Revising Kampflieder

Just under a week ago my Kampflieder had its first performance in Melbourne, by the Melbourne Metropolitan Sinfonietta with Elliott Gyger conducting. It was an interesting and worthwhile experience. While many of the players were quite new to this music, and circumstances (and resources) meant we only had a couple rehearsals, everyone gave it a really solid crack, and I got to hear more or less what I was aiming for.

Which means that while there were some things that didn’t come off due to lack of rehearsal time or whatever, there were a bunch of things that didn’t come off for precisely compositional reasons. So here’s my public self-criticism… Well, more like just a list of things I plan to change in the work for any potentially revised version – that hopefully will see performance one day.

This is purely personal affair, ‘for research purposes’ as we say, so readers can stop here less they get horribly bored!

In no particular order:

  • The overall stratification I was aiming for didn’t quite come across, since, at the end of the day, there was too much timbral and registral cross over. Given that such crossovers are more or less unavoidable once you have five different layers in the combinations that I had arranged, what is needed is another solution (though one that is not purely external to the essential idea of the work). My thoughts are that the groups need to be much more spatially separated, so that their independent identities can be better grasped. Simply spreading them further out on the stage may do it. Otherwise, the more extreme option would be to have a click-track that each performer listens to through headphones, which would allow the groups to be entirely spread out in the performance space. Too distant would lead to problems, but they could be quite far away from each other I think…
  • The opening ‘Rote Wedding’ attack in the strings should be taken up by the whole ensemble – needs more punch!
  • The percussion part needs considerably less resonant attacks – since the cymbals really cover so much spectral space, they tend to seriously blur out the timbral distinctions between groups, which reduces the stratification and makes everything sound even more like a mush.
  • Group III needs a longer build up to when they take over the Rote Wedding tune at bar 57 – so what they’ll do is start to double group IV from the last semiquaver of 51.
  • At bar D (bar 57), I should allow more space for the unfolding line in group III. To that end I’ll delete a bunch of entries from the other group and allow each group to articulate itself individually in relation to group III before they begin to stack on top of each other again.
  • Again on bar 57, group III: should have glissandi… as it stands it breaks the consistency of the whole section much too much.
  • At the end of the work: consider the Rote Wedding line in group V slowly accruing all the other groups… which would have a dramatic dimension to it, though may be a little too ‘obvious’.
  • On Group V when they take the Rote Wedding tune: the octave distribution should be rethought a bit, since sometimes the extremes of range in some instruments cut against the dramatic requirement of the punctuation.
  • At the moment of rupture at I (bar 140), the ff octaves in group V don’t articulate their oppositional relationship to the preceding material sufficiently: they felt, at least in the performance, more like a random insertion, rather than something necessary and negative. So this will need a bit of thought as to how to build up to this moment, or whatever…
  • In general the folk song solos leading up to I need to have much stronger song-identities much sooner than currently. As it stands they only just being to emerge before they get cut off: it doesn’t sufficiently affirm their important identities.
  • The super dense layering of 5 folk songs at letter M needs adjusting: all the folk songs should be in roughly the same time scale (within the orbit of the standard song tempo) so that they are all perceived as songs, rather than just 2 or 3 songs and then a number of clumsy accompaniments.
  • Also at M, the thirds in the trumpet and cor anglais should be inverted, with the cor anglais on top…
  • Group IV inversion… put accents on note changes, to give this layer more internal differentiation
  • Group III inversion… rethink a little the jumping octave materials…
  • Mutes for group V up to letter I? Particularly the trumpet.
  • Group V held octaves: tuba and contrabass should be greatly shortened and the glissandi should be quicker and more obvious, leaving the upper instruments to hold on for longer.
  • Percussion from D to I needs to be rethought: it is very important structurally to have the vibraphone nice and clear as a ‘cantus firmus’, so perhaps the rest of the percussion needs reduction.

