experience to which the names or the colors refer. However, we, know that the color was named on the basis o f a visual experience (which the simple speaker would define as 'pe rceptual reality') which is only tran�ated into wave-lengths by scientific experience. But let us assume that the wave,Jengths are something absolutely 'real'. There is no difficulty in stating that the undifferentiated continuum of the wave-length constitutes "reality'. Yet • science comes to know that reality after having divided ii into pertinent units. Portions or the continuum have been cut out (and as we shall see, they are arbitrary) so that the wave-length d (which goes from 540 10 490 millimicrons) constitutes a cultural unit to which a name is assigned. We also know that science has divided the continuum in such a way as lo justify in terms of wave-length a unit which simple experience had already cut out or its own accord and given the name /r,een/. The choice based upon naive ex perience ...,notorbitrory , in the sense that the exigencies or biological survival probably forced 1h31 unit to be tenned pertinent rather than another Gust as the fact that the Esltimos divide the continuum o f experience into four cullural units in place of the one which we call /snow/ is due to the facl that their vital relationship with ,now Imposes distinctions on them that we can disregard without suffering any notable damage ). But it wa.r arbitrary in the sense that 111other cuhure divided the same continuum in a different way, which means that the continuum is a content-stuff which can be cut into different fonnal systems. We are not lacking in examples: for the portion of continuum � (blue) Russian culture has two di fferent cultural units (corresponding to /,ol11buj/ and /si11i;n, while • Greco-Roman civiiiz.otion probably had only one cultural unit for the the vvlous names U1lo11c111/ /co,n,/111/) to indicate the purlion d , • and the Hindus combine under a simple term (and d1u1 a single cullur.tl unit) rht portion 11 • b. We can therefore say thal a aiven culture has divided the

..

..

78

A TIIEORY OF SEMIOTICS

continuum of experience (and it does not matter whether the conlinuum Is seen in terms of perceptual experience or defined by means of oscillographs and spectographs), making certain units pertinent and understanding others merely as variants, 'allophones'. Thus to single out a shade such as «light blue» and another such as «dark blue» means for an English speaker isolating a free variant, in much the same way as when two idiosyncratic pronuncia­ tions are sing)ed out from one phoneme which from the 'emic' point of view is considered a pertinent unit of the phonological system. All this leaves unsolved a question which will appear clearer when the units of two different semantic fields are compared in two different languages, Latin and English (Table 10), Table 10 Mouse Mus

Rat

which can be rendered as: "to the Latin word /mus/ correspond two different things which we shall call x 1 and x 2 " (Table I I). Table 11

IE] LG

On the other hand, since the existence of x 1 and x2 is only made evident by the comparison of the two semiotic systems, can we say that x 1 and x2 exist independently of the names which a language has assigned to them and which establishes them as cultural units and therefore meanings of a certain sign-vehicle? If we tum to colors the answer is simple. There is no reason why there must be a physical entity which begins at the wave-length 640 millimicrons and ends at the wave-length 590 millimicrons. In fact in Hindu culture the segmentation or the continuum occurs not at 640 but at S90 millimicrons. But why should there not be a cultural unit (and a unit of experience) which

79

77,,ory of Cod,i

,

• • goes from 610 to 600 millimicrons? Actually a painter with an extreme sensitivity to colors who possesses a more carefully graded system would answer that such a unit exists and is present in his own special code, where a specific name would correspond to that portion of the wave-length continuum. The problem concerning /mus/ is a different one. The zoologist would tell us that the x, and x, which correspond in English to /mouse/ and /rat/ exist as specific objects and that they can be analyzed in tenns of properties and functions. But what Foucault (1 966) has written on the 'epistemes' of different epochs and the variations in their segmentation of the universe, or what Uvi-Strauss ( 1962) has written on the taxonomy of primitive people, should suffice to make us aware that even on these points it is wbe to proceed with caution. Since, after all, a study of codes should not be concerned with x 1 and x2 , which are referents, it should be enouda to confinn that there exislS in English a semantic field governing rodenlS, which is more analytic than its equivalent in Latin, and that therefore for the speaker of English there exist two cultural uniu where for the speaker of Latin there exists only one. All this brings the problem of semantic fields back to the so-ailed Sapir-Whorf hypothesis and to the question of whether the fonn of communicative systems determines the world vision of a certain civilization. At this stage it does not seem appropriate to broach the question: it is enough to assume that (at least on the level of the segmentation of the experiential continuum into the form • of the content) .. there exists a fairly close in1eraction between the world vision of a civilization and the way in which it makes iu own semantic units pertinent. Given the elements in play - Y (material conditions of life), X (units of perceived experience), U (corresponding cultural units) and SV (the sign-vehicles which denote them) - it is not necessary to know at this point whether Y determines X, which generates U, assigning to it the name SV; whether Y strives to elaborate SV in order to segment the experience to which U corresponds; whether semiotic activity on a deeper level leads man to base his thoughts on SV, which not only produces U and X but directly condilions human being, to experience Y and so on. These are still extra·semiotic problems. It would be more interesting, from a semiotic point or view, to understand within which civilizations a semantic field functions and at what point it begins to dissolve in order to make room for another; and how, in the same civilization, two or more semantic Oelds can coexist although in opposition, when different patterns of culture are superimposed.

..

80

A THEORY OF SEMIOTICS

..

A typical example is provided for us by the series of definitions which Aulus Gellius gave to colors in his Nocus Allica� (ii, 26) in the second century A.D.: he, for example, associated the term Jrufu.1/, (which we would translate /red/) with fire, blood, gold and saffron. He stated that the tenn /xanlhos/ (color of gold) was a variation of the color red, just like /ki"os/ (which in the chain of interpretants reconstructed by philology must be understood as equivalent of our /yellow-orange/). He furlhermore considered, ... color red, /flavus/ (which we are also used to as altemalive names for the associating with gold, with grain and with the water of the river Tiber), and //ulvus/ (which is usually the color of a lion's mane). But Aulus Gellius calls the eagle, the topaz, sand, gold /fulva/ while he defines Jflavus/ as a "'mixture of red, peen and white"' and associates it with the color or the sea and of olive branches. finally he states that Virgil, in order to de£ine the color ..greenish" of a horse, uses the term /caerulus/, which is commonly associated with the color of the sea. The extreme confusion which strikes the reader in this one page or Latin is probably due not only to the fact that Aulus Gellius' field of colors was different from ours, but also that in-the second century A.O., in Latin culture, alternative chromatic fields coexisted owing to the influence of other cultures. Hence the perplexity of Aulus Gellius, who does not manage to amnge the material, which he takes from the works or writen of different epochs, into strict fields. As we have seen, the •actuar experience which the author could feel from looking at the sky, the sea or a horse is here mediated by recoune to given cultural units, and his world vision is determined (in a nther incoherent way) by the cultural units (with corresponding names) which he finds at his disposition.

-

We could therefore state that: (a) in a given culture thtre can exist contradictory semantic fields: this is an aberrant cultural occurrence which semiotics must take into consideration rather than try to eliminate it; (b) the

sam• cultural unit can itself become part of compltmtntary semantic fields

within a given culture. Carnap (1947, 29) gives the example of a double classification according to which animals are divided on the one hand into aquatic, aerial and terrestrial, and on the other hand into fish, birds and others. A cultural unit such as «whale» can then occupy different positions in the two semantic fields without the two classifications being incompatible. One must thus admit that the user of a language possesses within his 'competence ' the possibility of coupling a given system of sign-vehicles with various systems of meanings: {c) within a given culture a semantic field can

disintegrate with extreme rapidity and rtstructure itself into a new fi•ld.

