Introduction

In the context of increased interest in musical creativityFootnote ¹ and its social and distributed character, contemporary music offers particularly fruitful opportunities to investigate the detailed fabric of creative work. The chance to observe the live interactions between composers, performers, producers and technologists offers a direct insight into creative processes at a time when the roles of these principal creative agents have become more fluid in much contemporary musical production.Footnote ² This article focuses on a collaboration between the composer Jeremy Thurlow (henceforth Jeremy) and the violinist Peter Sheppard Skærved (Peter) in the making of Ouija, a work that consists of five movements for solo violin and precomposed sound files. The approach that we take brings together an analysis of the discursive interaction between the participants as they worked together on the piece; an analysis of the development of musical materials over the course of the collaboration; and a consideration of the embodied character of the performer's developing relationship with the music. Together, these perspectives provide a way to investigate and understand the intertwining of collaborative interaction with musical development.

The intimate, relatively compressed collaboration witnessed here, centred on a piece that contains a significant amount of guided improvisation and semi-indeterminate notation, brings into focus questions about creative development and the way in which collaborating musicians set up the conditions for its genesis. In discussing joint work and creative development, we situate this study in relation to two distinct areas of research: a number of recent studies of composer–performer collaborations; and a broader category of research that investigates the psycho-social conditions of creativity within groups. While our focus here is on the collaborative elements of the work and their significance, the genesis of the piece also involved a considerable amount of individual labour. Our intention in this study is not to downplay or deny the contribution of solitary work, but to draw attention to and analyse the subtle, transformational effects of what might be termed ‘sociable creativity’.

The dominant model across creativity research of the last 60 years has tended to adopt a singular, individualized and cognitive approach, with an emphasis on problem-solving.Footnote ³ An important recent extension to this cognitive framing of creativity has come from research with a focus on group work and the group itself as a creative force.Footnote ⁴ In tandem with sociological studies of creativity that emphasize the distribution of labour within different art worlds, this has substantially reframed the questions that are posed in relation to creativity – away from what happens within creative agents and towards what goes on between them. There is no single model of ‘between’, and as Howard Becker has written, the division of labour between the members of a group can take many forms: it need not be simultaneous or under the same roof, and many instances of collaborative work depend on a distributed succession of individualized contributions.Footnote ⁵ Much Western art music appears to follow a model of collaboration that is successive and linear, a composition developing as the autonomous work of a composer that is then passed on to the performer(s), with whom there may be only very limited exchange. Performers themselves may, and usually do, work in a highly collaborative fashion, but such collaboration is often characterized as realization or re-creation, rather than a contribution to, or involvement in, what is regarded as the ‘primary’ creative process.

In this study, however, we look at co-present collaboration as part of the shaping of the work (as well as acknowledging the significant involvement of sequential development), and it is in the examination of such face-to-face creative practices that Keith Sawyer's work has been particularly influential, adopting a more social and interactional approach to creativity.Footnote ⁶ Across a number of writings, Sawyer has explored the improvisational, collaborative and emergent qualities that are inherent in group creativity. In doing so, he argues that most forms of group creativity happen ‘in the moment of the encounter’ and therefore are necessarily improvisatory; that ‘creativity cannot be associated with any one person’ and is therefore necessarily collaborative; and that collective phenomena – being irreducible to the sum of their parts and inherently unpredictable in outcome – must be regarded as emergent.Footnote ⁷

Within this wider exploration of creativity, our work relates to a more targeted agenda – a developing body of research that since 2005 has examined musical collaborations between composers and performers from psychological, sociological/anthropological and practice-led perspectives. In an article that constitutes a precursor to the current study, Eric Clarke, Nicholas Cook, Bryn Harrison and Philip Thomas present an analysis of the rehearsal and first performance of a new work for solo piano (être-temps, composed by Harrison and performed by Thomas), focusing on a number of collaborative and distributed attributes of the project.Footnote ⁸ These include particular features of the way in which Thomas engages with the piece and prepares it for performance; the development of various interpretative strategies through the rehearsal period (particularly those concerned with rhythm and timing); questions of flexibility and fixity in the performance; and the relationship between notation, action and sound. Harrison describes Thomas as a ‘performer who […] sees questions as a productive rather than restrictive part of the learning process’,Footnote ⁹ and as a creative collaborator. A theme to which Thomas returns on a number of occasions is how he can maintain an open and inventive relationship with the piece and its notation, when he also has to spend long hours working painstakingly on detailed aspects of the material – as is vividly illustrated in an exchange between Thomas and Harrison during the rehearsal period:

Harrison: Otherwise you get used to the sound and keep imitating yourself in a slightly inaccurate way really.Footnote ¹⁰

Thomas's commitment to a deliberate provisionality, and to the sense of ‘liveness’ that he aims to create from that, is mirrored in Harrison's similarly unfixed and enquiring attitude towards notation. He acknowledges the formalized character of the notation, which is based on an unusual use of metrical grids, but regards this neither as the image of an authoritative sound in his head, nor as deterministic in its aims or consequences, but as a starting point for the performer's exploration and imagination: ‘I'm writing it to hear it as much as I'm hearing it to write it.’Footnote ¹¹

A similar study of composer–performer collaboration, involving the composer Fabrice Fitch and the cellist Neil Heyde, investigates analogous issues, but with a greater focus on instrumentality, sound and the affordances of particular technical features of the cello.Footnote ¹² Whereas être-temps was a finished score when Thomas received it, the score for Fitch's piece for speaking cellist entitled Per Serafino Calbarsi II: Le songe de Panurge (henceforth PSCII) evolved over a period from 2002 to its first full performance in 2006 through an intermittent dialogue between Fitch and Heyde. For reasons that are connected with PSCII's particular scordatura tuning, the piece plays particularly with the pitch relationships between different harmonics, and in exploring those possibilities Fitch and Heyde chanced upon a particular combination of harmonics with a specific pizzicato technique that subsequently became a central element in the further compositional development of the piece. Fitch and Heyde make reference to Helmut Lachenmann's description of composition as ‘building an instrument’, entailing ‘the building of “an imaginary instrument”, the exploration of whose properties (by the composer) brings about the piece itself’.Footnote ¹³ It is clear that PSCII indeed involves a kind of mutual tuning of the instrument and performer – both literally through the exploration of a particular scordatura and more metaphorically in the discovery and development of playing techniques. But more than that, the ‘instrument’ that is built crosses over between Fitch's and Heyde's apparent roles as composer and performer:

In a statement that both pays tribute to the crucial creative role of Heyde as the performer and at the same time reinscribes his own position as the original ‘source’ of the composition, Fitch writes:

As pieces for solo performer, neither être-temps nor PSCII engages with that very direct form of creative collaboration that is necessarily true of an ensemble. In two rather different projects, both of which involve new music for string quartet, Amanda Bayley and Paul Archbold have studied the interactions between composers and ensembles. Bayley's project has been concerned with the rehearsal of Michael Finnissy's Second String Quartet, and the commissioning and rehearsal of his Third String Quartet, both for the Kreutzer Quartet.Footnote ¹⁶ Using audio and video documentation, and a number of interviews with the composer, Bayley provides an account of rehearsals, three public performances, and reflections by the composer and members of the quartet on the performances. Bayley's examination of this collaboration provides one of the most detailed descriptions and analyses of discourse and interaction in rehearsal.Footnote ¹⁷ The rehearsal process is explored through a mixture of content analysis (outlining the proportion of time engaged in various forms of rehearsal activity such as playing and talking, with a breakdown of the topics for discussion between the composer and quartet members such as sound quality, coordination and social conversation) and more qualitative attention to what is said. The approach is developed from social psychological studies of rehearsals,Footnote ¹⁸ and from ethnomusicological writings that cover not only the process of making music but also the relationship of sound to discourse.Footnote ¹⁹ While the roles of composer and performers are distinctly separate here, and ‘the composition is in no sense collaborative’,Footnote ²⁰ it is clear that there is a significantly collaborative creative effort involved in bringing the piece to performance. A subsequent article arising out of the same project demonstrates that Finnissy (MF) and the leader of the Kreutzer Quartet (Peter Sheppard Skærved; PSS) engage in very much the same kind of creative interaction as that between Harrison and Thomas mentioned above:

If a deliberately ‘unattained’ quality, projected through a degree of asynchrony, is one of the aims in the Finnissy quartet, a serious and detailed attention to achieving finely tuned coordination in the context of great complexity is a central feature of the collaboration between Brian Ferneyhough and the Arditti Quartet in Archbold's documentary film of the rehearsal and performance of Ferneyhough's Sixth String Quartet.Footnote ²² The film documents the quartet's detailed preparatory work on the score and parts before the first rehearsal and the interplay between the composer and the four performers during the rehearsal period that immediately precedes the première at the Donaueschingen Festival. The quartet clearly enjoys and values working with Ferneyhough, as it has done over many years, but the relationship comes with its own internal dynamics, arising from underlying attitudes about the roles and responsibilities of composers and performers. One of the rehearsals with the composer exemplifies this when, in the middle of a passage of complex rhythmic interaction, Ferneyhough (BF) – who is directing the quartet in quasi-conducting fashion – goes ‘shhh’ to Irvine Arditti (IA), sitting just to his left. The quartet comes to a stop, and the following exchange takes place:

Behind the laughter and teasing lie the dynamics of authority and compliance, competence and intelligibility, and independence and cooperation between composer and performers, and the delicate negotiations that are involved. In the prevailing assumption that composers sit at the top of the pyramid of creativity, with performers below them, lies the source of the tensions that briefly poke through the surface of this interaction: the composer's reliance on the performers to bring his music to life; the performer's pride in fidelity to the score (‘I'm still fff !’) against the composer's authority (‘those who dare to contravene my instructions’); and the underlying anxiety that the composer might not really be able to hear his own music, or that the performer might be too literal in his reading of the notation. Collaborations between composers and performers are inevitably as much about the socially constructed roles that individuals inhabit more or less willingly and comfortably as they are about creative imagination and ‘free play’.

In an article that considers the collaborative work of a considerably larger ensemble and its conductor with the composer of a newly commissioned work, Clarke, Mark Doffman and Liza Lim analysed a number of critical rehearsal episodes that exposed similar interweavings of institutional, social, notational, instrumental and circumstantial factors in the creative process.Footnote ²⁴ Building on an increasing volume of recent publications,Footnote ²⁵ they offer a view of the distributed and ecological character of aesthetic production, pointing both to the ways in which people are deeply intertwined with opportunities and technologies and to the different scales of history and social order within which musical production takes place. Using video recordings of rehearsals and two performances, and interviews and discussions with the composer, conductor and members of the ensemble, the research focuses on three episodes in rehearsal and performance that reveal the network of creative forces that much of the time operates in rather hidden ways in the comparatively ‘seamless’ working practices of a professional ensemble. All three episodes involve opportunities and tensions that arise out of the interplay between fixed and improvised elements, and all three also expose the influence of microsocial forces (immediate interactions between co-present individuals) and macrosocial forces (larger and more long-term social factors that relate to roles, institutions and traditions) on the direct materiality of music-making and creative decisions. In one case, distinctly different views of the musical virtues of deliberate planning and spontaneous emergence shape the realization of a passage of semi-indeterminate notation. In a second case, an informal process of ‘rule-making’ and ‘rule-breaking’ that relates to a sustained passage of improvised ensemble accompaniment is called into question and becomes entangled with prevailing assumptions about instrumental roles and creative prerogatives. And in a third case, a combination of serendipitous personal history (the birthday of a key player), the relationship between permanent ensemble members as ‘hosts’ and a soloist as ‘guest’ and the specific affordances of the musical material give rise to a striking moment of complex social communication expressed through music. The creative ecosystem that the article explores is therefore revealed as a heterogeneous mesh of material, historical, ideological and ‘memorial’ forces operating at a variety of scales.

The current article also considers the ecology of collaborative creation in the specific context of a piece of contemporary concert music, but is complementary to the previous study in two respects: it focuses on the concentrated interaction of a composer with a single commissioned performer; and it combines qualitative and quantitative methods in analysing the creative development of the piece over the rehearsal period and across repeated performances. Our approach ranges across the physicality of an individual musician's creative engagements with his instrument; practical and conceptual aspects of notation; the discursive and social engagement of the composer and performer working together; and the historical and biographical threads that they weave. As in the earlier article, we argue that an ecological perspective provides a productive framework for such an account by drawing together what might otherwise remain separate domains of material culture, psychological process and linguistic and social interaction. Within the broad context of that ecological mesh, we make use of a deliberately simplified distinction between processes that are ‘inside the room’ – the immediate matters of notation, instrument, sound – and processes that extend ‘outside the room’, encompassing the influences, histories and aesthetic attitudes that the musicians bring with them to their creative encounter. And within this framework we also acknowledge that this is an ecology with a tangible outcome: a piece of music emerged from this collaboration, and our primary research questions reflect our concern to understand not just the details of this very particular ecology but its relationship to and effects on the musical outcome: (1) How do a composer and performer work together in the production of a new piece of music? (2) How is creativity enacted in face-to-face collaboration? (3) How is the collaborative effort reflected in material changes in the music?

The material

Musical material

Following a chance meeting with Peter in 2009, Jeremy was keen to write a piece for him when the opportunity and commissioning funds became available in early 2010.Footnote ²⁶ Drawing on the experience of a previous piece from 2007 entitled Endlessly Enmeshed, which combined Western classical musicians using conventional notation with two improvising Indian classical musicians, Jeremy's explicit aim was again to make use of fully notated and more improvised materials, but to push his own compositional practice in a new direction by incorporating computer-generated or computer-controlled sounds.

Despite the integration of improvised elements into the piece, Jeremy ‘very much wanted to write a piece which would feel like it had a character, a shape, a design, every time it was played, even though there's improvisation in it and therefore it's different every time’. He ‘wanted the piece to be a space that you could inhabit in lots of different ways, but it was still a [particular] space’.Footnote ²⁷ This posed a problem: how could the piece incorporate the flexibility of improvisation and at the same time retain a sense of its own shape?

There had to be some looseness, and some sense that all sorts of different possibilities could work, and for a while I just found that an impossible conundrum. I felt like a poet being asked to write a poem, but also being told ‘You can't actually specify what the words are in the poem.’ And I just thought: How can I do that? How can I make a poem if I can't specify which words they are and which order they go in? I really got stuck.Footnote ²⁸

The solution to this creative impasse came when Jeremy came across an account of the early twentieth-century Hungarian violinist Jelly d'Arányi on Peter's own website. From this he learnt that d'Arányi had been a medium,

and spoke to spirits, and especially musical spirits, and that gave me the idea that it might be interesting for the violinist to be a medium, and it would solve or help with two problems. One is: ‘Why are there these disembodied sounds coming out of loudspeakers?’, which I often find very problematic in electroacoustic music […]. But in this case I thought you have got a live violinist acting and moving, and there is a reason why there are voices, as it were – sounds that are disembodied. And we can relate to why they're disembodied because it's like what would happen in a séance or in a meeting with other-worldly spirits. So it helped with that, and it also gave a rationale for the improvisatory aspect, because a medium of course doesn't go in with a book, a play-script and say what he has to say: he has a conversation which evolves depending on what happens. […] And almost as soon as I read the bit about Jelly d'Arányi I had the idea. It just fell into place, so I was able to email Peter back quite quickly and say: ‘I think I've got an idea. Why don't we call it Ouija?’ […] And in his performances Peter's talked about Jelly d'Arányi and […] how she is connected to a sort of living tradition of past musicians, and memories of them, and feeling that they were still with her when she played, most of which I didn't know. So it worked out wonderfully. I just knew this little thing about Jelly d'Arányi but it was enough.Footnote ²⁹

