| FROM MASS TO LINE - TONALITY, HARMONY AND MELODY
Composing is essentially a problem of capturing force with form. Although a composer ostensibly "puts together" patterns of sound, like an architect who builds up the design of a cathedral, he is also like a carver who chips away at an existing mass in order to give it new, culturally significant form. Beethoven is one of many musicians for whom we have evidence of a composing process in which precise, but amorphous, ideas and lumps of sound are eventually crystallised into patterns of melody. The processes by which a musician puts together or chips away, as well as the materials which he assembles or "carves," are all influenced by the patterns of both his culture and the behavioural processes which he has learnt as an individual member of it. Underlying all these learnt and often conscious processes are the deeper processes by which composers and performers share with others the original experiences or ideas which inspired their work. A fine example of this reverse process, by which listeners are taken back to where the composer began, is to be found in the glorious opening of Bach's St John Passion: short rhythmic patterns merge into lines of melody, blocks of harmony, and finally huge tonal masses as the captured listener responds with his body, his heart, and his head, until he is enveloped in the experience of being from which Bach's musical ideas must surely have been crystallised. Although the circumstances of creation are different, chiefly because of different patterns in the division of labour, much Venda music is composed by the processes which I have just described, and its main purpose is to induce in performers and audience an expansion of feeling and greater experiences of being. It is true that cultural situations suggest word-phrases whose rhythms and speech-tone patterns generate and influence patterns of melody. But the influence of words on Venda musical composition is not as great as has been suggested for other African musical traditions, and it is only in certain types of Venda music, and in certain parts of their melodies, that words are important. It is for this reason that the Venda do not object to "distorting" the speech-tone patterns of words when they sing new music in the modern style, and any attempt to create church music that is heavily word-orientated would be contrary to the spirit of Venda music. The Venda concern with musical sound is well illustrated by their practice of converting a word-based response into a melody accompanied by vocables such as hee, ahee, as is common during domba (see also Songs 16 and 48). It is surely no coincidence that songs like Numbers 36, 37 and 47 were the most popular at the initiation schools which I attended and that they were sometimes performed without any words at all. Venda musical composition can be described as a process by which the particular experiences of culture are united with universal human experiences by the mediation of humanly created patterns of sound which capture universal force with cultural form. To mix the metaphors which I have already used, we could say that it is by building up patterns of tempo, meter, and rhythm that the Venda chip away at blocks of tonality and harmony to produce melody. In Venda Children's Songs I showed that the Venda may describe several different melodies as the same (Blacking 1967b:177-79), and that the same words may be sung to two different melodies, which are regarded as the same because they are "harmonically" equivalent (Blacking 1967b:109, 168). Elsewhere, I described how a controlled test revealed preferences for certain "harmonic" progressions in the performance of Venda ocarina duets (Blacking:1959c). My purpose here is to show how the melodies of the songs of girls' initiation are shaped according to preferred tonal and "harmonic" progressions. In order to avoid constant use of inverted commas during my discussion I shall use certain musical terms in a special way, to suit the context of Venda music.
The Venda have no words for scale or mode. They have the word mutavha, which is used for a complete set of divining dice, of metal amulets, or of reed-pipes, and also for a row of keys on a xylophone or hand-piano. Thus a mutavha may include more than one octave of a heptatonic or a pentatonic scale. The Venda state that heptatonic and pentatonic sets sound different, but they do not express the difference in terms of the division of an octave into seven or five intervals. The names of reed-pipes and xylophone keys express their recognition of the octave: the tones an octave above thakhula and phala are called thakulana and phalana, which mean "little thakhula" and "little phala." The Venda national dance, tshikona, is for them their most important music (Blacking 1965:37, 52): it expresses the values of the largest social group to which a tribal Venda belongs, its performance involves the largest number of people, and its music incorporates the largest number of tones. As many as 300 men may play at once, but even if they use only one set of twenty-four pipes and three drums, the music still covers a wider range of tone and timbre than any other Venda music. I have already shown how tshikona may be taken as a model to explain the characteristic melodic patterns of some children's songs (Blacking 1967b). In using it again as a model to elucidate Venda rules of tonality and harmony I must emphasise that I do not assume that other items of music are necessarily derived from tshikona. I suggest that tshikona and these other items of music are modelled on the same set of rules; but since tshikona demonstrates their application most dramatically it may serve as a powerful analytical tool. In particular, since the reed-pipes are named (Kirby 1933 and 1934:155-62), their musical function in tshikona may provide important clues to the identification of modes and their tonality. Given a fixed scale (mutavha) of twenty-four tones, one can obviously select from it seven heptatonic modes as a basis of creating melodies. If a person trained, say, in the Ambrosian or Gregorian systems of plainsong comes to listen to these melodies, he will be inclined to interpret them in terms of a known set of rules and will identify the modes by their dominant and final tones. Such evaluations can indeed be made of Venda melodies, but they would be irrelevant and incorrect in the context of Venda music. By common consent, the two most important tones in the Venda reed-pipe scale are phala, the keynote, and thakhula, "the lifter" which lies one whole tone above the keynote and "lifts" a melody back onto it. It has a similar function to that of the leading note in a European major or minor scale but in the opposite direction: it leads down a whole tone instead of up a semitone. Figure 2 is a transposed transcription of part of the music of tshikona, and A is the keynote (phala) and B the leading note (thakhula).
|
| Tshikona therefore begins on the tonic and moves away from it, to be brought back by its leading note at the end of each repeated phrase. Thus, in many melodies where there is a sequence of seven descending tones (as in Songs 39 and 46) we may be fairly confident that the first is not only the tonic of that particular row but also the final of the mode of which the whole melody is constructed. However, without knowing the harmonic progression of the melody we cannot be sure, because the use of a descending heptatonic tone-row in a melody does not necessarily mean that it is also the mode of the melody. For example, the tonic of such a tone-row could in fact be the leading note of the mode, and the six following tones merely passing notes. If during tshikona someone chose to sing the tone-row (a) in Figure 3 (below), it would be perfectly clear that he was not singing either of the prominent modes given as (b) and (c).
