For many people, the terms tone row and matrix are synonymous with post-tonal theory, but to this point, we haven’t even mentioned them. (If you have never heard of either of these terms, don’t worry. We’ll get there.)
We haven’t discussed them yet, because if you understand set theory, you will have no problem understanding what a tone row is and how we use them in analysis.
First, we need to reaffirm your understanding of set classes. To do this, let’s start by writing out each unique, normal-form pc set represented in the set class of (013). What factors increase the number of unique pc sets? What limits the number of pc sets?
Before you began, you probably realized that there would be twenty-four possibilities. Because each of the pc sets will be in normal form, we know that there is only one pc set for each possible starting pitch. We can represent these using transposition notation.
There are a further twelve unique pc sets that result from inverting our prime form pc set. Of course, for each of these, you should normalize the pc set after inversion.
Each one of these twenty-four pc sets contains a unique collection of pitch classes. Without using the word “inversion”, how would you describe the relationship between T0 and T0I?
In the above example, you could describe T0 and T0I as the reverse order of the same intervals, but the only obvious correlation between the actual pitch classes is that both of the pc sets contain 0 – which happens to be the interval of transposition for those two sets. (The interval of transposition is the “0” in T0.) If you look at each pair of transposed and inverted pc sets, you will notice the same thing; they always share the interval of transposition. If you wanted to represent this visually, you could plot each transposition and inversion on a chart in which this pivot pitch for each transposition became a center pitch and had the inversion and transposition branching out. Look at our trichord charted this way below.
Inv pc set | inv pc 1 | inv pc 2 | common pc | tran pc 2 | tran pc 3 | Tran pc set |
---|---|---|---|---|---|---|
T0I | 9 | e | 0 | 1 | 3 | T0 |
T1I | t | 0 | 1 | 2 | 4 | T1 |
T2I | e | 1 | 2 | 3 | 5 | T2 |
T3I | 0 | 2 | 3 | 4 | 6 | T3 |
T4I | 1 | 3 | 4 | 5 | 7 | T4 |
T1I | 2 | 4 | 5 | 6 | 8 | T5 |
T6I | 3 | 5 | 6 | 7 | 9 | T6 |
T7I | 4 | 6 | 7 | 8 | t | T7 |
T8I | 5 | 7 | 8 | 9 | e | T8 |
T9I | 6 | 8 | 9 | t | 0 | T9 |
TtI | 7 | 9 | t | e | 1 | Tt |
TeI | 8 | t | e | 0 | 2 | Te |
Here, the “common pc” column shows the pitch class that is common to both the inversion and the transposition pc sets. (Note that to create this chart, the inverted form of the pc set is written in descending form rather than ascending form, so you must read it backwards to find its normal form.) You can make a chart like this for any pc set, and you can see how helpful this would be if you were analyzing a piece of music that had this trichord present. Rather than only looking for intervallic patterns, you could quickly refer to your chart to identify whether a specific trichord belongs to this set class.
What if we were to expand this technique? What would the above chart look like for a pc set with all twelve pitch classes? Before we explore that, let’s set some ground rules. If you were to use normal form, there is technically only one prime form for an aggregate pitch set: (0123456789te). However, if we treat the intervallic pattern as a fixed part of the pc set – meaning we do not put the pc set into normal form – then we can create nearly a million different distinct combinations of all twelve pitch classes.
With this in mind, try creating the first two lines of the above graph (that would be T0/T0I and T1/T1I) for the following pc set: T0 = (0,2,4,6,8,t,1,3,5,7,9,e)
What do you notice as you work through this? Can you think of different methods for creating T1I rather than just mirroring T1? While the chart above was not too unwiedly for addressing all possible variations on a trichord, can you think of a way to create a chart that shows every combination (transposed and inverted) of an aggregate pc set that would take less space than the clunky chart above?
The answer to the pc set would take up too much screen for me to succinctly write in a table here, but in plain text the order would be:
What if rather than writing T0 and T0I in a straight line, we rotated the inversion pc set, T0I, downward at 90 degree angle to create a verical column? This would create the outline for a 12 by 12 grid with T0 as the top row and T0I as the first column.
– | T0I | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
T0 | 0 | 2 | 4 | 6 | 8 | t | 1 | 3 | 5 | 7 | 9 | e |
– | t | |||||||||||
– | 8 | |||||||||||
– | 6 | |||||||||||
– | 4 | |||||||||||
– | 2 | |||||||||||
– | e | |||||||||||
– | 9 | |||||||||||
– | 7 | |||||||||||
– | 5 | |||||||||||
– | 3 | |||||||||||
– | 1 |
This grid is a called a matrix. We could then fill in each row of our matrix with the transpositions of the original pc set and each column with inversions. For example, where would the other two pc sets that we figured out in our original example, T1 and T1I, fit in this grid?
Because the top row is the original pc set and the column is its inversion, it makes sense to place the next pc sets, T1 and T1I, in the row and column respectively that begin with the pitch class “1”. It would look like this:
– | T0I | T1I | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
T0 | 0 | 2 | 4 | 6 | 8 | t | 1 | 3 | 5 | 7 | 9 | e |
– | t | e | ||||||||||
– | 8 | 9 | ||||||||||
– | 6 | 7 | ||||||||||
– | 4 | 5 | ||||||||||
– | 2 | 3 | ||||||||||
– | e | 0 | ||||||||||
– | 9 | t | ||||||||||
– | 7 | 8 | ||||||||||
– | 5 | 6 | ||||||||||
– | 3 | 4 | ||||||||||
T1 | 1 | 3 | 5 | 7 | 9 | e | 2 | 4 | 6 | 8 | t | 0 |
Because we have done this correctly, you’ll notice the pitch classes in the bottom row line up perfectly with the final pitch class from our new column. Why?
The great thing about this system is that once each of the transpositions is filled in correctly, each column will be a correctly transposed version of the inverted pc sets. Once filled in, a matrix such as this shows every transposition and inversion of any aggregate pc set.
Before we go further, we should briefly define the genre of music most associated with set theory and matrices. (Although set theory can be used to study any type of music as long as it divides the octave into twelve pitches.) Serialism is any music in which some aspect of the composition is based on a pre-defined repeatable pattern; this can be the melodic intervals, harmonic intervals, harmonies, rhythm, or any other aspect of music that could be described in a series.
12-tone music is a sub-genre within serialism in which a fixed series of all twelve pitches is used to generate both the melodic and harmonic content of the piece. There are a variety of ways in which composers have employed this, but in its strictest form, all twelve pitches must be used before a pitch can be repeated. The series that determines the order of all twelve pitch classes is called a tone row, represented by our top row in the example above. To provide compositional variety, the tone row may be altered to any of its transpositions or inversions, and any of tone row may also be played in retrograde – all pitches in reverse order. So from this, there are four forms of a tone row.
Because any of these tone row orders can start on all twelve pitches, there are forty-eight possible arrangements of any tone row: twelve prime tone rows, twelve inverted tone rows, twelve inversion tone rows, and 12 retrograde inversion rows. We label these using the abbreviations above (P, R, I, and RI) followed by a subscript number to represent the transposition. (e.g. P5 or RI6).
Using this terminology, complete the tone row matrix above by filling in each row and then labeling. Is it necessary to go in a particular order? What is the easiest way to fill it out. Knowing that we use P, I, R, and RI to label each direction of the matrix, how would you differentiate each tone row? Would retrograde tone rows be labeled by their starting pitch or their corresponding prime label?
The correctly completed tone row matrix for our original row would be:
– | I0 | I2 | I4 | I6 | I8 | It | I1 | I3 | I5 | I7 | I9 | Ie | – |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
P0 | 0 | 2 | 4 | 6 | 8 | t | 1 | 3 | 5 | 7 | 9 | e | R0 |
Pt | t | 0 | 2 | 4 | 6 | 8 | e | 1 | 3 | 5 | 7 | 9 | Rt |
P8 | 8 | t | 0 | 2 | 4 | 6 | 9 | e | 1 | 3 | 5 | 7 | R8 |
P6 | 6 | 8 | t | 0 | 2 | 5 | 7 | 9 | e | 1 | 3 | 5 | R6 |
P4 | 4 | 6 | 8 | t | 0 | 2 | 5 | 7 | 9 | e | 1 | 3 | R4 |
P2 | 2 | 4 | 6 | 8 | t | 0 | 3 | 5 | 7 | 9 | e | 1 | R2 |
P3 | e | 1 | 3 | 5 | 7 | 9 | 0 | 2 | 4 | 6 | 8 | t | R3 |
P9 | 9 | e | 1 | 3 | 5 | 7 | t | 0 | 2 | 4 | 6 | 8 | R9 |
P7 | 7 | 9 | e | 1 | 3 | 6 | 8 | t | 0 | 2 | 4 | 6 | R7 |
P5 | 5 | 7 | 9 | e | 1 | 3 | 6 | 8 | t | 0 | 2 | 4 | R5 |
P3 | 3 | 5 | 7 | 9 | e | 1 | 4 | 6 | 8 | t | 0 | 2 | R3 |
P1 | 1 | 3 | 5 | 7 | 9 | e | 2 | 4 | 6 | 8 | t | 0 | R1 |
– | RI0 | RI2 | RI4 | RI6 | RI8 | RIt | RI1 | RI3 | RI5 | RI7 | RI9 | RI9 | – |
Finally let’s touch on some general important concepts and common mistakes that you will encounter when creating tone row matrices.
P
or I
respectively, followed by a subscript integer of the first pitch class for that row.As with most things, some enterprising people have created a shortcut to all of this work. There are now free-to-use matrix calculators such as the 12-tone assistant at this excellent website. These are great to speed up your analysis, but as a student, make sure that you understand the principles of why and how these work before becoming entirely reliant on them. Even when trying to use this website, you must understand exactly what output you are trying to achieve before you can choose the correct option.
Admittedly, it is unlikely that you will ever be required to create a tone row matrix without access to a calculator (unless you are taking a music theory exam.) But the tedium of transposing each of the rows by hand will help you notice patterns in the tone row and help you to remember the nuances of the tone row as you analyze. This often provides insight into the analysis that would be missed by those relying on a calculator to create the matrix.