These notes began life as a review of James Anderson Winn's Unsuspected Eloquence: A History of the Relations between Poetry and Music (Yale University Press, £16.50), so I had better say at once that I found this book consistently intriguing, informative and lucidly written. Anyone interested in poetry or music can read it with profit. Professor Winn begins with Homer and the recitation of the epics, but my comments are limited to music after Beethoven, and then largely to music after Wagner. This is in part a result of the limits of my knowledge, but it is also fair to say that Professor Winn does not allow himself enough space for the last hundred years of his history. I thought it would be useful to offer some comments on the important, early modern composers that Winn does not discuss at any length, Mahler, Debussy and Webern, for example. His accounts of Schoenberg and Stravinsky can be recommended, with some reservations (Schoenberg did not write the libretto for his monodrama Erwartung), but his wholly uncritical remarks on the 'stunning' music of Penderecki - who, to my mind, is a byword for the meretricious - suggest that his understanding of more recent music leaves something to be desired. This may be why he devotes fewer than three pages to the topic. I would like to suggest some ways in which this gap could be filled.

The surprising factor in the history of Western music is not its long dependence on words but the rapid development, in the eighteenth century, of those complex instrumental forms we now think of as its greatest glory. From this point music seems to alternate between abstraction and varying degrees of 'literariness'. In the previous century even so great a composer as Monteverdi had maintained that music must be subservient to poetry, but by the late eighteenth century Mozart felt able to put words firmly in their place as music's 'obedient daughters'. A generation of composers that had invented and perfected sonata-form, the symphony and the string quartet had every reason to be assertive. The four-movement symphony is, of course, the paradigmatic form of 'classical music', and yet, in its pure form, it was remarkably short-lived. The mature symphonies of Haydn, Mozart and Clementi adhere to classical proprieties but, in his Sixth and Ninth Symphonies, Beethoven severely compromised the purity of the form by introducing literary elements. It was this convergence of music and literature (as they saw it) that was seized upon by his romantic successors.

We are used to the idea of poetry aspiring to the condition of music, but the music of Berlioz, Liszt, Schumann and Wagner often seems to aspire to the condition of literature. As Professor Winn remarks, 'their music seems to cherish and prolong harmonic ambiguity in much the same way that the romantic poets had cherished and enacted syntactic ambiguity.' Keats's or Byron's or Shelley's pursuit of the unattainable seems to find musical expression in their vaguer, more wide-ranging modulations. Byron found favour with Berlioz, and later with Tchaikovsky, but the writers most admired by the romantic composers were Dante, Shakespeare and Goethe. The only musician who was thought to have equalled their achievements was Beethoven. His 'Pastoral' Symphony was by no means the first programme symphony (Dittersdorf, for example, had written a cycle of picturesque symphonies based on Ovid's Metamorphoses) but it was the first to gain a permanent place in the orchestral repertoire, and its example legitimised the nineteenth-century obsession with musical programmes. Berlioz's Symphonie Fantastique, Tchaikovsky's 'Manfred' Symphony and Liszt's symphonic poems follow from it. These hybrid forms were also a pragmatic response to the demands of larger, predominantly bourgeois audiences. It was thought unreasonable to expect such audiences to grasp abstract concepts: it was assumed that they would outline some sort of programme for themselves 'according to the grandiose, lively, impetuous, serenely soothing or melancholy character of the music'. Where Beethoven had neglected to supply programmes, nineteenth-century commentators were happy to invent their own. As a consequence of these developments, the view that music served only to express ideas and emotions - that is to say ideas and emotions that could be expressed in other than musical terms - became firmly established. Audiences still attempt to 'read' a symphony as they would a Victorian novel.

Liszt maintained that his symphonic poems represented 'a deeper union of music and poetry' than word-setting itself. He went on to elevate the 'tone-poet' over the 'mere musician': the former 'reproduces his impressions and the adventures of his soul in order to communicate them', while the latter is only an ingenious manipulator of notes who, 'since he speaks to men neither of his joys nor his sorrows, neither of resignation nor desire remains an object of indifference to the masses, and interests only those colleagues competent to appreciate his facility'. With its extraordinary denigration of technique and its invocation of an uninformed mass audience as the arbiter of musical taste, this passage uncannily anticipates Stalinist pronouncements. Liszt, of course, was a musician of formidable technical skill whose radical late works certainly cannot have been intended for a mass audience, but his symphonic poems just as certainly show the ill-effects of this philosophy. Composers have always derived musical ideas from extra-musical sources (paintings, books, dreams, love-affairs and landscapes) but the listener does not usually need to be aware of these sources to follow the progress of the music that springs from them; indeed the composer may actively discourage such awareness. Not so with the symphonic poem, which represents a more complete subjugation of music to literary concepts than had existed before. Because it follows a text without stating it, the symphonic poem must resort to more crudely illustrative effects than would be necessary in any actual setting of the text. This presented formidable problems of construction. Illustrative effects and story-telling could not be satisfactorily combined with symphonic forms that had their origin in an abstract musical logic, and even such late and sophisticated examples as Strauss's Don Quixote remain loosely episodic. Paradoxically, the worst passages in symphonic poems often result from the attempt to make abstract, philosophical ideas musically explicit: the bombastic conclusion to Tod und Verklarung compares very unfavourably with the episodes of the sheep and the windmills in Don Quixote. One might be inclined to say that the ultimate destiny of the form was to provide composers of film music with a convenient thesaurus of illustrative musical gestures, were it not for the career of Claude Debussy, a masterly 'tone-poet' of a new kind.

