For the sake of a song

Music and Dance Conventions



This chapter provides a framework for understanding the musical and dance conventions of the six wangga repertories presented in full in chapters 4–9. While each composer and performer has his own preferred style or combination of features, variability and innovation is constrained by the ceremonial functions of wangga, including its symbolic significance (discussed in chapter 1) and its role in coordinating group action (especially ritual actions by dancers and others).

The conventions adopted in four of these repertories—Barrtjap (chapter 4), Lambudju (chapter 7), Walakandha wangga (chapter 8) and Ma-yawa wangga (chapter 9—were previously treated in some detail in Songs, dreamings and ghosts (Marett, 2005, p 68), which explored how wangga performances enact broader social themes. In this volume we bring together a broader range of data, including two repertories never previously analysed, (belonging to the Belyuen singers Jimmy Muluk and Billy Mandji, presented in chapters 5 and 6), and in chapter 8 extra data is provided on the early Walakandha wangga songs composed by Stan Mullumbuk (Marett, 2007), and from additional 1986 recordings made available to us by Frances Kofod. Readers interested in acquiring further understandings of the conventions of wangga are referred to Songs, dreamings and ghosts and to Marett’s 2007 paper, ‘Simplifying musical practice in order to enhance local identity: The case of rhythmic modes in the Walakandha wangga (Wadeye, Northern Territory)’ (Marett, 2007).

This book documents the repertories by providing systematic detail on each song, to be read in conjunction with the primary data contained in the corpus of audio recordings. Each of the chapters 4-9 contains an introduction to the relevant repertory, its performers and the recordings, a track-by-track discussion of each recorded performance (including a song structure summary for each track), and a music analysis section summarising and discussing the main musical features of the repertory. The data presented in this chapter draws together these repertory-specific analyses within the broader framework of structural conventions used for the whole wangga corpus. We begin with a summary of the main musical formal conventions, which we explore in some detail through discussion of a single song performance.

Summary of musical forms used in wangga

Overall form and terminology

Wangga song items always begin with an instrumental introduction, which is followed by a number of vocal sections, which in most, but not all, cases are separated from one another by instrumental sections. Song items conclude either with a final instrumental section, or (in the case of Barrtjap alone—see further below), a coda. Wangga song items typically have two or three vocal and instrumental sections, but in some cases, particularly when there is dance, additional vocal sections can be added. At the end of each song item, it is the didjeridu player who finishes first (though the signal to finish comes from the singer, who will normally cue his didjeridu player by raising the clapsticks, or giving him a glance). As soon as the didjeridu player stops, the singer beats out the final terminating patterns and the song item comes to an end. The basic form of a song item is summarised in figure 2.1, based on a performance of Maurice Ngulkur’s song ‘Tjerri’ (chapter 9, track 13), which will be discussed more fully below.

Wangga Image Figure 2.1 Summary of musical and textual structure of ‘Tjerri’ as performed by Ngulkur (chapter 9, track 13), illustrating the terminology used throughout this volume.

All wangga songs conform, in general terms, to this format of alternating vocal and instrumental sections, with each vocal section typically repeating the text and melody of the first. [30] Vocal and instrumental sections, melodic sections, text phrases and their corresponding rhythmic modes (see further below) are marked in the song structure summaries that are placed after the notes for each track in part 2. [31]

The length and complexity of each section (whether vocal or instrumental) varies from song to song and even from performance to performance of the same song. For example, the length of instrumental introductions is usually determined by the amount of time the songman needs before he is ready to begin the vocal section, and this in turn may rest on any number of extra-musical factors: the dancers may not be ready, the singer may be finishing a cigarette or a conversation, or he may want to settle a performance down by having a long instrumental introduction. At the height of a ceremony he may chose to move quickly from vocal section to vocal section and begin each song with only the shortest of instrumental introductions. Instrumental sections and codas also vary in length according to a number of factors: whether there is dancing or not (un-danced versions tend to be shorter), and, if danced, the level of energy of the dancers and the overall excitement of the performance (when the dancing is going well the instrumental sections tend to be longer).

Certain songmen tend to compose particularly long complex vocal sections while others prefer them to be shorter and less complex. Jimmy Muluk and Tommy Barrtjap, for example, liked to compose complex vocal sections, while Maurice Ngulkur preferred shorter ones. Even within the repertory of individual songmen, some songs have relatively short vocal sections and others longer; and in some cases the length of vocal sections might vary even within the same song item. [32] In this situation, an experienced ceremonial song leader (such as Jimmy Muluk) keeps other musicians and dancers in synch with him by visual and aural cues such as raising the clapsticks or vocalisations to signal the impending end of an item. Textual instability of the type that results from a song not being regularly performed in ceremony may also affect the length of a vocal section.

Wangga Image Figure 2.2 Unknown dance at the tourist corroboree, Mica Beach, September 1972. Photograph by Allan Laurence, reproduced with the permission of Belyuen community.

In general, performances that include dancing tend to have longer items, which are extended by increasing the number of vocal sections. From the evidence of Barrjap’s performances (Marett, 2005, p 88), as well as those of Mandji and Muluk analysed in chapters 5 and 6, it is clear that tourist corroborees provide the context in which the longest wangga items occur (up to 17 vocal sections occur in one performance of ‘Puliki’ by Mandji). Perhaps western audiences have difficulty in engaging with conventional wangga performances that are over almost before they start. It may have been in response to such reactions that the Belyuen performance groups developed a number of relatively lengthy dance dramas to use in tourist corroborees, which often had little or no relationship with the accompanying sung text. At tourist corroborees dancers enjoy engaging with the audience—joking and playing up to audience expectations—and since this type of interaction seems to be what particularly pleases tourist audiences, it is in the interest of the group, which has after all been engaged for a paying gig, that they are prolonged for as long as possible.

