The Contribution of Musical Semiotics to the Semiotic Discussion in General

Jean-Jacques Nattiez

While in the first portion of this century semiotic research, essentially theoretic in approach, was the achievement of a few pioneers—Peirce, Saussure, Morris, Hjelmslev—and while it was slow, following the postwar years, at measuring its strength against empirical descriptions and concrete applications, in recent years the number of works placed explicitly under the semiotic banner has suddenly proliferated. The present North American Colloquium is caught in an irreversible movement of institutionalization which follows from this expansion of our discipline and finds expression through the creation of national associations and international meetings.

In the light of the present situation, the author of these lines does not hide his disquiet as these manifestations, far from representing a consensus on the object, the methods, and the limits of semiotics, display the often radically opposed orientations of theories and practices which have little in common aside from the affixed label. We know that it is impossible to list here all the arguments and the intellectual prerequisites which have led us to adopt our present stand. Instead of hoping to convince our reader, we have decided on an approach in which our previous stands, if today superseded, are revealed in the hope that, by laying bare the doubts that we have encountered, our present position may be more easily comprehended.

Since 1967, we have been working on the construction of a musical semiotics^1. and our investigations have always been related to the search for a specificity of semiotics:^2. a general semiotics or a musical semiotics sans rivage would only constitute a scientific imposture. Let us therefore proceed in the manner of negative theology and examine, in turn, what musical semiotics is not. To do this, we will examine critically some of the ideas currently acknowledged in the literature on general semiotics.

Musical semiotics is the study of the signs of music. The strict etymological definition of our discipline—the science of signs—could lead one into thinking that a semiotics of music should seek the types of signs of which music is constituted. This type of investigation is not new: to quote but one example, the Recherches sur l’analogie de la musique avec les arts qui ont pour objet l’imitation du langage is a work by the musicologist Villoteau, who, in 1807, made the distinction between the expressive or imitative-expressive means “which move our soul,” and the “meaningful signs’ which cannot recall to us the idea or the memory of things without recourse to reason and to reflexion” (1807: 32-33).

Thus envisaged, a musical semiotics encounters a difficulty of general semiotics: the latter has never been able to provide any stable and universally accepted definition of the different types of signs. Why? From St-Augustin to Condillac, from Port-Royal to the Encyclopedia, from Peirce to Saussure, the features retained in order to define a semiotic category are not the same. With Saussure, the sign, properly speaking and by opposition to the symbol, is conventional and arbitrary, yet it is possible to dissociate these two aspects and show that there are signs established by convention where the link between signifier and signified is not arbitrary (as is the case of certain roadway signs); moreover, an arbitrary sign is not necessarily unmotivated, as Saussure himself points out: all the words that derive from the same root are, in some way, motivated.

This first difficulty, ascribable to the incapacity of general semiotics to recognize the semiotic difficulties inherent to the constitution of its definitions, invites yet another problem: a semiotic category, although defined with precision in abstracto, will not necessarily designate the same phenomenon in different fields of application. Take for instance the Saussurian symbol: the example, relating to language, given in the Cours, is the onomatopea. To distinguish, in music, between symbolic and non-symbolic facts amounts to a particularly delicate endeavor: the sequence of sounds which imitates the songs of birds, the crashing of waves can be considered symbolic, as can the evocation of movement, of feeling. But are the latter the result of a natural association—therefore conceivably understandable by anybody—or is it something acquired in the frame of a given musical culture? From a typological point of view, these sonic phenomena may be assigned to different categories according to these two alternatives. In fact the entire classificatory issue does little to help us comprehend how these musical references to movements or feelings function. And what more have we gained by calling the Wagnerian leitmotive a signal—after Prieto, for example—when we know that bugle or trumpet calls, which represent another semiotic genre, answer the same criterion of communicational intention?

