“It is equally fatal intellectually to have a system and to have none. One must decide to combine both.”

—F. von Schlegel

Heinrich Schenker claimed that if you did not hear music according to his system, you could not be said to hear it at all. Moreover, his system was not elaborated with much consideration for more traditional ways of looking at, or listening to, music. He swept away as trivial and insignificant not only such notions as “modulation” and “sequence” but even “melody,” the common man’s way of recognizing and appreciating music.1 Schenker’s contempt for the layman is exceeded only by his contempt for all previous theoretical work before his own except that of Karl Philipp Emanuel Bach.

Schenker was the musical heir of the great Romantic literary critics of the early nineteenth century, like Friedrich von Schlegel, who conceived the task of the critic as being to convey the unity of the work of art. At the time of his death in Vienna in 1935 at the age of seventy-eight, he was ignored by most of the world of music except for a small group of distinguished pupils and admirers. Before Schenker, the analysis of a musical work was largely an articulation of its parts. Even today the most common method is still to identify the succession of themes and to note which ones appear more than once. A more technical analysis may articulate the harmonic scheme, listing the different keys to which the music moves and their relation to the main key of the piece (the tonic).

Schenker tried instead to show not how the piece may be divided up, but how it held together. A beginning was made toward this end in music criticism as early as E. T. A. Hoffmann, who observed how a work of Beethoven seemed to derive from a single motif, and traced this technique of composition back to Haydn and Mozart. For later nineteenth- and twentieth-century critics, however, analysis of motifs became only a new way of atomizing a work of music, and matters remained at this relatively primitive level until Schenker revived Romantic aesthetics and combined it with an anticipation of certain aspects of structuralism.

Schenker’s analyses contain the most important and illuminating observations made in this century about the music written between 1700 and 1880. Until now only the earliest and weakest of his books, a treatise on harmony, was available in English. A translation of the central theoretical work, Der Freie Satz, has never been published, nor are there any English versions of his analyses of Beethoven’s Third, Fifth, and Ninth Symphonies and the last piano sonatas.

For this reason two recent publications are welcome: the Five Graphic Music Analyses and, in Music Forum, Vol. II, a discussion of the Saraband of Bach’s C Major Cello Suite, one of the essays (and by no means the most significant one) from the volumes called Der Meisterwerk in der Musik.2 The Five Graphic Music Analyses is particularly important, although—since the book contains almost no text at all—it can hardly be said to make a beginning with the task of translating Schenker into English. These reductions of music by Bach, Haydn, and Chopin to skeletal graphs are Schenker’s last works.

In his excellent Introduction Felix Salzer maintains that, although Schenker was a specialist in eighteenth- and nineteenth-century music, his theory has much to contribute to the understanding of other periods as well. Schenker, however, was no mere specialist in these two centuries, but a firm believer that musical art of any consequence was confined to that period, when a developed and sophisticated form of tonality was the basis of music.

His book on Beethoven’s Ninth Symphony was dedicated to “Brahms, the last great master of German music.” “German” was an unnecessary qualification for Schenker, who considered the ability to sustain musical expression a proof of membership in the German race, even if one had foreign blood in one’s veins. Chopin, whom Schenker used to illustrate his theories almost as often as Beethoven, would, I presume, be an honorary German. This is like Schlegel’s “They say that the Germans are the greatest people in the world for their sense of art and scientific spirit: no doubt, but there are very few Germans.”

That the music after Brahms did not fit Schenker’s theories was only a proof to him of its inferiority. Stravinsky and Reger are both easily disposed of in this way and outside tonality was the outer darkness into which the degenerate composer was forever consigned.3

In Schenkerian analysis, every work of music is reduced to a simple line which is a step-by-step descent to the central or tonic note, and under each note of the line the harmonic functions are indicated by a bass. This line always outlines one of the intervals of the tonic triad (third, fifth, or octave). (In C Major, for example, the line may descend from E to C or from G to C; the octave descent C to C is rarely encountered.)


The fundamental line constitutes, for Schenker, the structure of every tonal work at the deepest level, and music that cannot be reduced to this structure must be judged incoherent and, indeed, ungrammatical. The “idea” of each work is not, emphatically, this fundamental line, but the elaboration of the line into the rich and individual superstructure that we actually hear. It is implied by all of Schenker’s writings that only genius can arrive at a musical work that is both grammatical and interesting; and within the terms set by Schenker himself, this is an inescapable conclusion.

What Schenker did was to extend the idea of dissonance from the individual moment to the level of the piece as a whole. Dissonance is simply an interval that requires resolution into a consonance, and the only consonances accepted in Western music since the fifteenth century are the intervals of the basic triad (third, perfect fifth, and octave) and the inversion of the third, or the sixth.4 All other intervals are, by convention, dissonant, and demand to be resolved into one of the consonances.

This concept was already extended in the eighteenth century when the chord and not the interval became the basis of harmonic thought. Dissonance now implied resolution into a triad, and the final resolution of every work of music was, of course, into the tonic triad.

The basis of Schenker’s system is that every note of a piece, whatever its immediate function, is considered as dissonant to the notes of this final, tonic cadence (except, naturally, for the notes of the cadence itself). Each note has therefore ultimately to be resolved into the tonic triad. An unresolved note is considered as in suspense, the tension it creates lasting until its resolving note finally appears in a context that emphatically displays its role in the large plan. The context is defined by the harmonic significance at each point of the basic line.

