A Socio-Historical and Contextual Analysis of Popular Musical Performance Among the Swahili of Mombasa, Kenya

Introduction
Taarab is the Swahili equivalent of the Arabic word tarab, which implies the concept(s) of entertainment, enchantment, emotion-filled movement, and delight (Anthony 1983; Askew 1992; Topp 1992). As used among the Swahili, taarab denotes the performance and singing of mashairi (poems) with instrumental accompaniment (Campbell 1983; Knappert 1979) and also carries the connotations of entertainment and expression of emotions. The Swahili are an African people of mixed descent living along the East African coast as well as in the interior. They are mainly Muslim and lead an urban lifestyle characterized by a mercantile economy. With a language that has become world-known, the Swahili have an elaborate cultural practice that draws from Arab, African, Indian, and European cultures. Taarab music is indeed a reflection of this complexity with its characteristics reflecting influences from Arabia, Africa, India, Europe, and the Americas. However, when one speaks of Swahili music, there is no doubt that taarab is a major part of that music.

Although taarab is often performed at Swahili weddings, the performance itself has very little to do with the bride and groom at the specific wedding. Rather, taarab performance is a social space in which local values, concerns, and relations are mobilized, discussed, evaluated, and reconfigured. Thus, apart from a song or two advising the groom and bride on how to live harmoniously in marriage, most taarab songs performed at a wedding will touch on different topics of life relevant to social contexts outside of the wedding. Some songs will touch on the concerns of humans in general, while other songs will touch on specific social issues of the local community. For instance, in one performance, one can hear a song about the human quest for control over the earth, a song about the beauty of Swahili women, a song about a changed political economy, or a song about the pain of losing a loved one. Whatever theme each taarab song represents, however, no performance is identical to another. At each performance, a song's meaning expands or changes depending on the images it expresses or the inferences the audience members may make from it. Given the use of metaphor and other symbolic tropes in song texts, a single taarab song may elicit numerous meanings and interpretations.

Taarab is unique among Swahili musical genres in that it is the only genre in which men and women now perform music together in public. In the words of Sheikh Ahmed Nabhany,¹ "these changes [of men and women performing together] come due to contexts. You will find women teaming up with men to sing but this is not Swahili culture or tradition. Indeed, it is against the teachings of Islam. These performers know it but still ignore it."² Sitara Bute, one of the prominent women taarab musicians in Mombasa, shared the same sentiments raised by Nabhany about Islam and gendered practices but saw the practice as part of a process of development. Thus in answer to the same question that I had posed to Nabhany on women and men performing together, she stated, "When we look at women's participation in singing, we find that if we strictly follow customs or religious practice, a woman is not allowed to sing. But now due to certain advancements or changes in social relations, we can now sing. In the past men did not mix with women publicly as they do nowadays. These days we see that just as the man has a job in the office so does the woman. So you see there are changes and I expect that they will continue in the days to come."³

Many Swahili people in the community have noted the social changes that Sitara mentions. In June 1997, I talked to one of the caretakers at the Muslim Women's Institute at the Islamic Center in Mombasa about the use of their large hall for social events. She mentioned that the hall is often hired out to individual families who use it for wedding receptions instead of holding them in the makeshift sheds made of canvas that are quite common in Mombasa. One of the reasons that she gave for many families choosing the hall (besides the fact that they could afford it) was that it enabled them to monitor who would come into the function. When asked if there were taarab performances that occurred in the hall, Sitara said that they do not allow taarab bands to come into the hall because they include both men and women performing together, which was contrary to what the center was promoting. Instead, the Institute may invite a female taarab musician who will lip-synch to her recorded music played on a cassette player. This is one way of trying to deal with the now common practice at Swahili performances in which men and women participate.

For Zuhura Swaleh, another prominent taarab musician based in Mombasa, these social changes are directly linked to taarab music, since she feels that "taarab lyrics give advice and encouragement to do things. They encourage [women] to believe that they can do that which another person can do."⁴ To see this music genre in these terms is to associate expressive culture with the social change that competes with the established tradition of expecting gender separation at Swahili weddings. Throughout my fieldwork I gathered information from men and women who, although agreeing that taarab performance was going against the expected gender practice, saw it as indicative of the changes the society was undergoing. To many, these changes were unstoppable, although some felt that there was a need for things to go back to the "good old days." I thus saw a dichotomy between culture as stated and culture as practiced. In this paper I try to foreground the importance of taarab not only as a forum wherein social practices are crafted, represented, and challenged but also as a way of interrogating the social processes of using music as an arena to understand a community's cultural and historical realities.

The Structure and Organization of Taarab
Performance of poetry among the Swahili has a long history dating as far back as the sixteenth century when written historical records for Swahili literary practices first appear (Allen 1981). That poetry is an important component of Swahili cultural identity and practices has been well documented (e.g., Allen 1981; Anthony 1983; Harries 1962; Knappert 1972, 1977; Mulokozi 1982; Mazrui and Shariff 1994; Shariff 1983). Indeed, much of Swahili's expressive culture is centered on this poetry. The Swahili divide their poetry into three categories—the shairi (a poem that has four lines in each verse), the utenzi (a long poem of three or four lines in each verse and mainly composed as an epic) and the wimbo (a three-versed poem composed to be sung).

However, these categories are not entirely distinct, as there are many overlaps between categories; indeed, if we heed the thoughts of earlier Swahili scholars, then we agree that Swahili poetry is composed to be sung (Abedi 1965; Harries 1962), and hence place all Swahili poetry under the wimbo genre. It is only after we include notions of the content of poems, where they are performed, and their length, that we are able to group them into various categories. Nevertheless, taarab technically falls under the third category of wimbo, with three lines in each verse and a fourth one that is usually the refrain (locally referred to as kiitikio, kipokeo, or kibwagizo). The composition of taarab texts adheres to the tradition of rhyme and meter that is followed in other forms of Swahili poetry. Here is an example of a taarab song that follows rhyme and meter that is present in much of Swahili poetry. The first verse of the song, "Singetema" by Zuhura Swaleh is as follows:

Takusema takusema tasema sitonyamaza
Na lawama na lawama waja mnganilemeza
Singetema singetema yamenishinda kumiza

The first line of this verse can be divided into 16 syllables of two hemistiches of 8 syllables each:

1. Ta ku se ma ta ku se ma 2. ta se ma si to nya ma za

Also note that all the end syllables in each hemistich have a rhyming sound (ma in the first and za in the last).

When sung, taarab may fit into the larger realm of African music through its performance and musical style despite the aforementioned influences from other cultures. The singing of many taarab songs does, however, betray its Arabic influence. Thus the majority of taarab singers have what Lomax calls "the bardic style of the orient" (1961, 443). Their voices oscillate between the spectrum of rubato, melisma, and tremulo with a clear emphasis on nasal singing, which is also present in Indian songs.

