SONGS OF THE KETU CULT OF BAHIA 53

SONGS OF THE KETU CULT OF BAHIA, BRAZIL

by
ALAN P. MERRIAM

In Afro-American studies, within which the following analysis falls, the anthro-
pologist can establish an unusual degree of control over the variables to be considered.
Here he has a basis for arriving at an understanding of observed change in that he cah
establish historical depth, a control situation in which the constant is the common
cultural heritage of the African slaves, and dependent variables represented by the
different socio-cultural environments which the Negro encountered as a slave in the
New World setting. Thus the requirements for a laboratory situation are met insofar
· as the always unpredictable human element may be counted upon to remain constant.
In this situation, then, the study of African cultures or cultural elements in the
New World may be placed against the cultural background of Africa itself, thus giving
the anthropologist an opportunity for controlled study of culture change. Further,
the various socio-rultural environments of which the Mrican Negro became a part
in the New World may be tested and evaluated in terms of their influence on the basic
Mrican culture, and thus a better understanding of the processes of acceptance, rejection,
and syncretism may be reached.
In the study of these acculturative processes, music is of considerable importance,
both because it is less susceptible to direct attack in the acculturative situation than
sbme other aspects of culture, and at the same time is particularly susceptible of objective
description. In the first instance, music seems to be patterned and carried on sublimin-
ally and thus to be less liable to outside or intrusive conditioning than are, for example,
food habits, or the construction of houses. Thus one deals with a phenomenon which,
in a cultural situation in which the members of one culture dominate those of another
as was the case in New World slavery, probably cannot be stamped out forcibly, and is
very likely to be retained in the new situation.
As an aesthetic aspect of culture which can be studied objectively, music is of
heightened importance in acculturation studies. While in other branches of the arts,
few techniques have as yet been devised for objective study, music can be objectified
in many aspects; analysis of time, tempo, interval usage, modal structure, and many
other characteristics may be noted quantitatively rather than qualitatively.
Thus, in these two respects music is particularly useful in acculturation studies;
its importance is further enhanced, finally, in that it is one of the creative aspects of
culture, and its study can therefore lead to an understanding of the basic human problems
of creative life, as well as the values, goals and meanings of the culture in which it
functions. Through music, then, one may probe the problems of cultural stability,
individual variation in performance, the reworking of old values in new situations, and
most particularly, the general processes of culture change.
The present study is directed toward establishing the music structure of an Mrican-
derived cult group of Bahia, Brazil. Such work is at present of considerable importance
to the entire study of the musical relationships between New World and African Negro
cultures; to date very few studies have been made. Before ethnomusicology can begin
to make its most significant contributions to the study of the New World Negro accul-
turative problems and the broader questions of cultural dynamics that derive from it,
a series of such works as that being undertaken here must be completed, so that factual
information will become available for comparative data,l
54 AFRICAN MUSIC SOCIETY JOURNAL

I
The question of the provenience of the Negro slaves taken to Brazil poses many
problems. For example, Ramos notes that, "from colonial times to the present day,
the popular terms of nago, mina, Angola and Mozambique have been employed to indicate
vaguely the portion of Mrica from whence the Negroes came. A still more popular
nomenclature developed in which such terms as India piece, Guinea black and coast. Negro
were used."2 To the difficulty of dealing with these vague terms similar difficulties
such as poorly kept records and the destruction of previous records must be added.
However, in resume, it is instructive to turn again to Ramos, who sums up as follows:
At the beginning of the slave trade, the largest number of those imported into Brazil were
from Angola, the Congo and Guinea. When more active communication began with Bahia3 the
leading source of supply was Guinea and the western Sudan. There began a remarkable influx
of Y orubas, Minas from the Gold Coast, Dahomans and various Islamized tribes such as the Hausas,
Tapas, Mandingos, and Fulahs.
Thus it was that three great Negro peoples entered Brazil. In the first group were the Sudanese
Negroes-the Y orubas and Dahomans together with the Ashanti. In the second group were the
Moslem Negroes-the Hausas, Tapas, Mandingos and Fulahs. The third group comprised the
Bantu family-the Angolas, Congos, Mozambiques and several lesser tribes. 4
In Bahia, the area with which this paper is concerned, "Negroes belonging to the
first and second groups predominated," while the "Yoruba Negroes were preferred."5
In view of the concentration of Negro populations in Bahia and in Brazil as a whole,
it is predictable that the cultures of these people should have, in great part, survived
the geographical transitions of demographic movement, and continued to exist in the
New World as a powerful integrating factor in the lives of the Africans. Religious
practices, as the cultural focus of West African life, about which centered a host of other
complexes including music and folklore among others, have been retained in the New
World, and especially in West Indian and South American areas in which the Catholic
religion provided a particularly fertile opportunity for syncretism between the two
religious traditions. In Bahia, and in other northern Brazilian cities, the African religious
system has been retained in particularly strong form to the present time, functioning
in cult groups as a living system of beliefs and practices.
The Mrican-derived religious cult group in Brazil is called the candombte, and is
based upon an order of orishas, or deities-West African gods whose efficacy has been
retained in the New World. Some of these gods are Osha/a, the most powerful deity,
Shango, the god of lightning, Ogun, the god of war and iron, Eshu, the trickster and god
of evil, Osbun, a water deity, Yansan, the god of wind and storm, Oshunmare, the personi-
fication of the rainbow, and many others. These gods are grouped in a pantheon, and
each is clearly identified by a complex of traits including sex, symbol, insignia, special
foods, colours, beads, bracelets, sacred days, and special calls. For Shango, for example,
according to Pierson, sex is male, the symbol is lightning, insignia is meteoric iron,
lance or hatchet, special or sacred food is cock turtle or he~goat; the colour is red,
beads are red and white, bracelets are made of brass, the sacred day is Wednesday, and
the special call hay-ee-ee6 • In addition, the gods of the African pantheon have come to
be identified with the Catholic saints-Ogun with St. Anthony, Oshossi with St. George,
· to name but two. These syncretisms are based both on the function of the deities in
the two systems, and upon the supposed functions of the Catholic saints as derived
from their pictured representation in chromolithographs. 7
Seven major, and two minor candombles are represented, of which three are sup-
posedly of Indian origin, and the remaining six, of Mrican origin. The ceremony in
the Ketu and]esha groups, (Nago in Portuguese), is derived from the religious practices
of the Yoruba people of Nigeria.8 The Congo Angola group takes its name from the
peoples brought from the southern and eastern fringes of the Congo basin. 9 The Gege
group is Dahomean in derivation,10 and the two minor groups, Egba and Efan, Yoruba.
 SONGS OF THE KETU CULT OF BAHIA 55