Rhythmic mode

Throughout this chapter we will pay particular attention to rhythmic mode, which is one of the main musical features used by songmen to differentiate their repertory from others. We have already mentioned the important topic of rhythmic mode, without yet fully setting out the rhythmic modal system. While the system of rhythmic modes is best appreciated with reference to individual repertories, [33] the principles shared by all repertories are basically as follows. Rhythmic modes are defined primarily by the combination of tempo (that is, the speed of the clapstick beating) with the patterns articulated by the clapsticks, which may be even or in a variety of uneven (gapped) patterns (see table 2.1).

Rhythmic mode  Tempo  Clapsticks 
Rhythmic mode 1  None (unmeasured)  None 
Rhythmic mode 2a [34]   Slow  Even 
Rhythmic mode 2b  Slow  Suspended 
Rhythmic mode 3a  Slow moderate  Even 
Rhythmic mode 3b  Slow moderate  Uneven (triple) 
Rhythmic mode 4a  Moderate  Even 
Rhythmic mode 4a (var)  Moderate  Even, suspended 
Rhythmic mode 4b  Moderate  Uneven (quadruple) 
Rhythmic mode 4b (var)  Moderate  Uneven (quadruple), suspended 
Rhythmic mode 4c  Moderate  Uneven (triple) 
Rhythmic mode 4d  Moderate  Uneven (quintuple) 
Rhythmic mode 4e  Moderate  Doubled 
Rhythmic mode 5a  Fast  Even 
Rhythmic mode 5a (var)  Fast  Even, suspended 
Rhythmic mode 5b  Fast  Doubled 
Rhythmic mode 5c  Fast  Uneven (quadruple) 
Rhythmic mode 5d  Fast  Uneven (triple) 
Rhythmic mode 5e  Fast  Uneven (sextuple) 

Table 2.1 Summary of rhythmic mode terminology, showing associated features of tempo and clapstick beating patterns.

Table 2.2 presents a different view of this data for the measured rhythmic modes (2-5), showing that some combinations (e.g., uneven slow) do not occur in the wangga repertories under discussion. Such combinations may occur in other repertories of didjeridu-accompanied public dance-song, however (e.g., slow uneven triple beating occurs in the muyil lirrga repertory) (Barwick, 2006). Differentiation of repertories and genres by characteristic clapstick patterns (and other musical features) assists in establishing and maintaining a clear aural identity, which Barwick has argued helps in ceremonial functions, especially in complex events like circumcision ceremonies, where there are likely to be multiple genres and repertories being performed (Barwick, 2011).

Tempo/beating style  slow  slow
moderate  fast 
even  2a  3a  4a  5a 
even (suspended)  2b  4a (var)  5a (var) 
uneven (triple)  3b  4c  5d 
uneven (quadruple)  4b  5c 
uneven (quadruple suspended)  4b (var) 
uneven (quintuple)  4d 
uneven (sextuple)  5e 
doubled  4e  5b 

Table 2.2 Measured rhythmic modes by tempo band and clapstick beating style.

Other factors help to further define rhythmic mode, including the metrical relationship between the voice and the clapsticks (Barwick, 2003; Marett, 2005). The didjeridu too has a role in defining rhythmic mode through the use of specific patterns associated with each mode. In measured rhythmic modes, the didjeridu plays almost as great a role in supporting the vocal line and defining metre as the clapsticks, while in the unmeasured rhythmic mode 1 (without clapsticks) it maintains a degree of independence from the vocal rhythm. Rhythmic modes are also associated with affect. For example, in the Daly region as in other areas across Arnhem Land (Anderson, 1992), unmeasured rhythmic modes are often associated with particularly serious songs. An association of fast tempo with heightened mood is reflected in the term lerri (Batjamalh for ‘happy’), which at Belyuen is applied to several songs in fast rhythmic modes (by Barrtjap, Mandji and Muluk).

It is the association with dance (which can occur in both vocal and instrumental sections) that most strongly underpins the system of rhythmic modes. Through synchronising their movements with the clapstick beating, dancers tread in the footsteps of the ancestral ghosts summoned by the power of song. Both men and women dance, though the men’s dancing is usually much more prominent, usually taking place in the centre of the dance-ground, and more likely to use mimetic movements referring to the content of the song texts. Men’s dancing is based around a stamping movement, usually using alternate legs, described in Batjamalh using the verb –mara ‘kick’ (see chapter 3). By contrast, women’s dancing, which usually takes place around the edge of the dance ground, employs totally different movements (described in Batjamalh using the verb –muy ‘sway’) with alternate arms marking the beat and little if any movement of the feet.