Points (a) and (b) will be dealt with in 2.8.4,since they are matters fora theory of codes. But they also have direct consequences as far as sign production is concerned; mainly in the rhetorical and ideological treatment or diJcourse. So they will also be more deeply considered in 3.9. Likewise, point

Thtory ofCodts

81

• of code-changing, which is a branch or the (c) is the concern of a theory theory of sign production. For an example of this see 3.8.5.

2.8.4. Contradictory semantic fields As for a suitable example or contradicting semantic fields, I shall consider the problem of antonymous terms as pairs of oppositions constitut• ,,.. ing a semantic axis.

..

Lyons ( 1968) classifies three types or antonymy: (i) complementary antonyms such as •masculine vs. feminine'; (ii) properly called antonyms, such as •small vs. large'; (iii) antonyms by convcrseness, such as 'buy vs. sell'. • Katz ( 1972), on the other hand, subdivides antonyms into: (a) contradic· tories, such as 'mortal vs. immortal', which have no possible medial ion between them; (b) contraries, such as 'superior vs. inferior', and 'rich vs. poor' which have ,.some possible mediation between or beyond them ; (c) converses, such as 'husband vs. wife' or "buy vs. sell', which, like the converses in Lyons, imply syntactic transformations and entail an inferential relation of the type 'if . . . then._ 1

Even a superficial glance at some pain of antonyms 1-vehicle and upon which rely all the other connotations. Consequendy, connotative markers are all those which contribute to the constitution of one or more other culturol units expressed by the preceding sign-function. 14 noted in 2.3 ,i propos of denotation and connotation, denotative markers differ from connotative marken only insofar as a connotation must rely on a preceding denotation; the former are not distinguished from the latter because or their greater stability; a denotative marker can be very shon-lived if the code that institutes it lasts only /npa« d'un ma/in (as many secret agents, changing their ciphers day by day, know Yery well); while a COMOtative marker can be stably rooted in a social convention, thus lasting as long as the denotation upon which it is based. So the following formal definition should mmce to distinguish denotative from connotative markers: (i) a denotative marker is one or the po,itions within a semantic system to which the code makes a sign•vchiclc • any previous mediation ; (ii ) a connotative marker is one correspond without of the positions within I semantic system to which the code makes • sign-vehicle correspond through the mediation of a preceding denotative • marker, thus establishing a correlation between a sign-function and a new semanlic unit. Nevertheless such a definition results in being unsatisfactory, whether • sign from the point or view or a theory or codes or from that or a theory or production, since it is difficult to distinguish a denotative marker from a connotative one. It Is easy to assume that the sign-vehicle /dog/ denotes a given animal through certain physical properties or zoological features, and • connotes, among many other lhinp, cfidelity». But what about the marker •domestic•? When in 2.10.2 the problem of the semm,e as 'encyclopedi1 item' is discussed, the difficulty or such a problem will become clearer. At present it would suffice to say that, widtin the framework of a theory of codes, the slraightforward distinction between denotative and coMotative markers still remains to be definitely established. Perhaps a tentative solution may be given through an empirical formulation of the problem into the terms or a theory or sign pruduction. From the point or view or I theoiy or sign production one should clearly distinguish denotation from connotation:

..

86

..

A TIIEORY OF SEMIOTICS

(a)

a denotation is a cultural unit or semantic property of a given sememe which is at the wne time a culturally recognized property of Its possible referents; (b) 1 connotation is a cultural unit or semantic property of a given sememe conveyed by its denotation and not necessarily corresponding to a culturally "cognized property of the possible referent. These two definitions allow us to understand why in the Watergate Model AB denoted «danger levelt and connoted «evacuation» or «flood•. In fact • «danger 1..e1t was a cultural unit corresponding to a supposed actual state of the water (even though this state, rather than constituting an actual event, was already the ..suit of a segmentation of the continuum performed by another science, i.e. hydrography), «Evacuation» , on the contnry, was not 1 property of the supposed ..rerent, but a meaning aroused by the signification of the content corresponding to the supposed referent ( ' 1 ), In any case it must be made clear that in the following pages denotatio,r will nor bt taktn as an ,qui•altnt ofarmlion. In the same way connotation w�I not simply be an equivalen t of intension. For intension and extension are calegories of a t-values theory, while denotation and connotation. in my sense, are categories of a theory of codes. Thus denotation in the present eontext is a semantic property, not a corresponding object. Denotation is the content of an expression, coMotation the content of a sign-function. 2.9.2. Denotation of proper names and or pu..ly syntactic entities Having made this much clear, one could now proceed to establish a compositional thtory of stmtmtl. But first of all one must eliminate some misunderstandings about proper names and sign-vehicles of purely syntactic systems that lack any apparent semantic content, such as musical sounds, for example. These problems must be clarified because in logical literature - for Instance - It is frequently asserted that proper names do not have a denotatum and therefo .. an extension. Within the framework of a theory of codes, to assume that an expression can and must have denotation means th•t this expression does actually have a corresponding content, which can be analyzed into more elementary semantic unils. The problem of proper names is similar to the problem of iconic signs, which are commonly supposed to refer lo someone without there being a precise code to establish who this person Is (for example, images of people). Above all, we must try 10 undcrstand what happens in the case of proper names referring to known historical personages. We shall see later that the

Theory of Codes

87

other cases are not different structurally. The expression /Napoleon/ denotes a cultural unit which is well defined and which finds a place in a semantic field of historical entities. lrus field is common to many different cultures (there can be a very great variety of connotations attributed by different cultures to the cultural unit «Napoleon,. but its denotations do not change). Thus the sememe «Napoleon» should have several markers including that of being a human person. It is because of this that it is semantically ridiculous to say "if Napoleon is an elephant" (see 2.S.3). Now let us imagine a case where the author of this book receives the sign-vehicle /Stefano/. The author possesses a competence, shared with many people from his own environment, which provides for a field of cultural units that includes his own relations and friends, and the sip,-vehicle /Stefano/ immediately denotes for him his own son. In this case we are dealing with a much more limited code than the one by which the message /Napoleon/ was • decoded, but the semiotic mechanism 1w not changed. Spoken languages can exist that have very few speakers (idiolects). A possible objection is that /Stefano/ can also denote other individuals. But here we are simply faced I with a case of homonymy. Homonymy often occurs in the use of language, and contextual situations exist which have to be specified, when term /x/, which can refer either to a meaning «X 1 » or to a meaning «X1 ». must be understood in one way or the other. The universe of proper names is simply a • poor universe in which there sre many cases of homonymy. The linguistically semantic universe of coMected cultural units (the named human beings), is, however, quite rich and every unit in it is isolated by very precise systems of opposition. Syncategorematic terms are homonymous in the same way. The /to/ of /to be/ is nol the ssrne as the /to/ of /lo you/. However, Ullmann (1962:122) states that a proper name out of context does not denote anything, while a common noun out of context always has a lexematic meaning. But no sign-vehicle denotes, unless ii is referred (on the basis of the context) to a • u an element of a repertoire of specific code in which it appears primarily sign-vehicles. The graphic sign-vehicle /cane/, if it is communicated to me out • can be eilher a Latin of context and without any indication of code, imperative, or an HaJian common noun («dog») or an English common noun. Thus there must always be a code indication which rde,s to a precise vocabulary. The vocabulary may also include a section on lint names which ., would tell me that a sip,-vehicle such as /David/ iJ a proper name and therefore coMotes a human being of the masculine 5tX, Where propa- names of u11kno...,, persotu are concerned one would

..