One basic lesson coming out of this, something I can’t correct in this score, but that I can work on in future pieces, is that an overemphasis on the ‘unforeseeable’, canonic approach needs to be rejected or at least heavily mediated by more planning of proportions and overall relations at all points in the work. This is not solely in order to ensure a more dramatically engaging temporal flow, but also (and most importantly from my contrapuntal perspective) so that the logical determination of one element on another is made sufficiently clear to perception. As it stands, in Kampflieder there is too much ‘contingency’ in how layers interact in their broad outline (where they start and stop, etc), and it undermines how layers interact on a purely logical level (the meaning of one layer in relation to another). So, in future, even in mensural canonic procedures, I will need to have a much clearer maths of how things unfold. None of this is to say that unforeseeable things will emerge in the course of composing – this aspect should, of course be welcomed, as long as it takes place within a logically consistent and compelling framework.


Kampflieder is finished

I finally finished Kampflieder, for chamber orchestra and it’s all sent off to the relevant people. It will be performed in Melbourne on March 18 by the Melbourne Metropolitan Sinfonietta with Elliott Gyger conducting, so if you’re in Melbourne around then, consider popping along!

Some basic reactions from looking at the finished score (I’ll put it up online once it’s been performed):

  • It will be interesting to see how the density of the piece for the first few minutes works. It’s pretty intense with very little let-up… We’ll see what effect that will have in listening.
  • The overall flow is odd, on the one hand it is underpinned by a very simple (and hopefully audible) process of expansion and contraction, on the other hand, there is a formal rupture almost 2/3rds of the way in and after that, what seemed like a continuous piece becomes very discontinuous, at least on on layer.
  • There’s two moments in the work where there are five different folk songs playing one over the top of the other in different keys. Firstly it’s five solo instruments playing fairly closely together, then secondly the whole ensemble comes together to play five (actually six…..) different songs at the one time spread out over the whole range of the group. In the latter case they’re all accelerating and decelerating at different rates as well. It’s something of an experiment and really it’s very difficult to get a clear idea in my head of how this will work, so we’ll have to see in performance!
  • Particularly in the opening section the balance of the groups is a question. Of course there is an imbalance amongst them (the size of the groups, the length of their phrases), but I want there to be a relative equality between these parts. Will it be that, for example, group V dominates?

Now I need to begin working on the chapter that explains what I did. Here are (some of) the basic elements of the work that I’ll need to outline:

  • Group counterpoint: the determination of the identities of the 5 main groups, how that influenced the selection of instruments – heterogeneous logics across groups. Fixed and variable parameters, and their ‘inversion’ in the second half of the work as a major structural determinant.
  • Mensural phrase- and dynamic-canon as method of creating long-range coherence and complexity in the work. Role of dynamics in creating group unity, reducing internal differentiation. Problems: high degree of unforeseeability of interactions in the long-run. Too much? Should future works be composed with a clearer scaffolding for each polyphonic line in each section, rather than letting each line unfold with only an initial sense of their synchronisation.
  • Macro-structural layer ‘Der Rote Wedding’… how its tempo is stretched and contracted, how it is deployed in the work. Ideas of ‘surface’ and ‘depth’. Problem with the tempo stretching, inexact.
  • Use of songs. First half: solos, slow emergence of tunes from more ‘abstract’ lines. Retrograde dynamics cannon.
  • Use of songs, second half: Tempo expansion and contraction, swirling time, primitivist vertical layering. Problem: performance of subdivision changes?
  • Use of songs, derivations: a lot of materials in the first half are loosely motivically derived from the songs in the solos…
  • Interesting elements, group IV: distribution of 24-tone scale across 5 octaves according to different densities, and the assigning of a varying number of pitches to each instrument in the group to control the degree of common pitches and overlap between them. (Problem encountered: contradicted by process of reduction of phrase lengths).
  • Interesting elements, group V: micro-polyphony of motives and linear directions, derived from El quinto regimiento. Not micro-polyphonic in Ligeti’s sense, since it won’t blend into a clear mass due to a) the heterogeneity of the instrumentation; b) the small number of instruments (5) compared to the different directions (3) and motives (3).