We think, with Molino (1975:45), that a certain importance must be attributed to Roman Jakobson’s remarks on the famous trichotomy of Peirce: “One of the most important features of the semiotic classification of Peirce resided in the perspicacity with which he has recognized that the difference between the three fundamental classes is nothing more than a difference of place assigned inside an all-relative hierarchy. It is not the absolute presence or absence of similitude or contiguity between the signifier and signified, nor the fact that the usual connection between these two constituents should be of the order of pure fact or of pure institutional order, which is at the basis of the division of the set of signs into icons, indices, symbols, it is only the predominance of one of these factors over the others”(1966:26).^3. As far as music is concerned, Molino has recently provided an illuminating illustration of this remark: “The sonic phenomena produced by music are indeed, at the same time, icons: they can imitate the clamors of the world and evoke them, or be simply the images of our feelings—a long tradition which cannot be so easily dismissed has considered them as such; indices: depending on the case, they may be the cause or the consequence or the simple concomitants of other phenomena which they evoke; symbols: in that they are entities defined and preserved through a social tradition and a consensus which endow them with the right to exist” (1975:45).

A semiotic approach to music made from the standpoint of sign typologies therefore seems ill-fated from the start, but the investigation is not lacking in positive facts:

1)first of all, it shows that the categories of semiotics are themselves symbolic constructions, in the general sense of the term^4. —i.e., an object which refers by association to some categories of thought which are not immediately given;^5. they are therefore appropriate to a semiotics;

2) the features that intervene in the definition of semiotic categories will change with the hierarchical weights conferred upon them by the theories and the various fields of application; it seems just to call such features variables;

3) what is true of the variables of semiotic categories could also be true of musical phenomena;

4) finally, if music can, in turn, be an icon, a signal, a symptom, a symbol, an image, or a sign, it is proof that music is first and foremost a symbolic fact.

If it is true that the sign—an “undefinable” category according to Granger (1971:72)—or, rather, the various types of signs have at least the common feature that they refer to something else, then we can envisage music as a symbolic phenomenon. In such a case, the goal of a musical semiotics is to inventory the types and modalities of symbolic references to which the music gives rise, and to elaborate an appropriate methodology to describe their symbolic functioning.

There cannot be a semiotics of music if musical meanings do not exist. The semiotic character of music has often been given a rather restrictive interpretation: a musical sign exists because it is a two-sided entity (signifier/ signified, expression/content); we must therefore unravel “what music is saying to us,” and if, in the process, music is found to be an asemantic art, then we must concede the impossibility of a musical semiotics.

We believe that such reasoning harbors at least three major inaccuracies:

1) The musical signifier is conceived after the linguistic signifier. Now, perhaps, it is here that a study of music may bring a new element to our knowledge of other domains, linguistic or artistic. As Molino pointed out, “The root of the fallacy is, in fact, to believe that language constitutes the model of all symbolic phenomena. In this, the study of music brings forth a rectification and makes an essential contribution to our knowledge of the symbolic: there is more in the symbolic than just the phantasmal concept” (1975:45). By trying to reduce all forms of meaning to linguistic meaning alone, it is precisely the latter which we forbid ourselves to understand.

Works in experimental psychology have shown that in the musical domain we must take care not to identify the musical signified with the linguistic signified. When, in their experiments, Robert Francès and Michel Imberty ask their subjects to translate into verbal statements what the musical excerpts mean to them, they know quite well that the musical meanings reach their consciousness in the form of vague sensations which the verbal word exceeds, by and large: “While attempting to say what the music just heard means to them, the subjects add to its meaning some additional conceptualized and referenced meanings which exist only in verbal language” (Imberty 1975:91). It is interesting to note that these observations concur with the conclusions reached by René Lindekens from his research on the semiotics of the photographic image (1971). Insofar as the only way to find out how the semantic content of music is perceived is to proceed with verbalization, the musical signified, as such, can never be pinpointed accurately; but we may consider that the statistical character of the experimental methods allows for a good approximation of it.

2) We think that the position criticized here is also erroneous with regard to the conception it implies of the symbolic nature of music. At times music is considered an ineffable language, of divine essence, capable of expressing the inexpressible; at other times it is considered a purely formal game of sorts, and then it is judged capable of references to the exterior world. In order to understand a debate as old as philosophy itself, let us go back to Eduard Hanslick’s famous essay, Vom Musikalisch-Schônen (1854). This author is often invoked when the issue is to deny music any power of evocation: “The beauty of a musical work is specific to music, meaning that it resides in the links between sounds, without any relation to a sphere of foreign extramusical ideas” (1854:10). Now, if we look more closely at the book, we find out that it is an “essay in the reform of musical aesthetics,” in other words a booklet published after Richard Wagner’s Das Kunstwerk der Zukunft; the purpose of Hanslick is normative, he defines what he thinks music must not be: as an art, it must not seek the essence of Beauty in the imitation and evocation of non-musical facts (think of the leitmotives!). But Hanslick does not deny for a moment that music can stimulate in us various impressions or feelings. It could even be argued that he anticipated the experimental approach to musical meaning: “Any feeling provoked by music must certainly be brought back to the manner—special to each feeling—in which the nerves are affected by an acoustic impression” (1854:85).

