“You Won’t See Me” – In Search Of An Epistemology Of Collaborative Songwriting


This paper proposes an observational methodology by which we may gain deeper understanding of the creative processes used by collaborative songwriters. Almost every aspect of popular music production and consumption has been discussed and analysed in scholarly work, but the creation of the song itself has rarely been subject to scrutiny. This is perhaps due to the fact that very little of the songwriting process can reliably be inferred by listening to an audio track or reading a score. Therefore, two methods are proposed in combination – interview-based and participatory auto-ethnographic observation. In both cases the songwriters themselves generate the qualitative data. The aim is to construct a framework that researchers may use to provide answers to the question “how shall we know the mind of the songwriter?” The methods proposed here have been tested with more than 20 professional songwriters between 2009 and 2013, and a selection of these observed co-writes will be published as case studies in 2014.

My current research focuses almost exclusively on collaborative songwriting. There are three reasons for this: firstly, there is a practical need to limit the scale of the work to a delineated area of creative practice. Secondly, as a musician and songwriter I have some experience of the decision-making processes involved in creating music, so there would be some benefit in studying these auto-ethnographically through personal co-writing as well as through interview or third-party observation (as almost all previous studies have been). Thirdly, collaborations provide evidence through process: by necessity, creative ideas must be communicated to the other collaborator, thus becoming manifest in the physical world in some musical or verbal form and, therefore, becoming observable. An auto-ethnographic approach has the added benefit of limiting any potential pollution of the creative process due to the observation effect.

In psychology terms, then, my approach is mainly rooted in Skinnerian behaviourism (Skinner: 1957, 180) – that is, my intention is to document and study observable behaviours rather than to infer thought processes or authorial intent (except when made explicit through behaviours). The auto-ethnographic part of the methodology also enables limited use of introspective methods, but these are only used when they describe observable behaviours in the songwriting team.

Existing approaches to methodology

Music psychologist John Sloboda (1985) identifies several of the difficulties associated with studying creativity in composers, discounting critical analysis (inferring creative processes from the completed work) as being concerned more with the musical relationships evident in the finished product than with the moment-to-moment psychological history of the genesis of a theme or passage (Sloboda: 1985, p.102). He identifies four possible methodologies: examination of manuscript; general and retrospective interviews with composers; live observation of composers; and observation of improvisatory performance. I discount the first methodology given the fact that notation/transcription of a popular song is a reductive post facto record of an audio object, as well as the last, given that songwriting is not pure improvisation – although any live observation may reveal an element of composition-by-improvisation. This leaves us with two possible routes: the interview and the observed songwriting session.

The interview

Interviews with songwriters abound in the music press and, although they may be limited in depth and technical detail, they provide virtually limitless genre breadth and cover a very large number of significant popular songwriters from the 20th and 21st century. Published interviews can be a useful research source because they usually share the question “where do you get your inspiration?”. This question and its variants, while generating many a banal or romanticised answer from some self-mythologising artists, is an important one for the study of songwriting, and perhaps for the study of creativity generally, but the researcher usually needs more detail than a mainstream media interview provides.

Frustratingly, given its potential status as a primary source, the published artist-songwriter interview has its limitations as evidence of the creative process. To demonstrate the problem, let us look at a particular kind of partnership – very prevalent now in the UK/Europe and US music industry – that is noticeably under-represented, even concealed, in the media. I call this the “Svengali” model, as used in the early 21st century by artists such as James Morrison, Avril Lavigne, Newton Faulkner, P!nk, Jake Bugg, Lily Allen, Take That, Beyoncé and James Blunt. In this model, the artist is marketed as a songwriter, but the actual practical task of writing the song includes one or more back-room professionals. The artists still have their authenticity to sell, of course, and thus have a vested interest in publicly diminishing the role of the Other. [1]

The following 2009 interview with James Blunt, discussing his platinum-selling hit You’re Beautiful, is a typical example. Blunt acknowledges the presence of song craft in his more experienced co-writer, but foregrounds the simplistic emotional lyric sentiment of the song as of more value; note his arguably dismissive use of the terms “finished the song off” and “tied up”:

It’s the spark of the idea that is necessary to make any great song… beyond that is the knowledge of how to craft it into a well-rounded song, and that took me many years to understand.  I saw an ex-girlfriend with her new man… and went back home and wrote the words [to You’re Beautiful] in under the time that I can perform the song – it came really fast. […] I then finished the song off with a friend of mine called Sacha Skarbek… he was, for me, a great educator in how songs should be tied up. And it took us all of an hour and a half maybe to tie those loose ends up. It was one of those songs that came from a spark, a really true emotion, and writing the lyrics took me all of a minute and a half.  (Hewson: 2009)

Significantly, Blunt implies in this statement that only one other co-writer finished the song; in fact, the album credits clearly state that You’re Beautiful was co-written (presumably in multiple separate sessions) with songwriters Skarbek and Amanda Ghost. Both of these individuals were experienced co-writers – Skarbek had co-written with Neneh Cherry, Samantha Mumba, Jason Mraz, Beverley Knight, and, since, Adele and Duffy. Ghost’s credits include Beyoncé, Boy George and The Prodigy. We can only guess at Blunt’s reasons for omitting Ghost from his description of the song’s creation, but such revisionist authorship claims are not uncommon among songwriting partnerships, especially when the song is successful.

In 2006, one of Blunt’s (alleged) collaborators, producer Lukas Burton, disputed the lack of co-writing credits he was afforded (Barnes 2006). Although Blunt later took legal action, Burton’s original claim had enough apparent merit to persuade the Performing Right Society to suspend royalty payments to Blunt (Smart: 2007). Burton had earlier described Blunt’s early songwriting skills as musically limited and implicitly in need of “professional” help:

His stuff was crude, occasionally laughably direct, and betrayed his relative lack of musicianship or discernible influence – it sounds unlikely but I think he genuinely hadn’t even heard of some people like Neil Young and Joni Mitchell let alone taken any of their music on board (a fact about which he was at least open and affable). (Barnes 2006)

The mainstream media, sharing the artist’s vested interest in the romanticisation of creativity, can collude in the concealment of the Other. At the time of the Blunt/Burton dispute, UK newspaper The Sun took a typically tabloid (and factually inaccurate[2]) position:

I really feel for James on this one. Since he hit the big time, everyone has been trying to get a piece of his pie and now a huge chunk of his royalties are being held back. I have had a good listen to James’s eagerly awaited second album, All The Lost Souls, which comes out at the end of September. And it’s very clear to me that James is more than capable of writing smash hits all on his own. (Author’s emphasis; Smart: 2007)

The backgrounding of the Other is perhaps in an artist’s economic interest, and the attendant romanticising of creativity makes a better press story than would be found in an exhaustively accurate description of how a song is actually created. These observations are not intended to be critical of Blunt artistically. Rather, my intention is to acknowledge Antoine Hennion’s view (1983: 159-193) that popular music production is necessarily contrived and involves many collaborators, and to note that, from the consumer’s perspective, some aspects of the process are more palatable than others. Romanticising creativity is culturally desirable.

The Blunt interview, and many others like it, serves to highlight the challenges of investigating the songwriting process through public interviews with successful contemporary songwriters: they may, when interviewed about their songwriting, intentionally or unintentionally provide misleading information. Artist-songwriters are obviously more susceptible to these pressures than non-artist songwriters, but such hypothetically less biased interviews are more difficult to find, there being less media interest in non-performing songwriters. Arguable exceptions are artist-songwriters who reflect on their craft when they are no longer at the peak of their careers, for example David Byrne (2012), but even these practitioners have an economic motive in the form of a back catalogue that still receives airplay and sales. This is not to accuse Byrne and his ilk of cynicism – like Hennion, he acknowledges that an ‘authentic’ pop persona is a necessary contrivance – but it does provide a challenge to the prima facie assumption that published first-hand accounts of the songwriting process will be reliable just because they are primary sources.

Conversely, an artist-songwriter has few positive reasons to tell the truth in interview – objectively to describe and analyse any mundane, uninteresting, unromantic, inauthentic or even random creative actions that may eventually manifest themselves in the song. Not only are these processes difficult to describe, they may also bore the consumer, and will therefore be avoided by all parties during an interview situation.

Studying the composer – observational methodologies

Perhaps unsurprisingly, given the difficulties of persuading composers to allow real-time observation as identified by Sloboda (1985), the retrospective interview accounts for a significant proportion of the methodologies employed by research into compositional creativity. There are some exceptions: Collins (2007: 239) uses the term “verbal protocol methodology,” i.e. interventionist first-person reporting of the composer’s decision-making, whereby the subject effectively narrates their own composing in real time. This provides primary source data at source but is obviously highly susceptible to the observation effect because it requires the composer to stop composing momentarily in order to provide the verbal protocol. Hayden and Windsor (2007) use, appropriately enough, a partnership methodology that they describe as collaborative qualitative enquiry. They support a partly auto-ethnographic method to address the objectivity-depth challenges of observing the manifestations of creative decisions:

[Rather] than… a conventional division of labour (researcher and informant), the first author is literally a co-researcher whose subjectivity is an essential and concomitant check on the second author’s more general and theoretical suppositions. It is all too easy for research and practice to become disengaged from one another when the latter becomes merely an object to be studied with little engagement with the role practitioners have in the realm of critical, as opposed to creative, endeavour. We would argue that the involvement of the practitioner, not only as a research subject, but also as a co-researcher, enables the research to gain a degree of validity… (Hayden & Windsor: 2007, p.32)

Hayden & Windsor’s methodology involved the co-researcher providing detailed retrospective notes on each composition to the researcher, who in turn provided the theoretical framework, using protocol analysis based on the work of Argyris & Schön (1974). Their operating context is musically very different from mine; their case studies are of instrumental art-music and the works are either visually-mediated (conventionally notated) or contain a substantial programmed or improvised element. They are not evaluating fully collaborative composition – rather, they research the way in which (primarily editorial) decision-making processes by specific composers are mediated by others. The works studied demonstrate impressive breadth, including a conventionally notated piece, a technologically-mediated work (which includes collaboration with a Max/MSP programmer) and a partly-improvised ensemble piece – which they categorise as directive, interactive and collaborative, the latter making use of “a collective decision-making process” (Hayden & Windsor: 2007, p.33).

However, none of these three categories describe collaborative songwriting fully and reliably, and Hayden and Windsor’s ‘classical music bias’ can perhaps be inferred by the fact that they ascribe so much importance to notation – even to the extent that they define a category based on its absence. Clef-based music notation is not a consideration for many songwriters.

Sloboda acknowledges the difficulties associated with attracting composer-participants as requiring a “rare degree of co-operation” and cites only one example from the contemporaneous psychology literature (Reitman: 1965). Csikszentmihalyi attests to a common disinclination for successful creative individuals to participate in research generally (Csikszentmihalyi: 1996, p.13). Perhaps partly for this reason, a large amount of observational research about composer creativity uses as its subjects non-professional composers, primarily schoolchildren or students. Riley (2012) describes the songwriting act as self-developmental, focusing on the educational benefits of the songwriting process; Cantor (2006) takes a similar education-based view by analysing the extent to which songwriting activities can inspire the songwriter. This learner-centred approach arguably supports the romantic view of songwriting as self-expression, self-development or therapy. It may be a useful area of educational scholarship but it does not deal with the central question of value in creativity. An individual might find songwriting (self-) helpful but if the song has no value to a listener it may not be ‘creative’. I would argue, after cognitive psychologist Margaret Boden (2004), that a truly creative artefact must be new, surprising and valuable. Newness is easy to measure – the song is only required to be non-plagiaristic of existing songs, but surprise and value must, by definition, be in the ear of the beholder. Therefore any research that purports to analyse songwriting creativity must deal with songs that are, measurably or potentially, of value to others.

Given that the song and its sound recording are separate objects, at least in legal and copyright terms, there is potential merit in the separation of composing and production activities. Folkestad et al (1997) differentiate the compositional and arrangement processes, describing compositional techniques as “horizontal” (completing the composition before transferring it to a computer-based medium) and “vertical” (merging composition and arrangement). Collins (2007) does not separate the two activities as clearly, frequently describing computer-based orchestral arrangement activities as compositionally creative, and indeed acknowledges the inseparability of composition and arrangement in his discussion of verbal protocol:

[Small-scale editing] activities (which may be construed by some as outside the domain of artistic, creative behaviour) took up a considerable period of time within the overall spectrum of the compositional period, as well as a considerable percentage of the verbal protocol. All these particular small-scale actions are part of the overall aspect of compositional flux; […] the verbal protocol highlights a constant dynamic where the composer is quite clearly working with his material, modifying it, altering, adding, deleting, trying things out. Thus at the micro-level of compositional processes and strategies, the verbal protocol clearly illuminates moments where the composer struggled with smaller-scale problem-solving issues. (Author’s emphasis; Collins: 2007, p.250)

There is some validity in Collins’s approach. He refuses to eliminate any activity that affects the outcome of the composition, perhaps because such elimination would be a necessarily subjective decision on the part of the researcher. Folkestad’s “vertical” and “horizontal” terminology provide an opportunity to attempt to isolate and subtract production and arrangement tasks such as quantising a drum part or choosing a keyboard sound. But given that these ‘production’ tasks themselves may influence future ‘songwriting’ decisions – (for example, an edit to a drum part might inspire a songwriter to make reactive edits to a lyric’s scansion) the argument in favour of separating the two activities looks weaker.

Tracking a real-time creative activity has a tendency to generate massive amounts of data, especially when applied to large numbers of participants. In my own work I have frequently encountered this logistical problem – one very simple co-written song (Bennett and Neate 2009) with a 120-word lyric, when tracked using computer-based revision history, generated more than 700 edits in the lyric alone; and in this instance, the size of the overall data set was relatively small because the songwriters were working asynchronously and remotely.

Heavily data-driven methodologies are suitable for latitudinal studies covering a large number of participants, where common compositional practices can be inferred; my case study approach substitutes breadth for depth by providing detailed analysis of individual songs’ evolution. The only songwriting processes that are not of interest to the study are cases where there is no evidence that a listener might engage with the eventual finished product – that is, songs written by songwriters who have not yet attracted the attention of an audience. This makes my threshold for defining creative value reasonably low, obviating any problems created by Richard Middleton’s well-reasoned criticisms of the practice of using sales data to provide evidence of popularity (Middleton: 1990, pp 5-6). Middleton argues that the manipulation of the market by those with vested interests (he cites hyping and payola; I would agree, and include A&R and in fact all forms of promotional activity) distorts the integrity of the charts as popularity statistics. But Middleton’s argument presupposes that there is a notional baseline that would be a positivist researcher’s ideal; in practice, popular music’s popularity has always been subject to such distortion and manipulation. Notwithstanding this (to use the forthcoming adjective literally) I contend that popular music will always show itself to be so through such metrics to some extent – and therefore, interviews with songwriters who have created economically successful work are of value to the research.

The proposed method, then, is to analyse primary and secondary interviews with collaborative songwriters who have created successful songs, to write songs with a number of different collaborators, and to cross-reference the resultant evidence across songs and partnerships using a selection of theoretical frameworks from existing psychological and educational scholarship. With this basic structure in place, I will evaluate the tools available in the recording and observation of the songwriting process.

Observing songwriting

Many of the processes of collaborative songwriting are recorded in some form as a by-product of the creative methods. For example, a lyric-first approach may be characterised by one collaborator acting as scribe for draft lyric ideas (concepts, phrases, rhymes or images) in a (paper) notebook or on a laptop. A beats-first approach may involve the collaborators auditioning drum loops and discarding ideas that are deemed unsuitable. Given that rejection/veto may be as important as creation of new ideas, our observation processes may need to record the contents of the cutting room floor as well as the final reel. It is therefore desirable for the research to select collaborators partly by the criterion of variations in their songwriting approaches.

My initial most attractive medium of choice was video, because it records physical/ergonomic, social, verbal and sonic interactions indiscriminately. But given that Sloboda’s “rare degree of co-operation” (Sloboda 1985: 103) in art music composers showed equivalent reluctance in songwriters I approached, video-based documentation was usually not possible. Further, professional songwriters are (by definition) likely to be significantly more economically motivated than art music composers or student composers, and have a vested interest in creating songs that appeal to a wide audience, so many were disinclined to participate in an academic study. The only participation incentive I could provide for collaborators were dependent on any perceived benefits provided to the co-writer (and the song) by my own creative input – and of course on a pre-agreed 50/50 publishing royalty split.

The “Save As…”[3] time-based study approach used by Collins  (2007) has the advantage of being reasonably technically easy to achieve. Iterative audio and MIDI [4] files can be generated quickly and relatively non-interventionally by the partnership. The lyric equivalent of this recording process is any word processing application that preserves every editing revision (my preference is for the web-based word-processor Google Docs, which allows synchronous or asynchronous collaborative editing and a complete revision history).

These technological methods were used whenever they did not risk diverting the songwriting process from its natural path as agreed by the collaborators’ personal artistic preferences and evolving interactions[5]. For example, it was not appropriate to use digital tools if the other collaborator wished to write with acoustic instruments and paper-based notebooks, and if such a collaborator also refused to allow audio, we would be left with only iterative demos, lyric sheets, informal notation (e.g. chord sheets), my own retrospective notes on the session, and retrospective interviews with the Other. However, when viewed in its entirety, I found that even this limited evidence base was surprisingly effective in highlighting creative behaviours within songwriting sessions.

So my initial intention – to use a single observational methodology across all partnerships and songs in the interests of consistency – was shown to be impractical, given the challenges identified in this discussion. Being unable to apply a single observation process to all collaborative situations allowed an unexpected methodological freedom – I did not need to participate personally in all collaborations, and could also interview songwriters without the need to collaborate with them.

Ontology – what can we (not) know about songwriting?

It is possible to say with some certainty what cannot be known, and in doing so I will begin by discounting a hypothetical point in time immediately before the songwriting process. The neurology of creative thought is a complex and specific area that overlaps with other fields, “causing difficulties in medical, academic, and philosophical discussions” (Mendes & de Carvalho: 2000, p.1) – but these pre-songwriting events are not the concern of this research. Rather, my intention is to capture and analyse the song-specific decision-making processes required to write a song, and thereafter to discuss the interrelationship between songwriting process and musical outcome – the journey from the blank sheet of paper to the studio-ready song, and, where necessary, onward to the demo or even the finished audio product. Sometimes creativity begins before this figurative or literal paper is seized; songwriters will often meet socially before beginning a co-writing session, and during this meeting a game plan or potential musical direction may be discussed, which sometimes included the establishment of musical or genre constraints upon the forthcoming co-writing session. As more observations and interviews have been undertaken, I have found it increasingly necessary to factor in these pre-sessional meetings, because it became clear that several creative decisions had already been implicitly made before the co-writing session began.

The other parameter that is unknown, or at least difficult to measure, is the musical/cultural context of each songwriter. It is not possible for anyone, even the songwriter themselves, to know the precise relationship between prior knowledge and compositional decision-making. Even if it were feasible to document the cultural and musical influences on a single adult in a lifetime, we could only speculate fruitlessly about which creative processes were or were not mediated by prior experience. Further, I would argue that due to creators’ unavoidable subconscious absorption of artistic conventions, all creative decisions have some element of prior cultural influence. Here, I take issue with Harold Bloom’s view, expressed in The Anxiety of Influence (1973) and referring specifically to poetry. He defines the successful poet as one who can succeed despite their influences [6]. Author Ian McEwan takes the opposite position, celebrating the evolutionary nature of his art and acknowledging that his own work is inevitably partly defined by the achievements of his predecessors:

We have come to take for granted … the vital and enduring concept of originality. It carries with it an idea of the new, of something created in a godlike fashion out of nothing… [But] the antithetical notion of artistic creation… is that no one escapes history.  Techniques and conventions developed by predecessors […] are available as ready-made tools and have a profound effect.  Above all, art is a conversation conducted down through the generations. (McEwan: 2012)

I contend that many songwriters are successful partly because of derivative elements of their work, or at least do not try to conceal them: influence is unavoidable and is actually desirable in the sonically self-referential context of popular music [7].

Let us now apply Bloom’s “anxiety” to our methodological challenges. Burnard & Younker acknowledge, perhaps with understatement, that reducing the timeframe of influence by working with student composers does not make the issue of personal musical experience any more manageable for the researcher;

[The] extent of any relationship between each student’s composing experiences at the situated level and what might be revealed through mapping their composing pathways is difficult to discern. Identifying influences – and the nature of those influences from musical experiences within each individual’s context – requires further investigation. (Burnard & Younker: 2004, p.71)

Even though it may be impossible to evaluate prior influence or cultural context, we can observe their manifestations as creative behaviours during a co-writing session.

The methods – interview and auto-ethnography

I have proposed two methods by which we may understand how songwriters collaborate to create valuable songs. The first is an interview with the songwriter(s), exploring three core questions: why do they choose to work with others; what processes do they use generally when writing songs; and, what processes did they use to create specific songs that were known to engage listeners. The last of these can never be proximately retrospective (that is, undertaken immediately after the co-write), because not every song will successfully engage listeners, but my research has shown that songwriters can reveal a surprising level of detail in interview about the way their most successful songs were written, even many years after the event. These bespoke, in-depth songwriter interviews are more valuable to the researcher than the shallower (and usually shorter) general interviews with songwriters in the music press and mainstream media. However, when a songwriter is not able or willing to undertake a bespoke interview, a less satisfactory but still valuable evidence base can be pieced together from multiple published interviews, with songwriters and with others in the production chain. In these cases, it is necessary to select case studies based on the amount and quality of source materials available, and such selectivity can yield successful songs with a rich evidence base. The advantage of both these interview-based approaches is that they solve the ‘value problem’ – that is, it is possible to learn about the creation of songs that are known to have engaged listeners.

The second method is to adapt Hayden & Windsor’s partly auto-ethnographic co-researcher model and to co-write personally with equally (or more) experienced co-writers. This approach needs to be selective of its subjects in order to address the question of value, and my subjects have been exclusively professional songwriters, defined as those whose songs have generated royalties[8]. These co-writes are documented as dictaphone-recorded audio files[9], which are then summarised in text. Where possible, the Other is interviewed immediately retrospectively, and the co-researcher makes their own notes as soon as possible thereafter. Depending on the co-writing process, some additional qualitative data may be generated (lyric revisions, chord sheets, early run-throughs, computer workstation files or rough demos before the song is completed) and these can add to the evidence base.

The interview method solves the value problem but lacks per-song depth. The auto-ethnographic method is not guaranteed to create works of value but provides the necessary depth, partly addressing the value problem through specifying professional participants. Both methods generate evidence that we can then triangulate with a critical analysis of the finished audio recording of the song.

“Linear event analysis” and the “time-dependence hypothesis”