•

A THEORY OF SEM IOTICS

88

,.,

�� �I

have to admit, however, that they connote but do not denote - reversing the opinion or J. S. Mill, for whom they could denote but not connote. These limilalions should be ad milled : proper names or unknown persons are sign-vehicles with an open denotation and can be decoded as one would decode an abstruse scientific term that one has never heard of, but that certainly must correspond to something precise. There is not much difference therefore between receiving the message /ascorbic acid/ and intuiting that it means a chemical compound {an imprecise connotation) without knowing which (no denotation), and receiving the message /David/ and knowing that ii must refer to a man (imprecise connotation) without knowing whom (no denotation). These are two examples of imperfect possession of the codes of a group. In the first cast I consult a chemist, in the second I ask to be introduced to David. But I could also want to know which is the position of «David» in a field or well-known cultural unils: David is the son or John and the brother or Sheila. Lei us now tum to the case or the signs in those semiotic systems that are purely syntactic and have no apparent semantic depth. Music is a typical example. Let it be quite clear that there is no question or defining what is the meaning of the graphic sign

1

This sign-vehicle denotes «note C» in the middle register of the pianoforte, it denotes a position in the system of notes; it denotes a class or sound events which have for interpretants mathematical values and oscillographic and spectrographic meuures. The problem is instead what it denotes and whether it denotes the note C itself, emitted by a trumpet. In this connection it must be said that the sign-vehicles of syntactical systems have denotations inasmuch as they possess some interpretants. Thus the note C of the central octave, or that emitted by the trumpet, denotes a position in the musical system that will be maintained despite various transpositions. It could be said that the physical signal lnote Cl denotes that position in the musical system which remains unvariable whether it is interpreted by the sign

or by the sign

B�i\);������� 3 e

II

�: ., e:

Th•ory of Cod••

89

In order to recognize the lnott Cl. the musician must hear it in relation to some other note and therefore to its position In the system (this relation may be purely mnemonic in the rare cues of so-called "absolute pitch"). One may object that lljelmslev (I 943: 99, 100) has carefully distinguished between semiotic and non1emiotic systems, or "'games". According to Hjelmslev '"the decisive point for the question of whether or not .... 1 sign is present is not whether it is interpreted, i.e., whether a content purport is ordered to it" (for there exist ..not interpreted but only interpretable systems") . but rather whether there are two planes and these planes are not conformal. "Two functives are said to be conformal if any particular derivate of the one functive without exceplion entea the same functions as a particular derivate or the other functive, and vice versa... In the case of "pure games"', as well as of music, formal logic and algebra, ••jf the two planes are tentatively posited .. the functional net will be entirely the same in both... Therefore these structures are not called 'semiotic' for they are interpretable but not biplanar (while lanJUages are biplanar and not conformal). lt should be enough to reverse Hjelmslev's position, therefore assenins that the decisive point of whether or not a sign is present is wh�thtr ii is interpreted ("interpretable' systems being only si&nal systems, ready to be inserted within a coding correlation), but one should also explain why and how such a counter-proposal can improve the theory. The reason is the following. To deny the nature of sign to conformal systems means to disregard a large portion of semiotic phenomena, first of all the entire range of so-calJed 'iconic' signs. It is not by chance that Hjelms!ev finds some affinity between so-called symbolic systems and these entities "that are isomorphic with their interpretation, entities that are depictions and emblems. like Thorwaldsen's Christ as a symbol for compassion, the hammer and the sickle as a symbol of Communism, scales as a symbol for justice, or the onomatopoeia in the sphere of language". We will see that all these signs, even if in some way 'conformal". are not monoplanar at all: simply the relationship between expression and content is ruled by ratio di/ficilis (see 3.4.9. and 3.6.). When speaking of this problem it will be shown why it is possible lo consider as signs even the units coming from conformal bu/ not monoplanar systems in which the expression form coincides at some extent with the content Corm. Hjelmslev classifies chess among "'pure pmes", but this example is misleading and throws a shade of perplexity on many other so-called 0symbolic" systems. In chess a given formal relationship between two .I correspond to an equivalent different pieces on the board dots not simply relationship on the content plane: a given mutual position, let us say, between the black Queen and the white Bishop (both beina conelated 10 rhe actual position of their own Kins and, in principle, of any other piece on the board) conveys - as the whole of its content - a series of optional moves, a set of possible responses, a chain of foreseeable (or unforeseeable) solutions

90

A THEORY OF SEMIOTICS

and therefore a series of new interrelational positions of the entire set of • piece in a given pieces. In other words, even if one maintains that a given position IRnotrs only itseU (a case of •monoplanarity'), one should agree that the same• piece connotts a series or possible moves - and in some way stands for them. Moreover, each position connotes di/{f!rtnt possibilities/or tad, of • the two players. Therefore the possible content of a single piece is I independent of the piece taken as expression. A chess game is a semiotic system with two planes and ils pieces act u functives of a sign.function. Insofar as every situation in a musical piece may (or may not) announce ... example a foreseeable but unpredicted musical solution, music offers another • of a semiotic system in which each situation could be differently interpreted.

'

..

2.9.3. The code and combinational rules

A sign-runction can be defined In itseir, and in relation to its combinational possibilities within a context. To maintain a sharp distinction between these lwo types or definition could clarffy many unsolved problems in semantic analysis. Al fi111 glance ii would seem that a theory or codes merely has lo consider the sign-runction in ilseir, for its combination within a con1ext is a mailer or sign production. But sign production is pennilled by rules previously established by a code, for a code is usually conceived not only as a correlalional rule bul also as a set or combinational ones. The facl thal an expression like /JohMy found sad/ is unacceptable depends on 1he code. Obviously a decision as lo whelher /Johnny found sad/ can in cenain circumslances be accep1ed as valid wdl depend on inlerprelive decisions connected to the practice of sign production. Thus the code states lhal /green colorless ideas sleep ruriously/ is semantically anomalous, but an • interpretive decision, within the context of a given text, can establish the legitimacy of such an expression - as when, for instance, It is viewed as a poetic device, a calculated and semantically interesting deviation from the nonn. Jn any case it would be very restrictive to assert that a code is only concerned with establishing the lsola1ed meaning or /sleep/ or or /ruriously/, without providing any indications abou t its combinational possibilities. Al Ibis point ii may seem necessary 10 conceive or a code as a double entity establishing on the one hand correspondences between an expression and a content and, on the other, a set of combinational rules. I believe, however, that the distinction between lhe deflnilion or the si(n·function in itself and the definition of the 1ig11·func1ion III a combina· tionol unit does not imply this double definition or the code. A code provides: (I) a 'restricted descriplion of a sign-runction so that It can be easily

77,eory of Code,

91

understood In its biplanar format independently of any context; (ii) a more complex definition which also foresees some nodal points In which the sign-function, in both its functives, can amalgamate with other sign-functions. In this way the notion of independent combinational rules can be avoided, for they are a part of the coded representation of the sign-function. Suppose for instance that the representation of /to love/ has a syntactic marker as V(x,y I - which specifies that the verb is transitive - and at least a semantic marker such as Action (A + human, 0 ± human), or 11..:tion -d A + h11111an -do • humlA Suppose then that the semantic representation of /to eat/ has semantic markers such as dactton -d A + humu