The Caboclo, Guarani, and Caboclo Guarani groups incorporate deities and practices
of indigenous Indian beliefs with those of African origin, particularly of Congo
derivation.11
Worship in all these groups is based on the world-view that the destiny of the Universe is in
the hands of deities that are everywhere the same, though the ·names they bear vary from region
to region and from people to people according to the language that is spoken. The destiny of man,
who is but a modest part of this Universe, is ruled by the same gods, but man enjoys the inter-
cession of a hierarchy of ancestral dead, who in death as in life continue to be preoccupied with the
well-being of the family to which they belong. Indeed, the gods appear to have given the ancient
dead a certain autonomy in regulating the moral code of their descendants ill the interest of human
well-being, though they have not abrogated their own powers to regulate the conduct of the living
members of each family. On the contrary, each individual has his or her god as a personal spirit.
Man is not, however, a passive agent in relation to his destiny, for through divination he can
discover the secrets of that destiny, his well-being, and his status in the group among which he
lives. 12
While marked deviations occur within the ceremonial for each cult group, or
"nation", some general terms and forms of ritual may be indicated here. The cult
centers found throughout the city of Bahia are called seitas, the sacred grounds on which
the house of worship stands, the terreiro. Each cult center contains several sacred spots,
the most important of which is the pegi, or altar to the orisha. Each seita is presided
over by a pai-de-santo, or mae-de-santo, terms which may be loosely translated as "priest"
or "priestess," respectively, although it seems best to retain the original term in dis-
cussion. The function of the pai- or mae-de-santo is to arrange for, and lead the worship
for the orishas; at the same time, the leader holds the secrets of worship, is an interpreter
of tradition, and directs the affairs of the cult group. Various helpers assist with ritual,
and these include the filhos or filhlfls de santo who are the initiates of the group; the
ogans-male members of the cult who assist in the ritual in various capacities; the
achogun, who performs the sacrifices; the musician, and others. In arranging for a public
ceremony, the officiating priest or priestess first brings together the initiates of the
group and begins the ritual of the particular ceremony in honour of the deity whose
feast is being celebrated. The despacho, which appeases the trickster deity, Eshu, so
that, though himself "sent away," he will permit the other gods to come, assures the
participants that he will not interfere with the remainder of the ceremony. Dances
and songs to the orishas follow, leading eventually to the possession of some dancers
which indicates that the god has entered into their heads. Possessed initiates are handled
with precise techniques throughout their possession and are eventually released from
the god with special ritual.13 Those possessed dance before the drums to the songs
specially reserved for the particular god represented, and finally, when "the gods no
longer show by their tension a further need to dance, the rite comes quietly to a close." 14
Animal sacrifice is also an important ceremony in the cult group, as is the ritual which
surrounds the entrance of the initiate into full cult participation. These Bahian cult
groups have been extensively described by several students.1 5
In the worship of the orishas, music has paramount importance, since it is by means
of songs and rhythms that the gods are "called" to participate in the ceremonies. Per-
cussion instruments play a dominant role in Mro-Bahian cult music; not only are the
instruments themselves important, but the musicians who play them fulfil special func-
tions within the hierarchy of the group, and may be regarded as professionals in this
particular branch of religious life.
The use of musical instruments tends to split into two patterns depending upon
the particular cult groups in which they function. For the Ketu, Gege, and Jesha
groups, the iron gong and drum are most often used, while in the Caboclo, Guarani
and Cabocle Guarani groups, rattle, calabash, and guitar, as well as drum and iron
gong are employed. Both percussion groupings seem to be used in the Congo Angola
songs.
56 AFRICAN MUSIC SOCIETY JOURNAL

"The drums employed in the cult rituals are termed hu ('the largest'), hunpri ('the
medium-sized one'), and le ('the smallest'),"16 and are used in sets of three, termed
terno. In the recordings 'a t hand, however, the most common unit is two drums and
iron gong. At the same time, since the two smaller drums usually play the same rhyth-
mic patterns, the important rhythms are present when only one of the smaller drums is
used with the large drum.U
The drums are ordinarily termed atabaque, and are of the African hollow-log type,
although at present they are most often made from barrel staves. The preferred material
for the drum-head is either deer or calfskin; the head is attached to the drum by means
of wooden circlets. The skin is then firmly secured by a special rope which is passed
alternately through slits made in the skin and around pegs of hard wood driven into
the body of the drum. Drums are played either by hand, or with drumsticks fashioned
from hard wood. Before the instruments may be put into operation they receive power
through a series of "painstakingly performed rituals." Drums are almost never played
by women.18
The iron gong, or agogo, is a piece of iron shaped in most instances somewhat
like a cowbell; this instrument is struck with a piece of iron or other metal.l 9
In the supposed Indian-derived cult groups, the gourd rattle is an important
instrument; this may be covered with netting to which are attached shells or other
materials that contribute to the sound. The larger calabash may either be played with
sticks or with the hands, the two methods producing very dissimilar results. Finally,
the Spanish guitar, played by conventional methods, is widely used. 20
The collection of songs from which the sampling is taken was recorded by M. ].
and F. S. Herskovits in 1941-42 in Bahia, Brazil; twenty nine of these songs-of the
Ketu cult-are analyzed in the following pages. All are of a religious character, including
not only songs for specific gods within specific cult groups, but also song cycles and
individual songs for various phases of cult ceremony. The analytical method used is
based upon the work of von Hornbostel, 21 which has been elaborated by Kolinski,22
Waterman,23 and to some extent the present author. 2<l The approach is aimed toward
obtaining as objective a description of a musical style as possible; to this end various
quantitative procedures have been adopted, although more subjective evaluations are
not completely disregarded. The characteristics, then, of the songs of the Ketu cult
group follow.