Also, we must not mistake a particular aesthetic conception, unique to an era or to a philosophy, with the fact that any music, once conceived, perceived, or analyzed, becomes the starting point for a series of symbolic references. Hanslick does not deny that music releases in us all sorts of associations; he asserts that musical Beauty must reside only in the sonic forms, and that the “pure and conscious contemplation of the musical work” will apply to nothing else but to that same formal organization.

Roman Jakobson sees in music a semiotic system in which the “introversive semiosis”—that is, the reference of each sonic element to the other elements to come—predominates over the “extroversive semiosis”—or the referential link with the exterior world (1973:99-100). It is our opinion that the concept of dominance used by Jakobson to characterize the semiotic specificity of poetry, figurative and abstract painting, pure and program music, must be extended to the aesthetic conceptions of these various artistic manifestations: then, the formal and asemantic theories of music proved in the works of Hanslick, Jakobson, Stravinsky, and Hindemith appear to be a cultural fact in which the extroversive semiosis has been minimized with respect to the internal interplay of sonic forms. This has not always been the case in music’s history. When Fontenelle apostrophizes: “Sonate, que me veux-tu?” his question is symptomatic of the disarray in which the theorists of the classical period were thrown when they found themselves confronted with the dilemma of pure instrumental music^6. —and there are societies (like the African Dogon and the Mexican Tepehuas) where music plays the same role as speech in interhuman relations.^7. The symbolic character of music is a semiotic fact which can be ascertained everywhere, but to which the various aesthetic theories, the compositional concepts, or the strategies of perception bring a variable or changing weight, and assign it to various levels in the hierarchy of acoustic components, according to time and culture.

3) The position questioned here thus runs the risk of mistaking semiotics for semantics. If we refuse to identify the musical signified with the linguistic signified used at the time of verbalization, then what is our general concept of meaning?

We will say that an object, whatever (a sentence, a painting, a social conduct, a musical work. . .), takes on a meaning for an individual who perceives it when he relates the object to his experience-domain, or the set of all other objects, concepts, or data of the world which make up all or part of his experience. To be more direct: meanings are created when an object is related to a horizon or a background. Now is the time to use, in perhaps its more fruitful manner, the theoretic contribution of Peirce: in his Essay on a Philosophy of Style (1968:114), Granger describes this phenomenon by means of Peirce's semiotic triangle, which he schematizes as shown in Fig. 1.

Figure 1

The sign S refers to an object O by means of an intermediate and infinite chain of interpretants I. These interpretants are the “atoms” of meaning by which we exercise our symbolic relation with the world.

The conceptual and verbalized meanings released in us upon listening to music are but a part of the interprétants associated with it. It is convenient here to use the expression “musical semantics” to describe the study of these interprétants are weighted. The objective of a musical semiotics is therefore sical experience—so long as we do not forget that a work of art is susceptible of many other types of references: when we analyze a musical score we favor such or such an aspect of the musical organization, we select this interprétant instead of that.^8. If we examine the Jakobsonian idea of a “musical introversive semiosis,” a particularly positive idea will impose itself: when we establish relations of identity, equivalence, or contrast between the internal sonic phenomena of a work we are indeed calling into play a semiotic process—that is, an organization preferred by the composer, or by the performer, or by the listener, or by the musicologist, depending on the way in which certain interpretants are weighted. The objective of a musical semiotics is therefore not only to describe the semantic references stricto sensu, but to give an account of any type of symbolic association that takes place with the musical material.

Notwithstanding all of this, the fact remains that musical semantics represents an appreciable portion of the semiotic program in our domain. The perspectives in semantic research fall into three main categories.

The first of these we will call, after Spinoza, Granger, and Ricoeur, musical “hermeneutics.” It is the interpretation of the meanings obtained from listening to music, the exegetic inspection of the proliferation of the interpretants, the establishment of subjective relations between the musical and individual experiences. As an example, think of the beautiful studies of Jankélévitch on Debussy (1968).