-do

+ o rganic. -human .••••

Al this point ii is easy to see why /John !OYes his father/ is semanticaDy acceptable and /John eats his father/ is semantically anomalous (except in a quite different cultural context in which even human beings are classified as possible food) ( • ' ). Insofar as a single sign-function can be ruled by many codes or subcodes, one ought to admil that every code establishes its own combinational modes. When speaking of a complex social type of competence such as a language, one should be thinking not of a single code. but of a system of interconnected codes. If somebody prefers to call such a system of systems of sign-functions 'a language', then the proposal could be accepted, provided that one is able to apply this word to each semiotic code without ambiguity or metaphorical transferences. 2.9.4. Requirements of a compositional analysis When considering the double delinitlon of a sign-function (in itself and In its combinational capacities) one realizes that the expression plane has a privileged status: any expression unit can be denned in itselr, not only Independently of its combinational possibilities but also of its material quality as a functive. Thus an expression (for instance the word /dog/ or a red nag on a beach) can be analyzed into its articulatol}' formants:/dog/: three phonemes, each of them formed by a bundle of pertinent features; a red nag: a geometrical form (resulting from an articulation of Euclidean elements) and a color, resulling from a given spectral situation. These expression markers

A TilEORY OF SEMIOTICS

92

•

remain the same even though the vehicle, ue not liken II vehicles (and therefore semiotic functiYes); they are stn,cturol proptrtit1 of the signal. When on the contrary the expression is considered in its combinational capacities, it acquires so-called syntactic markers. such as Singular, Masculine, Verb, Adjective, etc., which are ,rammatical properti'1 or the funclive. They may or may not be represented, 11 the content level, by corresponding semantic features USonne/ Is, syntactically speaking, Feminine In German while /sole/ in halian is Masculine; yet both expressions convey the same U semantic unit, which has no sexual markers) ( ). h Is now clear 1h11 the markers which must be considered relevant for a description of sign-function are only the markers or the funclivcs as such. The following discussion can thus dlsreprd the slNclunl muker, of the signal as such. They seem lo be more relevant lo a theory of sign production when this llller considers the 'labor' necessuy lo produce an ullerance (see 3.1 .). One may now outline I for,1 tentative 1111lytical model of sign-function. (I) The sign-vehicle pot1t11ts certain syntactical markers (such as Singular, Count etc.) which pennil its combination wkh other sign-vehicles, thus making some syntaclicaUy well-formed ..nrences acceptable even I hough they are semanlically anomalous (for instance /the train delivers a beautiful baby/), and making some other sentences unacxcptable even though, semantically speaking, they do make sense (for instance /it a, un autr,/) (1 4 ), (ii) A meaning u sememe lsformtd by semantic mukers of different kinds (denotations and connotations) which may be arranged hierarchically. Some of these markers may or may not correspond lo syntactic markers. (iii) In principle, no sl&n•funcllon Is performed by a simple syntactic marker, since the s)gn,funclion is established by the code bclwccn a given set of semantic markers and a given set of syntactic markers, both taken as a whole. This means that the sign-function /s not a marktr to m11rktr comlation; therefore the sign .function is not established on the grounds or a strict and 'n11ur11' homology between the two funcllvcs, but is the reNh of an arbitrary coupling ( • • ) . Therefore the schematic representation of the meaning o( a sign-vehicle (or or the sememe conveyed by a lexical item) should be as follows:

..

/1-v/ -sm-1S1---d 1 ,d2 .cl, --c 1 ,c, ,c, . . . (where /•v/ is sign-vehicle, sm arc the syntactic markers, 1S» Is the semcme

93

Theory ofCode,

conveyed by /s-v/ and lhe ds and the cs are the denolalions and connolalions which compose the sememe. Even if the representation of a sememe were lhal simple (and let us assume for the momenl thal it is) many problems would still arise as to the nature of the semantic components. Since a sememe is composed by a more or less finite and more or less linear set of elementary componcnls (denotations and connotations), lhe problems lhat musl be faced at Ibis point are: (i) whether these components can be isolated; (ii) whe1her or not lhey are a finite set of semantic 'universals'; (iil) whether they are theoretical constructs which do not need further semantic definition, or constructs of the type given by a dictionary, that is, words. definitions, purely linguistic constructs; (iv) whether !heir interconnection is sufficient to define a sememe and the way in which it can be inserted into a discourse, i.e. the way In which a given meaning can be contextually and circumstantially disambigualed. 2.9.S. Some examples of compositional analysis As regards point (i) Hjelmslev (1943) proposed the possibility of explaining and describing an unlimited number o( content.-entities by making use of a limited number of content-figurat, i.e. more universal combinatory features. Given four elementary features such as «ovine» and «porcine», and «male» and «female,, it is possible to combine them into the sememes «ram», «ewe», «pig» and «sow», these primary universal features remaining at one's disposal for further combinations. As to point (ii), according to Chomsky's first approach (196S), the syntactic mukers are undoubtedly a finite set of components on which the so-called 'subcateaorization rules" depend (for instance the subcatcgoriz.ation of verbs in Transitive and Intransitive explains why /John found sad/ is grammatically unacceptable). As for the semantic components, which give rise to the so-called 'selectional rules',Chomsk�states that ""the very notion of "lexical entry' presupposes some sort of fixed, universal vocabulary in terms of which these objects are characterized, just as the notion of 'phonetic representation' presupposes some sort of phonetic theory ..; thus selectional features are •universal ' and 'limited" or must be postulated as such. Unfortunately the only examples of such features so far distinguished are so •universal" that they are just able to differentiate a bishop from a hippopotamus but do not succeed in differentiating a hippopotamus from a rhinoceros. This difficulty rqards point (iv) and demands more analytical features. For instance, accordina to Pottier (196S) the sememe cfauuuil» can be analyzed by the scmes «pour 1'a11toir1, ·c,ur pitd(�J». «pour unt ptr1onnn, «avec dos1i�r» and uvtc bran. while the sememe «cona,w» has the first two semes, lacks the third and can or cannot have lhe list two. But since these r

94

A THEORY OF SEMIOTICS

'semes' are highly analytical, they fail to be 'univenar and. u reprds point (iii), need in lheir turn to be semantically analyzed. Greimas' 'structural emantics' (Greimu, 1966) seeks to establish semantic futures which are universal and are theoretical constructs which do not need a further analysis, or rather, which allow a further analysis but only in the sense that each feature1 posited as one amon1 the opposites relating to a dominating axis, can become the axis of an underlying opposition. Thus Greimas gives as 1n example the semic syllem or spatiality (Table 1 4). Tobie 14 spatiality Ill

I

+ dimensionality borizontality

verticality

�doJn)

prospectivity

�

(l�t)

11

• dimensionality I I

I

r-, surface

(wid•

.r)

•olume

.-1-,

(thick

thin)

latenlity (14',,.

I

ngrro�)

The bracketed words in italia are lexemes characterized by the presence of some semic clement : thus the couple lont/Jhort is chuacterized by the semea 1prospect irity1 horiz.ontality, dimensionality, spatiality». However, Greimu means by /ltximt/ the manifestation of an expression insofar as it is characterized by the presence or many semes; he calls on the other hand /1tmimt/ not the alobality or these scma. as I am doin1 in the present book, but a siven e{lttl dt 1tn1. or a particular ·readins' of the leximt. The limitation of this system seem, to be that the repertoire of these features ii not a finite one. One only has to consider the system of temporalily. or a syllem or values (Good, Bad, Acceptable, Unacceptable), in order to understand how such I system could develop like an ex pandina tbou,h llructurcd plaxy. Greimas• method ii very useful for explainina how a sememe can permit many rhetorical substitutions; thus to domon,trate that a lexeme like /tilt/ has 1 •noyeau ltmlqut" with ·nuclei' such 11 «exlrimiti» and «1phtroidl1t» helps one to understand why I here exist such catachreaes 11 /lltr d 'ipfotlt/, Jtltt du con 'IOyJ or /lltt dt pont/, dependin& on the scme made pertinent by the rhetorical 1ub1titution. Therefore a semantic repreacntation (especially if tryina to explain such problem, of sian production II aesthetic texts) must take markers of thll kind into account. But thla. ln icaclr. would not 1eem co be enoup(• • ).

