Over Time:
Theory
The theoretical modeling, analysis and redeployment of jazz improvisational,
and time-feel, mechanisms
Milton Mermikides
This collection of writings contains two theoretical papers and a compilation of case
studies. The first paper, M-Space and Expressive Contours offers a consolidated theoretical
model of jazz improvisation. This draws together a range of pedagogical sources and
concepts behind the diverse compositional portfolio (Changes Over Time: Practice). It also
serves as a necessary foundation for the second paper (Time-feel), which presents a model
of expressive micro-timing and forms the central theoretical contribution in the thesis.
Finally, a collection of analyses (Case Studies) drawn from a varied repertoire, and
Throughout the thesis, the term jazz improvisation is used repeatedly, so it’s advisable
at the outset to clarify this terminology. The area of research in this submission focuses
upon the popular genres of African-American born instrumental music including jazz,
blues, funk, soul and rock; but also extends to the variously related genres including
reggae, ska, fusion, hard rock and metal (with various levels of density) and the hip-hop,
rap, R&B and electronic dance styles. However, the musical mechanisms at work are
technological genres. Any one word would be too broad, or too narrow, for satisfaction,
but references will be made where appropriate and the moniker ‘jazz’ used as a
terminological pitfall: it may be argued that a distinction should be drawn between the
3
practice of free improvisation within a wider musical community – sometimes conducted
with little special training - and the enormously skilled discipline of the jazz heritage with
its demands on form, rhythmic accuracy, aural skills, ensemble techniques and evolved
language, on which this thesis focuses primarily. However, the techniques presented in
this thesis are found in, and are applicable to, all skilled improvisational scenarios from
traditional ‘standards’ improvisation to ‘free jazz’. It is telling that Hal Crook, an eminent
technique before addressing ‘free’ jazz (Crook, 2006). These various forms, although
various musical components (solo forms, melody, harmony, vocabulary, structure etc.),
rather than entirely different skill-sets.1 Terminology from the jazz idiom, which often
has little consensus on precise meaning, is used in this thesis and somewhat refined, and
some new terms have also been coined. A glossary of such terms is included in Section
4.3 (p 172).
The theoretical models in this thesis have been developed through years of jazz
training, performance, composition, teaching and audio production, and are informed
theoretical concepts are refined and continually reintroduced into the creative process
with varying degrees of premeditation and intuition. A critical reflection of the results
helps refine and develop these theories for further deployment. During this cyclical
process, music technology has provided an invaluable tool with which to analyse the
repertoire, test and employ theoretical models, and enhance significantly creative
sometimes employed, but these examples should always be accompanied with the
1A comprehensive and instructive survey of ‘free jazz’ is provided in Northern Sun, Summer Moon: Europe’s
Reinvention of Jazz (Heffley 2005).
4
reminder of the important limitations inherent in this system. Standard notation is not
without value, however due to its impoverished and discontinuous treatment of pitch,
timbre and rhythm, it should always be seen as a convenient approximate guide (masking
all manner of tacit stylistic implications), and not a definitive account of the deep
complexity of music that exists in the artist’s creative intention, the captured sound wave
Over Time: Practice are made, these may be explored as the examiner is compelled,
however the presentation of concepts in the theoretical writings has been made so that a
complete reading forms a framework and terminology with which to best approach the
portfolio.
All audio examples cited in the theoretical writings appears on CD1: Audio Examples
(where for example, CD1.3 identifies the third track of the CD). Some of these audio
examples are short and require attentive listening, so the ability to loop a track, and the
5
1. M-Space and Expressive Contours
The modeling and reapplication of jazz improvisational technique
Abstract
This paper presents, from a practitioner’s perspective, a consolidated model of jazz improvisation. This is
drawn from a range of theoretical and pedagogical sources, as well as the author’s own heuristic inquiry.
In this model, a musical object is seen as possessing an array of properties available for modification, and
existing at a point in multi-dimensional musical space (M-Space). Improvisation is represented as the
artful carving of trajectories through M-Space via corresponding gestural manipulations of consynchronous
musical parameters (expressive contours), which may form larger scale musical structures. This view of
improvisation offers practical applications for performance, as well as a framework in which to analyse
and appreciate the repertoire. Music technology is of great value in the facilitation of this model’s
employment in analysis, performance and composition. Also presented is the heterogeneous adoption and
reapplication of these concepts throughout a portfolio of stylistically diverse collaborative electroacoustic
works: ‘Changes Over Time: Practice’.
1.1 Introduction
technique. This approach demonstrates how the musical mechanisms found in the
beyond their presumed stylistic and practical restraints (Berio 2000, p 82-84). The
process from ‘behind the lines’, and an informed level of analysis. Furthermore, extensive
experience with music technology has enabled the design and employment of tools to
6
extent beyond the limitations of traditional practice. These resources provide an
opportunity to extend the creative reach of the improviser and, at the furthest extreme,
creative input from the composer, rather deferring musical decisions to various external
physical patterns.
This thesis serves to identify and demonstrate the utilisation of jazz improvisational
technique into a wider stylistic and theoretical context. This binding of a heuristic
music technology can inform the compositional process and improvisational practice in a
range of styles.
The model of jazz improvisation presented here, drawn together from diverse
theoretical and pedagogical concepts, has wide opportunities for application, and these
are outlined with reference to the Changes Over Time: Practice portfolio.
7
1.2 Background
The analysis and pedagogical focus of the jazz idiom before the 1980s was largely
limited to those musical features most easily described within the standard notational
system. Standard transcription, scale choices, harmonic extensions, melodic material and
formal structures took precedence over hugely important stylistic features such as
consideration and melodic interpretation. Improvising Jazz (Coker 1987) first published in
1964, represents a typical approach of its time: A comprehensive study of useful scales,
progression and chords, but ‘swing’ is incompletely defined (Coker 1987, p 45-9) and a
welcome mention of contour is frustratingly fleeting (Coker 1987, p 54-5). With some
notable exceptions2, jazz practitioners sided with the analysts’ bias, touting ever more
complex harmonic and scalar systems while largely ignoring all other salient stylistic
features.3 This can be attributed to the difficulty in analyzing certain idiomatic practices,
but may also the result of the incentive of jazz analysts and performers to achieve a sense
of academic parity with ‘classical’ music within the established analytical framework. It is
not that these other unarticulated aspects of improvisation were considered unimportant
listening and absorption, and not direct instruction. Jamie Aebersold’s introduction to a
popular Charlie Parker transcription, the Charlie Parker Omnibook, illustrates this:
2 Charlie Mingus’ 1962 lecture discussing the beat ellipsis (Mingus, cited in Berliner, 1994, p 151), solo
structuring in Clifford Brown’s playing (Stewart 1973), and Gunther Schuller’s 1968 (Schuller 1986, p 58)
linking of jazz improvisation with seed-pattern variations in African music (Jones 1959) are salient and
unusual analytical approaches for the time.
3The jazz community’s adoption of such texts as Nicholas Slonimsky’s Thesaurus of Scales and Melodic
Patterns (Ratliff 2008 and Slonimsky 1999), George Russell’s Lydian Chromatic Concept of Tonal Organization
(Russell 2008) and The Schillinger System of Musical Composition (Schillinger 1978) illustrate the bias of scalar
and harmonic devices over other important mechanisms.
8
Only a minimum of articulations have been put in this book.
We feel that jazz, being an aural art form, is often times best
imitated by listening over and over, and then playing the notes
the way you hear it on the record. This might seem like the
long way to do it, but experience has proven reliable. After all,
who would object to listening anyway? Listening is what
music is all about.
Aebersold & Slone, 1978, p iv
Ironically, many of the tools and vocabulary helpful to the academic understanding
and development of the jazz idiom were already underway in other genres4. The lack of
adoption of these tools by authoritative jazz practitioners of the time was a loss to all
genres concerned. Meanwhile, non-pedagogical research into the jazz idiom was confined
to the areas of harmonic observation, cultural history and heuristic enquiry, with little
rather than more analytical musical perspectives.5 Other research into improvisation (in
the general sense) has addressed relevant issues but is often divorced from the stylistic
practice within the jazz idiom, and largely from a non-jazz practitioner’s perspective.6
However, the advents of digital audio analysis, computer-based modeling systems and a
fresh, stylistically relevant approach have started to demystify and illuminate the
4 Folk music research, (in the areas of cantometrics (Lomax 1968) and melodic contour typology (Adams 1976)),
semiotic analysis (Cook 1994, p 151-82) and the field of notes inégales (Fuller 1980) were all readily amenable to
the study of jazz improvisation. The lack of adoption into jazz research of the concepts in electroacoustic
music particularly with a timbral, non-score based concept of the musical object, such as (Erickson 1975),
was also a missed opportunity.
5There exists a wealth of valuable cultural and phenomenological research from the 1970s including
(Nketia 1971), (Roberts 1972), (Wang 1973), (Titon 1973), (Nettl 1974), (Tirro 1977) (Sudnow 1978),
(White 1978) and (Sickler 1979).
6 The concept of generative improvisation (Clarke, 1988), aspects of timing (Gabrielsson 1988), tonal
theories (Lerdahl & Jackendoff 1983) and mechanisms of score interpretation (Sundberg 1988) are all
relevant improvisational research but were not immediately adopted in relation to jazz.
7Valuable and welcome research have emerged in such fields as Markov chain analysis of Coltrane solos
(Franz 1998) and broad improvisational processes and ensemble interaction theories (Hall 1992), (Larson
9
The emergence since the late 1980s of jazz improvisation writings from practitioners
these mechanisms beyond the immediate stylistic context. This paper combines the
theoretical concepts (found in the writings of Jeff Pressing, Eric Clarke, Trevor Wishart
particular emphasis on the crucial field of expressive micro-timing in time-feel) are, with
recorded repertoire are limited to the author’s chosen instrument, the guitar. However it
should be clear that the concepts presented are relevant to all instruments, and indeed, a
1998), (Monson 1991, 1994, 1996) and (Sawyer 1992). The hugely important and burgeoning field of
research in expressive micro-timing is explored and developed in Time-feel in this submission.
8 The new role of the jazz performer/educator/writer, usually associated with an institution such as
Berklee College of Music, offers a valuable insight to the jazz student and researcher. Key texts from top-
level practitioners who also have the ability to communicate clearly their concepts should not be ignored
by the research. See (Bergonzi 1992, 1994, 1996, 1998, 2000, 2002, 2004), (Crook, 1991, 1992a, 1992b,
1992c, 1995, 1999, 2006), (Goodrick 1987), (Hall 1991) and (Liebman 1991, 2000) for an introduction.
10
1.3 Chains of Thought and Musical Refractions
particular bias and interest in all subsequent creative endeavours. Although there are clear
New material is created by altering a selected set of musical parameters from previous
thought where every subsequent phrase relates to a preceding one in terms of a changing
transposition, metric placement and rhythmic subdivision (CD1.1). As is always the case
there are many important musical parameters - such as timbre, articulation and micro-
timing - present in the performance that elude the limitations of standard musical
notation.
Figure 1.3.1 (p 12) comprises one simple phrase that is transformed in terms of
easily visualised discrete pitches and rhythms. Phrases 2-34 can all be described as
rhythmic subdivision, with the exception of Phrase 30, which does not hold strictly the
9 See, for example, (Damian 2001, p 12-20), (Crook 1995, p 8-31) and The More Ways you Have of Thinking
(Berliner 1994, p 146-69)
11
Figure 1.3.1 Improvisation as transformation of coexisting musical parameters:
Phrase 1 is repeated with independent modifications of 3 parameters: Chromatic transposition, metric
placement and rhythmic subdivision (CD1.1).
Occasionally, phrases merge; for example, the last note of Phrase 5 is the first of
Phrase 6. This passage provides a clear, and easily notated example but this concept of
analytical power in a far wider range of complex examples. The following excerpt from
Swish (Mermikides 2008) illustrates the kind of conceptual process adopted in a more
complex manner (CD1.2). Figure 1.3.2 (p 13) presents a portion of the improvised solo,
10The term consynchronous is coined here, to describe multiple parameters that co-exist simultaneously
and continually along a time-line.
12
Figure 1.3.2 An illustration of the chains-of-thought improvisation methodology (CD1.2).
occurring phrase, and may themselves combine into larger, or break off into smaller,
phrases, and are labelled accordingly. The types of relationships between each phrase are
subset of musical properties of a phrase, or melodic fragment of the tune. This subset of
properties is then either fixed, or modified by varying amounts in the subsequent phrase,
13
4) Sufficient transformation may result in the formation of a new phrase unit for
further modification.
5) This analysis purposefully omits the complex interactions between performers in
an ensemble, whereby musical material created by one player may influence
the improvisations of any number of other players. The interaction between
players follows a similar methodology as described here, whereby musical
material is shared within a common ideas pool and modified between members
of the ensemble.11
6) A particular solo, performer’s identity or musical style may be described not just
by their melodic, harmonic and rhythmic vocabulary, but also by the types of
transformational processes and extent of variations employed.
7) Phrases G.1 and G.2 provide a glimpse of how technology may be employed as
part of the improvisational process, and how it may offer otherwise
unavailable transformational dimensions.
8) The precise demarcation of musical material into units for transformation, which
are called ‘phrases’ here, is a subjective exercise. Furthermore the definition
of a phrase unit may change in relation to the transformational process
employed. For example, a set of five notes might be considered a complete
phrase for a sequencing process, but a timbral modulation may also be
applied through those five notes, implying a smaller conceptual subdivision. It
is tempting (and sometimes useful) to describe musical fragments as cells,
which are combined into hierarchical phrases. But it soon becomes clear in
analysis that a cell’s autonomy is temporary, there are no uniformly indivisible
musical units; even a single note can be subject to all manners of
transformation and recombination, and parameters such as timbre do not
allow for such a convenient atomistic perspective12.
9) This example mainly presents a traditional improvisation where phrases occur in
a strict series. Contrapuntal mechanisms, ensemble interactions and the
smearing of a phrase (through electronics) allow phrases to coexist, and their
relationships to be parallel, rather than strictly serial.
See (Sawyer 1992) and (Berliner 1991, p 647-51) for theoretical, and transcription analyses of ensemble
11
14
10) There is a differing amount of variation from one phrase to the next; so a solo
may be characterised by the extent of relatedness (and how this measure
changes) between phrases. This concept of proximity, is discussed in Section
1.5 (p 25).
We may employ the above methodology, both as a form of retrospective analysis and in
terms of real-time improvisational choices. Figure 1.3.3 shows how a phrase offers the
performer a set of options for the continuation of an improvisation based upon various
transformational processes.
Figure 1.3.3 An illustration of musical refractions. In the course of an improvisation, a phrase is manipulated
by the selection of one of many transformational process (1-8 present a few of countless possibilities). The
resulting phrase is in turn open to further modifications. Improvisation is seen as the realization of a
pathway through the multitude of refracting musical possibilities.
Since there are a non-trivial number of precise transformational processes (let alone
combinations thereof), the above diagram shows but the briefest glimpse into the
15
improvisation may be seen as the artful carving of a pathway through this mesh of
Scofield’s Chank (Scofield 1998, CD1.3) on the other hand uses timbral modulation as an
Holdsworth’s playing (see Figure 1.6.1, p 56 and CD1.11), while Time-feel and Case Studies
rhythms.
improvisation into exclusively linear, serial and causal relationships between phrases, with
as this model is developed through this paper, it will be shown that these ‘pristine’
moments may in fact be accommodated into a theoretical model with the introduction of
surprise (p 25-55).
16
1.4 Limitation and Variation of Musical Topics
In the author’s own playing, the improvisational methodology outlined above grew
instrumental proficiency, jazz harmonic devices and vocabulary were developed, the core
Some external validation for this approach is received from both the pedagogical
and theoretical literature. Studies for a Bachelor’s degree at Berklee College of Music
beyond the typical ‘Learn your scales, transcribe and good luck’ advice. Instructors such
as Ed Tomassi, Jon Damian and Hal Crook would set particular challenges for students
such as:
1) Improvise through the tune using only chord-tones of the harmonic progression.
2) Improvise a short phrase. Rest. Improvise a short phrase. Rest. Improvise a long
3) Improvise a phrase that starts with the concluding material of the previous
exercise in forcing new ideas and avenues of exploration. Often the ensuing
improvisations are more successful than prior ‘free’ attempts, and the strictures of the
exercise rarely inhibit the potential musicality. Indeed, the educational benefit of
17
practicing improvisation within a carefully selected set of limitations is a central theme of
the work of jazz educators Crook, Tomassi, Jerry Bergonzi and Mick Goodrick. One
might think of this type of approach as the training of a particular type of skill: The
Developing proficiency in this area fosters a truly creative improvisation when the
authentically chosen material rather than pat phrases at the moment of improvisation:
In addition to the support from jazz pedagogical material, further validation, this
time from an academic theoretical standpoint, was offered by the paper Improvisation:
Methods and models (Pressing 1988) in which Jeff Pressing puts forward a model of
are defined. Alongside this set of variables, the amount of attention paid to each of these
variables is described by the currency of cognitive strength; this is unlikely to be more than
anecdotal reports on improvising seem to suggest that at any particular moment, the
creative improviser is thinking actively about one or two musical goals at the most
(Werner 1996, Nachmanovitch 2000 and Solstad 1991). Pressing’s model taken alone is
Nonetheless by defining this multi-level vision of music in which the improviser may
navigate, Pressing provides a very powerful conceptual vocabulary. Despite the difficulty
in proposed distributed cognitive strengths, the independent defining of attention to, and
values of, particular musical parameters and processes is illuminating. The staggering
18
developments in music technology now allow us to adopt this type of model in a
best informs the work in this submission. Bringing these theoretical concepts back to a
practical demonstration, here follows an example of the process. Here, a simple phrase
(Figure 1.4.1) is used as the starting point (a seed) for various subsequent improvised
phrases, which were then transcribed, within the limitations of standard notation.
Despite its simplicity, this phrase could be described in innumerable ways and levels
of detail such as: (i) a phrase starting on beat ‘4 and’. (ii) a three note rhythmic pattern
with no particular rhythmic placement, (iii) a melodic gesture, (iv) a broken chord
implying part of a Cmin9 or Ebmaj7 chord, (v) a phrase with a particular harmonic altitude
13 See Bäckman, K. & Dahlstedt P. (2008) for a recent significant development in this field.
19
Figure 1.4.2 Coexisting interpretations of Phrase α.
In this way, the phrase can be conceived as possessing many sets and subsets of
improvisation might develop with this concept in mind. Any number of the subsets of
musical characteristics may be used as a reference point for ensuing phrases. For
example, the starting beat may be fixed for a new phrase, the rhythmic pattern preserved
with new notes, or the melodic contour maintained but transposed and so on. Not only
can the concept of isorhytmos (fixed rhythmic structure) be explored but also isomelos (fixed
sequence of melodic pitches) (Persichetti 1961), isotimbre (fixed timbre), isopaesi (fixed
intensity), isomodos (fixed scale implication), isokinetos (fixed gesture) and isologos (a fixed
20
concept or pattern applicable across multiple parameters)14. From a jazz practitioner’s
(Pate, cited in Berliner 1994, p 385). A language naturally evolves from here to describe a
(displacement, distimbre etc.) Here follows some improvisations, each starting with Phrase α,
and with little preconception of what was to be played, other than the intent to vary the
Figure 1.4.3 Improvised continuations of Phrase α (CD1.4). Instances of Phrase α, and its close relations,
are labeled with solid and dashed outlines respectively.
Phrase 1 takes the rhythmic structure and general melodic shape of α as a constant,
The increased swing of the second phrase introduces us to the concept of expressive
micro-timing, an important focus of this submission, with its own dedicated paper
For clear examples of the compositional application of isokinetos and isologos see Primal Sound and
14
21
Phrase 2 uses the strict intervallic structure of α and employs chromatic
constant, but the placement of each note and articulation of the repeating melody is
rhythmic displacement and a simple polymetric implication. A shape on the fretboard is used
as a parameter of musical expression. The left hand shape is kept constant but played on
are seen as belonging to a larger set of notes, C jazz melodic minor. The introduction of
B-natural is a slight variation on the harmonic context and alters the ‘colour’ and harmonic
collapsed into a three-note harmonic structure that is repeated and transposed with
timbral modulation. A volume pedal is used to lengthen the natural attack time of the guitar
and interrupted with a phrase of contrasting rhythmic and melodic content, albeit with a
The initial phrase α is refracted through a few of many possible musical pathways, it
characterised as much by which parameters are held constant as by which are altered.
