G Models
POETIC-1195; No. of Pages 25
Poetics xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Poetics
j ourn al h omepa ge : www.elsevier.com/locate/p oetic
Categories and networks in jazz evolution:
The overlap between bandleaders’ jazz sidemen
from 1930 to 1969
Charles Kirschbaum a,b,*
a
b
Insper Institute of Education and Research, Sao Paulo, Brazil
CEM-Cebrap, Brazil
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 24 September 2011
Received in revised form 23 March 2015
Accepted 23 April 2015
This paper expands on Peterson’s process model for historical
changes in jazz. Peterson suggests that, given certain circumstances, musical genres migrate from ‘low-brow’ to ‘high-brow’. I
test this proposition for jazz by investigating whether bandleaders
were associated through the same sidemen (‘sidemen overlap’)
across time, and the underlying logics leading to these overlaps. I
confirm Peterson’s model to the extent that sidemen overlap shifts
from a ‘commercial’ logic to a ‘style-based’ logic. From 1930 to 1949,
sidemen overlap between bandleaders is mainly predicted by
recording session volumes (akin to ‘commercial logic’). From
1945 to 1969, style similarity emerges as an important predictor
of sidemen overlap. I extend Peterson’s process model by providing
a more nuanced account, based on social networks. I show
substantive collaboration across styles. As a consequence, stylistic
shifts are not as abrupt as originally depicted. I also explore how
past associations become increasingly stronger in terms of
explaining sidemen overlap. Furthermore, race emerges as an
important variable in explaining the same phenomenon. During the
thirties, non-African-American homophily is high. After this period,
African-American homophily increases steadily until the late fifties,
decreasing again during the sixties.
ß 2015 Elsevier B.V. All rights reserved.
Keywords:
Jazz
Distinction
Categories
Tie formation
Peterson
* Correspondence to: Insper Instituto de Ensino e Pesquisa, Rua Quatá, 300, Vila Olı́mpia, 04546042 São Paulo, SP, Brazil.
Tel.: +55 11 4504 2779; fax: +55 11 4504 2744.
E-mail address: [email protected].
http://dx.doi.org/10.1016/j.poetic.2015.04.001
0304-422X/ß 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
2
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
1. Introduction
Richard Peterson’s (1972) process model of style evolution successfully explained why certain
genres follow the trajectory from folk through pop to avant-garde phases in response to both
‘exogenous’ forces (including technological disruptive events) and ‘endogenous’ forces (such as
the emergence of a stronger professional identity). In Peterson’s account, shifts in styles are
comparable to disruptive events – analogous to Kuhn’s idea of scientific revolution – in which
younger musicians abruptly switch from one style to another, simultaneously promoting the
development of distinct social formations. Peterson explores jazz history to illustrate this process.
In this paper, therefore, I retrace the history of jazz in order to review Peterson’s model and
provide a more nuanced narrative. I shall claim that the shift from one style to another was not
abrupt. Much the opposite: there was significant collaboration among musicians playing a mixture
of styles.
A core assumption informing this paper is that musicians, when playing and recording
together, are motivated by choices and constrained by social forces. Adopting Peterson’s process
model, we would expect to observe that associations between musicians are closely linked to
style identification. Consequently, using a social network analysis to explore what impels
associations between musicians should afford us insights into how key social trends influence
musical production and how stylistic shifts take place. Over the course of this paper, I shall
focus specifically on understanding how bandleaders were connected to each other, revealing
and exploring the extent to which their recording sessions included the same (overlapping)
sidemen.
My analysis concentrates on a time period spanning from 1930 to 1969. In the early thirties,
following the Great Depression, the music industry resumed its growth and swing led jazz to enjoy
huge popularity. By the late sixties fusion and free jazz had emerged as important tendencies,
challenging jazz’s boundaries.
Throughout this paper, I test and discuss the following hypotheses in dialog with Peterson’s
model. First, similarity in style is positively related to sidemen overlap between bandleaders. This
hypothesis supports the idea that styles and social ties are closely linked, and might occur in higher
intensity once the field becomes more autonomous, but also more fragmented. This hypothesis was
supported for the period from 1945 to 1949, as well as later. Second, the number of sessions recorded
is a strong predictor of sidemen overlap in the early 1930s. This insight reinforces Peterson’s idea that
during this period jazz was strongly dominated by commercial interests. This pattern becomes
weaker during later periods. Third, time in the field was partially supported, suggesting points of
inflection. These variations in turn suggest inter-generational connections. Peterson’s model
receives a more nuanced account insofar as new styles do not displace older ones. Furthermore, while
associations between musicians are increasingly explained by a similarity in style, other
mechanisms such as popularity, race and past association remain important as ‘social glue’ and
sustain the jazz’s community cohesion.
Among the control variables, I have also explored how race, gender, nationality, instrument
similarity, and past associations affected the overlap between bandleaders’ sidemen. It is worth
noting that past association emerges as a strong predictor of sidemen overlap as we approach
the late sixties. Additionally, as emphasized by cultural studies scholars, race boundaries prove to
be an important factor in determining associations, though their effect decreases during the
sixties.
2. Theoretical framework
2.1. Bourdieu’s model of style emergence
Although Peterson’s original proposition does not focus on Bourdieu’s insights, I believe that the
latter’s model of style emergence shows close proximities to Peterson’s ideas on why musicians
abandon pop styles and espouse avant-garde alternatives. In the following text, I shall draw on
Bourdieu’s ideas of struggle and distinction within artistic fields. Bourdieu’s model associates style
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
3
adoption with generational cohort.1 Musicians from older cohorts have interests invested in older
styles and are thus constrained from shifting to new styles. A central idea for Bourdieu is the concept
of ‘consecration,’ which entails both institutionalization (e.g. prizes and academic places) and peer
recognition of individuals and categories (i.e. styles). Established artists may exploit the advantages
of having consecrated public identities. Despite the introduction of new styles, established musicians
rely on older audiences, whose durable and embodied tastes resist change.
A contrasting argument would suggest that young artists reproduce established styles. Younger
artists might be chosen by older artists as their heirs. This matches the mechanism of consecration
described by Bourdieu. When masters choose their students, they consecrate them as their natural
heirs. Consecration takes place bottom-up too. By recognizing their teachers as masters, students
support the consecration of the latter (Bourdieu, 1993: 57). As a result, we would expect a
reproduction of styles (materialized in practices, and embodied in the artists’ habitus) as new artists
look to affiliate to old masters in order to gain a foothold in the field.
Nonetheless, several mechanisms, if triggered, prevent this reproduction from happening in artistic
fields. Firstly, younger artists are not as able as older musicians to profit from established styles.
Certainly while older audiences still demand established styles, an oversupply of established
musicians is able to meet that demand, leaving little room for younger artists.
Instead of focusing on economic profits, some younger artists may aim to make ‘symbolic profits.’2
This is what Bourdieu called the ‘‘inverted logic of artistic fields.’’ Within an artistic field, the migration
toward the avant-garde promotes higher ‘field autonomy’: production is dictated less by commercial
(heteronomous) interests, while value becomes defined internally rather than externally.
By developing the genre toward less accessible codes, younger artists are able to couple innovation
with the development of a restricted field of production, destined to artists themselves as their audience.
Following the logic of distinction, the emergence of a closed network of artists provides an opportunity
for closure, while establishing this network as a newly emergent elite in the art world. While these new
artists are not always able to convert their symbolic capital into an economic variety, as established
artists are able to do, their expectations differ from older generations. Because younger artists are willing
to push for avant-garde styles, while established artists may work to protect the consecrated status quo,
a conflict between generations is likely to take place. Ultimately, we may observe a disruption of social
ties, leading to the emergence of disconnected collectivities (Bourdieu, 1992; Greenfeld, 1989).
Taking Bourdieu’s argument to its extreme, and assuming social ties to be epiphenomenal to the logic
of distinction, we should observe an increased coupling of ties to categories (see Lamont & Molnar, 2002;
Pachucki, Pendergrass, & Lamont, 2007 for an extension of this argument on identity boundaries). If we
assume that being associated with ‘high-brow’ individuals confers legitimacy, while the association with
‘low-brow’ individuals reduces it, then we should find that all individuals reject categories with a
relatively low status, but will be willing to embrace relatively high status categories. However, even if we
assume that individuals are strongly geared toward shifting to high-brow categories, this is unattainable
using Bourdieu’s approach. Properly consuming a high-brow genre may imply a degree of investment
beyond the subject’s current economic or educational position (Bourdieu, Darbel, & Schnapper, 1990).
Faced with the impossibility of embracing new high-brow tastes, individuals may then develop disdain
and rejection instead (Bourdieu, 2002). Conversely, artists associated with high-brow styles may find
they have less mobility across social circles (Lizardo, 2006).
Within a field, categories belong to a categorical set, organized and enacted through shared
schemata (DiMaggio, 1997). One major source of categorizations is the critics, who try to develop
coherent narratives through their usage of such categories. As the body of jazz critics became more
professionalized, so they categorized music on the basis of increasingly rationalized3 criteria.
1
Mannheim (1970) developed the ‘‘problem of generations,’’ trying to connect individual biography, biological time and
common aspects of socialization. We can see Bourdieu’s endeavor (1993: 52) as a theoretical option within the discussion on
Mannheim’s work, since the former underscores the moment when individuals are introduced into a field, weakening the
emphasis on biological age.
2
Heinich (1997) points out how this may become a normative expectation rather than just a pattern.
3
The Weberian concept of rationalization in music contains the idea of an increasingly sophisticated, yet path-dependent,
way of music writing (Weber, 1958). See Williams (1983) for a collection of essays that attempt to establish musicological
criteria as the only legitimate framework for jazz criticism.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
4
While the bureaucratic and commercial dimensions of an artistic field attempt to impose
categories in order to organize the distribution of goods (DiMaggio, 1987; in the jazz context, see
Szwed, 2000), artists may vary by resisting, complying with or even promoting existing
categorizations. Artists may espouse ‘categories’ as ‘manifestos,’ matching positions in a field
(Bourdieu, 1993). For instance, Dizzy Gillespie promoted ‘Bop,’4 Gunther Schuller promoted ‘Third
Stream,’5 Stan Getz promoted ‘Bossa Nova,’ and musicians like Ornette Coleman called one of his most
important albums, recorded in 1960, ‘Free Jazz.’6 By contrast, the category ‘West Coast’ is believed to
have been an invention of recording companies.
2.2. Networks and the role of interactions
While in the previous section I turned to Bourdieu’s model as a way of understanding the mechanisms
involved in stylistic adoption, in this section I revisit some of his other insights to explore how networks
relate to his field theory. Networks occupy an ambivalent place in his writings. Bourdieu (1986) suggests
that strong relationships are central to social attainment. His framework goes beyond the dyadic level to
include larger structures – such as when he refers to ‘social circles’ as close-knit groups with strongly
demarcated boundaries. Group membership may convey economic and symbolic gains. On the other
hand, Bourdieu deemphasizes the importance of interactions (Bourdieu & Wacquant, 1992), criticizing
the contemporary tendency of social network analysis to reify relationships as ‘ties’ (see Emirbayer &
Goodwin, 1994 for a lengthy discussion). As remarked above, Bourdieu’s writings frequently contain the
view that social interactions are epiphenomenal at best to the logic of distinction.7
Taking an opposite line, Lahire (2008) suggests that intra-individual variance could help clarify
some of the unexplained variance in Bourdieu’s studies. The rationale adopted by Lahire is that an
individual’s tastes are not univocally determined by his or her social position. Instead, they may
demonstrate different tastes in different settings. Individuals have a plethora of tastes, not necessarily
consistent with each other (Swidler, 1986, 2001). The same idea could be transposed to the
‘production-side’ of culture. Although public scrutiny makes it harder to mix genres (Lahire, 2004),
bandleaders might prefer ‘esthetic generalists’ (Pinheiro & Dowd, 2009). On an empirical level, this
would account for social interactions that go above and beyond the categorical coupling implicitly
assumed within Bourdieu’s oeuvre. The methodological implication would be to provide social
networks with analytical autonomy.
Several scholars have deployed social network analysis in order to understand how relations and
patterns of interactions constitute fields. DiMaggio (1986), for instance, collected data on respondents’
personal relationships in order to explore the relationships between field positions rather than
focusing on the individual level. Anheier, Gerhards and Romo (1995) expanded on DiMaggio’s
approach by compiling information on several types of relationships and explicitly incorporating
cultural and economic capital measurements into their analyses.
Although many interactions in the jazz world were confined to small audiences and went largely
unnoticed, many others were public,8 open to third-party observation and assessment, and easily
accessible when they produced enduring artifacts like ‘albums’ and ‘movies.’ To the extent that
audiences observed these sessions, there was a symbolic dimension underlying the public interactions
(Podolny, 2001).
In this paper, I focus on how different bandleaders might share the same (overlapping) sidemen.
My starting point is Bourdieu’s apprehension of fields as relational spaces (Bourdieu, 1985a; Martin,
2003). Put succinctly: a person’s structure of opportunity depends on her position in a field, and this
4
Fittingly Dizzy Gillespie wrote a book entitled To be, or not. . . to Bop (Gillespie, 2009).
Joining together the ‘First Stream’ (classical music) and the ‘Second Stream’ (jazz music).
6
Musicians may have felt ambivalent about the label ‘free jazz.’ It was taken either as a broad residual category used by critics
to lump together musicians who did not fit into the current canon, or as a label for stigmatizing ‘anti-jazz’ experiments
(Anderson, 2007).
7
Bourdieu also takes the American sociological interest in social interactions to be a byproduct of American ideology insofar
as the national culture assumes that individuals engaged in face-to-face interactions can transcend their original social
positions (Bourdieu, 1985b).
8
By contrast, informal jam sessions typically have small, restricted audiences (analogous to the idea of a ‘local scene’: Lena, 2012).
5
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
5
position is relationally defined, based on how actors are located vis-à-vis one another. More
specifically, how Bourdieu empirically operationalized ‘positions’ varied across his writings. For the
purposes of this paper, we can observe how Bourdieu used the idea of ‘audience overlap’ to explore
how different types of Parisian theaters related to each other (Bourdieu, 1980), assuming that theaters
promoting the same genres (high-brow or low-brow) were more likely to attract the same patrons. I
believe this insight can be expanded in the following ways. First, while Bourdieu explored how
theaters of the same kind presented overlapping audiences, the partial audience overlap between
theaters of different types is left unstudied. Second, while the low/high-brow distinction typically
acquires a privileged status in the Bourdieusian analysis, other factors may be at work that could
influence audience overlap (Abbott, 2005; Lopes, 2002).
Adopting an analogous approach, I propose that sidemen overlap among bandleaders may reflect
field relations among the latter, whether these are intentional or otherwise. They become intentional
whenever bandleaders consciously assemble sidemen who are expected to please common audiences
who, in turn, are expected to enjoy the same sidemen. This overlap may be driven by sidemen looking
to play in prestigious bands (De Nooy, 2002) or record labels allocating sidemen to bandleaders. At a
less conscious level, the allocation of sideman to bandleaders may be the outcome of fortuitous
structural dynamics (White, 1970).
2.3. Jazz evolution: decentralization, heterogeneity and multivocality
Throughout this section, I shall explore the emergence of new styles in jazz history and the
relationship among them during the selected time frame. I depict a process model with the following
components: (1) a descriptive account of the emergence of styles; (2) an analysis of the underlying
material and institutional influences that constrained and empowered the reproduction and emergence
of styles; (3) an interpretation of the social action of musicians that underlies the reproduction and
introduction of styles. My point of departure for depicting these processes within jazz is Peterson’s
(1972) model. After presenting his model, I shall turn to the existing jazz scholarship to expand on it.
Peterson suggests that the emergence of new styles in jazz followed a folk-pop-avant-garde path.
Along this path, Peterson places New Orleans under ‘folk,’ swing under ‘pop’ and bop, hard bop and
free jazz under increasingly sophisticated forms of ‘avant-garde.’ Peterson also acknowledges that
some innovations may diverge from this path. The ‘New Orleans revival,’ ‘cool’ and ‘fusion,’ for
example, might be anomalous attempts to resist the teleological movement.
Under Peterson’s model, the migration from folk to avant-garde can be explained by the following
conditions: material shifts in both technology and demand, combined with a shift in the logic of the
musicians. This account has two historical phases: the emergence of swing as the most popular genre
in the United States and, subsequently, the emergence of new styles within jazz after swing’s decline.
During the thirties, and especially after the United States started to present signs of recovery after
the Great Depression, there was a sharp rise in the demand for music. Swing became the most popular
style,9 combining both ‘sweet’ and ‘hot’ elements that afforded entertainment for young couples and
teenagers. The mass production of this music – observable in how standardized the big bands became
– was driven by both industrial and demand-side factors. The music industry favored standardization
because of the higher predictability it entailed (DiMaggio, 1977). Furthermore, production and
distribution costs forced recording companies to focus on high-selling musicians. The use of shellac
made the production and particularly the distribution of records an expensive business. The fragility of
the material meant special transportation was required. Consequently only major recording
companies were able to afford the associated costs. In addition, the US army used shellac heavily
during the Second World War, turning it into a scarcer and more expensive resource. To make the
industry as stable as possible, the major recording companies in the 1930s tried to shape consumer
tastes through radio and mass advertising, as well as by securing long-term contracts with musicians.
On the demand side, dancing required predictability, otherwise the average teenager dancer would be
unable to enjoy it (Wald, 2009). The emergence of swing as a popular style led to the concentration of
musicians in the major cities and their professionalization.
9
See Appendix 1.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
6
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
Within the music industry in general, Peterson and Berger (1975) show that from the thirties to the
sixties the industry underwent a reorganization, leading to its eventual decentralization. Firstly it
became cheaper to produce albums due to the advent of vinyl. Secondly radio regulation led to the
emergence of small local radio stations, enabling the diversification of tastes. Thirdly the combination
of musician strikes, cheaper technology and local radio stations meant small record companies could
form to meet smaller niches and hire musicians unwilling to accept the contracts offered by the large
recording companies. On the demand side, new generations of teenagers began consuming rock as the
most popular youth genre (Peterson, 1990).
With the fall of swing as the preferred musical genre among youths, musicians were led either to
shift genres (to rock or R&B), retire, explore established but shrinking niches, or attempt to adapt to the
new jazz styles. As remarked earlier, while older audiences still demanded well-known styles, an
oversupply of established musicians could still satisfy this market, leaving little room to younger
artists. In the history of jazz, this mechanism is illustrated when the big band business started to
crumble in the early forties, and several bandleaders saw their investments suddenly reduced to sunk
cost (Peterson, 1972; DeVeaux, 1997). Some well-established musicians were unable to acquire the
skills needed to play high-brow styles (e.g. ‘bop’ and later ‘free jazz’) and unable to learn the new
repertoire (Faulkner & Becker, 2009).
Bop was incubated during the forties when musicians would also play to audiences more
interested in listening to the music than dancing. At these venues – like small clubs, connoisseur
circles and jam sessions – musicians were able to offer less standardized music and introduce
innovations. Bop’s transition from its incubation period to wider legitimization in the jazz world
involved several enabling conditions. First, the strong industry clout that had kept swing dominant
became severely weakened. Critics who had once participated in connoisseur circles became widely
employed by jazz journals and magazines, and could support this shift (e.g. Down Beat). Second, some
musicians were willing to play less popular music for limited audiences, despite the meager earnings
compared to more popular genres – akin to Bourdieu’s idea of ‘inverted logic’ cited above.
We can extend and revise this model in several ways. First, it is difficult to establish any sharp
distinction between high-brow and low-brow styles. Moreover, it is essential to include racial
boundaries within any discussion of high/low-brow distinctions since this original model may have
overlooked them (Lopes, 2002; Phillips & Owens, 2004; Phillips & Kim, 2009). Second, the material
conditions were less severe than depicted in a model in which musicians are forced to accept only
symbolic goods over material goods. Furthermore, the institutionalization was not strong enough to
prevent boundary crossing. And third, it would be too simplistic to describe the kinds of logic
employed by musicians as hard-wired toward exclusion.
Several styles understood as ‘low-brow’ or ‘middle-brow’ introduced important musical
innovations that cleared the way for more sophisticated jazz musicianship. For instance, while
‘cool’ is regarded as a movement toward commercial music, we can also locate Miles Davis’ modal
experiments in this category. For a broader audience, this music could be softer and easier to consume
than bop. But for musicians, this music signaled important innovations in musicology.10 Hence we can
observe how the same cultural artifact achieves a multivocal status, open to interpretation in different
ways by distinct audiences.
Akin to the idea of multidimensional identity formation, we can identify various misalignments
between high-brow and low-brow elements (i.e. Pachucki et al., 2007). In several cases, race was used
as a boundary identifier, in conjunction with the low/high-brow distinction. During some periods of
the investigated time frame, interracial association between jazz musicians was strongly barred by the
Jim Crow legislation (Lopes, 2000). When this cleavage was not enacted by law, it would often be
reinforced by disaffected audiences (DeVeaux, 1997). Several white swing musicians would reinforce
this boundary, associating the original hot elements with black/low-brow, while the sweet (or
classical-trained) elements would be coupled to white/high-brow musicians. In the swing era
especially, this distinction obscured racial crossovers. On one hand, black musicians like Duke
Ellington would bring a synthesis of folk and classical music to jazz (Peress, 2008). On the other, white
10
As another example: while fusion is often conceived to be an attempt to tap a broader audience by mixing rock elements
with jazz, musicological analysis shows that much of this musical production introduced innovations from free jazz.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
7
musicians would often borrow from the African American vernacular and introduce these elements to
a broader white audience.
Over the course of jazz history we can observe a gradual inversion of the black/white, low/highbrow coupling. While during the twenties, the jazz era, association with the hot/African/low-brow
identity bundle was seen to reduce its legitimacy in the eyes of the record companies (Phillips &
Owens, 2004), during the thirties and forties, the feeling grew that an authentically American music
should not be subordinated to European high-brow music. As part of this re-evaluation, AfricanAmerican elements were extolled as a source of uniquely American music-making. During the
forties,11 the coupling of African-American to low-brow became weakened or inverted entirely. Lopes
(2002) suggests that during certain periods jazz intersected with broader political and cultural
movements. At these times, African-American bandleaders often overturned white segregation.
Eventually jazz musicians who were not associated sufficiently with hot/African elements fell into
near oblivion (Wald, 2009). Bop musicians would strive for recognition and resist the kind of ‘Uncle
Tom’ behavior often associated with New Orleans musicians. Several pundits would label bop
commercial and individualist, while promoting New Orleans jazz as authentic and collectivist, true to
African roots. Bop’s reaction to this stigmatization was to mobilize the avant-garde and race
discourses. While bop continued to introduce innovation into jazz, it also promoted African-American
musicians outside the ‘primitivist’ mold imposed on it.
Anderson (2007) suggests that during the 1960s African-American musicians were the most affected
by the sharp drop in jazz’s economic turndown. Most free jazz musicians were also African-American
and responded to this decline by turning to non-market sources of finance to sustain themselves
(including government patronage and pursuing academic careers). Hence even in an allegedly colorblind musical world, race may have mediated the shift toward avant-garde styles (Lopes, 2002).
For jazz musicians, material resources were not so restricted compared, for instance, to starving
avant-garde painters. Several examples during the fifties and sixties show that jazz was more
appealing to the broad audience than the model above describes.12 Given the strong support received
by jazz from the US government after World War II, for example, the genre was able to experience a
‘Golden Age’ in the mid-fifties (Anderson, 2007). Musicians like Brubeck were a success among college
students. Louis Armstrong’s ‘Hello Dolly’ competed with The Beatles, and bossa nova won Stan Getz
the Grammy Award. Following the same pattern, while critics in the classical music world were
usually highly qualified in musicological terms and linked to academic positions, in jazz this
institutionalization was never completed. Jazz magazines often combined musicologically-trained
professionals and less technical journalists. While technical criteria were promoted, subjective
appraisals were never abandoned. Similarly, while many critics would advocate that jazz should be
considered high art, regardless of how the audiences received it, many critics still argued that jazz had
to ultimately please its audience. The concurrent production of sophisticated and more commercial
pieces, as well as the partial institutionalization of criticism, led jazz along a ‘dual path’ evolutionary
process where high-brow and low-brow – and even pieces that combined both – were produced in the
attempt to reach a variety of audiences. An analogous process took place in the movie industry where
there was partial migration to high-brow forms (Baumann, 2007).
Finally, while we could infer from the model described above that new generations of jazz
musicians increasingly looked to produce sophisticated music for a limited audience, this notion can
be challenged on the grounds that, first of all, musicians with all kinds of style orientations would play
side-by-side, exchanging experiences, and, second, multivocality can be observed throughout jazz’s
entire history. For instance, various swing bandleaders were trained as sidemen in hot jazz bands.
During the thirties Benny Goodman’s sidemen would play at jam sessions after playing for swing
audiences (Wald, 2009). Conversely, several boppers played as sidemen for swing big bands, or tried to
create ‘dancing bop big bands.’ Along much the same lines, John Coltrane was able to shift to playing
R&B and come back to jazz (Kahn, 2003).
11
Probably due to a number of factors, including the support of left-wing New Deal intellectuals, the integration of AfricanAmericans with non-African-Americans in the US army during the Second World War, the subsequent fight for equality, and the
Civil Rights movement.
12
Refer to Appendix 1.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
8
As well as providing some context to the major jazz styles, the review of Peterson’s original model
in this section also suggests an ambivalence concerning the coupling of style categories and networks.
Firstly, we can envisage the forces that led to higher coupling between categories and ties. The
increasing fragmentation of styles, in tandem with the shift from racially-based social boundaries to
boundaries based on musical identity, lead us to expect a higher coupling of styles and ties.
Furthermore, if new styles are increasingly formed by younger musicians interested in ‘symbolic
profits’ (rather than economic profits), we could assume that these actors pushed jazz toward
developing a narrower field of production, aimed primarily at fellow musicians as the audience.
Following the logic of distinction, the emergence of a closed network of musicians would provide an
opportunity for closure, while establishing itself as a new emerging elite in the jazz world.
However, as we have seen above, several tendencies could also frustrate this network-category
coupling. Firstly, over the course of jazz history various low-brow elements have later been resignified
as high-brow elements, suggesting that during transitions musicians not only try to play several
styles, they also play with musicians of different styles. Additionally, following the partial
institutionalization of jazz criticism, we can observe less coercive power being used to enforce
style congruence or even to define clear jazz boundaries (e.g. Gridley, Maxham, & Hoff, 1989). We can
suspect that the coupling of the kinds of logic pursued by jazz musicians with a teleological narrative
may appear confirmed when the account in question privileges a specific jazz historiography, one
sustained by jazz scholars favoring the avant-garde (DeVeaux, 1991; Wald, 2009).
2.4. Hypotheses
If we accept that style not only encompasses ‘low-brow’ and ‘high-brow’ distinctions, but also
includes broader identity boundaries (as explored in Section 2.3), then we can state that in general:
H1. Similarity in bandleader styles is accompanied by a higher probability of sidemen overlap.
As seen in Section 2.1, Bourdieu’s framework also provides insights into how other variables might
explain the establishment of ties (i.e. ‘resources’ and ‘time in the field’). I chose to develop hypotheses
H2 and H3 in order to cover some of the most prominent variables in the Bourdieusian tradition.
Although individuals in the same position may not necessarily interact directly (Bourdieu &
Wacquant, 1992), in a highly connected field like jazz we can assume that bandleaders were
ultimately placed in a pecking order of status determined by their relative attractiveness to sidemen.
Moreover, leaders with a lower frequency of gigs would try to attract famous sidemen, who probably
played with important bands, into their own ranks. This idea parallels Giuffre’s premise that art
galleries compete for high-status artists (Giuffre, 1999).13 In order to capture this effect, I propose:
H2. The higher the total number of recording sessions of a given pair of bandleaders, the more likely their
sidemen will overlap.
As expounded in Section 2.1, bandleaders who entered the field during proximate time periods
were more likely to have overlapping sidemen, who would be able to adapt to similar stylistic
requirements and produce the same overlapping repertoire (Stinchcome, 1965; Swidler, 1986;
Faulkner & Becker, 2009). Therefore:
H3. Bandleaders with similar time in the field are more likely to experience higher sidemen overlap.
3. Data sources
3.1. Jazz recording sessions data
I consulted Tom Lord’s CD-ROM (Lord, 2005, version 5.0) for data on bandleaders’ recording
sessions. The Tom Lord discography was originally launched in 1992 in print form, and presented
13
In a less voluntaristic vein, we could also assume that due to the high asymmetries in gig volume, bandleaders who
concentrated higher numbers of sessions were more likely to hire musicians from other conspicuous bandleaders.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
9
similarities to previous discographies (e.g. Walter Bruyninckx’s work; see Kernfeld & Rye, 1995). Since
its release, this work has turned into an on-going project in which new musicians and recording
sessions are added with each new version. In 2002, version 3.3 was first launched on a CD-ROM
platform, containing 23,000 bandleaders and 132,000 sessions. In 2004, version 5.0 (consulted for the
present study) comprised 29,000 bandleaders, 155,000 recording sessions and a total of 900,000
musicians.
The database built for this paper includes a subset of the recording sessions stored in this CD-ROM.
The criteria applied for selecting bandleaders from this database was as follows. I ranked all available
bandleaders based on their volume of recording sessions. This sum covered the whole history of jazz,
from the early twentieth century until the turn of the millennium. From this unabridged list, I selected
the bandleaders who had at least twenty sessions recorded across their entire musical trajectory,
irrespective of when these sessions were recorded. This set of bandleaders amassed over 73,000
recording sessions, or 50.5% of all recording sessions included in the CD-ROM. Next, I selected the
leaders whose production included at least one session within the time frame from 1930 to 1969. As a
result, many leaders were not included since their sessions were all either after 1969 or before 1930.
Several bandleaders could be identified as not pertaining to jazz, leading to a problem of boundary
specification.14 I therefore triangulated this database with Scott Yanow’s style classification.15 The
artists who were misclassified (classified solely under other styles like R&B and rock ‘n’ roll) were
excluded. Some bands either had shared leadership or simply changed leaders over the course of their
existence. Consequently I produced a database matching sessions to leaders. The major sources of
information were the New Grove Dictionary of Jazz (Kernfeld, 2001) and biographies available on the
websites listed below. I ended up with 388,204 sessions16 distributed across 1112 leaders.17 Fig. 1
below depicts a simple count of sessions contained in the database. It provides a benchmark to check
against the jazz literature: we observe an upward production trend during the thirties and forties
(with a couple of interruptions due to two major strikes), a recovery in the late fifties, and finally a
decline toward the late sixties.
3.2. Style assignment methodology
Ideally it would be possible to identify styles for every recording session in the database. Such
information is unlikely to be available, however. Critics and jazz scholars assign categories (or styles)
to a subset of released records. Even the most comprehensive album database is only a subset of the
universe of records released, which is likewise a subset of all recording sessions registered in the
available discography databases. Many of these recording sessions might not even be available to
critics and jazz scholars. The album database consulted was the All Media Guide (AMG), compiled
under the direction of Scott Yanow, since it provides a comprehensive coverage of jazz albums and
respective styles18 (see Table 1 for a description of styles). In the following paragraphs, I explain how I
connected styles and recording sessions.
14
In a personal communication, Tom Lord told me that many non-jazz artists had been included due to pressure from CD-ROM
buyers. I also asked him what the criteria were for including artists in his database. My concern was that he could be excluding
important artists by setting arbitrary boundaries on what counts as jazz. In general, he tended to err on the side of inclusion. See
Kahl, Kim, and Phillips (2010) for recent research using Tom Lord’s data.
15
Scott Yanow has written for several major jazz magazines, including Record Review, JazzTimes, Jazziz, Downbeat, and Los
Angeles Jazz Scene. He has also written a number of books on jazz, including Jazz on Record (Yanow, 2003). He and a team of jazz
critics maintain a database of jazz albums on the All Media Guide website. See Data sources below.
16
Where a leader’s sessions were spread across a single day, these were grouped as a single session.
17
This figure corresponds to individual musicians, amounting to about 4% of the universe of 26,147 leaders (individuals and
bands) present on the CD-ROM.
18
Scott Yanow identified 59 singular sub-styles. Although this level of granularity was difficult for triangulation purposes, in
my database I included the album association and the lowest level sub-style available. Next I used Yanow’s nested structure in
my analysis in order to embrace a more coarse-grained approach, grouping these sub-styles in nine style families: Big Band/
Swing, Bop, Cool, Free Jazz, Fusion, Hard Bop, Latin Jazz/World Fusion, New Orleans/Classic Jazz, and Soul Jazz/Groove. This
coarser classification was easier to triangulate with alternative classification sources (e.g. Kernfeld, 2001). There was a high
match in classification with the exception of the ‘trad’ and ‘modern’ style labels, more typically used in Europe. For the former, I
used AMG’s ‘Big Band/Swing’ and ‘New Orleans/Classic Jazz’ labels, while for the latter I used the ‘Bop’ label.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
[(Fig._1)TD$IG]
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
10
3000
2500
2000
1500
1000
500
19
68
19
66
19
64
19
62
19
60
19
58
19
56
19
54
19
52
19
50
19
48
19
46
19
44
19
42
19
40
19
38
19
36
19
34
19
32
19
30
0
Fig. 1. Evolution of recording sessions.
Table 1
Style description.
Style family
Exemplars
Origins and musical characteristics
New Orleans/Classic Jazz
Original Dixieland Jazz Band’s
‘‘Dixie Jass Band One-Step’’
Swing and Big Band
Woody Herman’s
‘‘Woodchopper’s Ball’’
Bop
Charlie Parker’s ‘‘Bird and Diz’’
Cool
Dave Brubeck’s ‘‘Interchanges 54’’
Hard Bop
Fusion
Sonny Rollins’s ‘‘A Night at the
Village Vanguard’’
Hank Crawford’s ‘‘After Hours’’
Ornette Coleman’s ‘‘Tomorrow
Is the Question!’’
Miles Davis’s ‘‘Bitches Brew’’
Latin and World Fusion
Stan Getz’s ‘‘Desafinado’’
Born in New Orleans and spread toward
Chicago and New York
Collective improvisation
Hegemonic in the thirties
Combined ‘‘hot jazz’’ and orchestral
elements
Extensive use of big bands
Emergence of soloist
Use of routines (riffs)
Widespread dancing among young listeners
Born at jam sessions in the Harlem
Extensive use of soloist improvisation
Use of dissonance
Extinction of dancing
Use of small combos
Heir of Bop
Incorporates classic music elements, away
from blues roots
Heir of Bop, supposedly a reaction to Cool
Recovery of African and Blues elements
Inclusion of Soul and Funk elements in jazz
Return to collective improvisation
Disruption of Bop harmonic progressions
Incorporation of late sixties Rock elements
Borrowed some Free Jazz elements
Foreign elements incorporated, ranging
from Indian music to Brazilian Bossa Nova
Soul
Free Jazz and Avant-Gard
Source of examplars: allmusic.com.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
[(Fig._2)TD$IG]
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
11
Fig. 2. Evolution of styles as percentage of Sessions.
For illustrative purposes, I take a fictitious example of a bandleader who established her name as a
swing player, and then started to play a blend of swing and bop, finally playing only bop. Each of these
style assignments is associated with ‘style events’: specifically, released albums that have received a
style classification from critics. Correspondingly, the relevant albums might be her first debut album
in 1930, then her fifth album, blending swing and bop, released in 1945,19 and finally her tenth album
assigned only to bop, launched in 1947. Because these albums depict change in style classification, I
consider them ‘style events’ and associate the style category to the album’s respective recording
session date. As a result, I ended up with a database of ‘style events’ for each bandleader.20 Next I
assigned all sessions occurring between ‘style events’ under the earlier event’s classification. Fig. 2
illustrates how styles changed throughout the history of jazz, based on share of recorded sessions.21
3.3. Demographic data on musicians
Most of the demographic data came from the All Music database. On this site I was able to extract
the musician’s date of birth and gender. Based on the picture available for the musician, I identified
whether her or she was African-Descendent.22 The Tom Lord database provided each musician’s
nationality. Alternative sources included The New Grove Dictionary of Jazz, The Big Bands Database Plus
website, Jazz associations and musicians themselves. See Table 2 for a summary of bandleader
demographics.
19
Each event may undoubtedly be associated with multiple styles. As indicated above, the same album might be classified
under ‘Bop’ and ‘Swing’ at the same time.
20
I limited the temporal distance between one event and another to five years (the same temporal length used in the
construction of matrices: see Section 4.1). When this happened – that is, no album assessed in the AllMusic database for more
than five years – I considered that the leader had no ‘public identity’ for the time span in question and excluded him or her from
the analysis of the period. I applied this procedure in order to prevent overextending a given stylistic observation.
21
If a musician was affiliated to more than one style, his or her sessions were split equally between each style. It should be
noted, therefore, that this figure has an illustrative purpose only.
22
Whenever in doubt, I consulted the musician’s entry on Kernfeld (2001) for cues on ethnicity and ‘race.’
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
II
III
IV
V
VI
VII
VIII
1930–1934
1935–1939
1940–1944
1945–1949
1950–1954
1955–1959
1960–1964
1965–1969
Total number
of leaders
158
213
277
435
537
669
703
695
Non-African
descendents
Africandescendents
Non-African
Descendents%
AfricanDescendents%
103
144
184
255
318
396
407
410
55
69
93
180
219
273
296
285
65.2%
67.6%
66.4%
58.6%
59.2%
59.2%
57.9%
59.0%
34.8%
32.4%
33.6%
41.4%
40.8%
40.8%
42.1%
41.0%
Male
Female
Male %
Female %
140
18
88.6%
11.4%
194
19
91.1%
8.9%
250
27
90.3%
9.7%
391
44
89.9%
10.1%
479
58
89.2%
10.8%
596
73
89.1%
10.9%
635
68
90.3%
9.7%
634
61
91.2%
8.8%
U.S.
Non-U.S.
U.S. %
Non-U.S. %
134
24
84.8%
15.2%
162
51
76.1%
23.9%
205
72
74.0%
26.0%
344
91
79.1%
20.9%
413
124
76.9%
23.1%
502
167
75.0%
25.0%
507
196
72.1%
27.9%
472
223
67.9%
32.1%
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
I
G Models
Period
POETIC-1195; No. of Pages 25
12
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
Table 2
Evolution of bandleaders.
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
13
4. Analytical methods
In this paper my interest is to understand how sidemen overlap between bandleaders can be
explained by common characteristics (i.e. homophily) or by parameters capturing differences
between any pair of individuals. Formally we have:
Y ¼ b1 X 1 þ þ bn X n þ e;
where Y (the dependent variable) is a quadratic matrix (the same individuals in the rows and
columns), and X1 to Xn (independent variables) are also square matrices of size N. Each cell in the later
matrices contains a value xij which depicts the covariate value of individuals i and j of the relationship.
4.1. Dependent variable
4.1.1. Sidemen overlap
The association between bandleaders and their musicians is taken from recording sessions. If a
musician was invited to play, we must assume there was, to some degree, a relationship between them
(Feld, 1981). Nonetheless, it is sensible to take into account the size of the group when calculating tie
strength (Wimmer & Lewis, 2010). In big bands we may find quite large groups and musician hiring
may be casual: a sideman might simply be found at the local jam session venue and hired for a one
night presentation only. To measure sidemen overlap between bandleaders, therefore, I followed a
three-step analysis. First, I created a database associating every recording session to bandleaders and
sidemen. In this database, I recorded the musician participation as 1/N, where N was the total number
of musicians included in the session (‘musician-session-fraction’). Next, I created an affiliation matrix
for each period: the rows of the matrix contain all musicians in the database, while the columns
contain all bandleaders (‘bandleader-sidemen matrix affiliation’). The cells contained the sum of
musician-session-fractions within the period. Finally, this rectangular matrix was transformed into a
square matrix containing only bandleaders. In order to obtain this square matrix, I calculated the sum
of the cross-product of the bandleader columns.23
4.2. Independent variables
4.2.1. Style similarity
Based on the ‘style event’ database (see Section 3.2) for each period, I formed an affiliation matrix of
bandleaders and styles. Next, I transformed this rectangular matrix into a square matrix, containing
the identity between each pair of leaders. The matrix cells contain continuous measures of similarity
(Zegers & Ten Berge, 1985).
4.2.2. Styles
‘Styles’ matrices were also constructed (New Orleans, Swing, Bop, Hard Bop, Cool, Soul, Fusion, Free
Jazz, and Latin/World). For each style, I coded ‘1’ whenever both musicians played a particular style
and ‘0’ if either of them did not play it. For each style, therefore, I obtained a square matrix where ‘1’
denotes ‘‘both leaders play this style’’ and ‘0’ for all other cases. These matrices were used to further
explore the extent to which style similarity led to sidemen overlap. While ‘style similarity’ captures
the overall similarity, this set of matrices captures how affiliation to each single style can help us
understand sidemen overlap between bandleaders.
4.2.3. Diff. field age (ln)
In order to assess the difference in field age between two band leaders, I first computed the year
when they first recorded an album, including when this was as sidemen.24 Years in the field are
measured as the difference between the average year in the period and the year when the leader
23
24
See Appendix 2 for an illustrative example.
This data was not left-censored, since I had access to this information from Tom Lord’s CD-ROM.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
14
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
entered the field (both sets of data come from Lord’s CD-ROM). I calculated the log of this measure in
order to reduce its skewness.
4.2.4. Prod. of number of sessions
I calculated the product of sessions for each pair of leaders. Since the product could yield ambiguous
relations (for instance, if musicians A and B recorded 1 and 12 sessions, respectively, the product would
be the same as two musicians C and D who recorded 3 and 4 sessions respectively), I have also included
the difference of the number of sessions, captured in the matrix Diff. of number of sessions.
4.3. Control variables
4.3.1. Past association
As seen in Section 2.2, bandleaders are likely to play repeatedly with the same sidemen, regardless
of stylistic orientations. Other factors such as reliability, technical skills and good fit with other
bandleaders might explain a musician’s desirability (Faulkner, 1983; Jones, 1996). In order to capture
this effect, I created the matrix Past association following two steps. The first was to establish a
sideman-sideman association matrix. Based on the previous period ‘bandleader-sidemen matrix
affiliation’ (see Section 4.1), I constructed the square sideman-sideman matrix, where cells contain the
product-sum of musician-session-fraction. The second step was to match the previous period
‘sideman-sideman square matrix’ with the current period ‘bandleader association square matrix.’
Whenever the bandleaders’ dyad was also present in the previous period, I used the ‘musician-sessionfraction’ value as the past association.25
4.3.2. Gender
Male musicians predominate in the jazz field (Heckathorn & Jeffri, 2001; also see Table 2).
However, stronger female participation as bandleaders could be observed in swing, where singing was
more highly valued (Tucker, 2000). I created two matrices related to Gender. The Male matrix contains
information on whether both leaders were men. If true, the pair was marked ‘10 , otherwise, the pair
was marked ‘00 . The same logic was applied when creating the Female matrix.
4.3.3. Nationality
Because jazz since its origin has been a preeminently American genre, American bandleaders were
more likely to present shared leaders. The US matrix contains information on whether the musicians
were from the United States. If true, the pair was marked ‘1’, otherwise ‘0’. I applied the same logic to
constructing the NonUS matrix.
4.3.4. Neither African-descendent
As seen above (Section 2.3), white musicians may have preferred to play or been limited to playing
with other white musicians. If both leaders were not African-descendents, the pair was marked ‘1’,
otherwise ‘0’.
4.3.5. Both African-descendents
As seen above (Section 2.3), African-descendent musicians may have preferred to play or been
limited to playing with other African-descendent musicians. If both leaders were African-descendent,
the pair was marked ‘1’, otherwise ‘0’.
4.3.6. Instrument identity
Band formations could emerge as institutionalized combinations of instruments and skills. As a
result, bandleaders who play similar instruments are likely to bring in session musicians who play
complementary instruments. For each period, I coded all instruments played per bandleader.
I grouped the instruments into broader families of instruments (e.g. different kinds of flutes were
25
Because bandleaders are also included as sidemen, this matching process captures the changes from sideman to bandleader.
See Appendix 2 for an illustrative example.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
Avg.
1935–39
S.D.
Avg.
1940–44
S.D.
Avg.
1945–49
S.D.
Avg.
1950–54
S.D.
Avg.
1955–59
S.D.
Avg.
1960–64
S.D.
Avg.
1965–69
S.D.
Avg.
S.D.
Sidemen overlap
0.380
2.971
0.181
2.763
0.109
0.990
0.075
0.748
0.057
0.512
0.111
0.748
0.059
0.723
0.035
0.417
(D.V.)
Prod. of number
192.067 351.727 339.862 991.319 322.686 1248.293 251.142 756.529 132.257 328.376 177.613 318.784 121.480 258.840 75.598 160.355
of sessions
Diff. of number
14.579
14.882
21.474
34.536
22.250
42.827
18.887
27.896
12.688
17.162
13.033
14.490
11.566
14.253
8.872
11.520
of sessions
Past association
N.A.
N.A.
0.003
0.059
0.003
0.058
0.002
0.020
0.002
0.019
0.001
0.026
0.002
0.032
0.001
0.034
Both African0.075
0.263
0.096
0.295
0.130
0.336
0.181
0.385
0.140
0.346
0.159
0.365
0.191
0.393
0.175
0.380
descendents
Neither African0.521
0.500
0.471
0.499
0.406
0.491
0.327
0.469
0.245
0.430
0.361
0.480
0.316
0.465
0.337
0.473
descendents
Female
0.002
0.040
0.004
0.066
0.008
0.088
0.006
0.080
0.005
0.072
0.006
0.078
0.006
0.076
0.005
0.069
Male
0.912
0.283
0.867
0.340
0.827
0.378
0.843
0.364
0.859
0.348
0.847
0.360
0.852
0.355
0.865
0.342
U.S.
0.688
0.463
0.590
0.492
0.638
0.481
0.652
0.476
0.641
0.480
0.611
0.488
0.595
0.491
0.500
0.500
Non U.S.
0.028
0.164
0.052
0.223
0.040
0.195
0.036
0.187
0.039
0.194
0.047
0.212
0.052
0.222
0.085
0.279
Diff. Field Age Field
0.861
0.782
0.989
0.782
0.814
0.700
0.887
0.723
0.835
0.718
0.783
0.676
0.796
0.700
0.854
0.762
(ln)
Instrument identity
0.171
0.309
0.218
0.307
0.177
0.307
0.165
0.301
0.150
0.304
0.121
0.280
0.112
0.266
0.117
0.261
Style similarity
0.554
0.433
0.606
0.410
0.560
0.414
0.451
0.413
0.325
0.376
0.286
0.355
0.255
0.342
0.221
0.328
Both leaders play
0.257
0.437
0.174
0.379
0.141
0.348
0.116
0.320
0.105
0.306
0.079
0.270
0.068
0.251
0.049
0.216
New Orleans
Big band
0.411
0.492
0.580
0.494
0.567
0.495
0.452
0.498
0.273
0.446
0.217
0.412
0.135
0.342
0.073
0.260
Bop
0.005
0.073
0.045
0.207
0.082
0.275
0.073
0.261
0.062
0.241
0.048
0.213
Cool
0.001
0.037
0.020
0.139
0.024
0.152
0.017
0.129
0.011
0.105
Latin
–
0.009
–
0.025
0.001
0.039
0.002
0.049
0.003
0.051
Hard
0.031
0.174
0.082
0.275
0.131
0.338
0.153
0.360
Soul
0.012
0.108
0.039
0.194
0.065
0.246
0.069
0.253
Free
–
0.023
0.003
0.058
0.015
0.120
Fusion
0.008
0.087
G Models
POETIC-1195; No. of Pages 25
1930–34
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
15
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
Table 3
Variables’ descriptive analysis.
G Models
POETIC-1195; No. of Pages 25
16
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
considered simply ‘flutes’). In order to capture the similarity of played instrument, I applied the
‘identity’ algorithm proposed by Zegers and Ten Berge (1985).
Table 3 provides a descriptive summary of the variable matrices.
4.4. Models
A standard multivariate solution for explaining the number of sidemen overlaps among
bandleaders would be to apply an OLS approach. However, the OLS model assumes independent
observations. As a consequence, standard statistical approaches would produce misleading results
(Kalish, 2013). In this study, I applied the MRQAP algorithm (multiple regression quadratic assignment
procedure), developed by Dekker, Krackhardt and Snijders (2007) which does not assume
independence between observations. Its procedure is to ‘control for’ rather than model network
dependencies.26 MRQAP adopts a non-parametric procedure by calculating the correlation coefficient
between matrices, permuting the matrices of one of the matrices, storing the correlation coefficient
and permuting it again. As a result, it leaves the original network structure intact (Snijders, 2011). The
p-value is obtained by comparing the observed correlation to the distribution of correlations obtained
from the permutations (Gibbons, 2004). Because this approach does not assume independent
observations, it provides more conservative inferences. As in OLS logistic regressions, negative
parameters should be interpreted as lower probability.
The first model relates all independent variables, except the style matrices. Its results are
summarized in Table 4. The second model adds the matrices by style to the above model (presented
chronologically in accordance with the jazz historiography), while excluding the variable style
similarity. Its results are summarized in Table 5.27
5. Results
Hypothesis 1 states that bandleaders with similar styles should have higher levels of sidemen
overlap. Table 4 shows that ‘Style Identity’ presents significant and increasingly higher effects from
1945 to 1949 onwards. In a historical context, this might also indicate a field that became increasingly
diversified in its styles. Further, we might also infer that the field became more autonomous, as stylebased identity boundaries were reinforced.
It is worth comparing the results of Table 5 and Table 4 to test whether specific styles may have
emerged as strong factors in explaining sidemen overlap. For the purpose of this comparison, I
compared the Single Style parameters (Table 5) with the Style Similarity parameters for each period
(Table 4). This comparison is important in order to explore whether the importance of style similarity
was equal to all styles, or whether that effect was more important to some but not all styles. Taking
this comparison as a benchmark, we might identify the following styles as noteworthy in terms of
explaining sidemen overlap: New Orleans from 1935 to 1949, both Bop and Cool from 1945 to 1964,
and Soul from 1955 to 1969. Because the Similar Style parameter is not significant for 1960 to 1964, all
styles (except for New Orleans) appear as noteworthy during this period. This exercise permits us to
observe that some styles at some specific periods will emerge as more important predictors of
association. For instance, the reemergence of a style (New Orleans) or the very inception of a new style
(e.g. Bop) becomes stronger predictors.
Hypothesis 2 states that the likelihood of sidemen overlap will increase the higher the number of
sessions recorded for the bandleader dyad. Both sets of models (Tables 4 and 5) confirm this
hypothesis as the ‘Session Prod’ covariate is always positive and significant. Nevertheless, we can
observe a decline in strength over the course of the periods analyzed. This downward trend may have
occurred for a variety of reasons. As identity boundaries became more important and jazz less popular,
bandleaders may have been less attracted to hire sidemen based on their production volume alone.
26
My aim in this paper is to understand how attributes and categories influence tie formation, which does not involve network
variables (e.g., transitivity or centrality measures). For this reason, the MRQAP model is adequate (Snijders, 2011). This
algorithm is built into the UCINET package, version 6.285 (see Borgatti, Everett, & Freeman, 2002).
27
Correlations between variables for each period are available on request.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
Prod. of number of sessions
Diff. of number of sessions
Past association
Both African-descendents
Neither African-descendents
Female
Male
U.S.
Non U.S.
Diff. Field Age Field (ln)
Instrument identity
Style similarity
Adjusted R2
Observations
*
**
***
p< 0.1.
p< 0.05.
p< 0.01.
1935–39
***
0.426
0.041***
0.010
0.089***
0.040***
0.023*
0.097***
0.015*
0.036***
0.015*
0.004
20.20%
12,126
1940–44
***
0.213
0.027***
0.111***
0.018**
0.006
0.001
0.004
0.016***
0.038***
0.009*
0.011*
0.006
7.40%
25,356
1945–49
***
0.420
0.056***
0.159***
0.022***
0.032***
0.004
0.014**
0.058***
0.046***
0.017***
0.004
0.002
19.70%
39,228
1950–54
***
0.247
0.011**
0.195***
0.026***
0.005
0.000
0.003
0.020***
0.016***
0.004
0.001
0.011***
10.80%
77,488
1955–59
***
0.179
0.006*
0.200***
0.033***
0.010***
0.002
0.005
0.016***
0.020***
0.008**
0.000
0.019***
7.80%
145,270
1960–64
***
0.207
0.025***
0.207***
0.055***
0.020***
0.004*
0.002
0.054***
0.006**
0.016***
0.000
0.037***
9.70%
248,682
1965–69
***
0.165
0.008***
0.227***
0.018***
0.002
0.000
0.000
0.016***
0.000
0.002
0.004**
0.032
8.40%
263,936
0.171***
0.022***
0.323***
0.018***
0.002
0.001
0.001
0.024***
0.009***
0.007**
0.001
0.041***
13.90%
294,522
G Models
POETIC-1195; No. of Pages 25
1930–34
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
17
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
Table 4
MRQAP parameters per period, style similarity included.
*
p< 0.1.
p< 0.05.
***
p< 0.01.
**
0.011
0.088***
0.040***
0.024*
0.096***
0.015*
0.036***
0.015*
0.003
0.006
20.20%
12,126
1935–39
0.215***
0.027***
0.111***
0.016**
0.004
0.001
0.004
0.015**
0.038***
0.009*
0.012*
0.019***
0.002
7.40%
25,368
1940–44
0.420***
0.056***
0.159***
0.021***
0.032***
0.004
0.014**
0.058***
0.047***
0.017***
0.004
0.011*
0.004
0.005
19.70%
39,270
1945–49
0.249***
0.014***
0.196***
0.021***
0.002
0.001
0.009**
0.017***
0.017***
0.001
0.000
0.025***
0.018***
0.054***
0.019***
0.000
11.20%
77,558
1950–54
0.180***
0.004
0.199***
0.031***
0.010***
0.001
0.004
0.013***
0.021***
0.010***
0.001
0.011***
0.001
0.033***
0.035***
0.007**
0.013***
0.010**
8.10%
145,270
1955–59
0.202***
0.023***
0.204***
0.048***
0.021***
0.004*
0.002
0.052***
0.008***
0.013***
0.001
0.009***
0.020***
0.041***
0.047***
0.003*
0.026***
0.038***
0.011***
10.30%
248,682
1960–64
0.162***
0.007***
0.226***
0.015***
0.005**
0.000
0.001
0.016***
0.000
0.002
0.005**
0.001
0.004**
0.011***
0.010***
0.007**
0.031***
0.026***
0.028***
8.60%
263,948
1965–69
0.168***
0.022***
0.322***
0.012***
0.005**
0.001
0.002
0.024***
0.008***
0.006**
0.001
0.008***
0.010***
0.008***
0.001
0.002
0.017***
0.048***
0.013***
0.036***
14.20%
294,522
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
Adjusted R2
Observations
0.427***
0.041***
G Models
1930–34
Prod. of number of sessions
Diff. of number of sessions
Past association
Both African-descendents
Neither African-descendents
Female
Male
U.S.
Non U.S.
Diff. Field Age Field (ln)
Instrument identity
Both leaders play New Orleans
Big band
Bop
Cool
Latin
Hard
Soul
Free
Fusion
POETIC-1195; No. of Pages 25
18
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
Table 5
MRQAP parameters per period, styles included.
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
19
Yet, this effect is never weak or non-significant. This means that although the jazz field underwent a
significant shift to the avant-garde, popularity was always a strong driver of music association. As
expected, the control variable ‘Session Diff’ was usually negative and significant, which strengths the
idea that the higher the asymmetry in session production, the lower the odds of sidemen overlap.28
Hypothesis 3 states that bandleaders showing high asymmetries in the ‘time in field’ variable will be
more likely to present sidemen overlap. In general, this hypothesis is confirmed as we observe negative
and significant parameters. In the 1945–1949 period this parameter becomes non-significant.29 The
pattern is again reversed in the sixties as we observe a positive and significant parameter during the
1965–1969 period. When we observe a negative parameter, we may interpret that bandleaders who
entered in the field at the same time were employing the same sidemen in their recording sessions. When
this parameter becomes positive, we may interpret that older bandleaders are employing sidemen with
younger bandleaders. If generational coupling is no longer in effect, this might mean that old bandleaders
are seeking to renew their bands and quickly incorporate new skills. Conversely, young bandleaders
might be more prone to employ sidemen usually associated with older bandleaders, in order to gain
repertoire, experience and status. Future studies might be able to disentangle these two mechanisms.
It is worth analyzing the trends for the race, gender and nationality variables underlying sidemen
overlap. From the 1935 to 1939 period to the 1955 to 1959 period, when both bandleaders were
African-descendent the odds of sidemen overlap increased, declining sharply thereafter. Sidemen
overlap among ‘Non-African-American’ bandleaders was stronger than for the ‘Both AfricanAmerican’ parameter during two periods: 1930–1934 and 1940–1944 periods. National origin
mattered for sidemen overlap in the period under analysis as both US and Non-US parameters were
always positive and significant. Nevertheless these patterns were not constant. The US parameter
reached its peak in the 1930–1934 period when jazz was mostly recorded, while the US National
parameters (‘Both US’ and ‘Both Non-US’) reach their highest point between 1940 and 1944 owing to
the Second World War. After the war, these sets of variables become less determinant of sidemen
overlap, especially for Non-US, probably due to the increasing influxes of jazz musicians into Europe.
When both bandleaders were women, this accounted for sidemen overlap mainly during the 1930–
1934 period due to the presence of women in the swing/sweet bands.
‘past association’ was always positive and significant, but after 1950–1954 it became the strongest
variable in predicting sidemen overlap. It is worth noting that past association was always a stronger
predictor than style similarity or race.
6. Discussion
6.1. Extension of Peterson’s process model
Observing the change in the sidemen overlap patterns, we can suggest that the jazz field went
through different phases. Borrowing from the parlance of institutional theory, jazz experienced a shift
between distinct underlying logics (Thornton, Jones, & Kury, 2005). In the first phase, most notably from
1930 to 1949, bandleaders were predominantly led to overlap sidemen because of the number of
sessions (H2). Based on this evidence, we can suggest that this phase presented the highest drive toward
a commercial logic.
From 1945 to 1969, we can observe a subsequent – though partially overlapping – period during
which sidemen overlap was increasingly explained by style similarity (H1). We could infer that sidemen
overlap in this period also follows an identity-based logic, where style similarity yields higher odds of
association. Taking Hypotheses 1 and 2 in tandem, we can propose a few insights into how the jazz field
underwent this transition. As the commercial thrust of jazz lost its power, jazz musicians increasingly
looked to promote and protect their public style identity. This finding is counterintuitive insofar as we
might expect an ‘identity-based’ sidemen overlap to be a luxury afforded only during prosperous times,
28
However, in two periods this control variable assumed positive and significant parameters: 1935–1939 and 1945–1949.
This reversal might be accompanied by stylistic shifts in jazz. In the 1935–1939 period, swing observed an introduction of hot
elements, and upcoming bandleaders like Benny Goodman were able to supersede more ‘sweet’-inclined bandleaders.
29
This was when a new generation of boppers were gaining increased access to recording companies and were probably just
then shifting roles from sidemen to bandleaders.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
20
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
while, following the same reasoning, during economic downturns commercial interests should prevail,
expressed in the higher probability of sidemen overlap among wealthier bandleaders. However, the
findings in fact provide support for the Peterson process model where the musicians remaining in the
jazz field increasingly espouse an ‘inverted’ economic logic. This insight is also found in the Bourdieusian
research tradition (Bourdieu, 1983; Heinich, 1997).
It is worth examining the inversion in the effect of ‘Time in the field’ variable. While its effect was
negative, in the last period it becomes positive and significant. As suggested above, such inversion
suggests that older bandleaders were seeking new skills from younger musicians, as well as young
bandleaders were attempting to associate with musicians with larger repertoires and established
status. Taking Hypotheses 1 and 3 together, a higher emphasis on style led to higher inter-generational
collaboration (rather than higher intra-generational collaboration). This finding goes against
Bourdieu’s idea that young musicians would challenge and break with older musicians. Perhaps
because jazz was suffering from a strong centrifugal drive and increasing paradigm crisis, younger
artists saw on older peers a source of legitimacy – which is expressed on higher rates of intergenerational association. It would be interesting to explore whether such patterns are observed in
other fields that did not undergo the same fragmentation and crisis.
The examination of specific style parameters (Table 5) reveals a number of patterns that invite a
revision of Peterson’s model. Reinforcing the latter, we observe how bop (relatively avant-garde) was a
strong factor in explaining sidemen overlap (0.054 for the 1945–1949 period, higher than any other
style during this time span). However, in tandem with bop, we observe how cool (allegedly a deviant
‘pop’ style) was also a strong predictor of sidemen overlap (0.035 for the 1950–1954 period, higher
than all other styles during this time span). Furthermore, the parameter for New Orleans (classified as
a deviant return to folk), combined with its share of sessions, reveal that this revival was a strong
movement within jazz (0.025 for the 1945–1949 period, second only to bop). Although affiliation to
free jazz reaches a significant level in the 1960–1964 period (0.028, second only to hard bop), it does
not become the top predictor among styles (as seen with bop earlier). These findings might suggest
that the following dynamics were at work. First, while we observe a movement toward avant-garde
styles, we may also observe the introduction of new pop-oriented styles that were well-integrated in
the jazz field, given that jazz had by then developed into a middle-brow genre (Bourdieu, 1984). In
contrast to Bourdieu’s model, though, the co-existence of styles was also accompanied by an intense
association between musicians of different styles. Second, the recovery and revival of jazz roots took
place in tandem with the introduction of new styles, since musicians (e.g. Ayer) would take inspiration
from the traditional jazz vernacular. Third, the introduction of Free Jazz did not have the same success
as bop (a plausible analogy, given that both styles depicted strong thrusts toward avant-garde music).
While bop innovators counted on strong links to commercial jazz, and amassed vast experience as
sidemen and bandleaders, the free jazz innovators (specially the second wave) lacked the same
advantages. Furthermore, while bop became widely accepted in the jazz world, free jazz remained
peripheral. This suggests that while certain aspects of a middle-brow genre might acquire stronger
support, the drive toward radical avant-garde (free jazz) could face a ‘ceiling.’ It might be worth
exploring the conditions that brings this ceiling to comparable genres.
As mentioned in Section 2.3, being ‘American’ and ‘White’ was a strong predictor of sidemen
overlap within the periods of 1930–1934, 1940–1944, 1950–1954, and 1955–1959, probably for
different reasons. From 1930 to 1934, racial segregation was at its peak. In contrast, during the 1950–
1954 and 1955–1959 periods, cool/West Coast jazz was probably more associated with conservatorytrained jazz musicians.
All black dyads present significant and increasingly strong parameters from 1935 until 1959. After
1959 these decline in strength, though remaining positive and significant. This acceleration may have
been due to a combination of the following factors: the declining effect of Jim Crow legislation,
increased access to small recording companies in the forties (including ‘race’ records), strengthened
identity and self-recognition among African-Americans (led and epitomized by the Civil Rights
movement), and reaction to white-associated styles (i.e. West Coast jazz).30 The decline in this
30
It is telling that the peak of the ‘Both African Descendent’ parameter was during period VI, when jazz observed the strong
success of cool/West Coast genres and, as seen above, a new wave of all-white bandleader dyads.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
21
parameter is probably due to dynamics internal to the jazz world, including efforts to show that jazz
could be equally played by any ethnicity. While some musicians argued that certain styles were closer
to the African roots, several pundits would challenge this coupling of ethnicity and musical identity or
skills. John Coltrane was among the musicians opposed to any kind of segregation within the jazz
community. The magazine Down Beat too promoted ‘blind tests’ that challenged musicians to say
whether the player was black or white. Failure to recognize a musician’s race based on his or her sound
reinforced the argument against race-based boundaries. Yet the empirical findings of the present
paper do suggest that ‘race’ crosscuts Bourdieu’s ‘low-brow/high-brow’ cleavage (Lopes, 2000).
Turning our attention to national origin, we can observe that during the Second World War,
American and European musicians were relatively disconnected. While the war placed jazz musicians
on opposing sides of the battlefield, jazz itself was censored in many German-controlled areas. As the
war ended, we can observe a declining tendency for non-US bandleaders to overlap sidemen, while the
reverse is not true: the ‘Both US Bandleaders’ parameter remained strong. This may have two
explanations. First, European musicians may have attracted American sidemen in order to promote
jazz. Second, Europe was a safe harbor for African-American sidemen, still marginalized in the United
States in the period before the Civil Rights movement.
Finally, it is worth noting that Female homophily was significant in the 1930–1934 period, and, in a
weaker fashion, in the 1955–1959 period. Specifically in the case of women jazz musicians, during the
former period they were predominantly vocalists and frequently swing bandleaders (Tucker, 2000).
6.2. Further analytical extensions
Future research could explore the combination of effects depicted in this paper, namely: (1) race
and style, (2) style and number of sessions, (3) gender and style, and (4) style and nationality. The
relationship between race and jazz is also mediated through styles. While swing was associated with
white musicians in the twenties and thirties, New Orleans jazz was associated with black musicians.
The ‘discovery’ of early jazz by critics in the early 1940s is also associated with the idea that ‘authentic’
jazz should be rooted in African-American genres (e.g. blues), while swing was often stigmatized as an
inauthentic version. The collective improvisation characteristic of early jazz was evoked by left-wing
intellectuals in the 1930s as a virtue opposed to the dangers of exaggerated individualism. As a
consequence, when beboppers introduced sophisticated solo improvisations, and promoted soloists
as stars, they were accused of ‘individualism,’ ‘commercialism’ and ultimately ‘‘selling out true jazz’’
(Gendron, 1995; Panassie, 1973). Likewise, when ‘free jazz’ musicians pursued atonal voicing in their
music and associated with avant-garde contemporary artists, they were accused of being ‘too
European,’ as though betraying an externally-imposed identity.
Much less research has been devoted to the role of gender and nationality since jazz is traditionally
conceived to be American (Burns, 2004) and dominated by men (Heckathorn & Jeffri, 2001). These two
variables could be further explored in future research as both nationality and gender may well have
influenced both the formation of social ties and style adoption.
6.3. Geography, locale and record labels
Marquis (2003) has observed that ties are strongly influenced by territorial localization. The same
may be observed in the jazz field. This was partially captured by more inclusion of the ‘Both US
Musicians’ variable. However, within the United States jazz became gradually less clustered in New
York, as musicians in Europe spread beyond the London and Parisian centers. Geography may also
correlate with both ‘race’ and style homophily, given that local audiences might vary in terms of racial
and style acceptance. Following the same idea, it would be possible to determine how ‘locale’ (type of
venue) impacted the relationship between style and tie formation (Becker, 2004).
As suggested in Section 2.2, common affiliation to recording companies and labels may also present
an alternative explanation to tie formation for several reasons. First, sidemen overlap might be
affected by record labels who would often allocate sidemen to bandleaders. Also, more prominent
labels (e.g. Blue Note) might have attracted bandleaders with a higher potential. In a stable field,
where the ranking between labels is constant, this variable may be epiphenomenal in explaining
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
22
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
overlap formation. However, if style fragmentation was accompanied by the emergence and rise of
small labels to a prominent position, this might help us to understand better the turning points in
the field.
6.4. Dialog with adjacent theories
In the context of financial markets, Zuckerman, Kim, Ukanwa, and Rittmann (2003) were able to
show both theoretically and empirically a negative relationship between cognitive confusion and
stock evaluation. The authors’ rationale is based on the attention required by category-specialized
equity analysts to cover a given stock option. If a company tries to ‘‘be too many things for too many
audiences,’’ it will fail to get the same amount of attention as a company associated with a single
category. As a result, analysts penalize the company, passing on unfavorable recommendations to
their clientele.
Undoubtedly Zuckerman’s approach relies on the following assumptions: (1) stock analysts are
important gatekeepers in the financial market, (2) the financial market efficiently matches assets that
are amenable to consensual valuation techniques, and (3) focused cognitive attention is needed to
perform informed stock recommendations, leading to the relative stabilization of a category set.31
In a field where style categories are strongly structured, a multi-vocal bandleader might face strong
obstacles to the establishment of a public persona, though he or she will be able to tap from a wider
pool of resources. The approach I used here is related to the bandleader’s contributions insofar as it
tests whether relations are contingent upon style affiliation. Furthermore the findings in this paper
may indicate that the categorical imperative is contingent upon historically-constituted logics. That is
to say: at certain moments, multi-vocal categorical affiliation may not be subject to sanctioning so
long as the musician switches categories within the range of the jazz canon. When does sanctioning
become relevant? The establishment of strong gatekeepers may not be a sufficient condition, contrary
to what Zuckerman and his associates seem to assume: sanctioning may perhaps be more salient in
moment of crises when the lack of external structures leads actors to adhere to consistent lines of
action (Lizardo & Strand, 2009; Swidler, 1986). In the case of jazz, where musicians were able to
resignify low-brow elements into high-brow styles, critics were less capable to establish a stable
hierarchy of styles, preventing the enforcement of strong categorical boundaries (Rao, Monin, &
Durand, 2005; Wry, Lounsbury, & Jennings, 2014).
6.5. Methodological limitations
This paper has relied on the assumption that relationships between musicians are inferred from coparticipation in recording sessions. Certainly musicians played together and established relationships
outside recording studios. My major concern is whether, by relying on recording sessions, I am
introducing a bias. We can assume that whenever musicians enter a recording studio, they may have
some commercial expectations and may be more prone to maintaining their public identities. In
contrast, musicians meeting in informal settings may be free to experiment, or to play whatever the
audience requests, without having to worry that the session will be recorded for future reference. As a
result, the inferred relationships are based on those artifacts influenced (to varying degrees) by
commercial success and identity simultaneously.32
Another source of concern would be whether the non-observable sessions (not captured in this
dataset) would produce different results, and, furthermore, whether the sampling methodology
would become entangled with the field’s trends. The recording session database was susceptible to the
following biases: (1) it is influenced by technological shifts, and (2) it may be biased toward
commercial success.
In addition, one of the criteria I chose was to include bandleaders who recorded at least twenty
sessions throughout their lives. Although this threshold is not high, it becomes another source of bias
31
By contrast, other alternative markets are available for assets that do not fit well into any category, or just would not be well
appreciated on the open stock market. In such settings, categorical coupling would have much less sanctioning power.
32
See Ward (2009) on how cover music became less legitimate as pop styles emerged.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
23
favoring commercially-oriented musicians. This threshold might impact the periods in my sample
differently due to technological shifts. Musicians who lived during the thirties and forties might have
had fewer chances to record than musicians who performed during the fifties and sixties.
7. Conclusion
This paper builds on Peterson’s description of jazz history via a process model. To some extent, it
confirms his model when it depicts the weakening of the commercial logic, while the style-based
identity logic is strengthened. But it extends beyond Peterson’s model in a number of ways. First, as
suggested by several jazz scholars, race emerges as an important factor and should be included as a
variable along with style in order to capture the dynamics within the jazz field.
Second, it suggests how ‘time in the field’ asymmetries might both reinforce and expand the
generational narrative underlying the Peterson’s original process model to the extent that when
asymmetries did not lead to lower odds of sidemen overlap, this signaled important shifts in the jazz
field. Current evidence suggests that the shift from intra-generational to inter-generational
collaboration occurs in tandem with the emergence of style similarity as strong predictor of
musicians’ association, in contrast with what we would expect following Peterson’s or Bourdieu’s
insights.
Furthermore, I have shown that past associations became the most important factor in explaining
sidemen overlap, regardless of the dyad’s style. From a methodological perspective, this finding
reinforces the need to explore the social network effects in field dynamics in explicit fashion. From a
substantive perspective, it provides a more nuanced narrative to Peterson’s process model,
demonstrating that the emergence of new styles, rather than being accompanied by disruptive
dynamics, was grounded in intense cross-style collaboration.
Finally, I have established a potential dialog with adjacent theories (e.g. ‘categorical imperative’
and the role of critics) and shown that the sanctioning power of gatekeepers may depend not only on
field maturity, but also on the desire for stabilization.
Acknowledgements
I am grateful to Roberto Kirschbaum for technical assistance[1_TD$IF] and Cristina Sakamoto for help with
data collection. For comments on earlier drafts, I am indebted to Paul DiMaggio, Timothy Dowd, Sergio
Lazzarini, Karen Levi, Woody Powell, Patricia Thornton, Kewin Lewis,[2_TD$IF] Shyon Baumann and the Poetics
reviewers. This research was supported by grants from CAPES.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.
org/10.1016/j.poetic.2015.04.001.
References
Abbott, A., 2005. Linked ecologies: States and universities as environments for professions. Sociological Theory 23, 245–274.
Anheier, H.K., Gerhards, J., Romo, F.P., 1995. Forms of capital and social structure in cultural fields: Examining Bourdieu’s social
topography. The American Journal of Sociology 100, 859–903.
Anderson, I., 2007. This is our music: Free Jazz, the Sixties, and American culture. University of Pennsylvania Press, Philadelphia.
Baumann, S., 2007. Hollywood Highbrow: From entertainment to art. Princeton University Press.
Becker, H., 2004. Jazz placesIn: Music scenes: Local, translocal, and virtual, pp. 17–27.
Borgatti, S.P., Everett, M.G., Freeman, L.C., 2002. Ucinet for windows: Software for social network analysis. Analytic Technologies, Harvard, MA.
Bourdieu, P., 1980. The production of belief: Contribution to an economy of symbolic goods. Media, Culture & Society 2 (3),
261–293.
Bourdieu, P., 1983. The field of cultural production, or: The economic world reversed. Poetics 12 (4), 311–356.
