If you want to grow, you need to start measuring.
This has been at the back of mind for a while now. For the last 6 months, I’ve been deep into the art of learning… analytics and data science of all things.
But what is clear is that in order to improve quickly, you need to start measuring something.
Just look at business, sports, or medicine. They are all measurement based. They are scientific.
But composition hasn’t been very scientific until now, but I think it could be.
I am also not saying here that music is scientific. It really isn’t. But composition is not music, and it has the possibility to apply the scientific method to it.
What is science and scientific method?
If we step back and ask ourselves what is science and the scientific method I think we’ll have a clearer idea what I mean.
Science is not physics, or biology. It is more about an idea that we can observe the world, creating hypotheses, and then most important, testing those hypotheses with more observation.
What you observe, and how you test is critical. The goal is to actually move the ball forward, not confirm your preconceived notions.
One of the best ways to do this is to falsify your hypothesis. This is straight from wikipedia article on falsifiability.
A statement , hypothesis , or theory has falsifiability (or is falsifiable) if it is contradicted by a basic statement, which, in an eventual successful or failed falsification, must respectively correspond to a true or hypothetical observation.
When you see someone throw around the term pseudoscience, what they really should be saying is that the “science” in question lacks falsifiability.
The example they give is that all swans are white.
If you see a black swan, then you know the statement “all swans are white” is not true. It’s been falsified.
This doesn’t tell you about every possible color of swan. Just that they aren’t all white.
The difficult part is how to apply this to composition.
Getting Away from Useless Questions
How do I compose? How do I write a good piece of music? Who is the greatest composer? What is the secret sauce?
These all seem like valid questions to ask ourselves, and they are not useless completely. But after a while they become problems.
Let’s start off with “How do I write a good piece of music?”.
How do I write a good piece of music?
I want to pick this apart, and hopefully by the end, we’ll have something falsifiable, and something to measure.
The first problem I see is the word “good”.
Good is loaded with too many meanings, and so it’s about as meaningless as it comes.
We can take replace it with something else. Here are some ideas that come to mind:
- Competent (sounds like good, but more arrogant).
- Long (needs to be qualified).
- Theoretically sound (feels good, but it’s about as useless as good, and more arrogant than “competent”).
- With functional harmony
- With a melody that observes the traditional rules of voice leading
- That other people will recognize as correct given the traditional rules of voice leading, harmony, and melody (hmm, this gives me an idea about measurement. Measuring listener response).
- For a specific instrument based on the style of a specific composer (getting a little closer).
- Unique (this will probably be very difficult to measure).
Once again, piece and music. They are not too clear either.
Piece could be:
- Piece in small ternary form (I like this because it is flexible, relies on specific formal definitions, and but can be argued. Small ternary form is just a three part form that is about 16 measures with an A-B-A’ scheme).
- Maybe piece should be a smaller fragment. Maybe it should be chord progression.
- A progression that modulates, and uses specific harmonic devices such as modulation, applied dominant, modal borrowing. I mean we can go on and on.
If we put this together, I come up with a much bigger question, that has a much smaller answer:
How do I write a piece of music in small ternary form that others recognize as correct given the traditional rules of voice leading, harmony, and melody for a specific instrument based on a specific composer or style for a specific instrument that incorporates a progression that modulates, and uses specific harmonic devices such as modulation, applied dominant, modal borrowing?
Turning the question into a falsifiable statement
Now if I want to be scientific about this, I need to make it something I can test. Because the paragraph of text above is kind of difficult to read, I am going to put it in a list.
- Writing a piece of music in small ternary form.
- That follows the traditional rules of voice leading, harmony, and melody.
- For a specific instrument.
- Based on a specific style (read: the tendencies of a specific group of composers).
- That incorporates harmonic progressions that use specific harmonic devices (functionality, modulation, applied dominants, modal borrowing, standard schemas) and melodic devices (following tendencies, ornamentation, motives, specific kinds of repetition, tropes, etc.).
- Will cause others to recognize that piece as good.
1-5 are really just constraints. They limit the playing field.
The most important part of the question I think is the last part (number 6).
You can measure this. You can measure it by asking other people. This is actually a common way to measure.
Let’s give it a try. Here is a piece by Jeremy, one of my students.
“Piano in G” by Jim1989Jeremy is basically doing 1-5 above. Now it’s your turn. Is this piece good, given the constraints in 1-5.
Getting more fine grained
The problem here is that a piece is actually quite difficult to write (I sense another thing to replace good – “feel easy”).
It also takes a long time, and so the time from action to measurement is long.
If we focus on just two of the constraints.
That incorporates harmonic progressions that use specific harmonic devices (functionality, modulation, applied dominants, modal borrowing, standard schemas) and melodic devices (following tendencies, ornamentation, motives, specific kinds of repetition, tropes, etc.).
What’s cool about this, is that we can actually take out any one of those elements. And then ask more people the same question. We could if we wanted create a bunch of different versions of the same piece and then ask people to vote, and see which piece gets the most good votes and bad votes.
This get’s me excited because I can think of many things I would want to test here. And they are all falsifiable.
What if you don’t know how to compose anything yet?
Shameless plug here: Take my free course, The Vocabulary of Composition. This will give you the tools to put those ideas into practice.
So what if you can’t write a piece yet?
You need to start smaller. Start with something you can handle.
I think a great place to start is with chord progressions. They have built into them many of the things that you need to learn in order to write a piece that meets all those criteria above.
They are also the tastiest part of composing.
Who doesn’t love a good chord progression.
Falsifying that
If we want to take is full circle, I should falsify this last statement:
Memorizing chord progressions and learning to transpose them in your mind will reduce the time it takes for you to finish pieces in small ternary form?
That’s it for now. This post is an exploration of an idea, that is just starting to grow in my mind.
Luckily my father conducted medical research for years, so I will be developing this more, and eventually hope to start running studies on the improvement of composers.
Until then, start looking for ways to falsify your ideas… and sign up for my free course.
Hello again,
I might also suggest limiting what you wish to test to 2 or at most 3 things (but not more). So maybe try this: 1. the three part form ‘ternary’ form (or A B A) and they must use voice leading in creating the music too, is:
2. no better than writing music as you simply wish to write it (the ternary form is not allowed) and the person can write the notes of the melody or music in any way they see fit.
Then put the examples on line so that they can be automatically played and those who listen can give them a score between 0-10 (10 being excellent even beautiful music and 0 being not really music at all). Then take 10 examples of both (the people who write the music may remain anonymous with a ‘nickname’ (but known to Jon) and the people who rate the music can remain anonymous in the same way. Then see how the two types of ‘writing music compare score wise.
I would suggest that the music be written out as the example of Jeremy’s above (so Jon does not ‘expire’ in the process. But the null hypothesis is that neither will be considered better than the other; the ‘experimental’ hypothesis is that the 3 part (ABA ternary form) will be considered better. This would falsify the ‘null’ hypothesis if more people ‘liked’ the ternary 3 part form than people just writing music off the top of their heads. Or if people thought both types of music were the same in how they sounded (like both sides get 60% or about) then the null hypothesis would not be falsified.
So really testing what people think about the two ways of writing music. People that vote can also vote anonymously (for both styles – or at least Jon can present the votes anonymously). That way those who write (but their music is not given a ‘higher and better’ score (0-10) by those voting anonymously hopefully will not feel ‘attacked or hurt less’ by a lower score and may become a better composer out of it without being embarrassed.
Jim
Hi Jon,
First: I am your ‘Father’… (Jon cries out) …NO NOOOOOOOO! by the way I enjoyed Jeremy’s composition
These are very good intriguing concepts.
I might say I see your discussion as about ‘testable’ and falsifiable research using a research or an experimental hypotheses (write a piece of music in a small ternary form). The form is A-B-A’. So, in looking at it and listening to it: 1. it is a small ternary piece of music and, the ‘research or experimental hypothesis is not falsified.
This type of hypothesis is
1. The question or hypothesis is, the music form is to follow a small (this is certainly a short piece – not 3 minutes or 15 minutes. But short (so ‘small’ might need to be specified ‘as less than 1 minute or 2 minutes,’ etc.) But you did say ‘short’ and I think most people would agree that Jeremy’s piece was ‘short’.
Then it was to be in a ‘ternary’ form (or A B A). Again this was ternary, I agree. But if I was not also a musician (in 1977 I got a Bachelor’s in Music) ‘ternary form’ I might need to be explained (or the ‘definition of the ‘ternary form as A-B-A instead of something like A-B-C, or so on clarified). But since this is correct for ‘ternary form’ i music (see wikipedia) Jeremy succeeded and it is not falsified!
Now, as for the other parts (or other I might suggest are also all ‘experimental or research hypotheses’). This type of hypothesis in medicine is often put like this…
1: ‘if I do A’ (inject a strong opiate into a patient) they will: 2: feel less pain after surgery.
The null hypothesis would be
1: ‘if I do A (inject a strong opiate into a patient) they will: 2: feel just as much pain as if I did nothing (or injected inert saline into them).
Both of these could be measured using a 0-10 Numerical Pain Rating Scale (really an 11 point measure with 0 = to no pain at all and 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 amounts of pain with ’10’ being pain that is unbearable and excruciating so much so that they call 911 for an ambulance to take them to a hospital…
Other hypotheses that follow are: use the traditional rules of voice leading (not falsified – Jeremy clearly knows voice leading), ‘rules of’ harmony used (not falsified or ditto), and melody (ditto).
For a specific instrument. (In this case if ‘piano’ was specified it then the research hypothesis would not be falsified). But this was not specified here… so the next time the instrument would have to be ‘specified’. Like use: ‘Guitar, or Trumpet or a quartet (and name the instruments such as 1 violin, 1 viola, 1 cello and 1 Double Bass
Based on a specific style (read: the tendencies of a specific group of composers). ‘Name the style or have the student name the style (Jeremy sounded like ‘Mozart’ to me).
That incorporates harmonic progressions that use specific harmonic devices (functionality, modulation, applied dominants, modal borrowing, standard schemas) and melodic devices (following tendencies, ornamentation, motives, specific kinds of repetition, tropes, etc.).
All the above was done, including ornamentation of A (is that what A’ stands for)?
And finally,
Will cause others to recognize that piece as good. (again here I would use the 0-10 or even 0-100 for more refinement such as 0-10 only allows that it is a 7 out of 10 (10 being as beautiful as your favorite Bach, or Mozart or Bill Evens Jazz Piano Composition) and ‘0’ being ‘not a successfully composed piece of music… or ‘0’ could be not successful and ’10’ extremely successfully composed music…
I think these are vital ideas. There is of course always difficulty with the ‘measurement’ problem in some areas. But the ideas can give more structure to these seemingly ‘unmeasurable’ concepts. It is brilliant and could quickly speed up the process of learning to or how to compose. Applying the ‘methods of science’ and that is all science is really, implying ‘methods’ to rule out bias, or rule out failure to use voice leading in the melody or harmony… Yet keep in mind that Einstein was sitting in a building thinking about ‘gravity’ and then he imagined a ‘man falling off the building’ (and he said it was the most wonderful idea of his whole life)! Not the man falling but the recognition that when the ‘man’ fell off in that moment ‘there was ‘no gravity’… it took him another 8 (or 15) years I believe, to come up with the Special Theory of Relativity and then the General Theory of Relativity (particularly relating to gravity and the ‘man’ falling off the roof. By the way I put compositions by Bach and even some by the Jazz Pianist (including the ephemeral ‘improvisation’ in this same category. Einstein had to know or learn ‘mathematics’ first (and there are apocryphal stories that ‘Einstein’ was not so good at mathematics!) and some physics FIRST before he could have come up with the General Theory of Relativity! Inspiration as they say is ‘99%’ sweat and ‘1%’ a gift from heaven! This might not actually be a ‘falsified hypothesis but I know it to be very much closer to the truth…
Jim Brantingham PhD
Thanks. This gives me a lot to think about.
I guess at this point, I’m still trying to identify what the critical hypotheses are with teaching and learning composition.
If I look back over what I’ve personally created for Art of Composing, my original hypothesis was:
If I can understand the generally accepted music theories (based on the standard textbooks used at Universities), then I will have a stronger ability to compose.
But now I see that so many of the accepted theories are not very useful for the creation of music, just for the description of it. But there are some that are legitimately useful.
My goal now is to parse those out and determine the best method for learning so that creation is easier.
Just sitting here though, I am starting to see stuff that really needs to tested:
– Learning to compose in a classical style makes it easier to compose competently in other styles.
– Learning the rules of functional harmony versus memorizing standard chord progressions and transposing a la Partimenti tradition.
– Learning species counterpoint vs once again the Partimenti tradition.
– Using standard forms as a way to integrate the knowledge.
– Copying scores by hand (this is a big one for me, because I personally like doing it and have a gut feeling it is vital, but it is also very tedious).
There are obviously more, and I think the next article is going to be an exploration of the things that we can start to test.
But the biggest question in my mind is what outcome measures to use.
via GIPHY