PART I

Saying Nothing

1

Sirènes

Wordless voices

A low murmur in the bass, edged by a quiet tolling of the harp, a little swell of movement in the horns, and then, like a flicker of sunlight on water, a rapid arpeggio in the clarinet – all gently moving but contained, like the faint rise and fall of the calmest of warm seas emerging through the morning mist. The beginning of Sirènes, the third of Debussy’s three orchestral Nocturnes, does not so much represent a seascape as put our listening bodies into a gentle motion of rise and fall as if we were on water (see Example 1.1). But this realism of wave-like movement frames the appearance of a wordless call, a simple grace-note figure, given by eight mezzo sopranos divided into four parts. Using the same pitches as the clarinet, and like a distant echo of the horns, the women’s voices are part of the orchestral sound but also quite distinct from it. The insistent repetition of their two-note figure leads to a brief ripple of surface motion (the shimmer of violins, the harp arpeggiation, and the return of the horns). After the preceding two movements of purely orchestral music, this sudden appearance of voices is arresting. On the one hand, they are carefully integrated into the orchestral sound; on the other, we cannot help but recognise the distinctive timbre of the human voice. By blending them with the instruments, Debussy achieves a magical ambiguity between half-heard human voices and the orchestral seascape in which they float – a bit like a mermaid.

Example 1.1 Claude Debussy, Sirènes, bb. 1–4

As the movement unfolds, the blurring of timbral identity persists: the mezzos are associated with the cor anglais while the entry of the sopranos produces a bright splash of tremolando violins. Debussy’s control of doubling allows the voices to move, back and forth, between being part of an orchestral mix and assuming the foreground. Sometimes this doubling is exact, sometimes rhythmically displaced, but more often than not the voices have an independent line, weaving in and out of the orchestral sounds to create a kind of acoustic flickering. By the end of the movement, the vocal lines have contracted either to single held notes, with the open vowel sound now closed to a hum (bouche fermée), or at rest in simple wave patterns; the half-heard voices merge back into the orchestral whole, like half-seen forms beneath the surface of the water.

Debussy presents us with voices but framed as if not-quite human, voices which do not sing any kind of song since they have neither words nor the kind of melodic patterning that song implies. Constantly emerging from and disappearing into the orchestral ‘ocean’, these voices are as elusive and ungraspable as the mermaids they evoke.¹ In this way, Sirènes, just like the preceding two movements, is a study in saying nothing. Avoiding the daylight world of musical discourse, each of the three Nocturnes elaborates a moving tableau without a viewing subject. If you believe the titles and Debussy’s few words about these pieces, they have to do with clouds (Nuages), nocturnal festivities (Fêtes), and the rise and fall of waves, mixed in with a half-heard siren-call of wordless voices (Sirènes). But these are merely the veils (voiles) that cover the radical emptiness of these pieces – shocking, in terms of the compositional values of the late nineteenth century. Each one says nothing with engaging and seductive charm. As Vladimir Jankélévitch puts it, ‘the song of the enchantresses is a wordless song, the sirens say nothing and address no-one: the sirens do not look us in the eye; and, moreover, they have no eyes’.²

Sirènes was not Debussy’s first use of wordless voices within an orchestral texture (he had earlier experimented, in 1887, with a wordless choir as the ‘unarticulated voice’³ of nature in Printemps) but it is a striking essay in exploring the liminal overlap between words, the singing voice, and instrumental tones. The wordless choir produces all sorts of problems – practical ones (from the economics of concert promotion to acoustic balance in the concert hall), and cultural ones (by the 1960s, the wordless female chorus had become a staple of sci-fi film soundtracks but, even in 1900, the mermaid topic already bordered on kitsch).⁴ For all that, Debussy’s Sirènes demarcates an important cultural exploration of the resonant space between music and language. Its presentation of a pre- or a-linguistic voice connects to a broad body of music associated with the wordlessness of nature from Maurice Ravel to Kaija Saariaho.⁵

There is a long musical tradition of sounding the voice of nature by means of a wordless vocalise (often a solo soprano). In Debussy’s time it joined the decorated lines of the woodbird in Wagner’s Siegfried to the coloratura arabesques of the nightingale in Stravinsky’s Le Rossignol (1914).⁶ In the music of Olivier Messiaen the precisely notated presence of birdsong is often juxtaposed with wordless ‘singing voices’. In the Quatour pour la fin du temps (1940), the second movement (‘Vocalise pour l’Ange qui annonce la fin du Temps’) presents the angel’s announcement as an unending wordless melody ‘sung’ in octave unison by the violin and cello. And in the final movement (‘Louange à l’Immortalité de Jésus’), marked ‘extatique’ and ‘paradisiaque’, the heartbeat rhythm of the piano provides the accompaniment to the infinitely slow melodic lines of the violin’s wordless song. Elsewhere, the intensity of this wordless voice is given by the ondes martenot, an instrument that produces a sound halfway between human singing and something quite other. In the Turangalîla Symphony (1948), both the second movement (‘Chant d’amour I’) and the sixth (‘Jardin du sommeil d’amour’), use the ondes to ‘sing’ on behalf of the mute lovers. Its vocal origins can be heard in Messiaen’s earlier writing for voice, as in the first song of Poèmes pour Mi (1937), where the soprano takes off in extended melismas in the long Alleluia sections. Between the two it is not hard to hear the echo of Debussy’s sirens – on the one hand, the singing voice leaning over into wordless vocalise and, on the other, the imitation of a wordless singer in the ‘voice’ of the ondes martenot. Maurice Martenot’s instrument, invented in 1928, is, after all, predicated on playing between the oscillations of the waves (ondes).⁷

