Measuring Rhythmic Complexity in the West African Standard Pattern: An Empirical

Approach

By: Juan Diego Diaz, UC Davis

Paper presented at the fifth international conference on AAWM in Thessaloniki, June 26-29,

2018. (Please do not cite without permission).

Much of the groove-based musics found in West Africa, Central Africa, West-Central Africa,

and their Atlantic diaspora (aka the Black Atlantic) is temporally organized around short

asymmetric ostinati called timelines. The so-called West African standard pattern [2212221] is

one of them. This presentation focuses on one interesting property of this pattern; its rotational

possibilities. That is, the potential of other points in the pattern’s time cycle to be perceived as

reference points or downbeats.

Example 1. Circular representation of the West African standard pattern

This rotational potential is realized in specific traditional contexts (e.g. Nigerian Yoruba

drumming [2212212]; Candomblé drumming [2221221]; Central African Republic’s Pygmies

Bondo [1222122] and [2122212]) and in experimental Afro-diasporic musics from Brazil such as

Orkestra Rumpilezz’s “Das Arabias” [1222122] or Moacir Santos’s “Kathy” [1221222].

1
 Example 2. 12/8 Timelines in various black Atlantic genres

Waves of Metric Strength

At the previous meeting of the AAWM conference in NYC in 2016, I argued that Letieres Leite,

the composer of Orkestra Rumpilezz, deliberately aims to increase rhythmic complexity by

manipulating the standard pattern and other traditional timelines. The techniques he utilizes

include expanding, shortening, rotating, and staggering timelines. In examining this rhythmic

complexity, I considered four parameters: metric strength, syncopation, familiarity, and

ambiguity.

That study rested on a number of assumptions. The first was that the less a rhythm

matches metric beats, the more complex it is. In other words, that pattern misalignment increases

rhythmic complexity. If we take Rotation 1 (R1) as the reference, the complexity of rotations can

be measured according to the number of coincidences of that rhythm with rotation R1.

2
 Example 3. Seven rotations of the standard pattern notated in TUBS

The second assumption was based on the concept of metric strength, which pays attention

to the measurement of syncopation in relation to metric frameworks. R1 of the standard pattern is

characterized by the articulation of beats 1 and 4 and the syncopation of beats 2 and 3 in 12/8.

There is a flow from metric consonance to dissonance and back to consonance. Like in other

types of timelines, beat four is a point of resolution of accumulated rhythm tension. I represented

this with the help of Chris Stover’s (2009) waves of metric strength, with a shape that is wide

when a metric beat is articulated or narrow otherwise, as shown in Example 4.

Example 4. Waves of metric strength for the standard pattern. After Stover (2009: 141).

I then proceeded to build waves for seven rotations of the standard pattern, each

corresponding to an onset of the pattern and did this for 12/8 and 3/2 readings. I proposed that

the seven rotations of the standard pattern, resulting from using each onset as a downbeat, can be

3
classified in two groups or families: a family of rotations that is close to the standard pattern and

another one that is distant (see Figure 1). I assumed that listeners who have a high degree of

familiarity with the standard pattern in its basic form (R1), will find rotations with similar waves

to be less complex and rotations with different ways to be more complex.

Example 5. Waves of metric strength for seven rotations of the standard pattern in two meters

The third assumption was related to ambiguity and familiarity: familiarity with the base

pattern can increase complexity when a rotation of it triggers two competing interpretations of

(what seems to be) the same phenomenon.

I was able to verify some of these assumptions ethnographically while listening to music

based on these rotated patterns with some audience members in Brazil and Ghana. In “A Grande

Mae” by Rumpilezz for instance, (based on R1, called vassi in Bahia) people are able to identify

and clap the bell pattern throughout easily. But in “Dasarabias,” also by Rumpilezz, based on a

distant rotation (R5), they identified vassi and displaced the RTP to match the familiar gestalt of

R1, but clashing with the music’s RTP.

4
Goals

The main goal of this paper is to test empirically my assumptions of rhythmic complexity in

relation to my grouping of the standard pattern rotations. A broader goal is to examine the

rhythmic complexity of each of the twelve possible rotations of the standard pattern. The paper is

guided by the following question: How do musicians who are familiar with the standard pattern

in its basic version [2212221] react when the point of reference is shifted in time? More

specifically, how these shifts affect rhythmic complexity? Inspired by works on music cognition

and music perception (Fitch & Rosenfeld. 2007; Povel & Essens 1985; Shmulevich & Povel

2000), I engage this question from an empirical standpoint with three psychological experiments

on rhythm carried in the city of Accra, Ghana, one of the locations where this pattern is widely

used.