What is most striking about Schenker’s analytical system is his insistence that both listener’ and composer—consciously or unconsciously—have a sense of tonal forces that overrides the immediate, small-scale event and allows them to hear “at a distance” so to speak. For example, the basic phrase (Ursatz) underlying the whole of Chopin’s Etude in F Major, opus 10, no. 8, is:

Resolution at a distance would require enormous space to illustrate properly, and the shortest example will have to suffice. Measures 10 to 15 of the original are represented in the most complex of the series of Schenker’s analytic graphs by:

We can see here not only all the notes of the original resolved into the basic phrase (or its octave doublings), but also the relation of the high G of measure 11 to the high A that occurs four measures later. This is a relationship that a pianist with a sense of line naturally sets in relief, and is a direct part of musical experience. The large-scale resolutions take place according to the strict rules of counterpoint derived from the practice of J. S. Bach.

Schenker assumed that the contrapuntal technique of voice-leading (in which each note is part of an independent vocal or instrumental “horizontal” line in addition to combining into a simultaneous harmony with other notes) was valid not only on the level of the single phrase but underlay the general harmonic structure as well. He found that he could connect what he considered the basic notes at the points of structural importance, and that they formed a series of lines that conformed to the tradition of voice-leading as it had been elaborated in the fifteenth and sixteenth centuries, and systematized in the eighteenth.

These lines constitute the most complex of Schenker’s analytic graphs, the Urlinie Tafel. These in turn can be reduced in a series of stages to the simple cadential phrase outlining the tonic triad. This latter phrase is not to be understood as the structure of the work being analyzed, but as the structure of the tonal language.

In other words, for Schenker every tonal work is the elaboration of a simple cadence. Historically this point has much to be said for it (although Schenker’s way of thinking was preeminently anti-historical). The cadence is the determining element in Western music, at least from Gregorian chant until the early twentieth century. Not only classical tonality but the medieval modes are defined by the cadence, and the basic impulsive force of both Renaissance and Baroque music—the harmonic sequence—is generally a repetition of cadential formulas. The cadence is a framing device, and it isolates and defines a piece of Western music as the frame defines a Western painting. Unlike much of the music of Africa and Asia—and much of what is being written today from John Cage to rock—a work of European music from the twelfth to the twentieth centuries is conceived as a determinate isolated event, and the cadence fixes each performance in time.


If a work is essentially a cadence magnificently expanded, then it may be seen as a delaying action, or, in Schenker’s own terms, a tension sustained until the final resolution. What was original in Schenker’s approach was his insistence that the means of sustaining the tension be intimately related in all details to the simple cadence which defines the work. In this way, he was able to explain that sense of unity and integrity of the great works of eighteenth- and nineteenth-century music. A quartet of Mozart holds all its most violent and dramatic contrasts in one characteristic whole, while a quartet by Dittersdorf—with much more uniform material and texture—falls into a series of separate sections, jolly and tuneful as they may be.

It is a waste of time to ask if this unity that we seem to perceive really exists, or if the composer knew that he created this unity, whether or not he was able to put his awareness into words. These are not, of course, answerable questions even if the composer is on hand to incriminate himself. The unity of a work of art is the oldest critical dogma that we have, and every piece of music demands a perception of its unity in the absolute sense that that is precisely what listening to it means. That is, the unity is neither an attribute of the work nor a subjective impression of the listener. It is a condition of understanding: the work reveals its significance to those who listen as if even its discontinuities correspond to hold it together.

Some sort of symmetrical correspondence between detail and large structure is therefore integral to all Western art, and Schenker locates the basic correspondence in music between those tonal forces that enable a listener without perfect pitch to realize that the original key has returned and the rules of counterpoint that govern the individual phrase. Bach did conceive a saraband as an immense, single phrase, and Haydn could, indeed, think musically at a much longer range than any of his contemporaries except Mozart.

The reduction of a Chopin Etude to a simple contrapuntal graph that often resembles nothing so much as a phrase of Bach is not surprising when we remember how much Chopin revered Bach and how much he depended on him for his conceptions of harmony and form. Schenker considered tonality as God-given and established for all eternity (so much the worse for those heathen on other continents beyond the pale of revelation), but this should not obscure his discovery of how sensitive the greatest composers were to the implications of their musical language.

The limitation of Schenker’s approach is as evident as its cogency (although most musicians are capable of appreciating only one or the other, and discussion is blindly partisan). The most crippling omission is the rhythmic aspect: harmony and melody can be entirely reduced, convincingly, if at times somewhat speciously, to the simpler lines of the Urlinie, but rhythm plays not only a subordinate role but often none at all. This falsifies the musical thought of all the works with which Schenker dealt, most evidently those of Beethoven.

Resulting logically from the neglect of rhythm is the total disregard of proportions. Schoenberg once looked at Schenker’s graph of the Eroica, and said, “But where are my favorite passages? Ah, there they are, in those tiny notes.”5 It is not merely that one note of Schenker’s basic line may last one second and another a full minute in the complete piece, but that Schenker often minimizes the salient features of a work. This implies that there are important forces in musical composition that Schenker takes no account of, and they may often supersede the aspects of music with which he is concerned.

Basically these limitations arise because his thought—his way of looking at music—was ruthlessly linear, and while this may be an understandable and fitting mode for an age that saw the development of the twelve-tone technique and the serial music that Schenker so hated, it radically distorts the music of any period, including Schenker’s own. His neglect of rhythm, too, has only accentuated the nonsensical separation in theory of the elements of music, as if a tonal melody could exist without a rhythmic contour.

