My piece on the cognition of atonal music has appeared in Prospect. I’m happy to say that it’s part of the site’s free content, but here it is below in extended, pre-edited form. I’m glad to see that it is drawing some comments. Not everyone is going to like what it says, for sure. But Tali Makell gets the wrong end of the stick – I’m not attacking the music of Schoenberg and his school. I’m a fan of most of this music, and think Berg’s Lyric Suite is a masterpiece. And of course I never said that Schoenberg et al. used total serialism. It’s absolutely right that they used some of tonality’s organizing principles, and to great effect. But there are problems with that in itself, as Roger Scruton has pointed out, since some of these structural principles are a direct consequence of tonality itself, and lose their meaning when taken out of that context. Besides, my view is that Schoenberg’s serialism was here acting as a convenient compositional tool that made it a little easier for the composers to use atonality – basically it was a scheme that reduced the effort needed to avoid tonality, and not one that actually brought any profundity or musicality in itself. These composers put that back in in other ways – with dynamics and so forth. There was nothing inevitable about Schoenberg’s method, and the justifications for it given by Adorno, and indeed by Schoenberg himself, are largely bogus. But that doesn’t make serialism intrinsically ‘bad’ as a way of composing. It’s when serialism becomes total that the problems really start (both musically and ideologically).
I hope all of this will be made clearer in my forthcoming book The Music Instinct. I am aware, however, that I have not addressed either here or there the defence of serialism made by Morag Josephine Grant in her book Serial Music, Serial Aesthetics (CUP). Here she attacks Fred Lerdahl’s critique of serialism (and other modern compositional methods), which is based on his linguistic approach to music cognition. Lerdahl believes that music needs to have an audible ‘grammar’: as he says, ‘The musical surface must be available for hierarchical structuring by the listening grammar’. In short, if we can’t construct hierarchies of pitch, rhythm and so forth, then the ‘musical surface’ is shallow: everything just sounds like everything else. This is the thrust of my Prospect piece, and I think it is true: for music to be more than a collection of mere sounds, there needs to be some audible way of organizing it. Grant complains that Lerdahl is forcing music into a straitjacket: she says his ‘argument ‘ that musical language, like spoken language, is generative in structure excludes the possibility of other, non-hierarchical methods of achieving musical coherence.’
But Grant is not, as it might seem, rejecting the need for music to acknowledge cognition. Rather, she asserts that this was precisely the concern of the serialists. They, however (she says), felt that our perception was culturally conditioned, and that we have ‘the ability to develop or uncover previously suppressed abilities’. One of those is the recognition of the tone row itself: ‘the use of the row is itself a constraint, not just on the composer, but in the aid of comprehensibility as well.’ Lerdahl is unaware of this, Grant says, while serialism imposes a system precisely to aid cognition.
And this is where Grant’s thesis utterly falls apart. For it has been shown quite clearly now that serialism’s ‘system’ is not one that can be cognized. It exists only on paper; listeners simply don’t hear it. And the reason for this is that it is simply not the kind of system that the mind intuits: we don’t listen to music by remembering arbitrary sequences of notes, but rather – as Lerdahl says – we organize those notes into hierarchical patterns: hierarchies of pitch, rhythm, melodic contour and so forth.
Can musical coherence be achieved by non-hierarchical methods? I’m not sure anyone knows, but certainly Grant provides no evidence of this – it is an act of faith. And what, precisely, are those methods, if we must exclude the tone row itself as one of them? She doesn’t say. She does, however, say that ‘the intense concentration on the tiniest of fluctuations’ is ‘central to the hearing of serial music’. I’m not sure what this means: fluctuations in what? The fluctuations in rhythm are either rather traditional, as in Berg, or are so extreme (as in Boulez) that rhythm has no meaning. Fluctuations in pitch, as in deviations from the tone row, would not be heard even if they were permitted. Grant also says that serialism ‘has the ability to create structures specific to each of its utterances’. Again, make of that what you will. If it means that each composition, each tone row, creates its own private language, then we know what Wittgenstein had to say about private languages. Why can’t Grant just explain how we are meant to find coherence in serial music, other than by its utilization of traditional techniques in parameters other than pitch?