A century or so after Debussy’s sirens, we can hear their echo in Kaija Saariaho’s opera L’amour de loin (2000) in which wordless female voices are again blended into orchestral music that evokes the constant presence of the sea. Like Oltra Mar (1999), a preparatory work for the opera, it presents a music of the waves and tides in which the global containment of harmony and sonority, criss-crossed by repeated running figures and arpeggiations, combine to form a moving stasis, an elaboration of a single immersive space. Just as Debussy does in Sirènes, Saariaho blends wordless voices into the orchestral ebb and flow.⁸ The final movement of Oltra Mar, for example, foregrounds the ambivalence of the siren call – a beautifully seductive music, with its rich textural weave and its constantly fluctuating and modulating containment, it is also the evanescent scene of both arrival and of death. More generally, Saariaho’s music often takes place on the edge of words, exploring the gap between instrumental tones and vocal ones, and between the singing voice and the speaking voice. This is true not only of her vocal music, but also of instrumental works which ask for the performer to vocalise in various ways – witness the exploration of breathing, whispering, and speech in her music for solo flute.⁹

In Speakings (2008) for orchestra and electronics, Jonathan Harvey explores the liminal space between music and language by an astonishing use of the orchestra that continually borders on vocal tone and gesture. By transforming the instrumental tones electronically, ‘vowel and consonant spectra-shapes flicker in the rapid rhythms and colours of speech across the orchestral textures’.¹⁰ In Les septs paroles (2010) for orchestra, chorus, and electronics, Tristan Murail plays off a real choir onstage with a virtual ‘spectral’ choir via the electronics, synthesised voices that sing in microtones and at registral extremes, and which move freely around the performance space. The gap between the visible human choir and the invisible spectral voices foregrounds a play across the border between the human and the not-human, just like Debussy’s sirens.¹¹ Gérard Grisey’s Les chants de l’amour for twelve voices and tape (1982–84), uses words but stretched out in such a way that their function is sonic rather than signifying.¹² Music after Debussy has often played across this gap between conventional word-setting and a radical fragmentation of text that produces a kind of wordlessness even in the presence of words. György Ligeti’s Lux aeterna (1966) makes its minimal text effectively ungraspable through its complex multipart textures, reducing words to purely phonetic material in its slowly shifting clouds of sound. Ligeti’s chromatic micropolyphony might seem quite distant to Debussy’s sirens, yet both foreground the human voice detached from any direct linguistic and signifying function; what was an occasional device of music at the fin de siècle for evoking the presence of a supernatural voice thus becomes a staple of avant-garde music by the 1960s – witness the radically fragmented use of the voice in the work of Berio or Xenakis.¹³

The mellifluous charm of Debussy’s sirens might pose the modern listener less of a challenge than some of these more recent works, but it is perhaps no less problematic. Adriana Cavarero is not alone in warning us of the ‘misogynist overtones’ of this kind of presentation of a ‘seductive, carnal, primitive, feminine voice, which goes back at least to the Homeric Sirens’.¹⁴ Far from the beautiful creatures which tradition has made of them, Cavarero argues, Homer’s sirens, about which Circe warned Odysseus, are monstrous figures – half bird, half woman.¹⁵ Nevertheless, she continues, ‘the charm of the voice, rendered even more disturbing by the absence of speech, still calls men to a pleasurable (and often explicitly erotic) death . . . There is a feminine voice that seduces and kills, and that has no words.’¹⁶ But Cavarero is perhaps too literal here; the myth of the sirens is surely a myth about music and language, the pre-linguistic consciousness and a world made in language, not simply about men and women; the feminised voices of the sirens sing within us all. And as voice, the sirens are heard before they are seen; they are present first to the mind’s ear, which is why their appearance in music is far more unsettling than their rather awkward representations in nineteenth-century painting.

The myth of the sirens projects a profound anxiety, but it is the anxiety of an identity defined through language, one that fears the excess and fluidity of a voice (and thus a body, and thus a world) not ordered by language. The temptation of the sirens, from which Odysseus must steer clear, is a return to a pre-linguistic state. For Adorno and Horkheimer, in Dialectic of Enlightenment, the call of this elemental plenitude is profound but regressive. For them, the sirens are those found in Wagner’s Tannhäuser, or the flower maidens of Parsifal: ‘Their allurement is that of losing oneself in the past. But the hero to whom the temptation is offered has reached maturity through suffering.’ The Wagnerian hero, for Adorno, defends the achieved identity of the integrated bourgeois subject against the ‘narcotic intoxication’ of the sirens and their tendency to disintegration.¹⁷ And here we come close to Adorno’s problem with Debussy and those who followed in his wake – his stern refusal of tone (Klang) and an insistence on grammar that verges on a neurosis born of repression. He even seems to want to tell us as much in language that suggests a kind of self-mutilation: ‘Men had to do fearful things to themselves before the self, the identical, purposive, and virile nature of man, was formed, and something of that recurs in every childhood. The strain of holding the I together adheres to the I in all stages; and the temptation to lose it has always been there with the blind determination to maintain it.’¹⁸

We do not need to share Adorno’s anxiety to understand it; as Cavarero puts it, pleasure in the acoustic ‘destabilizes language as a system that produces the subject’.¹⁹ Undoubtedly, Adorno and Horkheimer’s position represents a fear about the loss of self, about ‘the risk of regressing to the joy of a pre-semantic stage for which it still feels nostalgia. It is as if the vocalisation of the in-fante called out behind the adult, returning the self to the pleasure of the ear and phonetic emission.’²⁰ But we should perhaps not be too quick to reject their position; Jankélévitch, no opponent of the pleasures of music, is clear that the siren voices in Debussy’s music are not to be followed and, in this, comes close to Adorno.