77,eory o/ Codn

9S

So it is necessary to enlarsc the notion of semantic muter, ewcn lhoulb this may be prejudicial to the postulates of u nivenolity and limi tioL 2.9.6. A rust approach to a definition or the sememe A si8"·vehicle denotes and coMotes various cultu..i units, and some of these exclude each other. This means that among the various denotations and coMotations that make up a sememe alternative, complemaillJy or mutually exclusive readinp may occur, thus producing semantic incompatibilities. While the decision as 10 which reading the sender of the message presumably chooses is a mane, or sill' production (and interp.. 1a1ion), a theo,y of codes must provide the atructu..i conditions for such a choice. Thus a theo,y or the interpretation and disambiguation or sememes relin on a theo,y or their compositional nature. /Mus/ can denote «living being, in ..,pect to the axis 'animate vs. inanimate', uodentt in respect to a zoological field, clwmfub in respect to the axis 'harmful vs. harmless' and so on. In other words a sil"·Vthicle s, may denote positions a, and II, in two different semantic axes and, because of these denotations, can coMote the contradictory positions 71 and 1, in another semantic axis, further connoting, throu_. 71 , f 1 md f I in two other axes. Table 15

__

connotations ,

denotations I

I __ I

sip-vehicles

Sa

Si

s,

s.

i

a,

02 OJ

61 /Ji

61

62

6J

This is equivalent to G"imas' r,m,rk ( 19116:JB) that "le lexime tsl I• litu de manifestation et de renconue de M'mes provenant souvent de cattgories et de aystimes 5'miques dilT,rents et enuetenant a1ue eux des relations hitru• chiques, c"est-'-di" hypotaxiques".

96

A TIIEORY OF SEMIOTICS

Thus s, branches out Into various positions, not necessarily mutually compatible, In difft1tnl semantic axes, fields or 111bsysttms. This m..ns that the codes proride the speaker with a competence whlch includes a large series of semantic fields. Thtst can shift in many directions, and match up In nrious ways, ., that, according to the above diagram, the following situations art possible: (i) a speaker A knows all the possible coded denotations and connotations of the sememe 1S, • conveyed by the sip-vehicle /s, / and therefore, when r presupposed in 1hc process of 1ign productlun Pwduc1inn or Foanini tign•l-unht on < expression• Producuon of con1inuum 1ign1l-chmen

Focusing on codes

Information theory phonctia, nriout physical acicncc, I

S)'stcm makin,; Arciculuion of �Syncm observing cxprcssion-uni1s Syucm changing Correlation r funcuvcs

Otbcr diaciplinct dealing w;th 1hc umc suh;cct matter.


I

ntMu u

l

..,._.'ID.TS _,.,,_.,,.

I

g

!�

219

Th,ory of Sip Production

The table records the way in which expressions are physically produced and not the way in which they are semiotically co"elated to their content� the latter is implied by two decisions that must be made either before or after ., . production of the expression. the For instance, in the case of recognition of symptoms, there ii undoubtedly a pre-established motivation due to a preceding experience which has demonstrated that there is a constant physical relationship between 1 given agent and a given result; it has therefore been decided, by con,.ntion, that these resultant objects must be correlated with the notion of that agent under any circumstances, even when one cannot be sure that an existing agent has ,eally produced the result. In the case of wo1ds (which may be classed among 'systematically combinable units') the correlation is posiled after the production of the physical unit and is in any case independent of ils form (this assumption being valid even if by unverifiable historical chance the origin of words had some sort of imitative motivation). For this reason such non-homogeneous objects as a symptom and a wo1d are posited in the same row; every object listed there can be produced according to its pre-existing expression-type (ralio fad/is) and this happens irrespective of the 1easons for which these objects were selected as the • content. All of them could be produced by a suitably expression of a given instructed machine which only 'knows' expressions, while another machine could assign to each expression a given content, provided II was instructed 10 correlate functives (in other words, two expressions can be diffuently motivated but can function in equally conventional fashion). On the other hand, all objects ruled by a ratio dl/Ttcilis are so motivated by the semantic format of their content (see 3.4.9.) that ii is irrelevant whether they have been correlated with ii on the basis of previous experience (as in the case of footprints, where die semantic analysis of the content has already been performed) or whether the content is the result of the experience of 'inventing' the expression (as in the case of painting,). Therefore the motivated way in which they have been chosen (see the further analysis of Imprints and projections below) does not affect their mode of • production according to a ntio diff,ci/i1; they are corrdated lo certain aspects of their sememes - thereby becoming expressions whose features are also content-features, and thus projected ..,,,anlic nuzrk,n (•• ), • ,hould be In this sense a machine instructed to p1oduce these objects considered to have also received semantic instructions. One might say that since It Is instructed to produce expressions, ll is bein1 fed with schematic semantJc representations (2 • ).

...

..

•

220

A THEORY OF SEMIOTICS

The items recorded in the row corresponding to the parameter 'type/ token-tatio' may look like 'signs', since to some degree they recall pre, existing sign typologies. Bui they ,re not; they are short-hand formulas Iha! should be re-translalcd so as 'to produce imprints', 'lo impose I vectorial movement' or •10 replicate combinable units' and so on. 'Imprints" or 'examples' musl, at most, be understood as physical objects which, because or certain or their characteristics (nol only the way in which they are made, but also the way in which they are singled out) become open 10 a significant correlation, i.e. ready to be invested with dignity or functive. In 01her words they are po1enliaJ expression features or bundles of features. According to the system into which lhey are inserted, they may or may not be able I� convey by themselves a portion or content. So thal although they can also act as signs, they will not necessarily do so. I t must be clear that the whole or Table 39 speaks or physical procedures and entities that are ordtred to the sign,func1ion but lhat could subsist even if there were no code to correlate them to a content. On the other hand, they are produced in order 10 signify and the way in which they are produced renders them able to signify in a given way. A ready-made expression like /cherry brandy/ Is the result or two procedures depending on a double type/token-ratio; it Is constructed rrom two combinational units ordered by a vectorial succession� likewise a pointing finger is both a vector and a combinational unit, while a road arrow is both a stylization and a vector. Therefore items like 'vectors' or 'projections' arc not types or signs and cannot be equated with typological categories such as 6indices' or 'icons'. For instance both 'projections' and •imprints' could appear to be icons but the former would imply an arbitrarily selected expression-continuum and the liner a molivatedly established one, while both or them (equally governed by a ratio di/Trci/is) would be motivated by a content-type (though imprints are 'recognized', while projections are invent· ed'). Imprints and vectors look like indices, but are in rac1 dependent on two r • different type/token-ratios. Moreover, certain categories (e.g. 'fictive sam, pies') come under two headings: they are the result... or a double labm, since something must be replicated in order to be shown (ostension). All these problems will be dealt with fu rther In the following parag,aphs. I have only anticipated some examples In order to slress the ract that one must not look at Table 39 in order to find types or signs. This table only lisu types or productive activity 1h11 can give rise, by reciprocal and complex interrelation,, to different sign-runctlons, whether they are codtd units or codini texts.