Having hit upon this creative solution to the problem of how the laptop music might be ‘motivated’, and how to conceive of the relationship between composer-defined and flexible or improvised elements, Jeremy decided to organize the materials of Ouija in the manner of a séance – as reflected in the titles of the movements: ‘Invocation’, ‘Among Voices’, ‘Sprite’, ‘Under the Shadow of Wings’ and ‘Among Voices II’. By the time of the first workshop in February 2012, Jeremy had prepared initial scored versions of ‘Invocation’, ‘Among Voices’ and ‘Under the Shadow of Wings’ on the basis of work that he had been doing since November 2011, and was in the process of writing further material that became ‘Among Voices II’ and ‘Sprite’. In an email to Peter and Doffman shortly after the first workshop, Jeremy commented that the piece would consist of seven movements:

As well as the movements we tried yesterday, there could also be a) a new mvt drawn from Paganini in the same way that Under the shadow is drawn from Bach [this became ‘Sprite’] b) a new mvt based on the idea of Peter ‘catching’ the tempo of the tape part, which suddenly changes every now and then like a will o’ the wisp c) a new final movement exploring the higher registers of the instrument, so far neglected [this became ‘Among Voices II’]. Not all of these will necessarily come to fruition, but if we consider all of them for now, the order might be: Invocation – Will o’ the wisp – Paganini – Among voices – Bach – new repartee – final mvt.Footnote ³⁰

By the middle of March, at the second workshop, the materials and overall shape of the piece were close to their final form and had been narrowed down to five movements. In the interim, Peter had worked on technical aspects of the music, was no longer sight-reading the notated material and was comfortable with the kinds of improvisation that the piece entailed.

In broad terms, Ouija employs three approaches to musical material and to the relationship between notation and improvisation. ‘Invocation’ (see Example 1) is the most conventionally and fully notated of the movements, with only the duration of the pause at the end of each phrase left unspecified – these pauses at first being filled with silence, and later with the initially ghostly sounds of other music emanating from the laptop. ‘Among Voices’ (see Example 2) uses a semi-indeterminate notation in which the pitches are specified, but little or nothing of the rhythm – although the score provides a passage of suggestive comments about how the rhythm of the movement might be approached. The movement involves a constant interplay between the violin and the laptop, with the violin line swimming in what Jeremy describes as ‘a shoal’ of other musical lines.Footnote ³¹ As its title implies, the fifth movement – ‘Among Voices II’ – adopts the same general approach. Finally, movements 3 and 4, entitled ‘Sprite’ and ‘Under the Shadow of Wings’ respectively, make use of material that has a specific compositional reference: Paganini in the case of ‘Sprite’, and J. S. Bach in the case of ‘Under the Shadow of Wings’ (see Example 3). In ‘Under the Shadow of Wings’, on which we focus in this article, the laptop music took as its source the Siciliano movement of the Bach Solo Violin Sonata in G minor (BWV 1001), while the violinist's music consists of a short phrase, loosely modelled on the opening of the first movement of that same sonata, followed by the instruction to ‘continue, improvising’. The score gives brief advice about the relationship between the violin and ‘tape’ (= laptop)Footnote ³² parts – mostly in terms of the flexible coordination of the two, including the statement that ‘the improvisation should feel, in some broad sense, in sympathy with the music of the tape (though this can be defined as freely as the player wishes: it certainly doesn't rule out playing notes which are dissonant with the tape part)’.

Example 1. Score of ‘Invocation’.

Example 2. Score of ‘Among Voices’, pp. 1–2.

Example 3. Score of ‘Under the Shadow of Wings’.

The laptop music for the three movements was developed in rather different ways. In the first movement (‘Invocation’), Jeremy's idea was to use the laptop to produce faint wisps of sound/music during the pauses following the third and fourth violin phrases and a ‘reaction’ at the end of the movement (see Example 1) – as if ‘called forth’ by the violinist's musical appeals. At the first and second workshops, Jeremy had not yet made these sound files, and Peter worked on the music in the absence of any answering ‘voice’. Shortly before the première, Jeremy put together a succession of sound files (from barely to distinctly audible) made out of snatches of recorded orchestral textures, transformed with granular synthesis and mixed with brief recorded extracts of Peter's own violin music from that movement. This was intended as no more than a first attempt at this music, to provide something for the première and likely to be superseded – but at the pre-performance rehearsal, and then in the première itself, both Jeremy and Peter agreed that it had worked so well that there was no need to consider any replacement.

By contrast, the laptop music for ‘Among Voices’ forms a much more substantial and closely integrated component of the movement, giving a sense of the violinist being immersed within lines of music (‘among voices’) that are a partial echo of the soloist's own material, and are therefore derived deliberately and quite directly from the violin part. The laptop music was generated by layering together a number of separate strands, each of which consisted of a rhythmic realization of the violinist's notated material (incorporating frequent tempo shifts) and a partial fragmentation of the scored material, so as to avoid and disguise any direct mirroring of the performer's line. Jeremy scored these layers using Sibelius notation software, and Peter then made digital recordings of these separate lines,Footnote ³³ which in the final version of the laptop music were combined with sampled string sounds controlled directly by the Sibelius-based MIDI files.Footnote ³⁴

Finally, ‘Under the Shadow of Wings’ used yet another approach. Starting with a fortuitously available commercial recording of the Siciliano movement of the Bach Solo Violin Sonata BWV 1001, Jeremy identified some relatively brief extracts that featured double-stopping and, using pitch transposition and tempo shifts coupled with granular synthesis, layered these elements to form a slow-moving homophonic texture, starting rather diatonically and becoming somewhat more chromatic as the movement progressed. Although the recorded source is a violin, the effect of the granular synthesis is to produce a texture that has the character of an instrumentally indeterminate harmonic wash. The remainder of this article focuses on ‘Invocation’, ‘Among Voices’ and ‘Under the Shadow of Wings’, representing as they do the three broad musical strategies of the work as a whole, and constituting the movements on which the most rehearsal and discussion took place.

Talk and collaboration

Jeremy and Peter established the collaborative momentum of Ouija over the course of two workshop days, separated by a month's interval. As already noted, prior to the first workshop Jeremy had prepared four movements to work on – ‘Invocation’, ‘Among Voices’, ‘Under the Shadow of Wings’ and ‘Repartee’ (which was subsequently dropped). Peter came to the first workshop having seen small snippets of some of the pieces but was effectively sight-reading the material, and although many of the final constituent elements were recognizable at the first workshop, there was still considerable uncertainty about the component movements and overall shape that the piece would take. Jeremy's and Peter's creative work on Ouija therefore occupies a collaborative middle-ground, in that while the overall framework and a significant proportion of the musical material is attributable to Jeremy's compositional perspective, the improvised nature of some of the movements and the relatively sparse violin part in others left much more to shared decision-making than would be the case in a fully through-composed piece. But the piece was also at a considerable remove from a completely improvised approach, in which all participants share similar creative authority. The materials and processes evident in the creation of Ouija specify a much more porous relationship between pre-given material and the in-the-moment qualities of improvised performance. In looking at the two days of workshops, our point of departure is to examine the dialogue between the participants and the ways in which their discourse participated in the collaborative process.