The depth to which the Sirens draw us is a somewhere which is nowhere; this depth is a snare and a deception; a false depth! Or if we dare transpose this to another dimension: the man tempted by the Sirens becomes something which is nothing; his becoming is a pseudo-becoming and opens out onto non-being [ . . . ] that is to say, he does not become.²¹

It is, for Jankélévitch, a closed nothingness: ‘an impasse, a dead-end road, a time without perspective and without hope’.²²

I do not hear this in Debussy’s Nocturnes. It may certainly be one aspect of the sirens’ voices to draw us towards an abyss, but there is surely another, which Debussy explores in both L’isle joyeuse and in La mer. Here, the call of the sirens is not a descent to nothingness, but the upward leap of a kind of aural/oral jouissance that characterises a state of being without language. What is L’isle joyeuse except a celebration of this libidinal excess embodied in the siren’s voice – one woven into the plural genesis of the work that includes Watteau’s painting Le Pèlerinage à l’Isle Cithère (1717), Debussy’s elopement to the island of Jersey with Emma Bardac in the summer of 1904, and (for Michael Klein), Chopin’s Barcarolle?²³ Debussy’s music here richly embodies what Cavarero calls the ‘libidinal register of the vocal’, the vocal excess largely regulated by the infant’s passage to adult language, but still alive in poetry where ‘the semiotic drives of the phonic find some fissures through which to invade language and disturb it with the agitation of its rhythms’.²⁴ Given that, in French, the ubiquitous play between la mer and la mère is no mere pun, but a recurrent topos of poetry and literature,²⁵ we should not be surprised to find Debussy’s explorations of this ‘sonorous, presemantic source of language’²⁶ in works connected to the sea.²⁷

Simon Trezise notes that ‘the main motifs of La mer are imbued with a quality of incantation, of ancient voices crying out from the depths of the oceans’, pointing to several writers who ‘refer to the first cyclic motif as the “call of the sea” or as a “melancholy call” ’.²⁸ More specifically, a number of commentators, including Pierre Boulez, have remarked on how this call of the sea relates back to that of the sirens.²⁹ Jankélévitch hears the three-note descending motif of the woodwind in the third part of La mer (b. 56) precisely as the whisper of siren voices (see Example 1.2) – ‘Viens à moi’ (come to me) they seem to sing, he suggests, in a ‘song of a seduction which leads us vertiginously, irresistibly into abyssal depths’.³⁰

Example 1.2 Claude Debussy, La mer, ‘Dialogue du vent et de la mer’, bb. 56–59

There is other musical evidence for considering Debussy’s first orchestral triptych as intimately related to the second,³¹ but what binds them together, beyond specific motivic or gestural similarities, is this shared aesthetic of saying nothing, or what Jankélévitch points to as a ‘seduction of underwater depths’, that runs through Debussy’s music from Sirènes to the water nymph Ondine in the second book of piano preludes.³² In other words, the sirens we hear in the last movement of the Nocturnes announce a key theme of music ‘after Debussy’, a theme certainly not confined to the relatively rare use of wordless voices. It is a topos that not only pervades Debussy’s instrumental music but also exceeds any suggestion of representation. The sirens’ voices link to something pre-semantic which evokes the in-fans, one who is without speech but not without voice. Indeed, the pre-linguistic infant is characterised by an excess of vocality whose loss, if not repression, is coterminous with the idea of maturity into adulthood – witness the stark warning of Hans Christian Andersen’s tale of the Little Mermaid for whom the price of entering the human (adult) world is not only the loss of her (siren) voice but the painful and bloody bifurcation of her fish tail into legs. On the one hand, for Adorno and others, the siren’s voice is a call to regression (as the sound of a pre-linguistic condition), on the other it remains haunting and enchanting precisely because of its a-linguistic nature. Jankélévitch uses the term balbultiant (babbling) for the ‘saying’ of music, underlining this connection with the world of the in-fans.³³ Infants babble – that is to say, they vocalise without semantic content. They say nothing while saying constantly, and in highly musical ways. Cavarero describes the wordless vocalisations of infant-mother exchanges thus:

This is an acoustic-vocal relation in which, importantly, what gets said is, as yet, nothing. There is not yet any signified in this voice – no reference through the linguistic sign to the noetic presence of an absent object. Materialized by the physicality of the vocal exchange, the only presence is the act of the relation.³⁴

I am not suggesting that music is equivalent to the babbling of infants or the vocal responses of their mothers (a practice linguistic theory denotes as ‘infant-directed speech’, or IDS), but rather that music amplifies a use of the voice that exceeds the signifying function of language. This is uncontentious in that music has always foregrounded vocal excess, from the Italian madrigal with its sobs and stutters, and cries of pain and pleasure, to operatic coloratura and vocalise.³⁵ My rather different point is that a music ‘after Debussy’ draws attention to the material medium of music’s ‘saying’ not to accentuate any linguistic or signifying content but, like the siren song or the infant-mother exchange, as a kind of touch or contact, the phatic condition of all speech.

It is not insignificant that, to demonstrate his insistence that artworks do rather than say, Jankélévitch points to Rimsky-Korsakov’s operatic heroine Fevroniya who sings like a nightingale, ‘and just as the nightingale produces vocalise, rather than lectures about vocalise, so Fevroniya knows only how to sing; her business is to do, not to discourse’.³⁶ He might just as well have pointed to the avoidance of all saying in the coloratura vocalises of Stravinsky’s ‘real’ nightingale in Le Rossignol, composed a few years later, or to Messiaen’s birdsong as the vehicle (and legitimation) for elevating wordless voices to the highest aural prominence and structural significance. If birdsong comes to occupy an important place in twentieth-century music it is less a matter of mimesis or symbolism and more a question of foregrounding a ‘babbling’ that threatens to drown out the orders of musical language. But there is nothing new about this topic; Elizabeth Eva Leach shows how the elision of birdsong and siren song was already made in medieval music theory. She sums up the anxiety of the (male) medieval scholastic thus: ‘If a man allowed inappropriate, especially effeminate music to act upon him without engaging rational judgment, the passive nature of his listening would feminize him’, an effect that in turn leads to a kind of ‘bestialization’.³⁷ There is a striking similarity here between the discourse of medieval monastic culture and Horkheimer and Adorno; the siren song is immoral and destructive because it threatens the singular achievement of rational humans over irrational animals, a self-divided antagonism couched in specifically gendered terms. Music theory thus assumes a task of regulative control that exceeds merely aesthetic concerns; it has to do with an ordering of the sensory by means of the rational.