These limitations, however, would not account for the extraordinary distaste that his work often provokes, chiefly among musicologists, and for the absurd disregard of his achievements that still lames critical writing on music. Nor would his manner of writing, brutally and repulsively arrogant, and his insistent German chauvinism explain why work of that importance should have remained relatively unknown and ignored except by an unhappy few during Schenker’s lifetime, and should continue to excite hostility during the more than thirty years since his death in 1935.

It is hard to take offense at Schenker’s suggestion that a new Beethoven would have to appear among the Germans just as Nature places elephants and crocodiles only where they will find the conditions to sustain life. Absurdities like this have been excised from the more recent edition of Der Freie Satz, and Schenker’s disciples tend to gloss over, and even dismiss, these aspects of his thought, but a true Schenkerite ought to insist that Schenker’s work hangs together as an organic whole. The paranoiac style of Schenker’s essays along with his insistence that his theory was a form of monotheism does reflect an essential characteristic of his ideas.

Schenker’s method is the uncovering of a hidden and secret form underlying the explicit one. He does not deny that the explicit forms exist (sonata form, rondo form, ternary form are almost as real for him as they were for d’Indy); he denies their importance. The implicit form is the only one that brings salvation, and it alone reveals the way the music was composed. The explicit form was imposed almost as an afterthought.

This absolute rejection of the explicit in favor of the implicit is a classic method of interpretation: the Marxian analysis of ideology, the Freudian theory of dreams and slips of the tongue, and the structuralist doctrine of myths. For Marx (in the 18 Brumaire, for example) the ideology of the different political parties was a mask for their allegiance to a particular class; for Freud, and for Lévi-Strauss, the dream or the myth is a disguise for an irresolvable and unacknowledgeable tension. In each case, the explicit meaning is generally false, and only the implicit meaning systematically revealed is given any weight. The initial reaction to each of these systems was one of rejection by the academic establishment: the Marxian dialectic met as much resistance from orthodox economists as psychoanalysis from the medical profession, and Lévi-Strauss is regarded by many of his fellow anthropologists as something between a charlatan and a poet.

The style of religious paranoia that occasionally appears in all these movements is both an answer to this ostracism and a provocation of it. Their organization into quasi-conspiratorial sects6 is related to the rejection of explicit meaning, and, in every case, this rejection is presented as a scandal. The scandal may be political, sexual, or purely intellectual, but it is always construed as an ethical attack on a conspiracy of silence. The explicit meaning which is to be cast out is conceived not only as mistaken or trivial, but as having been deliberately designed to mislead. The symbols of a dream are there to hide what they unconsciously betray, as ideology is intended not to enlighten but to conceal. For Schenker, all musical theory before his own was a deliberate conspiracy between mediocrity and non-Germanic musicians to betray the great tradition of music from 1700 to 1850 by concentrating only upon the superficial, explicit aspects of the great German classics. Even the works themselves, indeed, are conceived not as revelations, but as concealments of the underlying structure.

In Schenker’s analyses, this creates a difficulty when moving from the implicit to the explicit.7 He tried to cover the difficulty with an interesting and provocative theory of “improvisation,” but the awkwardness is always present. Schenker continued to use old-fashioned concepts like “first theme,” “counter statement,” and “bridge passage” in writing about sonatas, and even accepts these terms with all their nineteenth-century crudity. What remains unclear, however, is the transition from Schenker’s graph to the music itself with its admitted tunes, modulations, and so forth.

The transition from implicit to explicit always presents this difficulty, which is naturally less obvious when going in the other direction. It is easy enough to see the implicit meaning hidden within the exterior shell once the methodology has been learned, harder to decide why the inner sense should have taken just this outer form.

It should be emphasized that it is neither the truth nor the importance of Schenker’s deep structure that is in question but its status—its nature and its relation to the work as a whole. What does a Schenkerian graph represent? We cannot call it the form of a piece because it omits too many major forces of musical importance. We cannot even call it an adequate representation of the purely linear element of the form unless we are willing to claim that this is not essentially affected by proportion and rhythm. Nor is it, as Schenker claimed, a method of composition, although it plays a definite role in this process.

This ambiguity is a stumbling block found in any artistic analysis that takes the form of a hidden pattern, and yet it is hard to conceive of an analytic approach that could remain interesting while claiming that the meanings it uncovers are less significant than the explicit ones that have been evident all along. Such humility would appear to be self-defeating. The problem becomes clearer when we leave the isolatable world of music for literature, composed in a medium that connects at every point with a language used for everyday speech.


Ferdinand de Saussure, even while giving the lectures at the University of Geneva which are the foundation of structural linguistics, spent many years investigating the possibility of hidden anagrams in Latin poetry. When he died in 1912, he left ninety-nine notebooks filled with his speculations, but never published anything on the subject. (He was also never able to bring himself to publish—or, indeed, even to write—the great Course in General Linguistics on which his immense reputation rests. It is a compilation of students’ notes.)

The notebooks have been largely withheld from publication, perhaps because their character is as embarrassing as it is fascinating and provocative. A few selected passages of a general nature were published in essays by Jean Starobinski in 1954 in the Mercure de France, in Tel Quel No. 37, and in the Festschrift volume of 1967 for Roman Jakobson.8 In Change No. 6, which appeared a few months ago, Starobinski has edited Saussure’s analysis of lines 1-52 and 1184-1189 of Lucretius’s De Rerum Natura.