Although researchers have looked at language use as a collaborative outcome, for instance in improvised theatre performance,Footnote ⁴¹ there has been relatively little detailed work on the dialogic component of creativity leading up to the moment of performance.Footnote ⁴² In posing the question ‘How is creativity enacted in face-to-face collaboration?’, we look at the ways in which dialogue was used by the two musicians over the course of the collaboration. The starting point for our analysis is a set of figures that display the amount of time spent within the workshops on playing and discussing the piece and its performance. Figure 1 gives an overview of the proportion of playing and discussion in the three workshop sessions (over two days – W1, W2.1 and W2.2), providing a panoptic view of what took place rather than a detailed content analysis. In addition to playing longer or shorter stretches of the music (‘playing’), the major part of each workshop was devoted to discussion about the piece and its realization, and we have thematized this as follows: ‘composition-talk’ was conversation about the composition; ‘playing-talk’ centred on performance; and ‘making-talk’ was often about pragmatic aspects of the realization of the piece, including the rehearsal process itself. This characterization is summative rather than analytically detailed, and was coded at the level of pair-wise exchanges rather than at the sentence or phrase level, so as to convey a general sense of what was going on through a stretch of dialogue rather than to produce a detailed content analysis. In addition to these three categories, there were periods during the sessions when the conversation turned to topics that were of no relevance to the workshopping of the piece, and we have coded these periods as ‘social-talk’.

Figure 1. Pie charts showing the percentage of each workshop taken up with playing and talk about the composition, performance (playing-talk), practicalities of the rehearsal (making-talk) and general conversation (social-talk).

The interest in these summary charts lies in the changing proportions of playing and talking over the course of the workshops. By the afternoon session of the second workshop (W2.2), which took place in the resonant and atmospheric space of Robinson College Chapel, playing had become the dominant mode of work, following the earlier preponderance of discussion (W1 and W2.1). By contrast with the proximity and intimacy of Jeremy's relatively small and acoustically dry college room (where both W1 and W2.1 took place), the size, resonant acoustic and architectural character of the chapel powerfully afforded playing rather than talking. Not surprisingly, as Figure 1 illustrates, composition-focused talk became less a feature of the workshops as they progressed, since it was at the beginning of the workshop process that Jeremy explained the organization and ideas behind the work. For example, at the beginning of the first rehearsal, before the first play-through of ‘Invocation’, the conversation was largely given over to an explanation of the origins of the piece and the rationale for the phrase structure of the movement.

And, as might be expected, stretches of playing through the movements were often followed by talk that centred on interpretative aspects of the work. Following on from the first play-through of ‘Under the Shadow of Wings’, the conversation turned to Peter's getting to know the (unnotated) sound file:

This category of ‘playing-talk’ was demonstrably greater in the second workshop (both morning and afternoon) than in the first session, when composition-talk predominated.

The third type of content, which we have designated ‘making-talk’, consisted of conversation directed to the many practical issues of the piece without being directly focused on either the composition or its performance, and this became more noticeable in the third session:

Figure 2 is based on the same data as Figure 1, but indicates the moment-to-moment distribution of talk and performance across the workshops. It shows relatively long passages of composition-talk in the first workshop alternating with passages of playing through movements. Stretches of playing are often directly followed by conversations about the playing, and talk about the composition often returns once the interpretation or improvisatory playing has been discussed. Within the second workshop, there is a change towards less talk about the composition, and more about the playing and the making of the music.Footnote ⁴⁷ These different categories of exchange (excluding ‘social-talk’) together form a discursive register that we summarize as creative-talk, representing dialogue whose central focus was on the making of the piece.

Figure 2. Timeline of the workshops, showing periods of talking and playing, and an indication of the focus of the talk. Categories are: playing (P), composition-talk (CT), playing-talk (PT), making-talk (MT) and social-talk (ST).

Material changes in Ouija

In the quantitative analysis that follows, we focus on changes in timing in the opening passages of ‘Invocation’ (INV), ‘Among Voices’ (AV) and ‘Under the Shadow of Wings’ (USW). As already noted, these three movements provide an opportunity to compare music that is fully notated, rhythmically improvised and fully improvised. The three movements differ considerably in length (INV is between 2 and 3 minutes in duration, while AV and USW are both between 5 and 6 minutes), and the analyses that follow are based on the first 14 bars of INV, the first 24 bars of AV and the first 180 seconds of USW. The focus on timing excludes consideration of dynamic shaping, the relationship with the laptop part and the pitch content of Peter's improvisation in USW, but provides one particular perspective on change and development in the piece over time.

‘Invocation’

It is common in empirical investigations of performance to express timing variations as deviations from the notated rhythmic values, and to express these deviations proportionally in relation to the notated values.Footnote ⁷³ However, because the notated rhythmic values of INV were interpreted very freely by Peter, it is questionable whether this approach, which implicitly takes the score as a norm, is appropriate.Footnote ⁷⁴ One consequence of the proportional approach is that changes in the duration of short notes often appear large and changes in the duration of long notes disproportionally small (a lengthening of, for example, 30 milliseconds is a considerable proportion of a short note, but only a small fraction of a long note).Footnote ⁷⁵ An analysis of variations in local tempo at beat level avoids this problem of scale, but requires interpolation in the absence of note onsets on beats. The score of INV (see Example 1 above, p. 124) demonstrates that the music consists predominantly of groups of relatively short notes followed by one or sometimes two longer notes within a loose metrical framework, somewhat like a recitative. Our analysis therefore focuses on the absolute duration of successive units, alternating between units comprising one or more long notes and units comprising one or more short notes. This grouping of notes is shown in the numbering of the units in Example 4. Figure 3 shows the absolute durations of the inter-unit intervals in the first 14 bars of INV. The onset of units was manually identified in the audio recording of each playingFootnote ⁷⁶ using the audio analysis software PRAAT.

Figure 3. Absolute values of inter-group intervals in four playings of ‘Invocation’: W1 (A); W2 (B); P1 (C); and P4 (D). The data point out of range in panel D (unit 35) has a value of 14.46s.

Example 4. Start of ‘Invocation’, indicating the units identified for timing analysis.

The timing profiles feature obvious large peaks in duration (the last of which concludes the extract) that relate to the four notated pauses in the musical excerpt, and these are particularly pronounced in the two public performances (panels C and D). For Jeremy, these pauses are particularly significant for the meaning of this opening movement, as he explains on a number of occasions. Having ‘called out into the unknown’, the violinist should wait for the spirits to start to respond: ‘So [in] the pauses […] you are hoping something is going to happen; and eventually something will come back […]. You are the one who starts the conversation.’Footnote ⁷⁷ Over the course of these four playings, the duration of each pause in relation to the preceding phrase increases considerably, reflecting the increasing rhetorical importance of the ‘listening’ that follows each of Peter's ‘invocations’. Table 3 further illustrates this by showing the proportional value of the duration of the pause in relation to the duration of the phrase that precedes it, demonstrating the considerable increase in this value for P1 and P4 compared with that for the workshops.

TABLE 3 proportional value of the duration of a pause in relation to the total duration of the preceding phrase

Pause/phrase	W1	W2	P1	P4
1	0.162	0.233	0.326	0.345
2	0.099	0.146	0.428	0.510
3	0.174	0.199	0.372	0.909
4	0.143	0.163	0.312	0.243

Apart from the change in the duration of the pauses, the inter-unit interval (IUI) values change very little: cross-correlations between the timing profiles – excluding the peaks related to the pauses – indicate that in general the pattern of IUI values is very consistent across performances, with Pearson correlation coefficients all above .858 (p < .0001, N = 46).Footnote ⁷⁸ This consistency is based on the absolute duration of the musical units as already noted, and can therefore be attributed in part to variations in the notated duration of the units (which is constant for all playings). If the unit durations are normalized with respect to the notated durations (by dividing the measured durations by the score values of the units), the variation attributable simply to score durations is removed. The cross-correlations between normalized timing profiles are lower, but are all still strong (greater than .7). Within this overall figure, the normalized timing patterns for the two rehearsal playings are relatively strongly correlated with one another (r = .794, N = 46), as are those for the two concert performances (r = .813, N = 46), suggesting that Peter adopted distinct timing strategies (less and more rhetorical, respectively) under the two playing circumstances.