It is perhaps what is lost in the process of such rational control that Cavarero laments in language itself, in the way it ‘exploits, reduces, and regulates the marvellous exercises of the infantile voice. Stripped of its excesses and its imagination, the infant’s emission is frozen into the syllable and tones that language permits.’³⁸ But the exploration, within poetry and music, of the play of language between sens (sense) and son (sound) is no regressive return to an infantile phase; music, like the voice itself, is a ‘pivotal joint between body and speech’.³⁹ Which brings us back to the ambivalence of the siren’s voice. For Jankélévitch, music’s charm relates to Orpheus, for whom music ‘harmonizes and civilizes’, rather than the sirens whose goal is always ‘to reroute, mislead, and delay Odysseus’ on his journey ‘towards duty and truth’.⁴⁰ But George Steiner hears the voice of the sirens differently. The fable of the Sirens, he suggests, is about music itself, which ‘lies in wait for the speaker, for the logician, for the confidant of reason (Odysseus par excellence)’. He goes on:

The Sirens promise orders of understanding, of peace (harmonies) which transcend language. The language-animal, man, armoured in his will to power which is grammar and logic, must resist. He must deafen himself to the solicitations of the song. Otherwise he will be drawn out of himself – the ecstatic motion – to some irremediable sleep of reason.⁴¹

For Steiner, this siren call is present in the musicality of all language, a sound that ‘is always threatening to pull after it, with the force of the ebbing tide, the servile stabilities of sense’.⁴² From such a perspective, the sirens tempt the listener to a perilous place beyond language – an abyss, an island (as in Rilke’s ‘Die Insel der Sirena’), or a silence (for Kafka, it is precisely ‘The Silence of the Sirens’ that is so terrifying).⁴³

Aristotle makes the voice (the sounding body) secondary to the signified idea it carries.⁴⁴ It has no value other than to carry the semantic, otherwise it is no more than an animal noise (or, presumably, the babbling of infants). The history of philosophy from such a perspective, as we observed in the Prologue, is thus coterminous with the devocalisation of logos, a tradition that from Plato onwards equates a rise in signifying content (the ideal) with a reduction in the phonic material – an idea of language that Hegel carries through into his notorious gradation of the arts (the more material, the less spiritual). In Plato, suggests Laura Odello:

The timbre and the voice – that is to say, the sonority – are subordinated to the logical ideality of things: the Platonic gesture is as obvious as it is decisive for philosophy. The philosopher does not want to know anything about listening; hearing is of no use to the philosopher because the end as well as the beginning of all philosophizing resides in the intellectual understanding: sonorities are superfluous, secondary, unnecessary for the silent vision of the logos, of the idea, of the ideal signified, which also serves to grasp the truth.⁴⁵

It is not that philosophy is somehow wrong; merely that it presents only one kind of position and one kind of language use. It tries to remake Rodin’s La cathédrale with only one hand. Its opposite, ventures Cavarero, would be the central character in ‘Funes the Memorious’, a short story from 1942 by Jorge Luis Borges. For Funes, the world is unique in every instant and irreducible to the generic signifiers of language. But for all the epiphanic intensity of each and every experience as utterly new, to be so outside the collective logos is a kind of madness and an impossible way to live.⁴⁶ Is one position any more unbalanced than the other – the extreme empiricism of Funes any more than the extreme idealism of Plato?

Shipwreck and abyss

If the siren voices of a wordless, pre-linguistic nature entice the unwary listener towards danger, like a sailor towards the rocks, what exactly is that danger? What is the nature of this shipwreck and drowning of the self of which the myth warns us? It is a theme that pervades the work of Stéphane Mallarmé, whose poetry is haunted by the siren figure, the threat of shipwreck and drowning. For Mallarmé, as Jacques Rancière observes, the idea of poetry itself is a linguistic adventure on the edge of shipwreck, an idea that runs through Mallarmé’s Poésies from its opening poem, ‘Salut’.⁴⁷ Here, the poet toasts his fellow poets as the crew of a ship, fearless beneath their white sail and ready to ‘cleave wintry seas of blast and gale’, all the while acknowledging that the poem is ‘nothing’, no more than foam (écume) – both that of the glass he raises and that of the sea in which ‘plunges far away a body of sirens’. But the ship of poetry launched in the first poem, Rancière suggests, is sunk in the penultimate one. ‘À la nue accablante tu’ presents several images of shipwreck – the mast stripped bare, the gaping of the abyss, a drowning, and the flank of a siren in the white foam. And who is the siren? For Rancière, she is the figure of fiction itself, ‘the beautiful power of artifice’ that momentarily joins two worlds together.⁴⁸

But this is far more than a theme of Mallarmé’s work. As Roger Pearson has shown, the shipwreck first and foremost occurs in the syntax of Mallarmé’s poetry, famously so in the case of ‘A la nue accablante tu’ which has no punctuation and relies instead on spatial organisation and enjambement. The shipwreck, Pearson suggests, ‘is no mere analogy for the process whereby the “vessel” of language ceases to “convey” when stripped of its traditional representational function’; even the word shipwreck (naufrage) suggests a breaking up of the ship in the coming together of eau and rage.⁴⁹ For Pearson, ‘the “sirène” is at once an agent of death and a means of renewal and salvation: and the poem as siren song both destroys everyday language and offers the prospect of poetic beauty’.⁵⁰ The abîme to which the ship sinks is thus ‘the bottomless, empty abyss of a language that has cast off from the shores of reference’.⁵¹

This resonant imagery is central to Mallarmé’s late and most radical poem, Un coup de dés, published in 1897, the same year in which Debussy began the orchestral Nocturnes.⁵² No longer the master of his ship of language, it takes a dramatic shipwreck for the mariner-poet to find the transformation for which he voyages. The structure of language itself is here lured towards destruction by the siren song of a musicalised language, breaking up the vessel held together by the ordering planks of grammar and syntax, and giving way to something fluid and mobile – a polysemic world of infinite difference, beautiful but dangerous, that loosens the grip of the mastering ‘I’ upon language.⁵³ This is famously presented by Mallarmé’s dissolution of the usual visual order of the printed page. Un coup dés falls across the page like the sinking ship towards the abyss, its lines falling apart, breaking up into multiple strands and reconfiguring its semantic order as it does so. This dissolution is, for the poet as for the mariner, both a death and a setting free. At the poem’s still centre is the edge of the abyss, bounded on either side by the ‘comme si’ (as if) of all art, where the solitary act of writing encounters the juxtaposition of the blank white of the page and the infinite possibility of the sky (see Figure 1.1).

Figure 1.1 Stéphane Mallarmé, Un coup de dés, pages 7–8

Mallarmé’s metaphor of shipwreck is central to Pierre Boulez’s magum opus, Pli selon pli – portrait de Mallarmé (1959, rev. 1983). The third of its three ‘Improvisations sur Mallarmé’ takes Mallarmé’s sonnet ‘À la nue accablante tu’ as its starting point, a poem that similarly drowns linguistic grammar in its exploration of the shipwreck of language – key markers in the sonnet include naufrage (shipwreck), épaves (wrecks), le mât dévêtu (the stripped mast), perdition (destruction), se noyé (drowned), sirène (mermaid). Discarding all punctuation, it was famously singled out by Tolstoy as an example of a poem with ‘no meaning whatever’.⁵⁴ In Boulez’s ‘Improvisation’ the soprano does not sing the words of the poem – or at least, she sings only the first word, which, being ‘A’, is the default vowel sound of wordless singing. Her five entries in the piece are all extended vocalises on this single syllable, elaborating a musicalised voice at the edge of language. This, surely, is another instantiation of the wordless voice of the siren, shorn of all late nineteenth-century kitsch. Its lyrical and self-sufficient play might also recall the ecstatic wordless singing of Stravinsky’s nightingale in Le Rossignol. Its cyclic returns are like those of the tides, a coming in and out of presence that contrasts with the apparently timeless submarine space of the orchestral interludes. There is no saying here, no narrative, no musical adventure, just the movement of an aquatic voice against the stripped-back wreck of linguistic grammar. Sudden percussive attacks alternate with long resonances, silences are interspersed with the gentle shimmering of celesta, harp and mandolin. The long, drawn-out instrumental postlude of quiet noise and timeless resonance seems to take us back to the opening bars of Debussy’s La mer.

Mallarmé famously anticipates Derrida in the way he opens out a vast space that becomes visible only when the ship of language breaks apart. The white spaces of the page are, at one and the same time, the space of possibility and the dark abyss of language. Like Derrida, Mallarmé treats language negatively in order to show that its operations do not add up to a metaphysical closure (in which the word would denote the presence of a thing) but rather opens outwards and point beyond its own limits. In their different ways, both use language to delimit its own margins. Derrida’s terms are famously negative (erasure, deconstruction) in the same way that so many other key ideas of aesthetic and philosophical modernism are negatively defined (atonality, abstraction, negative dialectics). The sense of opposition arises from the instinct to clear a space by breaking up the restrictive nature of old forms – ‘to philosophize with a hammer’ as Nietzsche put it, anticipating Boulez’s composing with one in Le marteau sans maître. Instead of making a kind of linguistic enclosure, and insisting that only what lies within the walls of language is true and real, such an approach uses the walls to define the space outside of them (much like Kierkegaard’s metaphor of the border between two countries). Derrida’s philosophical language-work here thus converges with Mallarmé’s poetic work, and both meet Debussy’s music as it arrives from the opposite direction.

The abyss opens where language ends abruptly, like a cliff edge, dropping off into the unplumbed depths of the a-linguistic. The poetry of Mallarmé takes place on this cliff-edge, holding out its words over the edge, risking their footing within language by allowing their play with the sensuousness of rhythm and sonority. The risk of falling is thus a kind of loss of the linguistic self, guaranteed and shored up by language, the self that manages to carve a path across the ocean of the natural world for as long as it is held within the ship of language. But the shipwreck threatens all of that. The breaking up of the ship risks the drowning of the linguistic self – hence the omnipresence of the siren in Mallarmé’s poetry. As Stefan Hertmans suggests, such radical wordlessness is ‘a gorge, a gap, a void or a division’, an ‘abyss’, ‘a chasm one cannot look into without getting giddy’.⁵⁵ For Mallarmé, the poet stands on the vertiginous edge of this abyss, braving the terror of a fall in order to make appear the luminous silence to which poetry aspires.