The second is musicological reconstruction. It deals with the construction of the relations between the meanings induced from the texts or situations in vocal or dramatic music, and the musical forms chosen. Such analyses tell us nothing about how the works are perceived; the musical lexicons developed in Lavignac’sLe voyage artistique à Bayreuth (1897) or Chailley’s Passions de Bach (1963) are typical of a composer’s fancy. This type of research can be subdivided into more classes according to the methodology in use: the old studies of Pirro (1907), Schweitzer (1937), Lavignac, and more recently of Chailley, have in common that their authors conduct the analyses by going from the meanings to their musical embodiment; but it also proved necessary to proceed in the inverse manner, by going from signifier to signified, as in the works of Norma McLeod (1971) and Fritz Noske (1970). A synthesis of these onomasiological and semasiological orientations was proposed by Charles Boilès in his studies on Otomi music (1969). He later summarized the procedure as follows: “Two types of groupings were studied. The first grouping was arranged in sets according to designata common to all tunes placed in any given set. The other groupings were arranged in sets according to any musical similarities that seemed to be common to any set. Analysis of each group of tunes having common designata revealed certain phenomena that also occurred with great frequency in those tunes. The good fortune to discover musical vehicles that could be related to designata was further supported by results obtained from the second group” (1973:39).

The last category is experimental psychology, to which we have already alluded; for Francès (1958) and Imberty (1975) it consists, among other things, in preferring free induction over guided induction in experiments on verbalization. This perspective has often been the object of some rather violent criticism: either all the interest—let alone its existence—of a musical semantics is denied, or else the claim is that it only leads to some already known results. But no experiment has ever brought out total agreement in the answers obtained, and for this reason the individual testimony, profound, rich, and subdued though it may be, cannot present the same scientific warranties as an experiment, however regional and limited.

Its overall epistemological validity is not a safeguard against some serious difficulties. The inquiries are usually conducted over fragments of musical works, some very short in duration, in other words artifacts extracted from the musical flux: we fear that the answers obtained will not correspond to a true listening situation, and as far as we can detect, there is still no complete experimental semantic analysis of a complete work.

In these researches, the correspondence between the meanings presented and the musical material as such is never abandoned, but it is an aspect which has never been tackled systematically. The works in experimental psychology have at least established an important principle: “In the end, the features of the musical structures are pertinent to one determined (semantic) factor only, although they may also be observed in others” (Imberty 1970:92). This means that there is no stable relation between a particular musical feature and a given meaning; the relation is always the result of some complex combinations of variables.

A semiotics of music is constructed with the help of linguistic models. During the 1960s and especially in Europe, the semiotic project, reactivated most visibly by Barthes in 1964—even if other scholars (Buyssens, Prieto, Mounin) were already working on it, albeit more discreetly—manifested itself in joint relation to linguistics. Today, it is not at all obvious that linguistics should be a determining factor in the construction of any particular semiotics, and in order to understand how this pertains to musical semiotics, it may be useful to go back in time.

It is quite independently from Peirce that Saussure places the semiotic project in line with semiotics with a few famous quotes from his Cours de linguistique générale. But we must not forget that when he writes: “Linguistics can become the master-pattern of all semiotics, although language is only a particular system” (1922:101), it is because he envisages a synchronic linguistics which depends on the arbitrary character of the linguistic signs, and because, according to him, semiotics should first of all attend to the arbitrary signs.

From Sechehaye, an authorized interpreter of Saussure since he was an editor of the Cours, we can see in the C.L.G. a true course in general semiotics: “Language is only a particular case—albeit perhaps the most important—of a general case, and the problems concerning it must be considered before all as problems of semiotics (....) What is the special character of semiotics?.. . Any semiotics is essentially a science of values” (1917:13-14). Now this is the same notion of value^9. which, in the Cours, informs all the famous dichotomies: internal/external linguistics, signifier/signified, synchrony/diachrony, language/speech, syntagm/paradigm (Molino 1969: 341).

This interpretation of the Cours is similar to the semiotic conception of Hjelmslev. For him, the concepts of the Prolegomena to a Theory of Language “have a universal character and are valid for the system of signs in general (or for the system of figures which serve to form the signs)” (1943:130). For Hjelmslev, “language” is not limited to human language; it includes the set of all languages.^10.

