Advertising and Web Hosting on Woodwind.Org!

Klarinet Archive - Posting 000572.txt from 2001/05

From: (Tony Pay)
Subj: [kl] The 'cocktail party' theory of classical music
Date: Thu, 24 May 2001 16:05:34 -0400

Reactions to this, anyone?


Playing classical music 'stylishly' -- an evolutionary argument, and the
'cocktail party' theory of classical music

What does it mean, to play classical music in a way that is
'stylistically appropriate to the period in which it was written' (or,
as I shall shorten the phrase for convenience, 'stylishly')?

Do we need to use period instruments? Or can we, as Nicholas
Harnoncourt claims, do just as well on modern ones? Even on modern
instruments, how much of our playing can still be 'our own'? Are we
condemned to be always considering what 'they' did? Anyway, how much
can we trust that 'they' followed their own treatises, given that our
own, modern musical treatises fall so far short of capturing how we
ourselves play?

All of these questions have been given different answers by different
people, so that it might seem that there are no objective conclusions we
can come to. I would like to argue here, though, that there is actually
an absolute sense in which we can play classical music stylishly. In
fact, I think it probable that the most important difference between the
way classical music is unstylishly played today, and the stylish way in
which it was played by the best players of the time it was composed, can
be deduced in just two moves of the imagination. The first move is to
take seriously the question of why and how our auditory processing
systems have evolved to be the way they are; and the second move is to
notice how the scores of the great classical composers are constructed
in order to take advantage of that evolution.

I mentioned this argument briefly in the course of an article I wrote in
May 1996 for the journal Early Music. That article, called Phrasing in
Contention, was a thought-experiment investigating the change of meaning
of composers' slurs (also sometimes called 'ligatures') between the time
of Mozart and the present day.

Early Music is available in any good library, but in order to make the
article more easily accessible, I recently organised that it be put on
the web at

However, I have come to think that the 'first move', namely, to consider
the evolution of our auditory processing systems, is more important than
I made it seem. So, this much shorter piece is intended to throw into
stronger relief what is implicit in the idea, which is that the
'classical style' is not merely arbitrary or culture-dependent, but is
instead determined in a much deeper sense by our human nature, which of
course we share with people of quite different cultural backgrounds.
What happens to be stylish in classical music is therefore not merely a
way of playing that was adopted for a short time by musicians in the
eighteenth century. If it were only that, then discovering and
following the details of its practice would be little more than just
being fashionable, or trendy, in retrospect -- and indeed, so-called
'authentic' performances have often been accused of just such
trendiness. The aspect of stylishness that I am concerned with -- the
aspect that I claim is independent of any particular musical culture in
which it is used -- is instead a powerful and general attitude that has
universal value. It can be used to great effect in many other sorts of
music. It is often what is required to make complex twentieth-century
music work well.

In writing Phrasing in Contention, I first tried to imagine in
detail what the contemporary meaning of classical slurs had been,
through considering what sort of interpretation would make most evident
the structure of the best of the music of the time. Second, drawing at
least partly on personal experience, I discussed how it might be
possible to find a way of playing that would allow us to read and
represent the slurs in this way. It turned out that finding that
involved the adoption of a view of phrasing that I termed,
'beginning-oriented'. However, I argued in the article that adopting
this view of phrasing does not compromise our own naturalness and
expressivity. The argument is quite long; but that is because some
players find beginning-oriented phrasing counterintuitive, in the sense
that it seems to go against their natural instincts. Those players, in
difficulties themselves, are then surprised to discover that some
others, on the contrary, see 'beginning-oriented phrasing' as a
perfectly natural point of view, and cannot understand what all the fuss
is about. It was necessary in the article to talk at some length about
why there is a difficulty in moving from one viewpoint to the other.

I want here to explain what lies behind what I shall (rather
frivolously) call the 'cocktail party' theory of classical music. This
is a very simple idea. I hope that interested readers who would like to
know about the more complex details will follow the Web link (or look at
a copy of Early Music, May 1996) for themselves.

The 'cocktail party' theory of classical music derives its name from the
fact that we are able at a cocktail party to hear not only the
conversation in which we are engaged, but also, if we want, other
conversations that are going on around us. I would say that what is
required to perform the best classical music well bears several striking
similarities to what occurs at such a party.

The reason is that the individual lines of a classical piece are written
in such a way that each line is interesting in itself. Such lines often
contain motivic references to each other, and so reflect the underlying
unity of the composer's conception. The scores of the best classical
pieces demand that we should be able to pick out such lines from the
overall texture even as we respond to higher levels of organisation,
such as the harmonies, and on a higher level, shifts of tonality.
Perhaps no-one can hear all the individual lines simultaneously, but
each of those lines should be at least potentially available to us; the
possibility then exists that we may switch our perception between them,
even at a rate that creates for us the illusion of simultaneity. For
this to be even conceivable, the piece needs to be played in a way that
allows all the important individual lines to be heard. That is why, in
the preparation of a serious performance of a classical piece, a very
large part of the effort of both players and directors is devoted to
making the important parts audible.

But -- and this is the message of this article -- there is a great deal
to be learnt from the very simple example of what occurs in the normal
speech of people at a cocktail party. Understanding that phenomenon
makes the problem of balancing a performance much more tractable, and
explains why playing that follows the model of speech is much more

Speech has evolved under the twin constraints of the need to be audible
and the need to be recognisable, even in a noisy environment, so as to
maximise the probability that our ears and auditory processing systems
identify its message and pass it on to our other systems for possible
action. The survival value of clear communication can hardly be
doubted. You could almost say that the whole point of speech is that it
be both audible and recognisable. (Not to be able to hear, and
understand, "Look out behind you!" is to be at a considerable
evolutionary disadvantage.)