For unmeasured and slow rhythmic modes male dancers often perform unstructured and mimetic movements, the everyday actions of ancestors—walking about, hunting, tracking, standing on one leg. By contrast, for songs in the moderate and fast rhythmic modes, accompanying male dance movements comprise a highly structured sequence of runs, stamps and flourishes (Marett, 2005, pp 101-105). The precedents for these dances lie in the ceremonies of the ancestors: that is, humans dance as the ancestors dance, and in so doing become one with them. In several songs (e.g., Jimmy Muluk’s ‘Puliki’ and ‘Lame Fella’) setting the same text to a different rhythmic mode is associated with changes in the described activity of the ancestral ghost. For example, in Billy Mandji’s lengthy performance of ‘Puliki’ (chapter 5, track 2), vocal sections with slow beating refer to the slow rhythmic movements of the ancestral Buffalo swimming across the bay to Mica Beach, while fast beating is associated with the Buffalo dancing at Mica Beach.

Later in this chapter, we will discuss rhythmic mode and dance in more detail, including distribution of rhythmic modes within each repertory. See Appendix 1 for more detailed information on the characteristics of each mode, including its relationship to dance.

Musical example: ‘Tjerri’: a wangga by Maurice Ngulkur

In this section we provide further clarification of our musical terminology and definitions of song structure through close attention to one song, Maurice Ngulkur’s ‘Tjerri’ [35] one of the most interesting of the Ma-yawa wangga songs. ‘Tjerri’ speaks directly to the nature of Dreamings (ngirrwat). The composer and singer Maurice Ngulkur addresses the Sea Breeze Dreaming ‘Tjerri’ as elder brother (mana) and sings of how the Sea Breeze manifests itself right here and now, as it has always done, causing waves to break at the mouth of the creek at the Tjerri Dreaming site (kigatiya) (see further in the notes to chapter 9, track 13).

Wangga Image Figure 2.3 Men and boys at Belyuen dancing at the launch of Allan Marett’s book Songs, dreamings and ghosts, Belyuen, 2006. Photograph by Gretchen Miller, ABC Radio National, reproduced with the permission of Belyuen community.

Wangga Image Figure 2.4 A line of women dancing wangga at a circumcision ceremony in Wadeye, 1988. Photograph by Mark Crocombe, reproduced with the permission of Wadeye community.

Wangga Image Figure 2.5 Frank Dumoo executes a spectacular final flourish, dancing wangga at a Wadeye circumcision ceremony. Photograph by Mark Crocombe, reproduced with the permission of the Wadeye community.

Wangga Image Figure 2.6 Ambrose Piarlum dancing wangga at a Wadeye circumcision ceremony, 1992. Photograph by Mark Crocombe, reproduced with the permission of the Wadeye community.

Wangga Image Figure 2.7 Musical transcription of ‘Tjerri’ as performed by Maurice Ngulkur (rec. Marett Mar98-14-s08)]. Transcription by Allan Marett.

As already suggested, the musical form of ‘Tjerri’ is quite conventional. It begins with an instrumental introduction. [36] In this song, because the vocal sections are in rhythmic mode 1 (without clapsticks), the instrumental introduction is performed by the didjeridu alone. After a few seconds the singer begins vocal section 1, which comprises five text phrases sung within a single melodic section. The definition of a text phrase is somewhat loose: in the case of texts in human language, text phrases tend to comprise single units of meaning. In the text of ‘Tjerri’ we have defined karra mana tjerri and kinyi-ni kavulh (kagandja) as two text phrases, labelled text phrase 1 and text phrase 2 in the song structure summary of chapter 9, track 13, though they could have been classified as a single text phrase. In this case, the division into two was prompted by the need to treat the unit of text in text phrase 2 as a separate entity.

As usual in Ngulkur’s performances, the precise form of the text within the text phrase varies from vocal section to vocal section and even from text phrase to text phrase. This variability can be observed in the placement of the word kagan-dja ‘right here and now’ in the performance (see figure 2.3). In text phrase 2 of vocal section 1 he sings kagan-dja kinyi-ni kavulh ‘right here and now he is always manifesting himself’, while in text phrase 4 of the same vocal section he moves the word kagandja to the end of the text phrase so that it yields kinyi-ni kavulh kagan-dja ‘he is always manifesting himself right here and now’. In text phrases 2 and 4 of vocal section 2, kagan-dja is omitted altogether, so the phrase becomes simply kinyi-ni kavulh ‘he is always manifesting himself’.

VS   TP   A   B   C  
1   1-2   karra mana tjerri  kagan-dja kinyi-ni kavulh  
3-5   karra mana tjerri  kinyi-ni kavulh kagan-dja   purangang kin-pa-diyerr kagan-dja kisji 
2   1-2   karra mana tjerri  kinyi-ni kavulh  
3-5   karra mana tjerri  kinyi-ni kavulh   purangang kin-pa-diyerr kagan-dja kisji 

Figure 2.3 Text of ‘Tjerri’ as performed by Ngulkur in chapter 9, track 13, highlighting variability in text phrases. VS stands for ‘vocal section’, TP for ‘text phrase’ and the repetition pattern proposed for the units of text is indicated by the letters A, B, C. The variable text phrase B is highlighted in bold.

If we accept the basic equivalence of these variant forms of the text phrase, the textual form of each vocal section in this song is thus ABABC (as given in figures 2.1 and 2.3). Freedom to change the position of words or to truncate a phrase stems from the fact that the song was rarely, if ever, sung in ceremony in recent times and therefore did not require exact reproduction from performance to performance (Marett, 2005, p 200). It is probably for this reason that Ngulkur’s backup singer in this performance, Ambrose Piarlum, does not articulate the text very clearly, especially in vocal section 1, where he joins in mainly on the long notes.

In ‘Tjerri’, each vocal section comprises one melodic section, which in this case (but not in all wangga songs) is therefore co-terminous with the vocal section (as shown in figure 2.1 above). Melodic sections usually comprise a descent from the highest note to the lowest note. In most wangga the lowest note falls on the same pitch as the didjeridu. To facilitate comparison between songs, the pitch of the didjeridu is notated as C in all our transcriptions (with the original pitch of ‘Tjerri’, E flat, annotated above the stave in figure 2.2). Melodic sections may in some cases be further subdivided into subsections. The cadence points on F at the end of text phrases 2 and 4 in ‘Tjerri’ would mark such a division (and are indicated with dashed lines in figure 2.1). The melodic contour of ‘Tjerri’ is used in the majority of songs in the Ma-yawa wangga repertory, particularly those that are about Dreamings. [37]

Although not written out in full in the musical example in figure 2.2, the didjeridu always continues throughout the vocal section. In the case of a metrically free vocal section like that of ‘Tjerri’, where the metre is underlyingly determined by speech rhythm (hence the marking of parlando), the didjeridu tends to mark word and morpheme boundaries with pulsations that enhance the comprehensibility of the text. The astute reader and listener may also have noticed that word boundaries are frequently marked by rhythmic prolongation of the final syllables of words (sometimes performed over several pitches as a melisma). This enhances comprehensibility.

Wangga Image Figure 2.8 Maurice Ngulkur performing Ma-yawa wangga at Peppimenarti, 7 October 1998, during the period before the ceremony when dancers were still painting up. The didjeridu player is out of the frame to the right. Photograph by Allan Marett, reproduced with the permission of Wadeye community.

Each vocal section is followed by an instrumental section performed by the clapsticks and didjeridu, with some vocalisation by the singer. In instrumental section 1 of ‘Tjerri’, this vocalisation is a held note on the vowel ‘ii’; in instrumental section 2, it is a series of grunts (the latter not notated in figure 2.2). [38] Vocalisation within instrumental sections is another common but optional performance convention used by all singers from time to time, but most singers never deliver melodic text during an instrumental section. [39]

It is in the instrumental sections that the emphasis turns to dance, as the dancers enter to perform their rhythmically dynamic actions. For this reason, in instrumental section 1 of ‘Tjerri’ we can hear the mildly irregular metre of vocal section 1 replaced by regular even stick beating in a moderate tempo quaternary metre (rhythmic mode 4a). Even stickbeating patterns in moderate or fast tempo are used for non-final instrumental sections in all Ma-yawa wangga and Walakandha wangga [40] whose vocal sections are in rhythmic mode 1. The singer usually signals the end of an instrumental section with a clapstick pattern in the approximate rhythm ryqryqqQ. Because of its pervasiveness in the Walakandha wangga repertory, Marett has previously termed this the ‘Walakandha wangga cueing pattern’ but in fact it occurs in other repertories including Muluk’s (chapter 5) and the present Ma-yawa wangga (chapter 9) (Marett, 2007, p 71). [41]

In ‘Tjerri’ instrumental section 2 is the final instrumental section. To signal to the audience, didjeridu player and any dancers that this will be the end of the item, the lead singer Ngulkur commences a fast doubled beating pattern (rhythmic mode 5b), with interlocking beats provided by the backup singer Piarlum (in figure 2.2 we have notated this as an interlocking crotchet beat, performed by Piarlum, inserted into the quaver pattern performed by Ngulkur, the lead singer). In ‘Tjerri’, the crotchet tempo of the final instrumental section is approximately 130bpm (as compared to 136-140 for other fast tempo rhythmic modes in the Ma-yawa wangga repertory). Often final instrumental sections with interlocking beating in the fast doubled rhythmic mode (rhythmic mode 5b) are performed at a slightly slower tempo than is usual for the fast even rhythmic mode (rhythmic mode 5a), perhaps to make it easier for the backup singer to place the interlocking crotchet beats correctly, or perhaps because of the greater difficulty of performing regular fast quavers.

Cross-repertory variation

Song texts and vocal sections

Vocal sections vary a great deal in form from repertory to repertory and from song to song. Song texts may be entirely in spirit language or entirely in human language, or in a mixture of both. The number and complexity of text phrases and melodic sections too may vary. In some cases, vocal sections may contain contrasting types of text; see for example Muluk’s famous ‘Puliki’ song (chapter 5, track 1), where a portion of text in spirit language (text phrases 1-3) is sung to the first melodic section (that is, one full melodic descent), and text that is largely in human language (text phrases 4-5) is sung to the second melodic section.

As already mentioned, some singers (for example Barrtjap and Lambudju) have quite stable texts, which are sung more or less exactly the same way from vocal section to vocal section and from performance to performance. We have already noted in the above discussion of ‘Tjerri’ that other singers may display considerable instability, not only from performance to performance, but also from vocal section to vocal section of the same item. Textual stability assists the singers to maintain the strong vocal unison desirable in ceremonial performance and also makes it easier to co-ordinate the actions of the dancers. Once a repertory is not regularly sung in ceremony, singers seem to delight in ringing the changes on their texts (see for example Ngulkur’s three performances of ‘Walakandha Ngindji’ in chapter 9, tracks 1-3). This sort of instability is not to be confused with the highly structured variation of both text and music that an experienced singer like Muluk can use to great effect without diminishing the ability of the song to function ceremonially (as can be seen in his performance of five consecutive items of ‘’ in chapter 5, tracks 10-12).