I suppose one might interpret her remarks as saying that serialism encourages us to focus on each event as an entity in itself, not as something embedded in a hierarchical grammar. That’s possible. It might be interesting. And it is what I refer to below as sound art. If that’s the intention, it seems to me to be an explicit admission that ‘coherence’ isn’t the aim at all. And to my mind, coherence is the one characteristic that music should possess – I don’t care if you ditch tonality, or rhythm, or melody, or harmony, so long as what remains is in some manner coherent.
Incidentally, and in the hope of not sounding horribly patronizing, I have a lot of time for what Grant is up to more broadly. Defending Stockhausen is a noble cause in itself, even if I don’t buy it, and anyone who does so while listening to Scottish folk music and rap wins my vote.
Anyway, here’s the piece:
Writer Joe Queenan recently caused a minor rumpus in the austere world of contemporary classical music by complaining about how painful much of it is. He called Luciano Berio’s 1968 Sinfonia “35 minutes of non-stop torture”, Stockhausen’s Kontra-Punkte (1953) like “a cat running up and down the piano”, Krzysztof Penderecki’s Threnody for the Victims of Hiroshima (1960) “belligerent bees buzzing in the basement”, and Harrison Birtwistle’s latest opera The Minotaur “funereal caterwauling”. A hundred years after Schoenberg, he said, “the public still doesn't like anything after Transfigured Night, and even that is a stretch.”
Inevitably, Queenan was lambasted as a reactionary philistine. Performances of ‘modern’ works like this well attended, his critics said. And while Queenan took pains to distance himself from the conservative concert-goers who demand a steady diet of Mozart and Brahms, his comments were denounced as the same old clichés. The problem is that, like most clichés, they become such by frequent use. Sure, these works will find audiences in London’s highbrow venues, but the fact remains that Stockhausen and Penderecki, whose works now are as old as ‘Rock Around the Clock’, have not been assimilated into the classical canon in the way that Ravel and Stravinsky have. When someone like Queenan has earnestly tried and failed to appreciate this ‘new’ music, it’s fair to ask what the problem is.
David Stubbs considers this important question in his new book Fear of Music (Zero, 2009) but doesn’t come close to answering it. His speculative suggestion – that music lacks an ‘original object’ that, in visual art, can become the subject of veneration or trade – clearly has little force, given that it must surely apply equally to Beethoven and Berio. Indeed, Stubbs’ analysis is part of the problem rather than part of the solution. Like economists trying to understand market crashes, he wants to place all the motive forces outside the system: his gaze never fixes on the music itself. To Stubbs, your responses to music are a function of your context and perspective, not of the music. His comparisons of visual and musical art assumes an equivalence that allows no possibility of their being cognitively distinct processes.
He is in good company. Social theorist Theodore Adorno’s advocacy of Schoenberg’s atonal modernism was politically motivated: tonality was the bastion of bourgeois complacency. To the hardline modernists of the 1950s and 60s, any hint of tonality was a form of recidivism to be denounced with Maoist vigour; Pierre Boulez refused for a time even to speak to tonal composers. American composer Milton Babbitt’s provocative 1958 essay ‘Who Cares if You Listen?’ argued that it was time for ‘serious’ composers to withdraw from public engagement altogether, while offering nothing in the way of explanation for the public’s antipathy to ‘difficult’ music (his included) except a belief that they were too ill-informed to understand it. After giving a lecture on the music of Boulez and Elliott Carter, the eminent pianist and critic Charles Rosen responded thus to a question from the audience about whether composers have a responsibility to write music that the public can understand: “On such occasions I normally reply politely to all questions, no matter how foolish, but this time I answered that the question did not seem to me interesting but that the obvious resentment that inspired it was very significant indeed.”
No one can deny that audiences are conservative, whether they be Parisians rioting at the première of the Rite of Spring in 1913 or punks lobbing bottles at art-rockers Suicide on tour with the Clash. And since questions like this one are often a coded demand that composers start writing ‘real music’ like Mozart did, Rosen can be forgiven some impatience. Stubbs is justifiable indignant that even fans of conceptual art will parrot trite witticisms about the ‘cacophony’ of much experimental music.
