How would one go about composing a Tintinnabuli piece?

Question

I've been doing some research on the Estonian composer Arvo Pärt lately, known best for his simple yet highly emotive Tintinnabuli compositions. The theory behind it is incredibly dense and difficult to understand (at least for me). If I'm not mistaken, the whole theory behind it is that there are two voices: the T-voice (Tintinnabuli voice) and the M-voice (Melodic voice). The M-voice can be any pitch, and moves stepwise around the tonic note of the scale, and the T-voice is comprised of notes from the tonic triad and can jump octaves.

As per a thesis on this style, "According to Hillier, the relationship between the M-voice and T-voice is predetermined for every piece; moreover, some works are based on “some numerical pattern or by the syntax and prosody of a chosen text. Very often these two ideals are combined.”" (https://digital.library.unt.edu/ark:/67531/metadc271844/m2/1/high_res_d/thesis.pdf)

An entire Tintinnabuli piece is based on the harmonic fundamental and overtones of a ringning bell; for example, the hum tone of a bell is like a drone. The T-voice notes in most Tintinnabuli works come from the centric pitch of the piece, and often have similar characteristic of the minor chord produced by the main partials of the bell's harmonic series. These pieces are also very much based around symmetry and are usually written in Aeolian Mode, and have often been described as being Algorithmic (although I'd rather not generate music).

I'm having trouble figuring out how to actually write one of these pieces though. As I said, explanations of this can be difficult to follow, and there have been numerous theses and papers written by PhD students and academics. Can anyone help me to understand this on a more basic level? Thanks!

An example:

Answer 1

The Tintinnabuli process itself is quite basic (and eminently automizable), but it sounds complicated when described. Although Pärt generally begins with a stepwise melody that adheres to a mode, that isn’t essential to the underlying process. Similarly, although he generally chose a basic triad as the generator for the tintinnabuli voice, that too isn’t a fundamental part of it. Here’s the process:

Write a melody. If you want to stay very close to Pärt’s style, keep it rhythmically simple, intervallically conjunct and use an old church mode to pick the pitches (although, I know that he sometimes used harmonic minor scales, as in Fratres).

Pick a harmony for the T voice. Again, use a major or minor triad to be the most Pärt-like. However, in the pieces I’ve looked at, I’ve noticed there are plenty of pieces where this chord is not the tonic of the starting melody. In Fratres he chooses a chord that has one note that isn’t even in the mode of the main melody, and the cross-relations and dissonances this creates are a big part of his sound world.

There are two choices to make about the T voice. First, do you want it to be above or below the original melody? Second, do you want it to use notes that are one unit away from the main melody, or two units away? That part will make more sense when I describe the process itself. Also, I should note that these two decisions sometimes have a few additional complications that are part of why academic descriptions can get so complicated. For example, I might chose to generate my T melody above the M by two units, but then actually write it below the M. Let’s ignore that kind of complication for now.

Let’s say I’ve chosen to write my T voice using a c minor triad, above the M by one unit. Say the first note of the M voice is an A. That means the first note in my T voice needs to be the first note of a c minor triad that is above A. That would be a C. Say the second note of the melody is a B. That means the second note of my T voice will be a C just like the first note, because C is the first note of a c minor triad above B. Let’s say the third note of M is a C. Remember, the algorithm says that I’m using the first note of a c minor triad that is above the M voice’s note, so when the M voice is a note that is already in the c minor triad, I would never use the same pitch. In this case, the first note of a c minor triad that is higher than C is E♭ . So for an M voice that begins A–B–C, I’d have a T voice that is C–C–E♭.

Let’s do one more example using the same M voice. This time we’ll generate our T voice using the same c minor triad, but below by two units. The first note in the main melody is A, so we’re looking for the second note of a c minor triad below that. The first note below A would be G, but the second note of a c minor triad below A is E♭. Next M note is B, and again E♭ is two notes of a c minor triad lower. And, actually, we get the same note for our third pitch, because E♭ is still two c minor triad notes below C. Thus, the M melody A–B–C would give us E♭–E♭–E♭ as a T voice.

Of course, we could use both of these T voices and get a richer counterpoint. We could play with different generating harmonies; we could generate a “below” T voice, but then write it above the M melody; we could make two T voices, but use different generating harmonies for each; etc.

Let me know if this still needs clarification.

Answer 2

I use tintinnabuli as texture often.

To simplify, you just need to know how to build your “harmonic scene” and voice the stems - how many you want to. One of them must be tintinnabuli obligato - simply conduct your tonic triad for all the section, respecting counterpoint and not focusing too hard in dissonances formed by your T-voice with others. I like to swap T and others as phrases ask for (from alto to soprano, or sending to bass...).

For real, it’s a little more than this, but you can mimic the effect this way. To do as Pärt, unfortunately, only being Pärt himself. But it is an incredible texture in all kind of instrumentation.