542486 research-article2014

CUS0010.1177/1749975514542486Cultural SociologyMcAndrew and Everett

Article

Music as Collective Invention: A Social Network Analysis of Composers

Cultural Sociology 2015, Vol. 9(1) 56­–80 © The Author(s) 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1749975514542486 cus.sagepub.com

Siobhan McAndrew University of Manchester, UK

Martin Everett

University of Manchester, UK

Abstract Composers generally write music alone, and we commonly understand the great figures of classical music as singular geniuses. Even where composers’ social networks and friendships are of contextual interest, it is arguable that their association with other musicians arises because they choose to socialize with similar others. However, it is also possible that creative work, even for artists as solitary as composers, depends significantly on interaction and collaboration. Certain periods and places are considered hotspots of creativity where new musical ideas are shared and movements arise. In this paper we consider the case of British classical composition, both as an example of a music network, and to contribute to debates in music history.

Keywords art worlds, composers, composing, music, sociology, creativity, social network analysis, networks, sociology of knowledge

Introduction That networks may be important for classical composition may appear counterintuitive. Composers are generally thought to be more introverted than other professional musicians such as conductors and soloists. However, music scholars and commentators have long been aware of the importance of musical relationships in influencing musical output. The archetype is Ralph Vaughan Williams (1872–1958), who strove to promote Corresponding author: Siobhan McAndrew, Marston Research Fellow, Institute for Social Change, University of Manchester, 2.13J Humanities Bridgeford Street, Oxford Road, Manchester M13 9PL, UK. Email: [email protected]

McAndrew and Everett

57

English classical music and raise the standard of concert life in Britain. His close working friendship with Gustav Holst (1874–1934) and mentorship of younger musicians is well-known: The close friendship [between Vaughan Williams and Holst] is notable because the two composers subjected their work in progress to each other’s criticism … until Holst’s death in 1934 … As a teacher of composition, [he also] shared Parry’s gift for encouraging his pupils to be themselves … often a lasting relationship resulted, as with Gordon Jacob and Elizabeth Maconchy. (Ottaway and Frogley, n.d.)

In contrast, Kaikhosru Sorabji (1892–1988) was a composer who eschewed the classical music world to the point of self-sabotage: he remained an outsider as a critic and composer, owing to his anti-establishment views, private training, racial origins, homosexuality and self-described ‘mania for privacy’ … Because he was financially secure, Sorabji did not need an income from his music criticism and felt no pressure to make it tactful. (Rapoport, n.d.)

Sorabji is now best known for having written highly complex works, mostly for the piano, and many of which are difficult to programme. It was also commonly thought that he had banned performances of his works for some 40 years. These conscious decisions and his isolation, in the face of a mixed but often favourable critical reception, must have hindered his work in receiving due recognition. In his case, talent was not enough for conventional success. In this article, we investigate afresh the question of the social context of British classical music from 1870. We examine the network of British composers to address the following: to outline the possibilities of social network analysis (SNA) for research into music and culture; to describe the network of British composers and how it is structured; and to explain how particular movements may relate to individual and whole network-level connectedness. Our final motivation is the eternal interest in the question of why the great are great, and how context accounts for extreme achievement. Examples include quantitative research into inventors and entrepreneurs in the technological history tradition (Baumol et al., 2009; Lamoreaux and Sokoloff, 2005), and superstar effects in media, entertainment and CEO markets (Rosen, 1981). A recent analysis of collaborations in movie-making has established that network centrality is related to success at the Oscars, suggesting that selection into elite networks determines individual accomplishment (Rossman et al., 2010). We examine whether similar features hold for composers.

Social Networks and the Social Science of Music Why might networks matter for composers? Talent and status attract network connections, but networks may independently foster creative output. Composers acquire tacit as well as formal musical knowledge from networks of teachers and peers; they build on this to create their own styles and musical innovations. They use social networks to signal their own and others’ talent to patrons, agents, concert promoters and publishers.

58

Cultural Sociology 9(1)

Finally, composers may co-operate practically to stage concerts and promote each other’s new works, and share money and resources, including knowledge on employment opportunities, commissions and so forth. The social capital generated through network membership also gives rise to trust between members, and this facilitates musical and non-musical interactions alike. These networks enable what Allen (1983) has termed ‘collective invention’ among multiple composers which give rise to distinct musical movements and periods of intense creativity. Interpreting classical music as involving collective creation is more widely informative. Music is a form of knowledge and its performance a social action; understanding musical innovation is not only of interest to musicologists concerned with technical forms, but also to sociologists of knowledge and historians of innovation. Analysing music as being created by networks within ‘art worlds’ links to a tradition beginning with Becker (1982) and subsequent work by Faulkner (1983), Finnegan (1989), Crossley (2008) and Bottero and Crossley (2011). Following Faulkner and Crossley, we use the techniques of social network analysis (SNA) for research into the interactions and relationships associated with art worlds – a set of techniques relatively little-used by sociologists of music, or indeed in the arts and humanities. While network analysis is well-known to have a wide range of applications in the social and physical sciences (Borgatti et al., 2009), applications to music (and indeed the arts and humanities) are relatively few. SNA offers real opportunities to musicologists with interests in sociology or ethnography, and who often collect a great deal of data on musical relationships which would benefit from formal analysis. Moreover, it can be used alongside historical or text-analytic methods to consider the position of individuals embedded within subgroups and communities; their connections to institutions, events and organizations; how musical phenomena relate to connections; and possible mechanisms regarding the creation of music. Alternative approaches also offer insights on the importance of relationality and proximity for creativity and music in particular. Numerous spatial economists and economic geographers have identified how and why proximity and agglomeration matters for creativity, including sophisticated analyses of classical music in particular (Florida, 2002; Borowiecki, 2013a, 2013b; Comunian, 2012). Knowledge can be diffused more effectively when agents are closer together, while agglomeration enables economies of scale, economies of diversity in terms of the variety of inputs available, and economies of scope in terms of complementarity of outputs (for example, music teaching and performance for an individual musician). Spatial and network approaches share an interest in the importance of context for social action: proximity or connectedness is thought to enable the direct transfer of information or social influence between agents. Interdependence, whether across spatial units or between actors in a social network, has similar methodological implications; indeed, network autocorrelation models have their roots in spatial models (Cliff and Ord, 1975). The conceptual and methodological overlap also arises because the probability and frequency of tie occurrence strongly correlates with physical distance. Studies suggest that for people to establish and maintain relationships with others they must be located in the same place for a significant period, even when communication technologies and transport serve to ‘shrink distance’ (Butts and Carley, 2000: 2). The appropriate choice of method will generally be decided by the hypothesized data-generating process, adequacy of data, and whether the researcher’s

McAndrew and Everett

59

interest is in the properties of relationships between people or places. In our case, composers frequently formed their professional networks and key influences during formative periods – their time as students and at conservatoires in particular – while they often worked solitarily once established. We therefore consider network connections rather than the place of professional work to be of primary interest here. It should also be acknowledged that some anthropologists find formal network analysis to be exclusive and two-dimensional (Finnegan, 1989: 305), while others argue for the virtues of the cultural turn in SNA and for a variety of approaches to understanding place, network and group (Knox et al., 2006). Sociologists of music have also raised questions of how music should be examined in relation to its social context, debating whether music is ultimately a social or aesthetic phenomenon. DeNora (2003) argues that we should avoid examining the ‘sociology of music’, which implicitly posits that music is separate from society; instead we should investigate music as a social agent via ‘music sociology’. Marshall, however, calls for greater intradisciplinary research by sociologists, arguing in this journal that ‘the “sociology of popular music” has not been sociological enough’ (Marshall, 2011: 155). Regarding the role of networks, Lizardo (2006) has also suggested that cultural tastes might shape personal networks rather than the reverse; in other words, culture may shape networks rather than the converse. Clearly, it is conceptually difficult to disentangle music from its social context, and we do not answer these questions definitively here. Instead, our aims are to describe data on composers and to draw out patterns in composition, informed by historical and musicological scholarship on British classical music. There is also an extensive literature on the economics and sociology of creativity, with growing interest in network effects. Skills are learned and knowledge transferred via relationships, and creativity requires learning. Malcolm Gladwell’s bestselling Outliers has publicized the ‘10,000 hours’ theory, whereby genius is made, not born, via intensive and goal-directed learning (Ericsson et al., 2007: 119–20). For musicians, at least some of this learning is acquired via face-to-face teaching. In addition, musicians are often drawn together to be with like minds, learn from peers, and gain access to new ideas and thereby short-cut their path through the required knowledge. Composers also need to absorb knowledge in order to contribute original works to the canon – and with every successive year that passes, this ‘burden of knowledge’ only increases: information is both input and output of its own production process. In order to write today’s academic or news article, I need access to yesterday’s articles and reports. In order to write today’s novel, movie, or song, I need to use and rework existing cultural forms, such as story lines and twists. (Benkler, 2006: ch. 2)

Economist William Baumol summarized this as the need to ‘stand on the shoulders of giants’. However, intensive training may lead to what psychologists call Einstellung: inflexible thinking and overreliance on previous precepts, leading to stale ideas and suboptimal technique (Baumol et al., 2009: 714–15, 724–5). The ‘marginal person’ thesis accordingly suggests that people on the margins, who may participate in multiple intellectual domains, but are central to none, are more likely to introduce creative breakthroughs. They have different skills, permitting novel outcomes, and attempt riskier

60

Cultural Sociology 9(1)

work as a route to recognition. The paradox that networks may both enable and constrain was addressed by network scholar Ronald Burt in his study of idea generation in a large technology company: good ideas emerged from brokers at the intersection of different networks, but diffused most thoroughly within groups (Burt, 2004: 394). There is a third way in which networks may matter for creativity. Even informal networks constitute governance structures which mitigate against contractual risk, and provide what economists call an ‘appropriability mechanism’. The creative sector is highly prone to risk, since it is difficult to tell whether a new work or concept will succeed: the ‘nobody knows’ principle (Caves, 2000: 3). However, the status and pecuniary rewards of musical success can be high. Those who contribute ideas to a movement, or other social or practical resources, may benefit by having their co-operation recognized by reputation even if their own works do not succeed commercially; by contrast, those who draw on others’ contributions without reciprocating are liable to be sanctioned. Further, when choosing whether to collaborate or share opportunities with a fellow musician, reputations can be communicated between network members. In economic terms, the network serves as a non-hierarchical organizational form, governing the sharing of knowledge and other social resources, and the pooling of risk (Williamson, 1996: 115). In sum, networks will matter for music composition for three reasons. First, they transmit knowledge. Secondly, particular positions within a network, and different types of personal network, may be associated with more or less innovation. Thirdly, networks help reduce the risk and appropriate the benefits associated with developing new knowledge.

Networks in British Classical Composition Musicologists and historians have shown considerable interest in connections among composers. There has been some discussion of the extent to which British classical music of the 19th and 20th centuries has been associated with an ‘establishment’ or dominant cliques. Stradling and Hughes wrote a controversial text in 1993 (revised 2001), The English Musical Renaissance, 1840–1940: Constructing a National Music, which argued that a ‘super-league’ of Edward Elgar, Frederick Delius, Ralph Vaughan Williams, Gustav Holst, William Walton, Michael Tippett, and Benjamin Britten was consciously instituted by ‘a self-appointed and self-perpetuating oligarchy’ (Stradling and Hughes, 2001: 285, xv). In particular, they criticized the centrality of Vaughan Williams as an opinion leader and part of an axis at the Royal College of Music (Stradling and Hughes, 2001: 93, 101, 195). They also highlighted cases of apparently unjustly neglected composers, in particular Havergal Brian (1876–1972) and Samuel Coleridge-Taylor (1875– 1912). Both were of working-class origin; Coleridge-Taylor was additionally mixed-race with a Sierra Leonean father. While Coleridge-Taylor was highly successful early in his career, they suggest that he was increasingly excluded less because of his ‘racial origins … [than] his identification with Black causes’ (Stradling and Hughes, 2001: 248). Brian they describe as having been ‘gradually pushed into obscurity: the “unknown warrior” of English music’ (Stradling and Hughes, 2001: 139). The book was reviewed extensively by music scholars and historians who acknowledged its depth of original archival research but highlighted inattention to the music

McAndrew and Everett

61

itself, failure to set England in a European context, and a tendency to implicit character assassination (Frogley, 2003; Smith, 1993). Despite its flaws, one musicologist allowed that: it must be emphasized that some sort of establishment consensus surely did grow up during the period … [w]here they are mistaken is in attempting to single out the perpetrators and their misdeeds. (Onderdonk, 1995: 64)

That links existed between composers for the transmission of knowledge, generation of new musical ideas, and to further a common musical movement is, however, what we would expect. Accordingly, to investigate the relationship between network position and reputation, we examined the 505 biographies of British composers (and composers active in Britain) born between 1870 and 1969 included in Oxford Music Online, the portal which hosts inter alia the New Grove Dictionary of Opera and The New Grove Dictionary of Music and Musicians. Biographers follow a semi-structured template, providing information on nationality, education, and their life history where it captures factors pertinent to career success, before an analysis of their works. A large number of biographers were involved, so that we are less reliant on the interpretation of a single author; furthermore, links to composers could be mentioned in their own biographies, while they might also feature in others’ biographies. We originally chose this period to examine the ‘English Pastoral School’ in detail, together with composers active later in the 20th century. We define a network connection as existing when the biographer describes personal contact between two composers: it is not enough for an influence to have been cited, or for a composer to have simply attended a class given by another. We determine a personal contact to exist where biographers describe composers as family members, having corresponded, met in person, worked together, to have been friends, or to have had a teacher-student relationship. While biographers do not always provide detail on frequency or length of contact, we have little reason to think that they cite a relationship as existing on the basis of a single or inconsequential meeting. We treat the resulting list as a ‘whole network’ with a birth year of 1870 or after as the starting date and a requirement that the composer is British or active in Britain. We also included Irish musicians reaching 18 before 1922, when the Republic of Ireland seceded. Those emigrating before the age of 18 were removed, as were jazz musician-composers, whose career paths are typically very different. However, popular musicians were retained on the basis that their works often cross over to classical genres, or because they had extensive classical training – a group which includes Ivor Novello, Noel Coward, Elton John, Brian Eno and Paul McCartney. We retained British composers mostly active abroad, because their career might nevertheless be related to their position in British networks. Similarly, foreign-born composers active predominantly in Britain were retained because biographers often describe them as contributing to the British classical tradition. As a single, easily-available measure of success, we used the number of works each composer had had featured in the annual Promenade concerts from 1890 to 2011, available from a searchable online database at http://www.bbc.co.uk/proms/archive. This is not without problems, because songs are given equal weight to symphonies, and because

62

Cultural Sociology 9(1)

Table 1.  Summary of variables capturing composer characteristics. Composer characteristics

Mean or percentage (%)

Composer characteristics

Percentage female Percentage other than white Percentage born 1870–1889 Percentage born 1890–1909 Percentage born 1910–1929 Percentage born 1930–1949 Percentage born 1950–1969

13.5 1.2

Percentage British Percentage English

19.0 8.0 34.0 36.0 13.0

Percentage Scottish Percentage Welsh Percentage Irish Percentage Northern Irish Percentage other nationality Average number of (directed) ties with other composers Average number of Proms works Percentage receiving knighthood or damehood Number of composers

Percentage composer as primary occupation Percentage primarily instrumentalist Percentage primarily teacher Percentage primarily other occupation

82.0 7.0 10.0 10.0

Mean or percentage (%) 8.0 74.0 6.0 4.0 3.0 1.0 4.0 2.1 4.0 3.6 505

works performed repeatedly are given the same credit as a work performed once. Nevertheless, classical composers are often as interested in writing a number of individual works for few performances as a single hit. The data gathered from the biographical and Proms archive sources are summarized in Table 1. Examining the network as a whole via Figure 1, we find a core-periphery pattern, whereby some figures are central and many others marginal. This is typical, and not suggestive of exclusivity. 322 of the 505 (64%) are linked to the main network. The remainder are isolates with no links named in their own biographies (nor are they named in others’); or components of size 2, where a duo is named as linked solely to each other (such as Paul McCartney and George Martin). Notably, this single network spans the entire 1870–1969 period, suggesting that the British classical tradition of the 20th century can be examined as a single movement.