Music and language have often been compared, with varying degrees of
success. But the comparison I want to make here is much more
straightforward and basic than most such comparisons. My analogy is only
between the surface structure of speech itself and the surface structure
of classical phrases. The analogy is apt because the property of being
both audible and recognisable in a noisy environment is what is required
in both domains. Not all music relies on such clarity, it's true. The
intention of a composer may sometimes be to suppress the details of an
accompaniment. But in the music of the best classical composers, the
perception of individual lines is very often of crucial importance.

Now, one aspect of speech is that its syllables, in all languages, are
what I shall call, 'beginning-oriented'. That is, in some sense, they
begin more energetically, and lose intensity as they progress. This
particular type of modulation, a clear beginning plus a falling away,
contributes significantly to the audibility of spoken words. It is not
an all-or-nothing phenomenon: it admits of degrees. For example, if we
wish to make ourselves more heard or understood against the odds, we
automatically increase the degree to which what we say is
beginning-oriented. It is not time-symmetric either; a recording of
speech played backwards doesn't sound like speech.

However, that all languages have 'beginning-oriented' syllables doesn't
mean that 'beginning-orientation' is a simple thing to characterise. (I
have been purposely vague in describing it.) In fact, the lack of such
a simple characterisation was notoriously part of the difficulty in
programming automatic speech-recognition software. Researchers found it
very difficult to make explicit the rules that would allow their
programs to segment the continuous sound of speech into syllables, in
order to match them with the syllables that constitute dictionary words.
That humans find this problem trivial in practice shouldn't blind us to
the subtlety of what's involved in solving it. We just have very good
speech-production and speech-recognition software 'built in'. And
co-ordinately, we shouldn't be surprised that the corresponding musical
techniques are difficult to characterise in detail too, even though
young players often learn to execute them quite naturally.

It is actually the quality of being both easy to hear, yet difficult to
pin down in prescriptive detail, that makes this aspect of the
connection between speech and music profound rather than trivial. If
being 'spoken' had a simple, unequivocal character, the subtlety and
variety of musical expression would not be matched by the subtlety and
variety of speech, and the analogy would break down. 'Spoken' phrases
would just be the class of phrases that had that one particular
character. The truth is precisely the contrary: spoken phrases can be
subject to an infinite variety of nuance, and can actually be more or
less 'spoken', shading into the quality of being 'sung'.

And there is a further simple, yet overwhelmingly important bonus for
musical performance that we may obtain by following the structure of
speech. This bonus is the most important part of the 'cocktail party'
theory. It is that in the cocktail party, what allows us to be heard
against the background noise created by the speech of the other guests
is precisely what allows their speech to be better heard against us.

To appreciate this, recall that, at such a party, we can pick out a
nearby conversation that happens to be interesting, and pay attention to
it even as we continue to listen to the conversation of the person we
are at present engaged with. The audible syllable-beginnings of one
conversation don't consistently coincide with the audible
syllable-beginnings of the other, and so each syllable 'gives space',
allowing the other to be heard. Even if, by chance, two
syllable-beginnings do happen to coincide, the fact that
syllable-lengths vary, often quite widely, soon destroys the synchrony.

So, we see that if we adopt the analogous musical structure, the whole
texture of a classical score is clarified. If everyone 'speaks' the
music, to a greater or lesser extent, then much of the balance is
automatically taken care of.

Now, in music, the composer has control of the degree to which phrases
are synchronous. If we look at the music of the great classical
composers, it is easy to recognise examples of the use of both
synchronous and asynchronous phrasing between different instruments, in
order respectively to obliterate or enhance the perception of individual
lines. But this can only work if the performers have a common,
beginning-oriented view of phrasing that they generally follow.
Classical style must include that beginning-oriented structure as a
built-in normative element. We may not always follow it -- indeed, part
of the expressivity of any style lies in how in execution its norms may
be varied, or even contradicted. You cannot, though, effectively vary
what isn't clearly perceived to be there. (Again, both this and other
related issues are explored more fully in Phrasing in Contention.)

Of course, none of this is particularly original. There are many
players whose playing embodies precisely the attitude towards musical
performance that I have characterised. But there are others who find it
strange, and still others who think that it is merely a quirk of music
of the seventeenth and eighteenth century. ("That backward phrasing," I
once heard a colleague call it.) Perhaps by talking about it in a
different way I may have reduced that strangeness for them.

To sum up:

(1) The clear structure of classical music demands that we have
available the clear structure of 'spoken' execution.

(2) This structure admits of expressive variation, which is important
because a crucial part of musical performance is the expressive
variation of stylistic norms.

(3) But, you can't expressively vary what isn't perceived to be there.

(4) The structure of speech is designed, by evolution, to guarantee that
we perceive what is there; therefore, that structure is fundamental in
the performance of classical music; it is important in much other music

© Tony Pay, November 2000

I would like to thank Jan Schlapp for helpful comments on an earlier


_________ Tony Pay
|ony:-) 79 Southmoor Rd
| |ay Oxford OX2 6RE GMN artist:
tel/fax 01865 553339

... Adults are just kids who owe money.

Unsubscribe from Klarinet, e-mail:
Subscribe to the Digest:
Additional commands:
Other problems:

     Copyright © Woodwind.Org, Inc. All Rights Reserved    Privacy Policy    Contact