Generate a text only version of this page
Birkbeck, University of London home page
Help with using the Birkbeck web site
PORES - A Journal of Poetics Research

The Voice as Instrument: considering the semantics of text as an element of musical discourse

- Edward Nesbit


Last year I was a student on the Masters course in Composition at Guildhall School of Music and Drama, and one module on that course – the ‘Voiceworks’ module – involved the student composers collaborating with experimental poets, mostly from Birkbeck College.  It was in this context that I met the poet Steve Willey 18 months ago, never having engaged with experimental poetry before; we collaborated on the Voiceworks module for 2007-8, and again during Voiceworks 2008-9.  I will discuss the two pieces that we produced: Portmanteaux, which was premiered in Wigmore Hall in April 2008 and given a repeat performance at Openned Poetry Readings that May, and Soundings, which was premiered in Guildhall School of Music and Drama in April this year and performed again at Wigmore Hall a week later.

Before working with Steve, my exposure to the joining of words and music had come principally through the English art song tradition, which typically prioritises the words over the music as the primary locus of expressive content.  This tradition grows out of the tradition of the nineteenth-century German lied, where this hierarchy was even more extreme.  No less a poet than Goethe objected when the music compromised the audibility of the words, or did anything else which promoted it to being more than a kind if ‘backing track’.  This approach to the relationship of words and music is reflected in the attitudes of classical singers: they first read the words and engage with them fully before even looking at the music.  This approach seems to relegate the music to mere accompaniment, as a means of ‘prettifying’ the text.  However, it seems to me that this is not at all how a song ‘means’, and a large proportion of the expressive content of the song comes from the music.

While it was partly this dissatisfaction with the received classical tradition of word-music relationships which led me to seek a more equal relationship between the two in my collaboration with Steve, practical considerations also played a large part.  Music has a very established vocabulary of gestures with which to represent the images of conventional poetry.  Take, for example, Schubert’s two great song-cycles: Die Schöne Mullerin (The Pretty Milk Maid) is characterised throughout by arpeggios in the piano accompaniment which represent rippling water; conversely Winterreise (Winter Journey) is suffused with repeated chords which evoke a man walking.  When dealing with the more abstract world of experimental poetry I did not have to hand musical figures which could represent the text in this rather literal manner; it was necessary to find other ways in which text and music could correspond.

Musical example: Soundings
Hand-out: Mercurial ArticulationsIn Soundings, Steve and I attempted to translate poetry into music mainly in terms of the visual tensions of the poem.  At the bottom left of the panel is a wedge of very dense, almost wild, text, and out of (or into) it come three prongs, which Steve conceived as being in motion, rendering the whole panel dynamic rather than static.  Moreover, the two spherical objects in the corner complicate an otherwise clear structure.  It was this reading of the poem which we attempted to translate into music.  The piece is scored for two sopranos, two clarinets and two violas, which are arranged symmetrically along the length of stage: viola-clarinet-soprano-soprano-clarinet-viola.  The ensemble is initially divided into two halves, with the three performers to stage right performing fast, chaotic music, corresponding to the wedge of text at the bottom left of the poem, while the three performers to stage left performing slower, calmer music which corresponds to the three wedges of text in the poem.  Once this dichotomy has been established, the viola at stage right plays a glissando which is immediately taken up by the adjacent clarinet and proceeds to pass across the stage to the viola at stage left; this glissando was intended to represent the two circular objects at the corners of the poem.  As the piece goes on the glissando begins to pass across the stage more frequently and quickly, and as it does so starts to redistribute the material between the performers until the dichotomy of the opening is completely lost and they are all performing the same type of material.