One noticeable change across the playings is the relative duration of shorter and longer notes. In the absolute timing of the units, there is a clear distinction between shorter and longer units, in response to the rhythmic gestures notated in the score. This is reflected in a positive correlation between the notated values and performed durations (see Table 4). In a mechanical performance this value would be 1 (indicating perfect agreement between notated and played values), which is clearly not the case for P1 and P4. Instead, shorter notated units are played relatively long, while longer notated units are played relatively short. This is apparent from a negative correlation between the notated duration and the normalized played duration, indicating that with increased notated duration the proportional (normalized) duration of the played values becomes relatively short (see again Table 4). This negative correlation is particularly strong in P1 and P4. This analysis indicates that the overall temporal pattern of INV is fairly consistent over the four playings, as indicated by the high cross-correlations between timing profiles, except that the silences become longer in P1 and P4, and the contrast between short and long durations becomes progressively smaller. These changes result in a less metrical and more rhythmically homogeneous playing of the music (smaller contrasts between long and short notes), in which the silences play a significantly more prominent role.

TABLE 4 cross-correlation between score durations and absolute (left) or normalized (right) played inter-unit intervals (iui)

	Correlation with absolute IUI	Correlation with normalized IUI
W1	.922Footnote ***	–.501Footnote **
W2	.929Footnote ***	–.470Footnote *
P1	.822Footnote ***	–.705Footnote ***
P4	.823Footnote ***	–.709Footnote ***

* p < .01;

** p < .001;

*** p < .0001; N = 46

In Jeremy's and Peter's discussions of this movement, the interpretation of the pauses and other expressive gestures was addressed on a number of occasions. As already noted, in the first rehearsal Peter emphasized the variety of ways in which he could perform the opening gestures of the movement and the tone colour that he might adopt, while Jeremy returned to the idea of ‘invocation’, and encouraged Peter to think in terms of being a medium and of the attempt to start a conversation with the spirits. And it is this conceptual underpinning that seems to drive the development of the movement, as the following comments from the first workshop indicate.

Jeremy: Each of these phrases is actually sort of tentative […]

Peter: OK, so this is like there is a formal gesture that starts it and now we begin; then stuff happens after that.

And later:

Peter: I am very excited about this whole medium thing […]. There are two big André Jolivet pieces for solo violin […] [which have] that thing that Jolivet loved – kind of invocation actually. He believed it was the job of a player to summon up evil spirits, and Pan, and slavering gods and everything.Footnote ⁷⁹

‘Among Voices’

The semi-determinate score of ‘Among Voices’, shown in Example 2 (pp. 125–6), consists of a series of motivic units each made up of a group of short notes and a long note. As the performance notes in the score indicate,Footnote ⁸⁰ and as was reiterated by Jeremy in the first workshop, the performer is expected to phrase these units in various ways: ‘I would like the phrases to run across several bars […]. You know, sometimes it's four, sometimes it's five, three, and it doesn't have to start at the beginning of the line, it can overlap the line.’Footnote ⁸¹ These different phrase lengths can be achieved through tempo modulation and variation of the duration of the long notes (keeping some of them relatively short to preserve forward momentum), the results of which are shown in Figure 4. Owing to the strict alternation between a group of short notes and a long note, the odd-numbered units in Figure 4 always relate to the summed duration of a group of short notes (short-note units – SNU), while the even-numbered units indicate the duration of individual notes notated as long in the score (long-note units – LNU). The number of notes within each SNU varies from four to eight in a regular pattern across four-bar units: each four-bar unit starts with two bars, each containing five short notes and a long note; the third bar contains four short notes and a long note; and the final bar has between six and eight short notes and a long note.

Figure 4. Time intervals between successive units (groups of short notes, and individual long notes) in the first 24 bars for playings of ‘Among Voices’: W1 (A); W2 (B); P1 (C); and P4 (D). The start of the laptop music is indicated with an arrow. In A (first run-through) the laptop starts before the violin; in panel C (first performance) the laptop plays the wrong sound file, and has to be quickly restarted with the correct file – hence the two entries.

Figure 4 shows considerable changes in the timing profile of the music across the four playings. In panel A, which shows the very first play-through, the alternation of SNUs and LNUs is mirrored in a regular pattern of short and long unit durations – at least for the first 20 units (10 bars), after which it becomes somewhat more varied. The ratio of the summed duration of the short notes to the long note in each bar fluctuates around 0.487, indicating that the summed duration of the short notes is a little less than half the duration of the single long note.Footnote ⁸² This is in sharp contrast to the later playing, in which the duration ratio between the short and long notes is much more variable, the ‘zigzag’ pattern appearing only briefly and usually towards the start, and with numerous instances of the LNU being shorter than the SNU. The average proportion between successive SNUs and LNUs progressively increases, from just below 1 in the second workshop (0.978 – panel B) to slightly above 1 in the fourth performance (1.084 – panel D). The standard deviation of this proportion, which indicates its variability across a playing, is very small for the first run-through (0.152 – panel A), larger for the second workshop (0.459 – panel B), and largest for the two public performances (0.585 and 0.583 respectively – panels C and D), indicating that there is a marked increase in the variability or ‘playfulness’ of the phrasing in the concert performances. In addition to these local relationships, there is some evidence of larger-scale modulation of the durations of the units, indicating phrasing across numbers of units – including some evidence for phrasing across eight units relating to the four-bar structure of the music,Footnote ⁸³ despite Jeremy's comment that the phrasing need not respect the four-bar organization of the notation. For example, in the second workshop playing (panel B of Figure 4), unit 16 (LNU of bar 8) is distinctively long, and is followed by rather clear phrase arches across two spans of eight units (four bars) each. The fourth concert performance (panel D) shows similar patterns of tempo variation, but now much more variable in length – with phrase boundaries (signalled by long durations) at units 18, 24, 32 and 44 (LNU of bars 9, 12, 16 and 22).

The duration of the SNUs is only partially determined by the number of notes in each unit. The relationship between the number of notes and the unit duration as indicated by their correlation is stronger for W2 and P4 (r = .567 and r = .553 respectively; p < .01 for both) than for W1 and P1 (r = .302, n.s. and r = .456, p < .05 respectively; N = 24 throughout). This may be a by-product of the tendency to indicate four-bar phrases by a lengthening in W2, P1 and P4 of the fourth bar, which always contains a relatively large number of short notes. Nevertheless, even excluding every fourth bar, the correlation between the number of notes and unit duration is strongest for W2 and P4 (r = .530, p < .05; r = .466, p = .052 respectively; N = 18), and is non-significant for W1 and P1 (r = .219, r = .364; n.s. for both). It appears that Peter employs two timing strategies: one in which the duration of the units has a more direct relationship with the notation; and one in which this relationship is looser and more flexible. It is interesting in this respect that Peter observes after the first play-through that the groups with varying numbers of notes suggest an additive rhythm.Footnote ⁸⁴ However, variations in the duration of SNUs are more strongly correlated across playings than are variations in the average duration of individual short notes (see Table 5), which suggests that timing control was exercised at the level of the unit, rather than at the level of the individual note, despite the sense of an additive rhythm to which Peter refers.

TABLE 5 pair-wise correlations between snu durations (left) and the average duration of short notes per unit (right), calculated by dividing each snu duration by the number of notes per unit (n = 24 for every correlation)

	Duration of SNUs			Average duration of short notes
	W2	P1	P4	W2	P1	P4
W1	.366	–.106	.382	.226	–.214	.216
W2		.395	.549Footnote *		.207	.274
P1			.497Footnote *			.325

* p < .05

This analysis has demonstrated that the timing of rhythmic units in the movement changed considerably across the workshops and performances, reflecting discussions between Jeremy and Peter about the deliberate modulation of phrasing so as to emphasize the music's improvisatory character. There appears to be a distinction between more ‘notation-driven’ and more ‘in-the-moment’ strategies of playing, where the ‘in-the-moment’ strategy was driven by detailed local features of the musical material. This approach was manifest in playings that showed more diverse and varied phrasing, and a more improvised and unpredictable quality. Across the workshops and performances, the average tempo of Peter's playing varies somewhat, and the laptop joins Peter at a slightly different moment, with the consequence that the relationship between his part and the accompanying laptop is different on each occasion – an indeterminacy that is further emphasized by the option of using either one of two slightly different versions of the laptop music for each performance.