With the publication of No. 4 of Communication (1964), Roland Barthes adds a new argument: semiotics constructs itself from linguistics because, in the past half century, the latter has developed tremendously while the former has remained essentially programmatic. The 1962 position of Mounin is more or less identical: “Even if not formally constituted, this semiotics of the future delimits and defines itself little by little with respect to the notional discoveries of recent linguistics. And as the latter brings into display, scientifically, the defining characters of natural languages, we can verify whether these same characters are valid or not with respect to the definition of systems of signs other than natural languages” (1970:68)—although an essential difference remains: in order to justify “the reversal of the Saussurian conception,” Barthes does not hesitate to consider a priori all the systems of signs as languages, whereas, with Mounin, this comparative work is “propaedeutical,” or preparatory to assessing the true importance of the functional linguistic model. But in both cases the objective is semiotic: the objects studied by Barthes are languages constituted of two-sided signs, those of Mounin are signals produced with an intent to communicate.

The question is whether, in this recourse to linguistics, the semiotic character of the object has been preserved. In much of the European research in semiotics, including our own articles on musical semiotics up to 1973, it seemed that the fact of using the abstract categories elaborated by linguistics in order to describe languages was judged enough to turn any object into a language. We thus ran the “ontological” risk, so well denounced by François Latraverse, which “consists of inferring the worth of a method from the nature of what it manipulates” (1974:70).

Gino Stefani’s statement at the Belgrade Congress on musical semiotics in 1973, which bore a remarkable similarity to our thesis, defended in the same year, seems significant of this first stage in the development of musical semiotics: “The major influence on the new semiotic current in musicology can be imputed, up to now, to structural linguistics” (1975:12). “The more consistent contributions to a semiotics of music come from the application to music of linguistic methods and perspectives” (1975:13). Now, at the same time that general semiotics appealed to linguistics in order to free itself from the theoretical state in which it found itself, the linguistic models available in 1964 were perhaps not necessarily the best ones with which to analyze the linguistic and non-linguistic domains as symbolic facts. When Barthes asserted that “the development of mass media communications, today, contributes greatly to an actualization of this vast field of meaning, at the very time when the success of disciplines like linguistics, information theory, formal logic and structural anthropology provide semantic analysis ^11. with new tools” (1964a: 1), he was only deluding himself since:

a) While the concepts of signifier and signified, expression and content, can be used to operate a distinction in the domain, they are of little help in conducting a conclusive analysis.

b) The concepts of denotation and connotation, in the interpretation of Barthes, led him to a form of social psychoanalysis which certainly could do without linguistic concepts.^12.

c) The preferred linguistic model at the time in France was phonology, a model which did little to describe a particularly symbolic aspect of language.

d) As for information theory, Bar-Hillel has shown that it is inadequate to solve problems of semantics.

e) As for structural anthropology, today we can quote Dan Sperber: “In structural research the symbolic signifier, freed from the signified, is not much of a signifier except by a doubtful metaphor whose only merit is to elude the problem of the nature of symbolism, not to solve it” (1974:64). It seems difficult today to maintain an “identity of perspective between structuralism and semiotics” (Tremblay-Querido 1973:9).

f) There remains the question of the decomposition of the object studied according to the paradigmatic and syntagmatic axes: “These operations,” Barthes claims, “represent an important aspect of the semiotic enterprise” (1964b: 109). “Essentially, a semiotic analysis consists of distributing the facts inventoried according to these two axes” (1964b: 116). Indeed, we think that this operation brings to light some fundamentally semiotic phenomena— to which we shall later return—but if we re-read the context of these same two lines in Eléments de sémiologie, we realize that the semiotic justification is provided by recourse to the Saussurian model, not by the symbolic aspect which it brings into focus.

Then, the ontological illusion may well be succeeded by a sort of scientific illusion: the use of linguistics is no longer justified on the grounds that it effects a direct link with a semiotic fact, but because it fills a void (Barthes) and figures as a pilot science for the human sciences. In 1973, Stefani could say: “Any work on musical texts does not automatically belong to the realm of semiotics. Without mentioning other characters we feel that a dominant feature of our research is indeed the rigor which contributes to making semiotics into a science” (1975:9-10). We do not complain, but rigor alone cannot transform any musicological discourse into a semiotic object unless we identify science with semiotics, as Gardin suggested after Morris (1975:75) and as we thought for a time.^13. But in that case, as Ruwet pointed out (1975:33), there is no need for the term semiotics. In fact, there is a semiotic process in every scientific description (as we will show about music), but no science identifies itself with semiotics for that reason. The semiotic character must be thematized: this small difference is fundamental.