The origin of Saussure’s researches was an investigation into the phonetic structure of that early, primitive, and mysterious form of Latin verse called Saturnian. He became convinced that there was a regular repetition, a coupling, in fact, of the phonetic elements. Alliteration (the repetition of an initial sound) and rhyme (the repetition of a final) were only particular cases of a more general phenomenon, in which the interior syllables also had a function. His attempts to work out a regular law for these couplings in Saturnian verse proved unfruitful, but they suggested to him that the basis for them was a proper name (a number of the fragments of Saturnian verse that have come down to us are commemorative and taken from the sepulchral epitaphs of the Scipios).

From the proper name it was only a step to the idea of any word, its phonetic elements isolated, freely distributed and combined, providing the basis for verse, and Saussure turned to a study of Virgil, only to find the same couplings and the same musical suggestion of a hidden word influencing and even determining the phonetic structure of any given series of lines. This seemed to Saussure, as to anyone else, a deplorably and absurdly burdensome method of writing poetry, but the hypothesis explained the intricate phonetic symmetries that are a regular characteristic of poetry, and even—though less markedly—of prose.

The assumption that Saussure made was that the manifold phonetic symmetries of a passage in verse, including those not required by the rules of versification, were not determined at random by the sense, but combined into an ordered unity. This would correspond to the feeling shared by many readers of poetry that each part of a poem, closely read, has a specific and individual phonetic character of its own. This individuality of sound is the source of the most direct, most immediately sensuous pleasure that verse can give.

Saussure made the further assumption, a natural if more dubious step, that the unity obtained by the combination of the dominant assonances within a short passage had itself a clearly definable meaning—in other words, that the various syllables made up a word. This word was his “anagram,” a term with which he was eventually dissatisfied, substituting, in turn, “paragram,” “hypogram,” “logogram,” “paramorph,” and “anaphony”; the last named best expressed Saussure’s insistence that the game was played not with writing but with sound.

In all of his speculations on this subject, he relied principally on his ear, and his analyses constantly stress the setting-in-relief of his phonetic couplings by the accent of the verse and the stress of meaning within the line. He made it a condition not only that the different phonetic elements of his theme words should appear clearly emphasized within a very few lines, but that these lines should contain a definite model of the theme words in the form of a short phrase that began and ended as the word did. For example, “Aëriae primum volucræs tæ” is one model, or mannequin, as Saussure called it, for “Aphrodite.”

The analyses are fascinating to read because of this sensitivity: they represent the spontaneous reactions of a reader of poetry systematized and rationalized almost to a point of insanity. Almost—because at every point Saussure stopped to ask himself whether the whole investigation was not a magnificent delusion like the hunt for a Baconian cypher in Shakespeare.

In the first fifty-two lines of Lucretius’s poem, which contains the invocation to Venus, Saussure finds that the phonetic emphases combine to form the syllables of the Greek word “A-phro-di-te.” This is striking enough to shake any doubts of Saussure’s approach. Lucretius was a poet so steeped in Greek verse that it is more than probable that the composition of his invocation to Venus was accompanied by innumerable souvenirs of Greek verses to Aphrodite, echoes of lines and phrases remembered and half-remembered.

It is also reasonable that a poet should be almost pathologically sensitive to the suggestiveness of the purely phonetic aspect of words. The inherent improbability of Saussure’s theory begins for a moment to evaporate. In fact, if we accept Saussure’s analysis of Lucretius, we do not even have to ask whether the poet was conscious of his use of an underlying theme word; an unconscious intent is as natural and as credible in this case as a planned strategy.

Nevertheless, Saussure felt impelled to ask just that question: was the anagram a consciously applied technique of Latin verse? At this point, the latent paranoia of his program of research comes to the surface. If Saussure’s conjecture is right, the conspiracy of silence on the part of the whole body of classical literature is frightening. Finally, in desperation, Saussure wrote to a contemporary Italian who composed Latin verses, hoping to learn that the tradition of anagrams had been handed down secretly to the present day. We do not know if there was a reply, but Saussure never spoke again of his researches.

To conceal a word phonetically within a set of different words is a legitimate poetic effect; there is a famous example in Valéry’s Cimetière Marin:

La mer, la mer, toujours recom- mencée
O récompense après une pensée
Qu’un long regard sur le calme des dieux.

The second line literally illustrates the preceding word “recommencée” by hiding it and expanding it over the entire line, like a larger wave that builds itself up and breaks after a moment’s tension, broadening the faster and more regular rhythm that preceded it.

Such effects, based on the quality of the individual sounds, are more crucial in French verse, with its relatively uniform syllabic weight, than in English, which relies on the force of the rhythmic accent. They are perhaps even easier to achieve in Latin, with a syntax that permits so much greater freedom in the order of words. Both in French and Latin verse, passages are often emphasized by being constructed out of a specific nexus of sounds, and the importance given to such peculiar refinement is one of the reasons non-French-speaking readers find French poetry so difficult at first to appreciate. One classical scholar has remarked on the intricate patterns of sound that arise in Latin poetry from repeating the phonetic elements of the most expressive word in a passage, so that the sense of the word seems to radiate phonetically into those that surround it.9 With this explicit “radiating word” we are not far from Saussure’s implicit anagram.