The 'deeper union' that Liszt thought he had found in the symphonic poem was also Wagner's Holy Grail. He thought he had found it in the music-drama, that Gesamkunstwerk in which words and music were to be indissolubly wedded as they had not been since the days of Greek tragic drama. Like the Greek dramatists, Wagner would be both composer and poet, and thus there is a strong element of atavism in the Wagnerian revolution which appears as an attempt to bring the whole history of Western music full circle. In justifying his insistence that music without words was inadequate, Wagner turned to the example of Beethoven. Ignoring the eloquent, contradictory evidence of the late piano sonatas and string quartets, he insisted that Beethoven's career showed a progressive convergence of music and words, culminating in the Ninth Symphony. Obviously Beethoven had not gone far enough and Wagner felt obliged to provide a programme for the Ninth, based loosely on Goethe's Faust.

The Wagnerian enterprise might appear to be the negation of the classical symphony. Even when they were not initially related to words in the text his notorious leitmotivs were intended to express literary and philosophical concepts. Despite this they have a tendency to be absorbed into the glutinous mass of the Wagnerian 'endless melody', which is less an extension of earlier operatic practice than a monstrous distension of the classical sonata movement's development section. Wagner was confronted by an insuperable problem for, while music can be powerfully, if indefinably expressive, it cannot unambiguously depict anything: in music a storm at sea or a crisis in a love affair sound much the same. The leitmotiv could not solve this problem, and this gave rise to an extensive Wagner industry producing guides and leitmotiv charts so that earnest disciples might know at any moment precisely what the music signified. This apparatus is the measure of his failure. Debussy was to remark that a leitmotiv put him in mind of 'a harmless lunatic who, on presenting his card, declaims his name in song'. It is, perhaps, needless to say that without his music Wagner's dramas would be something less than historical curiosities.

The ultimate destination of the line that begins with Beethoven's Ninth, and passes through Lisztian symphonic poem and Wagnerian music-drama is the 'world-embracing' symphony of Mahler. This 'line' is most obviously revealed in the Eighth Symphony which seems to take Wagner's programme for Beethoven's Ninth as its point of departure: its vast second part is a setting of the final scene from Goethe's Faust. Mahler is the most literary of symphonists. I do not only mean that four of his symphonies employ voices to sing texts, or even that his purely orchestral symphonies have the character of heroic autobiographies. I mean that Mahler's whole style with its pervasive allusiveness, its use of parody, burlesque and near quotation, its ironic mixing of musical vernacular with high, romantic rhetoric, parallels the most advanced, modernist, literary techniques. 'Literary form' in Mahler does not serve any crudely programmatic purpose: such discrete, musical elements as funeral marches, Ländler, songs and the sound of cowbells serve a symbolic function. The disposition of musical symbols becomes the dominant ordering principle, a way of organizing meaning in the collapsing universe of the tonal system.

Composers have only begun to recognize the full potential of this technique within the last twenty years. Berio's Sinfonia is the clearest example. Its third movement uses the scherzo of Mahler's Second Symphony as a frame within which quotations from Debussy, Strauss, Ravel, Stravinsky and Schoenberg (to name but a few) are juxtaposed with solfège, spoken passages from Beckett and Joyce, and political graffiti from the Paris of 1968. Berio is not amusing himself at Mahler's expense, as some commentators seem to have thought; he is only making explicit what is already implicit in Mahler's music: 'Quotations and references were chosen not only for their real but also for their potential relation to Mahler'. Even the visionary political slogans are appropriate since Mahler is known to have sympathized with the Viennese workers' movements of his day.

By taking Wagner's saturated chromaticism a stage further, Mahler had arrived at the brink of atonality. Schoenberg was to take music over that brink. As he moved closer to atonality, abstract construction on a large scale became problematic for Schoenberg. He came to rely on poetic form, and particularly the poetic forms of Stefan George. When Schoenberg introduced a soprano voice in the third movement of his Second String Quartet of 1908, it was as momentous an event in the history of music as the entry of the baritone in Beethoven's Ninth, for at the same moment as the soprano enters, tonality is decisively abandoned. Dissonance has overwhelmed the sense of key and tonal cadence can no longer provide resolution or formal balance.