The interaction between the composed and improvised elements is well explained by Jeremy in an interview after the première, in which he comments on both the freedom of interpretation and the fixity of the composition:

I encouraged the player to think about phrasing and continuity and span, and I wrote all the notes out, so in terms of pitch it is not an improvisatory piece. […] Peter does actually do slides and bends and things on the notes and it sounds great, I'm happy about that. But I think he realized that the idea was that I had written the melody, and that the rhythm was one aspect of the freedom; but even more important than that was the phrase building. And you can do an awful lot with that that can be radically different every time. So in a sense you decide where semicolons or half-cadences are, and bigger cadences, and you decide where climaxes are; and you decide where crescendos and diminuendos are, and other aspects of the music go with that. But I've left that to the violinist […]. So that was one answer to the question of how you could improvise in some respects, but I could still feel that I composed the thing. [And Peter] said actually the shapes of the little phrases were suggestive to him, and the things I had not specified were also suggestive: he thought that he could make all kinds of things with this. And he suddenly was very imaginative and free; and he also said that he found that playing that movement felt particularly like improvising, which is interesting because as I say, every pitch is specified and in the right order.Footnote ⁸⁵

‘Under the Shadow of Wings’

‘Under the Shadow of Wings’ has only the most minimal of scores (see Example 3 above, p. 127), and in the absence of more extensive notation the analysis of performance timing that follows necessarily focuses on raw onset data. The timings of note onsets for the first 180 seconds were identified using Sonic Visualiser (as described above), and Figure 5 shows the detected onset times of notes on the x-axis, the distance between onsets indicating their inter-onset durations. Each playing starts with the opening gesture notated in the score: a broken chord (represented as two onsets) followed by a succession of fast notes.

Figure 5. Note onsets across four playings of ‘Under the Shadow of Wings’: W1 (A); W2 (B); P1 (C); P4 (D). Each vertical bar represents the temporal position of a violin note.

Figure 5 demonstrates a clear pattern of change across the playings. The first play-through (panel A) shows an alternation between a melodic gesture (consisting of one or two long notes, followed by a sequence of fast notes) and a longer pause. Peter explained in an interview: ‘I decided to [use] the material he gave me to play, which is about 17 notes, as a kind of mode of limited transposition. So I would worm my way backwards and forwards through that in different transpositions, inversions and things,’ and he would confine himself to ‘intervals or gestures, which were implicit in that inverted arch; in augmentation or diminution’.Footnote ⁸⁶ In the two concert performances (panels C and D), and to a lesser extent the second workshop, this strategy alternates with a more continuous manner of improvisation, in which event onsets are more evenly spread across time, there are fewer and shorter breaks between gestures, and also fewer fast notes – in other words a more even distribution of onsets. It seems that after the first playing, the improvisation becomes more attuned to the pace of change in the laptop music, in line with a key concept of the music. As Jeremy explained in the first workshop: ‘I am thinking of spirit voices out there [i.e. the laptop part] that you can tune into, you can have conversations with, you can kind of go with; and in this case there are sort of violin spirits out there and you should feel a warm envelope of violinhood as it were.’Footnote ⁸⁷

The data presented in Table 6 confirm this interpretation of the changes in rhythmic pacing across the performances. The table shows the median, lower and upper quartiles, and minimum and maximum values of the time intervals between successive note onsets, and demonstrates how the tempo of Peter's performance decreases over the four playings: by a factor of more than seven for the median (from 0.128 to 0.981 seconds); more than five and nearly nine for the lower and upper quartiles respectively (0.080 to 0.445; and 0.296 to 2.405); and a factor of nearly ten for the minimum (0.027 to 0.262). Only the maximum shows a different picture (effectively no change), indicating that the duration of the longest pauses between ‘utterances’ by the violinist remains essentially constant. The score itself (see Example 3, p. 127) provides two tempo indications: a general tempo for the movement (♪ = 60), which is also given as indicative of the laptop part; and a slower tempo (♪ = c.50) for the opening gesture of the violin part. If the median duration is taken as an approximation to the overall tempo of any playing, then P4 (corresponding to ♪ = 61) is closest to the indicated tempo, while all the earlier playings are significantly faster.

TABLE 6 summary statistics for inter-onset intervals (in seconds) in four playings of ‘under the shadow of wings’

Playing	Median	25%	75%	Minimum	Maximum
W1	0.128	0.080	0.296	0.027	17.387
W2	0.443	0.229	0.949	0.078	8.389
P1	0.650	0.395	1.612	0.148	15.044
P4	0.981	0.445	2.405	0.262	15.604

This analysis of onset timing indicates that Peter's approach to the movement changes appreciably over the course of these four playings, which is consistent with his own comment that while his approach to the movement met with Jeremy's approval right from the beginning, he nevertheless kept on developing and adapting his playing. ‘I think I probably hit something he liked fairly early on in the process; I just kind of wanted to make sure that he liked it slightly more each time.’Footnote ⁸⁸ He does not go on to explain exactly how he changed his playing, and there was little explicit intervention from Jeremy – almost his only comment being that Peter might leave a few longer gaps; but as this analysis has shown, the improvisation seems to become more homogeneous in its rhythmic characteristics in later playings, resulting in a more continuous interweaving of the two voices.

To summarize, perhaps the only general or unifying strategy across developments in the playing of these movements was a tendency to increase the homogeneity and integration of certain features of the performance, while increasing the variability and contrast of others. In ‘Invocation’, for example, the silences became more pronounced and contrastive, while the differentiation between short and long notes was reduced, with the result that the rhythms became more homogeneous. In ‘Among Voices’, on the other hand, the timing of composite units containing long and short notes became more varied and less stereotypical, while the duration of the short notes within each figure became relatively more uniform. Finally, in ‘Under the Shadow of Wings’, the improvisation became both more unpredictable, in the sense that it was less closely related to the notated opening gesture of the movement, and more integrated with the accompaniment, as the rhythms became less durationally contrastive, with a slower median tempo that was closer to the tempo of the laptop accompaniment.

Understanding, materiality and embodiment

We have so far discussed Jeremy's and Peter's collaboration in terms of their social and discursive interactions, and the material changes that took place across workshops and performances. But this is to neglect one of the most striking features of Ouija's development: the way in which it came to life through the instrument in Peter's hands – the way in which it became instrumental/embodied. While an increasing theoretical interest in the explanatory power of theories of embodiment has now built up a significant literature,Footnote ⁸⁹ there remain significant practical difficulties in analysing the ways in which embodied action may either constitute or reflect changed understanding – and what kind of ‘understanding’ that is. Variously referred to as tacit knowledge, procedural knowledge or ‘knowledge how’,Footnote ⁹⁰ the knowledge or understanding that is achieved and manifest through the exercise of a skillFootnote ⁹¹ has often been contrasted with the explicit, declarative or propositional knowledge that is exemplified in knowing, for instance, that Jelly d'Arányi died in 1966. It is clear that Peter's declarative knowledge and understanding of Ouija (including aspects of notation, narrative/poetic reference, phrase length, the content and sequence of the laptop part) develop through the workshops in many of the ways that we have already discussed – along with his procedural knowledge. As discussed earlier, Peter himself articulates a complex interweaving of declarative and procedural knowledge in describing the craft of violin playing both as a genealogical tree (an explicit history of teachers) and as a practice that involves a literal (as well as metaphorical) ‘laying on of hands’. The more embodied and material nature of that knowledge and understanding is expressed even more strongly later in the same interview, when he continues:

It is having the violin in his hands, the size of the instrument and the history of the other players who have also held the instrument – even the physical posture that those other players may have adoptedFootnote ⁹³ – that informs the way in which Peter approaches and makes the music. For him, this physicality plays a central role in the sound that he produces, his relationships with co-performers (most obviously the other members of his quartet) and even his relationship to the audience:

The instruments and physicality of playing together act as a bridge to the traditions of the instrument as well as serving the more immediate function of communicating with fellow musicians and the audience. While the instrument and body here act as a repository of history, there are also shorter scales of temporal engagement at work in Ouija, and in what follows we offer a description of the performer's emerging bodily engagement with the music over the course of the rehearsal period, as shown in a series of short video clips.

