Lesson 3
Syncopation, measured
Usually explained by playing you an example and hoping. It has a definition precise enough to compute, and computing it forces an admission that the examples never quite make: syncopation is not a property a rhythm has.
A definition you can run
Lesson 2 ended with a number for every position in the bar: how many levels of division mark it. Those numbers are all this needs.
Longuet-Higgins and Lee proposed the definition in 1984 and it is almost embarrassingly simple. A note is syncopated when it is followed by silence at a stronger position. Something happens on a weak position, and then nothing happens on the strong position that was coming. The size of the syncopation is the difference between the two weights, and the score for a pattern is the sum over all such pairs.
That is the whole thing. No list of syncopated rhythms, no appeal to what sounds off-balance.
Syncopation, counted
A note, then silence where something stronger was due.
- note 3 (w1) → silent 4 (w3) = +2
- note 6 (w2) → silent 8 (w4) = +2
- Onsets
- 5
- Syncopating notes
- 2
- Score
- 4
- Highest possible here
- 15
Solid rings mark notes that are followed by a stronger silence; dashed squares are the silences they point at. On the beats scores zero, which is what it should do. Every offbeat scores fifteen, and that is not a coincidence of this pattern: fifteen is the highest any of the 65,536 possible sixteen-step patterns reaches against this hierarchy, and the all-offbeat pattern is the one that gets there.
What the number does
Notes on the beats score zero, which is the first thing any measure of syncopation has to get right. The son clave scores four, from two pairs: the note at position 3 followed by silence on beat 2, and the note at 6 followed by silence on beat 3. Those are exactly the two places the clave leans.
Every offbeat sixteenth scores fifteen, and that number is worth more than it looks. Fifteen is not merely high; it is the maximum. Enumerate all 65,536 possible sixteen-step patterns, score every one, and nothing beats fifteen, with the all-offbeat pattern the one that reaches it. The measure has a ceiling and the most obviously syncopated thing you can write sits exactly on it.
One implementation detail that is really a musical decision: the reading here is cyclic, so the note at position 15 is followed by the downbeat of the next repetition rather than by nothing. Stopping at the barline would score that note zero, and a note pushed just before a downbeat is about the most common syncopation in the music. These figures loop, so cyclic is the honest reading, but it is a choice and a linear reading of a single bar would give different numbers.
The consequence nobody mentions
Read the definition again and notice what it requires. Every term in it refers to metric weight, and metric weight comes from a metre. So the measure cannot be applied to a rhythm by itself. It takes two arguments.
That is more than a technicality about the formula. Asking how syncopated a pattern is, with no metre specified, is not a hard question. It is an incomplete one.
Same notes, different answer
Change the metre underneath and the score moves on its own.
The bare skeleton of 3/4. Nothing to syncopate against itself. Beats of 3/4 at 0, 4, 8.
- Score in 3/4
- 0
- Score in 6/8
- 1
- Notes moved
- 0
- Bar length
- identical
Start on Three beats and switch between the metres. The notes do not move, the bar is the same length, the step rate is the same, and the pattern's own clicks are scheduled identically either way. In 3/4 it is the beat itself and scores nothing. In 6/8 it is syncopated. Then try Two beats, which is the same statement in reverse. Syncopation is not something a rhythm has. It is something a rhythm is doing to a metre, and with no metre in mind there is nothing to measure.
The beat click does change between the two, and it has to: it is the only thing supplying the metre. That is worth sitting with rather than treating as a flaw in the demonstration. The reason this figure needs a click at all is the same reason the score needs a hierarchy. Turn the click off and the pattern still loops, unchanged, and now genuinely has no syncopation to speak of - not zero, but undefined, because the thing it would be measured against is no longer there.
What that figure shows
The two metres are 3/4 and 6/8, chosen because lesson 2 established they share twelve positions and differ only in the order of the divisions. So the pattern can stay put while the hierarchy under it changes, with the bar length, step rate and onsets all held fixed.
Three beats is notes at 0, 4 and 8. In 3/4 that is the beat, and it scores zero. In 6/8 the beats are at 0 and 6, so two of those three notes are now off the beat, and it scores one. Two beats is the mirror: zero in 6/8, one in 3/4. Each is the plain unsyncopated skeleton of one metre and a syncopation in the other, and neither pattern was touched.
Those two were found by exhaustive search rather than picked by ear, and the general picture is stronger than the examples. Of all 4,095 non-empty twelve-step patterns, 61.5% score differently in the two metres. Metre-dependence is the normal case, not a curiosity at the edges.
There is a control preset for the same reason. Every eighth scores zero in both, so the claim is not that everything changes when the metre does. Some patterns are indifferent to it. Most are not.
Which puts the effect where it belongs
Lesson 1 argued the beat is supplied by the listener. Lesson 2 said the same about grouping. This lesson makes the consequence concrete: if syncopation is measured against the metre, and the metre is in the listener, then syncopation is something happening between the music and the person hearing it rather than inside the music.
That is why the beat click matters so much in the second figure, and why turning it off is worth doing once. The pattern still loops, the notes are unchanged, and the effect simply goes. Not reduced, exactly. There is nothing left for it to be relative to.
It also explains something performers know and notation does not record: a syncopated passage stops working if the listener loses the beat. The notes have not changed. The thing they were pulling against has gone, and a syncopation with nothing to pull against is just a rhythm.