Now that the “dissemination” of semiotics is, today, at a maximum, a history of semiotic theories and practices should evaluate, for all the different domains, the manner in which the symbolic character of the objects studied has been brought to light and evidenced by recourse to linguistic models. Quite possibly, as we will see, such a venture may take place despite all facts and not because a linguistic model has been used. Today, the divorce between linguistics and semiotics is consummated; it is only unfortunate that this separation was made on account of fashion—structuralism, said Lévi-Strauss, died in 1968—and not for epistemological reasons. Whereas in 1966 Nicolas Ruwet could write: “I will treat music as a semiotic system which shares a certain number of common features—such as the existence of a syntax—with language and other systems of signs” (1972:100), a little less than 10 years thereafter he does not hesitate to say: “I don’t particularly like the expression ‘musical semiotics,’ which I consider useless and perhaps even dangerous” (1975:33). This last assertion is undoubtedly excessive and unjust, but we ourselves have experienced this separation: when we set out to work, in 1967, by far the most vigorous example was the beautiful article by Christian Metz, “Le cinéma, langue ou langage?” First of all, we asked ourselves: “Is music a language?”—which is effectively a semiotic question—and this in turn led us to compare music and language; this “comparative semiotics,” as we proposed calling it, thus served as a propedeutical or preparatory basis to control the transplantation, into musicology, of the linguistic models.

Later on, we will show explicitly how the use of a particular model can contribute to the specifically semiotic description of a musical work. Insofar as these borrowings have been more or less closely related to the semiotic project—at least in Europe^14. —our discipline can certainly benefit from their positive contributions:

1.) The comparison between music and language. On this subject, both the scattered (Dufrenne 1967, Schaeffer 1966) and the systematic (Springer 1956, Nattiez 1975e:2nd part) remarks available today favor a better understanding of the specificity of music and language. Besides the fact that it serves as a basis for the importation of linguistic models (Nettl 1958, Bright 1963, Becker 1973), this comparison contributes to a new form of classification of the Beaux-Arts (Nattiez 1973b: 184). With the development of a systematic knowledge of music, we can hope, as Molino once suggested (1975:59), that the comparison will no longer be conducted one-way only (from language to music) but will also proceed from music to language. For example, it seems that what has been said of musical meaning could help examine in a different light the problem of linguistic semantics; of course, musical meaning is more connotative than denotative when compared to verbal meaning, but if we consider only the “core,” or the “cognitive” part of meaning, the one which lexicologists set up with degrees of variations from one dictionary to the next, thereby indicating the existence of a certain laxity, do we not prevent ourselves from grasping the specificity of verbal communication, which, more often than not, remains an absence of communicaton?

2.) The functional models. First, the application of the functional models: given the analogy between the note and the phoneme brought about by comparative semiotics (Jakobson 1936), some phonological principles may be used in order to reconstitute extra-occidental music systems (Nettl 1958; Bright 1963), or to analyze the structure of a dance (Kaeppler 1972); up to this date, the best elaborated methodology is that of Vida Chenoweth (1973), based on Pike’s tagmemic model. We agree with Stefani (1975:13) that the attempts to apply the Prague model to the analysis of musical works (Mâche 1971) remain rather vague.

Next, the functional point of view, in a more general way: this includes the attempts to apply Jakobson’s grid of linguistic functions (1963) to music (Lévi-Strauss 1964, Stefani 1969, 1972), as well as what Stefani (1975:13) aptly calls the socio-cultural functionalism (Bogatyrev 1936, Kluge 1967, Sychra 1948), where the influence of Mukarovsky’s functional aesthetics prevails. Finally, although their reference to the concepts of the Prague School is not always explicit, some American works (McLeod 1971, Boilès 1973) also enter this category.

3.) The distributional model. From the linguistics of Harris it can be said that Ruwet drew the fundamental principle of making explicit the analytical criteria (1966) and the idea of a taxonomic classification of a musical work (1962, 1966, 1967), which he relates to the paradigmatic procedures of analysis used by Lévi-Strauss in anthropology (1958) and by Jakobson in poetry (1963). Here we must distinguish between the parsing techniques, which are of linguistic origin, and the properly structural treatment of the units thereby obtained, which is of Lévi-Straussian inspiration—this being especially obvious in the analysis of the Prélude to Pèlléas (but also in the recent analysis, by Lévi-Strauss, of Ravel’s Boléro [1971:590-96]).