Measuring Composer Centrality Besides visualizing ties thus, we measure the connectedness of the composers three ways, using UCInet (Borgatti et al., 2002). The first, degree centrality (Freeman, 1979), captures information on how many ties to others each composer has, with data for the top-ranked 20 composers given in Table 2. This score is normalized, yielding a measure ranging from 0 to 1. For this we combine information on names given as contacts in each composer’s biography, but also judge that a tie exists where they are cited in others’ biographies (namely symmetrized or undirected ties).

63

McAndrew and Everett

Figure 1.  The British network of composers, 1870–1969 (components of size 2 and isolates removed). Table 2.  Twenty composers ranked by degree centrality. Rank

Composer

Degree Centrality

Rank

Composer

Degree Centrality

1 2 =3 =3 =5 =5 7 8 9 10

R. Vaughan Williams Gordon Jacob Herbert Howells R.O. Morris Lennox Berkeley Alexander Goehr Benjamin Britten Gustav Holst Matyas Seiber Alun Hoddinott

0.112 0.078 0.069 0.069 0.062 0.062 0.047 0.044 0.040 0.037

= 11 = 11 = 11 = 14 = 14 = 14 = 14 = 14 = 14 20

Alan Bush Edmund Rubbra John Ireland Harrison Birtwistle Michael Tippett Howard Ferguson Patrick Hadley Gerald Finzi Bernard Stevens Constant Lambert

0.034 0.034 0.034 0.031 0.031 0.031 0.031 0.031 0.031 0.028

Notably, no women feature in this group, and the majority appear to be composers active before the Second World War. By this measure, Vaughan Williams is the most central by some distance. Gordon Jacob, Herbert Howells, R.O. Morris and Lennox Berkeley, all important teachers, take the following four places. Leading contemporary composer Alexander Goehr, who met many older composers through his orchestral musician father, ranks alongside Lennox Berkeley. Benjamin Britten has been noted to

64

Cultural Sociology 9(1)

have ‘consciously distanced’ himself from leading composers of his own and earlier generations, but nonetheless features here (Brett, n.d.). This is partly because a number of young composers worked as his assistants – Imogen Holst and David Matthews, among others. Gustav Holst also scores highly, as does Matyas Seiber, another important teacher. Welsh composer Alun Hoddinott, who approached numerous distinguished composers at festivals as a student to discuss his early works, completes the top ten (Lewis, n.d.). Degree centrality depends simply on the number of connections each node has. An extension is eigenvector centrality (Bonacich, 1972), which ranks composers by how well they are connected to highly-connected others. Table 3 shows that by this measure Vaughan Williams continues to lead, and Jacob and Morris retain top-five positions. Finzi and Rubbra, both associated with the English ‘Pastoral School’, also rank highly. Four women achieve top-20 positions: the Welsh composer Grace Williams, Elizabeth Maconchy, Dorothy Gow, and the composer and conductor Ruth Gipps. Samuel Coleridge-Taylor also has a relatively high score, suggesting that he was relatively well-connected through his proximity to the highly-connected. Notably, Britten is not found here. The third measure we report here is betweenness centrality, in Table 4. Composers with high scores on this measure sit on many of the ‘shortest paths’ connecting others and are likely to have higher ability to control the flow of information or the exchange of resources (Freeman, 1977: 35). By this measure, Vaughan Williams is again the leader, with Lennox Berkeley also ranking highly. Britten resurfaces to score highly on this measure. Socialist composers Alan Bush and Bernard Stevens also figure, as does Humphrey Searle, a pioneer of serialism in Britain. Taking these results together, what is striking is the high centrality of Vaughan Williams. While Stradling and Hughes consider his role to have been partly malign, an alternative interpretation is that he invested heavily in the community of composers. Providing encouragement and mentorship can drain time and energy; this was a substantial contribution to the community of musicians. By comparison, Britten – the artist with the most works in the Proms, and who might be thought to attract many citations in others’ biographies as a high-status composer – is less connected, in particular to the highly-connected, although he exhibits ability to access and control information. William Walton (1902–83) achieved great success during his lifetime but does not feature here. While women do not feature in Tables 2 and 4, suggesting a lower absolute number of links and less of a brokerage role, a number feature high eigenvector centrality. It is not immediately obvious, therefore, that women are particularly segregated or excluded once they access the network, although their low overall representation is apparent from Table 1.

Communities of Composers within the Network Of additional interest is the clustering or grouping within the whole network. One method of doing so in UCINET is via the Girvan-Newman algorithm (Girvan and Newman, 2002) which splits the network into a number of separate groups by identifying edges which hold the network together. This method leads to the identification of five distinct

65

McAndrew and Everett Table 3.  Twenty composers ranked by eigenvector centrality. Rank

Composer

Eigenvector Centrality

Rank

Composer

Eigenvector Centrality

1 2 3 4 5 6 7 8 9 10

R. Vaughan Williams Gordon Jacob R.O. Morris Gustav Holst Gerald Finzi Edmund Rubbra Grace Williams Howard Ferguson Robin Milford Stanley Bate

0.678 0.408 0.378 0.285 0.270 0.229 0.209 0.203 0.194 0.187

11 12 13 14 15 = 16 = 16 18 19 20

Alun Hoddinott Elizabeth Maconchy Cedric Thorpe Davie Herbert Howells Dorothy Gow Ruth Gipps Inglis Gundry Samuel Coleridge-Taylor Constant Lambert Ivor Gurney

0.187 0.183 0.182 0.180 0.178 0.176 0.176 0.172 0.166 0.162

Table 4.  Twenty composers ranked by betweenness centrality. Rank

Composer

Betweenness Centrality

Rank

Composer

Betweenness Centrality

1 2 3 4 5 6 7 8 9 10

R. Vaughan Williams Lennox Berkeley R.O. Morris Alexander Goehr Benjamin Britten Howard Ferguson Herbert Howells Gordon Jacob Alan Bush Michael Tippett

0.196 0.157 0.124 0.122 0.115 0.106 0.105 0.101 0.099 0.087

11 12 13 14 15 16 17 18 =19 =19

Alun Hoddinott Bernard Stevens John Ogdon John Ireland Richard Rodney Bennett Patrick Hadley Robert Sherlaw Johnson Matyas Seiber Humphrey Searle Edmund Rubbra

0.086 0.076 0.064 0.058 0.057 0.053 0.050 0.049 0.045 0.045

communities, which appear most likely associated with educational establishments. We examine these in turn. Even for those largely self-taught, music colleges act as nexuses of talent through hosting concerts, through faculty providing private tuition alongside formal programmes, and as employers. Community 1 shown in Figure 2 is the largest by some distance and appears to be associated (although not exclusively) with the Royal College of Music. While the community might be considered to overlap with the ‘Pastoral School’ and the putative musical establishment (note Vaughan Williams’ position), it appears larger and more diverse than the Stradling and Hughes thesis would imply. For example, this community includes figures such as the film composer John Barry (1933– 2011), and Errollyn Wallen (b.1958), one of the few ethnic minority composers in our sample. Barry was largely self-taught while Wallen studied at Goldsmiths and Kings College London. This community also hosts composers from the whole period; and again, Coleridge-Taylor appears relatively central. We can also identify known cliques