Figure 1.4.4 (p 23) illustrates how the particular musical topics, guide the ensuing
improvisation.
22
Figure 1.4.4 An illustration of how the fixing and variation of musical topics may forge
improvisational continuations from Phrase α.
in the ever-growing jazz pedagogical material of the last two decades. Some pedagogical
texts give an overview of many transformational topics within one book15, while other
writers choose to create a series of volumes addressing each topic separately, of which
jazz pedagogical material has started to resemble a library of chess books with stacks of
general principle texts alongside titles dedicated to every conceivable opening, variation
of opening and style of end game. The study of these differentiated skills, as in both
15 See Crook (1991, 1995) and Damian & Feist (2001) for examples of improvisational meta-views.
16 To date, Bergonzi has written 7 volumes of the Inside Improvisation Series, each focusing on a different
topic. (Bergonzi 1992, 1994, 1996, 1998, 2000, 2002 and 2004) In fact these can be seen as studies in the
fixing of these featured topics (e.g. a particular melodic cell) and thereby exploring deeply other variables
(e.g. permutation, harmonic altitude, segmentation etc.)
23
chess and jazz improvisation, aims to offer the player informed options and intuition at
there will inevitably be associated parameters that alter, or have to maintain constant, as a
consequence of any given improvisational choice. For example the ascending chromatic
(These types of relationships are categorized in 1.6 Changes and Time: Expressive Contours p
56). The consequential effects of any particular musical choice may also be harnessed
musically18.
This section has introduced and given some examples of the concept of
according to various parameters and wanders from its starting origin, while remaining
varying set of dimensions. The next section will look more deeply into this idea of
extensively in Changes Over Time: Practice. This concept is refined through analysis, linked
with a broad time component (Expressive Contours 1.6 p 56), bolstered with important
stylistic mechanisms (such as time-feel, harmonic pathways and melodic shadowing) and always
applied practically.
17Psychological models of improvisation are explored in such texts as (Solstad 1991), (Juslin & Sloboda
1991), (Hall 1992), (Monson 1996), (Gustavsen 1999), (Rothenberg 2002), (Reason 2004) and (London
2004)
18 In String Theory and Event Horizon from the submitted portfolio, the natural bowing technique associated
with playing certain passages is picked up by the electronics of the Hyperbow, which in turn modulates the
electronic effects of other consynchronous musical layers.
24
1.5 Improvisation as Musical Mutation: M-Space Modelling
The improviser, given a starting phrase, is presented with a range of choices for
the theoretical interest, this concept may be applied pedagogically and in performance to
note separation. The values of these parameters are notated in Figure 1.5.1.
Figure 1.5.1 An analysis of Coltrane’s Acknowledgement in terms of the values of 3 parameters: metric
placement, rhythmic separation and chromatic transposition (CD1.1).
possible to visualize all mutations of Phrase 1 in the same conceptual space. Figure 1.5.2
(p 26) shows a three-dimensional space that represents all the possible variations of the
phrase, in terms of the three specific parameters. The individually numbered phrases in
25
the solo are plotted in reference to the individual phrases in the solo. With the caveat
that important articulation, dynamic and timbral elements are set aside for now,
Coltrane’s solo may be seen as constituting a small subsection of this clearly demarcated
musical space.19
Figure 1.5.2 Coltrane’s cube: The phrases of Coltrane’s Acknowledgement plotted in the three-dimensional
musical space of metric placement, rhythmic separation and chromatic transposition, with a few co-
ordinates illustrated with standard notation.
19 The phrases’ positions in three-dimensional space has been derived here from a standard notation
transcription. However, this multi-dimensional graphical and conceptual vision can accommodate readily a
continuous – rather than discrete - range of values in all three parameters including sub-notational, yet
perceptible and musically relevant, elements of micro-timing and intonation. In this way it is liberated from
the “finitistic” limitations of Wishart’s lattice (see 2.1 p 85).
26
Figure 1.5.3 Raup’s cube: An illustration of shells existing in genetic space. Three genes (spire, flare and
verm) contribute to the shape of the shell. The shaded area of the cube represents the shells existing as a
product of natural selection
(Illustration from Raup, cited in Dawkins 1996, p 192).
Figure 1.5.3 (Raup, cited in Dawkins, 1996, p 192) shows Raup’s cube; an illustration
where three genes contributing to the shape of a shell (spire, flare and verm) are laid out in
three dimensions. Every possible expression of these genes is laid out in multi-
dimensional space and the evolutionary pathway, through genetic variation, of a species
may be drawn. The shaded area represents the shells that have been found in nature as a
product of natural selection, and a subset of all possible species.20 Similarly, the phrases
in Coltrane’s solo represent the ‘naturally-occurring’ subset of all possible phrases within
Figure 1.5.2 (p 26) used a simply arranged set of transformed phrases however the
exact layout of phrases within that musical space is debatable. One could make a good
case that potential phrases existing along a particular dimension may not always have an
more radical mutation than a minim displacement; an octave transposition is perhaps less
For an overview of current research in the use of evolutionary mechanisms in music composition, see
20
27
extreme than a semitone, and a semitone less extreme than a quarter-tone for that
matter.21
Phrase α (Figure 1.4.1 p 19) may be imagined as existing at a particular point on an axis
of rhythmic placement within the concept of a cyclical bar (Figure 1.5.4 p 29). Because of
the pattern of strong and weak beats, a minim displacement is to be considered the
crotchet displacement is more distant, the D for example would now fall on beats 2 or 4,
rather than 1 or 3, a more significant change in character. Any quaver displacement alters
the phrases yet more extremely, removing the upbeat and interfering with any swing
quaver characteristics that may exist. Semiquaver shifts, and yet finer rational
This particular axis is represented here and usually conceived as a series of discrete
points along an axis representing integer divisions of the beat. In Section 2 (Time-feel p
82-129), this notion is challenged with the conception of this axis as a continuum rather
21This non-linear nature of musical proximity is noted in On Sonic Art in reference to the relationship
between frequency and consonance (Wishart 1996, p 71-73).
22 The layout of these phrases is not static, once a rhythmic displacement has been made, phrases are
reordered in terms of proximity. For example if Phrase α is displaced by a crotchet, its ‘nearest’ neighbour
is now a minim away.
28
Figure 1.5.4 An illustration of Phrase α existing within a range of proximate phrases,
the ordering of which is determined by the extent of their musical transformation.
With this concept of musical proximity in mind, more dimensions may be added and
a new musical space may be constructed for exploration. Orthogonal to this rhythmic
placement axis (Figure 1.5.4), a note separation axis may also be postulated, representing the
progressive elongation and contraction of Phrase α, with wider note separation in one
direction, and shorter in the other.23 In Figure 1.5.5 this axis is arranged with the
23It is useful to differentiate the concepts of note separation and note length. Note separation describes the
rhythmic distance between adjacent notes as opposed to their individual durations. This allows, for
29
emphatic top D used as a rhythmic anchor about which the outer two notes are stretched
or compressed. The individual notes may compress until they form a chord and then
Figure 1.5.5 Phrase α with some of its musical neighbours arranged in terms of
increasing and decreasing note separation.
In addition to rhythmic placement and note separation another axis may be added that
example, some measure of sparsity to be made as well as the concept of a phrase compressing to form a
chord of any duration, which has practical improvisational implications.
30
Dorian, as opposed to the chromatic transposition in Figure 1.5.1 p 25), higher in one
direction and lower in the other. Chromatic transposition within a tonal harmony creates
a non-linear pattern of musical distance. However within a modal setting, from which
Phrase α is derived, the hierarchical nature of scale degrees is less clear. The subjective
decision has therefore been made to arrange the proximity in terms of diatonic
Figure 1.5.6 Phrase α transformed through diatonic transposition in the key of C Dorian.
Given the definition of these parameters, variations of Phrase α exist in three
dimensions, with potential mutations of the phrase existing side by side in conceptual
31
space. A sense of musical proximity within these constraints may also be perceived
(Figure 1.5.7).
Figure 1.5.7 Phrase α existing at the centre of a three-dimensional musical space with variously proximate
neighbouring phrases. Phrase α is indicated in red and the musical distance between it and various close
neighbours is shown in green. The boundary of the orange sphere describes a boundary of equal proximity,
and contains phrases within this musical distance. The lower part of Figure 1.5.7 shows an impression of
Phrase α existing at a point within this musical space.
Now that the concept of proximity has been established, one might also imagine
32
those only achievable through electronic manipulations), points of symmetry, intonation,
1.5.8.
Figure 1.5.8 An impression of M-space: Phrase α (circled in red) sits at the centre of many simultaneous
dimensions of musical transformation. Twelve of these are represented in four three-dimensional subsets
(some of which are continuous rather than discrete values) with some proximate phrases indicated. A
phrase may move along any number of such transformational axes during the course of improvisation. In
the top right of the diagram a phrase (circled in blue) shows the result of a small move in all of these
subsets simultaneously (the modification is marked as a blue disc in each transformational subset).
It becomes clear that most jazz analysis limits itself to variation of only a few
simple vocabulary choice while ignoring characteristics less easily notated, such as
rhythms and pitches that fall between the cracks of standard notation, timbral gestures
to illustrate precisely in one diagram (Figure 1.5.8 is an illustrative attempt using two and
three-dimensional subsets for clarity). Similarly, the representation of three genes (or
33
musical transformations) with linear expressions is easy, but once we add several more,
phrase existing within a radiated sphere of closely related musical material may readily be
As more axes of transformation are added, an idea is built of any particular musical
object living at a particular point in a conceptual space of all its possible variations. A
point within a grand Musical Library of Babel 25 from which the improviser may explore in
together, are shown later as cloud-like structures, but the actual shape is harder to grasp.
(or M-Sphere here), a multi-dimensional sphere, exists in many dimensions and has a
conceptually challenging shape to consider. It may extend far along a few axes if stable in
24Research by the author, beyond the scope of this thesis, is currently underway in the production of M-
Space computer modeling and real-time computer improvisations. The musical distance between phrases
Px and Py is calculated using Euclidean geometry over m musical dimensions, thus:
m
d(Px, Py ) = # (x " y )
i i
2
i=1
25 See the Library of Babel, (Borges 2000). In this short story, Jorge Luis Borges describes a library of similar
length books, housing every possible permutation of the alphabet and a short list of punctuation marks. So
the library contains mainly nonsensical books, among all versions of the Bible, Great Expectations and On
! the Origin of Species – all with every imaginable alteration of plot, protagonists, obscenity, profundity and
editing. It is alarming that books could span clearly labeled volumes, and alternative encyclopedias could
exist. Stranger still, despite the awesome size of the library, it would have to be finite, encompassing the
limits of our imaginations. In lectures at the Royal Academy of Music, the author has postulated an (also
finite) CD library of Babel, whereby every permutation of the bits of a 16-bit 44.1kHz stereo information
within a CD capacity are housed. These include all one’s favorite – and least favorite - records, arranged
for every conceivable ensemble, standard of playing, and nuance of performance – amongst of course a
catalog of inconceivable noise and extremities of expression. Students were invited to contemplate which
few of this vast number of CDs they wished to recreate during their recording careers.
34
others, and if many parameters change then they are relatively constrained. This concept
is illustrated in Figure 1.5.9 (p 36), three M-Spheres of different radii are illustrated, each
with Phrase α at their centres. The surface of each sphere represents phrases of equal
musical proximity (from Phrase α). Since these are multi-dimensional shapes, proximity
and distance may be distributed unevenly across many dimensions. For example the
boundary of M-Sphere A may be seen as one small change in one dimension, while B
may represent one moderate alteration or a few small changes. The surface of M-Sphere
C includes phrases with for instance, radical changes on one dimension, moderate
alterations on several and small changes on many. A phrase exists within this complex
radiated sphere of proximate musical material, and once a selection has been made,
35
Figure 1.5.9 Phrase α illustrated in the centre of three M-Spheres of differing radii. The surface of
each sphere links phrases of equal musical proximity. Each sphere represents distance across multiple
musical dimensions, and musical proximity does not have to be distributed evenly across all dimensions.
phrase, with a major alteration in a single dimension, such as a huge timbral modification,
may be equally proximate as a repetition of the phrase with many slight alterations.
Techno (and associated electronic dance music styles) illustrates this idea clearly, a
while remaining an intelligible relationship to the original phrase (For one of numerous
examples of this, see extract Flaphead (Aphex Twin 1992) with an audio extract on
CD1.5). Funk, with its relatively immobile chord structure and repetitive hypnotic nature
On the other hand, jazz, for the unaccustomed ear, may vary too much from one phrase
to the next, becoming unintelligible. “Without certain musical glues, it all sounded like
noodles” (The Real Frank Zappa Show 1989). A stability of many parameters may often be
uninteresting, but can also engender a greater sensitivity to subtle musical changes, as
26In-depth multidimensional ensemble time-feel analyses of James Brown’s rhythm section performances
have been conducted (Mermikides 2005), and in samba music (Naveda et. al. 2010).
36
may be said of the minimalist movement. This interrelationship of parameters and the
Reintroducing the improvisational process from section 1.3, Phrase A.1 (Figure 1.3.2
p 13) may be seen as existing at a point in the multidimensional space of all possible
mutations. The labeling of the relationship between continuing phrases now makes sense
in terms of M-Spheres or fields of proximity, for example A.1, A.1.1, A.1.2, A.1.3, A.2 and
A.3 all exist within the same field: they are sufficiently closer to be recognized as similar.
In this case the relationship can be seen through standard notation, but the definition of
M-Space includes a host of musical parameters that escape easy notation, which
Every executed phrase implies a field of related phrases – along many dimensions of
transformation. The precise demarcation of these fields is subjective and may cross-fade,
37
Figure 1.5.10 An M-Space representational analysis of the Swish solo (CD1.2) Phrases exist in fields of
proximity, which may overlap, and the improvisation is conceived as an exploration of multi-dimensional
musical space.
Given a particular starting phrase, one may imagine a field existing of appreciable
‘natural’ continuations of the phrase. The language of evolutionary biology returns, just
38
phrases from a radiated sphere of proximity.27 These fields contract over time as the
listener for some duration, whereas more extreme mutations would have to occur
biology is useful, but not entirely analogous to the process described here. In this model
phrases do not have to reference the immediately previous phrase, we may skip generations,
any phrase in the past is fair game (as if from a storehouse of recessive genes), and the
performer is allowed to restart a pathway from any point in M-Space at will. We can
thereby imagine a cumulative proximity caused by the repeated referencing, along varying
model is that the repetition of similar phrases will expand the field of proximity, meaning
that a new phrase has to be relatively more different to maintain a similar level of
novelty. For example, a one-note solo would require a high level of rhythmic and/or
A contour of proximity over time may be described through a solo as the distance
traveled in M-Space from one phrase to the next. The itinerary of this ‘flight path’ is
27Unlike the evolutionary process, the improviser is not constrained by the inconveniences of
reproduction or limits of genetic viability, she is free to skip to the ‘hopeful monster’ (Goldschmidt cited in
Dawkins, 1996, p 87-88) at any point in the performance.
28 Incorporating a time function into a mathematical calculation of a distance:
m
2
d(Px, Py ) = ( f (T (Px, Py))) * # (x " y )
i i
2
i=1
Where T(Px,Py) is the time difference between phrase events Px and Py, and f is a (linear or non-linear)
function simulating memory response.
For a humorous illustration of this type of principle in a compositional context, see CD1.6, Johnny One-
29
!
Note (Keneally 2007).
39
Figure 1.5.11 The Swish solo (CD1.2) illustrated as a trajectory through M-Space.
The musical distance of each leap in M-Space can be measured and tracked over
time. Close repetitions of phrases would appear on a flat line while radically changing
Figure 1.5.12 An impression of a proximity contour, tracking the novelty of each new phrase in Swish.
The height of the contour represents the distance travelled in M-Space, a high flat contour would for
example represent a steady fast velocity through musical space, while upward and downward countours
would correspond to M-Space acceleration and decelleration respectively.
40
This raises a question of the existence of an optimal ‘musical’ distance in M-Space
from one phrase to the next. The skill-set of the proficient improviser includes the ability
to control musical proximity for artistic effect. In The Sound of Surprise (Balliett 1959),
Balliett’s “aural elixir” is the result of perfectly selected surprises and, as Borgo notes
(Borgo 2005), effective improvisation is not a random stumble through musical space,
randomness are addressed later in this thesis. Regardless, the concept of proximity allows
vocabulary, a series of phrases with only subtle changes creates a different musical effect
than a series of wildly disparate phrases.30 In other words, M-Space distance and velocity are
Rather than improvisation as a meandering drunken walk through this M-Space, large
scale improvisational strategies are possible, and occur often in the hands of skilled
30For an example of an unbounded improvisation refer to an extract from Derek Bailey’s Sheffield Phantoms
(Bailey 1975) on CD1.9.
41
practitioners. Listening to Jimmy Smith’s solo on The Sermon (Smith 1958) (Figure 1.5.13
and CD1.7) with the M-Space model in mind, it is easy to hear the separation of phrase
fields.
Figure 1.5.13 A standard notation transcription of an extract of Jimmy Smith’s solo on The Sermon
(CD1.7) with some salient time-feel features notated.
A first listen sorts these phrases into five main fields (A-E) with 2-5 phrases in each
(A1-3, B1-4 etc.) These phrases and fields are labeled and coloured respectively in Figure
1.5.14.
42
Figure 1.5.14 The phrases and fields of Jimmy Smith’s solo on The Sermon (CD1.7) labelled.
There are common features within each group that form a strong gravitational force
(or ‘Zappa’s glue’) between the phrases. This proximity means that they can tolerate and
43
indeed draw attention to any subtle transformations including editing of notes and
detailed variations of time-feel and inflection. The creation of proximate phrases fixes
groups of musical dimensions and thereby frees up other musical dimensions for
effective expression. Note also that exact repetitions of phrases in terms of notes still
include alterations in harmonic altitude, given their context, and small but effective
rhythmic and micro-timing variations. Figures 1.5.15-19 (p 45-49) illustrate the grouping
of these phrases within fields, with descriptions and illustrations of salient differences
between members of the same field. The diagrams to the right of each figure give an
44
Figure 1.5.15 Phrase field A: Phrase A1-3 share a very similar melodic structure and vary in terms of
placement of component notes, time-feel and harmonic altitude. An impression of relative distance in M-
Space is shown in the lower diagram.
45
Figure 1.5.16 Phrase field B: Phrase B3 differs from B1, B2 and B3 in its use of diatonic
transposition, yet, in its relationship to the anticipated Bb7 chord, maintains a harmonic altitude proximity
with B2 and B4 hence its illustrated position in the lower diagram.
46
Figure 1.5.17 Phrase field C and its component phrases. Phrase C3 is illustrated slightly closer to C1,
than C2 due to the dotted crotchet, as opposed to quaver, rhythmic displacement of the first 3 notes.
47
Figure 1.5.18 In Phrase field D the core G, D, G motif is kept constant but a changing upbeat
phrase, rhythmic placement, time-feel, harmonic altitude and articulation separate its component phrases.
48
Figure 1.5.19 Phrase E1 and E2 are loosely linked by a general melodic shape, use of chromatic
approach notes and expressive mechanism of falling behind the beat. The length of Phrase E2 and use of
transformations of similar low-level motives lends itself to further separation into smaller phrase units. The
proximal relationship of Phrase E2.1-E2.11 are represented in the right-hand diagram and provide the
vision of hierarchical structures in M-Space.
49
The most loosely connected group is Field E, the two members of which share
the same contour, use of ornaments and rhythmic expression derived from a progressive
falling behind the beat. Phrase E2 is also distantly linked to fields A and C with the use of
apparent that fields themselves exist in a nexus of proximal relationships with each other.
As was shown in reference to E2, phrases may also be conceived as housing similar
illustrated in Figure 1.5.20 (p 51). Fields A-E co-exist as part of the same solo, but their
relative proximity is also due to registral, timbral as well as temporal considerations. Not
shown in Figure 1.5.20 is the yet more complex interaction of fields and phrases between
other performers and musical objects. This is explored variously in Section 3 and Scores
and Notes.
Although Figure 1.5.20 is illustrated in two dimensions, one must be reminded that
the relative distances between fields, and between their constituent phrases, is the
many co-existing musical parameters). Once the concept of M-Space structures and their
relative positions has been grasped, the listening and analytical process becomes far
clearer. From the straight-ahead to the most avant-garde contexts, it becomes possible to
50
Figure 1.5.20 A multi-level depiction of The Sermon. Improvisation is seen as a configuration of fields
at varying distances and trajectories in M-Space, with each field containing a constellation of phrases.
Phrases, in turn, may be broken down into a nexus of smaller phrase units as is shown in reference to E2.
Phrase E2 has been placed closer to Fields A and C than B and D, to reflect features of E2.1 as discussed
on page 49. Fields themselves are linked together in terms of timbral, registral and temporal components
and may co-exist in a yet greater nexus of relationships with other performers or musical objects.
One can hear, for example, in Wes Montgomery’s solo on No Blues (Montgomery
1965, CD1.8), one very narrow phrase field being used repeatedly as a pivot to other
phrases are coloured accordingly and a pivoting pattern emerges so quickly and clearly
that the ensemble are compelled to share the motif. The pattern settles into a two-field
(blue and green) interplay before relinquishing the structure. The repetition of the blue
phrase field is so clearly defined that the other ensemble members are compelled to mark
it with their own musical material, in other words, the soloists M-Space structures has
environment.
51
Figure 1.5.21 A field illustration of an extract of Wes Montgomery’s solo on No Blues. (CD1.8)
Various phrases pivot around one very tight field (labelled in blue) and settle into a two-field exchange
(Notation transcription by Jeremy Poparad).
identifying a narrow field and then furtively exploring that space for extended periods
before moving to a new locale and repeating the process. Pat Metheny’s approach on
Unquity Road (Metheny 1976) (Extract on CD1.9) is less clearly delineated: Phrase fields
exist, but the transitions between them are often blurred, and referenced interchangeably
52
Figure 1.5.22 Pat Metheny’s solo on Unquity Road (CD1.8) merges and switches between phrase fields
fluidly. Fields have been identified by colour and smooth transitions illustrated by a cross-fade between the
relevant colours.
(Notation transcription by Jeremy Poparad)
passages into broad categories. These are grouped in terms of the relationships between
the phrases rather than the vocabulary itself. A pictorial comparison of five
improvisational strategies is presented in Figure 1.5.23 (p 54): nuclear, field series, pivot,
merged and unbounded of which Acknowledgement (CD1.1), The Sermon (CD1.7), No Blues
(CD1.8), Unquity Road (CD1.9) and Sheffield Phantoms (Bailey 1975) (CD1.10) are
53
respective examples. The categorisation of these strategies involves some subjectivity
(one man’s nuclear, may be another man’s unbounded improvisation) and there may be
borderline cases, but a clear terminology and framework in which to analyse, compare
Figure 1.5.23 Five improvisational structures: 1) Nuclear: phrases, with only occasional small
anomalies, fall within one close field with only minor variances (CD1.1) 2) Field Series: close phrases are
played a few times with variances before repeating the process at a different point in M-Space (CD1.7) 3)
Pivot: one particular narrow field is played often, acting as a springboard to various satellite fields (CD1.8)
4) Merged: fields are merged by the use of a transitional phrase of otherwise distinct phrase fields (CD1.9) and
5) Unbounded a series of phrases with little proximity of one phrase to any other (CD1.10).
interaction with human performers, is explored extensively in the Changes Over Time
portfolio.
54
Through a practice-based and pedagogical exploration of jazz improvisational
method this section has crossed paths with an elaboration of Pressing’s event-cluster model
(Pressing 1988). This meta-view of improvisation is made more powerful with the
proximity and velocity. The implications of a time element applied to this model is
55
1.6 Changes and Time: Expressive Contours
The previous section identified how subsequent phrases might develop in the course of
This holds analytical and practical power in many contexts, but there are some
limitations. M-Space modeling does not provide the clearest picture of how particular
parameters evolve over time, both within a phrase and through multiple phrases, nor
does it most readily show the interactions between parameters. Furthermore, as can be
1.6.1 (p 57), M-Space analysis can break down when phrase boundaries become too
merged or ambiguous. The phrases in the first seven bars may be seen as the toying
between two fields, a long note with a semiquaver offbeat kick and a diatonic descending
phrase. They are linked by a fragile timbre, made more ethereal with a subtle whammy-
bar vibrato. From bar 7 onwards however, the rhythmic density and phrase length
56
Figure 1.6.1 Transcription of Holdsworth’s solo on The Un-merry Go-round (CD1.11) with harmonic context.
There are however valuable expressive mechanisms at work in this passage that
should not escape analysis. It is clear that the expression inherent in this section does not
come from proximal relationships between clearly defined phrases, but from the
varying rhythmic density and melodic register that, together with an arresting tone,
combine with remarkable effect. Figure 1.6.2 (p 58) tracks melodic register (in red) and
rhythmic density (in green with a measure of notes per beat in the vertical axis) against
time. This analysis gives insight, which may escape conscious perception, into the
emotive power of the phrase. Melodic register tends to be positively correlated with
rhythmic density; however, this correlation is broken in bars 5, 11 and 14 where the line
57
lingers on medium-high register notes, and the expressive contours separate. Of
particular interest is bar 11 where the density plummets to reveal a slow microtonal glide
in melodic register. Despite its brevity, this passage lays the foundation for extensive
practical research in the effect of various linear and non-linear correlations, and
Figure 1.6.2 Holdsworth’s Un-merry Go-Round bars 5-14 (CD1.11): Melodic register and rhythmic density
(measured in notes per beat) tracked over time. Breaks in the generally positive correlation between the
two parameters occurs in Bars 5, 11 and 14.
This approach, of mapping particular parameters against time, can allow a vision of
the overall shape of a solo, or composition to emerge, as well as draw attention to the
interaction between parameters. It also lends itself readily to pedagogic practice, real-time
between phrases becomes impractical. This analysis can be seen as slicing M-Space into
individual dimensions, and tracking motion across these component layers over time as is
58
Figure 1.6.3 Illustration of a multi-dimensional M-Space view of an improvisation collapsed into its
multiple component parameters tracked over time.
In order to dig deeper still into the complex interplay between musical parameters,
an extensive analysis of a short active phrase is required. Phrase β (CD1.12) taken from
Standard Deviations (Mermikides 2008) is transcribed in Figure 1.6.4. Though short, this
complex and interactive contours available for expressive manipulation and analytical
consideration. Not all of these approaches will be relevant in every context, but an in-
depth analysis is taken to present a survey of the contours, and the subsequently derived
meta-contours.
59
Section 1.5 (p 25-55) put forward a conceptual model of grouped phrases within
overlapping phrase fields. A possible analysis of Phrase β is given below. This layout gives
revisited with modification over the course of the improvisation (Figure 1.6.5).
Observing one particular parameter, for example the pitch height (a lattice approach is
taken: ignoring microtones and micro-rhythms), and tracking it over time, presents a
This type of analysis with respect to melody has been researched in the field of jazz,
and other styles.31 In respect to jazz solos, the term tension and release curve is often used to
describe the shape of a particular parameter over time (Berliner 1993, p 57). Since the
some parameters, the author suggests the term expressive contour to describe the
31See melodic contours of jazz phrases (Coker 1987, p 57-61), jazz melodies (Goldstein 1993, p 13-14) and
folk music melodic contour typology (Adams 1976). The reversal of this relationship, whereby prescribed
contours (often derived from non-musical sources) are attached to musical parameters, is explored in
Changes Over Time (See Primal Sound, Head Music and Blood Lines), and with a historical precedent (See Villa
Lobos’ New York Skyline Melody (Frey 2010)).
60
modulation of a particular musical characteristic over time32. Opportunities for musical
expression exist along many parameters, and combinations thereof. For example,
tracking the intonation of the phrase, relative to an equal temperament, creates another
expressive contour. Note how this contour captures vibrato and bends and is
represented as a cyclical scale, wraps around vertically as a note passes the midway point
The dynamic contour, representing the volume (of each onset) of an improvisation
32Wishart’s gestural contours represent a similar idea of parameter control in the context of electronic music
(Wishart 1996, p 109-125).
61
Rhythmic density, the rhythmic distance between attack points may also be tracked
as before. It becomes clear that an expressive gesture may be formed by the change of a
Harmonic altitude may be defined as the chromaticism of the phrase relative to the
harmonic context. This is a complex field, with hosts of interacting factors (Liebman
1991). In this illustration, a simple measure is employed relative to the G7 context, this
runs from chord tones (CTs) (root, 3rd, 5th, minor 7th), to harmonic extensions (HEs) (9th,
11th, 13th) to common non-harmonic extensions (CNEs) (#9, #11) and the uncommon
62
A medium of expression is also possible in the varied employment of intervals.
Taking a measure of absolute intervals gives an idea of the predominant intervallic leaps
observation of a proximity contour - the extent to which a phrase is altered through the
solo. An impression is given in Figure 1.6.12 of how far one phrase moves to another in
M-Space. As new motivic material is introduced the contour rises and falls as this
These contours do not function entirely independently but exist in various correlations.
These relationships fall into three general categories: 1) Independent: contours that may
63
move freely against one another with no cross-effect (e.g. rhythmic density and harmonic
and melodic range) 4) A non-linear relationship between contours (e.g. melodic contour
Figure 1.6.13 Four types of relationship between contours: 1) independent 2) linear 3) exponential 4) non-
linear.
the effect of multiple parameter changes to be understood more clearly, and in turn,
There exists a complex resultant musical effect of these many consychronous expressive
64
contours may be conjectured. Examples of meta-contours might include 1) Activity: derived
from a combined measure of the contours of rhythmic density, intervallic change (a first
order differential of interval with respect to time) and proximity. 2) Intensity: determined
these meta-contours, an impression of these may be calculated (Figures 1.6.14 and 1.6.15,
p 65-66).
Figure 1.6.14 Activity contour, derived from an additive function of rhythmic density,
intervallic and M-Space proximity contours.
65
Figure 1.6.14 Intensity contour, derived from an additive function of dynamic, pitch and harmonic altitude.
In Hal Crook’s How To Improvise (Crook 1991, p143-5) some “solo curves” are
contours. So at a constant meta-contour level of, for example, intensity, a drop in one
66
expressive contour (e.g. dynamics) must be compensated for by a total increase in
parameters, is given attention in the submitted portfolio and referred to as slack theory.33 A
meta-contour of significant interest and research potential, developed by Rolf Bader and
David Borgo (Borgo 2005, p 92-121), is the “fractal correlation dimension”, which is a
frequencies and amplitude modulations. This hugely involved analytical process, taking
weeks of computer processing for each piece and including up to eighty transformative
and aligns quite convincingly with the subjective listening process. However clever, no
one contour can capture the pluralistic structures of music, but directed attention to a
salient set of contours can help explain the emotive effect of mechanism specific to each
musical context.
how a series of seemingly meandering phrases can create larger scale musical structures
or gestures, the shape of a solo. These larger scale structures are certainly appreciable to a
tuned-in audience, and the candid practitioner34, but without the vocabulary of expressive
ways:
2) the musical effects of particular contour types, drawn from eclectic sources, on
33The observation of expressive contours is also found in the dramatic arts including script-writing
(McKee 1999) and Zeami’s theory of Jo-ha-kuy in Japanese Noh theatre (Rimer & Yamazaki 1984).
34 See Composing in the Moment (Berliner 1993, p 192-220).
67
experience, 4) the construction of meta-contours by the manipulation of individual musical
studio composition and 6) the exploration of slack theory in the differentiated control of
Expressive contours may be seen simply as the introduction of the time component
into M-Space modeling35, but they also serve to bypass the difficulties in subjectively
identifying the proximity, and the demarcation, of phrases in general. Phrase demarcation
is by its very nature a form of post-hoc analysis; one must wait for a phrase end in order
M-Space moves during its execution. This is certainly the case for the listener and is
often true for the improviser.36 It is important to remember that, as is the case with all
experience, as both a performer and listener, one becomes more skilled at choosing the
most useful approach in any given situation, for example, an M-Space model is particular
illuminating in The Sermon (Figure 1.5.14, p 43 and CD1.7) but is inferior to a study of
contours in the case of the Un-merry Go-round (Figure 1.6.1, p 57 and CD1.11).
which of the many features deserve the most attention, and which are the most salient in
terms of our musical experience. As ever, good analysis should serve to illuminate, rather
35One could add time occurrence as just another dimension to M-Space, but the results are somewhat
unintuitive, difficult to visualize and implement.
36 The relationship between M-Space and expressive contours is illustrated in Figure 1.7.2 (p 73).
68
1.7 Applications of M-Space and Expressive Contours
in the author’s work in a variety of ways. These concepts have been developed through a
continuing cyclical pattern of practice, reflection, analysis, theory and reapplication and
Improvisational Strategies
for performance. The training and skill in order to control effectively this walk through
M-Space (via the limiting, and focusing of dimensions) is considerable and a never-
ending pursuit. In the model of improvisation given above jazz skill can be seen as the
created elements to create expressive gestures through certain parameters while keeping
others unaffected. ‘Making the changes’ (proficient negotiation of jazz harmony) for
example, may be seen as altering melodic and rhythmic content while maintaining
control over harmonic altitude and melodic range. Equally, a lack of improvisational
dynamic or harmonic control. Hence the incentive for the specific exercises of Tomassi,
Crook and Damian. This view acts as a framework not only to practice improvisation but
also a way to evaluate and appreciate the repertoire. One is given the ability to describe
cultural and individual styles as fields among various dimensions: The skills required in
the bebop heritage involves the fixing of rhythmic density (usually quavers) and a specific
realm of time-feel and timbre, while negotiating complex harmonies with mainly step-
69
wise and arpeggiated lines. BB King’s skilful variation of rhythmic placing, phrase length
and vibrato, despite a limited note choice and rhythmic density, is characteristic of his
style (for example Payin’ The Cost To Be The Boss King 2006, CD1.13). Wayne Krantz’s
playing on Is Something I Don’t Understand Yet (Krantz 1995, CD1.14) may be seen as the
limiting of motivic material (intervallic structures) and creating great interest in variation
through transposition, phrase length, metric placement and time-feel control) In short,
an awareness, and ability to identify, the limitating and variation of a range of topics fuels
Reaching Out
Music technology allows the improviser to extend her reach across several otherwise
the traditional saxophonist has stretched to the widest extent of the instrument’s timbral
spectrum, electronic effects expand hugely the available horizon (See CD1.15, an excerpt
The use of delay tools in improvisations allow the capturing, looping and warping of
live input to form a textural background to improvisations (See CD1.16, an excerpt from
Torus (Mermikides 2008) - a solo electric guitar (and electronics) improvisation, featuring
This can be seen as a technological smearing of motivic phrases, where a motif is frozen in
time and stretched and manipulated. Performers are no longer limited by a serial
procession of phrases, they may coexist in a parallel with newly initiated phrases and
form a quasi-ensemble relationship (See Omnia 5:58 (CD2.17) in the submitted portfolio).
70
Guided Improvisation, Flight Paths and Stripping Contours
Awareness of the multiple paths by which an improvisation may travel allows for
instructions such as ‘Explore timbre’ in String Theory (CD2.15), the harmonic pathways in
Event Horizon (CD2.16) or the modal specifications in Koshinokawa (CD2.9). Figure 1.7.1
shows an excerpt from Event Horizon where the intention is to allow the soloist to
improvise spontaneous long arching phrases. To facilitate the navigation through the
complex harmony, without having to resort to excessive preparation that might mar
spontaneity, a solution was devised. To avoid the undesirable ‘running of scales’ that
often results from improvised passages with non jazz-trained performers, harmonic
pathways are indicated in the score. The performer is allowed free reign to choose notes
from the provided scales within each harmonic context. However, when transitioning
from one chord to the next, smooth voice-leading and an unbroken line is required. To
achieve this each note is given one of three symbols to indicate whether it is sustained
into the next harmony, or altered by a semi-tone and in which direction. The technique
works extremely well, and provides a large range of paths through the chord
progressions without prescribing any particular one, and allows the soloist to form large
scale melodies without an extensive jazz training. This mechanism can be seen as the
maintaining of harmonic altitude, melodic register and phrase length through complex
harmonies.
Figure 1.7.1 Harmonic pathways in Event Horizon indicating methods for negotiating complex harmonies
while maintaining long phrases and melodic register. Symbols above each note indicate voice-leading
strategies via semitone ascent, descent or maintaining the note.
71
In other works more responsibility is offered to the soloist with greater jazz
experience. In Eleventh Light (CD2.8) the performer is given key locations in M-Space to
which the improviser must travel via a field series or merged improvisational pathway; the
soloist is given an itinerary through M-Space with fixed destinations but no specified
flight path. Intentional guidance of an improviser (from various disciplines) by the use of
instructions of which parameters should be fixed, and which must be varied, appear
throughout the portfolio and allow a work to retain an authentic spontaneity while
The use of electronics can expand greatly improvisational concepts. For example, in
I is a Robot (CD9.5), electronics are used to strip pitch and timbre information from a live
improvisation, replacing them with randomised timbres and pitches. This can be seen as
the stripping apart and reattaching of expressive contours, in this case, rhythmic elements
are hijacked from an improvisation and glued to another field in M-Space. String Theory
(CD2.15), on the other hand, uses the physical gestures of the cellist’s Hyperbow and
reattaches these to musical parameters that effect the timbre of the cello as well as other
The concept of M-Space may also be employed compositionally and not in the
including hierarchical phrase structures (as in Figure 1.5.20, p 51), spatial placement,
rhythmic elements, vertical stacking, form and time-feel. This may be seen as a
M-Space dimensions.
72
M-Space Modeling
arrives at the next phrase (P2) at some variable distance. The process is then repeated
along another (varying or similar) set of parameters to arrive at the next destination (P3).
A phrase may reference any previous phrase not just the most recent (P4 may reference
phrases have occurred in the past, albeit with a fading memory (Figure 1.7.2).
Figure 1.7.2 The relationship between M-Space, expressive contours and time.
A series of phrases over time will build up a nexus of proximal relationships - which
may interweave with the nexuses of other ensemble members - and carve differentiated
73
In some compositions there is an effort to provide glimpses of these separate
potential pathways of a phrase and their relationship to the memory of the seed phrase.
An extreme example of this is Omnia 5:58 (CD2.17) where the various effects on each
individual string, extremely spatialized in the hall, allow the listener to experience quasi-
improvised ideas radiating from one phrase. In other words, the electronics add a
fragmented multi-dimensional ‘glow’ to the musical object (Figure 1.7.3 p 75). As the
piece progresses, the fractured glimpses of past phrases are heard, heading through
independent pitch, timbral and spatial pathways. There is flexibility in the score so that
the performer can respond to these reflected musical strands, in the manner of Figure
1.7.2 (p 75), which in turn affects the character of the ensuing sections of the work.
Shutter Speed (CD2.21) takes a similar approach where one simple phrase is, with the
automated.
74
Figure 1.7.3 Each of the five cello strings in Omnia 5:58 (CD2.17) are sent along separate trajectories in M-
Space, a graphical representation is shown above a standard notational transcription of the effect of a
single chord.
75
Mapping
The Changes Over Time portfolio includes works that take the concept of an
and reverse engineers the process. Compositions are created from electronic systems that
use physical phenomena as source material to drive expressive contours and M-Space
work, many of which are inspired by scientific phenomena, and also to explore the
many musical decisions, and can listener to his own work as an impartial listener. Three
examples are shown below. Figure 1.7.4 (p 77) shows images from Primal Sound (CD2.1),
in which a contour – derived from the coronal suture of a human skull – is mapped onto
co-existing expressive contours. Event Horizon (CD2.16) employs the use of real-time
mapping in performance; the data from the Hyperbow is linked electronically to many
consequence, sculpt the various musical layers via the seven sensors on the bow. A
composition is created via the input of colour information, and DNA coding from
are mapped onto three parameters subsets in M-Space to form a complex, interactive and
emergent work that responds synaesthetically with the imagery (Figure 1.7.6, p 78).
76
Figure 1.7.4 Mappings in Primal Sound: the coronal suture (centre) is mapped to (clockwise from top
right) 1) amplitude against time, frequency against time in vertical (2) and horizontal (3) configurations and
as a control of event triggers and pan position.
Figure 1.7.5 Event Horizon physical gestures of the Hyperbow are linked to consynchronous musical
parameters, allowing the soloist to control various aspects of otherwise independent musical layer.
77
Figure 1.7.6 Microcosmos employs the mapping of colour and DNA data to three-dimensional musical
subsets.
The works in Changes Over Time which, as briefly discussed, range from ‘standard’
Space modeling and mapping, are covered in detail in Changes Over Time: Practice.
78
1.8 Reflection
This paper offers a broad framework in which to practise, develop and analyse
improvisation. This more complete view of improvisational mechanics also allows the
varied and subtle direction, through score notation or rehearsal, of guided improvisation
notation, where a performer is given a set of notes with which to play – but no other
‘learn your scales and good luck’ approach of much jazz pedagogy. The concepts
discussed in this thesis are amenable to a range of stylistic contexts and help guide
effective and compelling performance without the obligation to always employ the
overtly complex harmony and rhythmic devices often associated with contemporary jazz.
Harmony and rhythmic density represent just a fraction of all potential musical
time-feel, timbre, articulation, and the interaction between these parameters are as valid,
even if the scale material is pentatonic, and a standard notational approximation appears
simple. The concept of M-Space and expressive contours can explain the effectiveness of
evolution, simply extending the limits of various parameters, allowing direct control of
multiple expressive contours and the real-time capture and manipulation of musical
79
material. Music technology also allows the deployment of these improvisational
daunting, then the meta-view of improvisation presented in this paper can be utterly
overwhelming. How is one to begin formulating and constructing pieces using the
concept of M-Space, with its Babelian complexity? The author has found a solution in
technique to the compositional process. Can a piece be constructed with only one
specified expressive contour applied to many multiple parameters?37 What is the effect of
a group improvisation where electronics randomize pitch and timbral content, leaving
only rhythmic elements under the performers’ control?38 Can a satisfying chord sequence
contours?39 What is the result of mapping downward physical force of the bow to the
cello’s cut-off frequency, colour values and DNA codes to timbre, or blood cell
material, can a systematic cycle of the topic of improvisation form the structure of the
piece? Can geometrically derived coordinates and trajectories in M-Space form musically
form the melodic curve in a stylistically distant context? This type of targeted approach
practical experience, developed intuition and critical reflection, and is the case with the
Berklee improvisational exercises (see 1.3 p 11), the strictures imposed rarely inhibit the
potential musicality.
39Selfish Theme (2.7) employs a simple algorithm to derive a cyclical motion for outer two voices of a chord
progression, compositional freedom is allowed only in the inner voice writing.
40 See Event Horizon (CD92.16), Microcosmos (CD2.4) and Blood Lines (CD2.3) respectively.
80
The view of improvisation as transformations in multi-dimensional musical space is
so broad that it connects the mechanics and pedagogy of jazz practice with a diverse
modelling (Xenakis 2001) which, with its consideration of stochastic functions over
multiple – albeit discretely valued - musical parameters, has parallels with the concept of
19, 55), related directly to M-Space. 4) Moles’ and Schaeffer’s prescient graphical
representations of l’objet sonore (Holmes 2008, p 45-48) are conceived readily in respect to
employing isokinetos, isorhytmos and isologos, and 6) Dreyfus’ detailed studies of motivic
transformation in the music of J.S. Bach (Dreyfus 1996) are readily adopted in terms of a
the intuitive and idiomatic discipline of jazz practice, allows a wide range of
creative ethic.
parameters has been established. At this point, the discussion turns to one particular set
and rarely well understood. The next paper models various micro-rhythms that are
integral to the expression inherent in jazz, and are here collectively termed time-feel.
81
2. Time-feel
Abstract
This paper presents, with reference to current pedagogical and analytical approaches, a simple,
usable model of expressive micro-timing in jazz and contemporary popular music, variously referred to as
‘swing’, ‘groove’ or ‘feel’ and here collectively termed ‘time-feel’. Central to the model is the conceptual
separation of the mechanisms of swing (offset of the second quaver) from latency (the sub-notational
are suggested that create useful comparisons of stylistic and performer-based variations, as well as how
time-feel may be dynamically controlled during an improvisation as an expressive medium in its own
right. A formal mathematical model is shown that may be employed with great flexibility in the analysis
This paper is concerned with aspects of rhythmic feel in jazz, and other
contemporary musical styles (funk, soul, rock etc.). It aims to clarify salient characteristics
of rhythm that ‘fall between the cracks’ of the standard notational system and general
important aspect of jazz. By way of an introduction to the field, this paper focuses much
of its attention to the ‘jazz quaver’42 and offers a method of analysis that most readily
41Much insight has been gained from the author’s experience as a jazz performer, composer and in the
production and editing of a large range of contemporary and popular music projects.
42In the appendix of case studies, time-feel aspects of the semiquaver, the funk sixteenth note, are also
explored.
82
relates to the experience and skills of the contemporary performer, and the listener’s
experience.
There is a lack of clarity in the field that is likely due to no generally accepted
terminology and little consensus of opinion. Whereas the harmonic and melodic aspects
of the jazz language have been well documented and are easily accessible, specific studies
of jazz rhythm are relatively scarce and late to the body of research. Jazz musicians have
tended to only talk of the subject in vague terms, or not at all. When asked in a
masterclass at the Berklee College of Music (Boston, MA) about the influence of
Brazilian music, and the subsequent ‘straightening’ of the quavers in his music, Horace
Silver responded flatly “I don’t think about that stuff” (Silver 1994).
jazz skill:
83
Indeed, it seems that with the odd exception43 the greatest practitioners and those
most qualified to explain the subject are often either unwilling or unable to do so,
preferring to encourage learning through imitation, and using qualitative rather than
In short, it seems that those that fully understand time-feel won’t explain it; and
those that potentially could, don’t fully understand it. Part of the issue is that time-feel
characteristics do not lend themselves to standard notation, and are thus less amenable to
Wishart:
43See Crook (1991), Moore (1995), Mingus cited in Berliner (1996), and Govan (2010) for rare examples of
useful analytical discussion of time-feel in a pedagogical context.
84
Wishart goes on to suggest that aural rhythm allows the performer the “most
rhythm.” In contrast, the summative, ‘finistic’ and “economy” of notation loses these
Wishart’s contention, that the expressive complexities rhythm, and pitch are – at
points. Adding the simplified dimension of discrete timbre (as each instrument), Wishart
Figure 2.1.1 Wishart’s lattice. The representation of standard notation’s limitation as a lattice, due to
the ‘notational economy’ of ‘finistic’ division of pitch, timbre and rhythmic subdivision (Wishart 1996, p
26).
So when jazz, and other relevant music styles are approached through standard
notation alone, we arrive at a distilled, ‘lattice-filtered’ strain that has stripped out very
85
It becomes clear that the aspects of time-feel presented here are preserved through a
heuristic exploration of the musical style in question rather than through jazz scores. It is
learnt by ear rather than by eye. The advent of digital audio technology has allowed a far
deeper exploration into this field, as timing may now be accurately measured beyond the
How might it be possible to formulate a system for measuring deviation from the
‘Wishart lattice’ in a coherent and relevant way? In the recording studio, on what basis is
it chosen which way, and how far, to nudge a particular rhythmic placement, or which of
many alternate takes make the cut? Is it possible to form some kind of system that tallies
with the feel of a performance that sits ‘in the pocket’? How can a musician who does
not play metronomically be described as having ‘good’ time? The author’s experience of
performing, listening, composing and editing jazz and popular contemporary music has
allowed the creation of a model that addresses the salient features of these ‘deviations’ in
a simple and usable manner, so that the lattice is transformed into a continuum of
expression.44 Here follows a survey of related research in this field, and how the model
presented here differs from, and adds to, other approaches to the field.
The complexity of jazz rhythmic mechanisms that occur beyond the standard
notation lattice, often termed micro-timing, has inspired diverse analytical methodologies
and pedagogical approaches. Stylistically distant research into notes inégales, ritartendo,
classical ensemble asynchrony and time warping in computer music (Fuller 1980, Desain
& Honing 1990, Gabriellson 1988 and Dannenberg 1997 respectively) also hold
44The author has also conducted research in the continuum of pitch (Mermikides 2007) and timbre
(Mermikides 2006) which together with time-feel have built towards a concept of multi-dimensional
continuous musical space (See 1 M-Space and Expressive Contours).
86
relevance to the nuanced field. Enquiries into jazz micro-timing may be broadly
heuristic responses to the experience of hearing and performing jazz rhythm (Schuller
1968, Werner 1996, Sudnow 1995). 2) Pedagogical material to encourage and guide
exploration of sub-notational timing in jazz practice (Crook 1991, 1996, 1999 and Moore
1995). 3) Audio analysis, measurement and music software applications of swing 8ths (or
jazz quavers) (Cholakis 1995, Friberg & Sundström 2002 and Benadon 2006). 4) Tempo
of differing simultaneous tempi (Prögler 1995, Ashley 2002, Collier & Collier 2002, Folio
& Weisberg 2006 and Benadon 2009) and 5) Studies of ensemble synchrony, the internal
placement (Cholakis 1992, Millward 2001a, Millward 2001b, Tait 1995, Butterfield 2006,
Gouyon 2007, Hennessy 2009 and Naveda, Gouyon, Guedes & Leman 2010). Within
this context, this paper provides a model born of a practitioner’s experience, with
analytical utility and creative applicability. With the conceptual separation (and
individual performer’s musical material) from latency (the relative temporal placement of
an individual’s musical material against a relatively rigid mutually negotiated master time-
line) many of the mechanisms within the existing field of research are covered within one
relatively simple framework. The inclusion of weighting in the model serves as a useful first
step into the understanding dynamics and articulation play in the experience of time-feel,
towards which much valuable research is required. Important musical mechanisms such
differential elasticity, all emerge naturally from the model and are presented in 3 Case
Studies. The model has authenticity from a practitioner’s perspective and musical
87
intention as well as in the visceral experience of the listener. A purpose-built software
application is also presented which serves as a real-time pedagogical and analytical tool.
The Changes Over Time portfolio demonstrates the variously intuitive and conceptual
compositional contexts.
88
2.3 SLW Model of Time-feel
However, there exist clear identifiable features of expressive microtiming that can aid
The complex subject of time-feel may be approached from many angles, the model
presented here determines the salient features of time-feel as three conceptually distinct
between two camps. One group offers the theoretical simplification of the jazz quavers
as ‘triplet quavers.’ The other group, mainly jazz musicians, are aware of the range of
subtleties in the jazz quaver, and can readily recognize and imitate deviations of quaver
placement. However, familiarity with the subtleties of time-feel has not necessarily led to
loosely descriptive terms will describe particular characteristics of time-feel: ‘lazy eighths’,
‘light swing’, ‘pushing’ and so on. These terms (by no means standardised) are clearly
well-defined within the perceptions of individual experienced jazz musicians, but the
89
time-feel such as latency, or simply as a qualitative measure. The time-feel aspect of
contemporary music is conventionally tacitly absorbed, rather than explicitly taught. Its
salient properties are inherent within the explicit melodic material of the language, and
learned through the imitation of other player’s styles. The explicit melodic and harmonic
language of jazz obfuscates and mystifies this hugely important aspect of the idiom.
Note that although swing is often used to define time-feel in general, the SLW model
describes it as only the offset of the second quaver (or relevant metric subdivision, which
will be called the offbeat). Swing values exists on a continuum that includes straight 8ths
(50%)45, the traditional notion of jazz quavers (66.6° %), dotted eighths (75%) as well as
numerous points in between. In Figure 2.3.2 (p 91), this continuum is illustrated along
with some descriptions of degrees of swing have been added, but there is no
terminological consensus of opinion for such nuance and they are unlikely to hold over
all tempi and stylistic contexts. They are included here as an indicative guide and
45 Swing values below 50% exist, but tend to occur via inevitable fluctuations in performance (random
noise). It seems that once a swing value drops significantly and repeatedly below 50%, it is conceived as the
first triplet quaver (33. 6̄ % ‘scotch snap’) or the first semi-quaver (25%). In contrast, under some
conditions the concept of the quaver can still be maintained even with repeated swing values close to 75%
(See 2.10 p 117).
90
Figure 2.3.2 The continuum of swing points. Discrete nodes, such as triplets and
quavers are indicated with solid lines, and 5% differentials in dotted lines. The concept of jazz quavers falls
between the 50-75% range, within which some general and subjective descriptions given.
there are some exceptions: Hal Crook (Crook 1995, p52) suggests an exercise of playing
a repeated quaver on one note while gradually moving from straight (50%) to triplet
(66.6%) quavers, albeit with no scenic stop off points. The drummer Stanton Moore
(Moore 1995) describes and demonstrates examples of swing levels in various drum
styles but does not assign explicit values, describing his approach as “swimming” in
between swing values (Moore 1995 23:21-35). Guthrie Govan, a virtuoso of many
electric guitar styles, speaks clearly about the concept of swing levels as a fundamental
playing, Govan recommends that guitarists learn by playing along to records, rather than
going on between beat one and beat two”. In this listening process, Govan claims he was
91
exposed to “all these different levels of swing”, and goes on to proficiently demonstrate
examples of straight, light swing, and hard swing playing. Swing, he says, is not an “on-
off” but a “dimmer” switch, having varying levels and is something that you can only
learn by “enjoying it first, hearing it and responding to it” (Govan 2010, 5:08 - 5:42).
Traditional standard notation has dealt with the issue of jazz quavers by a blunt
(66.6%). On other occasions they are explicitly written as dotted quaver followed by a
sixteenth note, this is most likely due to notational economy as this hardness of swing
(75%) is unlikely to be desired in the ‘show-tune’ setting where this practice is commonly
In jazz pedagogical material however, these simplistic guides are rarely given, there
may be a few words in the introduction (Coker 1964, Aebersold & Slone 1978) but in
general jazz quavers are written as quavers and their stylistic execution is expected
The advent of MIDI, sequencing and computer notation brought the inadequacy of
computer performance and quantization are not without substance, but the underlying
inadequacy is in fact with a musician’s excessive attachment to the score, and more
fundamentally, a simplistic concept of rhythm. Computers after all, only follow human
92
incapable of independent interpretation, that has encouraged a better understanding and
to a predefined rhythmic nodes, started with a simple lattice approach to rhythm, but has
now developed to include varying levels and subdivisions of swing, limited degrees of
Figure 2.3.4 The evolution in Logic of computer quantization of swing, from none (left, Logic
Notator 2.0,1988), to discrete, but not explicitly defined, values (middle Logic Pro 6, 2004) to continuous
values and advanced options (Logic Pro 9, 2009).
This paper includes analyses on examples with either quaver swing or semiquaver
swing, but research suggests that statistically significant rhythmic nuances may occur on
different subdivisions simultaneously (Naveda et al. 2009). Analysis of this kind raises the
concept of swing telescopy; which questions whether swing ratios are maintained or altered
46See Cholakis’s DNA project for an example of a commercial database of performer-derived groove
templates (Cholakis 1992).
93
between the quaver and semiquaver levels, or between normal time, half-time and
other words, latency is defined as the placement of the performer’s crotchet (onbeat)
against a negotiated time-line (defined clearly in 2.4, p 100-105), and is the component that
(negative latency). Again, this element of time-feel exists on a continuum rather than
discrete points. Charlie Mingus described the beat as an ellipsis, rather than a fixed point
in time (Berliner 1994, p 96). Figure 2.3.5 (p 95) illustrates the SLW representation of this
latency depends on tempo and stylistic context, so the descriptions are only intended as
Positive latency is more common idiomatically than negative latency, and can be a
significant contribution to a player’s style. Tom Morello, guitarist from Rage Against The
Machine – a band revered for their eschewing of synthesizers and quantization in a rap-
rock context – has reported that when recording a certain track, the studio engineer
noticed that all his notes were similarly late and shunted them back into the ‘correct’
position. On hearing the results, Morello objected to the ‘fix’ (“That’s not my sound”)
and asked for the track to be reverted to its original position (Morello in Guitar
Techniques 1996).
47 See 3.4 (p 142-5) for a discussion of swing telescopy in the context of Jimi Hendrix’s Little Wing.
94
Figure 2.3.5 The latency continuum (top) illustrating behind the beat (positive) and ahead of the
beat performance (negative) latency. Descriptive terms are included as an illustration of the expressive
range available. A similar idea of a beat ‘ellipsis’ was made by Charles Mingus, who illustrated it graphically
(bottom) in masterclasses (Mingus, cited in Berliner 1994, p 96).
preference to rushing. There is more conceptual and aesthetic tolerance for positive,
95
sounds like they’re cool, they’re relaxed, they’re not in any
hurry to get to the next note, because the note they are doing
right now is so good.
Govan (2010) 4:29 – 4:52
This ‘preposterous’ lateness is, in this paper, given the term hyperlatency, to explain
musical moments where a note can under some conditions, fall near, and sometimes
beyond the semiquaver (25% latency) and still maintain the sense of being an onbeat.
Why such a hyperlatent note is not conceived, or notated as falling in its ‘proper’ place (e.g.
the second semiquaver) is due to elements of stylistic context, rhythmic preparation and
articulation.48
There is another more fundamental question still unanswered: Given a fixed tempo,
why is one player described as late, and not the other as early? The answer lies in the
concept of a hierarchy of rhythmic establishment, the idea that one performer holds
more (or all) of the authority in the definition of the time line than others. This mutually
why jazz mechanisms of falling behind the beat cannot be adequately equated with rubato,
latency is an expressive mechanism concerned with friction against a relatively rigid time
line, not the elasticity of the time-line itself. Tempo changes do of course occur in
contemporary popular music, both as rubato cadences (Ashley 2002) and as (usually
intuitive) ensemble tempo shifts between sections of a song (Millward 2001a). However
accelerado, tempo shifts etc.) from the subtle but important art of rhythmic placement
and movement against this framework. Counter-intuitively, it is the soloist rather than the
96
accompanist, or accompanying ensemble, who has the least say in the tempo, but the
There are moments in the jazz (and wider) repertoire when the concept of one
framework have been identified (Mermikides 2008 and Benadon 2009). These may occur
as short gestures or longer passages that include points in time when soloist and
ensemble synchronize at specific anchor points, before separating tempi again. At the
furthest extreme, the use of technology can easily allow the ‘Ivesian’ co-existence of
many unrelated tempi or the lifting of musical material from one musical context to
another, as in Zappa’s concept of xenochrony.50 In these cases, the notion of a single time-
line needs to be released, and analysis and appreciation should be made in reference to
multiple time-lines, which may or may not be related simply. The identification of
simultaneous tempi (Benadon 2009) and analysis of latency or note separation curves (see 3.5
Push-pull (p 146-8) and 3.6 Temporal Plasticity (p 149-67)) are useful adaptive tools in these
contexts.
Weighting is simply the strength of attack of the offbeat relative to the onbeat. In
straight-ahead jazz, the emphasis of the weaker beats occurs routinely on the crotchet
level (the hi-hat, finger-click in count-in and idiomatic walking bass-line all add weight to
beats 2 and 4, relative to beats 1 and 3)51. In addition, emphasis is commonly added, at
49 This is in direct contrast to the governance of tempo for the classical concerto soloist, where the
conductor relays tempo fluctuations to the ensemble, who follow obediently.
50 Dissatisfied with the recorded sound of his electric guitar solos in the studio, Frank Zappa took to
recording his guitar solos from live concerts. He would then in the studio, transplant these solos into
another piece with a similar tonality but differing rhythmic context. Zappa enjoyed the transformative
effect of this process, which he termed xenochrony, examples of which have become the standard solo in his
recorded compositions. Dweezil Zappa, who now performs his father’s music achieves the impressive task
of transcribing, memorizing and performing live, extensive xenochronous solos in his Zappa plays Zappa
project (Hrab 2010).
51 In contrast, classic funk music tends to have a very strong 1st beat, strong beats 2 and 4 and a weak 3rd
beat. Swing values (predominantly in the 55-63% range) occur at the semiquaver level, as well as latency
components. For excellent overviews and demonstrations of funk and related drum styles see Slutsky &
Silverman (1997) and Moore (2005) respectively.
97
differing amounts, at the quaver level, the ‘bop’ of bebop (Crook 1991, Crook 1994,
Berliner 1996)52. It is this element that is introduced as the third component in the time-
feel model. Weighting may be calculated objectively as simply the dB differential between
offbeat and down-beat quavers. However, to align with the listener’s subjective
articulation into one variable is inevitably simplistic53 but is included to highlight the
With the three components of swing, latency and weighting in place, a succinct
description of time-feel can be made, particularly with the inclusion of the standard
musical space (as laid out in M-Space and Expressive Contours), a particular time-feel may be
and weighting - (1) (63,0,0) 2) (58,12,7) and 3) (50,22,20) – as sequenced clicks and
melodic phrases.
52 In learning both classical and jazz guitar, the author notes that slurs were approached differently in the
two disciplines. On the beat for classical, off the beat for jazz, providing a significant musical difference,
even without swing elements.
53 A useful survey of jazz articulation techniques is found in John McNeil’s The Art of Jazz Trumpet (McNeil
1999).
98
Figure 2.3.6 The SLW model of time-feel represented as a three-dimensional continuous subset
of M-Space. Three SLW time-feels: 1) (63,0,0) 2) (58,12,7) and 3) (50,22,20) are illustrated. These are
sequenced as clicks, and then performed as melodic phrases, on CD1.17.
With this overview of the SLW model in place, a formal mathematical model is
(2.7-2.12). A selected collection of time-feel case studies and their integration with M-
99
2.4 Where’s the Beat? Determining The Master Time-Line
Each performer has an individual series of values for the quaver set, with X
!
quavers{x0,x1,…xX}.55
For performer e, the timings of the quaver set are defined (Figure 2.4.2).
{t 0e ,t1e …t Xe }
Figure 2.4.2 Quaver set for performer e.
54Rather than performers, these can include distinct musical strands e.g. each hand at the piano.
55This approach is constructed on the scenario of only (near) quaver subdivisions, but can easily
accommodate higher, and lower subdivisions as required. The model also starts with the very simplistic
approach of all performers playing a continuous stream of quavers, but this is easily developed to
accommodate the real world of music making (Section 2.12 p 125-9).
100
These can be thought of as a mutually negotiated time-value between the
"t xe
e=1
TX =
E
Figure 2.4.4 Master quaver set determined by averaged time values across e performers.
!
More realistically, this calculation may be modified to accommodate a hierarchical
rhythmic component. Let Rxe will be a figure between 0 and 1, representing individual e’s
E E
e=1
[( ) ] where " r
TX = " t xe * rxe
e=1
xe =1
Figure 2.4.5 Hierarchically weighted time values. A click track or drummer may have all the authority
in terms of the master time-line, or it may be distributed, or actively passed, between performers.
! !
In a scenario with a guitar duo, the determination of a master time point may
dynamically move between the two players (during alternating solos) or fall between the
points during more balanced sections, such as a harmonized melody. Note that if an
accompanist gains all the command of time line definition, he alone can make the tempo
fluctuate, while the soloist is free to explore all manner of subtle and ‘preposterous’
lateness. Figure 2.4.6 graphically represents this model with a representation of the
master time reference during a harmonized solo (Rx1 = 1/2, Rx2 = 1/2), Guitar 1’s solo
(Rx1 = 0, Rx2 = 1), and Guitar 2’s solo (Rx1 = 1, Rx2 = 0).
101
Figure 2.4.6 The same asynchronous events (tx1 and tx2) in three different moments of a
hypothetical guitar duo performance: a) Harmonized melody (Rx1 = 1/2, Rx2 = 1/2), b) Guitar 1’s solo
(Rx1 = 0, Rx2 = 1), and c) Guitar 2’s solo (Rx1 = 1, Rx2 = 0). These create three different calculations for
the master onbeat (Tx) and resulting descriptions of lateness or earliness for each guitarist.
102
Figure 2.4.6 demonstrates that the determination of lateness or earliness depends on
a hierarchy of importance of the performers. For asymmetry of latency to occur with two
scenario with three performers (guitar, bass and drums). The resulting descriptions of
three onset times ( t x1 ,t x 2 and t x3 ) are shown in reference to three different hierarchical
weightings a) Equal weighting (Rx1 = 1/3, Rx2 = 1/3, Rx2 = 1/3) b) Oligopoly (Rx1 = 0,
! Rx2 = 1/2)
Rx2 = 1/2, ! and c) Monopoly (Rx1 = 0, Rx2 = 0, Rx2 = 1).
103
Figure 2.4.7 Varying determinations of lateness depending on hierarchical context. Master time
point Tx (dashed line) is determined by tx1, tx2 and tx3 in a) Equal weighting b) Oligopoly, performers 1
and 2 share responsibility, and c) Monopoly where performer 3 has all the metric responsibility.
104
The performer with the least input into time-keeping has the most opportunity and
range to explore latency mechanisms, as their actions do not pull the perception of the
master time-line towards them.56 One can think of this relationship as three objects of
differing mass and the resulting relative gravitational pull between them. In this case the
soloist has no mass and no influence on the ‘gravitational centre’. In contrast, the
place offbeats as he chooses (determining his swing level). Given this scenario, the
drummer could theoretically command radical tempo shifts57, but there are of course
tight aesthetic and idiomatic limits particularly in the flow of a groove. It is the
expectation of generally metronomic tempo that most allows the freedom of time-feel
mechanisms. If a track were recorded to a click that would be the limiting factor, but
felt somatically, and every player is, for example, expected to feel a common time pulse
during stops and solo breaks. This unheard common pulse or master time-line,
perceptible to all the players and listeners during a performance, may be considered as
another - albeit inaudible - performer, just like the unbounced studio click.
56 Of course this, like most elements of time-feel, requires an element of repetition; one waywardly late
attack is unlikely to feel like anything but an anomaly, but a string of notes behind the beat becomes a
perceptible time-feel element.
57 Small but significant tempo shifts do in fact occur in metronomic styles like funk. (Millward 2001a)
105
2.5 Defining Swing: Offbeat Asymmetry.
Swing for each performer is defined as the timings of three adjacent quavers from
the performers quaver set {xo, x1, x2 … xx } starting and ending on a crotchet beat, for
example to, t1 and t2. Swing (S) is determined as the relative placement of t1 within the
Figure 2.5.1 Calculation of swing (S) as the ratio of the time length from first onbeat to offbeat (t1
–t0) relative to the time length between onbeats (t2 –t0) .
So, as seen in Section 2.4, the traditional concept of the triplet jazz quaver yields a
swing value of 0.66̄ , a straight quaver gives a value of 0.5, and discernible values exist
106
quavers {x0,x1,…xX}58, occurring at time points {t0,t1,…tX}, the mean swing value of
Another important component that can be derived from this model is the amount of
variance around mean time-feel values. This is experienced as the relative tightness or
looseness of a performance, the average variance from a mean value during a passage of
music. In the case of swing values the standard deviation of S within phrase P (δP) may
1
"P =
2X
( [ 2 2
S0 # µ P ) + (S2 # µ P ) +!+ (S X #2 # µ P )
2
]
Figure 2.5.3. The standard deviation of swing in phrase P consisting of X quavers.
!
58This calculation presumes an even number of continuous quavers starting on an onbeat. Section 2.10
discusses swing calculations in the case of missing onbeats.
107
2.6 Ensemble Swing and Swing Friction
(Berliner 1994, p 349-52). These elements can be explained and calculated from time-feel
swing of ensemble E during phrase P ( µsPE ) given mean swing values of each of e
!
1 E
µsPE = " µsPe
! E e=1
standard deviation (σSp). Figure 2.6.2 shows the calculation as the average deviation per
Figure 2.6.2 Ensemble swing standard deviation for E performers over X quavers in phrase P.
!
Varying swing values may occur simultaneously between ensemble members,
creating a time-feel dissonance. Swing friction (sfab) is defined as the discrepancy between
two performers’ (e.g. a and b) swing values. This may be calculated from mean values
over phrase P (Figure 2.6.3). The swing friction between a performer (e.g. c) and the
ensemble swing (SFc) may also be calculated for phrase P (Figure 2.6.4).59
59Examples of swing friction are found in analyses of Johnny B. Goode (3.3 p 139-141) and Just Friends (3.6 p
149-152)
108
sfPab = µsPa " µsPb
Figure 2.6.3 Swing friction between performers a and b over phrase P.
It is possible for performers to share swing values but not be aligned in terms of
!
offbeat (due to latency discrepancy), conversely swing friction may exist even when
offbeats are aligned. The interaction between swing, latency and offbeat is discussed in
109
2.7 Latency
Whereas swing defines the displacement of the 2nd quaver, latency describes the
discrepancy between an individual performer’s crotchet point (for example ta) and the
master crotchet (Ta - as defined in section 2.4). This calculation is made relative to the
duration of the master crotchet (Ta+2 – Ta) . It can be seen as the placement of an
framework. It is to this parameter that such terms as “lazy”, “behind the beat”,
“rushing”, “on top of the beat” and so on, allude (Figure 2.7.1).
Figure 2.7.1 Calculation of latency for a performer at onbeat a’as the discrepancy between
individual and master onbeats relative to master crotchet.
A performer with positive latency is playing behind, and negative latency ahead of,
the beat. A latency value of 25% is equivalent to the duration of the master semiquaver60
110
!
Note that since latency has been distinguished from swing, it does not consider the
individual’s offbeat (ta+1) even though a latent phrase displaces the 2nd quaver.
may be tracked and the mean value (µLp) calculated (Figure 2.7.2).
$ (t # T ) '2
X /2
*& (T 2 x # T2 x ) )
x=0 % 2 x+2 2x (
" LP =
2X
111
2.8 Weighting
relation to the offbeat. Often a time-feel is established by accenting the offbeat quavers
(Coker 1964, Crook 1991, McNeil 1995). This can be thought of as a double-time
rendition of the familiar jazz crotchet walking bass-line.61 Weighting (W) may be defined
as the dB level of the offbeat relative to the onbeat in each crotchet phrase.62 Positive
weighting is defined as a more weighted offbeat than offbeat. Mean and standard
deviation values for W may also be calculated (Figures 2.8.1 and 2.8.2).
1 X /2
WP = " dB where dBt1 ,t0 is the dB differential of t1 and t0
2X x=0 t 2 x+1 ,t2 x
!
!
1 X /2 2
" WP =
2 X x=0
(
$ dBt2 x+1 ,t2 x #WP )
61 Berklee professor Ed Tomassi would give succinct and effective (albeit simplistic) instructions for good
time-feel: Play legato, behind the beat, ‘straight 8ths’ and accent offbeats slightly. (Tomassi 1995)
62 As explained in Section 2.3, more sophisticated measures of weighting may be used.
112
2.9 Time-Feel Matrix
For mathematical analysis and integration with M-Space modeling the three
components of the SLW model may be consolidated into a 2x3 matrix for mean and
standard deviation values of swing, latency and weighting in phrase P (Figure 2.9.1).
#µ " SP '
% SP %
P = $µ L P " LP (
% %
&µWP " WP )
1 n
µ SJ = " µ S p
n p=1
113
Or, as a further example, the standard deviation of latency between phrases in solo J
(Figure 2.9.4).
n
2
$ (µ Lp # µL J )
p=1
" LJ =
n
nodes such as semiquavers and triplets. Although there may be perceptual limits to the
discernment of adjacent points in this cube, it must be noted that Figure 2.9.5 (p 115)
available for exploration. Time-feel can thus be characterised as occupying a small space
in the time-feel M-Space (A), moving between various points during the course of a
shape) within a particular field (E). CD1.18 plays a phrase with three different time-feels
as illustrated in Figure 2.9.5 (p 115). The first phrase stays in one time-feel position (A:
medium swing, behind the beat, light weighting) while the second moves through three
points (B-D) in time-feel space. The last phrase stays in one general position (light swing,
late and weighted) but has a slight looseness as illustrated by E’s relatively large shape.
63 See Wishart’s depiction of the standard notation lattice (Wishart 1996 p 33).
114
Figure 2.9.5 Time-feel represented as a three-dimensional continuum subset of M-Space. Time-
feel characteristics may be seen as tightly-defined areas (A), trajectories (B-D) or loose fields (E). CD1.18
plays a phrase with three different time-feels: 1) time-feel A (medium swing, behind the beat, very light
weighting), 2) time-feel B to C to D and 3) time-feel E, very behind the beat, slightly swung, weighted and
fairly loose.
115
2.10 Relationships between Swing, Latency and Weighting
Swing and latency are defined in this model as independent variables, it is possible
to perform, for example, even quavers with a positive latency (‘straight and late’) or
swung quavers on the beat. However there is a perspective from which they are
interlinked. Considering a time-feel of 65% and 50% swing (Figure 2.10.1a), if (T2 – T0)
= (t2 – t0) the offbeat (t1) will occur at 65% of the duration between master onbeats,
however a swing of 60% and a latency of 5% will create an offbeat at the same point in
time (Figure 2.10.1b, p 117). There is a continuum of swing and latency values that share
the same offbeat point, ranging from swung and on time to ‘straight and late’ (Figure
2.10.2, p 118). The two extremes of these time-feels have significantly different effects,
yet are difficult to explain without the differentiation of swing and latency provided with
the SLW model. Some of the jazz community’s difficulty in explaining clearly the
concept of swing may be due to this subtlety, examples a, b, c and d can all be said to
swing (in a general offbeat asymmetrical sense) but the means of placing the offbeat, and
the resulting effect, are significantly different and this differentiation is a key component
of jazz time-feel. Another implication of this model is that at higher tempos, when
performers are no longer able to maintain an asymmetric quaver pattern due to the short
second quaver, straight and late playing provides a technically easier alternative to
116
Figure 2.10.1 Four time-feels (a-d) sharing the same offbeat placement in relation to the master
timeline but with differing swing and latency values ranging from s heavy swing in time to ‘straight and
late’.
Given that there exists a series of time-feels that place an offbeat at 65% between
master time points, it is possible to build a library of swing and latency values for all
offbeat placements. Figure 2.10.2 (p 118) displays swing and latency values linked by a
common offbeat placement. Examples a-d, from Figure 2.10.1, appear along the 65%
diagonal line with on time swing playing in the top left and straight and late in the
bottom right. Parallel diagonal lines represent other levels of offbeat placement with
corresponding swing and latency values, and expressive potential.64 These are illustrated
64 3.6 Temporal Plasticity identifies the use of this time-feel mechanism in the playing of Pat Martino.
65 The issue of analysis and the limits of perception is discussed in Section 2.12 (p 125-9).
117
Figure 2.10.2 Offbeat placement values (the performer’s offbeat relative to the master timeline)
plotted against swing and latency. Examples a-d from Figure 2.10.1 (p 117) are plotted on the 65% offbeat
line. An oval area of common jazz quaver values is indicated.
There exists another important connecting mechanism between swing, latency and
weighting not yet discussed concerning displacement. A simplistic time-feel (swing 50%,
latency 0%, weighting 0%) may be displaced by a quaver and still retain its time-feel
pattern. However with the introduction of any weighting or non-straight swing values
exact quaver displacement (50%) alters the time-feel pattern. Figure 2.10.3 (p 119)
illustrates how only an unweighted straight time-feel can be displaced by a quaver and
still align with its original time-feel in terms of accent pattern or synchrony. In order to
imply a quaver displacement, while retaining the original time-feel, the accent pattern and
swing values must be adapted; delaying the phrase by 50% does not have the same effect,
118
Figure 2.10.3 The comparison of 50% latency on four different time-feels. Only in example a) is a
50% latency exactly equivalent to a rhythmic displacement.
The same issue arises at the semiquaver level, unless the time-feel is unswung and
(or a quaver depending on tempo) would normally start with an upstroke, so there is a
latency of 25%.
119
Figure 2.10.4 The significant difference between the transformation of phrase a) through b) 50%
latency and c) quaver displacement.
50% latency and a quaver displacement. The implication here is that time-feel gives
quavers an identity that is not simply about their approximate alignment with the lattice,
would offer a reason of how hyperlatency is possible: A phrase could be played 25% (or
even up to 50%) behind the beat - particularly if the latency is introduced progressively -
and still feel like it is preposterously late rather than simply a rhythmic displacement,
120
2.11 SLW-Coach: A Computer Application for Time-feel
Analysis
theoretical models. Various systems are employed extensively to inform the author’s
research into time-feel including 1) sequenced time-feel matrices (Figure 2.11.1) for
calculation of onset times (Figure 2.11.2, p 122) and 3) data algorithms for extended
Figure 2.11.1 Sequencing of time-feel matrices in Logic Pro 9. Latency and swing transformations
displayed on the horizontal and vertical axes respectively. Weighting levels exist in hierarchical folders
making up the three dimensions of the SLW model.
121
Figure 2.11.2 Time-feel sonogram analysis of Pat Martino’s solo on Just Friends (See 3.6 Temporal
plasticity).
Figure 2.11.3 Extract of data analysis of time-feel in James Brown’s Lickin’ Stick (Mermikides
2005).
These systems are invaluable in the formation and development of the SLW model.
However they are all forms of post-hoc analysis and so somewhat divorced from the
music making process. To remedy this deficiency, a software patch was developed by the
author (using Cycling74’s MAX/MSP programming language) that gives real-time data
using the SLW model. This has proven to be a very useful analytical and practice tool.
122
A master time-line is derived by one of the following methods:
- An externally generated midi clock signal which may be triggered and controlled
live.
To this time-line the following calculations are made from audio or midi sources,
- Mean swing, latency and weighting values for any of the performers within a
- The standard deviation of swing, latency and weighting for any of the
occurs, or may be triggered manually by a midi-event (e.g. midi footswitch) This stores
the data for the previous phrase for later retrieval and resets all values for the subsequent
these values that can also cross reference a database of styles and performers.
SLW-Coach has provided a pedagogical and practice tool to evaluate and develop
time-feel in the author’s playing, which acts an evolution of the improvisation studies
undertaken at Berklee College of Music (Boston, MA). The development of accurate and
123
consistent time-feel is a challenging but valuable exercise that trains technique and aural
skills. It is tools like these that complete the circle from practice to theory and back
again.
124
2.12 Real World Time-feel Analysis
The SLW model provides a variety of tools to analyse time-feel mechanics in the
recorded repertoire. However the complete model is only fully applicable in the analysis
of multitrack recording (an invaluable resource for contemporary music research) with
tracks recorded to click as the most amenable, or in carefully controlled studies. In the
absence of multitrack recording, or prepared live analysis, the model can be used like a
Swiss army knife, with the most relevant component applied to each given situation, and
with the occasional inevitable compromise. Regardless, analysis should be guided by the
listening experience, and serve as a way to explain and illuminate more clearly real
musical effects, rather than the uninvolved creation of data sets. Section 3 (p 131)
presents several case-studies showing the adaptive use of the model, in order to
illuminate the most salient and instructive time-feel aspects from each given context.
Care should always be taken to identify which time measurements are actually
perceptible from the high degree of detail obtainable through technology. The use of
sequenced time-feel patterns for aural discernment is a direct and useful way to identify
the listener’s limits of perception. Fernando Benadon suggests 50ms as a useful guide to
the limits of meaningful time discrepancy in the analytical process (Benadon 2009) 66. He
takes into account measurement error and the inevitable noise of human performance to
arrive at this figure. However, experience with digital recording suggests that latency
values from as low as 12ms are perceptible by some performers. The limit for most
listeners most probably falls between these extremes, and the extent to which meaning
may be derived from analysis depends on a host of factors including musical, context,
66Benadon in fact suggests this 50ms benchmark in reference to increasing time values rather than onset
times, as is the case here.
125
tempo and technique for deriving onset time.67 As a guide, Figure (2.12.1) displays the
nuance of detail, in terms of quaver swing and latency percentages, for a given tempo
and listener’s time discrepancy resolution. A listener with 30ms listening acuity, for
swing from 50-75%. This sort of guide is helpful in maintaining a realistic approach to
time-feel analysis, and guards against the inclusion of negligible data in analysis. However
all data that is perceivable is not necessarily relevant. A spurious performance ‘slip’ or
technical difficulty should not always be classed as expressive latency. Repetition, groove
playing, tight standard deviations and honest reflection on the part of the analyst helps to
Tempo
Figure 2.12.1 Quaver swing and latency perceptive limits for given listener perception (ms) and
tempo (bpm).
Figure 2.12.1 raises the issue of tempo. Can the internal ratios of time-feels be
maintained at all tempos with equal effect? Only up to a point. Cholakis (1995) notes
that swing ratios tend to decrease with higher tempos. As discussed previously, there is a
67 Incidentally, the case study of Django Reinhardt’s Swing 42 (3.1 p 131-33) demonstrates at least four
clearly perceptible levels of swing between 50-75% at a tempo of 204bpm implying a clear perception of
timing discrepancies of around 18ms.
126
technical limit to the shortness of a note performed, so straighter playing would be
expected as we reach progressively higher tempi (The offbeat placement relative to the
master time-line can be maintained by latency). However there is also a ceiling to the
listener’s perception of short durations (at 200bpm, there is only 50ms separating 50%
and triplet swing for example). So as tempos increase, both the ability to perform, and to
hear, swing decreases. It would be expected that at higher tempos, latency and weighting
would become the chief components of time-feel. This aligns well with Tomassi’s guide
for bebop time-feel (Tomassi 1995), perhaps the higher tempos of this style instilled,
through necessity, the idiomatic weighted straight and late playing. Further research with
the tools presented here may reveal some evolution of time-feel due to higher tempos, as
well as the South American straighter quaver influence68, if without Horace Silver’s help.
So tempo may impact swing values, but there are other stylistic considerations that also
do not align across tempi. The gypsy jazz quaver, in particular while comping, tends to
have an increasing swing ratio at slower tempi (Dunn 2005) which is not explained by
important and subtle relationships between tempo and time-feel values across idioms.
There exists a huge recorded repertoire ready to be researched that will inform greatly
There remains an analytical issue yet to be discussed, that is quite minimal but
onbeats? Figure 2.12.2 illustrates four examples of offbeat quavers, only the first of
which (a) is addressed in the current model. Swing has so far been defined as the
placement of the offbeat relative the length of the performers crotchet (t2 – t0) but in the
absence of a performer onbeat (crotchet), either, or both, sides of the offbeat, this
68South American music including Bossa and Samba do not in fact have absolutely straight quavers and
semiquavers, there are light swing elements as well as other multi-dimensional microtiming elements
(Naveda et al. 2009).
127
calculation cannot be made (b-d). The solution is simply to use the master crotchet as the
time reference (T2 – T0 ) which is assumed to be equal to the performer’s crotchet.69 With
a missing 2nd onbeat (t2), as in examples c) and d) swing may be easily derived from the
swing or latency? In the absence of the musical context the situation is ambiguous.
Unless there is compelling evidence in the rest of the ensemble, a series of slightly late
offbeat semiquavers (as in an idiomatic guitar ska rhythm) may be thought as anywhere
along the relevant offbeat placement contour seen in Figure 2.10.2 (p 118).
Figure 2.12.2 Methods for calculating a performer’s swing with missing onbeats.
69In certain situations, the use of the master crotchet as the time reference may be preferable even when
both onbeats are present, as in example a) This may well be the case when the master crotchet is
unequivocally stable as in a click track.
128
The SLW model was developed to highlight and explain the visceral experience of
good time-feel and provide a simple set of tools for its research and wide
implementation. Example analyses of the SLW model are presented in Section 3 Case
Studies. From the perspective of a practitioner and a listener, this approach aims to make
definable those musical mechanisms that are consistently reproducible by artists and
recognized by the appreciative listener. The measure of worth of music analysis is the
extent to which it identifies and illuminates an actual musical experience, and its ability to
improve pedagogical practice and reapplication, towards which goals this paper is
offered.
129
3. Case Studies
Introduction
This section presents selected time-feel and M-Space analyses in a range of stylistic
contexts. The salient features of each case study vary, so relevant parts of the model, and
different analytical approaches are used to most clearly underline the musical
mechanisms in question. The selection has been made to provide an overview of some
time-feel characteristics and analytical approaches possible, but the overwhelming wealth
within the scope of the submission. Extended research into each of these examples, or
indeed into the huge resource of recorded repertoire available, will uncover many more
time-feel features that can explain effective musical mechanisms otherwise filtered out by
standard notational analysis. In 4.1 (p 168-9) a list of omitted, and ongoing analyses by
130
3.1 59% Swing on Swing ‘42
Characteristic swing in Django Reinhardt’s Swing ‘42
Figure 3.1.1. A series of quavers from Reinhardt’s solo on Swing ‘42 (1:04-1:06). Solo extract
CD1.19.
A phrase from Django Reinhardts’s solo on Swing '42 (Reinhardt 1949) provides a
very clear case study for analysis of swing values (CD1.19 plays a short extract from the
solo for context, followed by just the phrase in Figure 3.1.1). Although short, this
continuous sequence of quavers is performed along one string and with alternating
discover the contributing time-feel components. Sonogram, and half-speed onset analysis
of quaver swing was performed using the standard SLW model, as well as mean and
Figure 3.1.2. Swing values for each quaver pair in Reinhardt’s phrase.
131
The mean swing comes out at around 59.3% with a standard deviation of just 1.6%.
At this fast tempo (≈204bpm) each crotchet lasts about 294ms, so there is a mere 50ms
separating offbeat placement for 50% and 66.6̄ % swing. Reinhardt manages to sit tightly
between these extremes, occupying a time zone certainly no greater than 30ms, even
allowing for measurement ambiguity. Can the listener hear the difference between
straight and triplet quavers at this tempo, let alone a value in between? It is easy to check
the significance of these results. CD1.20 includes the recorded phrase followed by midi
sequences of the same passage at 50%, 59%, 66.6̄ % and 75% with all articulation,
timbral and weighting effects removed (The same sequence of swing values is then
repeated with pitch elements removed, where differences between the swing levels
become yet more pronounced). The difference between 59% and 66.6̄ % is fairly subtle
but each swing value is clearly differentiable with attentive listening. Reinhardt’s quavers
are certainly not played straight or with triplet swing, but close to the centre point in-
between these values. The musical effect of this swing value (coupled with the simple
polymetry in the phrase) is at least as important as the note choices, but is lost to
standard notation. A transcription may provide the notes, but the musician is left to
The relevance of the swing curve, the changing swing values through the phrase, is
miniscule and of little perceptual significance other than perhaps the introduction of an
extremely subtle human looseness. CD1.21 plays the phrase with 59% swing, followed by
a faithful rendering of each swing value (rounded to the nearest 1%). In a blind test, the
author can just distinguish the difference between the two with some consistency, mainly
due to the more even quaver ‘kick’ at the end of the phrase (CD1.22 plays the last two
quavers at 59% then 54% swing) but this 5% distinction is difficult to hear in isolation
132
As this style is traditionally (but uncritically) associated with the triplet quaver, could
this lightness of swing be due to technical limitation? This notion can be dismissed given
examples of Reinhardt’s picking virtuosity in his recorded repertoire. And since 50%
swing is technically easier than a swung rhythm and the standard deviation is low, it must
be concluded that this swing value was, at whatever level of consciousness, intentional.
theoretical modeling, directed attention and aural testing training of the ‘theory-filtered’
results. Analysis divorced from music perception is of little value, but music technology
now allows a powerful tool to marry theoretical analysis with subjective experience.
Pedagogical tools such as SLW-coach allow the conscious practice of this, and any time-
feel, with real-time feedback. These unearthed time-feel characteristics may also be
actively directed in composition, and with great accuracy by the use of electronics.
133
3.2 A Little Drag
Swing, latency and hierarchy in Michael Jackson’s The W ay You M ake M e Feel
his track The Way You Make Me Feel (Jackson 1987) in preparation for the final
performances of his career. The song is characterised by an offbeat keyboard stab, and
Jackson’s direction to the music director and keyboardist Michael Bearden concerning
the placement of this offbeat provides a remarkable insight into his attention to time-feel
and means of communicating it to his band. The language employed is a neat illustration
Although there are many duple rhythms, the tune has a predominately shuffle
rhythm, and the introduction, the focus of this study, is at a slow tempo (≈82bpm). For
these reasons it would be tempting to consider this offbeat to be a 3rd triplet quaver,
however for the purposes of this study it will describe it as a 2nd quaver with 66.6̄ %
swing. The distinction has little bearing on the substance of the analysis, but this choice
has been made to keep terminology consistent within the thesis and explanations as clear
as possible.
The editing of this rehearsal is possibly jumbled, it seems that the chronological
sequence of rehearsal events is in fact first a comment about feel (appearing 32:39-32:52
in the film, that will be called Extract A (CD1.23) here) followed by a rehearsal of the
tune’s introduction (appearing earlier in the film at 32:05 – 32:29) here referred to as
Extract B (CD1.25).
Extract A starts with Jackson, apparently dissatisfied with Bearden’s feel on the
keyboard part, instructing him to introduce a ‘little drag’ and play ‘a little more behind
the beat’, ‘like you’re dragging yourself out of bed’ – a particularly evocative description
of the experience of latency. Bearden is more concerned with his keyboard sound and
134
Jackson is soon distracted by an incorrect chord change, but in the short time available
he demonstrates the desired placement three times (CD1.23). The placement of these
offbeats is calculated in relation to Jackson’s foot stamps and finger clicks which are
taken, with as much accuracy as possible in the limited conditions, as the master time-line
(Figure 3.2.1).
Figure 3.2.1 Jackson’s directed offbeat placements in Extract A with tempo and swing values
(CD1.23 and 1.24 plays Extract A, followed by aural testing of these values).
crotchet duration. So comparing the swing values to a baseline of the triplet quaver
discrepancies of 1.3̄ % (≈9ms), 3.3̄ % (≈22ms) and 5.3̄ % (≈44ms) are found. The
of the swing values of 68%, 71% and 73% are played sequentially (panned right) against
the baseline 66.6̄ % triplet quaver (panned left) while a click (panned centre) marks the
pulses. In the author’s experience the panning effect is barely noticeable at 69% but
increased swing value. Taken in isolation, a 71% offbeat placement, with no adjacent
135
onbeats (as in this ska type rhythm), may be considered to be heavily swung or straight
with high latency70 so context must dictate the best approach. In consideration of the
‘dragging’ implication, a good case can be made that these deviations are latency
mechanisms acting on 66.6̄ %. The musical implications are identical of course, but the
swing and latency in other musical examples. For comparison, Extract A is rewritten in
terms of a 66.6̄ % swing with latency as the deviating force (Figure 3.2.2). The formulas
for deriving swing (S) and latency (L) for the first crotchet are included for completeness.
Figure 3.2.2. Latency values, given 66.6̄ % swing from Extract A. Jackson’s vocalized keyboard
stabs are labeled (t1, t3, t5) whereas footstamps and finger clicks represent the master time-line
rehearsal of the same section of music with Jackson, Bearden on keys and Jonathan
136
Moffett on drums. The edit includes seven bars and 26 quavers so a much more useful
data set than in Extract A. An analysis is taken with Jackson’s foot stamps marking beats
1 and 3, and the drums indicating beat 2-4. Up-beat placement of the keys is determined
by the offset between these markers. Tempo, which is slightly loose (within a ≈3 bpm
range) is calculated, in all but one case, between the first beats of each bar. Figure 3.2.2
shows the quavers, with tempo fluctuations, offbeat placements and latency relative to a
66.6̄ % swing reference. All values are rounded to the nearest percentage, which is a fair
Figure 3.2.2. Latency transcription of Extract B, with all values rounded to the nearest percentage
(CD1.25).
For the first two and a half bars, the offbeat placement stays consistently around
the 69% mark, with a subtle 2% behind the beat feel. Jackson vocalizes this rhythm from
the third chord closely, aside from a quickly corrected anomaly in bar 2 beat 1. However,
Bearden stabs the third chord of bar 3 playfully, and remarkably, late (13% latency,
around 90ms), which causes Jackson to share a laugh with him, thereon the offbeat
placement settles in either side of the 73% offbeat placement mark (6% latency, around
40ms later than the triplet quaver). The musical context makes these stabs feel like late
swung quavers, rather than 4th semiquaver placements, even though the latter are closer
137
from a standard notation perspective. From bar 4 Jackson seems content with the
dragged feel and sings the string-line over the accompaniment. CD1.26 is a 2-bar
sequence of the keyboard part played at 67%, 69%, 71%, 73% and then 75% offbeat
placement. The 70-73% range would appear to cover Jackson’s desired feel, a no-man’s
land (or obtusely written) area of standard notation, but a perceptual and effective
The playfully late chord (bar 3, beat 3) seems to cause Moffett to drop back a little
tempo (2bpm). The analytical context is not adequately controlled, the discrepancy small
and the technological limitations far from ideal, so little should be read into the event.
However, if the drummer did in fact feel compelled to accommodate the significant
latency with a reduction in tempo, this would suggest that the governance of tempo is
not entirely his responsibility. Using the terminology of the SLW model (see Section 2.4,
p 100-5) this would imply that master timeline determination is not a monopoly but an
oligopoly: The latency of Bearden’s (implied) onbeat actually stretched the master time-
line, so Moffett did not accept total responsibility for its placement. In this example the
situation is tenuous, but the calculation may as well be completed: A 2% drop in tempo
was caused by a 13% latency, so for that bar, the hierarchical weighting of master time-
This case study highlights the potential value in the analysis of rehearsal footage (and
multitrack recordings) of particular artists. Research of this kind is a powerful tool in the
71In may also be the case, that there is a non-linear response to latency fluctuations with, for example,
accommodation being made only above a specific threshold.
138
3.3 Constant Friction
Swing friction in Chuck Berry’s Johnny B. Goode (1958)
In Time-feel the concept of swing friction was described as the differential of swing
values between individual performers (or groups of performers) (see 2.6, p 108-9). If the
of ensemble feel.
Chuck Berry’s Johnny B. Goode (Berry 1958) provides an instructive example of swing
friction. Berry, often considered the father of rock n’ roll, was instrumental in
‘straightening out’ the blues 12/8 shuffle rhythm into the archetypal electric guitar riff.
Johnny B. Goode features this ‘straight 8th’ guitar rhythm, as well as equally straight lead
playing juxtaposed with a stubbornly bouncy drum, bass and piano feel. Heavily swung
quaver values occur in the ride cymbal pattern, often near the 67% mark, a significant
deviation of over 52ms from the straight quaver at 170bpm. The guitar rhythm part
however remains resolutely straight rarely venturing beyond 52% swing. This already
large 15% discrepancy of swing value is exaggerated with the guitar part often sitting on
top of the beat (ranging between 0% and -4% latency) leading to a mean separation of
about 17% (≈60ms). The lead guitar is equally straight, although not pushed, and
occasionally falling behind the beat. Piano interjections are loose but quavers are
generally quite swung, mainly in the 60-67% range and repeated quaver triplets prevail.
The bass plays mainly crotchets, with the occasional quaver (usually ≈67%). A
representative extract from the track can be heard on CD1.27. Figure 3.3.1 (p 140) shows
a composite two bar template for the lead, rhythm, bass and drum parts, with time-feel
components added. There is a huge gap between the swing values of the guitars and bass
and drums. The vocal track tends to fall in between these two extremes. In order to hear
the effect of swing friction, CD1.28 contains electronic sequences of this section with
139
varying time-feel values: 1) as from Figure 3.3.1 2) all instruments at 67% 3) all at 52% 4)
all at a middle ground of 60% and 5) back to the ‘true’ values for comparison.
Figure 3.3.1. Composite swing and latency values for guitars, bass and drums in Johnny B. Goode
(CD1.28).
The sequences have been rendered with MIDI instruments on purpose; although
the section would sound better with human performers, but the elimination of the
inflection they would inevitably provide allows focus on the power - and limitations - of
the SLW model. Mean values for swing and latency have been provided, but the standard
different between players. Weighting elements also occur, (the cymbal has a slight
emphasis on offbeat quavers for example,) with both mean and standard deviations).
CD1.29 plays the sequence first as Figure 3.3.1 then with swing, latency, weighting
standard deviations from Figure 3.3.2 (p 141) introduced, which add a clearly-defined
subtle but appreciable difference between the sequences; attention to the cymbal pattern,
140
Lead Guitar Rhythm Guitar Bass Drums
Figure 3.3.2 Mean and standard deviation values of swing, latency and
weighting (measured as dB level) (CD1.29).
An averaging out of time-feel components over the entire track runs the risk of
over-generalization and may incorrectly group specific mechanisms that occur only
occasionally. There are for example, brief moments when the bass seems to join with the
rhythm guitar’s straight quavers. There is also the assumption, with a single matrix per
instrument, that all beats of the bar are the same, which ignores the emphasis on
crotchets 2 and 4 in the drums. Matrices could be provided for beats 1 and 2, and beats 3
and 4 separately, or even weighting at the crotchet level, for greater sophistication when
needed.
Despite these acknowledged limitations, the discretionary use of this type of analysis
141
3.4 Swing Blocks
Swing values as arrangement in Little W ing (1967)
Jimi Hendrix’s Little Wing (Hendrix 1967) is a seminal piece in the rock guitar
unorthodox guitar voicings and a sophisticated rhythmic feel. Although often included as
part of the electric guitarist’s pedagogy, it’s apparently moderate technical demands
convincingly. The introduction (CD1.30) is performed solo (with bell tones later added at
the onbeats of each bar) with quite a loose tempo (in the 65 to 72 bpm range) so it is
reasonable to take the performance as the master time-line, which focuses on swing
rather latency elements. Semiquaver swing values vary widely with a broad spread across
the 44-72% range with a wide standard deviation, so is it fair to call the swing ‘loose’ and
leave it at that? Not at all, a closer look at the values (Figure 3.4.1 p 143), and more
informed listening, suggests controlled groupings of similar swing values, here broadly
categorized as Straight (<53%), Light (53-56%), Medium (56-63%) and Heavy (>63%).
These values are not implicated with tight rigidity, some fields fall on the cusp between
two categories, and occasionally a member of a group may fall out of the range.
Furthermore, there are some ambiguities in measurement when embellishments are used,
or in the absence of adjacent onbeats. However the general shifts between swing value
groups are very clear, attention to which reveals an effective structural mechanism at
work. CD1.31 demonstrates a one bar rhythmic pattern played in the four swing
categories: Straight (50%), Light (55%), Medium (62%) and Heavy (69%). These are all
performed first with sequences electronic clicks and then live guitar for comparison.
142
Figure 3.4.1. Semiquaver and quaver swing analysis of Little Wing introduction (CD1.30). Broadly
defined swing categories are grouped together in discernible blocks.
Swing analysis has so far been made in reference to the semiquaver, but a quaver
swing analysis, the length discrepancy between adjacent quavers within a crotchet, has
143
also been conducted (lower part of stave in Figure 3.4.1). These values are generally
straight which leaves little to say on the matter. Furthermore most swing values at the
quaver level may actually represent rubato elements (an accelerando would create a
higher swing calculation, for example). Nonetheless, bar 7 beats 3 and 4, provide an
opportunity to discuss the topic of swing values existing on multiple levels, or swing
telescopy. For clarity, a simplistic approximation of this passage is taken, with Pattern A
possessing swing values of 60% (quaver) and 50% (semiquavers) (Figure 3.4.2a). Pattern
B has 50% quaver and 60% semiquaver swing values (Figure 3.4.2b).
Figure 3.4.2. Pattern A contains two equal length quavers each containing a pair of 60% swung
semiquavers. Pattern B has a swing of 60% at the quaver level, with each quaver containing straight
semiquavers. Timings at 60bpm are shown, only the 3rd semiquaver is different between Pattern A and B,
but the musical effect is significantly different (CD1.32).
144
Semiquaver placements of Pattern A and Pattern B would be identical, apart from
the 3rd semiquaver (timings for 60bpm are shown in the central portion of Figure 3.4.2).
The musical effect of this distinction is huge, CD1.32 plays Pattern A followed by
unfathomable rhythmic patterns, and in the use of double-time and half-time rhythmic
devices. Further analyses by the author have revealed for example in the playing of Wes
Montgomery the use of swung quavers followed by the use of straight quavers, to
particular artist. Little Wing however provides a clear example of widely varied swing
intuitive version of the ‘swing-latin-swing’ format found in some jazz arrangements) and
explains part of the virtuosity in its execution and the challenge in its convincing
reproduction.
145
3.5 Push-Pull
The simplicity of arrangement in Confortably Numb (Pink Floyd 1979) affords ample
opportunity for Nick Mason and Dave Gilmour’s effective use of latency in order to
build and release tension. In conversation with the author, Mason has said his time-feel
(in particular his ‘behind the beat’ playing) is often brought up in seminars and although
about actively, and this element of his style is an unconscious knack rather than a pre-
conceived goal. Can a time-feel analysis reveal anything of this ‘knack’? Analysed here
are the hit points of four drum fills in Comfortably Numb demonstrating the use of a
general latency curve. Drum hit-points tend to start ahead of the beat, fall progressively
behind towards the end of the barline, and then make up for any lost time with a
shortened last note, in short, a push-pull effect (Hrab 2009). In Time-feel the model of a
mutually negotiated time-line was introduced (2.4 p 100-5), where pulse is determined by,
somewhat rigid and does not need marking out for every beat, the responsibility for
time-keeping is, ideally, deferred to each musician’s inner clocks; if the previous 32 bars
have lasted about 1800 ms, ceteris paribus, so will the next. This is demonstrated in a
typical jazz solo break, when the rhythm section stops dead for one or two bars allowing
the soloist free reign in that prescribed time duration. Often quavers, or other simple
subdivision are continued by the soloist, but all manner of liberties may be taken in this
time, with the assurance that the rest of the ensemble will re-enter at the same onbeat,
due to the prior establishment of tempo. In the situation of a drum solo, the precision of
the musician’s inner clock can really be tested, sometimes to breaking point, depending
146
on the drummer’s penchant for obtuse rhythms, or indeed, cruelty to other band
members.
Comfortably Numb maintains a tempo of around 128bpm, it was after all, as Mason
explained to the author, Pink Floyd’s attempt to write a ‘disco tune’. This analysis focuses
on the last bar of four drum fills, occurring at the end of eight bar sections in the outro
guitar solo. The length of these bars as marked by Mason’s drums ranges between 1869
and 1905ms (to Mason’s recollection the track was not recorded to click), and in each
case they only vary by a few milliseconds from their respective preceding bars. Distinct
marker points of these four drum fills are analysed here (occurring around song position
5:02, 5:17, 5:32 and 5:47 in the original track, 0:32, 0:47, 0:52 and 1:07 on CD1.33.
CD1.34-37 plays each fill individually), and compared to the metronomically ‘correct’
positions derived from equal subdivisions of the available time length. This can be
considered as changing latency values, or indeed changing time lengths between note
onsets (‘note separation’). Swing values are not calculated; the changing latency limits
their usefulness and the listening experience of these moments is undoubtedly one of
stretching time before the looming downbeat, rather than a changing swing feel.
Figure 3.5.1 Four fills from Comfortably Numb compared. The horizontal time position for hit
points is shown for each fill relative to an equal division of the bar. The vertical position of each hit point
represents the degree of positive and negative latency. All fills share a subtle but effective push-pull effect
(CD1.34-37).
147
Figure 3.5.1 (p 147) displays for each fill the onsets of each marker relative to the
metric subdivision. The four crotchet fills are very similar but not precisely the same
length, and have been normalized so they can be compared on the same graph. The
horizontal position of each hit point shows its latency relative to an equal-division of the
bar. This latency (relative to the metronomically derived crotchet) is also represented
vertically, so that an overall shape may be observed. A progressive falling behind the beat
would be seen as a rising latency curve, with correspondingly increasing onset separation
(‘note length’), with the last being accommodatingly short. These four fills essentially
follow a similar pattern of note separation increasing towards the end of the bar and then
rapidly diminishing for the last marker (A flat contour would indicate similar, and an
upward slope increasing, note separation). The contours all show a slight push ahead,
the next barline with a curtailed last note separation: an expressive contour of latency.
The hit points of these four fills are sequenced in CD1.38, each preceded by its
This analysis goes some way to explaining the mechanics behind this aspect of
Mason’s playing and the emotive effect it has on its listeners. It also raises the question
of the plasticity of the crotchet (and other subdivisions). Does a gentle latency contour
allow a greater range of values for aesthetically acceptable latency values? In other words
if latency is increased (or decreased) gradually, can extreme values be tolerated (and
enjoyed) more than random stabs of onsets either side of the beat? One can visualize
this idea as a fabric that can be stretched slowly, but will snap if pulled too quickly. A
further example of this concept is found in the next case study, 3.6 Temporal Plasticity in
148
3.6 Temporal Plasticity
Pat Martino commands high status in the world of jazz, for his prodigious talents as
a teenage jazz phenomenon, numerous awards, recording output and his remarkable
Martino has been particularly praised, from a young age, for his ability to execute
long seamless improvised lines with a rhythmic feel which somehow manages to be both
precise and relaxed (Fewell 1996). This particular time-feel is evident on Martino’s
performance on the jazz standard Just Friends from his debut album El Hombre (Martino
1962, extract CD1.39). The tune is performed at around 230bpm, ≈260ms per crotchet,
so a keen listener able to distinguish 13ms time differentials may perceive 5% slivers of
the beat. This would imply the ability to distinguish at best five levels of swing from
straight to 70%, and a similar number of values from -10% to +15% latency.
A swing and latency analysis of the melody and first improvised chorus reveals
swing values mainly in the 50-63% range and latency (where possible to derive) is almost
exclusively positive (behind the beat), largely occurring from 0 to +15% range. The
calculation of latency, not always possible, was made in reference to the drums, who
A swing friction exists between the drums (with swing values of 62-74% and
his straighter quaver groups, brings his offbeat more in line with the absolute placement
of the drums’ second quaver. Latency is only written in when directly calculable, but
149
range. An analysis of the solo break (CD1.40) illustrates this mechanism (Figure 3.5.1).
Figure 3.6.1 Transcription of the solo break from Just Friends (CD1.40) with swing and latency
values annotated.
A glance into the analytical process for this passage is provided in Figures 3.6.2 and
3.6.3 (p 151). The sonogram has been annotated with these swing values, as well as the
drums’ ride pattern when distinct. Discrepancies between the drummer’s, and Martino’s,
Figure 3.6.2 A sonogram analysis of bars 1-3 of Figure 3.6.1. Swing and latency values, where
clear are annotated.
150
Figure 3.6.3 Sonogram analysis of bars 4-6 with swing values for guitar solo and ride pattern
annotated. The swing friction between these values is softened by Martino’s significant latency.
In this passage Martino’s swing value stays mainly between 50-63%, hardly crossing
softens this friction. Since we are dealing with a 5% benchmark, these values can be
approximated for the purposes of simple illustration. Consider a constant drum swing
value of 65%, and Martino’s characteristic swing values of 50%, 55% and 60%. Figure
3.6.4 shows values of swing and latency that match the drummers offbeat placement of
65%. Although these time-feels have the same offbeat placement, they have significantly
different effects, even at this brisk tempo. CD1.41 renders electronically a phrase in time-
72 An enquiry into the whole solo, in fact, shows Martino’s tendency to swing more on melody notes and
idiomatic bebop phrase endings than on long sustained lines.
151
Figure 3.6.4 Four swing and latency combinations (i-iv) that share an offbeat placement of 65%
(CD1.41).
of Martino’s feel, but it’s still not quite right. A closer listen to CD1.36 suggests that the
first, and the last quaver pairs are generally more swung than the central material, and
there are passages that are clearly straight. Limiting a reconstruction to the time-feel
values along the red line (the 65% offbeat ‘iso-contour’) the feel of the solo break may be
generalises some of the details of the passage (in particular the swung and late quaver pair
in bar 5, beat one) but much of the feel is captured, demonstrating the effectiveness of
152
Figure 3.6.5 A generalized reconstruction of the solo break (top line) from Just Friends using time-
feel values (i-iv) from Figure 3.6.4, all sharing 65% absolute offbeat placement (CD1.42).
How much of this sort of sophistication is apparent to Martino? The author was
Martino’s writings on music are lucid and intriguing, particularly after his recovery
when his representations and illustrations of harmony and the fretboard took on both a
conceptual and aesthetic beauty (Martino 2004, Martino 2005) so illumination on the
topic seemed likely. But, as is often the case with skilled practitioners, Martino does not
have the same clarity when it comes to explaining his rhythmic approach, however
73 Martino: Unstrung (2007) funded by the Wellcome Trust and released on Ken Loach’s Sixteen Films.
153
hand] if I have to be honest with you, is not even
Kirk…this side is the Klingon.
(Martino:Unstrung 2007)
So while Martino can write entire books on fretboard visualization, his rhythmic
approach is described as “thoughtless”, even though this feature of his playing is highly
specifically designed recorded session in order to dig deeper into his time-feel and
rhythmic mechanisms.74
a suitable candidate for analysis of his rhythmic flexibility at a ballad tempo, and as this
was a favoured track in gigs, his confidence with it was not in question. A metronomic
backing track of the 24-bar chorus was made with as little ‘leading’ rhythmic information
recording was prepared with a visual and audio quaver-click track, in this way it was
possible to track with a high degree of accuracy the rhythmic placement of the
A Gibson ES-175 fitted with heavy-gauge strings and a Roland GI-20 midi guitar
system allowed a confident tablature transcription (as each string’s data was sent out of a
unique midi channel), another method for retrieving micro-timing data, and the real time
writings (Nature of the Guitar and Sacred Geometry) and allowed him to witness, in real-time,
74 Footage from the recording session appears in the movie documentary Martino:Unstrung (2007 Sixteen
Films) (DVD1 45:08-46:03).
75 The real-time translation of music improvisation to digital visuals is also explored in Rat Park Live and, in
154
At the start of the session, in an effort to calibrate the system as much as possible,
Martino was requested to play onbeats to click for 32 bars to provide a helpful indication
of how much, or little, rhythmic characteristics may be attributed to random noise and
how much to purposeful expression. The results showed onbeats falling within 11ms
either side of the click, providing a useful guideline to extract relevant data.
Four takes of two choruses each were recorded, included here is a detailed analysis
The four page transcription of Martino’s first take includes four staves (Figure
3.6.6). The top stave displays the published lead sheet, and the second a transcription of
Martino’s interpretation. The third stave describes latency in reference to the crotchet (at
47.5pm), with values below the central line representing notes that fall behind the
notated rhythm, and above the line, rushing. When appropriate this stave switches to
note separation values (see 3.5 Push-Pull ). In these moments (Bar 11.2-12.4) more
meaningful information is gained from this method rather than referencing the master
time-line placement. The lowest stave shows a guitar tablature transcription, Martino’s
fretboard concepts being an important focus in his literature. (Martino 2005, Martino
2007).
155
Figure 3.6.6.a Transcription of Martino’s performance on Welcome To a Prayer. The top stave
displays the original lead sheet, below which is a standard transcription of Martino’s performance. A
latency contour and guitar tablature make up the lower two staves (CD1.43).
156
Figure 3.6.6.b Page two of Martino’s performance on Welcome To a Prayer. In bar 11 the latency
contour is transformed into a note separation curve to better accommodate the fluctuating rhythmic
material (CD1.43).
157
Figure 3.6.6.c Page three of Martino’s performance on Welcome To a Prayer (CD1.43). Almost all
melody notes are delayed.
158
Figure 3.6.6.d Page four of Martino’s performance on Welcome To a Prayer (CD1.43). A rare (and
minimally) anticipated melody note occurs in bar 21.
159
Between the Lead Sheet and Take One staves, there are a series of arrows that signal,
by their angles, the displacement of the rendered to the written melody. These are
divided into primary and auxiliary melody notes - the distinction being based somewhat
virtually all the melody notes are delayed, with only two rhythmically anticipated melody
notes, (bar 5 and 11 and these are auxiliary melody notes and only slightly early). In fact
the key melody notes are so delayed they are played, with considerable dissonance, on
the next bar’s harmony (See bars 4, 17, 21 in Figures 3.6.6.a, c and d respectively).
and Crook 1999, p 119) and can offer as much improvisational expression as a ‘free’
superimposing the written melody on top of the recorded solo. In this way the dialogue
What may also be heard is that the delay in the melody notes tends to be
These late primary notes are often repeated (bars 2, 3, 6, 11, 13, 14, 15, 22 and 23),
played with greater intensity (bars 1, 9 and 12) or played with rhythmic precision on
A musical effect is created by the varied delay of melody notes, and an expressive
contour may be traced to this end. This allows the observation of a hidden but powerful
musical mechanism: melodic shadowing76. Distance from melody (in this case rhythmic) may
change over time and this creates the opportunity for the jazz performer to impart
shows an analysis of the first eight bars of the performance and traces a contour
76This concept of dynamic melodic shadowing is explored technologically in Strike, Omnia and 11th Light
from the submitted portfolio.
160
describing the extent to which each melody note is displaced. The vertical height of the
blue contour indicates the distance that the melody note at that original bar and beat
position has been displaced. When a melody note is re-attacked, this is shown by
different shades of blue (as indicated on the legend), at progressively higher positions.
Figure 3.6.7 Melodic displacement expressive contour for the first eight bars of Welcome To a
Prayer, indicating the extent to which melody notes are displaced in the interpretation (melodic shadowing).
Figure 3.6.7 demonstrates the extent to which the melody is rhythmically displaced,
and since this changes over time, an expressive contour emerges. However, displacement
notes, interjected phrases, melody note repetition, articulation and time-feel mechanisms
all pull away from the written melodies by differing amounts in various musical
dimensions. The concept of M-Space (1.5 p 25-55) describes improvisation as the serial
control of the degree of proximity between any new phrase and a dwindling memory of
preceding material. The concept of melodic shadowing however introduces the idea of
illustrates the first five phrase units from Welcome To A Prayer as variously proximate
161
phrases to the ‘platonic ideal’ fixed melody units, with a changing set of salient
Figure 3.6.8 An M-Space illustration of melodic shadowing for the first eight bars of Welcome To a
Prayer. Melodic interpretation is seen as the improvised transformation, along multiple dimensions, of the
‘fixed’ musical objects (A-E). The resulting interpretations (A’-E’) lie in varied proximal relationship with
their referenced, but unheard, counterparts (CD1.44).
improvisation against the referenced melody (CD1.44) but what can be said of time-feel
in this performance? The slow tempo, variation of subdivision implications and stylistic
feel of Martino’s playing does not allow much value to come from a swing analysis. An
enquiry into latency should also be approached with care, and although recorded to click,
the rhythmic feel is in general nebular and without the sense of friction against a rigid
time-line usually associated with this mechanism. Indeed, the transcription of the
one must be constantly vigilant against pareidolia, the tendency to see patterns and
meaning where none exist. Awareness of perceptual limits, and blind aural testing of the
162
hypothesis-filtered results can help prevent these Type I false positive errors.
Furthermore an enamoured focus on just time-feel runs the risk of missing other salient
and recombine component musical parameters in order to understand further their role
in the listening experience is a valuable exercise. There are however two passages from
this excerpt that are clearly instructive and should be discussed. The first is the initial
phrase (Bar 0.2-1.2 in Figure 3.6.9 and CD1.45) in which the melody is pushed and
pulled micro-rhythmically either side of the metric quaver length. The second, a passage
where a series of notes bear little relevance to the master time-line, other than some
anchor points, and are best notated in terms of changing duration, rather than their
Figure 3.6.9 Latency contour of the first phrase from Welcome To a Prayer (CD1.45). The melody is
micro-rhythmically stretched to produce a hyperlatent last note.
163
Figure 3.6.9 analyses the first phrase in terms of latency to the master time-line. At
47.5bpm, each percentage of latency is more than 12ms and perceptible, anything above
3% are clearly significant and beyond any reasonable error margin. The final G is played
over 250ms late, and could have been notated as its nearest major subdivision, the 2nd
semiquaver of bar 1. If there were no preceding quavers, this would clearly be the most
sensible transcription, however the elasticity of quavers 1-5, and the rising contour
displacement. A listen to the recording with, and without the referenced lead sheet
superimposed, seems to support this position (CD1.45). The fact that the latency
contour is smooth and in one down-up motion (lower diagram, Figure 3.6.9, p 163)
implies that the melody is micro-rhythmically stretched, rather than metrically displaced
The concept of differential elasticity, the contoured micro-rhythmic pushing and pulling
of a phrase, is exploited to a much further extent in Bars 11.2-12.4 (CD1.46). The series
transcription rather clumsy and inaccurate. The phrase can be seen as having three main
anchor points relative to the written melody, the initial note (F# which is also used as a
pivot note), the Bb at the beginning of bar 12 (which creates the b9 interval with the A7
harmony) and the A at the end of bar 12 (which is resolved two octaves lower and re-
resolved in the next bar, this time with the preceding B-natural from the melody). These
three notes form the melodic arch of the phrase and occur reasonably close to their lead-
sheet counterparts. There are also subsidiary skeletal melodic reference points as
164
Figure 3.6.10 The melody in bars 11-13 is used as a skeletal structure to frame a series of notes
with little metric relevance other than stretching time to hit the middle and endpoint anchors. The lower
diagram tracks the interaction of the melodic register and metric length contours (CD1.46).
165
Plotting onset separation for the phrase (which is equivalent to note duration given
the assumption of legato playing), a series of notes is observed that is used to fill time to
hit the key anchor points circled. Although these notes are very loosely in the semiquaver
range, they do not have a metric sense, rather they are variably stretched in order to
target the melody anchor points. Since the relevance to orderly subdivisions is dissolved,
so too is the sense of latency as defined in Time-feel (Section 2). This passage cannot be
is completely rigid and unyielding, and Martino’s sense of form is clearly retained,
marking key melodic notes and re-joining the melody clearly in bar 13. Whereas rubato
concerns ‘stolen’ time, this mechanism only allows lending and borrowing of time, and
all debts must be repaid in full. The best definition is of a variable tempo
superimposition, like a tape speed being variably and purposefully manipulated in order
to target hit points in absolute time. This idea of tempo superimposition, be it a fixed
tempo (in a simple or complex ratio with the master time-line) or variably distorted,
explains much of the latency curves found in Figures 3.6.6a-d (p 156-9).77 A slower
duration). To illustrate the idea of simultaneous expressive contours (see 1.6, p 56-68), a
curve of melodic register has been superimposed on the lower diagram. These two
contours are independent (as defined in Figure 2.5.4, p 64) and can theoretically move
freely against each other, however the expressive effect of these two contours moving
77This concept correlates with Benadon’s description of shift (fixed tempo superimposition) and flux
(variable tempo superimposition) in reference to early jazz (Benadon 2009).
166
variously together and apart, is clear (see also Figure 2.4.7 (p 104) for a similar example),
167
4. Coda
4.1 Ongoing Theoretical Research
Case Studies presented a selection of time-feel and M-Space analyses. Omitted and
weighting in the James Brown rhythm section, the identification of expressive latency in
Herbie Hancock’s Watermelon Man (Hancock 1962), differential elasticity in Bob Marley’s
(Montgomery 1965), the use of latency in David Gilmour’s guitar style, a more complete
study into M-Space mechanisms in all the takes of Welcome To a Prayer (Martino 2001) and
research into microtones (the vertical dimension of Wishart’s lattice) has also been
microtones in the blues guitar styles, as well as the construction of music software
(Mermikides 2005, 2007) to explore, identify and aurally test a catalogue of microtones
and tuning systems. Progress has also been made along the third dimension of Wishart’s
lattice, in the exploration of the expressive use of timbre in rock and jazz guitar
with articulation in the creation of good ‘feel’ so there exists symbiosis in the parallel
development of these fields. The addition of continuous note duration values for onbeat
and off-beat (so that 100% denotes legato, and lower values, staccato performance) to
the SLW model, would provide a simple conceptual foundation from which to make in-
The ethos behind all the theoretical work in this submission is to provide a
168
and not an end in itself. Ongoing and upcoming compositional and performance projects
employing the concepts presented in this collection of theoretical writings are presented
in the conclusion of Changes Over Time: Practice and a reflection on their outcomes will
undoubtedly enrich the theoretical foundation for future works and musical experience.
169
4.2 Presentations
On Composing (2010)
Article in Computer Music Magazine Special: Making it February 2010. 5,000+ readership.
Lecture presentation exploring the electronic music and the contemporary realization of
Omaggio and A Pierre for the Luigi Nono South Bank Festival.
23 November 20007 Public lecture. York Gate Gallery, Royal Academy of Music.
170
The Science and Art of Tuning (2007)
Lecture presentation on the science and art of tuning, and intonation software Horatio.
7 February 2007 Public lecture at York Gate Gallery, Royal Academy of Music.
3 June 2007 Public lecture. York Gate Gallery, Royal Academy of Music.
26 November 2008 Public lecture in the Dana Centre’s Infective Art event, Science
Museum.
5 March 2005 Presented at York Gate Gallery, Royal Academy of Music as part
of the Composer:Performer Exchange series.
171
4.3 Glossary
There are several terms borrowed from the jazz vernacular that are used very
specifically within this submission. These, together with new terms coined by the author,
Hyperlatency, hyperlatent. The particular case when a rhythmic placement retains the sense
of being an onbeat despite its absolute position being near another significant subdivision.
Most often found in the 15-25% latency range.
172
Isokinetos, isokinetic. An improvisational or compositional passage characterised by the
maintenance of a specific expressive contour over multiple musical dimensions.
Master time-line. A reference pulse, achieved through the (often hierarchical) negotiation of
pulse.
173
Offbeat. The second subdivision of a given pulse (i.e. ‘1 and’, ‘2 and’ etc. in the case of a
crotchet pulse, and ‘1 e and uh’ in reference to a quaver pulse.
Offbeat placement. The swing of an individual’s offbeat placement relative to the master
crotchet (in the case of quaver swing). Simply calculated as the individual’s latency added
to swing. For example a 10% latency, 55% swing has an offbeat placement of 65%.
Onbeat. On the beat, the beginning of each pulse. The first onbeat of the bar is the
downbeat, and between each onbeat lies an offbeat.
Phrase. An identifiable musical unit that has a, at least momentary, sense of autonomy.
Phrases may exist in hierarchical relationships, and may be edited, spiced, recombined
and otherwise transformed into new phrases.
Proximity. In the context of M-Space, the distance between musical objects along a
variable amount of transformative dimensions.
Seed. A musical phrase or idea used as a starting point for multiple ensuing phrases.
Slack theory. The aesthetic concept that the maintenance, or relaxation, of a set of
expressive contours, allows more flexibility to others.
Straight-and-late. A series of quavers with swing values near 50%, and a positive latency
value, resulting in an offbeat placement greater than 50%.
Swing. The relative offset of the offbeat in reference to the duration between.
Swung. A series of quavers (or other subdivision) with a swing value perceptively greater
than 50%.
Swing friction. The differential of swing values between performers and or groups of
performers.
Temporal plasticity. A general term regarding the extensive use of flexible micro-timing
mechanisms against a relatively stable master time-line.
Topic. A musical feature or transformative mechanism that may be addressed during the
course of an improvisation.
174
Unbounded. An improvisation characterised by a series of unrelated, or distantly related
phrases.
Villa-Lobos lattice. A quantized pitch (horizontal axis) vs. metric rhythm (vertical axis)
expressive contour derived from a physical shape.
Zappa’s Glue. A term for the structural mechanisms that form a perceptual relationship
between phrases, or musical objects.
175
4.4 References
Aebersold, J. & Slone K. (1978) Charlie Parker Omnibook. New York: Atlantic Music Corp.
Ashley, R. (2002) Do[n’t] Change a Hair for Me: The Art of Jazz Rubato in Music Perception
19.3, 11-32.
Ayeroff, S. (2002) The Music of Django Reinhardt: Forty-Four Classic Solos by the Legendary
Guitarist with a Complete Analysis. Missouri: Mel Bay Publications, Inc.
Bailey, D. (1980) Musical Improvisation: Its Nature and Practice in Music. Englewood Cliffs,
NJ: Prentice-Hall.
_________ (2007) A Circular Plot for Rhythm Visualization and Analysis in Music Theory
Online 13.3.
_________ (2009) Time Warps in Early Jazz. Music Theory Spectrum 31.1.
Bergonzi, J. (1992) Inside Improvisation Series: Vol. 1 Melodic Structures. West Germany:
Advance Music.
_________ (1994) Inside Improvisation Series: Vol.2 Pentatonics. Rottenburg: Advance Music.
_________ (1996) Inside Improvisation Series: Vol. 3 Jazz Line. Rottenburg: Advance Music.
_________ (1998) Inside Improvisation Series: Vol. 4 Melodic Rhythms. Rottenburg: Advance
176
Music.
_________ (2002) Inside Improvisation Series: Vol. 6 Developing a Jazz Language. Rottenburg:
Advance Music.
Bilmes, J. (1993) Timing is of the Essence: Perceptual and Computational Techniques for
Representing, Learning, and Reproducing Expressive Timing in Percussive Rhythm. Boston: MIT
Press.
Bowman, W. (1988) Doctoral Research in Jazz Pedagogy: An Overview in Council for Research
in Music Education Bulletin, no. 96, 47-76.
Borgo, D. (2005) Sync or Swarm: Improvising Music in a Complex Age. London and New
York: Continuum.
Byrt, J. (2007) Elements of Rhythmic Inequality in the Arias of Alessandro Scarlatti and
Handel in Early Music 35.4, 609-26.
Chadabe, J. (1997) Electric Sound: The Past and Promise of Electronic Music. Upper Saddle
River, New Jersey: Prentice Hall.
Cholakis, E. (1992) DNA Grooves. Music Software. WC Music Research and Numerical
Sound.
_______ (1987) Improvising Jazz. Fireside edition. Englewood Cliffs, NJ: Prentice-Hall.
Coker, J., Knapp, B., & Vincent, L. (2007) Hearin’ the Changes: Dealing with Unknown Tunes
by Ear. Rottenburg: Advance Music.
Cook, N. (1994) A Guide to Musical Analysis. New edition. Oxford: Oxford University
Press.
Cox, C. & Warner, D. (eds.) (2004) Audio Culture: Readings in Modern Music. New York:
177
Continuum.
_______ (1992a) Creative Comping for Improvisation, Vol 1. Rottenburg: Advance Music.
_______ (1992b) Creative Comping for Improvisation, Vol 2. Rottenburg: Advance Music.
_______ (1992c) Creative Comping for Improvisation, Vol 3. Rottenburg: Advance Music.
_______ (1995) How to Comp: A Study in Jazz Accompaniment. Rottenburg: Advance Music.
_______ (1999) Ready, Aim, Improvise!: Exploring The Basics of Jazz Improvisation.
Rottenburg: Advance Music.
_______ (2006) Beyond Time and Changes: A Musician’s Guide to Free Jazz Improvisation.
Rottenburg: Advance Music.
Chowning, J. (1971) The Simulation of Moving Sound Sources in Computer Music Journal
1.3, 24-26.
Damian, J. & Feist, J. (ed.) (2001) The Guitarist’s Guide to Composing and Improvising. Boston:
Berklee Press.
Dawkins, R. (2006) Climbing Mount Improbable. Revised edition. London: Penguin Books.
Desain, P & Honing H. (1993) Tempo Curves Considered Harmful in Contemporary Music
Review 7.2, 123-138.
Dobbins, B. (1986) Jazz Arranging and Composing: A Linear Approach. Rottenburg: Advance
Music.
Dreyfus, L. (1996) Bach and the Patterns of Invention. Boston: Harvard University Press.
Fauvel, J., Flood R. & Wilson R. (2006) Music and Mathematics: From Pythagoras to Fractals.
Oxford: Oxford University Press.
178
Folio, C. & Weisberg R. W. (2006) Billie Holiday’s Art of Paraphrase: A Study in
Consistency in Jablonski, M. & Klein, M. (eds.) New Musicology (Interdisciplinary Studies in
Musicology) Poznan, Poland: Poznan Press, 249-77.
Franz, D. (1998) Markov Chains as Tools for Jazz Improvisation Analysis. MSc thesis. Virginia
Polytechnic Institute and State University.
Friberg, A. & Sundström A. (1999) Jazz Drummers' Swing Ratio in Relation to Tempo
presented in ASA/EAA/DAGA '99 Meeting, Berlin, Germany, 19 March, 1999.
Friberg, A. & Sundström A. (2002) Swing Ratios and Ensemble Timing in Jazz
Performance: Evidence for a Common Rhythmic Pattern in Music Perception 19.3, 333-49.
Fuller, D. (1980) Notes Inégales in Sadie, S. (ed.) The New Grove Dictionary of Music and
Musicians, 20. London: Macmillan Publishers Ltd.
Goodrick, M. (1987) The Advancing Guitarist: Applying Guitar Concepts & Techniques. Hal
Leonard.
Hall, J. (1991) Jim Hall- Exploring Jazz Guitar. Milwaukee: Hal Leonard.
Heffley, M. (2005) Northern Sun, Summer Moon: Europe’s Reinvention of Jazz. New Haven:
Yale University Press.
Holmes, T. (2008) Electronic and Experimental Music: Technology, Music, and Culture. Third
edition. New York: Routledge.
Huang, H. & Huang R. V. (1995) Billie Holiday and Tempo Rubato: Understanding Rhythmic
Expressivity. Annual Review of Jazz Studies 7, 181-200.
Hudson, R. (1994) Stolen Time: The History of Tempo Rubato. Oxford: Oxford University
Press.
179
Juslin, P. N. & Sloboda, J. A. (eds.) (2001) Music and Emotion: Theory and Research. New
York: Oxford University Press.
Keil, C. (1966) Motion and Feeling Through Music in Journal of Aesthetics and Art Criticism
24, 337-49.
Larson, S. (1998) Schenkerian Analysis of Modern Jazz: Questions About Method in Music Theory
Spectrum, 20.2, 209-241.
Lerdahl & Jackendoff (1983) A Generative Theory of Tonal Music. Cambridge: MIT Press.
Lewin, D. (1993) Music Form and Transformation: 4 Analytic Essays. New Haven and
London: Yale University Press.
Lewis, G. E. (2000) Too Many Notes: Computers, Complexity and Culture in Voyager in
Leonardo Music Journal 10, 33-39.
Lindsay, K. & Nordquist, P. (2007) Pulse and swing: Quantitative analysis of hierarchical
structure in swing rhythm in The Journal of the Acoustical Society of America 122, 2945.
Lomax, A. (1968) Folk Song and Culture. Washington: American Association for the
Advancement of Science.
London, J. (2004) Hearing in Time: Psychological Aspects of Musical Meter. New York: Oxford
University Press.
McKee, R. (1999) Story: Substance, Structure, Style, and the Principle of Screenwriting. Paperback
Edition. London: Methuen Publishing Ltd.
McNeil, J. (1999) The Art of Jazz Trumpet. Paperback edition. New York: Music Sales
America.
Martino, P. (2004) Sacred Geometry: Simplifying the Fretboard. Masterclass with Jude Gold in
Guitar Player Magazine, April 2004. New York, NewBay Media.
Martino, P. (2005) The Nature of Guitar: A Broader Interpretation of our Instrument in Music
Theory Online Volume 12.1, February 2006. Society for Music Theory.
Meadows, S. E. (2006) Jazz Scholarship and Pedagogy: A Research and Information Guide. 3rd
Edition. New York: Routledge.
180
Millward, S. (2001a) Feel-Good Factor: Giving Your MIDI Tracks a Live Feel: Part 1 in Sound
On Sound July 2001.
_________ (2001b) Feel-Good Factor: Giving Your MIDI Tracks a Live Feel: Part 2 in Sound
On Sound August 2001.
__________ (1994) Doubleness and Jazz Improvisation: Irony, Parody, and Ethnomusicology in
Critical Inquiry Volume 20.2. Chicago: University of Chicago Press.
__________ (1996) Saying something: Jazz Improvisation and Interaction. Chicago: University
of Chicago Press.
Moore, S. (2005) Take It To the Street. Book & CD. London: Omnibus Press.
Miranda, E. R. & Biles, J. A. (eds.) (2007) Evolutionary Computer Music. London: Springer.
Nachmanovitch, S. (1990) Free Play: Improvisation in Life and Art. New York: Tarcher /
Penguin.
Naveda, L., Gouyon F., Guedes, C. & Leman, M. (2009) Multidimensional Microtiming in
Samba Music in SBCM 2009 Brazilian Symposium on Computer Music, Recife. 7-9
September 2009.
Nyman, M. (1999) Experimental Music: Cage and Beyond. 2nd Edition. Cambridge:
Cambridge University Press.
Persichetti, V. (1961) Twentieth-Century Harmony: Creative Aspects and Practice. Peters edition.
London: W. W. Norton.
Pressing, J. (1982) Pitch Class Set Structures in Contemporary Jazz in Jazz Research 14,
133-72.
Prögler, J. A. (1995) Searching for Swing: Participatory Discrepancies in the Jazz Rhythm Section in
Ethnomusicology 39.1, 21-54.
181
A. (ed.) Generative Processes in Music: The Psychology of Performance, Improvisation, and
Composition. Oxford: Oxford University Press, 70-90.
Rimer, J. & Yamazaki, M. (1984) On the Art of the Noh Drama: The Major Treatises of Zeami.
Chichester: Princeton University Press.
Rossi, M. (2006) Contrast and Continuity in Jazz Improvisation in Journal of the Musical
Arts in Africa 3.1, 106-8.
________ (2008) Contrast and Continuity in Jazz Improvisation: A Diatonic and Multi-colored
Approach. Rottenburg: Advance Music.
Russell. G. A. (2001) Lydian Chromatic Concept of Tonal Organization: Volume One – The Art
and Science of Tonal Gravity. 4th edition. New York: Concept.
Sagi, M. & Vitanyi, I. (1988) Experimental research into musical generative ability in
Sloboda, J. A. (ed.) Generative Processes in Music: The Psychology of Performance, Improvisation,
and Composition. Oxford: Oxford University Press, 179-194.
Schillinger, J. (1978) The Schillinger System of Musical Composition. New York: Da Capo
Press.
Schuller, G. (1986) Early Jazz: Its Roots and Musical Development. Paperback edition.
Oxford: Oxford University Press.
Schwartz, E. & Childs, B. (eds.) (1998) Contemporary Composers on Contemporary Music. New
York: Da Capo Press.
Serafin, S. & Young, D. Bowed String Physical Model Validation Through Use of a Bow Controller
and Examination of Bow Strokes in Stockholm Music Acoustics Conference (SMAC 03)
Stockholm, 6-9 August 2003.
Sickler, D. (ed. Porcelli, B.) (1979) The Artistry of John Coltrane. New York: Big 3 Music.
182
Slonimsky, N. (1999) Thesaurus of Scales and Melodic Patterns. London: Music Sales Ltd.
Slutsky, A. & Silverman, C. (1997) The Funkmasters: The Great James Brown Rhythm Sections
1960-1973. Miami: Warner Bros. Publications.
Sudnow, D. (1995) Ways of the Hand: The Organization of Improvised Conduct. Revised edition.
Cambridge: MIT Press.
Tait, C. (1995) Audio analysis for rhythmic structure in Proceedings of the 1995 International
Computer Music Conference. San Francisco: International Computer Music Association: 590-
1.
Titon, J. (1977) Early Downhome Blues: A Musical and Cultural Analysis. Urbana: University
of Illinois Press.
Wang, R. (1973) Jazz Circa 1945: A Confluence of Styles in Music Quarterly 59: 4: 531-546.
Werner, K. (1996) Effortless Mastery: Liberating the Master Musician Within. Indiana: Jamey
Aebersold Jazz, Inc.
Wishart, T. (ed. Emmerson, S.) (1996) On Sonic Art Amsterdam: Harwood Academic
Publishers.
Xenakis, I. & Kanach, S. (2001) Formalized Music: Thought and Mathematics in Composition.
New York: Pendragon.
Young, D. (2003) Wireless Sensor System for Measurement of Violin Bowing Parameters in
Stockholm Music Acoustics Conference (SMAC 03) Stockholm, 6-9 August 2003.
Young, D. & Serafin, S. (2003) Playability Evaluation of a Virtual Bowed String Instrument in
Conference on New Interfaces for Musical Expression (NIME). Montreal, 26 May 2003.
183
Young, D., Nunn, P. & Vassiliev, A. (2006) Composing for Hyperbow: a collaboration between
MIT and the Royal Academy of Music in 2006 Conference on New Instruments for Musical
Expression (NIME). Paris, 4-8 June 2006.
Young, R. (ed.) (2002) Undercurrents: The Hidden Wiring of Modern Music. New York:
Continuum.
Zappa, F. & Occhiogrosso, P. (1990) The Real Frank Zappa Book. Reading: Picador.
Zinn, D. B. (1981) The Structure and Analysis of the Modern Improvised Line. New York:
Excelsior Music.
Online Resources
Cholakis, E. (1995) Jazz Swing Drummers Groove Analysis. [Online] Numerical Sound.
Available: http://www.numericalsound.com[Accessed 6 February 2010].
Frey, D. (2010) New York Skyline. [Online] Red Deer Public Library. Available:
http://www.villalobos.ca/ny-skyline [Accessed 6 February 2010].
Machover, T. (1992) Hyperinstruments - A Progress Report 1987 – 1991. [Online] MIT Media
Laboratory, January 1992. Available: http://opera.media.mit.edu/hyper_rprt.pdf
[Accessed 7 January 2010].
Machover, T. (2004) Shaping Minds Musically [Online] BT Technology Journal, 22.4, 171-9.
Available: http://www.media.mit.edu/hyperins/articles/shapingminds.pdf [Accessed 10
October 2009].
Machover, T. (2010) On Future Performance [Online] The New York Times, 13 January 2010.
Available: http://opinionator.blogs.nytimes.com/2010/01/13/on-future-performance/
[Accessed 6 February 2010].
184
Unpublished Materials
Fewell, G. (1996) Pat Martino’s Linear Expression presented at Berklee College of Music,
Boston, USA, 2 February, 1996.
____________ (2007) The Science and Art of Tuning. Lecture presentation, Royal Academy
of Music, 7 February 2007.
Brown, J. (1968) Say It Loud – I’m Black and I’m Proud. Album. USA: King.
185
Jackson, M. (1987) Bad. USA: Epic.
Martino: Unstrung. (2007) Film. Directed by Ian Knox. UK: Sixteen Films.
The Real Frank Zappa Show. (1989) Radio Broadcast. Frank Zappa and Michael Oliver.
BBC Radio 3. Broadcast 1 September 1989.
186