II
Tonal range
Nineteen of the 29 Ketu songs have ranges of more than an octave, and this wide
range may be considered characteristic of the group. Of the remaining ten songs,
eight have ranges of nine or more semitones but less than an octave, while one shows
an abnormally small range of but five semitones, and another of but seven semitones.
In general, however, the wide tonal range characterizes the Ketu group.
Melodic direction
Since 19 of the 29 songs show the final tone from two to fifteen semitones below
the beginning tone, it may be concluded that this downward melodic direction is repre-
sentative of the group. At the same time, eight songs have identical initial and final
tones, while in two other songs, the final tone is n;vo semitones above the initial tone.
Of those songs in which the overall direction is downward, the largest number
-eight-show the ending tone seven semitones, or a perfect fifth, below the initial
tone, while in three songs it is two semitones, and in three songs, five semitones below
the initial tone. In single songs, the beginning tone is 3, 10, 12, and 15 semitones above
the final.
 SONGS OF 'I'HE KE'I'U CUL'I' OF BAHIA 57

In eight of the 29 songs no tone is higher than the initial tone; the remaining 21
songs, however, show an upward movement in varying degrees within the songs.
In six songs the highest tone is seven semitones, or a perfect fifth above the initial
tone; in three groups of two songs, it is two, three, and twelve semitones above the
initial tone. In three songs each, the highest tone is five or nine semitones above the
beginning tone, and finally, in single songs, it lies ten, 11 and 14 semitones above the
initial tone.
The range from beginning to lowest tones is both varied and in general, wide,
although two songs show the two tones identical. In the largest single grouping-
eight songs-the lowest tone is seven semitones below the initial tone. In two groups
of four songs, it is five and twelve semitones lower, respectively, while in two songs
each, the lowest tone is three, nine, or ten semitones below the beginning tone. In
three songs, the difference of 15 semitones is extremely wide, and finally, in single
songs, the .two tones are separated by two, or four semitones.
A similar range is found when ending and highest tones are compared. In no
case are the two tones identical, although in one song they are but three semitones
apart. Eight songs~the largest single grouping-show the highest tone seven semi-
tones above the ending tone; in six songs, it is an octave higher. In four songs, the
highest tone is five semitones above the ending tone, and in three groups of two songs
each, it is nine, 14 and 15 semitones above the final tone, respectively. In the remaining
songs, the highest tone lies 10, 11, 16, and 17 semitones above the ending tone.
In seven of the 29 songs, no tone is lower in pitch than the ending tone; in six
songs the lowest tone is but two semitones below the ending tone, and in four songs,
three semitones below. Eight songs show the lowest tone five semitones below the
· ending tone; in one song, it is seven semitones below, and in single songs, four and
nine semitones below the ending tone respectively.
Intervals
Of a total of 1,475 ascending intervals, the major second is most frequently used
(38.6%), followed by the minor third and perfect fourth (21.3% each), and major third
(12.6%). The perfect fifth, minor second, major sixth, minor seventh, augrp.ented
fifth, major tenth, and octave combined account for only approximately 5% of the total
ascending intervals. Of 2,225 descending intervals, the major second accounts for
39.4%, the minor third for 22.7%, perfect fourth, 19.9% and major third, 14.5%. The
minor second, perfect and augmented fifth, augmented fourth, and major sixth together
account for less than 5% of the total.
Thus in both ascending and descending intervals, the major second is in highest
proportion (39%), followed by the minor third (22% ), the perfect fourth (20.6%),
the major third (13.5%), the perfect fifth (2.3%), and the minor second (1.3%), while
the augmented fifth, major sixth, minor seventh, augmented fourth, major tenth, and
octave together account for less than 2% of the total.
Descending intervals are in proportion, 58.5% ; ascending intervals, 41.5% .
Narrow intervals, those of a minor third or less, account for 32.3% of the total ascending
and descending intervals used, medium intervals, the major third, for 13.5% and wide
intervals, those greater than a major third, for 24.2 %· All songs use wide as well as
narrow intervals, although one song utilizes but ten ascending medium intervals, while
another uses one perfect fifth as its only wide interval. It is also to be noted that the
greatest proportion of skips of more than a perfect fifth is in ascending intervals.
Half-step intervals play a very minor part in the Ketu songs, accounting, as pointed
out above, for but 1.32% of the total intervals used. The absence of half-steps indicates
the anhemitonic model structures; no hepta modes are present, and the bulk of the
songs is pentatonic.
58 AFRICAN MUSIC SOCffiTY JOURNAL

Four or more successive tones sung on the same pitch are found in 13 of the 29
songs. Four, five, and six tones are found in two songs, but it cannot be said that these
patterns dominate any one song. The repetitive monotone may not be considered an
outstanding characteristic of Ketu music.
Combinations of intervals
Twenty different interval patterns are utilized in the Ketu songs. Of these, patterns
of thirds appear in 21 songs, and patterns of fourths in 19; no fifth patterns are present
while hybrid patterns appear in four songs.
Of the combinations of thirds, the triadic split fifth, Mf, is most often used,
appearing in 11 songs. 25 Also present are the triadic split fifth, mf, triadic split fifth,
mr, triadic split fifth, dim r, triadic split fifth, Mr, pendular thirds, mrfr, and pendular
thirds, mfrf. In more extended pattern are pendular thirds, mrfrf, pendular thirds
mfrfr, and pendular thirds, mrfrfr.
Of patterns of fourths, that most frequently encountered is two fourths, fr.
Also of importance are two fourths, rf, and interlocked fourths, f. Linear fourths, ff,
fourths, frf, and interlocked augmented fourth and perfect fourth, f, are found sporadi-
cally through the sampling.
The most frequently used hybrid pattern, the interlocked perfect fourth and major
third, f, is found in but three songs. The interlocked minor third and perfect fourth, f,
and the interlocked major third and perfect fourth, f, each appears in one song.
Patterns of thirds and patterns of fourths seem to be of equal importance throughout
the Ketu songs. Both may be considered characteristic of the group.
Meter and melodic rhythm
Meter may be established as "duple", or "triple", depending upon the number
of pulses, or accented beats which occur within the measure. Thus 2/ 4 and 4/ 4 time
are both duple; 6/ 8 time is also regarded as a duple meter becausr two major pulses
are discernable, although these are subdivided into groups of three. 3/4 time is a triple
meter. 6/4 time is equivocal; it may be conceived as either triple or duple depending
upon the accent. 12/ 8 time must probably be considered as duple meter; four major
pulses are present in each measure, although they are subdivided into groups of three
notes each.
Fourteen of the Ketu songs are in 6/4 time, while ten songs are notated in 4/4
time, three songs in 3/4 time, one in 6/ 8, and one in 12/ 8.
Melodic rhythm is for the most part in accord with the established meter of each
song, although some notable exceptions occur. In song 64A3, for example,26 the
melodic rhythm of two measures is in duple meter against the predominant triple meter
established in this case by the percussion instruments. The pattern is as follows:
64A3

8J d I
Other patterns of similar import are found throughout the collection, although such
superimposition of a diverse melodic rhythm upon an established percussion meter
never completely dominates a song.
The most common offbeat pattern occurs in the form of syncopation; this device
for heightening tension is found .in all the songs. However, further extension of the
offbeat pattern is also present; three types may be distinguished. Perhaps the simplest
type is the consistent anticipation of the established pulse by a fraction of a beat. In
this type of offbeat patterning the singer reaches the final note of the phrase a trifle
ahead of his expected arrival, giving the entire line a "rushed" feeling. Such a pattern
 SONGS OF THE . KETU CULT OF BAHIA 59

occurs in song 82A1, in which the anticipation is begun on the third beat ot the measure,
and not resolved until the last half of the sixth and last beat.

82A1

Offbeating of this type is also found in other songs.

A more important form of offbeating is that in which the melodic accent consis-
tently falls between the normal percussion beats over longer phrases than those noted
above. This form of offbeating is quite common in the Ketu group, and is clearly
illustrated in the following example: ·

62B1

In this case the melodic tension is heightened through the relatively long offbeating
which is finally resolved at the end of the second measure.
A device related to offbeat phrasing occurs when entire phrases are moved forward
or back in relation to the percussion beat, although the melodic line remains unchanged
as an entity. Most common here is the melodic displacement ot a beat or half a beat;
this may or may not be finally resolved. Song 46A1, however, illustrates the length
to which this device may be carried. In this song, the displacement is over half of a
6/8 measure; the two examples below show the melodic line both before and after
this displacement.

46A1
1 rJ? r1 tJ 1 J 1;J. .I

]} t t} Fir·
While this particular example is an extreme case, it serves to indicate the extent to which
offbeating may be carried, This general method of increasing interest in the melodic
line is utilised fully in the Ketu selections.
Melodic modes
The majority of Ketu songs utilize the pentatonic; 20 of 29 songs or 68% of the
sampling fall within this scale. Six of the songs use the penta sol mode,27 four the penta
do, and three each the penta mi and penta Ia modes. Three songs employ the penta re
type, of which one is structured 4:5:2, and one double mode, penta do re, is present.
Approximately 25% of the songs utilize the hexatonic scale; of these, four are
in the hexa re mode, two in the hexa Ia, and one in the hexa sol mode.
The two remaining songs are in the tetratonic; one is tetra re, and the other, tetra Ia,
is structured 3:4:6.
No songs utilize the heptatonic; this has been implied above where the absence
of half step interv;lls in the melodic line was noted.
60 AFRICAN MUSIC SOCIETY JOURNAL

Of the 29 songs analyzed, 18 show the subjective tonic and duration tones to be
identical, 28 and in the remaining 11 songs the two tones show definite relationship;
In six of these, the duration tone falls on the dominant of the subjective tonic scale.
in two songs on the third of scale. In two other songs, it is on the sixth, or relative
minor tonic ot scale, and in one song on the fourth of scale. In the single double mode,
the first duration tone coincides with the subjective tonic. Thus the discrepancy
between subjective tonic and duration tones is not as wide as it may first appear; it is
entirely plausible that the mediant, subdominant, dominant, or submediant should
receive more stress than the tonic tone.
In the rather large proportion of songs in the pentatonic, the Ketu group exhibits
a considerable degree of homogeneity.
Relationship of melodic pitch to percussion instrument pitch
It has been suggested by Waterman on the basis of investigation of the African-
derived Shango music of Trinidad, 29 that there the melodic and percussion instrument
pitches show at least a tenuous interrelationship. A siinilar, and perhaps mon; con-
sistent relationship seems apparent in the Ketu songs.
It was possible to isolate percussion instrument pitches in 26 of the 29 songs,
although it should be pointed out that in the tones of drum and gong, overtones are
both numerous and strong, making the isolation of the fundamental a difficult task at
best, and in many cases an impossible one.
In those songs in which the subjective- tonic and duration tones are identical, a
total of 34 percussion pitches could be determined. Ot these the majority fall upon
the tonic, third, fourth, or fifth of scale in the following proportions: tonic, 9 times;
third, once; fourth, 8 times; fifth, 7 times. In one case the percussion instrument
pitch falls upon the sixth degree of scale, and in another upon the minor third of scale,
but in a minor key. The remaining seven tones are not so easily reconcilable. One
falls upon the flat fifth of scale, and one upon the minor seventh. Two pitches are found
on the major seventh of scale, and three upon the second degree. In the cases of both
major and minor seventh, a leading tone is indicated but it would not se~m highly
plausible that this tone should be purposely stressed. In connection with these latter
seven tones, then, a relationship does not seem to be present. It may be regarded as
significant, however, that of 34 tones, 27 are in close relationship to the tonic triad,
stressing particularly the tonic, subdominant and dominant.
In those songs in which the subjective tonic and duration tones are not identical,
21 percussion pitches could be determined. Of these, three fall upon the tonic of the
subjective tonic scale, two on the major third, and six upon the fifth. One tone is found
on the second degree of scale; in other cases the relationship is not so clear. However,
of a total of 55 percussion pitches, at least 40 show definite relationship with the melodic
pitch; this figure seems too high to allow for a chance relationship. Without field
investigation of the problem, however, it is impossible to reach a definite conclusion,
although the relationship of percussion instrument pitch to melodic pitch seems to be
a distinct p ossibility.
Modulations and tonality
A well-defined tonality is present iri almost all the Ketu songs. This is indicated
by the relationship between the subjective tonic and duration tones discussed above.
No modulations are found in the Ketu songs.
Harmof!J, po{yphof!J and solo-chorus alternation
Sporadic harmony occurs in tWo songs; in other than these instances no harmony
is employed. In song 22B1, the harmony appears as the result of the overlapping of
two interpretations of the chorus melodic line; in this case, it seems to be accidental
rather than purposeful. Much the same case occurs in song 54A3 in which part of the
 SONGS OF THE KETU CULT OF :BAHIA 61

chorus apparently begins a phrase on the wrong note, causing brief sporadic harmony
in fourths. The difference is immediately rectifiedand is not repeated since its occurrence
was due to error rather than design.
All songs are of the leader-chorus type, and overlapping between leader and chorus
occurs slightly in 19 songs. In song 22B1 it has already been noted that the overlapping
occurs between two parts of the chorus and is thus a special case. It is in connection
with these songs that the only polyphony occurs in the Ketu music.
Formal structure
Three formal patterns are used in the structuring of the Ketu songs. Approximately
45% use the AA'A" pattern in which the same phrase is repeated with significant
variation. The same percentage of songs use the AAA pattern, in which a single phrase
is repeated throughout the song without significant variation. The remaining songs
are in the ABC pattern, in 'Yhich different phrases are utilized. In almost all cases,
these phrase patterns may be broken into smaller units, but these units are only significant
as they contribute to the larger groupings. 30
The shortest phrase is 2 bars in 4/4 time, or eight beats duration. The longest
phrase is 16 bars in 3/4 time, or 48 beats. Four songs use two phrase patterns of
differing lengths. In song 63A1, phrases of both 8 and 2 bars are present; in songs
63A2 and 63A3, phrases of four and two bars are present. In song 22B1 the normal
phrase length is 4 bars, but at one point a diminution of this phrase to three-bar length
occurs.
In many of the songs, the leader introduces, or otherwise aids the chorus with its
melodic line. In five songs, the chorus phrase is clearly introduced by the leader; in
ten songs, the two phrases are identical; in four songs, the leader assists the chorus
in the first singing of its melodic phrase; and in two songs the leader's phrase is related
melodically to that of the chorus. The leader does not assist the chorus in any way in
eight songs. In song 54A2, the two lines are reversed after they have been initially
introduced. In this case, the leader sings the opening phrase which is followed by a
phrase sung by the chorus. The repetition of the first phrase is then sung by both
leader and chorus, and the repetition of the second phrase by the leader. From this
point the two phrases have been reversed, and remain in this pattern throughout.
Ornamentation
Three ornamental devices are employed in the Ketu songs; these are the rising
attack, falling release, and portamento.
A rising attack occurs when the singer begins his intonation below the desired
tone, and rises to it in a short glissando. This technique is employed in 18 songs. The
falling release is, for practical purposes, the reverse of the rising attack; in this case
the singer sings a tone but lets it fall downward in a short glissando. All but one of the
songs use portamento in which the singer moves from one tone to the next without
making the differences sharply defined; in other words, all pitch gradations between
the two tones are employed in a glissando from one to the other. These ornamental
devices constitute important singing techniques in the Ketu songs.
Instrumentation
. Eleven songs are sung by a male soloist, the remaining 18 songs by a female soloist.
In all 29 songs, the leader is accompanied by a .female chorus. However, since the sex
of the singers usually depended upon the circumstances at the time of recording, undue
importance should probably not be attached to these figures.
Further accompaniment is provided by a percussion group in 28 songs. In two
songs the rhythmic accompaniment is provided by hand-clapping, and in two songs
by two drums. In the remaining 24 songs, the percussion unit is composed of iron
gong and two drums.
62 AFRICAN MUSIC SOCmTY JOURNAL

Tempo
A fairly restricted range of tempi within the various time signatures is usual in
the Ketu songs. In 4/4 time, tempi range from 128 oeats per minute to 188. In 6/4
time the variation is from 96 to 208 beats per minute; this is a considerable range. In
3/4 time, the slowest tempo is 224 beats per minute, and the fastest is 252 beats. In
the single 6/8 song, the dotted quarter notes move at 96 beats per minute, while the
tempo in the 12/ 8 song is 120 beats. -
Seventeen ofthe Ketu songs accelerate throughout. The largest increase is 56 beats
per minute from beginning to end; the smallest increase is 4 beats per minute. This
acceleration of tempo may be considered characteristic of the Ketu songs.
Percussion rhythm
The importance of the percussion instruments for the overall impact -of the music
of the Ketu group is stressed by the fact that 28 of the 29 songs employ a percussion
unit as accompaniment to the singers. Further, the songs themselves seem in great
measure to be dependent upon this unit to establish the basic metre and to continue it
with variation throughout.
In an extended treatment of the subject of percussion polyrhythm in connection
with the Shango cult music of Trinidad-a style historically related to that of the Bahian
music-Waterman has concluded that "music in this tradition makes use of patterns of
combinations of duple with triple time which include simultaneous and coterminous
duple and triple measures, duple accent applied to triple metre, and triple accent applied
to duple metre." 31 As a result of this duple-triple superimposition, the listener receives
an overall impression of a rhythmic gestalt formed from the resultant beats in what
Waterman terms the "8/8 ostinato." This pattern, although composed of both eighth
and sixteenth notes is reduced to a series of eight eighth notes in the 4/4 resultant
measure, within which accenting may vary. The pattern is as follows:

et JnJ JnJ Jn1J uAJ JnJ J1J0J JJJJ unJ 1

Furthermore, on the basis of superimposition of more than one metre, a 12/4 ostinato
pattern may be established within which the pulses are divided into four groups of
three notes each. Using these principles as a basic for discussion, we may turn tb the
percussion rhythms of the Ketu .group.
Song 20B2 illustrates some of the basic rhythms which are encountered through-
out the sampling. In this case, the melodic line is first established in duple meter,
which may be notated in 4/4 time. The percussion unit enters in the second measure
ot the song, the gong immediately establishing a triple meter in the first half of the
measure. This pattern is as follows : (Example 1), with the established duple meter of
the melodic line the gestalt thus becomes: (Example 2), etc. If this triple gong pattern
were continued, the 8/8 ostinato would then result: (Example 3). In this case, however,
the gong does not continue the triple patterning but moves into a duple pattern: (Example
4). Thus the resultant pattern of gong and melodic rhythm is as follows: (Example 5).
At the same time, however, that the gong is using a split triple and duple pattern, Drum!
has entered with a 12/8 beat which, since the basic meter is duple 4/4 time, must be
notated as four groups of three pulses: (Example 6). The resultant rhythm of gong
and Drum I then appears as (Example 7). In addition to these two percussion instru-
ments, however, Drum II appears initially in a duple meter: (Example 8). The final
resultant of the three percussion instruments, then, become a 12/8 or 12/7 ostinato
pattern as follows: (Example 9), although this pattern is altered throughout by a change
of rhythmic groupings effected by Drum II.
 SONGS OF THE KETU CULT OF BAHIA 63

j nJ J n J J nJ J n
Example 1 Example 2 Example 3

> "" ·
J t:IJ7JJJJ
Example4 Example 5

,.-!--, ...-!--, ~ ~
mmmm
Example 6 Example 7

> .,. 7 ,.

J hJ J t:IJ J hJ J'JJ J
Example 8 Example 9

The duple-triple superimposition is not the only percussion technique employed

in the Ketu songs. In those songs in which polymetre is not employed, the basic metric
scheme may be, and most often is elaborated extensively without changing the metre.
In song 63A1, for example, the following pattern appears:

Gong
Drum I

Drum II

In song 64A3, the tollowing patterns are used, in which the basic triple metre is
embellished particularly by Drum I.

Gong
Drum I

Drum II
64 . AFRICAN MUSIC SOCIETY • JOURNAL

Some other characteristics of the percussion rhythms, and the apparent functions
of the various instruments may be pointed out. Waterman notes that in the Trinidad
Shango percussion units "one drum-rhythm is a steady quarternote beat in 4/4 time."32
This drum, by and large, provides the orientation about which the other instruments
of the percussive unit, as well as the melodic line, are grouped. In the Ketu songs, this
stabilizing factor is instead provided by the iron gong which usually is the first per-
cussion instrument to enter, and which seldom changes its established metre in the
course of a song. Drum I usually enters either with the gong or a measure or two
later. Most often remaining within the 8/8 ostinato pattern, it may sometimes also
play the same rhythmic patterns as does the gong, and occasionally utilize some poly-
metre. Drum II, on the other hand, seems to provide most of the polythythmic patterns,
and at the same timcc may .also remain within a single metre elaborating the basic beat.
In almost all the songs, Drum II seems to be much freer in its choice of patterns than
either the gong or Drum I, varying its beat extensively.
It is clear that the songs of the Ketu group sampled here rely heavily upon the
percussion instruments to provide a steady beat about which the melodic lines may
be oriented. At the same time, the use of polymetre and the embellishment of rhythmic
patterns within one metre are carried to an exceedingly complex degree. The reliance
upon percussion and its importance, the use of polymetre, and the embellishment of
basic rhythmic patterns may all be considered outstanding characteristics of the Ketu
songs.
Summary
The char.acte~istics ot th~ so?gs ?f t~e Ketu group may be summariz~d as follow~.
Tonal range 1s w1de; melodic direction 1s generally downward. The maJor second 1s
utilized most frequently, followed by the minor third, perfect fourth, and major third.
Half step intervals play a minor part in the music; the repetitive monotone is likewise
not highly important. Combinations ot thirds and combinations of fourths are the
outstanding interval patterns. 6/4,4/4, 3/4, 12/8, and 6/8 time are used, in that order
of importance. Syncopation and offbeat patterns of wider extension are very noticeable.
Almost 70% of the songs are based on the pentatonic scale; also represented are the
hexatonic and tetratonic. In most cases, the subjective tonic and duration tones are
identical. The tones of the percussion instruments fall on the tonic, dominant, sub-
dominant and mediant of scale in most songs; this indicates the probability of a definit(l
relationship between melodic and percussion instrument pitch. All the Ketu songs
seem to have a well-defined tonality; there are no modulations. Sporadic harmony
occurs in two of the songs, overlapping in 19. All songs are of the leader-chorus type.
The AAA, AA'An, and ABC formal structures are utilized; the leader introduces or
otherwise aids the chorus in singing its melodic line in 21 of the 29 songs. Rising attack,
falling release, and portamento are used as ornamental devices. Songs are sung by a
male or female soloist, accompanied by a female chorus and a percussion group com-
posed of iron gong, drums, and in some songs, handclapping. Tempi are fairly restricted
except in those songs in 6/4 time in which a wide range of variation is noticeable.
Acceleration of tempo is common. Percussion rhythms are of extreme importance in
the songs, and both polymetre and the extensive elaboration of rhythmic patterns
within one meter are employed.

III
While it is by no means within the intention of the present paper to analyse the
relationship of Ketu song to the music of the Y aruba from which it is derived, it should
be pointed out that, within our present knowledge of the music of West Mrica and
specifically the Guinea Coast, the relationship seems incontrovertible. The difficulty
of making such comparison, however, is manifest in the fact that so far as is known to
the present author there does not exist any published analytical study of the structure
 SONGS OF · THE KETU CULT OF BAHIA 65

of Yoruba music; without such study there is, of course, no basic means of comparing
the .original musical style-Yoruba-and the derivation from it-Ketu. The importance
of such comparison for the study of cultural dynamics needs no further comment here;
the anthropological "laboratory situation"· noted in the opening paragraphs of this
paper provides an almost unparalleled opportunity for the study of culture change and
its various ·manifestations.
At the same tiine some general comparison can be made on the basis of the five
outstanding characteristics of the music of West Mrica populated by Waterman. 33
These five characteristics include the metronome sense, dominance of percussion, the
use of polymetre, the off-beat phrasing of melodic accents, and the overlapping call-
and-response pattern. Without entering into detail, it may be noted that the Ketu
songs show all these musical patterns and, in most cases, so strongly as to leave no doubt
whatsoever of the relationship between Ketu and West African styles. It is unfortunate
that the sharper comparison between Ketu and Y aruba cannot be undertaken at the
present time because of the limitations of our knowledge.

FOOTNOTES FOR KETU CULT OF BAHIA

1For the results of such comparison, see Alan P. Merriam, with the assistance of Sara Whinery and
B. G. Fred, "Songs of a Rada Community in Trinidad," forthcoming in Anthropos.
2Arthur Ramos, The Negro in Brazil, (Trans: Richard Pattee), (Washington: Associated Publishers,
1939), p. 10. .
3Gilberto Freyre, Tbe lt1asters and tbe Slaves, (Trans: Samuel Putnam), (New York: Alfred A.
Knopf, 1946), p. 311, says: "The proximity of Bahia and Pernambuco to the Mrican coast tended to
give to the relations between Brazil and the Dark Continent ·. an especially intimate character." Later
(p. 318) he continues: "It is, moreover, a curious thing to note that down to the end of the nineteenth
centuq the repatriation of Hausa and Nago freedmen from Bahia to Mrica took place, and it was such
freedmen-repatriates who founded in Ardra a city by the name of Porto Seguro. So intimate did the
relations between Bahia and these cities come to be. that heads of commercial houses in Salvador received
honorary distinctions from the government of Dahomey." See also: MelviUe and Frances Herskovits,
"The Negroes of Brazil," Yale Review XXXII (Winter, 1943), pp. 266-67.
•Ramos, op. cit., pp. 12-13. In another source, Ramos discusses the membership in these three
groups somewhat more fully. He says: "0 primeiro grupo foi introduzido inicialmente nos mercados
de escravos da Bahia ... Desses negros sudanezes, os mais importantes foram os 'yorubas' ou 'nagos' e
os 'geges' ('Ewes' ou 'dahomeyanos') e em segundo logar, os 'minas' ('Tshis' e 'Gas'), os 'hausas' os
'tapas', os 'bornus' e os 'gruncis' ou 'gallinbas'. Com esses negros sudanezes entrarem does povos
de origem berbere-ethiopica e influencia mohametana: os 'fulahs' .e os 'mandes' . Os 'bantus' foram
introduzidos em Pernambuco ... , Rio de Janeiro ... e Maranhao ... , focos primitivos de onde se
intradiaram posteriomente para varios pontos do territorio brasileiro. 'Bantus' foram os 'angolas', os
'congos' ou 'cabindas', os 'benguellas', os negros de Mocambique ..." Arthur Ramos, 0 Negro Brasileiro,
(Sao Paulo: Companhia Editora Nacional, 2nd E dition, 1940), I, p. 23. It may be noted in passing only
that Johnston says: " . . . the Portuguese seem even to have introduced Hottentots and Bushmen,
obtained probably from the region south of the Kunene River." Sir Harry H. Johnston, Tbe Nogro
in the New World, (New York: Macmillan, 1910), p. 83.
•Ramos, Tbe N egro ... , op. cit., p. 12.
6Donald Pierson, Negroes in Brazil, (Chicago: University of Chicago Press, 1942), p. 282.

7For syncretisms of this, and other nature, see MelviUe J. Herskovits, The Myth of the Negro Past,

(New York: Harper & Bros., 1941", pp. 213-35, passim. ·

8 Yoruba religious practices and beliefs are discussed by William R. Bascom, The Sociological Role
of th~ Yoruba Cult-Group, (Menasha: American Anthropological Association, Memoir 63, 1944), 75 pp.
See also: Edison Carneiro, Religioes Negras, (Rio de Janeiro: Civilizacao Brasileira, 1936), pp. 29-49;
and Arthur Ramos, Introduciio a Antropologia Brasileira, (Rio de Janeiro: Colecio Estudos Brasileiros,
1943), pp. 352-77.
•See Ramos, ibid., pp. 433-78.
1 0See Carneiro, op. cit., pp. 29-49; Ramos, lntroduciio . . . , op. cit., pp. 378-96. For Dahomean
religious life, see Melville J. Herskovits, Dahom~: An Ancient West African Kingdom, (New York:
J. J. Augustin, 1938), II, pp. 101-308. See also, Herskovits, "The Negroes ... ," op. cit., pp. 267-68.
66 AFRICAN MUSIC SOCffiTY JOURNAL

11 See Melville J. Herskovits, and Frances S. Herskovits, Afro-Bahian Religious Songs, (Album Notes,
"Folk Music of Brazil," Library of Congress, Album XIII), p. 2. A discussion of "The Coastal and
Amazonian Tupi," including the Guarani and Tupinamba, may be found in Julian H. Steward (Ed),
Handbook of South American Indians, (Washington: Bulletin 143, Bureau of American Ethnology, 1948),
ill, pp. 57-348. See also: Carneiro, op. cit., pp. 85-100; and Ramos, lntroducao .•. , op. cit., pp. 67-137.
12Herskovits, Afro-Bahian Religious Songs, op. cit., p. 9.
13The treatment of possession is discussed by Ramos, 0. Negro . , . , op. cit., pp. 249-84.
14Herskovits, Afro-Bahian Religious Songs, op. cit., p. 15. See also, Herskovits, "The Negro ..• , "

op. cit., pp. 271-79.

15 See those works cited above. See also, for general accounts: Manuel Querino, Costumes Africauos

no Brazil, (Rio de Janeiro: Civilizacio Brasileira, 1938), pp. 19-127; Ramos, 0 Negro ... , op. cit., pp.
35-247. A comparison with the cult groups of Recife may be obtained from Rene Ribeiro, Cultos Afro-
brasileiros do Recife: Um Estudo de Ajustamento Social, (Recife: Boletim do Institute Joaquim Nabuco,
Numero Especial, 1952), 150 pp.; with those of Maranhao, from Octavia da Costa Eduardo, The Negro
in Northern Brazil, (New York: J. J. Augustin, 1948), 131 pp.
16Melville J. Herskovits, "Drums and Drummers in Mro-Brazilian Cult Life," Musical Quarterly,

XXX (1944), pp. 480-81.

17 Melville J. Herskovits, Personal communication.
18Much detail on the construction and playing of drums, as well as the social position held by
drummers, is found in Herskovits, "Drums and Drummers ... ," op. cit., pp. 477-92, from which the
preceding material has been taken.
19A very early reference to the gong in Angola may be found in P. Cavazzi, Relation Historique de

I'Ethiopie Occidentale, (Trans: J. B. Labat), (Paris: Olarles-Jean-Baptiste-Delespine, 1732), II, p. 49.

"La Longa, est un instrument compose de deux petites cloches de fer, qui ressemblent a peu pres aux
sonnailles que l'on met au col des mulets ou des bestiaux dont on eleve des troupeaux. On les touche
avec un petit baton." On pp. 46 and 50 of this work, woodcuts illustrate the gong and other musical
instruments.
2 °Further description of the musical instruments used in Afro-Bahian cult music may be found in:
Nina Rodrigues, Os Africanos no Brasil, Revisao e Prefacio de Romero Pires), (Sao Paulo: Companhia
Editora Nacional, 1932), pp. 239-41; Querino, op. cit., pp. 105-90; Ramos, 0 Negro ... , op. cit., pp.
223-47; Oneyda Alvarenga, "A lnfluencia Negra na Musica Brasileira," Boletin Latino Americana de
Musica, VI (1946), 357-407, and many others.
21Erich M. von Hornbostel, "Die Probleme der Vergleichende Musikwissenschaft," Zeitschrift der
lnternationale Musikgesellschajt, VII (1905), 85-97.
22M. Kolinski, "Suriname Music," in Melville J. and Frances S. Herskovits, Suriname Folk-Lore,

(New York: Columbia University Press, 1936), pp. 491-740. ·

23
Richard A. Waterman, African Patterns in Trinidad Negro Music, (Unpub. diss., Northwestern
University, 1943).
2'Alan P. Merriam, Songs of the Afro-Bahian Cults: An Ethnomusicological Analysis, (Unpub. diss.,
Northwestern University, 1951); "The Bashi Mulizi and Its Music; An End-Blown Flute From the
Belgian Congo," (forthcoming in journal of American Folklore); Merriam and Linton C. Freeman, "Statis-
·tical Oassification in Anthropology; An Application to Ethnomusicology," American Anthropolugist,
LVIII (June, 1956), 464-72; Merriam, Whinery, and Fred, op. cit.
25
Abbreviations used: M-major; m-minor; dim-diminished; r-rising; £-falling. Explana-
tion of the various types of patterns may be found in Kolinski, op. cit., pp. 498-501.
26 Numbers used refer to the numbers of the original recordings.

· 27 Explanation of this particular type of modal analysis will be found in Kolinski, op. cit., pp. 491-93.
The figures used later in the text in relation to modal structure indicate, in the order listed, the number
of tones actually utilized, the number of tones encompassed in the cycle of fifths, and the degree of scale
on which the duration tone falls.
28Duration tone refers to that tone which is most heavily weighted throughout the song as defined

by actual duration count. The subjective tonic designates the subjective criterion of a key feeling, or
tonality, as observed by Western musicians in respect to Western musical traditions.
29Waterman, op. cit., pp. 104~06.

3 0The designations AA'A", AAA, and ABC, indicate the form used, but not necessarily the extent

of that usage. Each is to be considered as capable of infinite extension.

81 Waterman, op. cit., p. 103.

32lbid., p. 90.

83Richard A. Waterman. "Mrican Influence on the Music of the Americas," in Sol Tax (Ed),
Acculturation in the Americas, (Chicago: University of Chicago Press, II, Proceedings of the 29th Inter~
national Congress of Americanists, 1952), pp. 207~18.
 SONGS OF THE KETU CULT OF BAHIA 67

Note on the transcriptions and modal ana!Jsis:

Each song is headed with pertinent information in abbreviated form. Thus the
legend ±or the first song, "22B1: Yemanja: 2d: 140-168," indicates that the song is number
22B1 of the collection, that it is sung tor the god Yemanja, that the original pitch was
2 semitones lower (2u would indicate the original pitch to be 2 semitones higher),
:~nd that the tempo ranged from 140 beats per minute at the beginning of the song to
168 at its end. The basis tor the tempo figure is the quarter note; the songs are notated
in that key in which the majority of notes falls within the limits ot the regular staff lines.
The fermata (.) over a tone in the modal scale indicates that this is the final tone of
the song. A reversed fermata U indicates the beginning tone. The asterisk designates
the subjective tonic, while the whole tone indicates the duration tone. The sign r----1 in
the modal scale indicates important intervals used in the song; the 1> I show the
direction ot these melodic intervals.