Bourdieu, P., 1984. Distinction: A social critique of the judgment of taste. Harvard University Press, Cambridge, MA.
Bourdieu, P., 1985a. The genesis of the concepts of habitus and of field. Sociocriticism 2, 11–24.
Bourdieu, P., 1985b. The social space and the genesis of the groups. Theory and Society 14, 723–744.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
24
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
Bourdieu, P., 1986. The forms of capital. In: Handbook of theory and research for the sociology of educationGreenwood Press,
Westport, pp. 241–258.
Bourdieu, P., 1992. Les regles de l’art: genese et structure du champ litteraire. Editions du Seuil, Paris.
Bourdieu, P., 1993. The field of cultural production. Columbia Univ. Press, New York.
Bourdieu, P., 2002. Le Bal des célibataires: Crise de la société paysanne en Béarn. Seuil, Paris.
Bourdieu, P., Darbel, A., Schnapper, D., 1990. The love of art: European art museums and their public. Stanford Univ. Press.
Bourdieu, P., Wacquant, L.J.D., 1992. An invitation to reflexive sociology the purpose of reflexive sociology (The Chicago
Workshop). The University of Chicago Press, Chicago, pp. 61–215.
Burns, K., 2004. Jazz: A film by Ken Burns. PBS Home Video.
De Nooy, W., 2002. The dynamics of artistic prestige. Poetics 30, 147–167.
DeVeaux, S., 1991. Constructing the jazz tradition: Jazz historiography. Black American Literature Forum 25, 525–560.
DeVeaux, S., 1997. The Birth of Bebop: A social and musical history. University of California Press, Berkeley, CA.
Dekker, D., Krackhardt, D., Snijders, T.A.B., 2007. Sensitivity of MRQAP tests to collinearity and autocorrelation conditions.
Psychometrika.
DiMaggio, P., 1977. Market structure, the creative process, and popular culture: Toward an organizational reinterpretation of
mass-culture theory. Journal of Popular Culture 11, 436–452.
DiMaggio, P.J., 1986. Research in organizational behavior structural analysis of organizational fields: A blockmodel approach. JAI
Press, Greenwich, CT, pp. 335–370.
DiMaggio, P.J., 1987. Classification in art. American Sociological Review 52, 440–455.
DiMaggio, P.J., 1997. Culture and cognition. Annual Review of Sociology 23, 263–287.
Emirbayer, M., Goodwin, J., 1994. Network analysis, culture, and the problem of agency. The American Journal of Sociology 99,
1411–1454.
Faulkner, R.R., 1983. Music on demand: Composers and careers in the Hollywood film industry. Transaction Books, New
Brunswick.
Faulkner, R., Becker, H.S., 2009. Do you know? The jazz repertoire in action. The University of Chicago Press, Chicago.
Feld, S.L., 1981. The focused organization of social ties. American Journal of Sociology 1015–1035.
Gendron, B., 1995. ‘‘Moldy Figs’’ and modernists: Jazz at war (1942–1946). Jazz among the Discourses 31–56.
Gibbons, D.E., 2004. Friendship and advice networks in the context of changing professional values. Administrative Science
Quarterly 238–262.
Gillespie, D., 2009. To be, or not– to bop. University of Minnesota Press, Minneapolis.
Giuffre, K., 1999. Sandpiles of opportunity: Success in the art world. Social Forces 77, 815–832.
Greenfeld, L., 1989. Different worlds: A sociological study of taste, choice and success in art. Cambridge University Press,
Cambridge, UK.
Gridley, M., Maxham, R., Hoff, R., 1989. Three approaches to defining jazz. The Musical Quarterly 73, 513–531.
Heckathorn, D.D., Jeffri, J., 2001. Finding the beat: Using respondent-driven sampling to study jazz musicians. Poetics 28, 307–329.
Heinich, N., 1997. The glory of Van Gogh: An anthropology of admiration. Princeton University Press.
Jones, C., 1996. Careers in project networks: The case of the film industry. In: The boundaryless career: A new employment
principle for a new organizational eraOxford University Press, New York/Oxford, pp. 58–75.
Kahl, S., Kim, Y.K., Phillips, D.J., 2010. Identity sequences and the early adoption pattern of a jazz canon (1920–1929). Research in
the Sociology of Organizations 31, 81–113.
Kahn, A., 2003. A love supreme: The story of John Coltrane’s signature album. Penguin (Non-Classics).
Kalish, Y., 2013. Harnessing the power of social network analysis to explain organizational phenomena. Modern Research
Methods for the Study of Behavior in Organizations 99.
Kernfeld, B., Rye, H., 1995. Comprehensive discographies of Jazz, Blues, and Gospel. Notes 865–891.
Lahire, B., 2004. La culture des individus: dissonances culturelles et distinction de soi. La Découverte, Paris.
Lahire, B., 2008. The individual and the mixing of genres: Cultural dissonance and self-distinction. Poetics 36 (2–3), 166–188.
Lamont, M., Molnar, V., 2002. The study of boundaries in the social sciences. Annual Review of Sociology 28, 167–195.
Lena, J., 2012. Banding together: How communities create genres in popular music. Princeton University Press, Princeton.
Lizardo, O., 2006. How cultural tastes shape personal networks. American Sociological Review 71, 778–807.
Lizardo, O., Strand, M., 2009. Skills, toolkits, contexts and institutions: Clarifying the relationship between different approaches
to cognition in cultural sociology. Poetics 38, 205–228.
Lopes, P.D., 2000. Pierre Bourdieu: Fieldwork in Culture Pierre Bourdieu’s Fields of Cultural Production: A case study of Modern
Jazz. Rowman & Littlefield Publishers, Lanham, pp. 165–185.
Lopes, P.D., 2002. The rise of a jazz art world. Cambridge University Press, Cambridge.
Mannheim, K., 1970. The problem of generations. Psychoanalytic Review 57 (3), 378–404.
Marquis, C., 2003. The pressure of the past: Network imprinting in intercorporate communities. Administrative Science
Quarterly 48, 655–689.
Martin, J.L., 2003. What is field theory? American Journal of Sociology 109, 1–49.
Pachucki, M., Pendergrass, S., Lamont, M., 2007. Boundary processes: Recent theoretical developments and new contributions.
Poetics 35, 331–351.
Panassie, H., 1973. The real jazz. Greenwood Press, Westport, CT.
Peress, M., 2008. Dvorák to Duke Ellington: A conductor explores America’s music and its African American roots. Oxford
University Press, New York, Oxford.
Peterson, R.A., 1972. A process model of the Folk, Pop and Fine Art phases of Jazz. In: American music: From Storyville to
WoodstockTransaction Books, New Brunswick, NJ, pp. 135–151.
Peterson, R.A., 1990. Why 1955? Explaining the advent of rock music. Popular Music 9, 97–116.
Peterson, R.A., Berger, D.G., 1975. Cycles in symbol production: The case of popular music. American Sociological Review 40 (2),
158–173.
Phillips, D.J., Owens, D.A., 2004. Incumbents, innovation, and competence: The emergence of recorded jazz, 1920 to 1929.
Poetics 32 (3), 281–295.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001
G Models
POETIC-1195; No. of Pages 25
C. Kirschbaum / Poetics xxx (2015) xxx–xxx
25
Phillips, D.J., Kim, Y.K., 2009. Why pseudonyms? Deception as identity preservation among jazz record companies, 1920–1929.
Organization Science 20 (3), 481–499.
Pinheiro, D.L., Dowd, T.J., 2009. All that jazz: The success of jazz musicians in three metropolitan areas. Poetics 37, 490–506.
Podolny, J.M., 2001. Networks as the pipes and prisms of the market. American Journal of Sociology 107, 33–60.
Rao, H., Monin, P., Durand, R., 2005. Border crossing: Bricolage and the erosion of categorical boundaries in French gastronomy.
American Sociological Review 70, 968–991.
Snijders, T.A.B., 2011. Statistical models for social networks. Annual Review of Sociology 37 (1), 131–153.
Swidler, A., 1986. Culture in action: Symbols and strategies. American Sociological Review 51, 273–286.
Swidler, A., 2001. Talk of love: How culture matters. Chicago University Press, Chicago.
Szwed, J.F., 2000. Jazz 101: A complete guide to learning and loving jazz. Hyperion, New York.
Thornton, P.H., Jones, C., Kury, K., 2005. Transformation in cultural industries institutional logics and institutional change. In:
Organizations: Transformation in accounting, architecture, and publishing, research in the sociology of organizationsElsevier, Oxford, pp. 125–170.
Tucker, S., 2000. Swing shift: ‘‘All-Girl’’ bands of the 1940s. Duke University Press, Durham, NC.
Wald, E., 2009. How the beatles destroyed rock n roll: An alternative history of American popular music. Oxford University
Press, USA.
Weber, M., 1958. The rational and social foundations of music. (D. Martindale, J. Riedel, G. Neuwirth, Trans.)Southern Illinois
University Press.
White, H.C., 1970. Chains of opportunity: System models of mobility in organizations. Harvard University Press, Cambridge, MA.
Williams, M., 1983. The jazz tradition. Oxford University Press, Oxford.
Wimmer, A., Lewis, K., 2010. Beyond and below racial homophily: ERG models of a friendship network documented on
Facebook. The American Journal of Sociology 116 (2), 583–642.
Wry, T., Lounsbury, M., Jennings, P.D., 2014. Hybrid vigor: Securing venture capital by spanning categories in nanotechnology.
Academy of Management Journal 57 (5), 1309–1333.
Zegers, F.E., Ten Berge, J.M.F., 1985. A family of association coefficients for metric scales. Psychometrika 50 (1), 17–24.
Zuckerman, E.W., Kim, T.-Y., Ukanwa, K., Rittmann, J.V., 2003. Robust identities or nonentities? Typecasting in the feature-film
labor market. American Journal of Sociology 108, 1018–1074.
Data sources
Kernfeld, B., 2001. The new grove dictionary of jazz. Oxford University Press, Oxford.
Lord, T., 2005. The jazz discography (Version 5.0) Canada Lord Music Reference Inc., Taylor Way, West Vancouver, BC.
Yanow, S. All Media Guide. Available at: www.allmusic.com, accessed 2005/2006.
Charles Kirschbaum holds a Ph.D. in Business Administration (in Organizational Studies) from the Getúlio Vargas Foundation
(FGV) and completed his post-doctorate in Economic Sociology at Columbia University. He is Assistant Professor at Insper, São Paulo,
and associate researcher at Center for Metropolitan Studies. He is currently developing a research at public schools in São Paulo
municipality.
Please cite this article in press as: Kirschbaum, C., Categories and networks in jazz evolution: The
overlap between bandleaders’ jazz sidemen from 1930 to 1969. Poetics (2015), http://dx.doi.org/
10.1016/j.poetic.2015.04.001