In all of his work on the anagrams, Saussure continued to ask whether his results were due to chance. To this question, and also to the question whether the procedure was intentional or unconscious, no answer could in principle be given, whatever statistics were compiled. Alliteration, assonance, rhyme, rhythmic pattern—any form of phonetic repetition, in general, is part of the essence of poetry. Each system of versification sets up those repetitions which are obligatory (sometimes called canonic), and the others are freely used in subsidiary contrapuntal patterns with the main one. Accent and, sometimes, rhyme are obligatory or canonic in English; assonance and alliteration are subsidiary and, in a sense, free, but they are nonetheless a deeply ingrained part of English poetic tradition. That anagrams may be derived from these subsidiary systems is not surprising; phonetic repetition is not just by chance in a poem.

Or in prose, either, for that matter. Saussure found his anagrams, somewhat unwillingly, even in Latin prose (artistic prose, however, that of Cicero and Julius Caesar). But, as Starobinski pointed out, they may be a part of ordinary speech as well. How often do we not find a word or even an idea suggested by alliteration or rhyme, and allow our thoughts to be guided by sound as well as by sense? Saussure’s anagrams grow as much from language itself as from literary technique. The significance of Saussure’s ninety-nine notebooks is to show the intimate relationship between poetry and the processes of language and, above all, to demonstrate the power of a phonetic pattern to demand a meaning, the right to exist as a truly functioning part of language. It is difficult to read Saussure without hoping that he will prove to be right.

An ordered structure is a provocation, and we instinctively refuse to admit its lack of meaning. What Saussure was claiming was a significance for the structure of language independent of the message it contains. Saussure thought he was investigating not an attribute of language, but an esoteric technique of poetry. Ironically, what he found was an attribute of language which is a necessary condition for the existence of poetry.

We may rephrase Saussure’s questions so that neither intention nor chance will muddy the waters. Can the subsidiary, non-canonic techniques of phonetic repetition in a poem have a meaning of their own independent of the explicit poetic discourse? If Saussure’s own philosophy of language is right and meaning can only come into existence given an arbitrary convention that unites sound and sense, then the answer must be No. Can the non-canonic repetitions interact with the poetic discourse to form new meanings? That would mean taking the phonetic structure of poetry very seriously indeed. It would also entail defining the relation of implicit to explicit in a work of art.


The explicit, canonic structure of one of Shakespeare’s sonnets is fourteen lines of iambic pentameter with a rhyme scheme of three quatrains (each abab) and a final couplet. In his pamphlet on Sonnet CXXIX (Th’expence of Spirit in a waste of shame). Roman Jakobson (in collaboration with Lawrence G. Jones) has identified a considerable series of subsidiary patterns of a phonetic, syntactic, and semantic nature.

In his literary criticism, Jakobson relies largely on the principle of binary opposition which played so fundamental a role in his systematization of the study of the sound structure of a language. Completing Troubetzkoy’s work, he reduced the phonetic structure of any language to ten and only ten possible oppositions so that each elementary unit of sound, or phoneme, is analyzed as nasal or non-nasal, consonantal or non-consonantal, and so on. This principle of binary selection is essential to information theory, which conceives the transmission of a message as dependent on a series of successive choices between alternatives.

For this reason, perhaps, he prefers to analyze short poems that divide into four or five parts. He can then contrast the stanzas individually, pitting odd against even, inner against outer, anterior against posterior. The points he considers are of a precise technical nature often brushed aside by critics. The contrast of grammatical and nongrammatical rhymes (rhyme words with the same or different grammatical function) has been treated in English previously only by W. K. Wimsatt, I believe, although the relation of grammatical rhymes to the poetic language was already raised in the section drafted by Jakobson of the Theses of 1929 of the Prague Linguistic Circle. In the Theses, too, we find the principle developed later by Jakobson that the “purely” phonetic patterns in poetry are strongly bound to the semantic structure.

Here is the sonnet as Jakobson and Jones have arranged it for analysis:

I 1Th’expence of Spirit / in a waste of shame
2Is lust in action, / and till action, lust
3Is perjurd, murdrous, / blouddy full of blame,
4Savage, extreame, rude, / cruel, not to trust,
II 1Injoyd no sooner / but dispised straight,
2Past reason hunted, / and no sooner had
3Past reason hated / as a swollowed bayt,
4On purpose layd / to make / the taker mad.
III 1Mad[e] In pursut / and in possession so,
2Had, having, and in quest, / to have extreame,
3A blisse in proofe / and provd / a[nd] very wo,
4Before a joy proposd / behind a dreame,
   IV 1All this the world / well knowes / yet none knowes well,
   2To shun the heaven / that leads / men to this hell.

What Jakobson calls the “poetry of grammar” is most brilliantly shown in this great “generalizing” sonnet on lust by the remarks he and Jones make on the only two nouns that refer to man: taker in a swallowed bait that makes the taker mad, and men in the heaven that leads men to this hell:

Both…function as direct objects in the last line of the even strophes: II taker and IV men. In common usage the unmarked agent of the verb is an animate, primarily of personal gender, and the unmarked goal is an inanimate. But in both cited constructions with transitive verbs the sonnet inverts this nuclear order. Both personal nouns characterize human beings as passive goals of extrinsic, nonhuman and inhuman actions.

It is upon details such as this “grammatical metaphor” (as Jakobson has called it elsewhere) that the authors build their reading of the poem with its “semantic leitmotif” of “tragic predestination.” The ability of the grammatical structure of language to assume a poetic life of its own is fundamental to music, which imitates this aspect of language above all.

In their description of the verbal art of the sonnet, the authors are concerned to show the inner correspondences that work against the simple division into three quatrains and final couplet. Most convincing in this respect is the light shed on the central verses in this scheme. Lines 7 and 8 stand out in relief, as they are the only ones without grammatical parallelism, and line 8 is “built of five totally unlike grammatical forms.” This partially elucidates the means by which Shakespeare achieves the remarkable change of tempo in the center of the sonnet, with its sudden breadth and complexity of movement.

The prejudice against linguistic criticism of this kind is solidly, and to some extent reasonably, founded on a distaste for learning a new vocabulary, one which is at times unnecessarily technical for one’s purposes. But Shakespeare’s art consists—at least in large part—in an extraordinary feeling for the very stuff of language in all its aspects, and criticism has need of the tools that Jakobson offers. It is melancholy to read the pious horror (mostly British) at the invasion of an urbane humanistic discipline by the linguistic barbarians. Syntax is as relevant as irony for the understanding of literature, and Jakobson’s microscopic examinations come out of a long life’s delight in literature, and a breadth of interest unsurpassed since the deaths of Erich Auerbach and Leo Spitzer.

Those of us who cannot read a Slavic tongue must be content with Jakobson’s essays on English and French poetry. The most famous of these is the analysis of Baudelaire’s Les Chats (done in collaboration with Claude Lévi-Strauss), the most satisfying perhaps the closely reasoned treatment of Baudelaire’s Spleen. Equally important are the theoretical papers, above all the article “Linguistics and Poetics,”10 in which the largeness of vision is balanced by its clarity. In all this work, social and biographical interpretations are excluded. Jakobson has always insisted that language and literature must be understood as systems in their own terms before their interaction with other systems can be apprehended. It is doubtful whether, as a matter of fact, this is a possible or even desirable goal in all of the purity with which Jakobson has invested it, but it is unquestionably the best practical starting point for criticism.

As appears in one minor aspect of the Shakespearian pamphlet, Jakobson has been considerably influenced by Saussure, and he and Jones tentatively suggest an anagrammatic signature worked into the opening line:

Th’expence of Spirit in a waste of
   ksp sp.r Sha

in addition to a pun on Will in the concluding couplet, All this the world well knowes yet none knowes well,/ To shun the heaven that leads men to this hell; and in support of this suggestion they cite Shakespeare’s tendency to equate “will” and “well” in puns. (For a sonnet on lust, they might have added the Elizabethan sense of “will” as “carnal desire.”) This is not Jakobson’s first use of Saussurian anagrams. He finds anagrams for Spleen in all four of Baudelaire’s poems with this title, and an anagrammatic influence of the title throughout the poem called Le Gouffre.

The difficulty—at least in the Shakespeare—is not an inherent improbability, but the lack of coordination with the rest of the analysis. Sonnet CXXIX, written on the most personal of themes—fornication and its bitter aftermath—is “the only one among the 154 sonnets of the 1609 Quarto which contains no personal or corresponding possessive pronouns.” Nor is a fornicator even referred to except in a dependent clause as part of a simile and there in a passive role (see remarks on taker above), and in the superbly general word “men” in the last line. The entire sonnet is not only “generalizing,” but absolutely and even repressively impersonal.

If we accept the anagram and the pun on Will, they are—if conscious—sardonic jokes, and—if unconscious—the revenge of suppressed nature.11 In either case, we are left with a specifically personal meaning which deliberately undercuts the explicitly generalizing form of the sonnet. We need a very superior sort of irony to integrate the two.

The most disappointing part of Jakobson and Jones is the section “Odd Against Even.” They list every binary correspondence that comes to hand, and we sometimes have the impression of reading a series of notes left unused after an analysis was written. (There is even an apologetic note in the presentation, when they write, “By the way, the preposition in appears only in the odd strophes.”) It would be unfair to suggest that much of this detail has an interest more linguistic than literary: poetry rejects no aspect of language. But the presentation of this section does not establish throughout its relevance to Shakespeare’s verbal art.

All this would not matter if the significance of the odd-even opposition were acceptable. It is suggested that the odd strophes contain “an intensely abstractive confrontation of the different stages of lust (before, in action, behind), whereas the even strophes are centered upon the metamorphosis itself.” I do not believe that a reading of the sonnet bears this out: the heaven and hell of the fourth strophe, for example, are as abstractive a confrontation as anything in the first, although one leads to the other; and the opening line already describes the metamorphosis of “in action” into its aftermath.

In fact, the treatment of the three stages of lust is beautifully balanced, and the movement from one to the other is subtly modulated. Jakobson and Jones do not study this movement, so they miss the balance of the opening quatrains, in each of which the first two lines present two states and the last two lines characterize one at length (before action in lines 3 and 4, after in 7 and 8). They also miss the remarkable backward slide in time of line 10:

Had, having, and in quest, to have extreame,

which is prepared by the retrospective glance of lines 7 and 8, where the third stage is described by its horrified view of the first.

Jakobson has called upon Saussure’s authority to justify taking “the elements out of the order in time” in which they are presented in the poem. But to be effective, the correspondences so discovered should not be invalidated by an actual reading in time. A detail found in both the first and third strophes cannot be isolated when it is found in the second as well, and in linking the odd strophes Jakobson and Jones cite as correspondences some features which are pervasive; they bring together: I, Spirit (sp.r.t) in; III, In pursuit (p.rs.t). But they themselves later call attention to the paronomastic chain which unites the odd and even strophes and which pairs In pursuit with Past reason (the latter occurring twice in the second strophe, both times in the prominent initial position). Their neglect of this pervasive character makes the analysis of the expressive sound structure of the final couplet uncertain as well.