Using these two methods, then, I contend that the researcher can generate a rich evidence base from which to analyse collaborative songwriters’ creative processes. But what analysis should we undertake and how can we infer creative processes from the raw qualitative data? The famous songwriter question “which comes first, music or lyrics?” highlights an important part of the answer. The sequence of events in songwriting is crucial to understanding the creative mechanisms and skills that practitioners use. Writing a lyric to a supplied melody is a very different skill from composing a melody for a supplied lyric. Similarly, composing a melody atop a backing track (known as “top-line writing”) is different from inventing harmony or a backing track that supports a pre-composed existing melody and lyric. For example, one of the most important creative challenges for a songwriting team is to identify the lyric’s thematic meaning: the researcher must try to discover, for any given song, when in the process this lyric theme is agreed, and if not at the start, what prior creative decisions inform the negotiation that leads to the theme’s creation. All songwriting creativity – perhaps all artistic creativity – is iterative in the sense that all creative decisions are taken in the context of the part-completed work at a given point in its creation. Therefore, our analysis must take account not only of each decision the songwriters make but also of each prior decision that may have informed it. Creative decisions may manifest themselves musically, such as a sung melodic fragment, or verbally, such as an edit suggestion. All will be undertaken in the context of what has gone before, and the way the co-writer receives each creative decision. A co-writer’s behaviour in the songwriting session may be modified by the Other; in BF Skinner’s terms, collaborative songwriter creativity may be said to demonstrate operant conditioning based on verbal or non-verbal stimuli and real-time reinforcements (Skinner: 1957, p.29). A Skinnerian approach (where responses to a stimulus in a given context can be mediated by conditioning from the Other) is useful because collaborative songwriters usually act as both speaker and listener, and all utterances are (time-) contextual within the part-finished song.

In the context of this “time-dependence hypothesis” of creative songwriter behaviours, I propose the term “Linear Event Analysis” in order to demonstrate that the when of a negotiated creative decision may be as important as the what. The decision can only be fully understood if we know its timing in relation to its antecedents, and therefore a linear record of events represents essential raw data for the researcher. Both interview-based and auto-ethnographic evidence bases can be suitably linear, and we can infer much about why a particular creative decision was made if we know when in the song’s creation timeline it occurred. For an example of a secondary-sources case study that takes account of the sequence of creative decisions, see my analysis of the co-writing process for the 1950s standard “Mona Lisa” (Bennett 2013).

Conclusions and application of the methods

There is no template method for writing a song. Even a cursory glance at the available evidence confirms this. But perhaps there are commonalities between the decision-making processes at a macro level. Are there ways of negotiating with the Other that are common to all collaborative songwriters? How do songwriters communicate with each other, and are their verbal behaviours tacit or explicit as they approve some ideas and reject others? Are there models of decision-making and creativity that we can apply to all collaborative songwriting, regardless of an individual song’s method, genre, listener or creators? Using the evidence base generated by these proposed observation methods, future research may provide clearer answers to these questions, toward a better understanding of the creative processes that contribute to one of music’s most culturally and economically significant art forms.


Thanks to Prof Allan Moore at the University of Surrey (UK) and Dr Phillip McIntyre at the University of Newcastle (Australia) for their advice and support in the development of this methodology. Thanks also to all the songwriters who agreed to co-write, to be interviewed or to contribute generally to my songwriting research, including Jez Ashurst, Pete Astor, Dominik Boncza-Skryzynecki, Bill Bruford, Gary Clark, Chris Difford, Sarah Doe, Boo Hewerdine, Dr Mike Jones, Sam Kennedy, Andi Neate, Richard Parfitt, Davey Ray Moor, Eddi Reader, Margo Sky, Paul Statham, Jo Stevens, Mike Stock, Richard Thompson, Chris Turpin, Midge Ure, Andy West and Andy White. I am also grateful to Randy Talmadge, the son-in-law and publisher of the late Jay Livingston, for providing detailed information about the co-writing of “Mona Lisa”.


1 Throughout this paper I shall use the capitalised term Other to describe the co-writer, as a convenient shorthand for “another individual in the collaborative songwriting team.”

2 Of the six official singles that were taken from the normal and deluxe versions of the All The Lost Souls album (2007), only two were written exclusively by Blunt.

3 A “Save As…” research method refers to the practice of creating a separate version of a computer file every time it is saved, resulting in a record of many iterative stages of the creation of a computer-based musical work.

4 I acknowledge the outdated-ness of this technical term, based as it is on a declining serial protocol developed in 1983, and employ it here only as a convenient shorthand for the process of recording note-based information in software format.

5 It is impossible to eliminate observation effects from any observation of composers completely, and any attempt to do so (e.g., concealed cameras) would be unethical.

6 This scholarly view – that creative writing’s expressiveness is weakened by the influence of prior work – was prevalent in the mid-20th century. Barthes (1953) suggests that, in literature at least, extant creative work makes the act of writing arguably uncreative.