It is an approach that closely relates the work of Debussy and Mallarmé, a parallel between their aesthetic projects that goes far beyond Debussy’s few song-settings of Mallarmé or the Prélude à l’après-midi d’un faune. Jankélévitch identified in Debussy’s music a defining ‘geotropism’, by which he meant a recurrent tendency to fall. His exhaustive analysis of the astonishingly varied kinds of descent in Debussy’s music makes ‘La descente dans les souterrains’ by far the longest chapter of his Debussy book. Taking its title from Act 3.ii of Pelléas et Mélisande, Jankélévitch reads the whole of Debussy’s opera as a kind of descent: ‘At first sliding gradually then in an accelerating fall, the destiny of Mélisande and Pelléas descends vertiginously towards the abyss, beyond control or redress.’⁵⁶ It is a central image, of course, in La chute de la maison Usher (The Fall of the House of Usher), the unfinished opera based on Edgar Allen Poe’s short story on which Debussy worked for some ten years, on and off.⁵⁷ At the heart of Debussy’s music, Jankélévitch contends, there is an ‘obsession with the abyss’,⁵⁸ an emptiness which ‘is both the object of horror and temptation, and for that reason gives us vertigo’.⁵⁹ To be sure, Debussy’s music depicts no shipwrecks, though there are both journeys by boat (L’isle joyeuse) and some violent storms (La mer), but, as with Mallarmé, it is within the use of (musical) language itself that the abyss opens up, in the spaces that appear when music allows itself a kind of freedom from its earlier grammatical orders. It is on this level that Debussy’s music relates most powerfully to the boundless self-sufficiency of the aquatic – not because his titles tell us so, but because his music unfolds immersive musical structures whose harmonic and sonorous continuities have let go the firm footing of tonal ground.

Mallarmé was hardly alone in equating the allure of a musicalised poetry with that of the sea, or the adventure of language as akin to embarking on a voyage towards the sea’s distant horizon. ‘Music often takes me like the sea!’, Baudelaire declares in the first line of ‘La musique’. It is not just that music is experienced as a kind of transport to distant places, but that it moves in a quite different element. Its rhythmic transport is not confined to the hard enclosures of the land, but floats across the unbounded surface of the water. Such a poetic view of music being like the sea was in turn reclaimed by musicians as song texts and thus the site of new kinds of musical space. In Fauré’s early song, ‘Les matelots’ Op. 2, no. 2 (c.1870) to a text by Théophile Gautier, the poet voyages by means of language, like a mariner in his ship: ‘Sublime existence/ rocked within our nest/ we live over the abyss/ in the bosom of the infinite’ (Existence sublime / Bercés par notre nid/ nous vivons sur l’abîme / au sein de l’infini).

It is a theme that recurs periodically in Fauré’s work, right up to his late set of songs L’horizon chimérique, Op. 118 (1921), to poems by Jean de La Ville de Mirmont. All four of these songs exemplify Fauré’s use of the piano as a constant background that nevertheless constantly changes, a paradox created by means of an unchanging ostinato figure subjected to infinite nuances of harmonic shading. ‘The sea is infinite’ (La mer est infinie) is the opening line of the first song: ‘my dreams dance over its surface like drunken birds, carried on the vast motion of the waves, shaken and tumbled in the folds of the breeze’ – a metaphor embodied in the way that the vocal line is carried aloft by the constancy of the semiquaver accompaniment figure, endlessly varied in its harmonic folds (see Example 1.3). It shadows the voice but without interaction, response, comment or discourse, as a silent travelling companion like the sea to the sailor. This is a song that foregrounds an a-grammatical play on the liminal edge of sea and air, moving eccentrically outside of the straight lines of grammar – a release from the symbolic order, a play at the margins and thus a celebration of the power of music and its seductive pull towards an a-linguistic self.

Example 1.3 Gabriel Fauré, ‘La mer est infinie’, L’horizon chimérique, Op. 118, no. 1, bb. 1–7

The rhythm of the waves defines the movement of the second song, ‘I have set out’ (Je me suis embarqué), marked by Fauré in a constant dotted figure in the left hand of the piano, and syncopated chords in the right, which, in the opening bars of the song, leaves the last beat of each 3/4 bar empty, a rocking that embodies the singer’s sentiment that ‘the supple waves have taught me other cadences, lovelier than the weary rhythm of human songs’. This rhythmic effect is mirrored in the chromatic slippage of the harmony, a gentle disorientation of the tonally grounded self.

The desire of all these songs is for a different state, a different space of language that recalls the pleasure and freedom of the in-fans: ‘I want nothing but the sea, nothing but the wind / to rock me like a child in the trough of the waves’. Not a confusion or unstructured nonsense but a state of calm, self-sufficient speechlessness: ‘O moon, I want your clarity’, yearns the singer in the third song, ‘Diane, Selène’. The utter simplicity of the slow piano chords and the mute quality of their non-functional circling underline that this is all about not-saying. The piano speaks the interior silence of what the words lament; they voice the silence for which the words yearn. Each points to what the other cannot. But it is desire that ends this short cycle. ‘Ships, we would have loved you’ (Vaisseaux, nous vous aurons aimés) explores the tension between two worlds – between being landlocked by language and the unstilled desire for the great voyages promised by poetry: ‘For I have great unsatisfied departures within me’, is the gentle lament with which Fauré ends his song and, symbolically at least, his life as a voyager across the space of music and poetry.

Constellation

Mallarmé’s poem does not end with the drowning of the mariner on the bed of the ocean. Its still centre (the lost and lonely quill) is followed by a vision of a mermaid who slaps aside with her tail the hard structures (faux manoir) that had imposed a limit on the infinite (qui imposa une borne à l’infini). ‘Down falls the quill’ (choit la plume) in the face of what is revealed in its place. One polyphonic trajectory of the poem muses on the transformation of the mariner/poet in the face of this radical dissolution while another, separated by its larger, bold and upper-case typeface, slowly coalesces to deliver a single line with the force of an epiphany: ‘NOTHING . . . WILL HAVE TAKEN PLACE . . . OTHER THAN THE PLACE . . . EXCEPT . . . PERHAPS . . . A CONSTELLATION.’ The result of the shipwreck of linguistic order, the unstitching of its signifying forms, is a clearing of space within language itself. In place of the solidity of things, one can now see the patterning of relations between things, the spacing of form that is their constellation. It recalls the effect of lines by the seventeenth-century Japanese poet, Mizuta Masahide: ‘Now that my barn has burned down, I can see the rising moon.’