Video clips 1 and 2 show the first two minutes of ‘Among Voices’ – on the first occasion that Peter played it at the first workshop (clip 1), and at the première in Sidney Sussex Chapel (clip 2).Footnote ⁹⁵ A number of striking differences in the physicality of these playings of the music are evident – some of which are the consequence of Peter's first play-through of the music being in the relatively physically confined space of Jeremy's college room. While Peter moves a fair amount in clip 1, the movements are all of a similar kind – as if drawn from a limited repertoire. In general, these movements mirror the motivic profile of each unit of the piece, reflecting melodic and rhythmic properties of the music, and in some cases are partially determined by simple ergonomics (particularly the movements of the bowing arm). By comparison, clip 2 shows a much wider repertoire of considerably more dramatic movements (no doubt facilitated by standing rather than sitting), some of which trace much larger trajectories in space and extend over longer periods of time, reflecting or inducing an organization of the music into longer and more integrated strings of units. These movements convey a much more intense involvement with the music – a sense that these movements are making the music, rather than reflecting it; and at the same time they seem more free of the music, an example being the way in which Peter uses the opportunity of an open E string to take his left hand right away from the neck of the violin at about 1:40.

One obvious and significant difference between the two clips is in the social occasion and context in which the playing takes place: a private run-through in a relatively small room with only the composer and a researcher present in one case; and in the other, a public première in a large performance venue in front of an audience. Peter himself commented on what he described as his own inability to perform the music convincingly without an audience.

It would be wrong, however, to suggest that Peter's movements in the performance are simply the result of tacit and emergent processes of embodiment: there is also an explicitly choreographic element, as was clear from a comment of Jeremy's to Peter at the second workshop, following a play-through of ‘Invocation’:

Peter agreed, and though he had had no time in which to memorize the music, he tried out this more ‘dramatic’ and self-consciously choreographed approach with immediately different consequences for his posture and movement. Video clips 3 and 4 demonstrate the distinct character of these two playings, and the interpretative difference that is the consequence of this deliberate bodily strategy.Footnote ⁹⁸ With the séance narrative running throughout Ouija, each movement of the piece presents a clear opportunity for Peter to ‘act out’ some aspect of the drama – floating and swimming in a shoal of voices in ‘Among Voices’; responding to the sound and spirit of Paganini in ‘Sprite’; inhabiting and exploring the sound-world of Bach in ‘Under the Shadow of Wings’.

Nonetheless, while acknowledging the role of deliberate choreography, there is also a clear sense of the music's increasingly embodied presence (the music both taking over Peter's body and being taken over by it) in the second of the two workshops – particularly in the more theatrical and acoustically responsive space of the Robinson College Chapel. Our final example (video clip 5) shows Peter in Robinson Chapel playing ‘Among Voices’ for the first time in this space.Footnote ⁹⁹ After a straightforward start, the playing takes on a dramatically more physical and bodily character after about 30 seconds, first with a suddenly powerful rendering of two of the motivic units, followed immediately by an equally striking shift to a much more lyrical style accompanied by a distinctive rocking movement of his body. Although Peter was not yet deeply familiar with the movement, there is a palpable quality of involvement and bodily characterization that conveys his engagement with the music. Closing with an extended and intense decrescendo, and after a moment of dramatic silence, Peter walks slowly over to where Jeremy is standing with the laptop, and half-ironically remarks: ‘I was probably having way too much fun there, I don't know.’Footnote ¹⁰⁰ The comment encapsulates both Peter's acknowledgement of his own more passionate engagement with the music, and perhaps a genuine uncertainty about whether this approach was still consistent with Jeremy's conception of the movement. In this respect, and particularly when Jeremy responds: ‘No, I liked it, I liked it a lot actually,’ it also acts as an appeal by Peter to Jeremy's judgment and opinion, which in turn gets a reciprocal reassurance by Jeremy that he approves. This is an example of face-talk that signals increasing trust between the two musicians, in turn allowing them to appeal more directly to one another for judgment and opinion. In this episode of playing and talking, body, interpretation and working relationship come together in a way that seems to act as a turning point in the project.

In summary, we have suggested that embodiment performs a number of functions within the working relationship between the two musicians. First, the body acts as a conduit for knowledge in relation to instrumental techniques, and the tacit knowledge that connects musicians to a musical past – something that is also realized in less procedural and more explicit ways through the musicians’ dialogue (‘inside/outside the room’). In this respect, the body connects musicians, instruments and events over relatively long timescales. Over the course of the workshops and performances, however, and on a relatively short timescale, the body is a means to Peter's increasing absorption in the music,Footnote ¹⁰¹ and a manifestation of that changing relationship. It is in this enactive relationship that Peter both makes and finds a developing understanding of the material, an embodied complement to the shared discursive engagement that constitutes and intensifies the creative collaboration.

Conclusions

This article has examined the production of a piece of new music, using a combination of methods to explore collaboration and change in a creative partnership. We have sought to identify how a piece-in-performance emerges from collaboration by examining the development of musical materials through the embodied interactions and discourse of the musicians during their face-to-face work. The collaborative momentum of this project revolved primarily around two days of intense workshops between the composer and the performer; and while the completed score could be characterized as having a controlled indeterminacy written into it, it also made use of a sufficiently explicit notation and verbal instruction that the work could be performed without extensive exchange between the composer and the performer. What, then, was the role of collaboration and dialogue as a creative force in bringing this musicking to fruition, and how different was the process from the traditional sequence of composition and interpretation?

As we have shown, there were examples over the course of the workshops when stretches of dialogue led to direct changes in the musical material. Such conversational moments were marked by a degree of indirectness and a relatively open or fluid approach, allowing Peter and Jeremy opportunities to proffer suggestions and share uncertainties about the creative direction. Decision-making seemed often to occur at a very pragmatic level, and we have pointed to a number of specific moments when manifest creative change occurred through clear collaborative input. But conceptions of creativity place different emphases on different processes: Margaret Boden, for example, tends to see creativity as a relatively focused and deliberate transformational process, whereas other research points to the messier and more indeterminate nature of much creative work.Footnote ¹⁰² In this study, while there were moments of unambiguous change that occurred through interaction, perhaps of greater significance was the progressive accumulation of shared understanding, which took place in two ways. First, there is that category of talk that we have described as face-talk through which Jeremy and Peter demonstrated trust in each other and a sense of each other's competence, the latter also achieved through displays of compositional and performance prowess. This mutually sustaining interaction, constructed through competence and interpersonal trust, appears central to a collaboratively creative project. Secondly, however, the accumulation of understanding also occurred through the frequent sharing of musical and other references that we have characterized as ‘outside the room’. This form of discourse functioned some of the time to establish an understanding of musical materials by reference to other music acting, particularly for Peter, as a way of contextualizing and consolidating his improvisational strategies. But just as significantly, these links to a network of public materials (other music, literature, film, paintings) outside the room helped to foster the intimacy inside the room that an enterprise of this kind requires, building a shared world for the project to inhabit.Footnote ¹⁰³ Although there was clearly an increasingly public transformation over the course of the project (from the privacy/intimacy of Jeremy's room, to the public space of the Robinson College Chapel workshop, to the manifestly public première), in the workshops, public culture was used to intensify and stimulate the private workings of this collaboration, both in terms of praxis and at an affective level. In contrast to those psychological models that characterize creativity in terms of a surprising or innovative shift, the development that occurred between these two musicians is therefore incremental and cumulative, characterized by emergence – particularly the understanding (embodied and procedural, as much as propositional and conceptual) that was shaped by the participants’ interactions.