On the other hand, if we eliminate the opposition of odd and even, we can see that the four consonants of Spirit dominate the first twelve lines:

1 expence, Spirit, waste

2 lust, lust

3 Is perjurd (sp.r.rd)

4 Savage, extreame (str), trust

5 dispised straight

6 Past reason hunted (p.str.s))

7 Past reason hated,

8 On purpose layd (p.rp.s’….’d),

9 In pursuit (p.rs.t)

10 in quest, to have extreame

11 A blisse in proofe

12 proposd

This concentration is emphasized by the phonetic symmetries of the verses (e.g., the first word of line 7 is past, the final word bayt; pursuit of line 9 is balanced by possession; in line 6 reason and sooner exchange their three consonants).

In the last two lines, on the other hand, this particular nexus of sounds disappears, and the contrast of sonority is striking. The final couplet has no “p,” only one “r” (in world) and its three “t”s are all on weak beats and two of them are finals (yet and that), which further weakens their emphasis. The dominant sounds of the last two lines are the alliterations on “w,” “n,” “th,” “h,” the recurrence of “l,” “s,” and “n” in final position, and the prominent “sh” of To shun, a remarkable concentration of aspirates and soft consonants.

Jakobson and Jones remark on the density of texture in the final couplet12 but not on its hushed quality and on the almost complete absence of the explosives that pervade the first twelve lines. If one were to construct anagrams, it might justly be said that Spirit determines the sonority of the quatrains, and shame of the final couplet.

Jakobson and Jones do not convey the richness of meaning in the opening line and the extent to which it announces the tragic theme of the whole sonnet. They note double meanings in the opening line for shame (chastity and genitalia) and for Spirit (a vital power of both mind and semen), but do not give for the latter the common meaning in Shakespeare’s time of the soul, in particular at the moment of death as it leaves the body. This is a sense enlisted by the word expence which had for Shakespeare the now obsolete meaning of “loss,” as in Sonnet XXX:

Then can I drowne an eye (un-us’d to flow)
For precious friends hid in deaths dateless night,
And weepe afresh loves long since canceld woe,
And mone th’expence of many a vannisht sight.

Th’expence of Spirit is a metaphor for death, which is itself the oldest and most common of all rhetorical images for sexual intercourse. This philological background is a necessary supplement to Jakobson’s and Jones’s linguistic detail.

Jakobson has held with Empson that “the machinations of ambiguity are among the very roots of poetry,” as of any message centered on itself, and Jakobson and Jones remark that Shakespeare’s double-entendre does not interfere with the firm thematic construction of the sonnet. The double meanings they sketch are, as they say, only a kind of double-talk. But Empson’s stand is more powerful and more interesting, and his machinery enables us to integrate ambiguity into the sonnet as a whole instead of presenting it as a suggestive but unessential decoration. He claims that when the poetic context calls up two meanings of a word, it enforces a relation between the two which is understood in the context of the poem.

In Th’expence of Spirit this relation is, in fact, the theme of “tragic predestination” that Jakobson and Jones clarify. The grossly physical sense (semen) of Spirit is identified with the spiritual (soul) by the degradation of lust. The religious overtones of the last line (the heaven that leads men to this hell) take this up again: man loses his soul through the expense of Spirit and is damned. The paradox of a heaven that leads to hell mirrors the opposition compressed into Spirit.


As we have seen from Saussure, a formal pattern must signify something, but to achieve this significance, it needs a context. The context of a poem can only come from a “reading,” and many of Jakobson’s and Jones’s correspondences do not submit to—or, better, do not acknowledge submission to—their own specific “reading” which is essentially the thematic structure they assess so convincingly.

In other words, there is a hierarchy of significance in all poetry before Mallarmé, and an explicit sense to which all the other interpretations must pay homage. This central, explicit sense is generally not subject to much controversy. It is the meaning established by a reading in time, in a simple linear13 fashion starting from the beginning. This explicit meaning is the literal reading of Aquinas, and, when summarized, it is the simple thematic structure affirmed by Jakobson.

This is not to claim that non-linear interpretations and subsidiary, implicit patterns do not exist. On the contrary, their existence and their interaction with the explicit structure is the source of poetic strength, the verbal art of Jakobson’s and Jones’s title. But the criterion of relevance for these implicit patterns is the possibility of integration into the explicit one.14

Unfortunately for critics, some of these patterns will always prove to be irrelevant. In every sonnet, the phonetic structure is partially set in advance and therefore gratuitous. The semantic structure is mapped out to fit the rhyme and the meter. In a line like

Is lust in action, and in action, lust

the ideas are arranged to reveal the balance of the ten syllables and to isolate them: to some extent the content signifies the form and illuminates it, holds it up for inspection. The poet works to make the gratuitous form seem to be determined by the meaning. By definition he cannot completely succeed: if the phonetic structure seems totally dependent on the meaning, the work ceases to be poetry—that is, it ceases to be a privileged message, protected by its prosody from being confounded with an ordinary statement. The form must remain overdetermined.

The subsidiary, non-canonic phonetic patterns, in harmony with the principal one, are also partially gratuitous, can never completely fulfill the criterion of relevance. Their interaction with the semantic structure can never be complete. Above all, the semantic pattern itself will inevitably reveal the workings of organizing forces that are partially gratuitous: it, too, is set up like the rules of a game. These patterns also demand interpretation and yet cannot be made to yield absolutely to coherence. An analysis will always leave a residue, arabesques of phonetic and semantic patterns that imply a meaning and yet elude interpretation.