The freely atonal music that Schoenberg wrote between 1908 and 1914 is usually described as 'Expressionist', but this is almost as misleading as the term 'Impressionist' as applied to Debussy. It should not be forgotten that the earliest pieces to be composed in this style appeared before the publication of the first Expressionist poems. Schoenberg's literary tastes were Symbolist and fin-de-siècle. As late as 1914 we find him setting Dowson, and at the same time working on a vocal symphony of supra-Mahlerian proportions based loosely on Seraphita, Balzac's bizarre, Sweden-borgian tale of androgyny among the Norwegian fiords. Schoenberg's questionable taste in literature has discomforted his admirers, but in the case of the Second String Quartet there can be no doubt that he had found the perfect text for the occasion. Both poems are by George and the second begins 'I feel the air of other planets . . .'. Still more otherworldly in character is Herzgewachse, a setting of Maeterlinck composed in 1911. Both the freely developing form of the piece and its unique instrumentation (harp, harmonium and celeste) arise from the mood and imagery of the poem, not its strict metre and rhyme scheme. Schoenberg is concerned to find a disposition of musical images to match the disposition of verbal images, but music has a way of going beyond words, and the result is a kind of miniature cantata or study for voice and instruments in which Schoenberg's music, with its staggeringly difficult vocal part, effortlessly transcends the limitations of Maeterlinck's pale and crepuscular poem. As Charles Rosen says of Erwartung: 'the intense, relentless expressivity of each moment . . . is a formal device as well as an extra-musical significance'. The dissonant harmonic language is related to the symbolism of the text, but 'it is a mistake to think that one means or signifies the other'.

It remained for Schoenberg's pupils, Berg and Webern, to take the logical step of matching innovative music to innovative texts. Webern took to writing and setting his own brief, Expressionist poems and, somewhat later, found the ideal texts for his purposes in the poetry of Trakl. Berg turned to the eccentric Peter Altenberg. At the famous concert that took place in Vienna on 31 March 1913 and ended in a riot, the audience was as much provoked by Altenberg's cryptic, 'postcard' poems as it was by the extraordinary sounds emerging from Berg's orchestra. At first hearing the Altenberg Lieder seem to be clear examples of Expressionist word-painting, a principle announced immediately in the shimmering orchestral crescendo that precedes the line: 'Seele, wie bist du schöner, tiefer, nach Schneestürmen'. But closer study reveals Berg's 'obsession with infinitesimal organization'. An elaborate web of allusions and quotations binds the five songs together, and Boulez finds in this 'musical punning comparable to that found in written language'. The last of these songs is the most paradoxical. Lushly romantic in texture and melodic contour, it is underpinned by the abstract formal procedures of a passacaglia (and, in this sense it anticipates the way in which Berg would impose abstract forms on Buchner and Wedekind in his two operas). At its end the music seems to break free of tonality and the forms that go with it, dissolving into the total, keyless ambiguity of hesitantly descending whole-tone scales. But this innovation, so prophetic of later twentieth-century music, is also an utterly direct transposition into musical terms of Altenberg's concluding image of snow gently dropping into pools.

Talk of poetic form brings us to Debussy. To discuss Symbolist aspirations to the condition of music without serious consideration of the music that most clearly aspires to the condition of Symbolist poetry (as Winn does) is frankly senseless. When Debussy was not setting Baudelaire, Verlaine, Mallarmé, Louys or Maeterlinck he was setting his own Symbolist vers libre. His texts for the Proses Lyriques of 1893 show the influence of Henri Regnier, Maeterlinck and Laforgue. In his setting of these texts and the prose of Louys' Chansons de Bilitis he invented a new style of vocal writing which, sensitive to every verbal nuance, gives an effect of heightened recitation or (in Frederick Goldbeck's phrase) 'conversational psalmodizing'. The majority of Debussy's instrumental compositions are the subtlest of 'tone-poems', often with elaborate, poetic titles - Et la lune descend sur le temple qui fut or Les sons et les parfums tournent dans l'air du soir (this last borrowed from Baudelaire's Harmonie du soir). He was also one of the most incisive and elegant writers on music. Monsieur Croche, his critical alter ego, is close cousin to Valéry's Monsieur Teste. Because of this deep affinity with the most advanced poets of his day, the young Debussy grew to loathe the idea that music should tell edifying stories or deliver messages. Verlaine's Art poétique is also Debussy's. It is a paradox of the time that his interest in literature should have led him to insist on the primacy of music.