7 For further discussion of the sonically self-referential nature of popular song recordings, see Tagg (2013).

8 The selection threshold being membership of a collection society, typically the UK’s PRS for Music.

9 Participation consent was given verbally by the participants at the start of the session.


Argyris, C., 1974. Theory in practice : increasing professional effectiveness 1st ed., San Francisco: Jossey-Bass Publishers.

Barnes, A., 2006. James Blunt goes to war with his mentor over royalties – News, Music – The Independent. The Independent on Sunday.

Barthes, R., 1953. Writing degree zero, New York: Hill and Wang.

Bennett, J., 2013. Mona Lisa by Jay Livingston and Ray Evans – analysis of a co-write. Joe Bennett website. Available at: http://joebennett.net/2013/08/08/mona-lisa-by-jay-livingston-and-ray-evans-analysis-of-a-co-write/ [Accessed August 26, 2013].

Bloom, H., 1973. The anxiety of influence: a theory of poetry 2nd ed., New York: Oxford University Press.

Boden, M., 2004. The creative mind : myths and mechanisms 2nd ed., London?: Routledge.

Burnard, P. & Younker, B.A., 2004. Problem-Solving and Creativity: Insights from Students’ Individual Composing Pathways. International Journal of Music Education, 22(1), pp.59–76.

Byrne, D., 2012. How music works, Edinburgh: Canongate.

Cantor, J.S., 2006. Fearless Innovation–Songwriting for Our Lives: Inspiring Learners with Arts-Based Practices that Support Creativity. Multicultural Education, 14(2), pp.57–64.

Collins, D., 2007. Real-time tracking of the creative music composition process. Digital Creativity, 18(4), pp.239–256.

Csikszentmihalyi, M., 1996. Creativity: Flow and the Psychology of Discovery and Invention, New York: HarperCollins.

Folkestad, G., Lindstrom, B. & Hargreaves, D.J., 1997. Young People’s Music in the Digital Age: A study of computer based creative music making. Research Studies in Music Education, 9(1), pp.1–12.

Hayden, S. & Windsor, L., 2007. Collaboration and the composer: case studies from the end of the 20th century. Tempo, 61(240), p.28.

Hennion, A., 1983. The Production of Success: An Anti-Musicology of the Pop Song. Popular Music, 3, pp.159–193.

Hewson, J., 2009. Songbook: James Blunt. Songbook: James Blunt. Available at: http://www.skyarts.co.uk/music/article/songbook [Accessed February 2, 2010].

McEwan, I., 2012. The originality of the species. The Guardian. Available at: http://www.guardian.co.uk/books/2012/mar/23/originality-of-species-ian-mcewan?fb=native&CMP=FBCNETTXT9038.

Mendes, D.Q. & de Carvalho, L.A.., 2000. Creativity and Delusions: A Neurocomputational Approach. Arxiv preprint cs/0012020.

Middleton, R., 1990. Studying popular music, Milton Keynes [England]; Philadelphia: Open University Press.

Neate, A., 2009. The Witch’s Cat, Blue Elf records.

Reitman, W.R., 1965. Cognition and thought : an information-processing approach, Wiley.

Riley, P.E., 2012. Exploration of Student Development through Songwriting. Visions of Research in Music Education, (22). Available at: http://www-usr.rider.edu/~vrme/v22n1/visions/Riley_Student_Development_Through_Songwriting.pdf [Accessed May 23, 2013].

Skinner, B.F., 1957. Verbal behavior, Appleton-Century-Crofts New York.

Sloboda, J., 1985. The musical mind : the cognitive psychology of music, Oxford; Oxford University Press.

Smart, G., 2007. Now Blunt goes back to war. The Sun. Available at: http://www.thesun.co.uk/sol/homepage/showbiz/bizarre/article246437.ece [Accessed February 21, 2010].

Tagg, P., 2013. Music’s Meanings: a modern musicology for non-musicians, Mass Media Music Scholar’s Press.