Thtory of Sip Production

221

3.6.2. Rec0g11ilion

• object or event. produced by nature Recognition occurs when i given or human action (intentionally or unintentionally), and existing in a world or rac:1s as a fact among facts, comes to be viewed by an addressee as the expression or a given content, ellher through a pre-existing and coded 4 • correlation or through the positing of a possible correlation by its addressee. In order to be considered as the runctive or a s;gn.runction the object I must be considered as if ii had been produced by osJension, replica or event or invention and correlated by a given kind or type/token-11110. Thus the ,ct of recognition may re-constitute the object or event u an imprinl, a symptom or a clue. To interpret these objecls or events means to conelate them 10 a possible physical causality runctioning as their content, ii having being conventionally established that the physical cause acts as an uncorucious producer or signs. A,; we wUI see, the inrerred cause, proposed by means or abduction, is pure content. The object can be a fake or can be erroneou!ly interpreted as an imprint, a symptom or a clue, when in fact it is the chance product or other physical agents: in such a case the 'rec0g11ized' object expresses a content although the teferent does not exist. In the recognition of imprim,, the expression Is ready-made. The conl ), Since the experience of an event was constantly associated with a given imprinted form, the correlation, first proposed u the result of an inference, was then posited. In the recognition of symptom,, th• expression is ready-made. The content is the class of all poSJible causes (orpnic alterations). The ,. type/token-ratio is facilis, for red spots do not have the same semantic markers as measles, nor does smoke have ,, the same as Ore. Nevertheless within the sememic representation of their content there is, among the markers, both the description and the representation of the symptoms. This explains the way in which a symptom is correlated to the notion of Its cause; the notion of the symptom constitutes part of the sememe of the cause and It is thus possible to establish a metonymical correlation between the functives (by a pars 1010 procedure). The process is notion-to-notion (or unit-to-unit) and the effeclive presence of the referent is not required. There can be smoke even if there is no fire at all, which means that symptoms can be falsified without losing their significant power. The ratio being facilis, it would be incorrect to speak of a certain 'iconicity' of symptoms; they have nothing to do with their content (or referents) in tenns of similarity. Whm symptoms are not previously coded, their interpretation is a matter of complex inference and leads to the possibility of code-making. Symptoms can be used for mentioning (smoke means «there is fuet, red spots on the face mean «this child has measles,). In this case the mentioning procedure works as follows: by a coded and proved causality (contiguity) of the type 'effect to cause', an effective presence of the cau�ng whole is deduced. In the recognition of dues, one isolates certain objects (or any other kinds of trace which are not imprints) left by someone on the spot where he did something, so that by their actual presence the past presence of the agent .. can be inforred. It is evident that, when used for mentioning, clues work in exactly the opposite way from symptoms; by a coded and proved contiguity {of the type 'owned to owner') a possible presence of the causing agent is abductd. In order that the abduction be performed. the object must be conventionally recognized as belonging to (or being owned by) a precise class of agents. Thus if at the scene of a murder I fand a dental plate I may presume that, if not the murderer, at any rate someone who has no more natural teeth hu been there. If on the noor of I political party's office, recently broken

•

...

..

...

224

A THEORY OF SEMIOTICS

into, I find the badge or the rival organization, I may presume that the burgjan were the 'bad guys' (obviously clues can also be falsified, and in cases like this they usually are). As a matter or fact clues are seldom coded, and their Interpretation Is • frequently a matter or complex inference rather than or sip-function recognition, which makes criminal novels more interesting than the detection of pneumonia. One could say that Imprints and clues, even though coded, 11e 'proper names', for they refer back to a,;,,t,r agent. The objection does not affect the fact that they refer, in any case, to a content, far there is nothing to stop the class to which the expression ,efen from being a one-member class (see 2.9.2.). But in fact very seldom can imprints and clues be interpreted as the traces of an individual agent (indeed maybe never). When looking at the footprint on the island, Robinson Crusoe was not able to think about 111 individual. He detected «human being». When discovering Friday he was undoubtedly able to .xpress the index-sensitive proposition «this Is the m111 who probably left the footprint•. But even if he had previously known that there was one and only one man on the island he would not, when looking at the footprint, have been able to refer i t to a precise individual; the primary denotation of the expression would have been «human being• and the rest would have had to be a matter or inference. It is very difficult lo imagine an Imprint that mentions a referent without the mediation or a content (•• ). The only case would be that in which one sees a given individual in the act of producing a footprint; but in this case the footprint would not be 'recopized' u a sip, far it would not be 'instead or something else, but 'along with' It (see the case of mirror> in 3.5.5.) (• s ). The same happens with clues. Even if I know that only one particular man, among the murdered person's circle or friends, hu a dental plate, I cannot regard the object lert at the scene of the crime as a sign referring back to a «person x». The object simply mans .:person without teelh1, and the rest is once again a matter of Inference. On the contrary many clues are overcoded objects. Suppose that I find a pipe in the same place. What makes me sure that a man wu there? A social rule establishing that gentlemen smoke pipes and ladles don't (the opposite would happen If I found a bottle or Chanel No. 5). 3.6.3. Ostenslon Ostension occurs when a given object or event produced by nature or

11r,ory ofSign hod11cllon

225

human action (intentionally or unintentionally and existing in a world of facts as a fact among facts) is 'picked up' by someone and shown as lhe expression of the class of which it is a member. Ostension represents the most elementary act of active signification and it Is the one used in the first instance by two people who do not share lhe same language ; sometimes the object is connected to a pointer, at others it is regularly picked up and shown; in both cases the object is disregarded II a token and becomes, instead or the immediate possible referent of a menlion, the expression of a more general content. Many thing, have been said about slf11ification by ostension (see for instance Wiugenstein, 1945:29-30) and a purely ostensive language has been invented by Swift. ut me only remark here that in ostension there is always an implicit or explicit stipulation of p,rtin,nce. For example, if I show a packet of brand X cigarettes to a friend who is going shopping, this ostension can mean two different thing,: either «please buy some cigarettes» or «please buy this brand of cigarettes». Maybe in this latter case I would have to add certain indexical devices, such as tapping with the linger on the part of the packet which bears the name of the brand, and so on. Likewise, in other circumstances only a previous stipulation of pertinence makes clear whether,

.. when showing a packet of cigarettes, I mean «packet of ciguettes, or simply «cigarettes». At other times ostension may suggest an entire discoul'5C, as when I • show my shoes to someone not in order to say cshoes1 , but rather cmy shoes are dirty» or «please shine my shoes». In these latter cases the objec1 is not only taken as a sign but also as a referent and the indication constitutes an act of mentioning. As a milter of fact it is as if I were saying «shoes (ostension) + these (mention)+ shoes (referent)». This theory solves the problem of �ntrinsically coded acts or object' (see 3.5.8.) without implying that a part or all of the referent will constitute a part of the definition or the Sif11•function; the shoes are first of all viewed u an expression which Is made with same stuff as its possible referent. Therefore ostenslve si8JIS (depending on choice) ue homomateriol. In principle ootensive production should be considered as governed by a ratio diffic/1/s, for the shape of the exprse (a spalial direclion in the wrillen phrase and a lemporal one in lhe u11ered one) 1hat makes lhe contenl undersiandable; by changing round lhe proper names lhe enlire content is reversed. Again, a vectorialization is neither a sign nor a complete expression in ilself (except taken as an expression signifying a pure veclorial correlation as in /a:Jb/), but rather• a productive feature that, in conjunction wilh olhcrs, conlribules to the composilion of lhe expression (•• ). One could say that In some cases a veclor by iiself can give rise lo a sign-funclion; suppose thal I hum an upward pilch-curve; I can succeed in signifying «queslion• (or d am questioning» or «what?») by imposing a direc1ion on a sound-continuum without resorting to any other device. But this is a caseor coded slylization. Many veclors are governed by a very schemalic IVtio diffici/is so easily recognizable lhal, as happens wilh S1y1iza1ions, a sort of caLachresizing process lakes place and the ratio di/ficilis practically becomes a ratio facilis. The case of lhe inlerrogalive humming cited above is a lypical example of Ibis process. 3.6.6. Programmed stimuli and pseudo-combinational uni Half way belween replica and lnvenlion there are lwo kinds of productive operation lhal are nol usually considered as semiolically de­ finable. The first one concerns the disposilion of non-stmiotic dements inlended lo elicil an lmmediale response in lhe receiver. A flash of light during a theatrical performance, an unbearable sound, a subliminal excitation, and so on, are lo be listed among stimuli rather than signs, as was stressed in 3.5.5. But In lhe same paragraph we no1ed 1h11, when lhe sender knows lhe possible effecl of the displayed stimulus, one is obliged to consider his knowledge as a sort of semiutic com�te11ct, for to 1,im a given stimulus corresponds to a given foreseeable reaction that he expressly aims to elicit. In

