Syncopation
Syncopation
Recall that a measurement study involves no testing of a hypothesis. It simply involves making and reporting some measurements. An example of a measurement study is Huron & Ommen (2006) which looked at the historical changes in syncopation in American popular music. The study took advantage of newly released digital re-issues of early wax cylinder recordings, as well as 78-RPM and 45-RPM recordings of popular music from the early twentieth century. We limited our study to the first 50 years of commercial recording—namely the period from 1890 to the beginning of World War II (1939). The recorded sample of music included 437 musical works. For each work, we randomly selected a single moment (a particular second on the time counter) and transcribed the bar or measure in which the second appeared. From the transcribed notation of the rhythm we determined whether any syncopation was present, and if so, we categorized the type of syncopation.
How We Measured Syncopation
In order to measure syncopation, we had to make an operational definition. In metrically organized music, each moment can be regarded as occupying a particular point in the metric hierarchy. The highest point (labelled “1”) might be defined as coinciding with the downbeat. The second highest point (“2”) coincides with the third beat in 4/4 meter. Other points in the metric hierarchy can be similarly labelled as shown in Figure 1.
Figure 1
Numerical representation of the metrical hierarchy for 4/4 meter. The value “1” designates the highest value in the hierarchy.
Using these “metric hierarchy numbers” we can see an interesting pattern emerge for syncopated moments. For example, in Figure 2a (below), a syncopation occurs because the first note is tied to the second note. With each syncopated event, a note onset occurs at a relatively low metrical level (i.e., a higher number), and is sustained through a moment in the metrical hierarchy that has a higher metrical level (i.e., a lower number) and that does not manifest a corresponding event onset. In other words, there is a kind of metrical “hole” in the rhythm, which we dubbed a lacuna. The lacuna became our operational definition of a syncopated moment.
Figure 2
Examples of syncopation. As in Figure 1, numerical values represent positions in the metrical hierarchy.
What We Found
For each of our sampled musical moments, we counted the number of lacunae, and categorized the type of lacunae pattern. The results of our study are summarized in Figures 3 and 4.
Figure 3
The average number of lacunae (our operationalization of syncopations) per sampled passage. The graph is consistent with the (post hoc) notion that the density of syncopation increased over five decades.
Figure 4
The percentage of irregular syncopations by decade.
There are two conclusions from this study. The first conclusion is entirely expected—grandmother research: it appears that over the 50-year period, the amount of syncopation per unit of time nearly doubled, from just over one syncopation to nearly two syncopations per unit of time. The second conclusion was unexpected. The variety of patterns of syncopations appears to peak in the 1910s. In other words, musicians in the 1910s seemed to have tried all sorts of rhythmic syncopations, and then reduced the variety in the 1920s and 1930s. Until about 1920, there appears to have been an increase in both the quantity and variety of syncopations; after 1920, the variety seems to have dropped, but the density of syncopations continued to increase.
Now recall that empirical studies never prove anything. This is also the case here. It could be that our sample of 437 recordings does not adequately refect the music of the time. It could be that our random sampling of musical moments led to unrepresentative results. Or our rhythmic transcription of these passages might have been inaccurate. So we can’t conclude that “the density of syncopation increased over the 50-year period.” Instead, we should say, “the results are consistent with the idea that the density of syncopation increased over the 50-year period.” Similarly, we should say that “our study implies that the variety of syncopation patterns peaked in the 1910s.” Notice the use of the words “appears” and “seems” in the previous paragraph.
Methodological Observations
Notice that the study was not motivated by any theory. We did not use some theory to generate a conjecture, and then refine the conjecture into a testable hypothesis. The study was, however, motivated by a question: How did the use of syncopation change in American popular music over the first 50 years of sound recording? More specifically, how did the density of syncopation change? And what happened to the various syncopation patterns?
If you read the actual published article, you’ll see that we took care to avoid defining “syncopation.” Instead, we operationally defined a “lacuna.” Many people might be tempted to say that a lacuna is a pretty good way of defining a syncopation, however, we try to avoid essentializing the concept syncopation. Because we gave a precise definition of “lacuna” we could unambiguously count them: “this rhythm has a lacuna right there,” “this passage contains no lacunae at all.” Notice that Figure 3 (above) plots the number of lacunae (not the number of syncopations—whatever that might be).
In Figure 4, we plotted the “percentage of irregular syncopations.” Once again, in the published article, we operationally defined something we called “irregular syncopations.” Once again, we don’t believe that rhythms divide into regular syncopations and irregular syncopations (as though these are “natural kinds” of some sort). This was simply a provisional way of distinguishing commonplace syncopated patterns from less common ones, and then seeing how often the less common patterns occurred in different decades.
Notice that our study tells us nothing about causality. It has no story to tell about why something might have happened. Also notice that the study did not invite failure; we did not test an a priori hypothesis or theory. Having presented our observations, the study might inspire other researchers to form theories regarding why this happened.
Reference:
David Huron & Ann Ommen (2006). An empirical study of syncopation in American popular music, 1890-1939. Music Theory Spectrum, Vol. 28, No. 2, pp. 211-231.