66

Cultural Sociology 9(1)

John Barry

Errollyn Wallen

Roger Quilter Norman O'Neill

Ivor Novello

Edmund Rubbra

Cyril Scott

Henry Balfour Gardiner

Elizabeth Maconchy

Gerald Finzi Ralph Vaughan Williams

Herbert Howells

Grace Williams

Samuel Coleridge-Taylor

Benjamin Britten Peter Warlock William Walton

Kaikhosru Shapurji Sorabji

Figure 2.  Community 1: A group associated with the Royal College of Music with important composers labelled.

such as the ‘Frankfurt Group’ of Norman O’Neill, Roger Quilter, Henry Balfour Gardiner and Cyril Scott within this community. The second community (Figure 3) is centred on figures attached to the adult education institution Morley College. Seiber was an influential teacher here; it is interesting to note that Vaughan Williams, Holst, Michael Tippett and Cornelius Cardew, who also had teaching posts at Morley College, are not located in this subgroup, suggesting that the communities are not determined purely by institutional affiliation. Alan Bush (1900–95) is also relatively central to this community; a communist embargoed by the BBC during the 1940s, he founded the Workers’ Music Association and composed works on political subjects (Christiansen, 1995). He was also associated with the Royal Academy of Music until 1978 where, among others, he taught Michael Nyman, known as composer of the soundtrack for The Piano (Mason et al., n.d.). Similarly, the communist composer Bernard Stevens (1916–1983) is also located here, despite having trained with Cyril Rootham, R.O. Morris and Gordon Jacob, all relatively central figures (MacDonald, n.d.). The third community (Figure 4) is largely centred on the pre-war RAM, where Benjamin Dale (1885–1943) and York Bowen (1884–1961) were influential. A Welsh cluster is also visible following Ian Parrott’s move to the University College of Wales, Aberystwyth, where he taught David Harries and William Mathias. The fourth community (Figure 5) identified by the algorithm is apparently centred on the composer and teacher Lennox Berkeley, who studied at Oxford and with Nadia Boulanger in Paris, taught at the RAM, and was also father of noted composer Michael

67

McAndrew and Everett

Antony Hopkins Alan Gibbs

Malcolm Lipkin

John Mayer

Roger Smalley

Simon Holt

Francis Routh Norman Del Mar

Anthony Gilbert

Matyas Seiber

David Cunningham

Douglas Young Peter Racine Fricker Justin Connolly

Anthony Milner

Angela Morley

Michael Nyman

Hugh Wood Minna Keal Alan Bush

Iain Hamilton William Alwyn

Edward Gregson

David Matthews Roger Steptoe Giles Swayne

Doreen Carwithen

Rachel Portman

Colin Matthews Nicholas Maw

Piers Hellawell

James Wood

Figure 3.  Community 2: A community associated with Morley College and the BBC.

Stanley Marchant David Dorward William H. Harris

John Gardner

Ernest Walker Herbert Murrill

Christopher Steele York Bowen

Kathleen Dale

Arnold Richardson

Derek Holman David Harries Benjamin Dale Ian Parrott

Mansel Thomas

Patrick Piggott

William Mathias

Figure 4.  Community 3: A community centred on the pre-war Royal Academy of Music.

Berkeley. Many of the composers in this community are still active. We can also identify some local clusters within this community: first, a Scottish cluster of James MacMillan, Sally Beamish and Martin Dalby, head of music for BBC Scotland from 1972 to 1991.

68

Cultural Sociology 9(1) Peter Wiegold Jonathan Lloyd Edwin Roxburgh

Richard Barrett Graham Fitkin

Roger Redgate Arthur Hutchings

Ivan Moody Peter Wishart

Rebecca Saunders Nigel Osborne

Philip Alexander Hammond Brian Ferneyhough John Tavener Kenneth Leighton Clifford Crawley Geoffrey Burgon Stephen Oliver Raymond Warren Judith Bingham Christopher Headington Betty Roe Eleanor Alberga Michael Hurd Bernard Rose John Manduell John McLeod Eric Fenby Tim Souster Lennox Berkeley Richard Stoker Michael Berkeley Christopher Brown Graham Whettam Peter Dickinson Richard Rodney Bennett Francis Burt Christopher Gunning David Bedford Paul Patterson John A. Speight Debbie Wiseman Christopher Bochmann Buxton Orr Mike Oldfied Michael Young Howard Ferguson Cornelius Cardew John Casken James MacMillan Benjamin Frankel Christopher Hobbs John Joubert Sally Beamish Michael Parsons Howard Blake Michael Alcorn Martin Dalby John Cale John White Howard Skempton David Byers

W.H. Bell

Dave Smith

Benedict Mason

Figure 5.  Community 4: A community associated with the post-war Royal Academy of Music.

Secondly, the cluster of W.H. Bell, John Casken, John Joubert and Michael Alcorn is identifiable through their sharing a non-metropolitan status based largely in university departments. Bell emigrated to South Africa at the age of 39 in 1912, where Joubert was among his students; Casken trained at Birmingham with Joubert, later spending his career at Huddersfield, Durham and thereafter Manchester; Alcorn studied at the Universities of Ulster and Durham and is currently Director of the School of Music and Sonic Arts at Queen’s University Belfast. Post-war composers faced a more diverse set of opportunities than those before the war, particularly outside London and in university rather than conservatoire settings. That an increasing proportion of the younger composers have pursued doctoral study in music is also suggestive of a movement of composers from freelance to academic employment. A final subgroup of interest within this larger community is that centred on Cornelius Cardew (1936–81) and a number of experimental composers associated with the Scratch Orchestra (Pisaro, n.d.). This was founded by Cardew in 1969 with Michael Parsons and Howard Skempton, evolving from his class at Morley College. The orchestra had an anarchist structure and was open to those with no musical training. It folded by the mid1970s, but is considered to have spurred the careers of a diverse range of talents, including Gavin Bryars and Christopher Hobbs among others (Prevost, 2001: 25–8). Hobbs in particular is known for his musical ‘readymades’, using patterns from objects such as a knitted sweater or Sudoku puzzles as a template (Anderson, n.d.). This latter group suggests that musical movements are a phenomenon associated with groups at a lower level of aggregation than the community, but where bonds are weaker and more extensive than within close friendships.

69

McAndrew and Everett John Buller John Woolrich William Lovelock

Andrew Ford

Havergal Brian

Simon Bainbridge

Eric Fogg

Robert Simpson

Philip Cashian George Lloyd Edward Cowie Mark-Anthony Turnage John McCabe Silvina Milstein Judith Weir Nicholas Sackman George Benjamin Thomas B. Pitfield Oliver Knussen David Fanshawe Janet Beat Philip Wilby Ronald Stevenson Julian Anderson John Lambert Jonathan Dove John Ogdon Alexander Goehr Robin Holloway Param Vir Peter Maxwell Davies Arthur Butterworth Geoffrey Poole Avril Anderson Richard Hall Francis Shaw Bayan Northcott Robert Saxton Delia Derbyshire Harrison Birtwistle William Sweeney Jonathan Harvey Erika Fox Robert Sherlaw Johnson Hilary Tann Philip Grange

Julia Usher

Elisabeth Lutyens

Anthony Powers

Enid Luff Janet Graham Anthony Payne

Dominic Muldowney

Andrew Toovey Christopher Fox

David Blake Bernard Rands

Gordon McPherson Thomas Simaku Peter Aston Roger Marsh Wilfrid Mellers Vic Hoyland Reginald Smith-Brindle

Vivienne Olive

David Sawer

Figure 6.  Community 5: A community centred on the post-war ‘Manchester School’.