Elizabeth McCombie, in her book-length study of Mallarmé and Debussy, points to the ‘grandeur and finality’ of this ‘symphonic conclusion’ to Mallarmé’s poem, with its ‘visually and semantically dramatic ascent towards the shimmering “CONSTELLATION” of the last page’.⁶⁰ In this, she compares Mallarmé’s poem to Debussy’s Jeux, but to my ears it is surely the ‘grandeur and finality’ of the ‘symphonic conclusion’ to La mer that offers the most resonant musical parallel. On one level, the chorale ending evokes a musical gesture all but worn out by the time of Debussy’s work, suggesting a kind of symphonic language Debussy otherwise distances. But, as in Mallarmé, the ‘grandeur’ of the ending is less the arrival of some thing than the luminescence of appearing itself, less the proclamation of some discursive conclusion than a resonating of the space that has been opened. Malcolm Bowie draws an explicit link between Un coup de dés and La mer: both place centre stage, he suggests, the contradiction between the sea as a ‘spectacle of endless possibility and plurality’ and as an ‘emblem of intellectual defeat and of humankind at the mercy of an inhospitable world’.⁶¹ To be sure, ‘Dialogue du vent et de la mer’ does not depict any shipwreck, but the violence of the sea and its ‘in-human’ aspect have been heard by many commentators. Like Mallarmé’s poem however, Debussy’s music ends not with some a-linguistic abyss but with the shimmering ‘constellation’ of its closing bars.

McCombie’s reference to Jeux reminds us that Debussy’s music, like Mallarmé’s poem, presents a mosaic-like patterning of recurrent but endlessly varied figures whose unfolding is without obvious teleology. Taken as a whole, the six movements that make up the Trois nocturnes and the Trois esquisses symphoniques of La mer form a kind of constellation of related but separate movements, underlined by some specific motivic connections across separate works as well as within them. The sense of a larger set of pieces is reinforced by further suggestive links to both D’un cahier d’esquisses and L’isle joyeuse from 1903.⁶² Simon Trezise, for example, underlines how D’un cahier d’esquisses is not only in the same key as La mer (D♭) but also ‘makes extensive use of a rhythmic figure that dominates the cello theme of the second principal section (first movement): there is, therefore, a possibility that this piece, about whose genesis little is known, is a spin-off from the composition of La mer’.⁶³ Aside from such specific material points of contact, it is not insignificant that both works are described as esquisses.

Analysts have made much of Stravinsky’s debts to Debussy in this respect, seeing the French composer as the origin of the idea of ‘block form’ which comes to define, through Stravinsky, an attitude to musical time in the twentieth century quite different to the linearity of an earlier age. The metaphor of the ‘block’ emphasises something defined by discrete and hard edges, as in Stravinsky, but the structural effect in Debussy is more fluid – associative rather than oppositional, working with allusive links rather than hard juxtaposition. Messiaen, who learned from both composers, is rather less like Debussy than he is often assumed to be, but at times displays a similar kind of musical logic. In the ‘Regard du silence’, for example, No. 17 of the Vingts regards sur l’enfant Jesus, he presents a kaleidoscopic play of crystalline chords to form a structure that is neither discursive nor propositional but constellational. As Messiaen’s Vingt regards exemplifies, the new formal principle is also echoed on a macro level in terms of large, multi-part forms: it is striking how often, in music after Debussy, composers have found a way out of the teleological structures generated by tonality by reaching for forms made of multiple parts, works often consisting of several, shorter movements, rather than the narrative linearity that is the legacy of the sonata, symphony, or concerto – witness many works of Messiaen, Saariaho, Dutilleux, and Boulez. Debussy’s two books of piano preludes might be heard in this way, neither an arbitrary and disconnected set of pieces nor a strictly ordered sequence like the four movements of a symphony, but a constellation of individual pieces that make a larger pattern in relation to one another. The same may be said of Debussy’s song collections, so often, as David Code has noted, presented in groups of threes.⁶⁴

Given the ephemeral nature of its material, music might be thought to lend itself well to a kind of sense-making that is made of nothing but relations – that projects the constellation of its dynamic parts, rather than anything of fixed substance. The charme of music, in the words of Jankélévitch, ‘thus stems from a kind of infinite totalisation and it is in this sense that it is literally made of nothing, that it comes from a nothing!’.⁶⁵ The very idea of a ‘spectral’ music is rooted in the idea of constellation – to understand sound as a complex structure of overtones is to understand it not as an opaque concrete object but as a set of relations defined by interval and intensity; in the possibilities made available by computer technology, it is a constellation that can be expressed precisely in terms of number. But in their working out, spectral compositions proceed by exploring not linguistic structures but spatial ones, defined by global, spatial conditions of relationality rather than musical objects. They amplify the patterns that emerge between sounds and within sounds once the structures of language have been dissolved.

The constellation is a resonant image for music on many levels; rhythm, melody, harmony, and form exist only on account of interval and the patterning of intervals, a network of differences like language itself. Such a patterning reveals itself in music only through time, as a continuously unfolding edge of the temporal moment, running like an elongated wave from start to finish, yet leaving its trace as it goes. Not for nothing did Henri Bergson look to music for an embodied image of his idea of the durée as opposed to an atomised idea of time based on the separation of spatial co-ordinates. Try to grasp music as a set of things or objects and you end up only with ‘bassoon’, ‘B♭’, ‘minim’, not the music formed from the constellation of all its parts.