The Experiments

The experiments examine three measures of rhythmic complexity: perceptual (how complex a

person thinks a rhythm is); metric (how difficult it is to reproduce the underlying beat of a

rhythm), and performative (how difficult it is to reproduce a rhythm against a given underlying

beat).

In order to study issues of syncopation and metric strength more thoroughly I examined

the 12 possible rotations of the standard pattern (including those with non-articulated

beginnings). Because metric context (whether explicit or implied) is central to the perception and

cognition of rhythm, the experiments test each rhythm alone and in two relevant metric contexts,

12/8 (which is the emic perspective among my interlocutors) and 3/2 (often implied by

musicians’ phrasings or built into accompanying ostinatos). In sum, the stimuli for each

5
experiment consists of three sets of rhythmic sequences:

(1) The 7 rotations with articulated beginnings played by themselves—a subject will not be

able to distinguish between rotations 2 and 3, for instance;

(2) The 12 rotations played in conjunction with a beat stream corresponding to a 12/8 meter

(i.e. numbers 1, 2, 3, and 4 in the slide)

(3) The 12 rotations played along with a beat stream corresponding to a 3/2 time signature

(i.e. numbers 1, 2, and 3 in the slide).

All these possible combinations can be seen in Example 6.

Example 6. TUBS representation of 12 rotations of the standard pattern in 12/8 and 3/2

Stimuli presentation

Each of these rhythmic sequences was transcribed in Sibelius 7.5 software and played as MIDI

through headsets to each participant. The stimuli presentation was manually controlled by the

experimenter. Each pattern was repeated four times and after 2 to 3 seconds, each participant

responded to a question or performed a previously assigned task. Patterns were presented in

random order, different for each participant. But the experiments always progressed from

6
perceptual, to metric, and to performative. The tempo was kept stable at 100 BPM (600ms) for

the dotted quarter-note in 12/8, which corresponds to 75 BPM (450ms) for the half note in 3/2

meter. 100 BPM is a common tempo in genres where the standard pattern is used. It is also an

ideal tempo for experiments on rhythm as it is within a range of greatest pulse salience

(Toiviainen and Snyder 2000:2) and falls into the indifference interval—where people’s

judgments of a duration are less biased (Fraise 1963).

Participants

Figure 1. Ghanaian participant in experiments of rhythmic complexity (Accra, 2016)

The participants are twenty Ghanaian music students and lecturers who are familiar since

childhood with the standard pattern in its basic rotation [2212221]. I recruited them at the Music

Department of the University of Ghana, where I worked in the 2015-16 school year, and

performed the experiments between 2015 and 2017. In order to isolate the variable of familiarity,

7
thirty-three music students at King’s College London who are unfamiliar with the standard

pattern were used as control subjects.

Experiment 1: Perceptual Complexity

In this experiment, subjects judge the complexity of 31 rhythmic sequences. They are organized

in the following manner: 1) the seven rotations that have articulated beginnings (1, 3, 5, 6, 8, and

10) without accompaniment; 2) the twelve rotations accompanied by a 12/8 beat; and 3) the

twelve rotations in tandem with a 3/2 beat.

Each subject and the experimenter sat in an empty room. After the presentation of

the stimulus, participants were asked to judge the complexity of the stimulus on a 5-point scale

where 1 represents the simplest and 5 the most complex. Subjects were invited to define and

judge complexity in their own terms. They were allowed to listen to a sequence more than once

and to adjust their response before the next stimulus. Each subject was presented with three

random stimuli (one unaccompanied, another one along with a 12/8 beat, and the last one in

combination with a 3/2 beat) in order to get them acquainted with the procedure and to calibrate

their scale for their judgments. This experiment lasted approximately fifteen minutes. Figure 2

shows the mean values of perceptual complexity for the 31 rhythmic sequences.

8
 Figure 2. Histogram with average results of perceptual complexity experiment

Experiment 2: Metric complexity

In this experiment, subjects judge the complexity of 14 rhythmic sequences. The sequences were

organized in two subgroups: 1) the seven rotations with articulated beginning in combination

with a 12/8 beat; and 2) the same seven rotations in combination with a 3/2 beat.

After the presentation of the stimulus and a short pause, participants heard the rhythm

repeated four times, but this time unaccompanied, and were asked to clap, tap, or sing the beat

along. If the subject failed to perform the task correctly, the sequence stimuli-attempt was

repeated up to five times. The experimenter assigned a number from 1 to 5 corresponding to the

number of times the subject took to perform the task correctly. If the subject did not succeed at

the fifth presentation of the stimuli, number 6 was assigned for the calculation of average scores

of complexity. Failed attempts were also counted in a separate list. Subjects had the opportunity

to practice three times beforehand to become familiarized with the rhythms and tasks. The entire

experiment lasted approximately ten minutes. Figure 3 show the average scores of metric

complexity for the 14 rhythmic sequences. Figure 4 shows the total number of failed attempts in

this experiment for each rotation.

9
 Figure 3. Histogram with average scores of metric complexity experiment

Figure 4. Histogram with number of failed attempts of metric complexity experiment

Experiment 3: Performative Complexity

This experiment tests complexity of the same 31 rhythmic sequences examined in the perceptual

experiment: 1) the seven rotations that have articulated beginnings without accompaniment; 2)

the twelve rotations accompanied by a 12/8 beat; and 3) the twelve rotations in tandem with a 3/2

beat.

Participants first heard the seven rhythms without accompaniment and were asked to

clap, tap, or sing the rhythm afterwards—here the critical point was to begin the rhythm at the

10
same point as heard. For the rest of the sequences, after the presentation of the stimulus and a

short pause, participants heard the beat alone, and were asked to clap, tap, or sing the rhythm. In

all cases, if the subject failed to perform the task correctly, the sequence stimuli-attempt was

repeated up to five times. The experimenter assigned a number from 1 to 5 corresponding to the

number of times the subject took to perform the task correctly. If the subject did not succeed at

the fifth presentation of the stimuli, number 6 was assigned. Failed attempts were also counted in

a separate list. Subjects had the opportunity to practice three times beforehand to become

familiarized with the experiment. The entire experiment lasted approximately twenty-five

minutes. Figure 5 shows the average scores of performative complexity for the 31 rhythmic

sequences. Figure 6 shows the total number of failed attempts in this experiment for each of the

12 rotations.

Figure 5. Histogram with average scores of performance complexity experiment

11
 Figure 6. Histogram with number of failed attempts of performative complexity experiment

Discussion

Let me recapitulate the main idea that I put to the test with these experiments. I proposed that the

seven rotations of the standard pattern with articulated beginnings could be subdivided into two

families: close (R1, R6, R8) and distant (R2, R3, R5, R12). I assumed that people with high

familiarity with the base rotation (R1) would perceive and score lower rhythmic complexity in

rotations of the close family and higher in the latter family.

As shown in Table 1, the empirical evidence shows that: 1) in average the close families

have lower values of complexity than the distant family in all measures of complexity; and 2)

Although the two families rotations do not cluster neatly at either extreme of the spectrum, they

tend to verify the hypothesis.

12
 Table 1. Order of complexity for the three experiments (rotations with articulated beginnings)

If we consider the 12 rotations, we notice that rhythms with syncopated beginnings tend to have

higher values of complexity, at least in average (see Table 2). These results align with those of

others who have proposed a direct relationship between syncopation and rhythmic complexity

(Fitch and Rosenfeld 2007). Interestingly, the mean values of complexity of rotations with

syncopated beginnings were markedly higher in the performative experiment than in the

perceptual experiment. There is a difference between what people think is rhythmically complex

and how well or badly they did when that complexity was tested through a specific task.

Generally speaking, participants thought they could do better than they actually did. This can be

more clearly seen in Table 3 where the average values of the three experiments are grouped by

wave family. This table also shows that, in average, rotations with syncopated beginnings have

higher values of complexity than those with articulated beginnings—a fact that is particularly

true for the experiment in performative complexity.

13
 Table 2. Order of complexity for the three experiments (all rotations)

Table 3. Mean values for all experiments with results grouped by wave family

Control Subjects (KCL)

I have conducted experiments with control subjects and, so far, only have sufficient data for the

analysis of perceptual complexity. In October, 2017 I conducted a collective experiment of

perceptual complexity with 33 undergraduate music students at KCL. Their average age was 20

years old. 19 of them had no previous exposure to traditional African or Afro-derived musics.

The rest had played some Afro-diasporic genres such as bossa nova or salsa. These experiences

14
may have given them some familiarity to other timelines such as son clave, but not to the

standard pattern, according to their responses to a survey. Figure 7 show the mean values of

perceptual complexity for the 31 rhythmic sequences.

Figure 7. Histogram with average results of perceptual complexity experiment (control group)

In the same manner as I did with my Ghanaian subjects, I consolidated the results of the

experiment with the control group from KCL to see the distribution of the families of rotations.

In Table 4 we can see that the close family moved from low to medium complexity values. This

isolates the variable of familiarity with the base rotation, and gives a first basis for empirical

verification of the hypothesis of family rotations. Finally, in Table 5 we also see that, in average,

rotations with syncopated beginnings tend to score higher in all measures of complexity.

15
 Table 4. Order of complexity for the three experiments (all rotations). Control group.

Table 5. Mean values for all experiments with results grouped by wave family (control group)

A Son Clave Principle?

Lastly, I’d like to go back to the waves of metric strength of the 12 rotations to reflect on one

particular inconsistency I found between my metric and empirical analyses. Figure 8 shows the

waves and points of mismatches in both metric contexts as well as onset mismatches in relation

to R1. You may have noticed that in this chart R7 has maximum misalignment with R1 by all

counts and is thus the most distant rotation to R1.

16
 Figure 8. Waves of metric strength for 12 rotations of standard pattern with various kinds of

metric and pattern mismatches.

However, this maximum distance to R1 does not correspond with maximum scores of

complexity in the consolidated results of the Ghanaian experiments as shown in Table 2. R7 is

actually has one of the lowest scores among the non-syncopated rotations—reflected by its

position towards the left of the chart among the yellow results. R7 even scores lower than many

syncopated rotations in the perceptual experiment.

In the circular notation of the standard pattern and clave shown in Example 7, we can see

that R7 starts half way through the 12-beat cycle in reference to R1. In other words, R7’s

beginning is located diametrically opposite to R1. Godfried Toussaint (2010:5) has written that

rhythms with a pair of antipodal onsets (like R1 and R7) neutralizes the rhythmic oddity property

of the standard pattern. Said differently, it reduces the perceived complexity of both rhythms

with respect to each other.

A more intuitive way to see the relationship between R1 and R7 is by comparing it to son

clave, which traditionally is used with two diametrically opposite beginnings as shown in the

17
Example 7. The use of these two rotations of clave are common place in salsa, Afro-Cuban

music and other diasporic genres. Are R7’s relatively low scores of complexity the result of a

deep rooted timeline logic where a pair of rotations with diametrically opposite beginnings are

perceived as related? Do R1 and R7 behave analogously to 2-3 / 3-2 clave orientations? To my

knowledge, the 3-2 / 2-3 clave orientation principle is less prominent in Ghana than in some

places of the diaspora like Cuba. I hope to have made a strong case that a combination of

analytical, empirical, and ethnographic approaches can provide answers to this and other related

questions.

Example 7. Circular representation of clave and standard pattern

Bibliography

Agawu, Kofi. 2006. “Structural Analysis or Cultural Analysis? Competing Perspectives on the

Standard Pattern of West African Rhythm.” Journal of the American Musicological

Society, 59(1): 1-46.

Anku, Willie. 2000. “Circles and Time: A Theory of Structural Organization of Rhythm in

African Music.” Music Theory Online 6(1).

Arom, Simha. African Polyphony and Polyrhythm: Musical Structure and Methodology.

Cambridge University Press.

18
Diaz, Juan Diego. 2017. “Experimentations with Timelines: Strategies of Rhythmic

Complication in Afro-Bahian Jazz.” Analytical Approaches to World Music, 6(1).

Fitch, Tecumseh and Andrew Rosenfeld. 2007. “Perception and Production of Syncopated

Rhythms.” Music Perception: An Interdisciplinary Journal, 25(1): 43–58.

Fraisse, Paul. 1963. Psychology of Time. New York: Harper.

Povel, Dirk-Jan & Essens, Peter. 1985. “Perception of Temporal Patterns.” Music Perception: An

Interdisciplinary Journal, 2 (4): 411- 440.

Shmulevich Ilya & Povel, Dirk-Jan. 2000. “Measures of Temporal Pattern Complexity.” Journal

of New Music Research, 29(1): 61-69.

Stover, Chris. 2009. “A Theory of Flexible Rhythmic Spaces.” PhD dissertation, University of

Washington.

Toissaint, Godfried. 2010. “The Rhythm that Conquered the World: What Makes a “Good”

Rhythm Good?” Percussive Notes, 48(6): 52–59.

19