The final community (Figure 6) has the former ‘Manchester School’ at its centre. This had a ‘devotion to the ideas and techniques of Viennese modernism’ (Cross, n.d.). The composers Peter Maxwell Davies, Alexander Goehr and Harrison Birtwistle met in Richard Hall’s composition class at the Royal Northern College of Music in the 1950s, and also became close to the pianist John Ogdon and conductor Elgar Howarth. Alexander Goehr, raised in a musical family in London, has been described as ‘their intellectual leader’: ‘[I was] trying to move the others towards some sort of an artistic movement. I was trying to state what was and what was not real modern music’ (Cross, n.d.). While students, they formed the New Music Manchester group to play their own compositions alongside works of the continental avant garde. Also located here are composers linked to the ground-breaking music department at the University of York, established by Wilfrid Mellers in the 1960s. Peter Aston and David Blake were early appointments, as was the contemporary composer Bernard Rands, who worked there from 1970 to 1975 before moving to the US. The British serialist Elisabeth Lutyens (1906–83), known for her critique of the English Pastoral School as ‘cow-pat music’, and one of the older members of this community, is also located here with a number of her students (Saylor, 2008: 54). A composer of note on the fringe of this community is the aforementioned ‘unknown warrior’ Havergal Brian (1876–1972). He received much attention during 2011 following the programming of his Gothic Symphony at the 2011 Proms. Of working-class origin, he received the patronage of a Potteries industrialist early in his career, as well as encouragement by composers Edward Elgar, Frederick Delius, and Granville Bantock; conductors Thomas Beecham and Henry Wood; and critic Ernest Newman. However,

70

Cultural Sociology 9(1)

Brian’s personal life was complex: he moved to London with his mistress at the age of 37, eventually having ten children in all. To support his family and first wife he was drawn into journalism, finally taking a post in the Civil Service but nonetheless continuing to compose, writing 21 symphonies after the age of 80. Despite his ‘outsider’ status his continued efforts were championed by Robert Simpson at the BBC, with some BBC premieres and broadcasts. A group of aficionados has continued to draw attention to his output, and some hand-written manuscripts have been transcribed by hobbyists using Sibelius software, which has a ‘playback’ feature allowing the works to be heard. In 2011 the Gothic, thought to be the largest symphony ever performed in terms of forces (for large orchestra, quadruple chorus, children’s chorus, and four brass bands) and one of the longest ever written, was performed at the Proms. Many of his most important musical links predate our sample, except for the critical link with Robert Simpson, and his musical instincts as a composer and critic ran counter to contemporary trends. Indeed, in 1925, when asked to provide a summary of his career for a directory of contemporary British composers edited by the composer Joseph Holbrooke, he replied, ‘I have no wish to be in any book … the “Hidden Hand” [of the music world] is too much for me, and any discussion of my work is useless’ (Stradling and Hughes, 2001: 202). His longevity and productivity in pensioned retirement also enabled him to effectively live two careers. His own temperament, and artistic and life choices, undoubtedly affected his career and reception. A final means by which we can potentially identify a core sphere of influence within the whole network is through calculation of a ‘k-core subgraph’ (Seidman, 1983), namely a graph of those who have at least a certain number (in this case four) contacts with others in the sample. This result is of interest (Figure 7) in that it does appear to summarize the core of 20th-century British composition efficiently. Holst and Vaughan Williams are represented, as are ‘Pastoral School’ composers Edmund Rubbra and Gerald Finzi. Again, Samuel Coleridge-Taylor is featured, further reducing suspicion that he was marginalized, as are Elizabeth Maconchy and other female composers, known to historians as a group of close friends mentored by Vaughan Williams (Doctor, 1998). This subgraph also throws into relief the social distance between Vaughan Williams, Walton and Britten, raising the issue of whether success requires a measure of independence from powerful others. Serialist composer Humphrey Searle – a non-obvious link – bridges Vaughan Williams and Walton. Hoddinott and Tippett bridge the pre-war and post-war worlds. The ‘Manchester Group’ is also represented here. No composer born after 1940 is represented, suggesting either that their biographers are less concerned with mapping the musical friendships they have, or that the nature of careers has changed – composers are more likely to study and work overseas, and to have careers with attachments to universities outside London where networks are sparser.

Composers’ Centrality and Prestige Finally, we examine the association between centrality and musical productivity, using the number of works composed by each artist programmed up to 2011 at the Henry Wood Promenade Concerts as a measure of output. Henry Wood (1869–1944) began conducting this series in 1895; in 1927 the BBC became the concerts’ patron. Wood was

71

McAndrew and Everett

Peter Maxwell Davies Richard Hall

Harrison Birtwistle John Ogdon Bernard van Dieren

Alexander Goehr William Walton

Michael Tippett

E.J. Moeran Peter Warlock Alun Hoddinott

Benjamin Britten

Constant Lambert Humphrey Searle

John Ireland

Grace Williams

R.O. Morris

Ralph Vaughan Williams Gordon Jacob

Samuel Coleridge-Taylor Fritz Hart

Gustav Holst

Dorothy Gow

Edmund Rubbra

Elizabeth Maconchy

Imogen Holst Gerald Finzi

Robin Milford

Figure 7.  Four-core subgraph of 20th-century British composers.

committed to increasing the ‘serious’ component which he achieved over the half-century he ran the Proms: throughout his career he sponsored no fewer than 717 first performances of works by 357 composers. Figure 8 depicts a target diagram of composers with those with higher eigenvector centrality at the centre. Each node is scaled to reflect the number of works each composer has had featured in the Proms. Vaughan Williams, who is the most central by this measure, is not quite the most successful in terms of number of works performed; Britten with 94 outranks both Vaughan Williams (71) and Walton (52). We have already noted a certain social distance between the three. Indeed, the editors of a collection of Britten’s letters summarized how Britten had an uneasy relationship [with Vaughan Williams]. He had little enthusiasm for Vaughan Williams’s works and remembered what he took to be a generally unsympathetic attitude on Vaughan Williams’s part during his student years. (Mitchell et al., 2004: 625)

Walton’s biographer notes that Britten and Walton first met over lunch in 1937, after which Britten wrote: ‘He is charming, but I always feel the school relationship with him – he is so obviously the head-prefect of English music, whereas I’m the promising new boy’ (Lloyd, 2001: 200). Relations between them were ‘strained’ although Walton acted as Britten’s witness at his conscientious objection tribunal in 1942, for which Britten was very grateful (Lloyd, 2001: 201). This was partly because ‘[n]ot all who knew [Walton] found him easy to deal with or appreciated his shafts of acid’ (Lloyd, 2001: 262).

72

Cultural Sociology 9(1)

Eric Coates

Peter Maxwell Davies

Benjamin Britten

William Walton

Ralph Vaughan Williams

Montague F. Phillips Hermann Lohr

Figure 8.  Target diagram of composers by eigenvector centrality.

It is plausible that Britten and Walton were careful to maintain distance from other central figures, and that their personal and musical relationships were more intentionally those which furthered their own work. By comparison, Vaughan Williams appears highly community-minded. While biographers of his students and friends may well have ‘overcited’ his influence and interest to add status, he was undoubtedly generous in his friendship, mentorship and service to an extraordinary extent. The graph also illustrates how some peripheral figures succeeded in having a large number of works performed, notably Eric Coates (1886–1957), Montague Phillips (1885–1969), and Hermann Loehr (1871–1943). Coates composed the music for The Dambusters and ‘By the Sleepy Lagoon’, the theme for Radio 4’s Desert Island Discs. His biographer describes how ‘his success derived from his recognition of, and adjustment to, new trends’ (Self, n.d.). Phillips wrote a number of ballads for his soprano wife, as well as light pieces for orchestra; his serious works were less successful (Lamb, n.d.[b]). Loehr was also a balladeer, having composed the wildly popular Little Grey Home in the West among others (Lamb, n.d.[a]). Their success appears to relate to their specialism in lighter and more popular works; this suggests that network centrality here is associated with serious music. It is also telling that Coates was a founder member of

McAndrew and Everett

73

the Performing Rights Society, established to protect intellectual property in recorded and broadcast music. In sum, success at the Proms appears to be associated with relative centrality, but the most central figures are not necessarily the most successful, and some marginal figures were financially successful. We can model this more formally using regression analysis, using the number of works featured in the Proms as the dependent variable and the following predictors: birth cohort, sex, the number of professional roles used to describe the composer (ranging from one to four), nationality, and measures of centrality. Because 252 of the 505 composers have had no work featured in the Proms, this creates a selection or unobservable heterogeneity problem for conventional linear modelling. We only observe Proms success for those composers who cross the threshold of having any works featured at all – perhaps because they possess a certain level of talent, motivation, or recognition. Application of a Heckman correction and use of finite mixture modelling are two of the most common methods to deal with this. The diagnostics for sample selectivity bias were mixed when applying different versions of the Heckman correction. Because of this, and because of its flexibility with regard to the distribution of the dependent variable, we therefore estimate a finite mixture model, also known as a semi-parametric heterogeneity model, or latent class model (Cameron and Trivedi, 2009: 621; Van Dijk, 2009: 1). Finite mixture models assume that two or more distinct distributions exist – such as less talented and more talented composers – which we observe with error and furthermore may overlap. Accordingly, it allows for population heterogeneity, without a sharp dichotomy between groups. The model separates the distributions semi-parametrically, with the explanatory variables predicting relative success within each component. Since our outcome variable – the number of works performed in the Proms – is a count, one with many zeroes and which is over-dispersed, we specify the component densities as negative binomial. We then fitted finite mixture models with an increasing number of components, and examined model fit statistics. The Akaike and Bayesian Information Criteria (AIC and BIC) indicated that two components provided a better fit than one or three. As reported in Table 5, with the two component model, we find that an estimated 13 per cent of composers are predicted to fall into the high-achieving group, while 87 per cent are lower-achievers. The entropy score of 0.87 is reasonable (1 equates to perfect separation of the components), suggesting that the two-component model is relatively successful at separating the more successful from the less successful. We look first at results for the less successful group, Component 1. Here, being born in the three later cohorts compared to the earliest is significantly associated with less success, while members of the second birth cohort are not significantly different from those born in the earliest. This likely reflects the creativity of the English musical renaissance and formation of the British canon. Being Scottish, Welsh or Irish rather than English is not associated with more or less success; however, being an immigrant composer is associated with less success, perhaps because programmers perceive the Proms as having a ‘British’ remit. Being recorded as having more professional roles (for example, as a teacher or performer in addition to being a composer) is associated with less success, possibly because composers with more roles have more distractions from composition, or because those who are less talented tend to diversify in order to build a career. Being female is associated with less success, but the results are shy of

74

Cultural Sociology 9(1)

Table 5.  Finite mixture model predicting Proms success. Component 1 (lower success) Constant Born 1890–1909 Born 1910–1929 Born 1930–1949 Born 1950–1969 Scottish, Welsh or Irish Non-British Number of professions Female Degree centrality ImlogitTtL (mixing probability parameter) Log alpha (over dispersion parameter) Estimated class proportion (k) Entropy AIC BIC (sample size adjusted) Log-likelihood N

p value

Component2 (higher success)

p-value

0.935 −0.352 −1.454*** −1.158** −0.919* −0.467 −2.251*** −0.409** −0.512 2.610*** 1.905**

0.137 0.617 0.000 0.002 0.041 0.261 0.000 0.005 0.094 0.000 0.005

3.865*** −0.940* −1.114** −1.026* −1.169*** −0.249 −0.861 −0.351* −0.051 0.632**

0.000 0.027 0.005 0.012 0.000 0.372 0.234 0.017 0.896 0.001  

0.452*

0.014

−1.309

0.355

0.870 0.S15 2004.417 2028.577 −979.208 505

0.130

           

Notes:* Significant at the 0.05 level; ** significant at the 0.01 level; *** significant at the 0.001 level. Base category is born between 1870 and 1889, English/British, and male.

significance at the conventional levels. The relationship between degree centrality and success is positive and statistically significant, even for this group in whom biographers may be thought to have less interest. Turning to the ‘star’ component, again, being a member of the three later birth cohorts compared with the first (the peers of Vaughan Williams and Holst) is associated with significantly less success in terms of Proms works. For this component, being an immigrant composer is not associated with less success. Perhaps immigrant stars are treated differently, or there may be a small-n problem. Having more professional roles is associated with significantly less success. Being Scottish, Welsh or Irish rather than English is associated with less success, but again the results are shy of significance. As before, women appear not to be particularly penalized: their lower representation in the Proms is most likely due to their relatively lower representation in the cohorts which are most productive by this measure. This finding of no gender penalty is perhaps surprising in the light of popular concerns regarding the underrepresentation of female composers, and we explore it in considerably more detail in a related book chapter ([add reference] 2014). Finally, degree centrality is again strongly associated with success. We remain circumspect with respect to causality, because it is likely that networks are at least partly

McAndrew and Everett

75

endogenous – that is, that the attributes of the composers determine their connections. Even so, it remains plausible that networks aid success rather than (or in addition to) success attracting connections, or the reporting of connections. The difference in the effect of centrality between the two components strengthens this interpretation: the effect of centrality for the ‘journeyman’ component is significantly larger than that for the ‘star’ component (difference in coefficients = 1.978, p = 0.002), suggesting that the result is not an artefact of biographers giving more information on the relationships of the best-known.

Discussion Our first aim was to describe the structure of networks among British composers from 1870. The main component (excluding isolates and pendants) consists of 322 composers, which spans the entire period. This suggests that classical music in Britain forms an unbroken single tradition within which distinct movements are situated. We should note that a number of important popular musicians – including Paul McCartney, Elton John, and Noel Coward – exist outside this component and have achieved enormous commercial success through popular music, light music and ‘classical crossover’ output. By examining the cohesive subgroups structure, we find a single connected four-core which contains acknowledged leaders of British classical music. Cohesive parts of the network at a higher level of aggregation can be revealed by using the Girvan-Newman method, which uncovers pedagogic networks, and also provides more specific information on the subgroups to which specific marginal figures are attached. We suggest that the development of musical movements may be associated with relatively cohesive parts of the network at a finer level of granularity than the community level, and further, that they are not associated with the four-core. Finally, we model the factors associated with Proms success. We can distinguish a smaller group of ‘star’ composers from ‘journeymen’ composers. Centrality is associated with success, and particularly so for the less stellar group. For both, those born in the latter three birth cohorts are significantly less successful than the cohort born between 1870 and 1889. We suggest that the cohort and network effects here relate to music being a form of knowledge, and that these findings are similar to studies in the technological history tradition. Using this evidence, the revisionist story of the discriminate exclusion of particular composers on the basis of their class, sex and ethnicity is not borne out. We also reaffirm Vaughan Williams’ contribution to British classical music with quantitative evidence. Access to elite networks depended both on ability and personality; while many talented marginal figures were undoubtedly simply unlucky in that they possessed all the ‘right’ attributes but somehow did not break through, others were marginal partly through personal choice and self-imposed isolation. Some composers chose more commercial paths with less need for network support; others chose to compose music which was difficult to programme or publish. We also investigated the embeddedness of the ‘English Music Renaissance’ and whether social networks were associated with lock-in to a particular musical style to the detriment of overall standards. In additional analysis (not reported here) we find that

76

Cultural Sociology 9(1)

cohort effects on connectedness itself are insignificant. This suggests that the English Music Renaissance was a product of a particular historical context rather than of network closure. It is not unique to Britain that many composers sought ideas and originality from ‘national music’ and proselytised in favour of particular forms. While it is entertaining to repeat Lutyens’ description of the pastoral school as ‘cow-pat music … folky-wolky modal melodies on the cor anglais’ (Taruskin, n.d.; Stradling and Hughes, 2001: 100), the music of the Edwardian and interwar composers was of a sophistication that provided a clear break with the mid-Victorian period. While it could be argued that the ‘oligarchy’ cited by Stradling and Hughes was not situated among the composers themselves, but in the wider music world (BBC producers, concert promoters and agents, the orchestras, critics, publishers and record companies), our suspicion is that this array of additional actors in itself suggests plurality rather than hegemony. In sum, the network structure suggests relative openness. We might also draw some conclusions from our analysis for the value of music departments in higher education institutions. Composers active before the Second World War were often freelance, depending on a variety of income sources and living precariously. Those active after the Second World War benefited heavily from the expansion of higher education associated with the Robbins Report of 1963, both as students and as faculty. Higher education expansion led to the establishment of vibrant music departments and centres of contemporary classical music outside the metropolis, diversifying and enriching cultural life throughout the UK through creating viable classical music ‘ecologies’ outside London. Many composers were able to delay their reliance on music-related employment until completion of doctoral research, in the course of which further connections were established and personal styles developed. A number of composers now earn their primary living as academics. These infrastructural investments are likely to have generated numerous spillovers in addition to the provision of music education and conduct of musicological research: music students learn tacitly how to become working performers and composers through the informal networks they build at university as well as via their formal education, while composers have had some recourse from the market, allowing time to compose works which would otherwise not have existed. While some popular commentators argue that new music by composers based in the academy can be too academic, elitist and concerned with formal innovation, the economist Tyler Cowen argues that ‘only the modern world can support so much fascinating production for the tastes of a small minority’ (Cowen, 1998: 155).

Conclusion This analysis provides a first examination of this dataset. Our next step will be to gather more detailed information on the characteristics of each composer: whether they attended a conservatoire, their socio-economic background, and the genres in which they worked (symphonic music, opera, church music, light music, film, other). We will also create additional measures of success from data on recordings, and a measure of historians’ interest. Further extensions could involve adding composers born from 1840 – to capture the influence of Stanford, Parry, Delius and Elgar – or a comparison with composer networks in other countries.

McAndrew and Everett

77

Potential criticisms of the analysis presented here fall into two broad domains: methodological and disciplinary. Regarding methodology, it might be argued that the network structure we have described is an artefact of collecting data from a directory of biographies: the most talented students have the most prominent teachers. Further, biographers subconsciously rationalize essentially random success by identifying likely friends and teachers. Cohesiveness between composers may well arise less for learning and collaboration and more because of status and value homophily. A final issue is that biographers cannot report all relevant social ties; they may also fail to report antagonism provoked by their subject. Our choice of object of interest also raises a number of questions regarding the delimiting of the object and disciplinary approaches alike. In examining networks of composers, we draw inspiration from the concept and metaphor of ‘music worlds’ as outlined in the introduction to this volume, which in turn acknowledges acknowledges the influence of Becker’s Art Worlds (1982). To reiterate, music is created via collective activity and new music via collective invention situated in interaction networks. Our particular interest was in the creation of original works, and its parallels with scientific research and technological innovation; in addition, we limited our frame to composers rather than members of the classical world more broadly for tractability of data-gathering and to benefit from a long-run perspective. In this regard our approach fits securely in the new economic sociology tradition, with its interest in relationality, SNA, the interlinking nature of economic and social incentives for behaviour, and methodological parsimony. Nevertheless, we should recognise that a fully sociological account – indeed, a cultural sociological account – would surely take as its object of interest the wider classical music world in richer, more localized, and temporally-focused context, involving the wider world of conductors, concert hall and opera house intendants, promoters, publishers, agents, producers, patrons, critics and others. Such an approach, whether involving a large-scale assembly of historical evidence for a fixed point or successive points in time (for example, following Crossley, 2014), or a present-day ethnography (O’Shea, 2014), might, in combination with SNA methods, reveal more clearly wider-range sources of structural inequalities and the means whereby cultural hierarchies are formed. However, while necessarily imperfect and limited, our results nevertheless do provide significant evidence to debates in music history via our testing of a ‘middle range’ theory (Merton, 1949) of collective musical invention. Extensions of this work along the lines suggested, whether by ourselves or others, should generate further findings and allow refinement of theory. We accordingly consider SNA highly valuable for describing the case of British music in detail, in particular the relative positions of individual composers and different groups, and for allowing us to examine the association between centrality and success. We hope that historians, musicologists and sociologists of music will appropriate these methods in combination with their detailed knowledge of forms and contexts to provide more definitive answers. Funding This research was funded by ‘Music Communities’, AH/J006807/1, an Arts and Humanities Research Council project funded under the Research Councils UK cross-council ‘Connected Communities’ programme. The authors would like to thank the AHRC for their support. Datasets for this article available on request from the corresponding author.

78

Cultural Sociology 9(1)

References Allen RC (1983) Collective invention. Journal of Economic Behavior and Organization 4(1): 1–24. Anderson V (n.d.) Hobbs, Christopher. Grove Music Online. Available at: http://www.oxfordmusiconline.com/subscriber/article/grove/music/2082352 (accessed 29 August 2012). Baumol WJ, Schilling MA and Wolff E (2009) The superstar inventors and entrepreneurs: How were they educated? Journal of Economics and Management Strategy 18(3): 711–728. Becker H (1982) Art Worlds. Berkeley: University of California Press. Benkler Y (2006) The Wealth of Networks: How Social Production Transforms Markets and Freedom. New Haven: Yale University Press. Bonacich P (1972) Factoring and weighting approaches to status scores and clique identification. Journal of Mathematical Sociology 2(1): 113–120. Borgatti SP, Everett MG and Freeman LC (2002) UCINET for Windows: Software for Social Network Analysis. Cambridge, MA: Analytic Technologies. Borgatti SP, Mehra A, Brass DJ and Labianca G (2009) Network analysis in the social sciences. Science (13 February) 323(5916): 892–895. Borowiecki KJ (2013a) Geographic clustering and productivity: An instrumental variable approach for classical composers. Journal of Urban Economics 73: 94–110. Borowiecki KJ (2013b) Agglomeration economies in classical music. Discussion Papers of Business and Economics 2013, Department of Business and Economics, University of Southern Denmark. Bottero W and Crossley N (2011) Worlds, fields and networks: Becker, Bourdieu and the structures of social relations. Cultural Sociology 5(1): 99–119. Brett P, Doctor J, LeGrove J and Banks P (n.d.) Britten, Benjamin. Grove Music Online. Available at: http://www.oxfordmusiconline.com/subscriber/article/grove/music/46435 (accessed 27 August 2011). British Broadcasting Corporation (2011) The Proms archive. Available at: http://www.bbc.co.uk/ proms/archive (accessed 1 May 2011) Burt R (2004) Structural holes and good ideas. American Journal of Sociology 110(2): 349–399. Butts C and Carley K (2000) Spatial models of large-scale interpersonal networks. Working Paper, Center for Computational Analysis of Social and Organizational Systems. Available at: http:// www.casos.cs.cmu.edu/publications/index.php. Cameron AC and Trivedi P (2009) Microeconometrics Using Stata. College Station, TX: Stata Press. Caves R (2000) Creative Industries: Contracts between Art and Commerce. Cambridge, MA: Harvard University Press. Christiansen R (1995) Obituary: Alan Bush. The Independent, 3 November. Available at: http:// www.independent.co.uk/news/obituaries/obituary-alan-bush-1537087.html (accessed 18 August 2011). Cliff AD and Ord JK (1975) Model building and the analysis of spatial pattern in human geography. Journal of the Royal Statistical Society Series B 37(3): 297–348. Comunian R (2012) Exploring the role of networks in the creative economy of North East England: economic and cultural dynamics. In: Warf B (ed.) Encounters and Engagement between Economic Cultural Geography. London: Springer, pp. 143–157. Cowen T (1998) In Praise of Commercial Culture. Cambridge, MA: Harvard University Press. Cross J (n.d.) Manchester School. Grove Music Online. Available at: http://www.oxfordmusiconline.com/subscriber/article/grove/music/49722 (accessed 28 August 2011). Crossley N (2008) Pretty connected: The social network of the early UK punk movement. Theory, Culture & Society 25(6): 89–116.

McAndrew and Everett

79

Crossley N (2014) Networks of Sound, Style and Subversion: The Punk and Post-Punk Music Worlds of Manchester, London, Liverpool and Sheffield, 1975–1980. Manchester: Manchester University Press. DeNora T (2003) After Adorno: Rethinking Music Sociology. Cambridge: Cambridge University Press. Doctor J (1998) ‘Working for her own salvation’: Vaughan Williams as teacher of Elizabeth Maconchy, Grace Williams and Ina Boyle. In: Foreman L (ed.) Vaughan Williams in Perspective. London: Albion Press, pp. 181–201. Ericsson KA, Prietula MJ and Cokely ET (2007) The making of an expert. Harvard Business Review 85(7/8): 115–121. Faulkner RR (1983) Music on Demand: Composers and Careers in the Hollywood Film Industry. New Brunswick, NJ: Transaction Books. Finnegan R (1989) The Hidden Musicians: Music-Making in an English Town. Cambridge: Cambridge University Press. Florida R (2002) The Rise of the Creative Class: And How It’s Transforming Work, Leisure, Community and Everyday Life. New York: Perseus Book Group Freeman LC (1977) A set of measures of centrality based on betweenness. Sociometry 40(1): 35–41. Freeman LC (1979) Centrality in social networks: Conceptual clarification. Social Networks 1(3): 215–239. Frogley A (2003) Rewriting the Renaissance: History, imperialism, and British music since 1840. Music and Letters 84(2): 241–257. Girvan M and Newman MEJ (2002) Community structure in social and biological networks. Proceedings of the National Academy of Sciences of the United States of America. 99(12): 7821–7826. Gladwell M (2008) Outliers: The Story of Success. London: Allen Lane. Knox H, Savage M and Harvey P (2006) Social networks and spatial relations: Network as method, metaphor and form. Economy and Society 35(1): 113–140. Lamoreaux NR and Sokoloff KL (2005) The decline of the independent inventor: A Schumpterian story? National Bureau of Economic Research Working Paper No. 11654. Lamb A (n.d.[a]) Phillips, Montague F. Grove Music Online. Available at: http://www.oxfordmusiconline.com/subscriber/article/grove/music/21579 (accessed 29 August 2011). Lamb A (n.d.[b]) Löhr, Hermann. Grove Music Online. Available at: http://www.oxfordmusiconline.com/subscriber/article/grove/music/16886 (accessed 29 August 2011). Lewis G (n.d.) Hoddinott, Alun. Grove Music Online. Available at: http://www.oxfordmusiconline.com/subscriber/article/grove/music/13118 (accessed 28 August 2011). Lizardo O (2006) How cultural tastes shape personal networks. American Sociological Review 71(5): 778–807. Lloyd S (2001) William Walton: Muse of Fire. Woodbridge: Boydell. McAndrew S and Everett M (2014) Symbolic versus commercial success among British female composers. In: Crossley N, McAndrew S and Widdop P (eds) Social Networks and Music Worlds. Abingdon: Routledge. Marshall L (2011) The sociology of popular music, interdisciplinarity and aesthetic autonomy. British Journal of Sociology 62(1): 154–174. Mason C, Cole H and Watson D (n.d.) Bush, Alan. Grove Music Online. Available at: http://www. oxfordmusiconline.com/subscriber/article/grove/music/04432 (accessed 29 August 2011). Merton R (1949) Social Theory and Social Structure: Toward the Codification of Theory and Research. Glencoe, IL: Free Press. Mitchell D, Reed P and Cooke M (eds) (2004) Letters from a Life: The Selected Letters and Diaries of Benjamin Britten, 1913–1976, Vol. 3. London: Faber and Faber.

80

Cultural Sociology 9(1)

Onderdonk J (1995) Review: The English Musical Renaissance, 1860–1940: Construction and Deconstruction. Music Library Association Notes (2nd Series) 52(1): 63–66. O’Shea S (2014) The art worlds of punk-inspired feminist networks: A social network analysis of the Ladyfest Feminist Music and Cultural Movement in the UK. PhD Thesis, University of Manchester, UK. Ottaway H and Frogley A (n.d.) Vaughan Williams, Ralph. Grove Music Online. Available at: http://www.oxfordmusiconline.com/subscriber/article/grove/music/42507 (accessed 27 August 2011). Pisaro KG (n.d.) Scratch orchestra. Grove Music Online. Available at: http://www.oxfordmusiconline.com/subscriber/article/grove/music/47477 (accessed 29 August 2011). Prevost E (2001) The arrival of a new musical aesthetic: Extracts from a half-buried diary. Leonardo Music Journal 11: 25–28. Rapoport P (n.d.) Sorabji, Kaikhosru Shapurji. Grove Music Online. Available at: http://www. oxfordmusiconline.com/subscriber/article/grove/music/26247 (accessed 27 August 2011). Root D (ed.) (2011) Oxford Music Online. Available at: http://www.oxfordmusiconline.com (accessed 1 May 2011). Rosen S (1981) The economics of superstars. American Economic Review 71(5): 845–58. Rossman G, Esparza N and Bonacich P (2010) I’d like to thank the Academy, team spillovers and network centrality. American Sociological Review 75(1): 31–51. Saylor E (2008) ‘It’s not lambkins frisking at all’: English pastoral music and the Great War. Musical Quarterly 91(1–2): 39–59. Seidman S (1983) Network structure and minimum degree. Social Networks 5(3): 269–287. Self G (n.d.) Coates, Eric. Grove Music Online. Available at: http://www.oxfordmusiconline.com/ subscriber/article/grove/music/06001 (accessed 30 August 2012). Smith M (1993) Review: The English Musical Renaissance, 1860–1940: Construction and Deconstruction by Robert Stradling; Meirion Hughes. Tempo (187) (n.s.): 27–8. Stradling R and Hughes M (2001 [1993]) The English Musical Renaissance, 1840–1940: Constructing a National Music. Manchester: Manchester University Press. Taruskin R (n.d.) Nationalism. Grove Music Online. Available at: http://www.oxfordmusiconline. com/subscriber/article/grove/music/50846 (accessed 31 August 2011). Van Dijk A (2009) Essays in Finite Mixture Models. Amsterdam: Rozenberg Publishers. Williamson OE (1996) The Mechanisms of Governance. New York: Oxford University Press.

Author biographies Siobhan McAndrew is Marston Research Fellow at the Institute for Social Change, University of Manchester. Her research areas include sociology of music, sociology of religion and cohort analysis. Publications include Social Networks and Music Worlds (edited with Nick Crossley and Paul Widdop; Routledge, 2014) and ‘Modernization and the Gender Gap in Religiosity: Evidence from Cross-National European Surveys’, Koelner Zeitschrift fuer Soziologie und Sozialpsychologie, 65(1): 259–283 (with David Voas and Ingrid Storm; 2013). Martin Everett is a Professor at the Mitchell Centre for Social Network Analysis, University of Manchester. Research areas include methods for social network analysis, analysis of centrality, positional analysis and cohesive subgroups, and negative tie networks. Publications include Analyzing Social Networks (with S.P. Borgatti and J.C. Johnson; SAGE, 2013), ‘Extending Centrality’ (with S.P. Borgatti), in P.J. Carington, J. Scott and S. Wasserman (eds) Models and Methods in Social Network Analysis (Cambridge University Press, 2005) and UCINET for Windows: Software for Social Network Analysis (with S.P. Borgatti and L.C. Freeman; Analytic Technologies, 2002).