One thing that I think will be interesting to look at will be my approach in this piece of determining the sets of proportions for a particular parameter, and then permuting those proportions randomly to get the actual result. (Which is to say that, for instance, very simply, a certain section might have 8 triplets, 5 semi-quavers, 3 quintuplets and 2 sextuplets, and then I permute this set randomly to get the basic rhythmic structure of that section). This contrasts with my earlier practice in for instance a new day in the desert where I tried to define complex tendencies by superimposing several linear processes.

Of course I have both approaches in Kampflieder, but the former increasingly predominates, which is logical, considering I was thinking more and more about group identity, rather than processual development, and I am moving away from such a processual-allegorical formal approach. The other problems I had with the approach in desert was that it was really messy (I really didn’t know how processes would end up, especially once several were stacked on top of each other), and slow moving (the creation of these processes was done by guess-work and trial and error). Now, I think both approaches are relevant. These latter problems can be solved by better maths – something I’m working on through the Khan Academy. The question of process and allegory can be solved by creating relations between continuous and discontinuous transformation, and by composing works in a less linear fashion (i.e. not starting at the start). The problem of how to permute sets of proportions in a less random fashion, I’m not sure… Anyway, this will be the subject of a future post I’m quite sure.

Times in Kampflieder

In a previous post I ‘decided’ on the basic metric stratification in Kampflieder. However, I’ve since rethought this aspect. With a more rigorous approach to group definition, I’m not putting so many eggs in the sub-division basket, since the differentiation of the groups is more thorough and rests on many different parameters. This allows me to relax the stratification of the individual sub-divisions to be kinder to the performers…

Part of the problem was that if each group has a different subdivision, and within each group there are different sub-divisions, I would need many different nested tuplets. Which would perhaps be fine if writing for MusikFabrik or whoever, but I want the performance by the Melbourne Met to sound good.

This is not a matter of compromising – at least I hope not. It’s more a matter of making sure that where a certain parameter has to be somewhat restricted in its range, others are broad enough to create an expansive vision.

Regardless, if I really wanted to push the metric stratification further, I would need to have multiple conductors or click-tracks. Something I might pursue in the next big work.

In any case, what I will do quickly here is outline where I’m at with the macro- and middle-range temporal structures. This should give a feel for the ‘poly-temporality’ of the music (Coulembier).

Macro-level time

Firstly, there is what could be called the primary structural layer. It is the layer that gives the major sections to the work. This is based on the song ‘Der Rote Wedding’ by Hanns Eisler. This is slowed down massively at the start with around 28 semiquavers (at quaver=120, or 28 semi-quaver triplets at quaver=80, which is the actual tempo at the start of the piece) between beats of the song. This has a more or less exponential crescendo to the end of the first section of the song, where there is around 4 semi-quaver triplets (at quaver=80) between beats. This then brings in a new section in the work with a new tempo (quaver=90). During this section the structural song slowly expands in time (and deceleration) until it reaches about 31 semi-quavers (at =120) at the beginning of the third section, which takes a tempo of =120 and across which the structural songs accelerates until it is less than a quaver between beats at the very end of the work. At each change of section, a different group takes up this fundamental layer (while maintaining its own material), firstly it’s group IV (string quartet), then group III (percussion, trombone, bs clarinet), and lastly it’s group V (mixed).

Two things to note: while the space between beats of Der Rote Wedding expands and contracts across the work, and fits within different ‘surface’ tempi, it nonetheless is based on a consistent tempo which is quaver=120. Which means that only in sections (such as the final one) that have this tempo does the structural song have a simple relation to the metre.

The other thing is that while there is a (more or less) directly linear/exponential process of expansion/contraction of the beats of the structural song, because this song has its own syncopations and melodic rhythms, what actually enters the score, and what will be heard, is something less obvious. The overall process of acceleration or deceleration should be audible, but only after some time and with some red herrings thrown in.

What I like is the idea that the macro-level can move between a kind of deep-structure to a surface-level and back again, occupying at different points, different perceptual levels and interacting in different ways with the other elements of the work.