Across the wangga corpus, the majority of songs have one or two melodic sections per vocal section (the example of ‘Tjerri’ discussed above had only one), although more melodically complex vocal sections are not uncommon. For example, vocal section 1 of Ngulkur’s song ‘Malhimanyirr’ (chapter 9, track 17) has four melodic sections (in this case, formed by two repetitions of a two-melodic section text).

Vocal timbre

Vocal timbre (or voice quality) is an integral quality distinguishing the different repertories, especially at Belyuen, where singers consciously imitate the voice of the their ancestral singers (whether alive or dead). Thus, when we played back Elkin’s 1952 recordings, our collaborators at Belyuen were unable to distinguish the recorded voices of Tommy Barrtjap and his teacher Jimmy Bandak (Marett, 2000; 2005, p 68). We have also observed similarities in vocal quality between apprentice and master in Roger Yarrowin’s performance of Lambudju’s song ‘Mubagandi’ (chapter 7, track 25), and Colin Worumbu Ferguson’s performance of songs sung by Billy Mandji (see notes to Worumbu’s performance of ‘Puliki’, chapter 5, track 4). While each lineage at Belyuen might be said to have a characteristic vocal quality (piercing, with highly controlled vocal ornamentation for Barrtjap’s lineage 1; a lighter, sweeter tone for Lambudju’s lineage 2; a strong and darker-toned husky quality for Muluk’s lineage 3, and a forceful rounded quality for Mandji’s lineage 4) [42] some songs, especially those of Billy Mandji (for example, ‘Happy (lerri) song no. 2’ in chapter 6, track 5), are notable for their systematic alteration of vocal quality within the vocal section, with prominent timbral changes giving the impression of a conversation. Marett states that reproduction of the voice quality of deceased songmen and ancestral song-giving ghosts adds to the songs’ power to create a liminal space between the worlds of the living and the dead (Marett, 2005, p 68).

At Wadeye, by contrast, there is no conscious effort to imitate the voice of the originator of the song (Marett, 2005, p 68). Rather, at least in the Walakandha wangga tradition, in ceremonial performances there is a preference for strong vocal unison from a rather larger number of singers (up to four or five). Perhaps this is due to singing groups of similar size in the two Wadeye sister traditions lirrga and djanba. Certainly in very large ceremonies a larger group of singers is more likely to be heard, and the content of their songs to be understood.

The power of the voice is evident in other traditional practices in the Daly region and indeed in much of Aboriginal Australia. Traditional owners often call out in their ancestral language to introduce strangers to ancestors residing in the ‘sentient landscape’ (Povinelli, 1993) of their homelands. Voice, like sweat, [43] has the power to penetrate bodies and landscapes, to integrate the strange and to summon the company of the ancestral ghosts to ceremony (Marett, 2003).

Melodic considerations

In this section we give more detail on the use of melody and melodic mode in defining wangga repertories. We define melody as a formed by a fixed sequence of pitches that recurs regularly in each vocal section.

Sharing melody across repertories

It is quite uncommon for a melody to be shared across repertories. Rather, each repertory has one or more characteristic melodies that form part of its aural identity. When Maurice Ngulkur took the Walakandha wangga song ‘Walakandha No. 2’ (chapter 8, track 23) as the model for his song ‘Walakandha Ngindji’ (chapter 9, tracks 1-3), one of the most significant changes he made was to change the melody from one typical of the Marri Tjavin Walakandha wangga to one emblematic of Marri Ammu people, their country and their Dreamings (Marett, 2005, pp 137-150). Such performative gestures rest upon a convention that is widespread throughout the parts of Aboriginal Australia where song traditions are still strong, namely that melody encodes relationships between songs, totemic ancestors, country and kinship (Keen, 1994; RM Moyle, 1979, p 71; Toner, 2001). Given the power of melody to articulate such relationships, there was no more potent way for Ngulkur to bring the song into the Marri Ammu cultural sphere.

The melody/repertory relationship

In wangga generally, repertories commonly use more than one melody, but some repertories have a more restricted number of melodies than others. As already mentioned, in the Ma-yawa wangga (chapter 9) most songs that refer to Dreamings (ngirrwat) or Dreaming places (kigatiya) in Marri Ammu country share a melody, and most of the early Walakandha wangga songs use very closely related if not identical melodies. The Walakandha wangga repertory has a high proportion of shared melodies: 25 of its 34 songs share a melody with at least one other. By contrast, the Belyuen repertories use many different melodies, though there are still one or two examples of songs that share a melody, such as ‘Tjinbarambara’ and ‘Wak’ in Muluk’s repertory.

The only repertory containing songs that may be sung to more than one melody is the Walakandha wangga repertory, such as ‘Truwu’ and ‘Mirrwana’. Perhaps it was this precedent that inspired Ngulkur to set the text of the Walakandha wangga song ‘Walakandha No. 2’ to his own Marri Ammu melody in ‘Walakandha Ngindji’.

The melodic mode/repertory relationship

It is common for songs within a repertory to share a melodic mode. All repertories include at least two different melodic modes (see table 2. In Songs, dreamings and ghosts, Marett demonstrated that there is a relationship between melodic mode and lineage (Marett, 2005). Marett has also shown how in performance singers can manipulate melodies in order to articulate the intersection or divergence of clan or language group interests (Marett, 2005, pp 117-121, 137-150).

Barrtjap’s songs, for example, were for decades performed by a single patriline, passing from Barrtjap’s classificatory father Jimmy Bandak to Barrtjap himself, and then to Barrtjap’s sons, Kenny and Timothy. This straightforward transmission of the repertory has led to significant musical homogeneity. All but one of Barrtjap’s songs have melodies that use a descending dorian series (Bb A G F Eb D C). This remains true today, even though, as we saw in chapter 1, these songs are now sung by singers from outside the Barrtjap lineage, such as Roger (Rossie) Yarrowin and Colin Worumbu Ferguson. The reason why this melodic homogeneity continues is probably that no new songs have been added to this set for at least a decade.

On the other hand, a complex history such as we find for Lambudju’s repertory may result in a diversity of melodic modes. Marett has suggested that older songs inherited from Lambudju’s father’s brothers use all or part of a descending lydian series (C B A G F# E D C), while songs of Lambudju’s own composition tend to use a dorian series (or part thereof, in the case of pentatonic or hexatonic melodies) (Marett, 2005, pp 194-195). A similarly complex musical history may underlie Jimmy Muluk’s use of a mixolydian modal series for two of the older songs in his repertory, while most are in a major mode, with one sharing chromatic tendencies with two of Lambudju’s songs. The distribution of melodic mode in Mandji’s repertory is similar to Muluk’s, perhaps reflecting their common musical history. Within the Walakandha wangga repertory (chapter 8) most songs composed by Stan Mullumbuk, together with those by members of the Kungiung, Dumoo and Kundjil families, have melodies in the dorian mode, while all songs composed by Mullumbuk’s younger brother Philip Mullumbuk are in a major series.

dorian   major   lydian   mixolydian   chromatic  
Barrtjap  most  one 
Muluk  most  old  one 
Mandji  most  one 
Lambudju  newer  older  two (new) 
Walakandha  older  newer 
Ma-yawa  most  some 

Table 2.3 Distribution of melodic mode across the six repertories.

Rhythmic considerations

In this section we will examine the use of tempo, clapstick beating style and rhythmic mode across repertories, and show that certain rhythmic modes or combinations of rhythmic modes occur more frequently in certain repertories, suggesting that rhythmic mode is an important component of the aural identity of the repertory. The following consolidated data draws on the detailed information set out in the musical analysis summary for each repertory (see the relevant chapters for more detail). Additional information on the characteristics of each rhythmic mode is given in Appendix 1.

Because of significant discrepancies between the early and late repertories of the Walakandha wangga (see further discussion in chapter 8) we deal with them separately in this section.

Cross-repertory use of rhythmic mode and its components

For the purposes of clarifying cross-repertory variation in use of rhythmic mode, we have found it useful to separate the rhythmic mode into its two components, tempo and clapstick beating style, to account for the fact that certain singers perform certain clapstick beating styles more commonly in one tempo than was usual in another repertory. For example, in final instrumental sections Muluk liked to employ moderate doubled beating (rhythmic mode 4e), while other repertories (such as Barrtjap and the Walakandha wangga) preferred fast tempo for their doubled beating (rhythmic mode 5b).


Collating data from the various performances in chapter 4-9, we can observe that different repertories consistently use slightly different absolute tempi for the various tempo bands (see table 2.4). For example, we can observe that Mandji’s slow tempo songs (with a tempo range of 46-48bpm) are considerably slower than those of Lambudju (64-69bpm). Some repertories, such as Barrtjap’s, use a relatively wide range of tempi within a tempo band, while others are more restricted. Lambudju’s unique use of four rather than three tempo bands means that each tempo band is relatively smaller in range.

tempo band 
Slow  Slow Moderate  Moderate  Fast 
Barrtjap  58-65  117-120  126-144 
Muluk  50-55  110-113  126-140 
Mandji  46-48  110-116  130-140 
Lambudju  64-69  99-107  110-116  120-125 
Walakandha (early)  60-72  112  120-136 
Walakandha (post 1986)  55-65  133-142 
Ma-yawa  117-125  134-146 

Table 2.4 Tempo bands by repertory, showing range of absolute tempi used across each repertory (expressed in beats per minute).

If we examine the relative frequency of use of each tempo band by repertory, calculated by collating information on the usual tempo band for each song in the repertory (see data in the relevant chapters), [44] significant inter-repertory differences again emerge (see table 2.5). For convenience of discussion we count the lack of clapstick beating as its own tempo band category.

Tempo band clapsticks   Barrtjap   Muluk   Mandji   Lambudju   Walakandha (early)   Walakandha (late)   Ma-yawa  
none  18  32  13 
slow  19 
slow mod  17 
mod  13  40  25  33  23  11 
fast  76  35  60  36  55  68  76 
TOTAL  100  100  100  100  100  100  100 

Table 2.5 Proportional use of the various tempo bands across repertories (preferred tempo for each repertory highlighted in bold font). The numbers represent percentage values based on the number of times a particular tempo band occurs per song section in each repertory.

Key observations include the lack of slow tempo songs in the Ma-yawa repertory (see chapter 9) and the lack of moderate tempo songs in the late Walakandha wangga repertory. Each of these recent Wadeye repertories uses only three tempo bands, whereas the early Walakandha wangga conforms to the pattern of most Belyuen repertories in using four clapstick tempi (none, slow, moderate and fast tempo bands). Lambudju’s repertory is the only one to use all five tempo bands, and the only one to include slow moderate tempo. Muluk’s repertory is the only one to use the moderate tempo band more frequently than the fast.

Significant differences between repertories emerge when we consider the relative use of the various tempo bands by song section.

  • Vocal sections: Overall, fast rhythmic modes are slightly preferred in vocal sections, with the next most common being rhythmic mode 1. Both repertories of Walakandha wangga display a clear preference for rhythmic mode 1 (no clapsticks). Muluk prefers slow tempo for his vocal sections, while Lambudju shows an equal preference for moderate and fast. Barrtjap displays an overwhelming preference for fast tempo in vocal sections. Mandji and Ma-yawa also prefer fast, though like Muluk and Lambudju they have a more even distribution of songs across the various tempo bands.
  • Internal instrumental sections: All repertories prefer fast rhythmic modes, except for Muluk, who prefers moderate, and Lambudju, who again shows equal preference for moderate and fast tempi.
  • Final instrumental sections (Barrtjap coda): All repertories prefer fast tempo, which in danced performances usually constitutes the climax and allows the most virtuosic dancers to show their ability to maximum effect.

Clapstick beating style

The other component of rhythmic mode, clapstick beating style, also shows some interesting variability across repertories (table 2.6). For convenience of discussion we count the lack of clapstick beating as its own clapstick beating style category.

clapstick beating style   Barrtjap   Muluk   Mandji   Lambudju   Walakandha (early)   Walakandha (late)   Ma-yawa  
none  19  30  14 
even  25  44  60  42  33  40  18 
even (suspended) 
uneven (triple)  31 
uneven (quadruple)  38  23  22  14  57 
uneven (quadruple suspended) 
uneven (quintuple)  10 
doubled  17  33  33  30  11 
TOTAL  100  100  100  100  100  100  100 

Table 2.6 Proportional use of the various clapstick patterns across repertories (preferred tempo for each repertory highlighted in bold font). The numbers represent percentage values based on the number of times a particular clapstick pattern occurs per song section in each repertory.

Two clapstick beating styles occur in all repertories: none, and even beating. Even patterns are preferred in all repertories except Barrtjap’s and the Ma-yawa wangga, which prefere uneven (quadruple) beating patterns (qqqQ). Barrtjap and Mandji employ the most number of different patterns (six) and Barrtjap’s is the only repertory to include the uneven quintuple beating pattern (qqQqQ). Muluk uses five different beating patterns and the remaining four repertories (Lambudju, the early and late Walakandha wangga, and the Ma-yawa wangga) employ four each, though in each case a different combination.

More differentiations appear when we consider separately the distribution of clapstick beating styles across each structural division of the song.

  • Vocal sections: Muluk, Mandji and Lambudju show a preference for even beating, while Barrtjap and Ma-yawa prefer the uneven (quadruple) pattern, and as already noted above, both eras of the Walakandha wangga prefer rhythmic mode 1. Note that it is only in vocal sections that suspended patterns occur (in Barrtjap, Muluk, Mandji and in both eras of the Walakandha wangga even suspended patterns occur, and in Mandji we find the sole example of uneven quadruple suspended).
  • Internal instrumental sections: we again see Barrtjap and Ma-yawa grouping together in their preference for the uneven quadruple pattern, while all the remaining repertories show a strong preference for even beating.
  • Final instrumental sections (Barrtjap coda): Again Barrtjap and Ma-yawa prefer the uneven quadruple pattern, while Lambudju and Mandji prefer even beating. Only Muluk and the two sub-repertories of the Walakandha wangga have a different preference for final instrumental sections (or coda) from internal instrumental sections. In both cases the preference clapstick beating style changes from even beating to doubled beating. Although this similarity was not singled out by Marett in his 2007 paper on the relationship between the Walakandha wangga and Muluk’s repertory (Marett, 2007), it lends additional weight to his argument that Muluk’s wangga was a compositional model for the Walakandha wangga repertory.

Combinations of rhythmic mode

We will finish our exploration of rhythmic mode across the wangga corpus by considering the proportion of songs in each repertory that occur in mixes of rhythmic modes.

Firstly, analysis reveals that the majority of songs (59%) in the corpus of 103 songs display some mixing of rhythmic modes, whether within sections of an item or between different performances of the same song. This is not evenly distributed between repertories, however. The repertories of Mandji, Lambudju and Ma-yawa have a preponderance of songs in the same rhythmic mode throughout (55%, 85% and 69% respectively), while the repertories of Barrtjap, Muluk and the Walakandha wangga have a majority of songs mixing rhythmic modes (constituting a massive 96% of songs in the late Walakandha wangga, due to the preponderance of vocal sections using rhythmic mode 1).

There are various ways in which different rhythmic modes can be applied to the same song. These are not mutually exclusive; indeed, Muluk’s repertory in particular exhibits many different ways of combining rhythmic modes.

Presenting the same text in different rhythmic modes in successive items occurs only in the repertories of Muluk, the early Walakandha wangga, and the Ma-yawa wangga. [45] In Muluk’s repertory the songs concerned are ‘Rtadi-thawara’, which occurs with the vocal section set in three different rhythmic modes (2, 5c and 5b), ‘Lame Fella’, which occurs in two different rhythmic modes (2 and 5a), and ‘Lerri’, which occurs in three different rhythmic modes (2, 4a and 5b) (see table 5.1, chapter 5, for more detail). In the early Walakandha wangga, the songs ‘Walakandha no 8’ (rhythmic modes 4c and 5a) and ‘Walakandha no. 9’ (rhythmic modes 4a and 1) each occur in two different rhythmic modes, while in the Ma-yawa wangga it is ‘Wulumen Tulh’ (rhythmic modes 5c and 1).

Across all repertories, songs with vocal sections in rhythmic mode 1 and the slow rhythmic modes have fast or moderate tempi in the instrumental sections, required to allow the dancing characteristic of the style. There is also frequently a difference in rhythmic mode between internal and final instrumental sections (or coda), irrespective of the rhythmic mode used in the vocal sections (this feature occurs in almost half of the songs in the corpus, 49 songs out of 103 songs, and is especially common in songs using rhythmic mode 1 for the vocal section/

Presenting the same text in different rhythmic modes in different vocal sections within an item is unevenly distributed across the various repertories, with this feature being used in various ways in the Barrtjap, Muluk, Mandji and Ma-yawa repertories, but being absent from the Lambudju repertory and both eras of the Walakandha wangga repertory. In Barrtjap’s case, six songs start out with the vocal sections set to the fast doubled rhythmic mode (5b), followed by one or more vocal sections in one of the other fast modes, most commonly by 5c (fast uneven quadruple) but in one instance by 5a (fast even) (see table 4.2 in chapter 4 for details). In the Ma-yawa, both performances of the song ‘Walakandha Ngindji’ (chapter 9, tracks 1-2) have the first vocal section in rhythmic mode 5c, and the second vocal section unaccompanied (rhythmic mode 1).

Mixing of rhythmic modes within a vocal section occurs most frequently in Muluk’s repertory, where rather than using modes that contrast entire vocal sections, he instead uses suspension of the beating accompaniment (rhythmic modes 4a (var) and 2b) in a systematic way to differentiate the initial vocal section from subsequent vocal sections. For example, in the slow versions of both ‘Lame Fella’ and ‘Rtadi-thawara’ (chapter 5, tracks 13 and 15), he suspends beating throughout the first entire first vocal section (rhythmic mode 2b throughout), whereas in the second and subsequent vocal sections begin with the usual (non-suspended) slow beating (rhythmic mode 2a). In the five items of Piyamen’ga (chapter 5, tracks 10-12), we can observe a similar systematic use of the suspended form of the moderate tempo rhythmic mode (4a (var)). In these cases, the first vocal section in each item begins with the non-suspended form (rhythmic mode 4a), whereas the second and subsequent vocal sections begin with the suspended form, rhythmic mode 4a (var) (see chapter 5 for further details).

In Muluk, Mandji and the early Walakandha wangga repertories we find a number of songs that use contrasting rhythmic modes in different melodic sections or text phrases within the vocal section. For example, Muluk’s song ‘Puliki’ (chapter 5, track 1) uses slow even beating (rhythmic mode 2a) in the first melodic section, and suspended slow beating in the second melodic section. We have already discussed above a number of instances in the Barrtjap, Muluk, Mandji and Walakandha wangga repertories in which suspended beating for part or all of a text phrase in the slow or moderate tempo bands (rhythmic modes 2b and 4a (var)) occurs in combination with the non-suspended forms (rhythmic modes 2a and 4a).

It is important to note that combinations of rhythmic modes within a vocal section only occur between the unaccompanied and slow rhythmic modes, or between the suspended and non-suspended forms of the slow and moderate rhythmic modes. The great variability in rhythmic mode we can observe in Muluk’s repertory is perhaps in part attributable to his preference for moderate and slow rhythmic modes, that is, the modes with the most combinatorial possibilities. There are no instances in the corpus of vocal sections that combine moderate and fast rhythmic modes, or indeed slow and fast rhythmic modes. Indeed, one can imagine that this sort of rhythmic modal variability would be rather difficult to manage, both for musicans and dancers.

A few songs in the Muluk and early Walakandha wangga repertories, along with two songs in the Ma-yawa wangga repertory, have the feature of mixing rhythmic modes within a single instrumental section. This occurs in rhythmic modes 4* and 5*, in which internal instrumental section begins in doubled beating (rhythmic mode 4e or 5b), to be followed by a return to the even form of the same tempo band (4a or 5a). This distribution pattern confirms the influence of Muluk’s repertory on the Walakandha wangga, and perhaps a subsequent influence of the Walakandha wangga on Ngulkur’s performances of his Ma-yawa wangga repertory. [46]


We have gone into some detail on the differential use of rhythmic modes and their combinations across repertories, observing that different songmen have developed significantly different rhythmic modal profiles. We argue that, in combination with other factors (such as vocal timbre, use of a particular ancestral language, use of various melodic characteristics), these rhythmic modal characteristics contribute towards the aural identity or gestalt of the repertory, which allows it to be identified quickly in a complex ceremonial soundscape, and facilitates its function as a powerful vehicle for effecting social change through communication and danced interaction with the worlds of the living and of the dead.

Figure 2.2: Musical transcription of ‘Tjerri’ as performed by Maurice Ngulkur (rec. Marett Mar98-14-s08)