-

242

A THEORY OF SEMIOTICS

other words, there 11 a sip,-runction by which the stimulus Is the expression plane of a supposed effecl functioning as its content plane. N..ertheless the effect or a stimulus is never completely predictable, especially when inserted among other more specifically semiotic dements within a text u a pseudo-sign. Suppose lhlt a speaker Is daborating a persuasive discoune according to the rules orjudiciary rhetoric and trying to arouse in his addressees reellnp or pity and compassion. He can utter his phrues in a throbbing voice, or with barely detectable vibrations that could suggest that he is templed to cry. The,e supra-segmental reatures could obviously be either paralinguistic devices or mere symptoms indicating his emolional state; but !hey might also be slimuli he inserts into the discourse in order to provoke some degree or identification in his listeners and to pull them toward the same emotional state. He is using these devices as programmd stimulations but docs nol know exactly how they will be received, detected, interpreted. The speaker is thus hair way between the execulion or certain Nies or stimulation and the displaying or new unconventlonaliud elements that might (or might not) become recognized u semiolic devices. Sometimes the speaker is not sure of the relation between a given stimulus and a given presupposed response, and he is more making than performing a tentative coding of programmed 11imuli. Therefore these devices stand between replica and invention; and may or may not be semiotic devices, thus constiluling a son of ambiguous threshold. So that even though the expressive string of programmed stimuli can be analyzed into detectable units, the corresponding content remains a nebula-like conceptual or behavioral 'discoune'. The exprmion, made of analyzable and replicable units (governed by a ratio /aci/is)may then generate a vague discourse on the content plane. Among such programmed stimuli one might list: (i) all the programmed synesthesiae in poetry, music, painting. etc.; (ii) all so-called 'expressive' sip,1, such as those theoriud by artists like Kandinskij, i.e. visual configurations that are conventionally supposed to 'convey' a given feeling directly (force, grace, instability, movement and so on) and that have also been studied by the theorists of Eiir/iilrluni or empathy; insofar u these devices hold a motivated relationship with psychic forces or 'reproduce' physical experiences, they should be dealt with in the paragraph concerning projections (3.6.7.); insofar as they are displayed by a sender who knows their emphatic effect, they are programmed stimulation (and therefore precoded devices) of which, however, the result (on the content plane) is only partially foreseeable; (iii) all production of substitutive stimuli described in 3.S.8.; (rv) many projections, about which more will be said in 3.6.7. Anyway one should carefully distinguish between this sort of pro-

..

-

2A3

�ory ofSi,n Product/on

grammcd stimulus and the more explicidy coded devices used to express emotions, such as body movements. facial expressions, and so on, now ., precisely recorded by the latest researches in kinesics (Ekman 1969) and in p1ralinguistics. • Another kind of spurious semiotic operation is pstudo-combination. The most typical example is an abstract painting or an atonal musical composition. Apparently a Mondrian painting or a Schoenberg composition is perfectly replicable and therefore appears to be composed by systematically combinable unilS. These unilS are not appartndy endowed with meaning but they do follow combinational rules. Nobody can deny that there is an expression system even thouglt the content plane remains. u it were, open to all comm. These examples are lhus more open sigmI/ u,ctures than sign-functions; for this my reason they appear to invite the attribution of a content, thus issuing a sort of interpretive challenge to their addressee (Eco, 1 962). Let us call them visual or musical propositional functions that can only 'wait' to be correlated to a content, each being susceptible of many different correlations. Thus when hearing a post-Webemian sound cluster one dececu the presence of replicable musical units combined in a certain fashion and somelimes one also knows the rule governing this kind of agreption or material events. However, che problem seems to change when one is dealing with • happenings and abstract expressionist painting.,, random music, John C11e's 10 on. In these cases one can speak of cextural clouds which 1.::k any predictable rule. Can one then continue to speak of a pseude> 1 ). In cases of ratio faclliJ mappins presents no problem; • ' I it simply involves Ute reproduclion of a properly using the samesortofm11erialas thal prescribed

..

'

246

A THEORY OF SEMIOTICS

by the type. In the case or a phonm,e the type ml)', for instance, prescribe 'labial+voiced' (thereby Implying: by means or human phonallon), thw establishing how to produce a (bl .

___ ______ J •.� -----....;i�-.... � Tabla 40

• YI

• Y2

X2. ___.____...,____,._ Xi

(

I

TYPE

• Xs

\

• v, /

TOKEN

The notion of mapping is somewhat more problematic in cases of ratio di/ficilis, because the type or a r•tio di/ficilis is a content unit, a sememe, and its properties are semantic marken, and are not in principle linked to any particular expression continuum. So what does one mean by mapping the pertinent properties or a glass or wine within another material so as to produce the recognizable wet imprint or a glass or wine upon a cable? Fonnulaling the question in this way miglu make ror a puuiing answer, but this is because or one's 'rererential' bias. As a mailer or racl the imprinl or a glass or wine does not have to possess the properties or the object «glass or wine• but it does have lo possess those or the cultural unit «imprint or a glass or wine•. And in this case the semantic representation or the entity in question entails no more than four semantic markers, Le. ccircln, creda, «length or the inradius (or diameter)• and «web. To map these muken within another malerial simply means to realize the geometrical and chemical intrrpn111nt1 of the sememes •circle», «red,. •diameter X• and ,web . This done, the mapping p,ocess is complete, and the realization or a token or the content type a comparatively easy matter. In this sense one Cllllnt>I maintain that the imprint of a hare's paw is an iconic feature in the same way as is the im38" of a hare. In the fonner case the content type is culturally established, whereas in the latter one it is not (except in cases or stylization). The only probIm, would appear to be: ill what sense does a circle or a given diameter realized upon a table map the semantic maJkers ccircln and •diameter X.? But on second thoup,ts, that question is not so different rrom asking ill what sense a labial and voiced consonant maps the abstact phonological lype 'labial+voiced' in sound. In the !alter case the answer seems easy enoup, :

