Music Taxonomies: an Overview
Franco Fabbri, University of Turin
This paper was presented as a keynote at the conference “Musique Savante/Musiques
Actuelles: Articulations. JAM 2014: Journées d’analyse musicale 2014 de la Sfam (Société
Française d’Analyse Musicale)”, Salle Igor Stravinsky, Ircam, Paris, 15‐16 December
2014. It is based on a research project submitted to the European Research Council. An
abridged version of the first part was also presented at the 18th International Conference
on Popular Music Studies, Universidade Estadual de Campinas, Sâo Paulo, Brazil, on July
1st, 2015.
1. State‐of‐the‐art
Concepts like ‘genre’, ‘style’, ‘form’, ‘mode’ – and their equivalents – have been used for
centuries in many cultures to classify music, by creating ‘types of music’ characterized
by recurrences and similarities. One of the purposes of such taxonomies, according to a
tradition originating from Aristotelian philosophy and developed throughout European
music history (Fabbri 1981, 1982, 2007a, 2007b), as well as in other music traditions,
has been to devise norms connecting the way music events are made with their meaning
and social function; another fundamental purpose has been to facilitate discourse about
music, by making it easier to recognize music events, describe, and ‘point at’ them.
Such musical taxonomic concepts have a great impact on everyday music life. Peo‐
ple choose a certain radio station (or listen to a certain web radio) because they like the
selection of genres or styles it broadcasts; concert venues and clubs are specialized ac‐
cording to genre (their location in the urban space, the profile of their customers, the
way in which the space is designed, and the acoustics of the halls are genre‐oriented);
record shops have shelves where similar music is assembled together (by genre), and
record companies created genre‐specific labels since their inception in the late Nine‐
teenth and early Twentieth century: their organization as corporate companies is genre‐
oriented (Negus 1999, Taylor 2014). As files with audiovisual content started circulating
over the Internet, the need arose to attach digital ‘tags’ to them, to indicate their content
and make it easier for users to find them. Soon, researchers in computer science started
1
working on algorithms enabling the automatic recognition of music properties by the
scansion of audio data, as one of the most challenging tasks in the Music Information Re‐
trieval (MIR) research field; with the expansion of interactive features in so‐called
WWW 2.0, and the huge success of social media, MIR researchers started considering
‘social tagging’, ‘folksonomies’ (Lamere 2008), and other user‐generated sources of in‐
formation as a complement to the analysis of audio content (Sordo, Celma, Blech and
Guaus 2008). Genre, style and other taxonomic concepts are now largely used in Inter‐
net, and big enterprises like Amazon, iTunes, Spotify, Last.fm, and others, improve their
business by suggesting customers and listeners (‘users’, in Internet jargon) to buy or lis‐
ten to music similar to previous purchases or listenings, by means of music recom‐
mender systems (Celma 2010). All musical taxonomic concepts play a central role in
music making (see how ‘form’ and ‘mode’ are articulated in many different music cul‐
tures around the world), in music theory and criticism, in any kind of verbal communica‐
tion about music. Music categorization contributes substantially to give music events a
socially recognized meaning (Fabbri 2007a, Brackett 1995 and 2002, Tagg 2013).
Research on music categorization has been done in the domains of the Social Sci‐
ences and Humanities (musicology, popular music studies, ethnomusicology, philoso‐
phy, semiotics, anthropology, sociology), of the Physical Sciences and Engineering (com‐
puter science: for a survey and an extended bibliography see Sturm 2012), and of the
Life Sciences (cognitive sciences and neurology). Scholars in each domain have acknow‐
ledged the existence of studies in the others, only to a limited degree, and – seemingly –
more as a form of academic correctness than out of a need to understand more deeply
the scientific problems involved.
Scholars in the Social Sciences and Humanities have of course acknowledged the
paramount importance of Internet in the shaping and usage of current music categories
(especially genre), but very few of them have dealt with the technical aspects of such
usage (tagging, ‘folksonomies’, recommender systems, datasets and data mining, statist‐
ics). The lack of technical expertise leaves musicologists, media scholars, sociologists,
etc., at the mercy of sensational news released on Internet, offering solutions to long‐
standing research issues. In January 2014 an article posted on the Google Research Blog
proclaimed that according to ‘data available on Internet’ jazz was (or is?) the most popu‐
lar genre in the 1950s (Cichowlas and Lam 2014). An odd statement for a music histor‐
2
ian’s ears, considering that in the 1950s jazz was more and more becoming a genre for a
niche of expert listeners, while rock ‘n’ roll emerged in that decade in the USA, and other
genres became very popular at the same time elsewhere (chanson in France, for exam‐
ple). A closer examination of the dataset used for that research, would reveal that it was
full of anachronisms (today’s labels, like ‘adult ballads’, used retrospectively) and wrong
entries (e.g. Serge Gainsbourg as a representative of jazz in the 1950s!). However, no
music historian complained publicly, and the news was spread all over the mediasphere.
On the other hand, most scholars in computer science, especially in the MIR (Music
Information Retrieval) field, seem to feel that their own musical competence, usually fo‐
cused on the Anglo‐American popular mainstream which dominates Internet usage in
the West, be enough to build datasets and models for their research (which is necessa‐
rily concentrated on recorded music). From the first experiments (Tzanetakis 2002) to
our days, the community focused on musical genres has compared different algorithms
in the MIREX contest (Downie 2008, 2010) but always using static datasets with clear
boundaries. Datasets are crucial for MIR research, and the choice is often limited by fac‐
tors that are not under the researchers’ control, like copyright ownership. Therefore, the
fact that datasets are biased (which is the main cause for odd results like those men‐
tioned above) is due to reasons that go beyond the musical expertise and/or taste of re‐
searchers. It must be said that many MIR scholars since the early 2000s have included
references to musicology‐oriented papers (see, for example, McKay 2004, Basili, Serafini
and Stellato 2004, McKay and Fujinaga 2005, Guaus 2009), but most of them have lim‐
ited themselves to picking up hints at very specific methodological issues. As a matter of
fact, tagging protocols (ID3) used by some of the most popular applications (like iTunes)
are now very old and inefficient: they are restricted to a limited number of genres (de‐
rived from the Anglo‐American popular mainstream), obliging to categorize music not
belonging to that mainstream as ‘world music’ without any distinction; they do not allow
a correct management of chronological data, and the way pieces of classical music are
described (usually referred to as ‘songs’, as this is the name for an audio file in Internet
jargon) is simply disastrous. MIR researchers are not directly responsible for such a
situation (commercially available music files are tagged mostly according to protocols
developed initially in 1996 by a computer programmer with no specific music exper‐
3
tise), but the fact that no satisfactory standards have been proposed to date is discon‐
certing.
Cognitive scientists and neurologists have offered empirical results and theories
that were very soon included in approaches to categorization by other disciplines (like
musicology and computer science), with concepts like prototype, schema (see Lakoff
1987, Lakoff and Johnson 2003, and comments by Eco 1999), and others – e.g. mirror
neurons (Rizzolatti 2005), which aren’t directly related to categorization, but have left
their mark on many music studies in the past ten years (Molnar‐Szacaks and Overy
2006). At the moment, such research seems to be in a phase where cognitive scientists
are more eager to give than to receive suggestions (see, for example, Levitin 2008),
probably due to the complex nature of their experiments, the cost and bulkiness of the
equipment involved (MRI scanners), and the fact that they are more often dealing with
fundamental aspects of categorization, rather than with the very specific details of music
categories.
Communities are a crucial concept in all disciplines involved in this research. It has
been noted that although communities are included in the definition of other fundamen‐
tal concepts – from ‘meaning’ and ‘cultural unit’ (Eco 1976), to ‘genre’ (Fabbri 1981,
1982), and even to ‘nation’ (Anderson 1983) – little has been done in order to give a de‐
tailed description of what it has to be meant by the term, whenever the concept is ex‐
tended beyond its original formulation in sociology or anthropology (Kaufman Shelemay
2011). What kind of community is a ‘virtual community’? Can the idea of an ‘imagined
community’ be applied to the fans of a certain artist, or genre? Do communities overlap?
2. Some hints from my approach to the problem
It is a matter of fact that musical genres exist in many cultures. Nineteenth century posi‐
tivist musicologists even treated them (or what was then conceived as genres) as living
entities. Such biological metaphors – implying concepts of birth, infancy, growth, ma‐
turity, death – were soon abandoned, in the wake of the hegemony of formalist musicol‐
ogy, and its devaluation of any concept related to function, context, community. Genres,
however, continued to survive and are still flourishing: their existence today found in
the domains of culture, of commonsense, that is, in the semiosphere, in the discourses
4
and practices of musicians, critics, fans, concert promoters, record industry executives,
sales people, web designers, and so on. One of the main challenges for scholars is to re‐
late such existences to an understanding of the meaning and working of music taxono‐
mies: ethnographies are useful, but without theory they are blind.1
As cultural units (and not metaphysical categories), genres are rooted in history:
which would imply that for each genre that comes to our mind there must have been a
time when it didn’t yet exist. This might be an obvious observation, but one that does not
seem to have troubled many of the scholars who have dealt with the subject. On the con‐
trary, I believe that no genre theory – be it a ‘strong’ theory or a simple description of
how the concept is used in contemporary communities – can be valid if it doesn’t take
genre formation and diachronic processes into consideration.
According to several theoretical approaches, which, in my view, should be seen to
overlap and/or complement (rather than contradict and oppose) each other, the ‘birth’
of a genre can be located in the establishment of conventions within a community, in the
‘semiotic act’ of naming, as well as in the acknowledgement of ‘family resemblances’, in
the acceptance of prototypes. Such processes, however, do not take place in a void, but
within a system or network of existing genres: this also implies that some or all of them
can be activated, or catalysed, or polarized by existing genres, to which the new genre is
opposed, or put on their side as a variant. For example: is popular music a genre? Ac‐
cording to colloquial usage and to some definitions, yes, but when referring to it as a
superordinate category, which includes many genres, one may prefer to call it a type of
music. Also, in discourses where distinctions between genre and style are not relevant,
type may be a useful synonym for both.
2.1. Rules or Black Boxes?
I have to admit that, to date, my suggestion of a kind of ‘complementarity principle’
among theories (or between two main theoretical categories) is just a little more than
wishful thinking. Proponents of general or music‐specific taxonomic theories, in fact,
seem to be convinced that theories based on conventions (on the one side) and theories
1
This is my basic critique of otherwise excellent works, such as Holt (2007). For a review, see Fabbri
(2008).
5
based on prototypes, family resemblances and schemata (on the other side) do actually
oppose and exclude one another. There seems to be something in common between
such diverse theories as George Lakoff’s (and other cognitive scientists’) theory on
prototypes and ‘basic level recognition’ (Lakoff 1987), Pierre Bourdieu’s habitus (‘prin‐
ciples which generate and organize practices and representations that can be objectively
adapted to their outcomes without presupposing conscious aiming at ends or an express
mastery of the operations necessary in order to attain them’, Bourdieu 1992, 53), and
Daniel Levitin’s neo‐Kantian ‘schemas’ (or schemata, Levitin 2008): they all strongly op‐
pose rules and property‐based definitions (to the point, in Levitin’s case, of ridiculing
them),2 and are in favour of explanations based ultimately on neural phenomena, on the
formation of cognitive and behavioural habits hardwired in human bodies. In terms that
those authors would probably not accept or have accepted, this means bringing such
phenomena well below the domain of semiotics. At least in Bourdieu’s case, the dis‐
missal of rules was part of an anti‐structuralist stance (aimed especially at French struc‐
turalist anthropology, i.e. Lévi‐Strauss);3 and in Levitin’s case, with his grotesque de‐
scription of property‐based definitions, I would claim a misunderstanding of semiotics’
basic principles (such as coding). In Lakoff’s case, moreover, a thorough commentary, as
well as a revision of semiotic theories, has been provided by Umberto Eco in his Kant
and the Platypus (1999). In all cases, even in Eco’s theory of ‘cognitive types’, a kind of
black box is invoked: all we seem to be allowed to know is that there is something func‐
tioning at the interface between perception and cognition, and that it influences our be‐
haviour. It is suggested that our neural system, with specific associations of neurons ‘fir‐
ing’, is at the physical base of such a black box. It is also implied that we ‘learn’ how to
recognize and behave (Bourdieu is the one who worked more on this issue, though he
admitted that it was a very complex matter), but it is unclear what learning processes
are involved, and whether cognition plays any part in it.
Most of these explanations (whose authors have been particularly keen in re‐
proaching other theories for the same reason) tend to be ‘static’ in one way or another. It
is clear that we are not born with a schema to recognize heavy metal (Levitin’s exam‐
ple), and that at some point we can form one, based on family resemblances or proto‐
2
3
Ibid., pp. 142–3.
See Jenkins (2002).
6
types. But how does it happen? Is it possible to improve our schemata so that we recog‐
nize heavy metal more promptly? Is it possible to forget them? And how was heavy
metal first recognized (as there must have been someone who had this experience be‐
fore any other)? Moreover, does the expression ‘heavy metal’ exist only to allow a lis‐
tener to recognize specific pieces of music? What about guitarists who want to learn to
play like one of the prototypical heavy metal guitarists: must they develop specific
schemata? Similar to or different from an average listener? What about the manager of a
stadium where a heavy metal gig is planned: will their considerations about seat place‐
ment, security, and so on, be part of a distinctive schema? Or of a habitus? Or are such
considerations in the domain of cognition, of ‘conscious aiming at ends’?
How we recognize things is very important, as is how we make sense of such re‐
cognition. Unfortunately, albeit all the more interesting, concepts (cultural units) such as
genre extend over both domains. Humans use genre – genre names, especially – to talk
about music. For many, this is one or maybe the only way to verbalize their own experi‐
ence of music. Some people are definitely able to talk about a genre without even being
able to recognize a piece of music that ‘belongs’ to it. Genres are also about beliefs and
lies: as such, they can be an object of semiotic study. But can semiotics make sense of the
diachronic aspects of genre? Isn’t a semiotic approach to genre (as a cultural unit, or ‘a
semantic unit inserted into a system’, see Eco 1976, 67) ‘static’, as a number of scholars
maintain (see, for example, Santoro 2010, 27‐28)?
Often what emerges from such criticism is a distorted image of semiotic concepts:
not distorted enough to make the destructive rhetoric clear even to those who are not
familiar with the discipline, yet just right to accommodate common sense prejudices
against a field of study that hasn’t been fashionable in the past decades. Semiotic codes –
according to that rhetoric – are like commandments, they are agreed upon deliberately
by fixed communities and are there for good. Codes (or norms, or conventions) are bind‐
ing, they are not negotiable, no conflicts about their meaning are possible. It is implied
by such criticisms that semioticians (and their associates: linguists and language phi‐
losophers) don’t know how to handle collective processes: only sociologists know what
a socially shared norm is and how it works. I would argue that this is not so. Yet, I do not
imply by this that sociological approaches to genre are wrong.
7
Another advantage of semio‐linguistic approaches to genre, opening new, interest‐
ing perspectives to the otherwise ‘mysterious’ act of initial codification, is pointed out by
the philosophical study of convention made by David K. Lewis (1941–2001) in 1969.
Aimed at a rigorous definition of one of the fundamental processes that make language
possible, the study is based on a class of games (coordination games) that were over‐
looked by game theorists at the time. With examples of growing complexity, Lewis
shows how conventions can be established without ever stipulating them explicitly.
Therefore, the conventional nature of language, and of any code (or set of norms) estab‐
lished conventionally, doesn’t imply that at any point there be a clear agreement, or that
the involved parties (or community) shall declare the acceptance of the convention(s).
Of course, the existence of a convention can be recognized and even officially acknow‐
ledged, but recognition happens when the convention is already in place, and is based on
the actual functioning of the convention itself. There is no black box, though. Here is the
first rough definition of convention given by Lewis:
A regularity R in the behavior of members of a population P when they are agents in a re‐
current situation S is a convention if and only if, in any instance of S among members of P,
(1) everyone conforms to R;
(2) everyone expects everyone else to conform to R;
(3) everyone prefers to conform to R on the condition that the others do, since S is a co‐
ordination problem and uniformity to R is a proper coordination equilibrium in S (Lewis
2002, 42).
Given the other, more refined definitions in the book, Lewis himself warns against the
risk of hiding ‘the concept beneath its refinements (Ibid.)’. The rough definition is en‐
ough for our purpose,4 especially if we complement it with an extract from the book’s
‘Foreword’, by Willard Van Orman Quine:
We have before us a study, both lucid and imaginative, both amusing and meticulous, in
which Lewis undertakes to render the notion of convention independent of any fact or fic‐
4
I made a reference to Lewis’s definition when my 1981 paper on genre theory was first published in
Italian: Fabbri 1996, n. 5, 16.
8
tion of convening. […] in the course of the book the reader comes to appreciate conven‐
tion, not analyticity, as a key concept in the philosophy of language.5
The final comment, made by the quintessential analytic philosopher Van Orman Quine
about one of his pupils, is meaningful. But are we allowed to extrapolate Lewis’s ideas
from the philosophy of language to genre theory? Yes, without any doubt, I would say. In
fact, Lewis does the opposite: he attaches to the philosophy of language a theory based
on the observation of ordinary recurrent situations, such as: who should recall first if a
telephone call is unexpectedly cut off? Such coordination problems are essential in
music practice. Although Lewis’s examples do not cover music activities, one of the
earliest examples in the book could easily be translated into musical terms:
An example from Hume’s Treatise of Human Nature: Suppose you and I are rowing a boat
together. If we row in rhythm, the boat goes smoothly forward; otherwise the boat goes
slowly and erratically, we waste effort, and we risk hitting things. We are always choosing
whether to row faster or slower; it matters little to either of us at what rate we row, pro‐
vided we row in rhythm. So each is constantly adjusting his rate to match the rate he ex‐
pects the other to maintain (Lewis 2002, 5).
In a recurring music event, maximizing the pleasure of each of the participants, or ensur‐
ing everyone has the best understanding of what’s going on, or minimizing the amount
of information that must be processed to obtain pleasure and understanding may be
perceived as coordination problems; conforming to genre conventions, to use Lewis’s
words, is ‘a proper coordination equilibrium’ in that recurring event.
What is especially fascinating in Lewis’s theory is that it makes sense of the ‘sub‐
conscious’ aspects of categorization and behaviour (see Bourdieu’s ‘practices and repre‐
sentations that can be objectively adapted to their outcomes without presupposing con‐
scious aiming at ends’), but without making use of black boxes. Or, instead, looking into
them.
In order to make a comparison let us return to some other theories. According to
prototype‐based theories, genres are defined by resemblance; in my opinion, it may well
be that crucial resemblances are established by convention. Moreover, I would argue
5
Willard Van Orman Quine, ‘Foreword’, in Lewis 2002, xi–xii.
9
that although prototypes and family resemblances are fundamental in the recognition of
stylistic features by individuals, what makes them meaningful in collective music prac‐
tice is that they are conventionally formalized; that is, that a community socializes such
recognition and takes full advantage of it. Of course an individual can learn a schema to
distinguish ‘heavy metal’ as the genre where music resembles pieces by Led Zeppelin,
the ‘quintessential heavy metal band’ (Levitin 2008, 142).6 But, which piece(s)? And who
developed that schema first? And how is it transmitted to a whole community? And is
that resemblance enough to define all aspects of the genre? Levitin himself acknow‐
ledges that prototypes and family resemblances are not satisfactory explanations for the
many nuances of music competence, though he doesn’t seem worried by the circularity
of arguments such as ‘we say that something is heavy metal if it resembles heavy metal’.
Looking for diachronicity on the basis of a theory that is exemplified by such statements
can be a hopeless task.
On the other hand, convention (as described by Lewis) is a process situated in
time: there is a time before a convention takes place, and a time after; it is also possible
to describe the process by which a convention ceases to exist (when, for example, it is
recognized that a certain regularity in the behaviour of the population is not a coordina‐
tion equilibrium anymore, and members of the population stop conforming to it). Look‐
ing for a theory that helps to explain how genres are born, we also need a theory that ac‐
counts for ‘dead’ genres.
2.2 Naming genres
Of all conventions that define a musical genre – that is, of all regularities in the behav‐
iour related to music events to which members of a community conform – naming con‐
ventions deserve special attention. In many historical cases I can think of, it seems that
the naming of a genre is a kind of sanction, of ratification that other conventions exist
and have been acknowledged. In such cases practice anticipated naming: that is, a gen‐
eral acceptance of styles, social practices, functions, etc., under a specific name, followed
years, decades, maybe even centuries of similar music activities, as with fado, flamenco,
6
By the way, what about the fact that Led Zeppelin were initially dubbed a ‘heavy rock’ band, and
that ‘heavy metal’ later became a name for the genre, or maybe simply another genre?
10
tango, the blues, jazz, rebetiko, up to rock ‘n’ roll. Often, music historians wonder if a cer‐
tain genre existed before any reference to its name could be found: I would suggest that
in many cases some of the most relevant conventions defining a genre tend to operate
before a name for the genre is agreed upon, but that the ‘act’ of naming makes other
conventions more ‘visible’ and helps to create new ones. Though apparently pointing in
the opposite direction, the way film historian Rick Altman describes the early develop‐
ment of cinema could also be interpreted in the light of the above‐mentioned theoretical
scheme:
During the last years of the nineteenth century and the first of the twentieth, it simply
cannot be said that there was such a thing as ‘cinema’, clearly separate from other phe‐
nomena. On the contrary, what we now retrospectively think of as cinema was at the time
recognized as several quite different phenomena, each overlapping with an already exist‐
ing medium. This multiple identity is made manifest by contemporary terminology, which
applied to moving picture projection such shared terms as ‘views’, ‘pictures’, ‘electric
theater’, ‘living photographs’, and ‘advanced pictorial vaudeville’, each self‐evidently iden‐
tifying the new technology with an already existing medium (Altman 2004, 19).
In the case of cinema, names based on existing media polarized practices around each of
a number of established conventions prevented the main common element (moving pic‐
ture projection, i.e. ‘cinematography’) to emerge both as a name and as the activity
around which new conventions could be formed. From this example we may concede
that rather than identifying previously unnamed practices (and conventions), a new
naming convention can, so to speak, rearrange the whole field, by deleting previous
names. That is how ‘cinema’ worked (and, maybe, ‘jazz’). Names are crucial (see Marino
2013).
However, while I would insist that a name may be a sufficient condition to define a
genre (though a named genre that keeps together any kind of music events, without any
kind of codified regularity among them, is hard to imagine), it is not a necessary one. A
genre with no name may exist: that is, a set of music events that conform to some con‐
ventions established within a community, while the community itself has not given it
(and/or does not want to give it) a name at all; unless we decide that ‘music that does
not belong to any genre’ is actually a genre name, such a genre exists (a few festivals
11
celebrate it every year in some European countries.7 And, of course, many genres with a
name probably existed without one (for a short or long period), before getting one.
2.3 A definition (and further suggestions)
To sum up some of the issues discussed so far, and in order to investigate the diachronic
processes by which genres are created, changed and deleted, we need to consider the
definition of a musical genre as ‘a set of music events regulated by conventions accepted
by a community’. This is a compact form of a definition that has circulated for thirty
years,8 which benefits from the refinements I suggested above, namely that, first, ‘con‐
vention’ must be intended within the framework of David K. Lewis’s study, second,
prototypical effects and family resemblances can be intended as socially acknowledged
by convention, so they do not need to be referred to specifically, third, norms or rules
can also be omitted in the definition, because although laws and regulations act as regu‐
latory principles and may help in defining genres, they perform such a function within or
among genres only if they are made relevant by convention (for example, a juridical
norm can exist on paper, while being ignored by a community), and, fourth, ‘community’,
in the strictest sense, is the ‘population’ where a convention is established. Usually, the
community accepting genre conventions is the union or stratification of many diverse
communities. Can such a short definition be of any use for the study of music(s)? Can it
shed some light on the work of scholars? We will now see.
What does it mean for a music historian to study the birth of a genre? It means,
first, to examine any kind of document (manuscripts, scores, newspapers and maga‐
zines, letters, marketing material, recordings if available, etc.), with the help – when pos‐
sible – of direct witnesses, for the earliest traces of the genre’s name; second, to investi‐
gate, similarly, the genre’s community, and evaluate recurring behaviours, norms, codes
and prototypes (within the framework of other existing genres, so evaluating opposi‐
tional functions); and, third, to confront the chronologies of naming and other conven‐
tions, and to formulate hypotheses on a possible pre‐history of the genre. Of course, this
is what historians usually do.
7
8
I am thinking of the Angelica festival in Bologna, for example.
It was first formulated in Fabbri 1980.
12
But there is a strong tradition in musicology that favours notated music above all
other sources (not just because scores are more easily found,9 and, in the end, what
scholars often come up with is an evolutionary description of styles, rather than a his‐
tory of genres. There is much to learn from the methods of the École des Annales, as
suggested by Rick Altman in his study of silent film sound:
Just as the French ‘new’ historians of the Annales school found the texture of history in the
daily fabric of common lives, rather than in the exceptional existence of the ruling class, so
I find the most important lessons in the most common uses of sound located in the most
banal practices. Who plays the piano? Where is it located? Is it used just for the film or for
other parts of the program as well? And what kind of music is selected? (Altman 2004, 6)
The ‘daily fabric of common lives’ is made with conventions, and, as Altman sug‐
gests when discussing methodology, it is essential that historians look at past practices
as they were, without being influenced by their subsequent evolution (ibid., 17).
Musical historical investigation, I would add, requires a continuously shifting per‐
spective, an uninterrupted comparison among conventionally accepted practices before
and after a genre is born.
Investigating the present, which is what music sociologists, music anthropologists,
cultural scholars do, demands a similar approach. In order to be more than snapshots of
passing instants, demographic studies and ethnographies must give historical sense to
their specific context and circumstances, in particular to their time and place. And this
means, at least, to provide links, hooks to the past and the future, and to other places and
cultures. There is a kind of study of the present (whatever this is: I know that ‘present’
isn’t an easily definable concept!) that I would like to see: a real‐time monitoring of new
genre names and new conventions. For many contemporary genres, this might be per‐
formed on the Internet: a chronological listing of the number of web pages containing a
certain genre’s name (obtained from a search engine) is a very rough approximation to
9
An example: scholars of the Italian language and popular music were informed by a record producer
that the neologism cantautore was first used in the summer of 1960, in the press material of the Italian
subsidiary of RCA, to promote singer-songwriter Gianni Meccia; the word – according to interviews –
was also included in the liner notes of one of Meccia’s singles. None of the documents was ever
found. Just imagine how difficult it could be to find similar documents for earlier genres. See Cartago
2005, 317-328 and Tomatis 2009.
13
what I intend as a web‐based research method, but software engineers could develop
ways to trace new genre names before they become commonplace.
New conventions are a tougher matter, of course, but a large enough team could
successfully create the proper mesh to collect ‘early warnings’ for practices that seem to
emerge as new. To some extent this is what already happens sometimes in scholars’
blogs and mailing lists, like those set up by the International Association for the Study of
Popular Music (IASPM). However, a far quicker method needs to be found: when a
scholar writes to a mailing list asking for more information on some new music prac‐
tices, a new genre has existed long enough to make the request outdated.
Perceiving genres as sets of music events regulated by conventions accepted by a
community is not an obstacle to understanding diachronic processes, like those that can
be metaphorically described as the birth, change or death of a genre. Rather, such a defi‐
nition can help by describing more accurately than others (or better than no definition,
anyway) diachronic processes, and, thereby, may suggest methods for music historians,
sociologists, anthropologists and other scholars, be they interested in music(s) of the
past, or in current music practices. Here ‘community’ and ‘convention’ are the key con‐
cepts: an interdisciplinary approach to their understanding is fundamental for the study
of genre.
3. Conclusions: a research project
Having worked on genre theories for quite a while, and having faced the problems em‐
erging from intra‐disciplinary approaches that often seem to be scarcely attentive to the
real content of research in other disciplines, it occurred to me that something radically
new was needed: not just an interdisciplinary study, but a study aimed at creating
bridges among disciplines, in order to build common knowledge, and a shared language
to describe genre‐related phenomena. I will finish my overview by illustrating some de‐
tails of a research project.
The project’s aim is to combine the perspectives on music categorization of many
and sometimes ‘distant’ disciplines, like musicology, popular music studies, ethnomusi‐
cology and music anthropology, semiotics, sociology and computer science, capitalizing
on previous research, and learning from its successful results and from its flaws. Nothing
14
similar has been attempted to date. What is needed is a comprehensive interdisciplinary
model for the understanding of music categorization, with a degree of formalization that
will allow computer scientists to translate it into flow diagrams and programs, while re‐
specting the complexity of phenomena taking place in many different musical and soci‐
etal contexts. In the language of computer science this would be defined as the definition
of an ontology, that is, a full description (naming, typology, properties) of the concep‐
tualizations of phenomena and their interrelations in a domain of discourse (Gruber
1993). From an anthropological and sociological perspective, the target would be a
‘thick description’ (Geertz 1973) of music events, allowing the understanding of music
practices in the framework of their ‘living’ context. From a semiotic and anthropological
perspective, the study would focus on conventions agreed upon within communities
(Morgan 2014), and the musical, linguistic, behavioural, proxemic codes, etc., emerging
from that work (Fabbri 2007a, 2007b). For musicology, the challenge is to relate cate‐
gorizing processes and naming conventions to features of musical structure, and to their
historical and social meaning.
If successful, the project would have a considerable impact both on scholarship in
the disciplines involved, and on industrial sectors and media, for which music categori‐
zation is vital: the live music industry, the phonogram industry, radio and television, In‐
ternet. While a better comprehension of music categorization processes is expected to
lead to a further integration of musicological disciplines, it could also improve greatly
the precision of music recommendation systems, offering important new resources to
web‐based music applications.
1. Categorization theories
In the first phase of the research project, different approaches to music categorization
should be compared. As I have shown, the debate about categorization processes per‐
formed by humans in classifying music is still an open issue. Two main theories can be
distinguished, the first one based on conventions, and the second based on prototypes;
some studies (Guaus 2009) debate about whether the categorization process is a ‘posi‐
tive’ attitude (i.e. one is aware of some given rules or prototypes and decides whether a
given music event belongs to a category or not) or ‘negative’ (i.e. one is not sure to which
15
category a given music event does belong, but is sure about which categories it doesn’t
belong to).
2. Genre descriptors
A genre descriptor is any information that may help in the categorization of music. At
first glance, two main families of descriptors can be observed: descriptors related with
different facets of music, that is, descriptors related to melody, harmony, rhythm or tim‐
bre, or other ‘intra‐musical’ parameters, and descriptors derived from the social implica‐
tions of music (that is, ‘para‐musical’ factors), like, fashion, age, economic interests, and
use in social networks, among others (Sordo 2012). Not all musical genres are charac‐
terized by the same type of descriptors. Some genres may be characterized by some
musical features, while others are determined by the calendar (e.g. Christmas music in
Western cultures), or other industrial criteria (Pachet and Cazaly 2000). It must be con‐
sidered that a descriptor that can be crucial for one specific category can be considered
as ‘noise’ for other categories. Including all descriptors in all categorization processes
can lead to a decrease in the accuracy of the model, unless proper weighing factors are
introduced.
3. Time evolution
As I have discussed above, categories in music are in constant evolution. Many musical
genres emerge every year (for example in so‐called ‘electronic music’), some others die.
As has occurred since early times, music’s evolution has been directly related to tech‐
nologic development. So, in a given historical period, some contemporary music catego‐
ries evolved in parallel paths. Even the creation of music genres can be related to spe‐
cific technology, so specific music descriptors derived from that may not represent a
musical category but a musical period.
The goal is to connect diachronic development in music categorization to music
histories, in order to build a model of how genres work through time. How do changes in
music practices and technologies, as well as in the media landscape, affect categorizing
processes? How can categorization be understood in its wider cultural and historical
context?
16
To answer these questions, further investigations into past categorization are
needed. Sources can include music publications from the past, including magazines,
music press, popular press, newspapers; TV and radio broadcasts; critical essays and
music books; advertisements and press releases; etc. Enquiries must encompass differ‐
ent music cultures and their specific practices, in order to challenge any simplistic An‐
glo‐American‐centric perspective. Also, folksonomies and labelling practices (Lamere
2008) – which have developed strongly thanks to the Internet over the past twenty
years – must be addressed from the point of view of their diachronic and diatopic devel‐
opments.
4. ‘Borders’
If one asks performers/composers which genre labels could be applied to their activity,
probably they will start by presenting similarities to one or another strict convention
(e.g. musical genre) or prototype (e.g. an artist name, or an existing work). In general,
only few musicians agree that they belong to one of the established music genres. Topo‐
graphic metaphors are often used in describing genres, and since such metaphors are
two‐dimensional (while genres aren’t), ‘borders’ have to be described as ‘fuzzy’; most
musicians would probably like to be geo‐located close to these fuzzy ‘borders’ between
three or four ‘countries’. There appears to be a contradiction between the attitudes of
various musical communities (for example, musicians versus listeners/fans) towards
the strictness of genre ‘rules’.
5. Knowledge
In machine learning and statistical analysis areas of knowledge, different strategies and
algorithms can be applied. In the MIR community there is a long tradition for organizing
contests to evaluate the results of different algorithms in relation to different fixed
datasets. Evaluation is based on accuracy (precision, recall, f‐measure, etc.) and comput‐
ing time. Some of the best‐rated algorithms are built using, for instance, neural net‐
works, which do not allow an easy comprehension of the model itself. Nevertheless, to
my knowledge, there are no initiatives attempting to identify the key parameters that
can explain the evolution of musical genres over time.
17
6. Datasets
As mentioned earlier, the choice of the dataset is crucial for MIR research: biased
datasets bring to biased results. Similarly, all musicological, ethnographic, sociological,
surveys (etc.) are subject to errors due to the lack of representativeness of samples,
weak methodological control, and so on.
The problem of musical categorization can be approached at different levels of de‐
tail. For example, trying to understand the taxonomy proposed by the musical industry
(record companies, iTunes, Amazon, etc.), is not the same thing as trying to explain the
evolution of electronic (dance) music in Northern Europe during the last five years.
Since a high level of detail is important, descriptors derived from audio or social sources
must have an equivalently high degree of specificity (Herrera 2002). Moreover, not all
music listeners move in the same social circles. Some of them buy CDs, others make use
of streaming services provided by different companies, while others only go to concerts.
Descriptors derived from social sources may often include naive generalizations and
need to be treated with particular care, because some collectives may engage in certain
social activities while others do not. A descriptor dataset is biased in its construction.
7. Labels and tags
Since the age of cassettes, people have been able to create their own musical collections
according to specific criteria. In some cases, these homemade compilations or playlists
are selected according to non‐standardized labels such as ‘music for driving’ or ‘house
cleaning’ (Sordo, Gouyon and Sarmento 2010). Fixed taxonomies are not always useful
to the end user beyond buying music in a physical or online store.
Would it be possible (as suggested by Marino 2013) to make a meta‐taxonomy of
labels? For example: those based on (prosaic) function/use (‘music for doing x by’, e.g.
elevator music, seduction music, music to impress people by, various types of music for
dance, background music, foreground music, film music, concert music, etc.); those
based on musician understanding of difference; those defined by user group/population;
those based on aesthetic preference/values; those based on cultural habitat; those based
on musicological observation; those based on computer analysis.
Cultural and linguistic factors have to be taken into close consideration; until re‐
cently, research on social tagging has been focused on English (mostly American Eng‐
18
lish) terms. But genres ‘exist’ (i.e. are used as concepts) in many cultures, and are label‐
led with terms belonging to many languages. Are such labels translatable? And, con‐
versely, do the same transnational terms (like ‘pop’, ‘rock’, ‘opera’, ‘Lied’, etc.) have the
same meaning in different cultures?
8. Categorization and music theory
The part of the study dealing with music theory will face several aspects of technical
elements. Musical languages, depending on musical genre or situation, do identify simi‐
lar technical elements with different terms, revealing ideological orientations (Tagg
2013): a study of those technical elements, their occurrences and their names will lead
to the creation of a metalanguage and, possibly, to a reform of musical terminologies. It
is also important to investigate how similar elements – melodic, harmonic, timbral,
rhythmic, in the mix – ‘declined’ in different situations and played a part in the creation
of different musical genres; these ‘musical vocabularies’ will need to be analysed.
Another fundamental aspect – never studied before – is the categorization of the
broad field of ‘classical’ music: music genres do exist in euro‐classical music too, but they
seem to be overwhelmed by the huge amount of genres in popular music. In the classical
sphere, musical and para‐musical activities varied considerably, depending on the venue
and the genre, and audiences do tend, still today, to distinguish between genres in what
we now call ‘classical’ music.
The final interdisciplinary model will result from the integration of models derived from
the salient points above, with the aim to overcome differences in disciplinary metalan‐
guages, which are obviously apparent now. The project, as stated earlier in this paper, is
aimed at ‘building a bridge’ between musicology (with related SSH disciplines) and
computer science: both are expected to benefit greatly from the construction. The pro‐
posal of a new tagging protocol for audio files is a feasible outcome of the project: if ad‐
opted, it would certainly affect the circulation of music currently disadvantaged by im‐
proper tagging (classical music, popular music and traditional music from non‐
Anglophone countries), and favour the usage of digital resources in music education.
19
Acknowledgements
I would like to thank warmly Marta García Quiñones (Barcelona), Enric Guaus (Bar‐
celona), Jacopo Tomatis (Turin), Jacopo Conti (Turin), Gabriele Marino (Turin), Goffredo
Plastino (Newcastle‐upon‐Tyne), and Philip Tagg (Huddersfield) for their precious sug‐
gestions and corrections. Of course, the responsibility for any remaining mistake or lack
of clarity is mine.
References
Altman, Rick, 2004, Silent Film Sound, New York: Columbia University Press.
Anderson, Benedict, 1983, Imagined Communities, London and New York: Verso.
Basili, Roberto, Serafini, Alfredo, and Stellato, Armando, 2004, ‘Classification Of Musical
Genre: A Machine Learning Approach’, ISMIR.
Bourdieu, Pierre, 1992, The Logic of Practice, Stanford: Stanford University Press (first
edition in French 1980).
Brackett, David, 1995, Interpreting popular music, Cambridge: Cambridge University
Press.
Brackett, David, 2002, (In search of) musical meaning: Genres, categories and crossover,
London: Arnold.
Cartago, Gabriella, 2005, ‘Cantautore, canzone d’autore e le voci della popular music
nella lessicografia non specializzata’, in Lingua letteraria, delle arti e degli artisti, Flor‐
ence: Franco Cesati Editore, 317‐28.
Celma, Óscar, 2010, Music Recommendation and Discovery. The Long Tail, Long Fail, and
Long Play in the Digital Music Space, Berlin and Heidelberg: Springer Verlag.
Cichowlas, Alison, and Lam, Tony, 2014, ‘Explore the history of Pop ‐‐ and Punk, Jazz,
and
Folk
‐‐
with
the
Music
Timeline’,
http://googleresearch.blogspot.it/2014/01/explore‐history‐of‐pop‐and‐punk
jazz.html, last accessed 19 October 2014.
Downie, J. Stephen, 2008, ‘The Music Information Retrieval Evaluation Exchange (2005‐
2007): A window into music information retrieval research’. Acoustical Science and
Technology 29 (4): 247‐255.
20
Downie, J. Stephen, Ehmann, Andreas F., Mert Bay and M. Cameron Jones, 2010. ‘The
Music Information Retrieval Evaluation eXchange: Some Observations and Insights’.
Advances in Music Information Retrieval Vol. 274: 93‐115
Eco, Umberto, 1976, A Theory of Semiotics, Bloomington: Indiana University Press.
Eco, Umberto, 1999, Kant and the Platypus: Essays on Language and Cognition, London:
Secker & Warburg (first edition in Italian 1997).
Fabbri, Franco, 1980, ‘Musica tra le musiche’, Laboratorio musica, 14‐15, luglio‐agosto
1980, Milan: Ricordi‐Arci: 42.
Fabbri, Franco, 1981, ‘I generi musicali. Una questione da riaprire’, Musica/Realtà, 4,
Bari: Dedalo: 43‐66.
Fabbri, Franco, 1982, ‘A Theory Of Musical Genres. Two Applications’, Popular Music
Perspectives, ed. D. Horn & P. Tagg., Göteborg and Exeter: IASPM: 52‐81.
Fabbri, Franco, 2007a, ‘Browsing Music Spaces: Categories and the Musical Mind’, Criti‐
cal essays in popular musicology, ed. A. Moore, Aldershot: Ashgate: 49‐62.
Fabbri, Franco, 2007b, ‘Tipi, categorie, generi musicali. Serve una teoria?’,
Musica/Realtà, 82, Lucca, Milano: LIM‐Euresis: 71‐86.
Fabbri, Franco, 2008, ‘Genre in Popular Music, by Fabian Holt’, Popular Music, 27(3):
490–92.
Fabbri, Franco, 2012a, ‘How Genres Are Born, Change, Die: Conventions, Communities
and Diachronic Processes’, Critical Musicological Reflections, ed. S. Hawkins, Alder‐
shot: Ashgate: 179‐191.
Fabbri, Franco, 2012b, ‘Rebetiko as a Testing device for Genre Theories and Musical
Categorizing Processes’, Journal of Mediterranean Studies, 21/2, Malta: Mediterranean
Institute, University of Malta: 321‐332.
Geertz, Clifford, 1973, The Interpretation of Cultures: Selected Essays, New York: Basic
Books.
Giltrow, Janet, and Stein, Dieter (eds.), 2009, Genres in the Internet. Issues in the theory of
genre, Amsterdam and Philadelphia: John Benjamins Publishing Company.
Gruber, Thomas R., 1993, ‘A translation approach to portable ontology specifications’,
Knowledge Acquisition, Volume 5, Issue 2: 199–220.
Guaus, Enric, 2009, Audio content processing for automatic music genre classification: de‐
scriptors, databases, and classifiers, PhD thesis.
21
Herrera‐Boyer, Perfecto, 2002, ‘Setting Up an Audio Database for Music Information Re‐
trieval Benchmarking’, ISMIR.
Holt, Fabian, 2007, Genre in Popular Music, Chicago: University of Chicago Press.
Kaufman Shelemay, Kay, 2011, ‘Musical Communities: Rethinking the Collective in
Music’, Journal of the American Musicological Society, Vol. 64, No. 2: 349‐390.
Lakoff, George, 1987, Women, Fire, and Dangerous Things. What Categories Reveal about
the Mind. Chicago and London: The University of Chicago Press.
Lakoff, George, and Johnson, Mark, 2003, Metaphors We Live By, Chicago and London:
The University of Chicago Press (first edition 1980).
Lamere, Paul, 2008, ‘Social Tagging and Music Information Retrieval’, Journal of New
Music Research, 37:2, 101‐114, DOI: 10.1080/09298210802479284
Levitin, Daniel, 2008, This Is Your Brain On Music, London: Atlantic Books.
Lewis, David K., 2002, Convention, a Philosophical Study, Oxford: Blackwell (first edition
1969, Cambridge: Harvard University Press).
Marino, Gabriele, 2013, ‘“What kind of genre do you think we are?”. From technique to
lyrics, “genre definers” within music intermedial ecology’, paper delivered at the 12th
International Congress on Musical Signification (ICMS12), Université catholique de
Louvain, Louvain‐la‐Neuve, April 4, 2013.
McKay, Cory, 2004, Automatic Genre Classification of MIDI Recordings, McGill University,
Montreal, Master thesis
McKay, Cory, Fujinaga, Ichiro, 2005, ‘Automatic Music Classification and the Importance
of Instrument Identification’, Proceedings of the Conference on Interdisciplinary Musi‐
cology, Montreal.
Molnar‐Szakacs, Istvan, and Overy, Katie, 2006, ‘Music and mirror neurons: from motion
to ’e’motion’, Social cognitive and affective neuroscience 1.3: 235‐241.
Morgan, Marcyliena H., 2014, Speech Communities. Key Topics in Linguistic Anthropology,
Cambridge: Cambridge University Press.
Negus, Keith, 1999, Music Genres and Corporate Cultures, London: Routledge.
Pachet, François, and Cazaly, Daniel, 2000, ‘A taxonomy of musical genre’, RIAO.
Rizzolatti, Giacomo, 2005, ‘The mirror neuron system and imitation’, Perspectives on Imi‐
tation. From Neuroscience to Social Science, ed. S. Hurley, N. Chater, Cambridge (MA):
MIT Press, Vol. 1: 55‐76.
22
Santoro, Marco, 2010, Effetto Tenco. Genealogia della canzone d’autore, Bologna: Il Mu‐
lino.
Sordo, Mohamed, Celma, Óscar, Blech, Martin, and Guaus, Enric, 2008, ‘The quest for
musical genres. Do the experts and the wisdom of the crowds agree?’, ISMIR: 255‐260.
Sordo, Mohamed, Gouyon, Fabien, and Sarmento Luis, 2010, ‘A Method for Obtaining
Semantic Facets of Music Tags’, 1st Workshop On Music Recommendation And Dis‐
covery (WOMRAD), ACM RecSys, Barcelona
Sordo, Mohamed, Gouyon, Fabien, Sarmento, Luis, Celma, Òscar, & Serra, Xavier, 2013,
‘Inferring Semantic Facets of a Music Folksonomy with Wikipedia’, Journal of New
Music Research, Vol. 42, No. 4: 346‐363.
Sordo, Mohamed, 2012, Semantic Annotation of Music Collections: A Computational Ap‐
proach, PhD dissertation, Universtat Pompeu Fabra, Barcelona.
Sturm, Bob L., 2012, ‘A survey of evaluation in music genre recognition’, International
Workshop on Adaptive Multimedia Retrieval.
Tagg, Philip, 2013, Music’s Meanings. A Modern Musicology for Non‐Musos, New York and
Huddersfield: The Mass Media Music Scholars’ Press.
Taylor, Timothy D., 2014, ‘Fields, Genres, Brands’, Culture, Theory and Critique, 55:2,
159‐174, DOI: 10.1080/14735784.2014.897242
Tomatis, Jacopo, 2009, ‘Ideology, neologisms, neo‐realistic aesthetics: Definition of a
musical genre. The case of Italian cantautori’, paper presented at the International
Conference on Popular Music Studies, Liverpool, UK.
Tzanetakis, George, and Cook, Perry, 2002, ‘Musical genre classification of audio signals’,
Speech and Audio Processing, IEEE transactions on 10 (5), 293‐302.
23