Overall, the four major qualities cited in definitions of African music can be equally useful in analyzing taarab. These are: a) call and response, b) percussiveness, c) syllabic singing, and d) short musical units that are repeated in small variations (Chernoff 1979; DjeDje and Carter 1989; Merriam 1982). For purposes of consistency let us use the same song by Zuhura Swaleh to exemplify the performed structures of taarab. The song "Singetema" represents the style of taarab that incorporates local rhythmic patterns often used in chakacha and msondo performances:⁵

Takusema takusema tasema sitonyamaza
I will talk about you I can't keep quiet
Na lawama na lawama waja mnganilemeza
Even if people castigate me
Singetema singetema yamenishinda kumiza
I would not have talked, I can't take it any more

Refrain:

Mtu lake ni kusema nami tasema sitonyamaza
A person ought to speak up and thus I won't keep quiet
Singetema singetema kweli yamenishinda kumiza
I wouldn't have spoken, I can't take it any more

Nakusema nakusema nasema japo sitaki
I'll talk about you though I don't want to
Penye wema penye wema wajua hapafitiki
Good deeds can't be concealed
Singetema singetema kumiza hayamiziki
I wouldn't have talked, I can't take it anymore

La kutengwa la kutengwa hutengwa lisofwatika
Bad things have to be shunned
Lingapangwa wenye kupanga huchoka
Even if one tried to mend it they would be tired
Nimeshindwa nimeshindwa kumiza yasomizika
I am unable to take it any more

Nimekoma nimekoma umechoka moyo wangu
I have had it, my heart is tired
Nakusema nakusema ingawa sio dawa yangu
I will talk about you though it's not my style
Nayasema nayasema heri apendayo Mungu
I talk in God's will

When we look at the four key points for analyzing African music mentioned above, we find that this song "Singetema" fits and illustrates them very well. First, the vocal part of the song includes a soloist and a chorus who both exemplify the call and response quality with the soloist being the lead voice as the chorus responds to his/her text. Second, the instrumental accompaniment includes the harmonium, tabla, bongos, an electric guitar, a chapuo drum, and a tambourine, thus forming the major percussive ensemble. Third, when this poem is sung, the singer pronounces the syllables as singular short sounds that cannot be extended as one would in English words. Thus the singer will sing the first word in the first verse as follows: si nge te ma as opposed to siiingeeeehteeehmaaah, which would be possible in English words. Poetically, the song is divided into four stanzas of three lines each and a chorus that is repeated after every stanza. And fourth, the instrumentalists play a basic tune that they repeat over every stanza but with slight variations. All these parts for the song "Singetema" can be illustrated as follows:

Structure of "Singetema" by Zuhura Swaleh

Instrumentation

34

25

25

25

Vocal

33/a

12/b

34/a

12/b

34/a

12/b

34/a

20b

Text

vs. 1

Ref.

vs. 2

Ref.

vs. 3

Ref.

vs. 4

Ref.

Melodic phrase

A

A'

A

A

A'

A

A

A'

A

A

A'

A

Note: Numbers denote time in seconds; a denotes vocal part sung by soloist and b the choral part; vs. denotes a verse in the poem; Ref. denotes the responsorial refrain; A is the melodic phrase; and A' is the second melodic phrase.

Where A is the melody played by the harmonium, which in this song marks the beginning of the song and the beginning of each subsequent stanza, a denotes the vocal part that constitutes the singing of a single stanza by the soloist, and b the responsorial chorus sung after the instrumentation and soloist sections. Throughout the song, there is a drum pattern with a polyrhythmic texture of triples and duples. The first drum pattern, performed on a tabla, plays the triple, while the second, performed on a dumbak, plays the duple. A tambourine completes the rhythm with a triple pattern that also acts as the pulse marker.

In the entire song, the harmonium not only plays the accompanying melody along with the soloist, but also the response that comes after the soloist, just before the chorus joins in the repeated response. This responsorial chorus is sung in unison with the soloist and the other members of the group and is repeated after every stanza. Each stanza in the song carries a meaningful idea. This is divided into lines that are broken into two hemistiches of eight syllables whose divisions are repeated end sounds (e.g., in the first line ma and za are the end syllables of the two hemistiches respectively and are repeated in the whole stanza and the choral part). The choral part is distinct in the number of syllables it carries. Certain speculative conclusions can be made to account for this variation in length between the chorus and the other lines in the stanza. First, the chorus cannot be strictly considered part of the technical composition of the whole poem. The poem falls under the wimbo genre that, as a rule, comprises three lines in a stanza. Therefore, the chorus lines need not conform strictly to the rules governing the other lines. Secondly, since the chorus marks the end of one sung part before the accompanying harmonium plays the basic theme, it is lengthened to prepare for change into the next stanza.

The melodic phrase, which is repeated in the different parts of the song, expresses the key idea of the line and follows a similar pattern of subsequent lines and stanzas. Musically speaking, the first part of the song that is performed by the harmonium A can be represented as follows:

A

1     2       1     2       3     3       3     4

1, 2, 3, and 4 represent distinct melodic sections in this musical part. 4 marks a descent in the melody that ushers in the next part of the song, which is the section sung by the soloist. In this part the melody, which is represented by A', follows a pattern in which the harmonium plays along with the soloist's sung part, repeating the same melody (as that of the soloist) after each line in each verse, except in the last caesura of the second line in which the melody of the first line is repeated. Marking each caesura off with a letter, I represent this second melodic phrase (A') as follows:

Takusema takusema [1] tasema sitonyamaza [1]
Na lawama na lawama [2] waja mnganilemeza [1]
Singetema singetema [1] yamenishinda kumiza [1]

Since the middle line has a low pitch compared to the other two, the assumption here is that the adoption of 1 instead of 2 in the repeated part of the second line is intended to create a smooth blend between this second line and the subsequent third, hence ensuring a harmonious melody throughout the song. Note that the letters of illustration correspond with those issued in illustrating the first melodic phrase performed by the harmonium. Both these melodies are similar to each other.

One general characteristic of taarab songs is that their melodic structure is dictated by the poetic construction of the text as well as by the style of taarab that the musician wishes to follow. In the example above, the chorus or refrain is responsorial and is performed in unison by the soloist, other back-up singers, and the harmonium, and it is repeated after each stanza. The song begins with a drum pattern, followed by the harmonium that plays the choral melody, which is then followed by the soloist. Some structural differences may be noted between different bands. Taarab bands that make claims to an Arab musical identity will emphasize stringed instruments such as the oud and the violin, over percussive instruments such as drums and tambourines. Bands that tend more towards local African musical styles emphasize percussive instruments over string instruments.

Despite these specific differences, most, if not all of the taarab songs that I have listened to include an electric keyboard or harmonium, a set of drums, and a tambourine as the basic instrumental accompaniment. The keyboard plays a major role in not only providing the song's melody but also in acting as an accompanying "voice" by repeating the soloist's sung part or the refrain. Further, the keyboard, playing this role of another "voice," often indicates the end of one song and the introduction of another, especially at live performances where a single song may incorporate parts of more than three songs. One notable characteristic of chakacha music is the repetition of short phrases which are sung in a call and response pattern where the leader sings a single line to which the audience responds with another line or two. Whenever a taarab band chooses to incorporate chakacha melodies in its performance repertoire, the keyboard acts as the leader by "singing" the first line of a well-known song to which the audience members respond with the appropriate lines. Indeed, according to Swahili oral history on the development of taarab music in Kenya, foreign instruments were adopted into the local music ensembles due to these instruments' ability to "sing" with the musicians.⁶

Taarab Music in Kenya
In Kenya, the popularity of taarab is widespread both in urban centers, where the sale of audiotapes thrives (Campbell 1974, 38), and in rural areas, where the music is performed at weddings and other social functions. Furthermore, there are weekly programs devoted to taarab music on Kenyan radio and television. Weddings, however, remain the most prominent arena for the performance of taarab in Kenya.

For a long time now, taarab has seemingly remained a musical style associated with Swahili people and has often shown little following among communities other than the Swahili.⁷ In 1993, Malika Mohammed's taarab performance for three consecutive sessions in fully packed auditoriums to Nairobi residents suggested that this seeming lack of a following in other Kenyan communities (outside of the Swahili area) may be attributed to the poor distribution and marketing of the music rather than its intrinsic value and popular appeal. Mombasa Swahili taarab groups mostly record and distribute their music through Mbwana Radio Service in Old Town, Mombasa. This distribution network limits their market to Mombasa residents and others who know about the Mbwana Radio Service. The limited distribution network is evident in Kenya today given the fact that there are no taarab records in the Kenya Broadcasting Corporation's music library in Nairobi that are dated beyond the mid-1980s, when major record labels stopped cutting Swahili songs on vinyl records (except for Malika, who ventured outside of Mbwana Radio Service and recorded with a popular music band called Them Mushrooms in Nairobi in 1993). The recording of music on audiotapes rather than vinyl records is important because Kenyan radio stations, like many other radio stations in East Africa, only play music recorded on vinyl records or compact discs, not on audiotapes. Therefore, music that is recorded on audiotapes is never available to its listeners via radio, leading to a decline in the playing of taarab music (especially new releases) on radio.

Besides the marketing problem that makes it unavailable to larger audiences, much of taarab's text is composed in Kiswahili that draws from archaic vocabulary and often employs metaphors to disguise and hide meanings (Askew 1997, 7). For one to clearly understand the meanings of these texts, one needs to be well acquainted not only with the language but also with the basic modalities for deciphering obscure underlying meanings. Some of my friends and colleagues in Kenya were curious to know how I was able to understand the texts in taarab songs, which they often found inaccessible. However, not all taarab songs elicit such complexity in meaning and thus some songs have become national hits in Kenya given their easily understandable message and good marketing strategies.

Just as taarab music seems to have been neglected by music promoters in Kenya as part of the national repertoire of local music, it is only recently that this music has caught the interest of researchers. When I started working on popular music in my graduate work at Kenyatta University in Nairobi in the late 1980s I had very little to fall back on in terms of literature on popular music. Indeed, I had to try and fit popular music research into a literary course of study, something that had not been done before in the department of Kiswahili and African Languages in which I was enrolled. Furthermore, research carried out in the music departments in the public universities in Kenya was mostly centered on "traditional" music.⁸ As we shall see in the next section, this neglect of popular music was a reflection of the development of music research in much of Western scholarship, which in turn influenced scholars in Kenya, who had received that same Western training.

Taarab within a Context of Swahili Music Ethnography
In the past decade, ethnomusicological research among the Swahili of the east African coast has progressed from an analysis of musical practice as a distinct social phenomenon (Campbell 1983) and as a historical performance (Franken 1986) to a process of localization through women's participation (Topp 1992), as a language of communication (King'ei 1992), as political history (Fair 1996, 1997), and as a producer of national culture (Askew 1997). This progress has often reflected advances in ethnomusicology, anthropology, and other disciplines, which increasingly utilize performance as an important analytical device. The absence of an analysis of taarab music in the works of both Campbell and Franken reflects the primary concerns of ethnomusicology and anthropology of music in the late 1970s and early 1980s when both these authors were in graduate school. At that time, much ethnomusicological research was geared towards "traditional" and "folk" music rather than popular music wherein taarab fits.

Yet Swahili musical practices, like Swahili cultural identity, are a reflection of the complexity that surrounds their social and cultural activities and the meanings they carry. Looking for "traditional" music genres leads to two challenges. The first lies in the attempt to retrieve "traditional" music among a community whose multicultural identity would not elicit a musical tradition devoid of influence from other cultures.⁹ In such a situation one would have to depend on informants' memory or would have to study performances that do not reflect that traditional nature of music in its strict sense. Secondly, an attempt to categorize musical practice into independent entities is an exercise that is futile in a multicultural society such as the Swahili. Embodying African, Arabic, Indian, and Western styles of singing and composition, Swahili musical styles today cannot fit into bounded categories of definition. Rather, they reflect the contextuality, fluidity, and transformations that have become the defining factor of Swahili cultural activities and processes.

I discuss taarab's musical performance here bearing in mind that it has borrowed heavily from other local musical genres, as has been well discussed by other scholars (e.g., Campbell 1976, 1983; and Franken 1986). In its performance, taarab incorporates styles associated with other musical genres of the Swahili such as vugo, msondo, and chakacha. Vugo is a musical genre that is centered around the procession that moves from a groom's home to the bride's to present the mahari (dowry). The musical performance entails the hitting of buffalo horns (locally known as vugo, hence the title of the genre) with sticks in triple rhythms that compliment the duples played by the accompanying drummers. Short lines of folk songs mixed with excerpts from taarab songs are repeatedly sung in these processions (Franken 1986, 149).

Like vugo, msondo derives its name from a local drum that is used in this genre. Mainly in Mombasa and Pemba both men and women perform it as part of wedding celebrations. Participants will form a semi-circle facing the drummers and leave space between them for dancers. Short musical lines are sung repeatedly as the women in the audience often dance the popular hip gyration dance associated both with this genre and with chakacha. The drummers perform triple rhythms that have a pattern of twelve pulses. Chakacha is second to taarab in popularity within the realm of Swahili music. It is performed during the all-night session that marks the climax of a Swahili wedding. Many Swahili families nowadays invite taarab bands to perform at these all-night sessions instead of having chakacha. In chakacha, the msondo drum is one of the key instruments that is used and is accompanied by other instruments such as the cone-shaped vumi drum, a two-sided drum that resembles an hour-glass called chapuo, a trumpet, and brass plates. Interlocking rhythms of duples and triples constitute the performance of chakacha. Fast beat rhymes with the intense hip gyration dance movements are the notable characteristics of this performance. These genres are rarely performed in isolation, as much of their rhythms are incorporated in taarab. During my field research in Mombasa, many of the musicians with whom I spoke agreed that taarab combined different rhythmic and melodic structures from these Swahili musical genres as well as from other musical traditions from other continents, such as Latin American, Indian, and Arabic styles.

Taarab Music in Historical Perspective
Taarab developed in East Africa out of the local musical styles of people living along the Swahili Coast as well as the musical styles of other people from cultures in Asia, Europe, and America, with whom the Swahili interacted. Even in cases where very little cultural influence was visible between the Swahili and members of these other cultures, their musical styles found their way into taarab. Thus, despite the absence of close cultural contact between Indian populations and the Swahili, for instance, major Indian cultural influences are evident in taarab and in other forms of Swahili popular culture. Indian films, which form a strong basis for the tunes in some taarab songs, are a major attraction in Mombasa. Writing about this phenomenon, Janet Topp rightfully says, "some taarab musicians copy the melodies and rhythms of Indian ghazal-like film songs. This Indian-style taarab is most prevalent in Mombasa, where the two main exponents are Juma Bhalo and Maulidi Juma" (1992, 288).

The disadvantage of such a practice, as some Swahili informants explained, is that the popularity of the songs only lasts as long as the movie does. Indeed, Maulidi Juma was quick to tell me that he was working towards reducing the number of Indian tunes he included in his compositions, because he "needed to maintain [his] own style."¹⁰ Another common practice I noted during my field research in Mombasa is that some of the taarab musicians in Mombasa have taken to copying musical tunes and styles performed by prominent taarab groups in Tanzania. Some of these Mombasa musicians include Sitara Bute and her Diamond Star Group, Rukia Mohammed, and Mohammed Yusuf Tenge. Taarab styles associated with Tanzanian groups usually feature guitar rhythms and dance tunes that are common among Tanzanian dance bands (Askew 1997) rather than the usual organ or keyboard tunes associated with taarab in Mombasa and Zanzibar (Ntarangwi 1995, 1998).

Many scholars who have written on the history of taarab have said that it was first introduced to Zanzibar in 1870 by Sultan Barghash, who had invited a troupe of Egyptian musicians to play at his court (Graham 1992; Saleh 1980; Topp 1992). At the time, Zanzibar was the center of a significant commercial empire, controlled by Omani Arab Sultans, which linked India, the Persian Gulf, Arabia, and East Africa with European and U.S. traders.

The story states that Barghash was so impressed with the Egyptian musicians that he decided to send Mohammed bin Ibrahim, a musician from Zanzibar, to Cairo to learn to play the ganuni (zither), one of the key taarab instruments. After accomplishing his mission, Mohammed returned to Barghash's palace, where he became Barghash's personal poet. Mohammed also taught a few of his friends how to play the ganuni, a skill that they later put to work when they teamed up to start the first taarab group, Nadi Ikhwani Safaa, which was formed in 1905 in Zanzibar (Topp 1992). Possibly, some of the members of the group that performed in the Sultan's palace played with Nadi Ikhwani Safaa when performing at friends' and relatives' places (ibid., 72). From Zanzibar, taarab is said to have spread to other parts of the East African coast, including Mombasa and Lamu. Over a long period of time, taarab music was gradually modeled to fit local musical styles and taste, hence explaining the often notable differences in taarab performance and production among various Swahili communities spread along the East African coast and its environs.

That taarab has often been defined as an importation of Arabic music from Egypt (see, for instance, Saleh 1980 and Khatib 1987) follows from what Askew (1997) calls a "Zanzibar-centric" approach to Swahili culture, which seeks to locate all the "civilized" Swahili cultural phenomena to Zanzibar or its Arab settlers of the seventeenth and eighteenth centuries. In my discussion of taarab music I wish to show that taarab in Zanzibar is one among other styles of Swahili music in East Africa. Each style developed following the musical context of each area. Thus Tanga, Mombasa, Zanzibar, and Dar es Salaam have styles of taarab that reflect the specificity of their local musical contexts. For instance, Mombasa taarab clearly reflects the fast, rich rhythms of local ngoma¹¹ such as chakacha and vugo performed in and around Mombasa. Tanga taarab has a clear emphasis on guitar rhythms, following the beni (brass band) tradition of the area (see Ranger 1975 for a discussion of this music tradition); and Zanzibar taarab has a clear emphasis on Egyptian orchestra-style ensemble with more string instruments than percussion. Granted these regional differences, attributing the "origin" of taarab music to Zanzibar is erroneous. Indeed, taarab as a Swahili musical genre dates further back than mentioned in this historical account. A popular Swahili musical practice associated with the kibangala (a seven-stringed lute) was performed in Lamu for a long time prior to the beginning of taarab music in Zanzibar.¹² The kibangala was used in the performance of a music genre referred to as kinanda that also included drums and string instruments (specifically, two drums and a lute) as part of its ensemble. Although this kind of music was regarded as provocative in the Muslim world, it was very popular among the local Lamu people. A picture taken around 1907 in Lamu, by a British photographer, shows women dancing to kinanda (Graebner 1991, 187) which Nabhany says was usually danced by men in pairs in front of an audience until they were very tired.¹³

One of Zanzibar's earliest musicians and founding member of the first taarab group to perform at the turn of the century (Nadi Ikhwani Safaa), Shaib Abeid (1890-1974) remembered having been taught taarab songs in the Lamu style (Topp 1992, 73). Mohammed Kijumwa, a poet and performer from Lamu, started with performing the kibangala, which he carried with him to Zanzibar when he was invited by the then Sultan of Zanzibar to go to the island and train and lead an orchestra and dancers. A photograph taken in Zanzibar in 1907 shows Kijumwa playing the kibangala (Graebner 1991). Given these stories, it is doubtful that taarab started in Zanzibar; one can only speculate that the earlier historical account was in keeping with a "Zanzibar-centric" account of Swahili musical history. If anything, the term taarab used to designate a musical style does not enter the Swahili vocabulary until the 1930s, through the English-Swahili dictionaries (Graebner 1991, 181). This is the same time that Egyptian musicians such as Mohammed Abdi-Al-Wahhab and Umm Kulthum were quite popular and had influence on taarab. In short, a musical style whose current name only enters into the musical scene in the last sixty years shows the possibility of different musical styles developing in different regions with different names. Thus the term taarab is only a recent reference to musical performances that existed before the late seventeenth century in which some scholars have placed the development of taarab music in East Africa. Kinanda or kibangala are some of the terms previously used to refer to musical performances that are now commonly referred to as taarab.

Taarab music may have been associated with interaction among different cultural groups, but the most influential mark it has created in East Africa is its ability to bring many women to the limelight in popular music. Indeed, most of the successful women musicians in East Africa in this century have been taarab musicians. For a society that not only discourages women from singing in public but also from mixing with men in public, to be successful in taarab music is quite an achievement for women. Much of taarab's history rests on women musicians. Siti binti Saad is one such musician whose involvement and contribution to the shaping of Swahili popular music is a well-known historical fact.

Taarab and Islam in Kenya: Are They Compatible?
One of the major defining factors of Swahili culture is Islam. In the absence of a tradition of adherence to Islam, the Swahili would no doubt be defined quite differently. It is a truism that Islam is a way of life for those who practice. That notwithstanding, we find an interesting divide in perceptions about taarab between conservative Muslims and their liberal counterparts in Mombasa. The former are openly opposed to the performance or consumption of taarab by Muslims, while the latter will condone taarab as long as it does not interfere with their religious practices. In this latter case many of the people I talked to in Mombasa about this tension stated that only when taarab was instrumental in moving people away from their concentration upon Allah, especially during Ramadan and prayer time, was it considered negative. When I asked Amina Fakii, a radio program producer with Kenya Broadcasting Corporation's Swahili Service, to tell me something about taarab, she said she had nothing to say about it. She gave me the analogy of asking a Muslim to talk about alcohol—it would all be negative. She conceded that she dislikes taarab because it is not in keeping with religion. I consider these to be sentiments that represent those of conservative Muslims in Mombasa. Liberal Muslims, on the other hand, often see nothing wrong with taarab music. Again one must look at culture as stated and culture as practiced to understand the tension that may ensue between groups against and for taarab music. These views can best be understood when juxtaposed with the general views expressed about music in Muslim communities. Citing Lois Ibsen al Faruqi's delineation of the hierarchy of performance genres in Muslim societies, Marjorie Franken says the following about music and religion in Muslim societies:

Some forms of [music] are considered halal, that is, legitimate and even beneficial, but there are also forms of music that are haram [polluted] and these are to be avoided. Muslim law, scholarship, and popular consensus all agree that some musical forms are acceptable without question, but below these come forms which are permitted, but not necessarily beneficial. At another level below this type of music is an unnamed category that represents a "gray area," where individuals are allowed to make their own judgements. Finally come the haram categories of music at the very bottom of the hierarchy, where the bad context(s) and associations at this kind of music are positively harmful (Franken 1991, 164).

From this hierarchy we have ten examples of performance genres that I will list here as given by Franken:

a) Qur'anic chant (qira'sh)
b) Religious chants (adhan, tahlil/talbiyyah, takbirat madih, tasbith, and tahmid)
c) Chanted poetry with noble themes (shi'r)
d) Family/celebration music (lullabies, women's songs, wedding songs, etc.)
e) "Occupational" music (caravan chants, shepherd's tunes, work songs, etc.)
f) Military music (tabl khanah)
g) Vocal/instrumental improvisations (layali, avaz, taqasin, istikhbar, etc.)
h) Serious metered songs (dawr, muwashshah, tasnif, etc.) and instrumental music (bashraf, da'irah, sama'i, dulab, etc.)
i) Music related to pre-Islamic or non-Islamic origins
j) Sensuous music associated with unacceptable contexts

From this list of performance genres we can place taarab music in different levels because it does not fit in any one restrictive category. As I have shown, taarab incorporates many styles from multiple cultural contexts and would thus partly be placed in categories d), g), h), i), or j), but to categorize it as either halal or haram depends on the perception of the person involved. This is why there are differing views on the relationship that taarab has with Muslim culture among many Swahili people in Mombasa. For those who hold with conservative Islamic ideology (like Amina Fakii), there is hardly any relationship between taarab and religion. For one, taarab includes the playing of drums (ngoma) and other percussive instruments that are usually associated with evil spirits. Again, some of the lyrics found in taarab are sensuous and hence likely to make people have undesirable thoughts. Thus instead of people concentrating on God, taarab lures them to think of things that are not godly.

Sometimes, however, taarab may serve as a precursor to a religious function, as was the case on January 10, 1997, when Sitara Bute was entertaining Muslims in Majengo-Msaji in Mombasa to usher in the holy month of Ramadan. This performance drew opposition from conservative Muslims. About twenty Muslim youths led by the local Imam, Sheikh Mohammed Idris, disrupted the performance amid shouts of "takbir! takbir!" (God is great). The youths, who had come from the nearby Sakina Mosque after twareh (special prayers before the beginning of Ramadan), had the music stopped, as the Imam grabbed the cordless microphone seeking to know why Muslims were enjoying music during the eve of the month of Ramadan. As the youth group got rowdy, threatening to lynch the participants if they did not stop, Sitara Bute abandoned the performance and ran off-stage. Later the next day, Sitara Bute was quoted as seeking to know the Muslim law that was used as the basis for stopping her musical performance. She said that they were all Muslims and that when the moon is sighted nobody is allowed any form of entertainment after midnight. But their performance was scheduled for the hours between 7 and 11 p.m. and thus was not contravening any law (Sunday Nation, 12 January 1997). According to strict rules stipulated by Islam, however, it does not matter what time the performance was scheduled to take place; the fact is that for Muslims to perform taarab is undesirable. This intolerance of performance or consumption of taarab has led many Muslim leaders to continually campaign against the airing of taarab music on radio and television. Some of the people I talked to about this campaign stated that the erratic airing of taarab music on Kenyan television pointed to the success of this campaign. Others felt that it was due to a lack of interest in Swahili cultural practices that can be discerned in many of the national cultural programs. So far, three radio stations have weekly programs dedicated to taarab music that reach most of the country's population.

A Description of Taarab Performance in Mombasa
As in many other Swahili communities, weddings in Mombasa provide the best stage for the performance of taarab. Due to financial constraints, many Swahili families are unable to have wedding celebrations that last more than two days. In the past, Swahili wedding celebrations would last up to seven days, and on each day there was a form of celebration that involved feeding and entertaining guests at the wedding home. It is thus common to find a taarab band hired to perform both for the afternoon session after the formal marriage ceremony at the mosque as well as at the all-night session (kesha) when the bride is to be shown to the crowds. The choice of a taarab band to be hired for these performances may reflect the host families' musical taste (including their own cultural identity), the fame of the band in the community at the time, or the band's affordability. Many informants I talked to agreed that when a host family hires an expensive taarab band, they are making a public statement about their social status.¹⁴ For example, a family that identifies itself with Arabic culture will have a taarab performance that not only concentrates on Arabic musical styles but also entails listening rather than dancing, especially if the guest list includes both men and women, whereas a family that identifies itself with African culture based on the various groups living in Mombasa will have an audience more given to dancing than listening. Whenever a taarab band is invited to perform at an individual family's home, the band members know in advance the cultural identity that the hiring family articulates, and thus the band is ready to perform music that is appropriate for that particular context.

Afternoon sessions usually last about three hours, while night sessions last about eight hours and start around 10:00 p.m. Each band has a performance calendar in which the bandleader marks the days they are scheduled to perform each month. After practice sessions (usually held at the homes of bandleaders), the band members are reminded where and when to meet for the performance sessions. Transportation to and from the venue where the band is to perform is organized and paid for by the host family or their representative. If the band arranges for its own transportation, the costs incurred will be included in the overall charge made to the host family for the entire performance.

When the band arrives at the venue of the performance the members are welcomed into the home by a person appointed as an usher for the day's celebrations. Usually, the band members, along with other guests and relatives, are served a meal of spiced rice mixed with beef (pilau). Children sit together on a mat spread under a shed made of jute sacks, while men and women sit separately and are served in large trays from which all eat. Young men and women are frequently the servers and older women cook the food. When everybody has eaten, the band members start to position themselves at their designated places at a special spot facing the host family's main house. If it is an all-night performance session the band is stationed adjacent to the bride's seat so that as people dance and listen to the music, they can see the bride clearly.

People, mostly women, trickle in and sit in a semi-circle next to the band with some space in between them and the band. This space is reserved for dancing. Soon the organist starts the day's performance with some testing of the instrument before following it with a short instrumental interlude, usually called bashraf. After the interlude, other members of the band take their positions and the band starts to play one of the songs on schedule. The lead singer, in consultation with the organist/keyboard player, decides what song to sing, although during the performance there are numerous requests made by members of the audience for specific songs. This performance also gives the band an opportunity to introduce some of its new compositions as well as advertise other releases, which are usually available on audiocassettes at Mbwana Radio Service.

Whenever the singer wants to introduce a new song in the existing schedule, s/he notifies the other members by just mentioning the title. Many times the scheduled list of songs to be performed is changed when audience members make special requests for songs. Other times the singer and the instrumentalists will gauge the mood of the crowd and perform certain songs that they deem appropriate for that particular context. At the end of the day, a successful taarab performance is one that has been able to balance the choices of the band with the needs of the audience members. Thus, although the performance is for the celebration of a wedding, very little focus, if any, is projected onto the bride and groom. The taarab band has the mandate to entertain the audience, although at times some musicians may overstep the boundaries of that mandate.¹⁵ Most songs are composed from poems written by people not part of the taarab group. Thus each taarab group has a poet/composer who will write their poems and their names are hardly known to the public.

Compared to recorded songs available on audiocassettes, taarab songs at weddings are usually longer and often the band tends to play more dance rhythms—depending on the audience members' reaction to the music. At many of the taarab performances I observed in Mombasa, whenever a song drew a large crowd of dancers, the musicians would elongate the parts of the songs to which the people were dancing. Such parts would include a 3-5 minute guitar and drum ensemble that is not included in the recorded version of the song. Thus a song at a live performance is usually longer (up to 10 minutes long) than one recorded in the studio (usually about 4 minutes long). There are two major reasons for this difference in song length. First, at the live performance, the taarab band's performance is centered on the musical interaction it has with the audience members; thus, a song has longer instrumental sections that the audience dances to compared to the studio recordings. Second, in the studio, where most Mombasa taarab bands record their music, the instruments and studio equipment are owned and controlled by the studio owner. This allows very little room, if any, for musicians to influence the length of a recording. Furthermore, the musicians only use the recording studio to make recordings and not for rehearsals. Some musicians own instruments that they may use at live performances but not during studio recordings where the studio equipment prevails. As mentioned above, at live performances some taarab songs combine melodies from other genres such as chakacha and msondo. Music from these other genres is usually recorded and sold as distinct, which further shows the difference between live performances and studio recordings.

Live performances also feature songs by other local or international bands, which may be special requests from members of the audience.¹⁶ When such a song is performed, the audience members sing in unison with the band singers. Sometimes, the performance of another band's song may include just portions of the original composition with the rest of the song being on-the-spot innovations to blend with that specific context. Thus as the music tempo picks up, the instrumentalists may insert other local song melodies to the delight of the audience members. Occasionally, a member of the audience will dance towards the band and give money to one of the performers and then dance back to her original spot. This act of tipping is important in any taarab performance for a number of reasons: first, it is one way that the band can make some extra money; second, it shows that the person tipping likes the lyrics sung by the musician s/he tips; and, third, it is a way of sending a message to another person in the audience. I discuss these reasons at length later in the paper.

The women in the audience are elaborately dressed in shiny dresses; some have their heads covered, while others do not. Some are in dresses of similar color or pattern. Others have wrapped themselves in leso/kanga with the words that usually are printed on these cloths conspicuously visible. These cloths are also used in dancing when tied around the hips to enhance the dancer's hip gyrations. Sometimes the dancing turns into a display of one's prowess in the style or mirrors the instructional dance performed by young women of marriageable age in preparation for correct hip movement during coitus. During a hip-gyration dance the performer stands with her feet slightly apart, knees bent, arms bent at the elbows and held up close to the torso with the head pushed back slightly. Then she gyrates her hips whose movement is enhanced by the leso tied around them. She may do this while lowering her body closer to the ground and slowing down the tempo to the delight of the onlookers.

Men hardly dance at these performances, always staying in the periphery and watching as the women dance and get involved in the performance. At some of the performances I attended in Mombasa and the surrounding area I observed a few occasions in which a man would go up to the musicians and tip them while another would be dancing at the periphery. Many of the rhythms used during these taarab performances draw from women's dances and musical genres, which might explain partly the overwhelming majority of women dancers at such functions. Moreover, in Swahili communities in Mombasa the part of wedding celebrations that includes cooking, entertaining, and cleaning up is reserved for women and men often feel odd participating in these performances. Moreover, much of traditional Swahili music is divided along gender lines and thus men and women have songs and performances specific to their gender. Taarab as a popular musical genre, however, does not clearly reflect this dichotomy. Yet due to the association taarab has with that part of a wedding where women are mostly involved, dancing and participation is often seen as entirely dominated by women. Further, none of the dance rhythms in the taarab music performed at many weddings corresponds to songs associated with men, but rather clearly incorporates rhythms found in many women's musical performances. This may thus explain why women form the majority of audience members at taarab performances as well as being the prominent dancers.

After each performance, members of a taarab band count the money that they have accumulated through tips from the audience members and share it out according to each band's procedure. The larger amount paid to the band by the host family for the performance is shared later after the whole amount has been paid to the band. According to some of the musicians I interviewed, the money is shared equally with all the members after the band has paid for any hired equipment and other miscellaneous costs. Others said that the money is shared out according to positions in the band, where the owner or bandleader gets a larger share. I found that this arrangement entirely depends on the kind of relationship the band members have with each other. Some members may have standing fees that they charge for each performance while others share whatever remains after the band's costs of performance have been paid.

Given the monetary benefits that come with such a performance for a taarab band and especially since there are few families who can afford to hire a taarab band, competition for invitations to perform at weddings is part of Mombasa's taarab culture. In many cases band members will accuse rival groups of trying to curtail their chances of getting invited to perform at weddings. Occasionally I heard of cases where some members of taarab bands would state that a member or members of another band had put a spell on them so that they would not perform at weddings. Individual musicians would also claim that a rival musician had been to the mganga (witch doctor) to bewitch the musician so that he/she could lose his/her voice or be unable to play his/her specific instrument. How then do so many bands in Mombasa conduct their business given this precarious context of operation? Let me now turn to the taarab groups in Mombasa and the relationships among them.

The Relationship Among Taarab Groups in Mombasa
One notable quality about taarab groups in Mombasa is that, despite their large number, there are few musical instruments. This small number means that the few instruments there are serve all the Mombasa taarab bands. Some groups, such as Zein Musical Party, Mohammed Khamis Juma Bhalo, and Zuhura & Party, have their own instruments. The rest of the groups usually hire their instruments from Mbwana Radio Service, where they also make the recordings of their music that are sold at the same premises. At times members of these different groups team up and perform together, especially on foreign tours as has been the case in the past few years. During this period, Maulidi (of Maulidi & Party), Sitara (of Diamond Star), Zuhura (of Zuhura & Party) and Mohammed Adio Shigoo (of Maulidi & Party) have teamed up and made tours to Europe in 1992, 1994, and 1996 under the management of taarab promoter and researcher Wenner Graebner of Germany.

This teaming-up does not necessarily reflect an existing healthy working relationship among these groups since there is always a spirit of competition with outright animosity expressed in public through their songs. For instance, Sitara and Maulidi in some of their compositions have engaged in this kind of tension where they attack each other. The following songs, "Wembe wa kutu" by Sitara and "Nataka sema" by Maulidi are such examples (translations by author).

NATAKA SEMA (I WANT TO TALK) by MAULIDI

Refrain:
Today I want to talk
I want to talk
Today I want to talk
I want to talk today, my friends,

I have words to say, listen residents
About a wise person whose needs are unending
S/he is inside a well but begs for a mouthful of water

Surely this world is full of surprises
Do not blame me people I wish to tell you
There are ten people sleeping in one bed

The person is full of greed s/he doesn't know what s/he is doing
That's why s/he is always blaming for lack of the right thing
S/he has abandoned steak for 1/4 kilo of bone

When this happens there is an uproar
And two are running away
The one being chased is behind the chaser ahead

There is a witch who is never reputable
For lack of mercy people are perishing
When they are well the sick complain

WEMBE WA KUTU (BLUNT RAZOR BLADE) by SITARA

Let them talk about me I do not really care
I don't want anything from any one and I don't copy their style
I have chosen a blunt one and that is their blade

Refrain:
A pot sits on three stones I will cook their tongues¹⁷
That blunt one I know their blade
On three stones I see their pot
They mix everything, concocting their instigation
But whatever they wish will never happen

Nasty words is their business
Ridicule and laughter is their character
What's bad to you is good to others

You see them searching in their meetings
They talk and disrespect others
There is one God and ours is not theirs

Maulidi's song (released after Sitara's), talks about Sitara, who was a member of his group until late 1994. He apparently regards her as greedy for leaving his group (which he compares to steak) to form her own group (which he calls 1/4 kilo of bone). Sitara, for her part, says that what Maulidi and his group do is talk about her and that she has a blunt blade for them (meaning they are so low they are not worth a sharp blade to be shaven or cut to size). This competition and the singing of songs that challenge other individuals or groups is in the Swahili tradition of expressive culture, but many taarab bands use these songs to express their rivalry. This rivalry is intense, at times leading to accusations of witchcraft and bad blood. In 1994, I attended a taarab performance in Mwatsafulu in Likoni (the hometown of New Star taarab group), Mombasa, where Maulidi & Party were performing. As the group started playing, there was general consensus among the musicians that their instruments were not producing the kind of music that the group normally performs. This was immediately attributed to witchcraft by the local band, New Star. Individual players and singers may also blame each other for any misfortune befalling their performance. For example, there was a rumor in Mombasa in 1996 that both Sitara Bute and Malkia Rukia, who once performed with Maulidi & Party, put a spell on each other's voices and thus were not singing as well as before.¹⁸ Such allegations indicate the rivalry that exists among taarab groups and musicians.

When it comes to making recordings, many taarab musicians in Mombasa may ask the few well-known instrumentalists available to play the accompanying rhythms for them. This may mean that Mohammed Adio Shigoo, an organist, and Rajab Omar, a rhythm guitarist, may give instrumental accompaniment to Sitara and Diamond Star, Maulidi and Party, and Mohammed Yusuf Tenge. This sharing of instrumentalists may thus account for the consistent use of one particular person's name (especially the singer) to identify the musical group. Apart from New Star, Diamond Star, and Johar Orchestra, the rest of the groups are identified by the names of the main singers and who are usually the bandleaders. The name of the poet who composed the lyrics and those of the instrumentalists are never mentioned on the audiocassettes of their recorded music, which may also account for the common practice of singing songs done by other groups. It is my argument that this individualistic practice of identifying the group with a singer's name is influenced by commercialism, where the audiocassette, as well as the invitation to perform for a social function for a fee, are identified with a single person.

Taarab, Performance, Voice, and Social Commentary
During the performance of taarab music, members of the audience as well as the musicians themselves pay a lot of attention not only to the music and textual message but also to the context of the performance. Therefore, the way the music is performed (be it in a slow or fast tempo, for instance), the implicit interaction between singer and text, the interaction between musicians and audience members, and the interaction between audience members, are crucial in expressing the different meanings and interpretations the songs are given. By highlighting this performative notion of taarab, I extend the analysis of taarab performance beyond textual analysis so as to include that part of the performance that "communicates non-textual messages" (Askew 1997, 13). This is particularly important because at a single taarab performance multiple messages are often communicated through multiple media. As Askew puts it, "meaning [in a taarab performance] is located at the intersection of text, performance, embodied action (e.g., tipping audience members), and the intimacies of local knowledge . . ." (1997, 17).

When a singer performs the text of a taarab song, s/he is not confined to rendering a specific message because the symbolic and metaphorical nature of the text subsequently opens the song up to numerous meanings and interpretations. It is this nature of multiple meanings that makes taarab particularly popular and relevant in numerous contexts because a single song can be used by many people to comment or reflect upon numerous social events. Thus while I agree with Kelly Askew that "texts and performance lose some layers of meaning when heard and viewed in isolation, at a distance from the community that provides performers, context, audience, and communal norms against which behavior is measured . . ." (1997, 18), I extend my argument to note that given the non-specific reference in taarab texts, the meanings given those texts are always emergent and relational. Therefore, it is not necessarily the local knowledge of who composed the song, why it was composed, when it was composed, and what the song is supposed to mean, that constitutes the richness of the meanings of its performance. It is the association between events in any community or household with the interpretations of the meanings carried by the song (independent of previous interpretations) that gives taarab songs their power and meaning. Thus an audience in Nairobi can interpret a taarab song sung and composed in Mombasa to fit that particular Nairobi context with no reference to any meanings associated with the song in Mombasa. I argue that it is the potential of Swahili taarab songs to be "read" and "heard" differently not only at different places (contexts) but also at different time periods, that makes Swahili songs such an important cultural repertoire. Malika's song, "Vidonge," had a large appeal to upcountry Kenyans to the extent that it numbered among the top ten songs in the country for two months in 1993. In my opinion this is despite a lack of knowledge of the song's original context of composition or the specific nuances carried in the lyrics. The audience members at any taarab performance are thus able to continuously use the songs by manipulating them to fit their specific social contexts. For instance, for a song entitled "Chungu" by Zuhura, one can make multiple interpretations of the intended message. Here is an example of the first stanza in which Zuhura says:

To ask is not taboo please tell me
I have exhausted my knowledge I do not know what to do
how does a pot without a crack leak?

The general message relayed in this stanza is that of a situation, a person, or object that is meant to be without fault, complete, and well guarded, but which is actually faulty. This song has been interpreted variably in Mombasa. There are interpretations that equate the pot with a woman who was secluded but who became pregnant out of wedlock; others equate it with a company/firm whose operations were meant to have been top secret yet many people knew about those operations; and others equate it with a highly respected politician who was rumored to have been involved in questionable business transactions. It is this boundless nature of the meanings that can be generated from a single song that makes taarab a crucial tool for social commentary. Even in situations where a song's composition is triggered by real life situations of which the community is aware, that initial meaning competes with other meanings that are constantly generated from the same song. The essence of a taarab song is in what it "says" to a consumer rather than the meanings it retains from its original composition.

Indeed, many taarab songs do not make specific reference to people or events and thus are suitable for these multiple interpretations. In view of this multiplicity of interpretations, a taarab performance is therefore a "dialogic process between and among performers and audience members" (Askew 1997, 17). Through such a process, both men and women are able to reflexively and actively comment on and review their social interactions, expectations, aspirations, and practices (see, for instance, Abelmann 1996; Turino 1995; and Bruner 1984, for a discussion of how history, identity, and textual meanings are contested in everyday cultural interactions).

Besides the interpretation that is accorded the song text, other facets of the performance are crucial to a better understanding of it. For instance, even when the message does not specifically refer to a person or episode, the words in the text may beg for a particular meaning or interpretation, as seen in the following song entitled "Jana Nilala" ("Yesterday I Slept") by Sitara:

Yesterday I slept, dreamt of you, and thought about you,
because your love torments me.
Don't walk provocatively, you torment me
Don't look at me provocatively, you torment me

Many of the people in Mombasa whose opinion I sought regarding the possible interpretations of this song in Mombasa agreed that the song should have been sung by a man because the message is directed to a woman.¹⁹ Indeed, one of my informants mentioned that the song as sung by Sitara, a woman, suggested a lesbian relationship since "normally" a woman would not tell another woman she is walking or looking at her provocatively. This particular interpretation shows that in singing a taarab song there is a close relationship between the singer (voice) and the message of the song. The singer's voice is thus not neutral but situated and gendered.²⁰ The singer's voice represents a male or female perspective, which often directs the reading of the song.

The voice also serves as an identity marker for a taarab group. In Mombasa out of about eleven taarab groups, eight use the name of the lead singer to identify their recorded music and the group's name. Although there is a notable degree of consistency in the vocal style of Swahili taarab singing that emphasizes nasal singing, a singer's voice is expected to reflect his/her gendered voice mode. Thus, for instance, some informants consistently stated that Mohammed Yusuf Tenge was not a good singer because of his use of a high pitched voice that these informants called a "female" voice (sauti ya kike). Such an observation clearly shows a knowledge of certain cultural stipulations of voice types suitable for men and women; whenever one does not follow his or her stipulated type, then one is deemed as deviating from the norm.

On a performance level, voice acts both as a social medium of communication as well as a musical instrument, and plays an important role in constructing the meanings and interpretations of Swahili taarab songs. During a live performance, a taarab song's message is heightened by the way that the singer uses her/his voice, such as in variation of pitch and tone. I noted for instance that while performing her song "Vidonge," Malika used her voice craftily to enhance the meaning of the song. She would pause between syllables and words in the song to emphasize the meaning of the previous or subsequent line. In the third line of the song Malika states, "jamani wivu mwaona mi silali peke yangu" (you are jealous I do not sleep alone). She sang this line in a slightly higher pitch than that used in the rest of the verse. Such a pitch is one that resembles the phrasing of a question in a regular conversation, and was so used in order to emphasize the act of jealousy and to subsequently tell off those involved. She also occasionally repeated certain lines in the song to emphasize their meanings.

The singer often brings out the extra-linguistic meanings of the song by inserting non-linguistic sounds and nuances (cf. Erlmann 1996; Sugarman 1997; Turino 1995; Watermann 1990). The singer's body language during the performance also contributes to the meaning of the song as s/he gestures with the hands, the entire body, or just the face. Thus performance is a communicative process that could not be fully grasped from a textual analysis of the song.

Audience members also contribute to this meaning construction in the performance as they sing along, dance, request particular songs, or tip the performers.²¹ By tipping and dancing, audience members at taarab performances inject and display the various meanings attributed and attributable to each of the songs. Usually, an audience member will stand and either dance for a while and then proceed to tip the musician of her/his choice, or stand and go straight to tip the musician then return to one's spot. The most common practice is the former through which the audience members, especially women, display their clothes, jewelry, and hairstyles. Moreover, when a person stands to dance to a song or to tip a musician, that person is either showing their preference for that song, taking the opportunity to register her/his presence at the performance, or is identifying with the message entailed in the lyrics. In many instances all the above situations intersect.

Identifying with the lyrics of a taarab song and the subsequent relaying of its message to a person present in the audience is what makes taarab performance a rich metaphor for social commentary. Other members of the audience will often "read" this message, as this practice is one well-known way to make use of the common Swahili practice of indirect social confrontation. The singer becomes the voice of the listener stating what the latter wants relayed to another. The listener will make the intended message clear by repeating or singing loudly particular lines in the text or by throwing an accusing glance at the target person. For instance, if person A has a disagreement with person B, a taarab song with a message that corresponds to that disagreement may be used by A to express to B her feelings and opinion on that particular disagreement. If the disagreement were based on an action that could be interpreted as driven by jealousy, a song that castigates jealous people (mahasidi) would be appropriate for the above-mentioned communication. To relay this message expressed in the song, A will stand up, make sure that B sees her (either by turning and looking in B's direction or by drawing attention by dancing), then proceed to tip the musician singing the corresponding lyrics. A may also sing along loudly some particular words in the song that castigate the actions of B. In this way, A is using the musician as a medium/mediator to communicate with B.

There are many songs whose messages directly castigate unbecoming social practices. Age is an important indicator of one's expected social practice among the Swahili, as it is among other communities in Africa. There are social rules that govern one's behavior in respect to one's age. For instance, since age comes with different responsibilities for both men and women in Swahili communities, whenever an individual behaves in a way deemed out of step with his/her age category, s/he is castigated. This is exactly what Zuhura's song entitled "Nyanya" (grandmother) represents; it is a castigation of older women who are struggling to be young. In part, the song states:

You are cutting off your ear lobes what have they done to you?²²
You want youth and it's not possible
You are cheating yourself Hassan's mother

Your days are over yet you don't seem to realize
You force yourself to be young
But you can't bring back your past

Zuhura performed this song at a wedding in Kisauni in February 1997 to a capacity crowd that included many women of all ages and a few men who as usual stood or sat behind the women who were seated in a semi-circle next to the band. On hearing the melody that marked the beginning of the song a group of young women seated together in the middle of the audience ululated and cheered in unison. As Zuhura sang the text of the song, these young women were singing along and waving their hands in the air, making it clear that they were in agreement with the message. Someone from the crowd shouted above the singing, "Afadhali Zuhura uwaeleze, wamezidi!" (Tell them Zuhura, they are too much!). One young woman got up and danced her way to Zuhura and danced around for a while before tipping her and dancing back to her spot amid cheers from her colleagues with whom she had sat. An older woman shouted from one corner, "Hakuna moyo kongwe!" (There is no old heart!) when Zuhura sang the last stanza of the song that states:

Many girls of great worth can't get partners
Forcing them to mess themselves up
Why can't you accept you are old?

As the song came to an end, the young women who were cheering requested for the song to be played again. Before Zuhura could decide whether to play it or another song, the hostess came and whispered something to her. Then Zuhura announced that the song was not aimed at every grandmother but rather to those who were unbecoming (wenye vishindo). It was evident that someone had complained to the hostess about the meaning of the song, and for the hostess to ensure that her wedding party was not marred by controversy she requested Zuhura to clarify the specific person(s) targeted by the song. This episode clearly shows how taarab songs can create immediate forums for social contest on practices and their meanings.

Besides the texts of songs that make reference to a social practice that is seen by some of the musicians and their fans as unbecoming, the melodic structure of a song may point to the nature of the song's meaning as well. Songs that castigate unbecoming social behavior are usually performed in fast beats with more emphasis on the drumming patterns of triples and duples that are played in interlocking patterns. This style is similar to that of "women's taarab" in Zanzibar (Topp 1992; Topp Fargion 1993). Here musical rhythms associated with women's songs and performances are incorporated into taarab accompanied with texts that "use strong and often abusive language with the intention of hurting and degrading the subject in the eyes of the public" (Topp Fargion 1993, 118). In contrast, songs with messages about God or about a person's lover, where sentiments of endearment are expressed, are performed in slow rhythms. Other scholars have shown how music structure and performance has been used to investigate social structure (Feld 1984) as well as to show distinct social identity (Turino 1995). In this way, therefore, we can see that one can draw certain conclusions about a society from an observation not only of the content of its music, but also of the structure of its music. Such an observation cannot, however, be made independent of other factors such as context, social meanings, and the processes of making music in that particular culture.

Conclusion
Ethnomusicologist Veit Erlmann states that "performance with its potential for ambiguous representation and expression, mediates between the transformative action of everyday consciousness and the socio-cultural form" (1996, 28). Swahili taarab music is performance that enables both men and women to express and represent the relationship between socio-cultural form and social practice by using it as an arena where the social is performed in order to analyze and reconstitute the cultural. Each participant at these performances is positioned in such a way that whatever s/he does contributes to the creation of socio-cultural meaning of the relations that exist between performers. Thus, whereas the song lyrics may be referring to some ambiguous event, person, or trait, they are used as diagnostic of larger cultural structures within the society. I have thus shown through this paper that popular music is a useful tool in exploring and analyzing a community's culture and history because music is considered an important window into a culture. Also by using music, I am taking up analytical tools used by the Swahili themselves, and have thus avoided objectifying Swahili culture.