71r•ory ofSftn Production

247

there are certain sound parameters which pennit the realization and recognition of the replica (as to how the realization of a parameler is recognizable, this sends us back to basic perceptive requirements tha t. aswas noted in 3.4.7. and 3.4.8., are postulates rather than theorems for a semiotic theory). Thus one need only repeat that (as was underlined in 3.4.l.) various expressions may be realized whether in aa:ordance with spatial parameters ot phonic parameters in order to justify Jisting the replica of a circle in the same theoretical row as the replica of a phoneme. The only difference is that the sound features governing the reproduction of a phoneme are not content marken, while the spatial features governing the reproduction (even if virtual, u In the recognition of imprints) of a geometrical figure are. This - as we have seen - is exactly the difference between ratio facllis and ratio dif/ici/is. Now if one considers Table 39, one nolices that all the cases of ratio dilfici/i1 concern content types in which the most bnportant semantic markers are toposensiti•e, I.e. figural or vectorial properties. This brings us back to the p1oblem outlined in l.7.2.: not every semantic marker can be verbalized. When semantic markers can be verbalized they have undoubtedly acquired a maximum of abstraction; previously culturalized and frequendy expressed through verbal devices, they can even be arbitrarily corrdated with other non-verbal devices (for example a geometrical form in a road signal meaning cstop» ), and through the mediation of verbal habits they can easily be detected. In these cases it is true that, as Barthes and other theorists say, non-verbal semiotic systems rely on the verbal one. But there are markers that cannot be verbalized, at least not completely, so that they cannot be conveyed by a metalinguistic definition verbally expressed. The spatial disposition of the imprint of a hare's paw caMol be verbally meta-described. It is, however, hard to assert that is has no cultural 'existence', and the proof is not in the fact that it can be 'thought' (which would be an extra-semiotic and somewhat mentalistic argument) but the • fact that it can be ;,uerprettd in many ways. For instance one can conceive of an algorythm which, when fed into a plotting machine as input, would produce as Its output a drawing of a hare's paw. The facl that this drawing is more schematlcal than a real Imprint Is a further proof of the present thesis: the cultural notion of such an imprint (a sememe) is neither the same as its perceptual model nor as the corresponding object. The process from perceptual modd to semantic model and from semantic model to an expressive model governed by arotio difficiis,may be represented as in Table 41 .

248

A THEORY OF SEMIOTICS Tabl1 41

Mappi"I by abstraction

Mappi"I by tlmilitudo

) ______ \

•-��---_,.___,...cx2 ».------+----...••/x2 /

x,

PERCEPTUAL MODEL

SEMANTIC MODEL

EXPRESSION

In other words, given •tNrctprual model as a 'dense' representation of a given experience, assigning to the perceived object x the properties x 1 , x2 , x3 1 x. . . . "n • that perceptual model gives rise to a semantic model which pmerves only, let us say, three of the properties of the dense representation. It is not said that all those selected semantic markers are necessarily vorbalizable items; many or them may be toposensilive relationships. At this point it would be possible to expre" this semantic model (a sememe) by means or an expressive device. If the markers of the sememe were non�toposensitive, the correlation content-expression could be an arbitrary one. Since, however, in this case some markers are toposensitive, the correlation is motivated, and must follow in principle the rules governing every type/token-ratio, i.e. rules or transfonnation. Let us now add something about the double mapping outlined in Table 41. The first kind or mapping (from percept to sememe) does not need to be semiotically explained: it follows the rules governing every phenomenon or abstraction - both in conceptual and 'visual' thinking - and is therefore a procedure depending on the mechanisms or human intelligence (which is not to say tha1 even this procedure could not be seen as a semiotic one, but rather that the definition or this problem constitutes one or the 'political' boundaries or semiotics - Ke Introduction). The second kind or mapping should be identical to that which governs the production or a triangle that is Jimilar to another, given certain spatial parameters and conventions (such as that size is inelevant. but sides must be proportional and angles 'equal'). Let us call this procedure a tra11ifonnatio11: "every biunivocal correspondence or points in space is a transformation. What concerns us is the existence of particular transfonnations that leave certain prominent properties or the geometrical entities to which they are applied unchanscd" C• • ). This concept or transformation fits cam or token-to-token

11,eory ofSi,n Production

249

reproduction as well as those of type/token-ratio perfecdy (this being one of the postulates of semiotics). But it also explains cases such as the production (even if virtual) of an imprint, which is why in 3.6.2. even imprints were said to be cases of transformation. But in cases of type/token-ratio, mapping by similitude takes place between an expression type (and thus the model of an object) and an expression token (and thus another physical object). In the case of the imprint, on the other hand, we are considering similitude, establiJhtd between a semantic model and its phylica/ expression. We are once apin concerned with the difference between rotio foci/is and rotio dif{,ci/is. At this point two problems arise: (l) how to 'map' from a content-model into an expressive one, i.e. from a non-physical reality into a physical continuum; (il) how various kinds of mapping may be listed according to a degree of conventionality reached by the content-type and its toposensiti,e complexity. If, in Table 39, imprints (even if accidentally replicated rather than recogniud) were not classified as straightforward transfonnations under the heading of Invent/om, this was for a good reason. In the case of an imprint the content-model already exists. It has, in one way or another, been culturally established. When replicating an imprint one is mapping from something known. And there exist similitude Nies establishing how to embody in a material continuum certain semantic toposensitive properties of a sememe (as in the case of the glass of wine). So the mapping procedure by ftlt/o difficilis in Table 41 ls not so dilferent from that perfonned in cases of rorio facilis (Table 40). This mapping is undoubtedly morillatrd by the sememic representation of the supposed object but is at the same time ruled by mapping conventions. The main problem arises when trying to detenninr how it is possible to map onto an expression continuum the properties of something which (because of its cultural oddity or fonnal complexity) is not

..

yet cu/rurally known.

It must be stressed that one ls not hero concrmed with the representation of a golden mountain, or a mut with ten eyes and seven legs. It is very easy to infer the nature of unknown rlements from the addition of known ones, just as language manages to express unheard-of events by articulating recognizable units. But the real puzzlina problem is not so much how one may represent a man wilh ten eyes and seven legs as why one may vltually represent (and ncop,izt u reprex11ted) 1 given man with two eyes and two lop. How lt is possible to represent I man standing and a lady silting

250

A THEORY OF SEMIOTICS

under a tree, a calm landscape with clouds and a com-field behind lhem, a given light and a given mood - as happens In Gainsborough's Mr. and Mn. AndreWJ? Since 1his com plex conlent is nol a unll but a discount (and lhe painting is not a sign but a tut). and since that content was not previously known by the addressee grasping for the first lime from an expression for which no typt pmiously existed, how is ii possible to define this phenomenon semiolically? The only solu1ion would seem \.. to be thal painting b nol a semiotic phenomenon, because there is nellher pre-eslablished expression nor pre-established content, and thus no correlation between runclives 10 permil signifaca1ion; thus a painling should appear a 'my$lerlous' phenomenon whichposin runc1ives inslead or being posilcd by 1hem. • Neverlheless, if such a phenomenon seems to escape the conelalional deOnilion of sip1-func1ion', it certainly does not escape the basic definition or a sign IS something which rtandJ for some1hingelse: for Gainsbofough's painling is exacUy this, somelhing physically prtunt which conveys 10mething obsent and, in certain cues. could be used in order to mention a state or lhe world. 1

•

3.6.8. lnvenlion as code-malting Wilh lhis exsmple we have arrived 11 1 crilical polnl In the present classiflc11ion or modes or sign-produclion. We now have 10 define I semiotic mode or produclion in which somelhing is mapped from somelhing else which was nol defined and analyzed before the act or mapping 100k place. We are witnessing a case in which a significant convention is posited at the very mcr•tnt in which bolh the runctives of the correlation are invt11ttd. Bui for lhe semiolician lhis lallor definlllon has I ralher familiar ring to II. II curiously resembles lhe problems (so vigorously rejec1ed by 11 leasl lhree generalions or linguislJ) surrounding lhe origins or language or the historical rise of semiotic convenlions. Now if such a problem can be rejec1ed when ii is proposed from an abslracl and roughly archeological poinl or view, it cannol be escaped when approached from lhe viewp oint or I phenomenology or modes or sign· produc1lon. Lei us lherefore assume 1h11 lhe problem or these lransforma­ lions listed IS invenlions and based on I ratio diffici/i, (depending on a loposensilive conlenl model) nises lhe question or the activily or codr­ makin1 (1