Conventionally when writing vocal music a composer knows exactly which words he is setting to music in any given passage; Steve and I decided to do things rather differently, however.  Once Steve had written the poem and we had collaborated to invent the structure which I outlined above, I then wrote the music with no further reference to the text, producing a piece in which the vocal lines had no text attached to them but rather functioned as instruments in a purely musical texture.  With the exception of taking into account practical considerations (a singer is not as versatile or agile as an instrument) I drew no distinction between a soprano and a clarinet.  Only when I had done this did Steve and I reconvene to decide which words should be fitted to which note; thus, rather than setting text to music, we set music to text.

By working in this way we ensured that the poem was relevant conceptually and structurally, but I wanted to find a way to render the words more than just ‘filler’.  The American musicologist Carolyn Abbate has written that in vocal music the text and the music constitute two distinct narratives which at different points in a piece can support each other or contradict each other.  Moreover, and more importantly for our present purposes, text and music come in and out of focus, and at any given point in a piece they can be equal partners or one can take precedence over the other.  This idea is very clear in opera of the eighteenth and early nineteenth centuries, which alternates between recitative, in which musical content is kept to a minimum in order that a large number of words can be sung quickly while remaining audible, and aria, where more elaborate music takes priority over – and sometimes renders inaudible and therefore irrelevant – the words.  While composers have deconstructed this rather rigid duality over the course of the last two centuries, the principle remains that in much vocal music the hierarchy between words and music is dynamic and this can be used as an important structural device.  Lacking the compositional resources with which to react to the words of the poem on an individual basis, I employed Abbate’s ideas as a device to render the words an integral element of the piece.

In the opening paragraph of Soundings, for example, the words “Dawn in hell is hot”, taken from the three ‘prongs’ in the poem, are presented with very slow music and are therefore readily audible.  Alternating with this, the text taken from the wedge, set to quick music and with large leaps in the vocal line, is (certainly on a first listening) largely inaudible.  The point is less that the words ‘mean’ – with so much aural information being transmitted to the listener it seems unrealistic to expect that any direct sense is made of the words – than that the fact that the words are conveying information is a defining element of the musical texture and therefore in itself means.  Thus the two sections are distinguished not only by the fact that one is fast and one is slow but by the fact that the text of one is audibly present while the text of the other is effectively absent.

Although the two musical textures – fast and slow – initially alternate with one another, they quickly come to be layered on top of one another and are heard simultaneously.  While the musical layers are sufficiently different to be heard as distinct elements, the text is completely lost.  Conversely, the point in the piece where the two voices come together for the first time is musically a moment of resolution; textually, it is the first moment when the two sopranos are singing the same words, which results in a sudden sense of clarity.  While the words themselves do not denote resolution (unlike the music, the poem is not teleological and therefore has no resolution as such), the very fact that they are audible contributes to the resolution of the tension of the earlier music.

This process of bringing words in and out of audibility is carried out more explicitly in Portmanteaux, which is scored for Soprano, two Clarinets, Viola and Vibraphone.  In Portmanteaux there are literally two separate poems.  At any given point in the piece one poem is being sung by the solo soprano; the other poem is also set as a vocal line in a conventional way, but the resulting melody is given to instruments rather than the voice, rendering the words literally and absolutely inaudible.

Musical example – Portmanteaux

The only time that the second poem gets a significant airing is in the slow section towards the end of the piece, with the words ‘For your company my silent monument I do pay’.  There are many features of the music which point to this moment as being distinct from the rest of the piece: the music is slow for the first time; the harmony is very different from what has been heard before; and the vibraphone is playing for the first time.  As in Soundings, however, the fact that the words are clearly audible here for the first time in the piece is perhaps the defining feature which distinguishes this passage from the music which surrounds it.  To return to Abbate’s ideas, the dynamic hierarchy between text and music has suddenly been wrenched to a place no hitherto unheard in the piece.

To conclude, I would like to emphasise that these thoughts represent work in progress and constitute no more than a provisional position: Steve and I intend to work on a third piece to act as a companion piece to Portmanteaux and Soundings, and I anticipate – and indeed hope – that in this third piece our approach to the relationship between text and music will be radically different.


PORES webjournal, Professor William Rowe,  email: