Back to Schedule for Joint Sessions
Notions of "representation" permeate discussions
of music in highly diverse intellectual quarters. The question
of what (if anything) music represents has been a preoccupation
of aesthetic theories of Western music for well over a century.
The question of how music may represent anything external to itself
has engendered much recent discussion, particularly in parallel
with the rise of semiotics and cognitive studies. These subdisciplines,
although entirely distinct from each other, give far greater attention
to mental processes than earlier theories which concentrated exclusively
on the music itself.
An apparently different set of questions emerges in computational
studies, where the question of how music itself may be represented
for symbolic processing reveals the binding influence of all representational
systems on the results of queries dependent upon them. Such questions
have had significant influence on the growth of cognitive theories,
because actual applications reveal how devoid of value the results
of psychologically deficient representations may be. At the same
time they force the enquirer to choose between performance and
notation as a basis for theorizing about music. A similar dichotomy
is evident when confronting temporal processes with intellectual
constructions of them. Inherent contradictions call into question
the validity of such constructions. Four papers are offered as
a basis for broader discussion.
The question whether and what represents was enormously exaggerated
through much of the twentieth century. Here are four intellectual
obstacles and some proposed rectifications:
1. The idea of absolute music-an absolute category of music which
refers to nothing or refers only to itself. (Carl Dahlhaus has
already done much to dismantle this too-neat package, showing
us that the impulse to resist mimesis and verbal paraphrase in
the nineteenth century was itself a sign of transcendence.) Many
sign systems explore elaborations of structure which supersede
their original representational functions. We can understand musical
abstraction as a development comparable to these.
2. The idea that we can not incorporate brute cause and effect
schemas in our understanding of subtle musical representation.
Our physical responses to music are not fundamental to what music
means, but fundamental to how music means.
3. The idea that music has no meaning because we can't translate
it and the idea that we should not try to. This notion accords
an excessive privilege to language in deciding what counts as
meaning. Yet, what can not be translated (e.g., a poem) may still
be described.
4. The idea that differences between the understanding of the
producer of music and the understanding of the receiver of music
preclude musical communication.
Compared to other arts music is customarily viewed as paradigmatically
temporal and non-representational. While the first characteristic
is generally lauded, the second is frequently a source of some
embarrassment for the scholar who deals in representation. I propose
that temporality and problems of representation are intimately
linked and that rather than regret music's representational deficiencies
we might more profitably look to musical practice for a critique
of the notion of representation in general. I will relate both
of these propositions to questions of musical analysis because
it is here that problems of representation often seem most acute.
From a temporal or processive point of view music will be understood
as actual event-evanescent, transitory, subject to passage. From
this perspective "present" does not mean "standing
before" us stable and complete; rather, "present"
will mean ongoing, in the process of becoming, as yet incomplete.
To analyze process we must arrest it in order to name its parts
and judge their relationships; but so arrested, it can no longer
be a process.
In order to fix or arrest the event, we are forced to make a model
based on whatever characteristics interest us. In short, we must
construct some sort of re-presentation. But if the temporality
of an event is essential to what the event is, our abstraction
from process may, in fact, be misrepresentation. Through our efforts
at representation, we run the risk of losing a clear idea of the
actual event to which the representation might correspond. This
is always a problem for intellectual analysis if we take time
and passage seriously. Although music is not more temporal and
not necessarily less representational than many other of our activities,
music by its very resistance to our attempts to hold onto it offers
valuable opportunities to question our faith in representation.
Encoding schemes which represent selected features of music
symbolically are essential to all software for processing music.
Although some debate is directed towards the selection of features
which may influence outcomes, little debate is addressed to the
more fundamental question of which domain-sound or notation-shall
form the basis of a representation scheme. The choice has profound
consequences for the results of later processing, however. The
relationship between notation and sound is necessarily selective
since not all features of sound are indicated in common notation;
conversely many features of notation have no meaning in sound.
Much speculative music theory of the past century is tacitly grounded
on notational representations of music, whereas musical data used
in computer-assisted analysis may come from either domain. Since
notation is itself a method of representation, encoding schemes
based on it are necessarily meta-representations and theories
based on it are in a sense secondary. Yet the ephemeral nature
of sound introduces much instability in evaluations based on it.
The dichotomy between sound and notation in computer applications
may itself represent a different order of problems implicit in
historically grounded musical analysis, for it would appear that
at certain junctures of musical history (e.g., the early Baroque)
a system of music-making based on signs on the page operated in
direct opposition to a system of sounds in the air.
Representing Music: A View From Cognitive Science
Lawrence Zbikowski (The University of Chicago)
Early work in cognitive representation (from the 1950s) invested
heavily in a computational/ symbolic approach and projected impressive
returns within five to ten years. As Hubert Dreyfus and others
have shown, such was not to be the future of cognitive science:
the challenges of representing the outside world together with
the complex interactions of the human nervous system (including
the brain) presented formidable problems not easily addressed
by the computational/symbolic approach. Music is a case in point:
isolating "appropriate" musical phenomena for study
is hardly a simple task. A thorough understanding of how humans
process musical information must extend beyond the auditory system
to include the physiological response to music and music making,
and some account of how musical structures are correlated with
those of other expressive domains.
Recent work in embodied cognition (e.g., by Lakoff and Johnson)
together with work in perceptual symbol systems (Barsalou et al.)
suggests more promising avenues for modeling cognitive representation.
In this paper, I shall explore what this research has to offer
music scholarship by considering a number of issues associated
with musical representation, including the following:
1. the problem of "representing" a non-linguistic and
ephemeral cultural product;
2. the correlation of representations of music with representations
of other expressive domains;
3. the explanatory limitations of musical representations; and
4. whether music is a representational system distinct from natural
language and capable of unique representations of non-musical
phenomena.
By way of illustration, the opening of Schubert's "Erlkönig"
will be used.