But the understanding of the cognitive mechanisms of music that has emerged in the past several decades implies that it is not enough to tell ingrates bemused by Stockhausen to knuckle down and try harder. Many musicologists accept a definition of music as ‘organized sound’ (ironically, since this was the description used by avant-garde electronic pioneer Edgar Varèse for his own musique concrète, a paradigm of all that is seen as distressing about ‘modern’ music). Yet sound does not become organized merely because the composer has used a system to arrange it. Sound is structured into music not on paper, nor even in the mind of the composer, but in the mind of the listener. So music is sound in which the organization is audibly perceptible, not just that in which it is theoretically present.
Our brains use rules of thumb, both learnt and innate, to arrange an acoustic signal into a coherent entity: to pick out key, melody and harmony, to identify rhythm and metre, and to create a sense of structure and logic. The traditional music of just about every culture on earth builds in elements that assist this decoding process. When we encounter unfamiliar music, we may need to adjust our decoding rules, or learn new ones, before we can truly hear it at all.
Chief among these rules are the ‘Gestalt principles’ identified by a group of German-based psychologists in the early twentieth century. Initially identified in visual processing, these principles help us make good guesses at how to interpret complex sensory stimuli. We make assumptions about continuity, for example: the aeroplane that flies into a cloud is the same one that flies out the other side. We group together objects that look similar, or that are close together. Although the Gestalt principles are not foolproof, they make the world more comprehensible. Both in sound and in vision, the ability to interpret sensory data this way must have had clear evolutionary benefits.
In music, this means that melodies that move in small pitch steps tend to sound unified and ‘good’, while ones with large pitch jumps are liable to seem fragmented and harder to make out. Traditional melodies in diverse cultures do indeed proceed mostly in small rather than large pitch steps. Regular rhythms also contribute to coherence, while erratic ones are apt to confuse us.
The composer’s job is to manipulate the expectations that these principles produce – enough to avoid predictability and create a lively musical surface, but not so much as to lose coherence. Out of the interplay between expectation and reality comes much of music’s capacity to excite and move us. But what happens if these rules are seriously undermined? In Boulez’s Structures I or Stockhausen’s Klavierstück VII, say, there is no discernible rhythm, and the ‘melody line’, if one can call it that, is as jagged as the Dolomites. In this situation, we can develop no expectations about the music, and this absence of an audible relationship between one note and the next cuts off a key channel of musical affect. What remains may be a temporarily diverting sound, but the indulgent listener risks becoming like the sentimental audiences about whom nineteenth-century music theorist Eduard Hanslick complained, wallowing in the sonic surface while oblivious to the musical details.
And yet how can Structures lack structure? It is one of the most ‘structured’ pieces of music ever written! It was composed using the technique of ‘integral serialism’, in which musical parameters such as pitch, dynamics and rhythm are prescribed along the lines of Schoenberg’s ‘twelve-tone’ method, introduced in the 1920s. In Schoenberg’s original formulation, this approach dictated only the choice of pitches, and it was meant to eliminate all vestige of tonality – the anchoring of a piece of music to a tonic centre, which enables us to assign it a particular key. Schoenberg considered that tonal music – which meant all Western music until that point – had become tired and formulaic, and serialism was supposed to offer a systematic way of composing atonally.
The composer first chose a tone row: all twelve notes of the chromatic scale, lined up in a particular order. This was the composition’s basic musical gene: the piece was made up of repetitions of the tone row in strict order, sounded simultaneously or in succession. Individual notes could be immediately repeated, and could be used in any octave. And various mathematical permutations of the tone rows, such as reverse order, were also permitted.
The twelve-tone method ensures that no note is used more often than any other, so that none can acquire the status of a tonic merely by repetition. By the 1950s serialism had became, in many leading schools of classical composition, the only ‘respectable’ way to compose; anything hinting at tonality was considered passé and bourgeois. Yet Schoenberg not only failed to justify his horror of tonality – a composer like Béla Bartók displayed remarkable dissonant invention without abandoning it – but more importantly, he never truly came to terms with what its abandonment implied for both composer and listener. Since atonality has no tonal ‘home’, there was nowhere to depart from or return to, so that beginning, endings, and the entire matter of large-scale structure became problematic. As Roger Scruton says, ‘When the music goes everywhere, it also goes nowhere.’