Since they bear titles like La Mer, Nuages, Fêtes, Sirènes and Images it is all too easy to misinterpret Debussy's orchestral scores as programme music, or even as 'Impressionism', but we should note the vagueness and generality of these titles: in true Symbolist fashion there is 'nothing in them weighty or fixed'. These masterly scores have nothing to do with romantic heroism or rhetoric, they are concerned only with poetic image. The question of 'Impressionism' is revealed as a sham question if we remember that Debussy chose for the cover design of La Mer, Hokusai's famous print of the Great Wave: it would be hard to think of anything further from Monet. Like poets from Mallarmé to Ashbery, Debussy is concerned to escape 'paraphrasable meaning', in his case the kind of meaning that had been imposed on music by Wagner and generations of Beethoven commentators. He does so by piling on tonal ambiguities, by resurrecting antique modes and directing his attention to the question of timbre (including such remote timbres as those of Javanese gamelan), by forswearing symphonic development (which he dubbed 'the laboratory of the void'), mechanical rhetoric and any musical autobiography. Hadn't Verlaine enjoined poets to 'Take eloquence and wring its neck'? Debussy follows the path of reticence and 'ironically deprecated anguish'. The result is a kind of inverted eloquence. Hence, Pelléas et Mélisande - an opera of half-tones, unanswered questions and averted faces, and music in which (long before Webern) silence attains a new expressive value. Debussy's life-long sympathy with eccentric and isolated individuals, whether General Lavine or Roderick Usher, springs from the same source.

To say, as Schoenberg did, that Debussy's harmonies are 'without constructive meaning' or, as Winn does, that his methods 'cannot sustain real development' is to miss the point. Debussy would rather trust himself to Verlaine's 'bonne aventure'; he asks 'no one for advice except the wind that passes'. In place of 'pretentious forms', he elevates the principle of ornament that he found in Bach, Rameau, Monteverdi, Palestrina and gamelan. The flute monody that opens Prélude à l'après-midi d'un faune is an historic declaration of intent. The strenuous oppositions and tensions of major and minor, of tonic and dominant are dismissed in this liberative exploration of the possibilities of pure sound. Debussy's prelude cannot be mistaken for programme music - although its concluding pages do sound very like a meditation on Mallarmé's marvellous final line: 'Couple, adieu; je vais voir l'ombre que tu devins' - rather, it is a new kind of musical response to a new kind of poetry, a poetry which, perhaps, can only be adequately interpreted by music. Averse to all direct statement, Mallarmé believed that 'Les monuments, la mer, la face humaine, dans leurs plenitude natifs' were better served by evocation, allusion and suggestion than by description. So Debussy's score is the evocation of an evocation, and nothing inhibits the free elaboration of musical ideas. Mallarmé was delighted by it.

Discussion of the relations between Debussy's music and Mallarmé's poetry should not end with Prélude à l'après-midi d'un faune or even with the Trois Poèmes de Stephane Mallarmé of 1913. Composed at about the same time as these settings, the enigmatic poème dansé, Jeux, presents a very close correspondence with Mallarmé's elusive imagery and involuted syntax. Here all is elegance, concision and ellipsis. The very opening chords - ominous, aloof, scintillant, erotic - seem to be 'announcing a rose in the shadows', and the form of the piece is so unorthodox it resists description in conventional terms. It is a form that instantly renews itself - a great architecture composed of fragments and evanescence, in which coherence is achieved through discontinuity. Or, to put it another way, distant constellations reveal themselves through windows suddenly thrown open, mirrors expose their vacancy, fans palpitate in unknown hands, ecstatic flights are abruptly cut off . . . Jeux is a masterpiece of musical Symbolism. Boulez has spoken of its 'dream of vitrified improvisation' - a dream that is surely fulfilled in his own five-fold portrait of Mallarmé, Pli selon

Given his insistence on impersonality, Debussy's eventual arrival at the abstraction of the late sonatas was to be expected. Yet the Cello Sonata of 1915 is almost the first piece of 'absolute' music he had written since the String Quartet of 1893, and it was obviously hard to relinquish the habit of literary allusion: originally the central serenade of the sonata was to have been called Pierrot fâché avec la lune. I do not think that the allusion is to Schoenberg's Pierrot Lunaire, but to the Pierrot of Jules Laforgue to whom Debussy had paid homage in his Proses Lyriques. It is not merely the mood of the Cello Sonata's serenade that seems Laforguian: the allusions to musical vernacular in this movement seem to correspond to Laforgue's use of slang in Les Complaintes, and Debussy's irregular metres, his apparently 'free' association of musical ideas can be compared to the sophisticated vers libre of Derniers Vers.

This essay will be concluded in the next issue of PN Review