This gives art its privileged position. We are not allowed to claim, because of Sonnet CXXIX, that Shakespeare thought sexual intercourse disgusting, immoral, or degrading. This would be like rushing onto the stage to warn Othello that the handkerchief was planted. The sonnet is not a personal communication, however much feeling and experience went into it. The richness of meaning in the sonnet depends on the partial release of language from its normal function of conveying information. A poetic “message” is not tied to a specific receiver, and its direction remains open; in Schlegel’s words, “A poem is written for everyone or no one—the poet who writes for someone in particular deserves to go unread.” If, however, the significance of the “message” is now tied asymmetrically to the sender, this freedom is equally menaced. A personal message to everyone or to no one is a voice crying in the wilderness, and excites an absurdly misplaced pathos.

This privileged status makes art dangerous: these arbitrary conventions that struggle into significance and that signal to the reader that the message is a fiction allow things to be written that would be otherwise intolerable. The unspeakable may be whistled. A characterization of fornication that has the intimate emotional power of Shakespeare’s would have been unthinkable in late sixteenth-century prose, even in a sermon.

When the poetic “message” is freed from too intimate a tie with a specific sender or receiver, the latent meanings in the text are released and come alive, and the sonnet is open to the reader to interpret as irresponsibly as he pleases. No control is possible. But if the status of a work of art is threatened by being taken too seriously, it is also endangered by being reduced to “a superior amusement,” as T. S. Eliot modestly called poetry. Unless the implicit patterns, the intricate correspondences that Jakobson delights in uncovering take their place within a “reading,” they lose the significance that only that framework can give, and, in turn, cease to contribute to that framework. They tend to become facts of language, not of poetry. The analysis of the most complex and various patterns can only impoverish a work if they do not come together, if there is no focus.

Jakobson has, indeed, always insisted that the apparently autonomous phonetic and grammatical structures must be related to the structure of meaning. But the relation he generally proposes is a static one, less dynamic than a simple reading. His implicit correspondences sometimes add emphasis to the thematic structure, or at other times are a decorative counterpoint to the explicit form. Rarely do they substantially affect or aiter the explicit reading; we start from the text and move to the discovery of the interior symmetries. What is lacking is the continuous movement back and forth between the whole text and the interior forms, a movement which makes possible the fullness of poetic criticism.

As we read, we create—by imagination, instinct, reason, or whatever—a frame from which the poem takes its meaning. Criticism is only an extension of this process, and is therefore, as Walter Benjamin said, the necessary completion of a work of art. The fundamental critical tradition since the sixteenth century, philology, is the creation of a historical context. It appears inevitable that linguistics, with its new-found power, will replace philology, and Jakobson prefers a context as little tied to history as possible. This releases Jakobson’s correspondences from any fixed relation with tradition for the most part, and the new freedom is welcome: not all poetic technique is founded on convention, based on precedent. But the loss of historical significance is not easy to accept. It makes the poetry paradoxically less immediate, less likely to disturb.

Like literature, music can disturb and even shock. It has all the attributes of language except one. It has a grammar, a syntax, accent, tone of voice, syllables, and phonemes; it, too, is formed into sentences, paragraphs, chapters—what it lacks is a vocabulary. Musical phonemes act directly without first being strained through an abstract system of denotation. Music is a mimetic art in so far as it imitates language and language’s poor relation, gesture. It may be called pre-verbal but post-lingual. (What rudimentary vocabulary music has is mostly one of gesture, not of language.)

Most accounts of the expressive character of music have been largely attempts to identify the vocabulary; that is why they appear so simpleminded. The expressive force of music rests principally in its grammar. The capacity of the grammatical structure of a language to assume a meaning of its own (described by Jakobson), the power of any ordered pattern to appear significant, to demand interpretation (which fascinated Saussure)—these attributes of language are annexed by music and developed with an intensity unparalleled in any other art. Music is made up in great part of relationships like those that Jakobson analyzes as Shakespeare’s verbal art, but the musical structures are far more sophisticated, efficient, and powerful.

The relation between explicit and implicit in music is therefore more difficult to clarify. For example, the distinction between canonic and noncanonic organization necessary for prosody is almost useless in music. Those conventional forms so often taught in music courses are largely a fiction, invented long after the fact. The fugue was a free form for Bach, and Mozart and Haydn never heard of sonata form and certainly had no idea of the standard pattern that has been taught under that name since the middle of the nineteenth century. Schenker was correct in maintaining that (at least since 1600) the most important “rules” of music are simply the rules of counterpoint and the laws of tonality. It is to these rules that Schenker reduces a work of music, ignoring the phrase-structure—which for most listeners contains the explicit sense of the work, and he turns his principles of organization into a hidden, esoteric form.

This is so because his method takes the form of a gradual reduction of the surface of the music to his basic phrase,15 and the analysis moves in one direction, away from what is actually heard and toward a form which is more or less the same for every work. It is a method which, for all it reveals, concentrates on a single aspect of the music and, above all, makes it impossible to bring the other aspects into play. The work appears to drain away into the secret form hidden within itself. That is the impasse of every critical method which places the source of vitality in an implicit form.

Criticism cannot do without these implicit forms, inner relationships, hidden significances. But they must be so presented that they not only reflect the work but also reflect back upon it, and at an oblique angle so that they can receive more in return than their own images. Criticism is not the reduction of a work to its individual, interior symmetries, but the continuous movement from explicit to implicit and back again. And it must end where it started—with the surface.

This Issue

June 17, 1971