Tonality is also one of the pillars of music comprehension. Far from being a decadent Western device, it is used in just about every musical tradition in the world (it does not rely on Western scales). Cognitive studies have shown how tonality provides a sense of location in pitch space and a way to organize the sequence of notes. It is the removal of this, far more than any considerations of harmony and dissonance, that many listeners find disconcerting in serialism.
This is not to say that atonality in general, and serialism in particular, is doomed to sound aimless and incomprehensible. There are plenty of other parameters, such as rhythm, dynamics and timbre, that a composer can deploy to create coherent structures. Schoenberg often did so masterfully, and Alban Berg’s Lyric Suite (1925-6) is so beautifully wrought that one would hardly know it was a twelve-tone composition at all. But as integral serialism and other techniques progressively and systematically subverted other means of providing audible organization, so it was unsurprising that audiences found the music ever harder to ‘understand’. The serialist’s rules are not ones that can be heard – even specialists in this music can rarely hear tone rows as such. Boulez’s serial piece Le Marteau sans Maître was widely acclaimed when premiered in 1955, but it wasn’t until 22 years later that anyone else could figure out how it was serial: no one could deduce, let alone hear, the organizational ‘structure’. One can hardly blame audiences for suspecting that what is left is musically rather sparse.
This is not to imply that music must return to tonal composition, with its cadences and modulations (although that is to some degree happening anyway). But ‘experimental’ music can only qualify as such if, like any experiment, it includes the possibility of failure. If musical composition takes no account of cognition – if indeed it denies that cognition has any role to play, or determinedly frustrates it – then composers cannot complain when their music is unloved.
Sadly, although these difficulties afflict only one strand of modern classical music, the fact that it was once dominant means that all the rest tends to get tarred with the same brush. Its critics often fail to differentiate music lacking clear cognitive ‘coherence systems’ from that which has new ones. What Javanese gamelan experts Alton and Judith Becker say of non-Western music pertains also to much contemporary experimental music: “it has become increasingly clear that the coherence systems of other musics may have nothing to do with either tonality or thematic development… What is different in a different musical system may be perceived, then, as noise. Or it may not be perceived at all. Or it may be perceived as a ‘bad’ or ‘simple-minded’ variant of my own system.” Often the only thing that stands in the way of comprehension, even enjoyment, is a refusal to adapt, to realise that it is no good trying to hear all music the way we hear Mozart or Springsteen. We need, in the parlance of the field, to find other ‘listening strategies’. Gyorgi Ligeti’s works can be appreciated as some of the most thrilling and inventive of the twentieth century once we realise that it handles time differently for instance. Musicologist Jonathan Kramer distinguishes it as ‘vertical’ rather than ‘horizontal’ time: musical events do not relate to one another in succession, like call and response, but are stacked up into sonic textures that slowly mutate and take on almost tangible forms.
It would arguably benefit all concerned if some experimental music, like much of Stockhausen’s or Boulez’s oeuvre and certainly the ambient noises of John Cage’s notoriously ‘silent’ 4’33”, were viewed instead as ‘sound art’, a term coined by Canadian composer Dan Lander and anticipated by the Italian futurist Luigi Russolo’s 1913 manifesto The Art of Noises. That way, one is not led to expect from these compositions what we expect of music. For if music is not acknowledged as a mental process, sound is all that remains.
Added note: The comments continue on the Prospect site, and make interesting reading. Of course, there are the inevitable blogosphere crazies. Will Orzo thinks he should tell me about this field called music cognition, in which people study other people’s responses to music. Thanks Will – hey, maybe I should use some of that work in my book on music cognition! Seriously, though, anyone who actually knows this field, as opposed to having looked it up on Wikipedia, would see straight away that this is precisely what I’m drawing on in my claims about how atonalism is perceived, especially the work of Fred Lerdahl, Carol Krumhansl and David Huron. If I am regurgitating anyone’s opinions, it is theirs. If you want references, look up my article in Nature last year (453, 160).
As for Joe Schmoe – anyone figure out who he’s ranting against? Sometimes it seems to be Stubbs, sometimes Adorno, sometimes Babbitt, sometimes me. An angry man. But incoherently so.