Sawyer's research on group creativity has highlighted how emergence is one of the defining features of collaboration – the recognition that ‘the whole is greater than the sum of its parts’.Footnote ¹⁰⁴ However, a number of refinements to this familiar formula are suggested by our study. As Sawyer acknowledges, the momentary interactions within an improvised, collaborative framework are never simply of the present: there is a wealth of tacit understandings and reference points that scaffold the process through pre-existing knowledge and via structuring processes that emerge internally within the flow of the creative work. But Sawyer sees these as secondary to the improvisational present, a hinterland that lies behind the real work in the moment, thus underplaying a central aspect of joint work – namely the reflective understanding on the part of the collaborators about the creative context in which they are working.Footnote ¹⁰⁵ Collaborators are not simply known to each other as individuals; they also inhabit generic roles (in this case those of composer and performer) with particular cultural histories that are brought to bear in the work. They interact with one another as composer and performer – in both an episodic sense (being familiar, or not, with this composer) and a more generic or semantic sense (a more or less shared sense of composers’ roles more generally). In this study, much of the engagement between the participants hinges on their desire for creative interaction within, and perhaps at times against, the knowledge and experience they have accrued as composers and performers.

So even in highly improvised situations, such as a workshop conversation or an improvised performance, history matters, both in the sense that an aesthetic object or performance participates in a genre (that is, it is part of a significant ‘large-scale’ or broad social history) and in the way that an activity of this kind makes and feeds upon its own micro-historical context. What is particularly salient about context in the course of a possibly unpredictable creative process is not simply what context may explain about the unfolding creative work, but also how context (micro or macro) is used in this unfolding to generate new insights, solidify agreements, nudge the collaboration in new directions or suggest an alternative to an unproductive trajectory. We make no claim that all creative collaborations would enjoy the same degree of reference to other works, performers, composers and so on that we have observed here, since much of that referencing is a product of Peter's particular immersion in, and passion for, the history of his instrument and its repertoire. Nonetheless, the principle remains that a form of signifying – pointing to the context of the work and to the collaboration itself – is central to the twin goals of making music and developing a productive and enduring collaborative relationship.

While there is no escaping the profound influence of the history of musical materials and musical roles (the sedimented roles of ‘composer’ and ‘performer’) suffusing the whole project, there is also the risk of over-stressing the macrosocial and institutional forces – of appearing to espouse a type of determinism in which the entirety of the interaction and collaboration is understood in terms of institutional and historical power. In the immediacy of face-to-face and moment-to-moment working, those macrosocial forces recede into the background, only to reappear in sometimes sudden and unexpected ways – as when Peter alludes to the long lineage of violin players to which he feels connected, or reinforces Jeremy in his role as composer just four minutes into the first workshop: ‘You tell me, you're in charge, you're the boss.’Footnote ¹⁰⁶

A second refinement to Sawyer's perspective is the recognition that collaborative work seldom takes the form of an equal and constant input from all collaborators throughout the lifetime of the project. A collaborative ‘deficit’ may be the consequence of inequalities of status that result in a more hierarchical set of working relationships, such that collaborative good faith may be quite attenuated at certain points. Equally, as is illustrated in this project, there may be considerable changes over time in the depth of collaboration, as a consequence of intrinsically different phases of a creative project. The great majority of Jeremy's compositional work had been accomplished before the first workshop, apparently placing primary creative responsibility almost entirely in his domain;Footnote ¹⁰⁷ but by the time of the first performance, there was a palpable sense that the work had passed very much more into the performer's territory.

A final point in relation to language and interaction – and one that seems central to understanding creative collaboration – is the way in which participants are involved in a social process that extends considerably beyond what is narrowly required to achieve the musical goal. In collaboratively creating a piece of music, significant work also goes into establishing, maintaining and developing a working relationship. We have mentioned the way in which face-talk is implicated in the creative process, but there is a degree of creativity involved in the construction and maintenance of the collaboration itself, over and above its ‘products’. For every component of a collaboration geared towards making materials and identifying or solving creative problems, there is an accompanying interactional dimension that is dedicated to the shared task of establishing and maintaining appropriate social engagement.

Beyond the recognition that collaboration has become a more central feature of contemporary music, there is also an implicitly positive gloss that is often attributed to collaborative work, which should not go unquestioned. While the animated engagement between Peter and Jeremy and the positive creative outcome constitute a fruitful collaborative project, in what ways did the piece actually develop through these interactions? From our detailed timing analysis, it is clear that there were shifts, sometimes quite marked, in how the piece was realized in performance; and some of the material (for instance in ‘Under the Shadow of Wings’) developed a much more integrated relationship with the compositionally fixed laptop part, with Peter's improvisation becoming more nuanced and selective as he got more attuned to Jeremy's ideas and more familiar with the sound files. Similarly, in comparison with the initial read-through, ‘Among Voices’ manifested a very different quality by the time of the performances, with the improvisatory framework realized through longer phrases that became more differentiated and less predictable, in somewhat the same way that Peter's approach to ‘Invocation’ became more dramatic and rhetorical.

However, not only were the musical materials changed by the collaborative process in particular ways, but these same materials also afforded different opportunities for the collaborative process itself. One of the intriguing features of the collaboration was the way in which the movement that lies midway between explicit notation and free improvisation seemed to elicit the most intensive exchanges in the collaboration. The most improvised movements (‘Under the Shadow of Wings’ and ‘Sprite’) seemed to afford interchange only at a relatively broad and general level; and the most fully notated movement, ‘Invocation’, not surprisingly drew the collaborators into more standard topics of clarifying and realizing the notation – though not exclusively. But it was ‘Among Voices’, where the presence of a loose notational framework acted as a kind of anchor or partially fixed point, that gave both participants something to work around. Peter remarked that Jeremy was at his ‘fussiest’ with respect to ‘Among Voices’; far from being a negative comment, this indicates the degree to which the structure of the piece affords a more sustained and dialogic interaction. Indeed, in an interview after the second performance, Peter reflected on the sometimes counter-intuitive relationship between notational specificity and freedom:

And, contrary to a negative view of the specialization and consequent separation of the two spheres of composition and performance that took place in the latter part of the nineteenth and into the twentieth century, Peter expressed a fascination for the interpretative space that this opened up for the performer:

We end by returning to Ingold's proposal to read creativity ‘forwards’, understood in the light of Jeremy's reflection on the whole experience of working with Peter. To read Jeremy and Peter's collaboration ‘backwards’ – that is, to assign it value on the basis only of the outcome – would be to miss the point. It is in a ‘forwards’ reading that the value of this joint work can be seen. The creativity of the collaboration is not so much a matter of innovation as of developing a shared, complex realization of a piece. Many of the creative changes that we have documented in this article could also have taken place under the more standard and sequential circumstances of a finished piece (however ‘open’ or ‘closed’) that is passed on to a performer; and despite Peter's own strongly expressed preference for working with composers wherever possible,Footnote ¹¹⁰ we make no claim for the special virtues of collaboration in terms of its outcomes. But the particular circumstances of this project allowed us to witness creative processes that are also features of less obviously collaborative circumstances. Some of Peter and Jeremy's collaborative decisions undoubtedly resulted in concrete changes, but many of the ‘changes’ are better understood as shifts of emphasis and understanding – a developing sense of comfort, confidence and identification on the part of both musicians that configures creativity not only in terms of production but also as collaborative empathy, mutual understanding and the realizing of opportunities: