Enriching the Blend: Creative Extensions to Conceptual Blending in Music
Danae Stefanou,*1 Emilios Cambouropoulos*2
School of Music Studies, Aristotle University of Thessaloniki, Greece
[email protected], [email protected]
In this paper we critically investigate the application of Fauconnier &
Turner’s Conceptual Blending Theory (CBT) in music, to expose a
series of questions and aporias highlighted by current and recent
theoretical work in the field. Investigating divisions between
different levels of musical conceptualization and blending, we
question the common distinction between intra- and extra-musical
blending as well as the usually retrospective and explicative
application of CBT. In response to these limitations, we argue that
more emphasis could be given to bottom-up, contextual, creative and
collaborative perspectives of conceptual blending in music. This
discussion is illustrated by recent and in-progress practical research
developed as part of the COINVENT project, and investigating
structural and cross-domain blending in computational and social
BACKGROUND & CRITICAL
A. Conceptual Blending in Music
(Fauconnier & Turner 2001) is a cognitive theory whereby
elements from diverse but structurally compatible mental
spaces are ‘blended’ giving rise to new or enriched concepts.
The blending paradigm extends Lakoff & Johnson’s (1980)
influential theory of Conceptual Metaphor (CMT), to suggest
multiple integrations operating across different conceptual
spaces, rather than unidirectional cross-domain mapping
between a source and a target domain. Blending has been
discussed extensively with regard to several fields, but has
primarily been applied to language and mathematics. The
theory has also been criticized as a ‘theory of everything’
(Gibbs 2000), given that its potential scope is so broad.
Consequently, research in blending often focuses more on
designing case-specific experiments for particular applications
of the theory, and on building constraints and optimality
principles for narrowing its scope (e.g. Bache 2005).
With regards to music, conceptual blending has been
predominantly theorized as the cross-domain integration of
musical & extra-musical domains such as text or image (e.g.
Zbikowski 2002 & 2008; Cook 2001; Moore 2012), and
primarily discussed from a musico-analytical perspective
focusing on structural and semantic integration, between e.g.
musical and textual rhythms, verbal and musical meaning etc.
Blending as a phenomenon involving “intra-musical”
elements (Spitzer 2003, Antovic 2011) is less straightforward.
In principle, one of the main differences of blending theory
from CMT is that it may involve mappings between
incongruous spaces within a domain (e.g. conflicting tonalities
in a musical composition). In this case, ‘intra-musical’
conceptual blending in music is often conflated with the
notion of structural blending (e.g. Kaliakatsos-Papakostas et
al. 2014, Ox 2012) and Fauconnier and Turner’s theory is
primarily applied to the integration of different or conflicting
structural elements, such as chords, harmonic spaces, or even
Nevertheless the recourse to intra-musicality, and its implicit
identification with structure, is not a neutral gesture; in fact it
brings a number of questions to the fore, including:
• What is a musical concept?
• What constitutes structural blending in music and how
does it relate to / differ from cross-domain blending and
• What can blending theory tell us about music not only as
top-down formalized structure but as an emergent,
data-driven, creative activity?
In their work on algebraic semiotics, Goguen & Harrell
(2010) have distinguished conceptual blending from structural
blending; they have used the latter to enrich the linguistic
notion of blending with “structure building operations” (291),
which include the composition of syntax, narrative etc. In this
respect, their approach is one of the few cases that have dealt
with blending from a bottom-up, creative perspective of
generating novel blends rather than analysing existent ones
(see next section).
In defining musical concepts and the process of their
formation, it is common practice to rely on divisions between
what is ‘music in the proper sense’ and what falls outside of
its scope. Such divisions, as we shall argue in the next section,
are perhaps one of the reasons for the relatively limited
application of blending theory in music thus far.
B. Revisiting the Intra- / Extra-musical Divide in Musical
“The intramusical (simply referred to, in music parlance, as
‘music’) is captured either in the inscription of notation, or in
specifically quantifiable, audible phenomena. Only what avails itself
of the assignment of specific musical values (i.e., pitch [and pitch
relations], meter, tempo, dynamics, instrumental voicing) is
proclaimed internal to the proper concerns of music. All else is
extramusical.” (Kim-Cohen 2009, p.40)
The passage above illustrates the problems of an age-old
division between “absolute” and “programme” music
(Hanslick 1891), also manifested in “absolutist” versus
“referentialist” accounts of musical meaning (Deutsch 2013,
p.589). Several recent theories attempt to break this binary
down, or develop it into somewhat more encompassing,
tripartite models of musical meaning and conceptualization.
Koelsch (2013), for instance, adds a further category,
“musicogenic” meaning, which refers to physical, emotional
and “personality-related” responses to music. At the same
time, he essentially retains the traditional definition of
intra-musical meaning as the product of “structural relations
between musical elements” (p.xi) and considers extra-musical
meaning as “iconic, indexical or symbolic” (p.xi, p.157).
Koelsch’s tripartite model is not without its parallels.
Brandt’s (2009) typology of formal, emotional and referential
levels of musical conceptualization can be neatly mapped onto
the former categories of intra-, musicogenic, and
extra-musical meaning respectively. A similar division is
found in Kühl (2007), but, unlike Brandt, the tripartite
typology of emotion, cognition and kinaesthetics is not
depicted as a linear progression, but rather as a cycle (for a
detailed evaluation of tripartite models such as these, as well
as an overview of terminological issues relating to levels of
conceptualization and the number of conceptual spaces at play
in this process, see Antovic 2011).
Fauconnier and Turner’s theory, however, rests more on
the assumption of conceptualization as an ordered, if not
entirely linear, progression, and essentially posits three stages
(or “levels”) of blending: Composition, Completion and
Elaboration. As Bache (2005) notes, progression between
these stages is depicted as a directional process, which
emphasises the subconscious, automatic aspects of integration
over the conscious process of dis-integrating elements in order
to reach an optimal selection of features for blending. In
response, Bache proposes an alternative three-level model,
starting from a low-level process of “Binding” conceptual
spaces (Level 1), then moving on to more abstract
“Construction Building” (Level 2), and finally, reaching a
“Partial Selection” Blend (Level 3), achieved through a
number of consciously imposed constraints on the lower-level
A fundamental limitation in all of the above divisions is
that they are retrospective analyses of progressions between
different levels of conceptualization; in other words post-hoc
accounts of concepts that have already been formed and are in
common use. Hence, even though Fauconnier and Turner
(1994; 2002) have theorised blending in relation to concept
formation, emergence and invention is largely studied by
looking at concepts as the already formed products of
blending processes, rather than setting up processes that may
lead to a new formation. As Schorlemmer et al. (2014) note,
CBT is “silent on issues that are relevant if conceptual
blending is to be used as a mechanism for designing creative
systems… [It] does not specify how novel blends are
constructed” (2014, p.2). Studies like Goguen and Harrell
(2010, see previous section) are a promising step in this
direction, which in recent years has been followed more
systematically in some domains, particularly the intersection
of computational creativity and poetry (see e.g. Corneli et al,
forthcoming). To our knowledge, however, there is very little
evidence of similar research dealing with generative, creative
theorisations and applications of blending in music. The fact
that “composition”, a term assigned to one of the most
complex high-level functions in music is used in CBT to refer
to the most basic-level process of subconscious binding
between domains also indicates that the ordering of levels in
CBT might require some adjustment and relativisation if it is
to be applied to musical creativity.
A further problem with retrospective applications of the
blending model (i.e. identifying a pre-existent blend, then
examining how it was constructed) is that they are less
relevant to High-Context situations (Hall 1992), where
structural and semantic relations are formed ad-hoc and are
extremely case-specific. However, for socio-centric,
contextualist theories of musical meaning, pretty much all
music could be considered high-context, so long as it is
inextricably tied with the subjective, variable nature of
real-time performance and interpretation. Schutz (1951) goes
so far as describing music as “a meaningful context which is
not bound to a conceptual scheme”. In practice, this could
mean that musical conceptualization is much more dependent
on dynamic, context-specific processes than it is tied to fixed,
The issue at hand, then, is to develop ways of overcoming
an analytical bias on fixed content theorisation and top-down /
post-hoc formalisation, and to also look for bottom-up,
creative applications of blending in the context of
compositional and performative musical processes.
II. BLENDING AS MUSICAL CONCEPT
INVENTION: TWO EXAMPLES
A. Melodic Harmonisation & Structural Blending
Structural blending processes appear in music across
several formal musical levels, from the level of individual
pieces harmoniously combining music features of different
pieces or styles, to the level of entire musical styles having
emerged as a result of blending between diverse music idioms
(for instance, jazz can be seen as ‘blend’ of african music,
european harmony and american pop, or, bossa nova as
combining samba and jazz; more generally, fusion music
‘fuses’ musical characteristics of different idioms/styles). Can
such blending be considered as conceptual blending or is it a
different type of blending? What is a music concept in the
context of structural blending?
Goguen (2003) suggests that structural blending is different
from conceptual blending: “Whereas conceptual spaces are
good for studying meaning in natural language, they are not
adequate for user interface design and other applications
where structure is important, such as web design and music.
For example, conceptual spaces and conceptual blending can
help us understand concepts about music, but semiotic spaces
and structural blending are needed for an adequate treatment
of the structure of music, e.g., how a melody can be combined
with a sequence of chords.” (p.9). Conceptual blending is
good for blending concepts about things (i.e. conceptual
spaces that describe high-level language-related descriptions
of things) but less adequate for blending the structure of
In the context of the COINVENT project (Schorlemmer et
al., 2014) a model is being developed that is based on
Goguen's proposal of a Unified Concept Theory (Goguen,
2006), inspired by the category-theoretical formalisation of
blending (Goguen, 1999). As an illustration of the model’s
potentialities, a proof-of-concept autonomous computational
creative system that performs melodic harmonisation is
developed. In this section, we present and discuss a couple of
creative examples that have arisen in the context of the
COINVENT melodic harmoniser.
In the current project, music concepts are taken to be
generalisations of harmonic entities and relations, derived
from a corpus of harmonic annotated data via statistical
learning. This data-driven approach ensures that learned
concepts adequatelly reflect characteristics of different
harmonic idioms. From each independent harmonic space (e.g.
modal, common-practice tonal, jazz, atonal, organum, etc.),
represented by a set of characteristic annotated music pieces,
important harmonic concepts (e.g., chord types and categories,
chord transitions, chords at phrase endings, note connections
of successive chords, etc) are automatically extracted and
encoded. This structural information sometimes corresponds
to standard musicological linguistic terms (e.g. ‘cadence’,
‘perfect cadence’, ‘dominant’, ‘leading-note’ etc.), bringing
the learned musical concepts closer to the standard notion of
‘concept’ in the domain of cognitive linguistics. In any case,
the important aspect of this approach is that manual
hand-coding of structural concepts is avoided, and emphasis is
given to bottom-up data-driven knowledge acquisition.
Another important aspect of the adopted methodology, is
context-sensitivity. The acquired structural descriptions are
relative and meaningful within the context of a particular
corpus of musical works. Music is defined in a circular
manner as something that specific human cultures identify as
being music (no general definition of music exists); specific
music contexts define relative musical concepts. The adopted
corpus-based learning methodology is one way to respect
contextuality, flexibility and adaptability of harmonic
descriptions; such automatically derived ‘ontologies’ may be
employed in conceptual blending.
Most of the research in conceptual blending in the domain
of music involves explication – particular music passages or
pieces involving, for instance, music-image or music-text
blends, or even structural blending between chords (as in an
excerpt by Stravinsky, analysed by Ox, 2014). All such
studies provide a rich interpretation of the selected music
examples via Conceptual Integration Networks (CINs). It is,
however, anything but trivial to reverse such processes so as
to employ the constructed CINs with a view to generating new
music examples. Constructing a blending framework that can
be used for the invention of new concepts and new musical
structures is a much more complex procedure; an abstract
language-based CIN outline is not sufficient. What is required,
is rich ontologies for the input spaces (including redundancy),
a strategy for constructing the generic space (i.e. what the two
input spaces share), and then to find efficient strategies to
combine ‘weakened’ descriptions of the input spaces that
avoid inconsistencies/contradictions and at the same time
preserve important properties of the input spaces (using
possibly priorities/saliences and other heuristics). Some
preliminary examples in this vein are given below.
Take, for instance, blending chords and more specifically
blending (prefinal) chords in the context of cadences
(Cambouropoulos et al. 2015; Zacharakis et al. 2015). It is
possible to interpret a certain chord as a blend between other
established types of chords. For example, let’s examine
briefly the augmented sixth chords, and more specifically the
German sixth chord (Figure 1). It is established that
augmented sixth chords have a strong predominant function
(Kostka-Payne, 2000, p. 385) and can be seen both as having
a secondary dominant character (Piston, 1978) and a phrygian
cadence character (Aldwell & Schachter, 2010). A Conceptual
Integration Network (CIN) can easily be constructed to
illustrated the German Sixth as a blend, with the Secondary
Dominant (V7/V) and the Phrygian cadences as input spaces,
and a general cadence description as generic space (brief
verbal descriptions can be introduced in the CIN such as
‘leading note’ for the secondary dominant or ‘descending
semitone to tonic’ for the phrygian input space). This may
look good and may highlight the character of a given chord,
but it cannot be used to generate new blended cadences or
The current research project is geared towards generation
and creative production, not simply explanation. Conceptual
blending is employed as a means to construct novel melodic
harmonisations and, even more so, new harmonic spaces that
can harmonise new unseen melodies. In this sense,
constructing rich ontologies for chords and chord progressions
in the above example, i.e. chord blending in the context of
cadences, and specifying precise mechanisms for blending is
anything but trivial. Multiple representations of constituent
chord types, roots, bass notes, chord note doublings or
ommissions, chord transitions and voice leading, relations
weights/priorities for all these properties (i.e., which are more
important/salient) are necessary, for a plausible chord
blending mechanism to be devised and implemented as a
computer program. Appropriate search strategies are
paramount in any attempt to create ‘meaningful’ cadence
blends, i.e. blends that preserve salient properties of the two
input spaces. A more detailed account of how the
COINVENT blending core model can create useful chord
blends is given in (Eppe et al, 2015) and an empirical
evaluation of the algorithm’s output is presented in
(Zacharakis et al. 2015).
Figure 1. Blending between the secondary dominant and
phrygian cadences (both ending on the dominant) gives rise to
the augmented sixth chords, such as the German sixth.
If blending of chords is a relatively complex procedure,
employing blending in melodic harmonisation is much more
so. A melody embodies a rich set of musical concepts that
relate to scales, tonal centres, motives, cadential patterns,
phrase structure, rhythmic characteristics, implied harmony,
and so on. Harmonising a given melody within its implied
‘natural’ harmonic space involves primarily exploratory
creative processes (finding a novel solution within a given
harmonic space), whereas a foreign harmonic language
triggers the need to combine different musical spaces leading
to novel harmonic concepts (combinational creativity). A
number of different harmonisations of a single melody are
given in Figure 2; the harmonisations are created
automatically by the COINVENT melodic harmoniser (at this
stage, chord types are computer-generated - voice leading is
added manually). The creative system is expected to be able
to adapt/adjust existing harmonic systems to foreign (possibly
incompatible) melodic structures by means of transformation
and/or invention of new harmonic concepts (more details in
Cambouropoulos et al. 2015).
Bach Chorale melody harmonised in medieval Fauxbourdon
Bach Chorale melody harmonised in Renaissance modal style
Figure 2. Two different harmonisations of a Bach Chorale
melody (chord types generated by melodic harmoniser) – reprint
from (Cambouropoulos et al. 2015).
It is maintained that a melodic harmonisation assistant that
facilitates conceptual blending should allow a modular highly
structured representation of harmonic concepts in an explicit
manner at various hierarchic levels and parametric viewpoints.
In this study, these harmonic concepts are not manually
conctructed, but, instead, are induced via machine learning
from harmonically annotated datasets. Five constituent
structural components of harmony are explicitly represented:
• Harmonic pitch space: scales, pitch hierarchies in scales,
consonance/dissonance, chord types.
• Chord transitions: Learning of chord transitions from
corpus data in one or more idioms/styles.
• Cadences: Learning of chord transitions that end phrases
at various hierarchic levels.
• Modulations: Changes of harmonic pitch spaces that
characterise a certain style.
• Voice leading: general characteristics of the way chords
are realised and connected in a given idiom.
Once structural characteristics of diverse harmonic idioms
are induced in an explicit modular fashion, then blends can be
created that combine different harmonic aspects from
different harmonic spaces. For instance, modal chord
transitions may be combined with tonal cadences (see
example in Figure 3), or, more daring blends may be
generated that combine, say, messiaen-like octatonic chord
transitions with tonal voice leading and modal Renaissance
cadences. Such harmonic blending experiments may produce
novel harmonic spaces that can generate new interesting
Figure 3. Bach Chorale melody harmonised in medieval
Fauxbourdon style with inserted tonal cadences (cf. Figure 2a) –
reprint from (Cambouropoulos et al. 2015).
B. Blending in the Context of Social / Distributed
The importance of collaboration and social interaction in
problem-solving as well as more open-ended creative tasks
involving structural blending has formed the basis of a series
of recent studies, particularly in the fields of mathematics and
computational creativity. Inspired by Tim Gowers’ (2009a;
2009b) work on collaborative mathematics, Pease (2012;
2014) and Corneli (2014) employ qualitative methodologies
to look at the process by which novel solutions, often the
result of blended thinking, emerge in social situations (such as
an online community of bloggers or peer-to-peer environment
contributors, sharing solutions to a mathematical problem).
In music, Georgina Born’s notion of “distributed
creativity” (2005, p.34) and its subsequent elaboration by
Clarke et al. (2012; 2013) extends the understanding of
musical creativity towards a more pragmatic, process-based
understanding of music as an activity distributed across
several different agents and attached to everyday activities.
This is not without its contingences, particularly when
considering music’s double role as a social practice and as an
autonomous art object in some cultures. In the majority of
Western Art Music contexts, for instance, musical
performance itself is of secondary importance, and emphasis
is placed on fixed-content musical works; consequently, the
social distribution of creativity is seen as an extraneous
side-effect of the musical process (hence the performance
ideal of Western Art Music largely assumes a Low-Context
distribution of musical meaning). By contrast, in situations
where music is understood as a more flexible,
context-sensitive continuum of actions, as in various kinds of
improvised or indeterminate composition (e.g. using graphic
scores) the unique creative content of every performance can
be examined according to more case-specific rules.
In such situations, multiple agents are engaged in a
real-time as well as post-hoc interplay of subjective questions
and answers: “what is this? is it good/right? is it bad/wrong?
is it even music?”. As a result, the very question of ontology
for every improvised soundwork is both formative and
dependent on the process of performance. As Russell (2009)
summarizes, “this approach is not ontology understood as the
deduction of reality from logical categories: it is the deduction
of those categories from reality.” (Russell 2009, 78).
How do musical structures emerge and combine in such
contexts, and what is the role of conceptual blending in an
open-ended, indeterminate music-making situation? In
May-June 2014, a loosely structured qualitative study was set
up and carried out at the School of Music Studies, Aristotle
University of Thessaloniki, to explore social creativity &
cross-domain musical blends, supported by the COINVENT
http://folioharmonies.wordpress.com and student participants
were invited to contribute as Authors, documenting their
responses to an open task.
Participants were given two kinds of sources: (A) an
example of a post-1945 graphic score (Folio: December 1952
by Earle Brown, which bears no verbal instructions and uses
abstract visual symbols instead of conventional musical
notation (B) a set of harmonic space paradigms, drawn from
examples used in the COINVENT harmonisation trials. These
included sample chord progressions and harmonic reductions
of composition segments by five prominent early 20th-century
composers, and suggestions for extending the harmonic
framework beyond these paradigms (e.g. free harmony).
The task was to collaboratively compose and/or improvise
a novel piece, putting these two kinds of sources to use in any
combination, following discussion and rehearsal. This latter
aspect was emphasized as equally if not more important than
the collaborative end-product itself. Setting up and
documenting an open-ended process with unknown outcomes
was one of the study’s key features, aiming at gathering a rich
set of process-based, context-specific documentation data.
Participants formed four ad-hoc groups of two to three
students each, and collaboratively composed and performed
four new pieces. A summary of end products as described by
participants themselves upon completion of the study is
presented in Table 1 below.
Table 1. End Products in FolioHarmonies
a reworking of
of ‘noise’ and
idea of people
1952 / Great
3/2D in 5-10
2 pieces in 1,
starting with a
as a starting
in real time
What kinds of data can an open-ended experiment like this
yield, and what can it tell us about musical concepts and
conceptual blending in music? Working with participants who
had no prior conceptions or experience in free improvisation,
composition or open score performance, and focusing on how
they approached the open-ended tasks of combining two
dissimilar sources into the composition and performance of
original piece of music, enabled us to examine the emergence
of novel blends in a social setting. Questions regarding
ontology, structure, style and evaluation (“what are we
making?”, “what context are we making it for?”, “why are we
making it like this?” and “how do we assess it?”) were
formulated and answered on an ad-hoc, context-specific,
Upon a first-level analysis of participants’ communication
patterns, shared problem-solving patterns also emerged across
all four groups. Although not always in the same linear order,
all groups followed strategies that could be summarised as
follows (Stefanou 2015):
1. Narrowing the problem space (e.g. from an open “what
if...” or a more case-specific “what to do with these two
sources” to a directional “how can we use source A [the
harmonic spaces] to interpret source B [the graphic score]”
and “how do we make this work?”)
2. Assigning functions and/or meaning to the set material
(e.g. using particular elements in the graphic score as
durational markers, or assigning narrative significance to
3. Mapping sonic elements onto visual ones, and vice versa
(e.g. creating subscores and testing them via different
4. Defining end-product ontologies (agreeing on what the
resultant piece should be described as, and what its constituent
Three out of four groups (Groups 1, 2 and 4) also produced
visual work in the form of ‘study scores’, aside from their
sonic end-products and performances. In the example below
(Figure 4), the graph depicts physical motion of the
performers in space (red line), with sonic events marked as
numbered rectangles on the Earle Brown score. Horizontal
rectangles in this version were interpreted as phrases and
vertical ones as cadences, drawn from the given harmonic
spaces. The marked course was to be followed by a ‘lead’
singing performer, while two other performers on pianos
played melodic and harmonic segments in two clashing
The cross-domain mapping between visual and sonic
spaces was extensive, and permeated all levels of the process.
One of the most significant, and somewhat unexpected
conceptualisations to have emerged during the study was the
dual metaphor “HARMONIES ARE TEXT” / “GRAPHIC
SCORE IS SOUND”. Across all four groups, participants
used references to reading, vision, and texts when referring to
source B (the harmonic paradigms), while consistently
evoking hearing, listening and sound in relation to source A
(the graphic score example). Harmonic paradigms were
conceived as something to be “read”, “said” (see example 1)
and made quasi-verbal sense of. Moreover, they were
metonymically associated with the composers of the examples
they were drawn from (mentions of “octatonic harmony”, for
instance, were quickly substituted by references to “the
A further instance of conceptual metaphor in operation
could be articulated as “GRAPHIC SCORE IS SPACE”.
Despite assigning different status to the two source materials
(as e.g. prompts, scripts, maps, targets, or sources) all groups
tackled the graphic scores in terms of a space that had to be
“navigated” (Groups 1 and 4), a surface that had to be
“mapped” (Group 2), or a framework that the harmonies had
to “fit into” (Group 3), consistently employing spatial
metaphors and in some cases (as in Figure 4 above)
translating these into literal motion in the performance space.
In terms of the transformation and structural integration of
the given harmonic paradigms, it is interesting also to
compare how the exact same material received different
handling and was therefore imbued with entirely different
meaning across groups. A modal mixture exemplified in bars
3-16 of Bartok’s Romanian Dance no.4, for instance, was
employed as a structural and narrative device by Group 2, and
was used as a kind of main theme, introduced by the leading
performer at the start of the piece, and strategically reiterated
at the end, to bring the harmonic spaces of all three
performers to a convergence. By contrast, Group 1 (see
example 1 below) re-framed the mixture entirely,
conceptualising it as “oriental”.
Figure 4. Sub-score developed by Group 2 for Me, You, Them.
Example 1: conversation segment, Group 1
By creatively integrating two types of inputs – an
indeterminate graphic score and a set of determinate harmonic
spaces – each group produced something more than a hybrid,
exceeding and transcending the two sources. This was largely
assisted by the introduction of new elements as ‘constraints’,
to optimize the process. The two types of input might appear
to be drawn from the same broad domain (music), and both
were presented as source material from which a composition
could be made. Nevertheless, they were in fact associated
with two distinctly different idioms or even genres
(indeterminate / open score music in the case of source A, and
harmonic composition in the case of source B). As a stylistic
and/or structural constraint, most participants resorted to a
third kind of source, which was not given from the onset.
Such parameters could be thought of as ‘extra-musical’, but
were described and handled by participants as anything but
that; they included noise, pre-recorded sounds and
mixed-media (Group 1) narrative and spatial / theatrical
motion (Group 2, see Figure 4 above), videos (Group 3).
Player 2: I won’t tune the C differently. Because it sounds nice as
it is written. Actually you know what? With a guitar we can say
[sic] Bartok. And the other could be e.g. pentatonic, like you also
said. Also thirds would be nice.
Player 1: With thirds yes, this could work. Shall we try it?
Actually shall we try doing it first, to see how it sounds if we both
have Bartok going on? On both guitars. Or we may even include
another two guitars. And we have the electric one too. […]
Player 1: […] Now it has this kind of, I don’t know, a Chinese
quality to it.
Player 2: So yes, let’s add an extra layer… like something oriental,
you know, Chinese.
While the material gathered during this study allows for
much deeper analysis across several levels, overall the
prioritisation of process and social context enabled the
formulation of radically relativised ontologies and shared
concepts to describe such ontologies. It also fostered a
dynamic, multi-level approach to blending, from the level of
harmonisation and melodic-harmonic relations, to that of
overall forms and end-product pieces.
III. CONCLUSIONS & NEXT STEPS
What becomes evident during a preliminary evaluation of
the above examples is that, in applying conceptual blending
models to compositional / creative musical processes, a lot
needs to be specified, particularly if we are to move beyond a
post-hoc explication of existent musical structures and onto
the invention & creation of new blends.
In this research we have been going in two directions
a) exploring structural blending in music, in the context of
b) investigating how cross-domain blending and conceptual
metaphor are implicated in collaborative musical creativity
situations in humans
We have also been looking at issues of terminology in the
applications of CBT to music so far, particularly with an eye
to better situating such research in contexts that are both
inclusive (i.e. not using unnecessary or aesthetically biased
divisions between conceptual categories) and specific (i.e.
formulating a given scope as precisely as possible, so that the
appropriate kinds of constraints and optimality principles can
be identified and applied in the construction of new blends).
The idea that emergent structural blends do not have to be
classified as intra- or extra-musical, but at the same time, need
to be described more precisely in terms of the level at which
they operate and the context / framework in which they can be
considered as blends, is key to this effort. The rather vague,
and historically loaded metaphor of music as an exclusive
core around which other domains orbit independently (Spitzer
2003) appears less and less relevant to an investigation of
structural blending and concept invention in music. By
contrast, further research on the types of blending observed in
bottom-up creative processes might have significant impact
on our understanding of how novel structures and concepts
emerge in music, how they are dynamically re-framed and
re-situated in high-context situations, and how we
conceptualize these transformations across different styles,
idioms and genres.
The project COINVENT acknowledges the financial
support of the Future and Emerging Technologies (FET)
programme within the Seventh Framework Programme for
Research of the European Commission, under FET-Open
grant number: 611553.
Aldwell, E., Schachter, C., & Cadwallader, A. (2010). Harmony and
voice leading. Boston: Cengage Learning.
Antovic, M. (2011). Musical Metaphor Revisited: Primitives,
Universals and Conceptual Blending. Available online at SSRN
http://ssrn.com/abstract=1763503. Accessed 14 May 2015.
Bache, C. (2005). Constraining Conceptual Integration Theory:
Levels of Blending and Disintegration. Journal of Pragmatics,
Born, G. (2005). On Musical Mediation: Ontology, Technology and
Creativity. Twentieth-century music, 2, 7-36.
Brandt, P. (2009). Music and the Abstract Mind. Journal of Music
and Meaning, 7, 3.
Cambouropoulos, E., Kaliakatsos-Papakostas, M. & Tsougras, C.
(2015). Structural Blending of Harmonic Spaces: A
Computational Approach. Proceedings of the 9th Triennial
Conference of the European Society for the Cognitive Science of
Music (ESCOM), Manchester, U.K
Clarke, E. (2012). Distributed Creativity in Musical Performance.
Talk given at the Keble College Creativity Series, University of
ce. Accessed 29 June 2014.
Clarke, E., Doffman M. and Lim, L. (2013). Distributed Creativity
and Ecological Dynamics: A case study of Liza Lim’s Tongue of
the Invisible. Music & Letters 94(4), 628-663.
Cook, N. (2001). Theorizing musical meaning. Music Theory
Spectrum 23/2: 170-195.
Corneli, J. (2014). Peer produced peer learning: A mathematics case
study. PhD thesis, The Open University.
Corneli, J; Jordanous, A; Shepperd, R; Llano, M. T.; Misztal, J.;
Colton, S. and Guckelsberger, C. (forthcoming). Computational
Poetry Workshop: Making Sense of Work in Progress.
Proceedings of the 6th International Conference on
Computational Creativity (ICCC) 2015.
Deutsch, D. (2013). The Psychology of Music. 3rd ed. London:
Eppe, M., Confalonieri, R., Maclean, E., Kaliakatsos-Papakostas, M.,
Cambouropoulos, E., Schorlemmer, M. and Kühnberger, K.-U.
(2015). Computational invention of cadences and chord
progressions by conceptual chord-blending. Proceedings of the
24th International Joint Conference on Artificial Intelligence
Fauconnier, G. and Turner, M. (1994). Conceptual Projection and
Middle Spaces. Research Report. Available online at SSRN.
http://ssrn.com/abstract=1290862. Accessed 14 May 2015.
Fauconnier, G. and Turner, M. (2001). The Way We Think:
Conceptual Blending and the Mind’s Hidden Complexities. NY:
Gibbs, R.W. (2000). Making good psychology out of blending theory.
Cognitive Linguistics 11 (3-4), 347-358.
Goguen, J. A. (2006). Mathematical Models of Cognitive Space and
Time. In D. Andler, Y. Ogawa, M. Okada, and S. Watanabe
(eds.), Reasoning and Cognition, Interdisciplinary Conference
Series on Reasoning Studies, volume 2. Keio University Press
Goguen, J. A. (2003). Semiotic morphisms, representations, and
blending for interface design. In Proceedings, AMAST Workshop
on Algebraic Methods in Language Processing (pp. 1-15).
Goguen, J. A. (1999). An introduction to algebraic semiotics, with
application to user interface design. In C. L. Nehaniv, editor,
Computation for Metaphors, Analogy, and Agents, volume 1562
of Lecture Notes in Computer Science, pages 242–291.
Goguen, J. A. and Harrell, D. F. (2010). Style: A Computational and
Conceptual Blending-Based Approach. In S. Argamon et al
(eds.) The Structure of Style. Berlin: Springer Verlag.
Gowers, T. (2009a). Is massively collaborative mathematics
ive-mathematics-possible/, March 2009. Accessed 14 May 2015.
Gowers, T. (2009b). Polymath1 and open collaborative mathematics.
ollaborative-mathematics/. Accessed 14 May 2015.
Hall, E. T. (1992). Improvisation as an Acquired, Multilevel Process.
Ethnomusicology 36 (2), 223-35.
Hanslick, E. (1891). On the Beautiful in Music. Trans. Gustav Cohen.
Kaliakatsos-Papakostas, Maximos; Cambouropoulos, Emilios,
Kuhnberger, Kai-Uwe; Kutz, Oliver; Smaill, Alan (2014).
Concept Invention and Music: Creating Novel Harmonies via
Conceptual Blending. Proceedings of the 9th Conference in
Interdisciplinary Musicology – CIM2014.
Kim-Cohen, S. (2009). In the Blink of an Ear: Towards a
Non-Cochlear Sonic Art. New York: Continuum.
Koelsch, S. (2013). Brain and Music. Oxford: WileyBlackwell.
Kostka, S. M., & Payne, D. (2000). Tonal Harmony, with an
introduction to twentieth-century music. 4th edition.
McGraw-Hill, New York.
Kühl, O. (2007). Musical Semantics. European Semiotics, vol. 7.
Bern: Peter Lang Publishing.
Lakoff, G. and Johnson, M. (1980). Metaphors We Live By. Chicago:
The University of Chicago Press.
Moore, Allan F. (2012). Song means: Analysing and interpreting
recorded popular song. Aldershot: Ashgate.
Ox, Jack. (2014). Analogy and Conceptual Blending are Part of a
Visualization Toolkit for Artists and Scientists: Introducing the
Cognitive Space Transfer. Proceedings of the IEEE VIS2014 Arts
Program VISAP’14: Art + Interpretation. Paris, France,
Pease, A., Budzynska, K., Lawrence, J., and Reed, C. (2014) Lakatos
games for mathematical argument. In S. Parsons, N. Oren, and C.
Reed, eds. Fifth International Conference on Computational
Models of Argument. Frontiers in Artificial Intelligence and
Applications. Amsterdam: IOS Press.
Pease, Alison and Martin, Ursula. (2012). Seventy four minutes of
mathematics: An analysis of the third Mini-Polymath project. In
Proceedings of AISB/IACAP 2012, Symposium on
Accessed 29 June 2014.
Piston, W. (1978). Harmony (revised and expanded by M. DeVoto).
New York: Norton.
Russell, B. (2009). Towards a Social Ontology of Improvised
Soundwork. In Mattin and Anthony Iles, eds. Noise and
Capitalism. Donostia: Arteleku Audiolab, 73-95.
Schorlemmer, M., Smaill, A., Kühnberger, K.-U., Kutz, O., Colton,
S., Cambouropoulos, E. and Pease, A. (2014). COINVENT:
Towards a Computational Concept Invention Theory, 5th
International Conference on Computational Creativity (ICCC),
10-13 June 2014, Lubljana, Slovenia.
Schütz, A. (1951). Making Music Together: A Study in Social
Relationship. Social Research, 18(1), 76-97.
Spitzer, M. (2003). Metaphor and Musical Thought. Chicago:
University of Chicago Press.
Stefanou, D. (2015). Investigating Social Creativity in Collaborative
Musical Situations. Proceedings of the 1st International
Conference in New Music Concepts, Treviso, Italy, March 2015.
Stefanou, D. (2014). FolioHarmonies: COINVENT Graphic Score /
http://folioharmonies.wordpress.com. Accessed 14 May 2015.
Zacharakis, A., Kaliakatsos-Papakostas, M. and Cambouropoulos, E.
(2015) Conceptual Blending in Music Cadences: A Formal
Model and Subjective Evaluation. (submitted).
Zbikowski, L. (2002). Conceptualizing Music. New York: OUP.
Zbikowski, L. (2008). Metaphor and Music. In Raymond W. Gibbs
Jr., ed. The Cambridge Handbook on Metaphor and Thought.
Cambridge: Cambridge University Press, 502–524