Middle-level time

Next there is the deployment of the different groups over time. I chose to do this by way of a ‘mensural canon’ – not of their micro-rhythmic structures, but of the medium-range time of their entrances and exits. Something like a ‘phrase rhythm’. This was conceived in bars, or part-bars. And these were the proportions:

Group I: 7

Group II: 6

Group III: 5

Group IV: 2

Group V: 4

So this meant that for a four bar ‘phrase’ of group V (heterogeneous mix of winds, brass and strings), group IV (string quartet) only has a two bar, and group I (solo horn) has seven full bars.

The opening sequence of proportions goes like this:

Group I: 7 – 7/8 – 1 ¾ – 7 7/8 – 1 ¾ – 2 5/8 – 8 ¾ – 2 5/8 – 3 ½

Group II: 6 – ¾ – 1 ½ – 6 ¾ – 1 ½ – 2 ¼ – 7 ½ – 2 ¼ – 3

Group III: 5 – 5/8 – 1 ¼ – 5 5/8 – 1 ¼ – 1 7/8 – 6 ¼ – 1 7/8 – 2 ½

Group IV: 2 – ¼ – ½ – 2 ¼ – ½ – ¾ – 2 ½ – ¾ – 1 – 2 ½ – ¾ – 1

Group V: 4 – ½ – 1 – 4 ½ – 1 – 1 ½ – 5 – 1 ½ – 2 – 3 ½ – 2 – 1 ½

The numbers apply to different types of structure, in a repeating sequence: full group – solo – rest. Over time, the full group sections reduce while the solo sections (where only one instrument from the given group play) increase (the rests remain fairly constant). This means that the texture progressively thins out across the first two sections. I’ll talk a bit about the meaning of this and how it relates to the songs of Kampflieder in a later post.

On thing to say about this is that while the number sequence above is based on a number of fairly simple additive or subtractive numeric processes, this comes (in a very Ferneyhough-type way) to interact with the bar structure itself which, while based on 4/8, has many little ‘kinks’ within it, which affect the smooth function of both medium-level rhythms and micro-level rhythms. This is evident if you’re working with the bar as your basic unit. So, if a phrase length is 3 bars and those bars are all 4/8, then it’ll be fairly long, but if the three bars are 3/16, 4/16 and 5/16 (as sometimes happens in the work), the phrase will be substantially shorter. Again, like the macro-process of the structural song, what you’ll eventually here is the underlying tendency, which does not have a one-to-one relationship with its micro-level instantiations. Well that’s the idea, the aim is that these tendencies will be audible, but at the same time, but that the local-level stuff is not perceptually redundant…

There’s a lot more to talk about, especially regarding the micro-level temporal structures (rhythms), which I’m currently working out. But that’ll be another post.

Defining groups in Kampflieder

I had a great lesson with Richard Barrett a few days ago up in The Hague. As always, it was very stimulating, and I came out with renewed energy and a bunch of stuff to work on. As always, he encouraged me to push further my basic ideas, rather than allow my natural desire for compromise to get the better of me.

On that point, Richard suggested that my thinking of counterpoint qua stratification-imitation could be in danger of losing its coherence and intensity if I try to find a classical ‘balance’ between these two elements. The point, in his eyes, at least at this moment, would be to radicalise the contradiction and present it – quite clearly – as a contradiction, which could be formally expressed as a rupture. As with our discussion of ‘tension’ in our last lesson, he firmly suggested I not try to fit my new ideas into old, Romantic poetics such as formal closure. New music is free to pursue whatever thought experiments it wishes, and does not have any historical or heteronomous authority to answer to.

We spoke primarily about my new piece Kampflieder that I began working on earlier this month.

I had already done a bunch of thinking about the basic conception of Kampflieder: there would be 5 groups each with a different number of instrumentalists (that is, groups of 1, 2, 3, 4, and 5 instrumentalists); these would be heavily stratified and characterised differently; the work would be woven out of transformations of songs from the Spanish Civil War. Beyond that, I had begun characterising these groups and setting up the metric and temporal/formal structure of the work.

Yet Richard impressed upon me the need to deepen and clarify my basic propositions. In particular, what he suggested was to really think through the basic characteristics of the groups and then decide on the instrumentation that could best carry this out. This is not entirely different to what I had already done, but I had gone about it in an unsystematic way, which means I missed opportunities for making the ideas really clear, and for having maximum possible manipulations of these ideas throughout the work.

What Richard suggested was to come up with a list of parameters that could give unity to a group of instrumentalists, then to outline for each one a continuum from 1-5 where 1 is least unified and 5 is most unified. For example, for register, we could say that all instruments in the group would play:

  1. at opposite extremes of entire ensemble range, 6+ octaves (minimum identity)
  2. across 4-5 octaves
  3. across 3 octaves
  4. within 2 octaves
  5. within 1 octave or less (maximum identity)

Now, if a group had a register identity value of 1 (that is, the instruments play across, or at a distance of, 6+ octaves), it would be in need of an identity value of 5 in another parameter to allow for that group to really be perceived as a single ‘group’ – perhaps complete rhythmic identity (rhythmic unison) would be the way to go. If, on the other hand, a group had a register identity value of 5 (that is, all instruments play within the one octave or less), it would be able to have a much lower identity in other parameters and nonetheless retain its cohesion as a group – for example, it could be entirely timbrally heterogeneous.

So, I went about rethinking my basic groups with this in mind. I chose 5 fairly basic parameters that seemed pertinent to what I wanted to focus on. I decided that each group would require both a 1 in just one parameter and a in one other. The other three parameters would be assigned numbers from 2-4 (that is, neither maximal nor minimal identity). Eventually, after much juggling things, I came up with this table:

No. members Register Rhythm Timbre Harmony Melodic contour Instruments
1 3 1 4 5 4 Hn
2 1 5 2 4 3 Vln1, CBsn
3 4 4 1 3 5 Perc, Cl(+bscl), Tbn.
4 2 2 5 3 1 Vln1, Vln2, Vla, Vc
5 5 3 2 1 3 A.Fl, Cor A., Tpt, Cb, Tba

Naturally, there wasn’t a simple linear process from deciding on number to deciding on instrumentation. Possible instrumental combinations arrived in my imagination as I laid out numbers, and when I put in possible instrumentations, I had to rejig the numbers, and so on.

So, for example, group 2 is maximally rhythmically unified but minimally registrally unified – with a high degree of harmonic unity, a low degree of timbral unity, and a medium degree of unity of melodic contour. This is expressed by a contrabassoon playing in its low register, and a violin playing in its absolute upper register.

Or another example, group 5 is maximally registrally unified, which allows it to have a minimal degree of harmonic unity (which is to say it will be very polytonal within itself) and to have a very low degree of timbral unity. This is expressed by an alto flute, cor anglais, and trumpet all playing in their lowest register, and a contrabass and a tuba playing in their uppermost register.

What I like is how this process has lead me to find ways of uniting, for instance, a contrabassoon and a shrieking-high violin, or finding ways of making a timbrally homogeneous bloc (group 4: a string quartet) have enough internal heterogeneity to make it interesting.

Groups 1 (solo) and 3 (with percussion) have their own peculiarities and would perhaps need their own blog post each, but the basic thinking behind the constitution of the groups should be clear.

From here there is also the question of how each group will develop, and how the overall work will develop. Logically, this will involve manipulations (or complete inversions!) of the identity-degrees of the various parameters in each group, imitation of various parameters across groups, and then of course degrees of self-identity of the songs that form the basic materials of the work. Hopefully I’ll have time to write all this up too.

Lest anyone think this is all academic, it’s worth confessing that indeed none of this formal thinking is self-justifying. The use of thinking in these terms is to provoke me to go beyond my intuitions, beyond what one is ‘supposed’ to do in a composition, and to stop me from compromising (too much) on this vision. The goal is a maximum possible friction in the work, maximum intensity of expression of each group, maximum expression of their possible interrelation, and a maximum impact for the Spanish war songs. More on this too, at some point…

Metre and metric stratification in Kampflieder

I’m (still) in the planning stages of Kampflieder for chamber orchestra (ok ok, I’m a little behind schedule, but I’m doing my best alright!). I’m up to that bit where I work out the basic metric approach of the piece and each of the lines (here ‘groups’) within it. Since I haven’t written for a large ensemble since developing my current compositional approaches (and since I’m dealing with a group of musicians who, while no doubt very skilled, aren’t specialists in complex new music), I’ve deliberated over this quite a bit.

After outlining some basic possibilities in a previous post, and thinking each of these through a bit, the basic question came down to whether the common metric structure would consist of relatively stable durations (e.g. 6/8, 6/8, 6/8, etc) as in my third motet, or would be more fluid as in much of my other work recently, like Si el clima. The advantages of the latter are that I can define some common tendencies across the whole ensemble (i.e. expansion and contraction of overall time), I can easily have much more irregular rhythms, and it provides me with a compositional provocation at each new bar. The advantages of the former are that I can much more easily create sturdy tempo strata, and thus also mensural canons.

I sat down and worked out a whole bunch of different linear relations to these two metres as I have done in the past.

Regular metre

Metric functions2

Different possible approaches to rhythmic stratification with a consistent metre (see bar 2 for the simple capacity for mensural canons). The second from the bottom replicates a shifting metre (see below example) in one line and is in a way related to the ‘hyper-metric’ approach I suggested in previous posts; the bottom gives a ‘relative’ or ‘spatial’ or ‘fluid’ notation that is more indeterminate but allows for very fluid accels or decels in one line.

Irregular metre

Metric functions1

Different approaches to rhythmic stratification with an irregular metre. Note the second from the bottom which is essentially the substitution of a consistent metre.

(Naturally each of these should be explained, and I should explain why I’ve selected these particular approaches and not others, and why I’ve made certain distinctions between approaches that in some ways could be said to be the same… That will have to wait for a future post, or publication, or for the dissertation itself).

I ended up deciding that the former would be the best in this instance, at least in part because I think this will be easier on the musicians (this isn’t to say that there won’t be irregular measures thrown in as an interruptive device). But as I said above, it also affords some particular opportunities (there’s also the challenge of bringing rhythmic dynamism to the whole-bar tuplet strata – something that I don’t think that Carter ever fully dealt with).

So, here’s the thinking for the metric stratification between the four main groups in the initial set up in Kampflieder (the fifth is a ‘shadow’ group intended to give resonance, and will adopt the metric approach of whatever group it is shadowing at any moment) – this is all subject to change throughout the work, but this is more or less the setup at the beginning.

Line 1: Group 5; Line 2: Group 4; Line 3: Group 2; Line 4: Group 1. (I'll explain more about the groups in a future post).

Line 1: Group 5; Line 2: Group 4; Line 3: Group 2; Line 4: Group 1. (I’ll explain more about the groups in a future post). I’m not 100% decided on 7 as the basic metre, but it does afford a very non-polyrhythmic stratification (that is, it won’t sound like hemiolas or whatever).

An important thing to note here is that I’m thinking of the whole-bar tuplets as basic tempi, not audible polyrhythms (which means that each unit of subdivision will itself be subdivided, sometimes with nested tuplets). The idea is that groups will coordinate within themselves so that members of a given group are in time with each other, and they’ll watch the conductor (and listen to what else is going on) so they are aware of the general time of the ensemble (e.g. down-beats, obvious cues, etc). Of course this means there will inevitably be a degree of indeterminacy with regard to the precise coordination  of relations between groups (especially in a non-specialist ensemble). However, this is fine, since in general, the groups are conceived as distinct, independent strata. Where important interrelations between group-strata are composed, they will need to be marked obviously and coordinated by the ensemble, but outside these moments, a degree of accidental coincidences and non-coincidences between groups on the very micro-scale doesn’t worry me, since they will likely not be particularly audible in the complex stream of polyphonic information.