When we apply Ruwet’s method of analysis to the analysis of monodies, we engage in the parsing of the musical syntagm into units which are paradigmatically associated, if identical or transformed according to a series of well-defined rules. If we look at it more closely, it will be seen that the procedure raises a number of questions, which have been tackled in various critical contributions (Arom 1969, Nattiez 1974c, Lidov 1975); these testify to the interest of the method. Being especially concerned with the question of making explicit every analytical gesture or criterion, we have developed this first aspect of Ruwet's enterprise^15. by applying it to the analysis of Intégrales by Varèse, the Intermezzo Op. 119 No. 3 by Brahms, Syrinx by Debussy (Nattiez 1975e: part 3, chapter II), and to Density 21.5 by Varèse (1975d). The same perspective can also be ascertained in the works of Arom (1969, 1970) and Levy (1975) on non-European music, Stefani (1973) and Morin (1976) on J. S. Bach, Guertin on Messiaen (1975), and Hirbour-Paquette on Debussy (1975).

4.) The generative-transformational model. If we except a recent study, still unpublished, on the tonal harmonic system (Jackendoff and Lerdhal 1975) and a grammar of the melodic aspect of J. S. Bach chorales (Baroni Jacoboni 1975), most applications of the Chomskian model are restricted to the analysis of folk and non-European music. We cannot fail, however, to note the number of works which indulge in the metaphorical use of certain Chomskian concepts: deep and surface structures (Blacking 1971), transformations (Herndon 1975, Cooper 1973), phrase or kernel sentences (Bent 1974:43, Treitler 1974:66). Generally, the generative ventures fan out in two directions: for generating pieces that belong to a given style (Sapir 1969; Lindblom and Sundberg 1970, 1973, 1975; Sundberg 1975; Asch 1972; Becker 1974), or to formalize musical theories verified according to an hypothetico-deductive procedure (Lidov 1975, Ruwet 1975).

Since the time of the first articles by Ruwet (1962), and given the influence of linguistics since then, we can consider that an autonomous current of research and musical analysis has abundantly developed, representing an original perspective in contemporary musicology.

A semiotics of music must be based on the acknowledgment of the symbolic character of the musical phenomena. It is still largely questionable whether the use of such linguistic models fosters a specifically semiotic approach to musical works, however. Speaking about “examples of semiotic analyses” Stefani specifies “analyses that are based on making explicit criteria rigorously adhered to” (1975a: 15). If the explication of the criteria is indeed a necessary condition of scientific seriousness, is it enough to make semiotic objects out of musical works?

In the first paragraph, our conclusion was that a sign, or a symbolic fact, exists when there is some kind of reference or “renvoi.” The very fact that a work or a human conduct becomes the object of an analysis implies that we have associated it with a certain number of interprétants. Simply put, the analyst is a specially attentive manipulator of interprétants, but one who, operating in the isolation of his workroom, should in principle be conscious of the fact that the scientific prehension of the object studied is not immediate, that in the passage from the work to its meta-language many aspects have been left aside. The composer, the performer, the listener also organize for themselves the interprétants of the work which is created, played, or listened to in a specific manner, and this without their being necessarily aware that a symbolic process is at work.

By making explicit the criteria of a musical analysis, the latter is transformed into a semiotic enterprise from the fact that it now enables us to describe and show, through such an analysis, just which interprétants have been retained, and how.

But the same enterprise is also semiotic in another way: created by an individual, the work is transmitted—at least in occidental music—by means of a written document, the musical score, to a performer who interprets it, translates it for the benefit of listeners. In this way, “a network of exchanges takes place between individuals” (Molino): the work cannot be dissociated from the one who produces it and the one who receives it. This is why “it is impossible to analyze, to reduce to units and organize a symbolic ‘object’ without going back to the three dimensions which the object necessarily presents” (Molino 1975:47). Molino calls these three aspects, after Gilson and Valery’s terminology, the p oie tic (production pole), the esthesic (pole of reception), and the neutral (level of the material object, as heard and produced) aspects of a work.

