Engineering and Music
"Human Supervision and Control in Engineering and Music"

Orchestra Concert
Ensemble Concert
About us

James L. Alty :   Musical Composition as an  Engineering Activity

1. Introduction
Those who are not primarily concerned with artistic creative endeavour often view the creative process involved in, say, the writing of a poem or of a piece of music as a somewhat mystical process. Themes and even whole works are thought of as emerging in a semi-uncontrolled fashion, and at unplanned times, on the unsuspecting artist. The reality is somewhat different. There are, of course, times when inspiration will suddenly direct an artist in a particular path, or present a solution piecemeal to a compositional problem, but much of artistic endeavour is carefully planned, and artists and composers who do not approach their art systematically, often produce works which are fault ridden and inconsistent. It is the objective of this paper to provide an insight into the musical compositional process and to point out that much of this process has its parallels in Engineering Design. Engineers, usually, are involved in creating an artefact – they are producing a practical solution to a problem. Scientists on the other hand are mainly concerned with satisfying curiosity. The musical compositional process is therefore much nearer to Engineering. A product is being designed. Since the paper is aimed primarily at engineers and scientists, musicians may occasionally need to forgive some oversimplifications. 
2. What is Musical Composition?
Musical composition is basically the creation of a series of sounds that create an acceptable experience in the listener. The term “series of sounds” is used to avoid restricting the output to recognisable musical notes perhaps from established musical instruments. The term “acceptable experience” allows for many forms of musical appreciation that may be pleasing, calming, disturbing, exciting or alarming. However, because the main audience of this paper is intended to be Engineers and Scientists who, in many cases, will not have experienced the musical creative process in any extended sense, it will restrict itself to the creation of fairly traditional music. 

Whatever the form of the creation or acceptability criteria, being auditory in nature, music is basically a temporal medium. The listener usually experiences it as a stream. This is in contrast to many visual media (movies excepted) where the eye can re-visit and re-examine the experience over time. Music is therefore experienced within a narrow window frame of time, which is continuously moving forward. It is this fact that determines, more than any other, the form of most musical compositions. Because of this time dependence the listener has to rely heavily on short-term memory to make immediate sense of the stream. Although psychologists argue over the nature of this short-term memory, most will agree that it does exist and that typically a human being can hold 7+- 2 chunks of information in it-. This has important consequences for composition. It means that tunes are relatively short, that there is much repetition, and that development (as the music progresses) usually involves controlled transformations. A very simple tune (like a hymn or popular song) need only involve a few short lines with some simple relationships between them. If, however the composer is writing a more extended piece of music (say a few minutes or longer) then other organising principles have to be employed to hold the piece together. 

Unfortunately, most composers do not keep detailed records of the compositional process, and indeed the interplay between inspiration and the rational development of ideas discourages this. Although composer’s sketchbooks do provide some insight into the process usually the information is incomplete (for example, Beethoven had a number different completely worked endings for some of his symphonic movements, and by studying these, it is possible to understand what problems he was wrestling with and why he chose the particular final solutions we now hear). I therefore though it would be an interesting idea to try to keep detailed notes of the compositional process when writing a work in order to share the compositional experience with others, and to try and compare the process with an Engineering approach. This is what will be reported in the remainder of the paper. No claim is made that the resulting musical work is a great work, though I have tried to produce an acceptable one. 

3. Constraints and Musical Composition
There is a need for mechanisms for organising a composition over and above immediate short-term memory constraints. Since composers are communicating in some way with their audience and wish to be “understood”, there must be some higher-level rules (either shared by the audience, or picked up during the listening activity) governing the compositional process to assist the audience in following the music. The rules may be explicit or implicit. An explicit rule might be the splitting of larger works into manageable sections or movements (in terms of reducing the continuous workload on the listener this seems eminently sensible). An implicit rule might be “ a short sequence of notes when played backwards or upside down will still have an identifiable relationship with the original set of notes in the mind of the listener”. 

Composers have always worked with musical constraints. Some are imposed upon them because of the time of writing, or conventions of the time. For example, certain modulations were required and others simply not allowed in Mozart’s time. Other constraints are necessary to establish the composer’s individual voice or style (Alty, 1995). Further constraints are applied because of the nature of the work, the occasion and the instrumental forces used. Sometimes overt constraints are deliberately adopted perhaps to show the ingenuity of the composer (theme and variations being a good example). 

In the twentieth century, composers have deliberately adopted constraints to escape constraints. For example, to avoid the constraints imposed by conventional harmony, Arnold Schoenberg adopted Serialism – another form of constraint (see Ford, 1997). Other composers have spent much design time in creating new approaches (like grammars) often employing numerical ratios as a basis for constraining temporal duration, rhythm or timbre. This is why new works are often misunderstood. The audiences have not become used to the “new” language being employed by the composer. In this paper, in order to avoid this difficulty, the traditional constraints of Key signatures will apply. 

4. The Initial Compositional Constraints
The opportunity arose (and a particularly apt one) when a friend told me she was studying the piano and had reached a level of Grade IV or V, and I suggested that I should write her a short piano work. The obvious constraints at the outset were as follows: 
  1. The work should be reasonably short (say of 2-3 minutes duration) 
  2. It must be for the piano. This immediately imposes a huge set  of constraints on what combinations of notes are possible 
  3. It must not involve stretches of more than one octave for one hand (imposed by the lady herself) 
  4. No nasty key signatures! i.e. not too many sharps or flats 
  5. It should be playable at Grade V level 
  6. It should be reasonably accessible to the average listener

I also added a further self-imposed constraint. Since the lady’s name involved the sequence of letters CAFFE, I decided to use the notes CAFFE as the basis for the whole work! Composers often do this sort of thing – Bach and Shostakovitch are obvious examples. 

5. Overall Planning Constraints
A work lasting a few minutes will require some structure. It is too long for a single theme, since a typical verse of a song only lasts about 10 – 20 seconds. A Rondo structure was therefore adopted, a structure commonly used for compositions of this nature. In a Rondo there is a basic theme (say A) and this is successively contrasted with other derived themes (say B and C), so a typical rondo might have the structure ABACADA, or perhaps ABACABA. The essential characteristic is one of an anchoring theme interspersed with related, contrasting themes. 

Repetition is, of course, an essential in composition since it assists memory. However, too much repetition can be boring. The composer always treads a knife-edge between the two. I therefore decided that each time the A theme returned it would become more “grand” in terms of the chords and backing techniques used. It would start timidly but end majestically. I also decided to repeat the initial Rondo theme at the outset to fix it in the listeners mind. In the end the following sequence was adopted. 

A (timid) – A (filled out) – B – ending 1 – B – ending 2 – A (slightly grander) – C (vaguely fugal) – climax – A (very grand). 

The key signature of the opening theme (CAFFE) is clearly F major (which happily only has one flat). I decided to place theme C in A minor to provide contrast (no sharps/flats). 

6. The development of the initial A theme.
The opening was already set – CAFFE. Actually this sounded quite pleasant on the piano with even timing for each note. A common technique in composition is to keep the same pattern and repeat it at different parts of the scale. However, the pattern repetition technique is more complex that it appears. The Major Scale is defined by a sequence of semitone intervals. It is the sequence 2:2:1:2:2:2:1. Figure 1 shows how this happens on the piano and why C major has no “black” notes. If you start the sequence on F then the B has to be flattened to maintain the sequence. Keeping within the key signature enables the composer to create anticipation in the listener. 

Figure 1 Semi-tonal intervals in the major scale
Figure 1 Semi-tonal intervals in the major scale 

So, when patterns are moved round the scale, they should normally maintain the key signature, which often means changing the semitone pattern. For example consider the first two bars of Frere Jaques – CDE(C); CDE(C); EFG; EFG; The pattern CDE is being repeated as EFG, but the semitone interval pattern has changed from 2:2 to1:2. Actually maintaining the 2:2 interval would sound odd. 

Figure 1 The construction of the A theme
Figure 1 The construction of the A theme 

In order to minimise short-term memory load the CAFFE pattern is repeated twice at different points in the scale, starting on A and then F. Thus the pattern CAFFE is repeated with A as the starting note (AFDDC) and again starting on F, (FDB(flat)B(flat)A). This also has the advantage that the move forces the listener to expect a following F major chord (a sort of resolution). This is one of the bonuses of working within tonality. Note the four-note sequence which appears in the bass. In the A theme it is little more than decoration. It will become important later. 

The listener now anticipates a resolution to the F major chord so the next sequence deliberately plays with the listener’s anticipation by suspending its resolution, alternating between what are called the SubDominant and Dominant chords (Bflat-D-F and C_E_G) before ending in F major. This is deliberately designed to give the listener an emotional “kick” and a sense of closure. The A theme is now complete. 

7. The A Theme Repeat
Figure 2 The Repeat of the A theme
Figure 2 The Repeat of the A theme 

The A theme repeat involves filling out the single notes CAFFE etc. into chords making then sound a little more weighty. In the base, the four-note figure - a fall of four notes in sequence GFED – begins to become more important.  It is slipped in here to prepare the listener for its subsequent use. It is repeated in step with the CAFFE theme as it moves down the scale (EDCB(flat) with AFDDC, and CB(flat)AG with FDB(flat)B(flat)A) and the suspension is more weighty using inverted triadic chords together with slightly syncopated rhythm in the base (Syncopation involves stress off the beat). 

8. Introducing the B theme
We have now had enough of the A theme (the listener should know it by now!) so it is time to introduce a contrasting theme. This is derived from the A theme and uses the four-note run that was introduced earlier. 

Figure 3 The B theme is developed
Figure 3 The B theme is developed 

We take the first two falling notes (Figure 3) of CAFFE (CA) and play the four-note run (GFED) in the base. We repeat the falling notes this time as two chords (CA and AF) but repeat the run one tone down (FEDC). A rising chord is then played which is actually an inversion of the CAFFE pattern (AB(flat)DF). The whole pattern is then moved up the scale by two tones (a third). In maintaining tonality of F major, many of the relationships subtly change. For example CA (one and a half tone difference) becomes EC (two tones different). The theme is concluded (Figure 4) with a downward run to C major (which leads us back to F major. If Figure 4 is examined you can see it is very carefully constructed using the four-note runs. 

Figure 4 Conclusion of Theme B
Figure 4 Conclusion of Theme B 

9. Repeat of Theme B
Figure 5 Conclusion of Theme B Repeat
Figure 5 Conclusion of Theme B Repeat 

Theme B is repeated exactly as before. However, the conclusion needs to be something different. The listener needs a bit of tension here so, the second repeat of theme B involves a similar but more tense conclusion section which leads into a minor climax using the syncopation idea introduced earlier. 

10. Repeat of Theme A
Theme A comes back exactly as in its second repeat. 
11. Introduction of Theme C
We have now had Theme A and Theme B, both similar in many respects. What we need now is a real contrast yet which is somehow related. I decided therefore to switch to a minor key and use the theme in a fugal manner. A fugue is really a sort of round like Frere Jaques. A simple theme begins and additional versions are introduced “out of step” in time with the original. 

The key change is easy. We just go straight into A minor with a similar run to that which ended the A Theme. The fugue then starts in the treble: 

Figure 6 The Fugal-like Theme C
Figure 6 The Fugal-like Theme C 

The whole theme is derived from the CAFFE theme (slightly modified) and the four-note sequence. For example, in the second two bars there are two crotchet (black notes on their own) four-note descending sequences intertwined, playing along with quaver (black notes connected by lines) four-note sequences. The second part of the fugue has this sequence (Figure 7) playing along with the first part (from Figure 6, now in E minor). 

Figure 7 Second part of C theme playing with first part.
Figure 7 Second part of C theme playing with first part. 

Part 2 introduces the syncopation idea. 

12. From Theme C to Theme A and Conclusion.
The fugal subject makes its third entry in B minor. However, at this point the constraint of work difficulty begins to worry me. This work is not supposed to be beyond Grade V. My pianist might not be able to cope with the complexity that is inevitably developing. 

The next step, bringing in a fourth entry of the fugal subject in F sharp major (three sharps) will probably be the last straw! I decide therefore to stop development and get back to theme A. 

Figure 8 The Conclusion of theme C, and tension building for the return of theme A
Figure 8 The Conclusion of theme C, and tension building for the return of theme A 

A six note oscillating figure is introduced in bar 63 and, with a number of sharps and flats, A minor is re-established. The tension is meant to rise rapidly here so that when theme A returns in a heroic form it seems as if it is rescuing us from this tense situation. The oscillating figure develops into two four note sequences moving in contrary directions (though the last note of one is not played), whilst a set of triads in the base follow a four-note descending sequence over a pedal note (AE). This is repeated twice 

Figure 9 Re-entry of the A theme and Conclusion
Figure 9 Re-entry of the A theme and Conclusion 

We move into F major. Out of the base a rising figure of three notes repeats itself and fully establishes the key of F major. The Theme A reappears in a “grand manner” (with apologies to Elgar!) over a moving bass line (incidentally playing four-note sequences) and comes to a maestoso conclusion. 

13. Conclusions
Firstly, it is now obvious to me why composers do not faithfully write down their thoughts. It is hard work and gets in the way of the creative process! 

Secondly, I hope that even those who are not familiar with musical notation have managed to extract the idea of the processes involved. There is much use of symbolic manipulation according to a set of rules - breaking up sequences and re-using them, inversion, fragmentation etc. Of course, the paper only described one way of doing it. Every composer will have a different way of doing things and a different set of rules. However, the paper is intended to show that the compositional process, whilst relying heavily on inspiration, does involve much symbolic manipulation rather like that carried out by Mathematicians and Engineers. Where the process differs is in the evaluation function. Much Engineering involves interplay between aesthetic considerations and basic functions. I am claiming that in the same way, a composer moves between basic transformations and evaluation of those transformations. The process by which a composer decides that this musical manipulation works and that one does not, however,  is difficult to rationalise. This is why composition, in the end, is such a personal process. 

Alty, J.L. (1995), "Navigating through Compositional Space: The Creativity Corridor", The LEONARDO Music Journal, Vol. 28, No. 3, pp 215- 219. 

Ford, A., Illegal Harmonies, Hall and Iremonger, Australia.