But while music is essentially constellational, defined by its patternings, it resists being reduced to endless différance. It is striking that the development of the ‘Method of Composing with 12 Tones’, developed by Schoenberg and his circle in the decade after Debussy’s death, is entirely based on this idea of interval, on the gap between things, and yet is quite opposite to the musical tradition that emerges with Debussy. The difference has to do with a system based only on intervals (the abstract law of number) and a kind of ‘logic of sensation’ that emerges between things. In serialism, the identity of any musical object is conferred by the system; in Debussy, the constellation of relations arises from the sensuous given (donné) of the musical materials. They move in opposite directions: one makes musical sounds the carrier of the abstract system, the other starts with sounds in motion to discover a constellational form.

It is telling that Debussy, trying to articulate his discovery of a new kind of harmonic grammar to Ernest Guiraud in 1889, should reach for an aquatic metaphor: ‘In submerging tonality (en noyant le ton), one should always proceed where one wishes, one can go out and return by whatever door one prefers. And our world, thereby expanded, is capable of greater nuance.’⁶⁶ Debussy’s phrase ‘en noyant le ton’ is just as well translated as ‘in drowning tonality’⁶⁷ – not just an immersion but a kind of re-making, such as follows the drowning of the mariner in Un coup de dés. For Debussy, this new ‘aquatic’ space of harmony brings with it a kind of ‘polymorphous’ quality (to borrow Freud’s term); compared to the phallocentric law of tonal direction, in Debussy’s world one can enter or exit harmonic space in multiple ways. The aquatic metaphor captures well how Debussy dissolves the hard structures of tonal grammar and reorganises them by a different, more fluid logic.

Musicology often remains tongue-tied in relation to this different logic. More than a century on, it still falls back on phrases about Debussy’s ‘static’ harmony, even though this music is palpably full of movement and, in the face of Debussy’s astonishingly rich and multiple harmonic connections, often resorts to the negative and utilitarian label of ‘non-functional harmony’. On the one hand, it’s a reductive and ridiculous label, suppressing bewilderment that music might deploy triadic harmony to any other end than that of tonal progression towards closure. On the other hand, the negative label unwittingly underlines that Debussy’s triadic but ‘non-functional’ harmony constitutes a kind of tonality under erasure; it undoes the terms of tonality precisely in order to point elsewhere, to open to a different space.

The discourse of the young Boulez, in the 1950s, could hardly be further from the world of Debussy; its bristling array of quasi-mathematical techniques, its aesthetic of asceticism and apparent rejection of pleasure, hardly recall Debussy. And yet the soundworld of the later Boulez suggests a quite different picture and a proximity between the two composers that music history has too often obscured. In the closing section of Répons (1984), the shimmering arabesques of arpeggio figures in harp, pianos, cimbalom, vibraphone, celesta, and their transformed electronic echoes, suggest a cultivation of the sensuous aspects of the sonic that relates directly to passages in Debussy or Ravel.⁶⁸ Hearing the later works in this way invites a rehearing of the earlier ones – not just the more obviously indebted juvenilia but classic serial works of the 1950s, like Le marteau sans maître and the Mallarmé-based Pli selon pli.

Boulez was explicit that in Mallarmé’s Un coup de dés he found a rich model for the structures of his own music. Peter O’Hagan cites a letter from Boulez to Stockhausen of September 1957 in which he states that ‘Constellation’, which forms part of the Third Piano Sonata, ‘is a little like Mallarmé’s Un coup de dés with structures playable in all directions following the layout of the page’.⁶⁹ Influential in this respect were Mallarmé’s posthumous notes, published by Jacques Scherer in 1957 as Le ‘Livre’ de Mallarmé.⁷⁰ For a composer exploring the permutational possibilities of the number twelve, the resonance with the throwing of two dice is obvious (2 x 6). Mallarmé’s text consists of 24 pages (the number in a printer’s cahier), each of which has space for a maximum of 36 lines; 24 is also the number of syllables in a couplet of alexandrines.⁷¹ But the possibilities suggested to Boulez by Mallarmé’s own fascination with numerology are perhaps less important than the larger poetic move here, of stepping away from representation through a play of language itself. Mallarmé’s move from book to album (a collection of loose sheets) can certainly be found in Boulez too, in the mobile forms of the 1950s, insisting on the piece as a kind of ‘unfolding’ of the text rather than a ‘folding up’ and binding together as the making of books is usually taken to be.

Nowhere is this idea more obvious than in the Third Piano Sonata (1955–57), on which Mallarmé was a key influence.⁷² In Constellation-Miroir we find an exemplary case of the coming together of quasi-mathematical wonder at the system of relations and the embodied pleasure of sound. The Third Sonata exemplifies the idea of an open work, a text that remains a ‘work in progress’ (a notion Boulez borrowed from James Joyce). In its five-movement plan, Constellation-Miroir forms the central point.⁷³ The other four movements have titles that refer to ways in which music takes on structures and modes of address drawn from language: Antiphonie, Trope, Strophe, Sequence. The same literary and linguistic ideas abound within the movement; the structural plan for Strophe is based on sonnet form; the four sections of Trope, which can be played in a variety of different orders are titled ‘Text’, ‘Parenthesis’, ‘Gloss’, and ‘Commentary’. These four sections are not bound in to each other, but printed on four loose leaves to facilitate their re-ordering by the performer. Additionally, they use red and green print to differentiate between ‘blocks’ and ‘points’ of material. The fascination, shared with Mallarmé, with the idea of constellation, mirror, symmetry, and mobile forms is surely to be understood as part of a wider shared aesthetic outlook, not merely as an arbitrary borrowing of techniques. As Peter O’Hagan underlines, for Boulez, by the mid-1950s, the triumvirate of Mallarmé, Cézanne, and Debussy were simply ‘the source of all contemporary art’.⁷⁴