As soon as an analysis explicates its own criteria, it cannot fail to encounter these three dimensions, because the reasons for considering particular units of a musical work to be paradigmatically equivalent are based on a phenomenon of perceptive association, on a knowledge of the equivalences allowed by the composer, or on both at the same time. One may wonder, then, about the necessity for a “neutral” description of the object, especially when the analysis is being referred constantly to the two extreme poles of the trichotomy. In fact, the poietic and the esthesic dimensions are not necessarily bound to correspond, and it is one of the contemporary myths of semiotics to conceive as normal a state of equilibrium between both. The aim of a neutral level is to inventory, as exhaustively as possible, the set of all possible configurations: bear in mind that a work is never perceived in the same manner from one time to the next, let alone by two individuals at the same time.

In this sense, Ruwet’s principle of paradigmatic parsing, based on the dialectic repetition/transformation, is semiotic, less because it is inspired by linguistic procedures than because it helps inventory all possible relations between the units of the same piece. When he stressed the “impossibility of representing the structure of a musical piece by means of a unique diagram” (Ruwet 1972:134), he was not only raising a technical and material problem of analysis, but also showing that the multiplicity (properly infinite) of interprétants attached to a work forces us to present many different configurations or possible modes of organization of that work, according to the explicit criteria that have been defined.

Hence, when Stefani cautions us in a recent article (1976:49) to adhere to a monoplanar semiotics and deal only with the signifier, thereby leaving semiotics aside, he is only the victim of a semiotic concept still confined in the scheme expression/content, and even if the functional aspects of his analyses are not exactly comparable to concepts, they are not very far indeed from the Saussurian signified. Now, we think that Peirce’s semiotic triangle breaks up the Saussurian dichotomy since, in the process of apprehending symbolically the musical material, the classical signifier is itself sprinkled with interpretants.

We will now give an example (Fig. 2) of the necessity for examining an object from the point of view of the three dimensions. The example is borrowed from an article by one of our collaborators, Gilles Naud, on Xenakis’ composition: Nomos Alpha. The work was selected because Xenakis has published the mathematical program which served to compose the work. The following passage (Naud 1975:71) is poietically divided by Xenakis into two objects: s2a and s3a. The neutral analysis, conducted on the basis of paradigmatic principles, identifies three distinct objects: α, β, and γ. The (esthesic) experiments on the auditive perception of this piece show that the same passage was categorized by a first category of listeners into objects (1), (2), and (3), which concur with the neutral segmentations. A second group of listeners identified objects (4) and (5), which coincide with the poietic data; a third group of listeners identified (6) only, and thus did not distinguish between objects α, β, and γ. Had we preoccupied ourselves with only the poietic data, it would have been impossible to explain groups I and III of listeners, and a minimal neutral segmentation is also necessary in order to explain group I’s perception.

This theory of the neutral level has met with much opposition, and, surprisingly, on behalf of researchers who should appreciate its necessity. During the past fifty years, linguistics has multiplied the modalities of an immanent description of languages. What happens today, after so many years of neutral descriptions, is that we can no longer remain at a stage where the phenomena are tackled “in themselves and for themselves,” as Saussure would have it: they must now be envisaged in their relation to the poietic and the esthesic poles. An essential difference, however, with the reducing attempts (social character, psychology of the sign) denounced by Saussure in a famous page of his Cours, is that the only way to relate a poietic and an esthesic characterization to the works studied is through a neutral and immanent description of the object—this has been the fundamental attainment of structuralism—and for two principal reasons:

1) the musical works have an autonomous material reality which cannot be reduced to the poietic and esthesic aspects;

2) the links between the works and the strategies of production and of reception cannot be specified unless we first dispose of a description of the work itself.

Now, we may assign to their proper place in our semiotics the techniques imported from linguistics into musicology—not to mention the semiotic character of the inventories of paradigmatic equivalences: they provide models for inventorying, classifying, and organizing the constitutive elements of a work on the neutral level, such that the latter may in turn supply a basis from which the relations to the poietic and the esthesic dimensions can be effected. The external data supplied at these two extreme poles may then reilluminate or organize differently the combination proposed at the neutral level. It is easy to see that a work of art is never static, but is part of an unending process: and a semiotics of music must especially endeavor to bring out this symbolic dynamism, this swirl of interpretants.