The mode given in Figure 3 (b) is the chief mode of tshikona, and its final, a', coincides with the keynote (phala) of the reed-pipe scale. Its associated mode is pitched a fifth below, following the Venda rule of harmonic equivalence in which every tone has a companion tone (Blacking 1967b:168-69). If the tshikona pattern follows a general rule of Venda music, it can be said that in chords of the fourth or fifth the lower tone is the root of the fourth and the upper tone the root of the fifth-which is the reverse of the situation in European diatonic music.
b) c)
The importance of the harmonic progression can be seen when tshikona is compared with khulo, from the great song of the domba initiation school. Khulo is an adaptation of tshikona for girls' voices; but it is a transformation and not a transposition, as the changing position of the tritone in Figure 4 reveals. The differences between the modes combined in tshikona and khulo are eliminated at the level of the common harmonic progression which they share. In khulo this is emphasised by the addition of extra voice parts, called u bvumela. In khulo the secondary companion mode of tshikona (the G-mode, B in Figure 4) has been selected as the primary mode, possibly because of its greater suitability for voices. When, according to the rule of harmonic equivalence, it is accompanied by its own companion mode, it follows that the tritone will occur at a different point in the sequence of chords. Since tritone chords are generally avoided in vocal music, the second chord of khulo is an octave. If khulo were a transposition and not a transformation of tshikona, its chords and the position of its tritone would be as in Figure 4c In the tshikona pattern every tone has two companions, the tones a fifth below and a fifth above. The chords in which a tone is "on top" are those in which it is in its strongest position tonally. Thus the chord (A/D) with the keynote (A) at the beginning of the tshikona pattern is stronger than the chord (E/A) in which it is accompanied by its secondary partner. Conversely, the first chord with the leading note (F/B) is weaker than the final chord (B/E) of the pattern, in which the leading note is given full strength to "lift" the melody back onto the keynote. In the chords of the harmonic progression common to both tshikona and khulo, the process may be summarised as follows: the overall tonic is D, but tonality shifts regularly between D and its leading note E. During every two such tonal shifts there is one harmonic progression from a strong tonic to a strong leading note. Thus tonally and harmonically the strongest parts of the pattern are the first and last chords, in which the tonic and the leading note respectively play their tonal roles. This is illustrated in Figure 5. The music of domba illustrates the working of another rule which applies to several Venda melodies. The master of initiation begins the solo call on the E of the last chord of khulo and descends a fifth to the A of the fourth chord. The khulo pattern continues and is repeated, and the master begins the next solo call on the final chord. The rule is that the tonality of the call is centred around the leading note, whilst the chorus establishes the tonic or final of the mode. This is perfectly logical when one considers that very often only the chorus responses of the songs are sung, particularly when novices are practising ndayo movements. Thus the chorus must establish the tonality of the mode, for otherwise a performance could have no established tone-centre. It may be argued that in using tshikona and domba as models for understanding the songs of girls' initiation I am imposing on the music a preconceived system which will distort its meaning as much as would the use of European musical theory. To this I can only say that after trying to understand the melodies from many points of view I found that the use of these models was able to reduce numerous inconsistencies; and secondly, as I have already mentioned, it is not the structures of tshikona and domba as much as the rules that underlie those structures which I expect to find present in the music of the girls' schools. One example will illustrate the kind of difficulty that arises in trying to analyse Venda melodies on the basis of their surface structure and show why the deeper harmonic progressions are more important to the Venda than the precise composition of the tone-rows used. As a result of changes in speech-tone, following changes in words, or of limitations in the range of a voice, an apparently pentatonic melody may be converted into a hexatonic melody. For example, in Song 44 may become
To describe the melody as basically hexatonic would be misleading, and to describe its tonality as pentatonic would be wrong, because the harmony from which the extra tone is derived is inconceivable within the limits of a pentatonic scale. The B's should really be regarded as passing notes, and we could say that the basic mode is but the tone row is In other words, the apparent simplicity of a melody may conceal the greater complexity of the musical structure on which it must be based. I have illustrated this in connection with Venda children's songs (Blacking 1967b), where I showed that the patterns of some four-tone melodies presuppose the existence of the idea of two heptatonic modes moving together at the interval of a fifth (i.e., as in tshikona). At this point it will be helpful to summarise some of the rules of harmony and tonality, using the numerical system set out previously (Rules of Venda Music I, II and III). 4.0.0. The Venda have no word for 'scale', but the word mutavha refers to a set of twenty-four heptatonic reed-pipes, twelve pentatonic reed-pipes, or a row of keys on the xylophone or hand-piano. 4.0.1. The tones of each mutavha are named. 4.0.2. In the mutavha of heptatonic reed-pipes used for the national dance, tshikona, the keynote is called phala, and the tone above it, which acts as leading note, is called thakhula, "the lifter." 4.0.3. In tshikona the keynote begins each pattern and the leading note ends it. Thus, tonality moves from tonic to leading note and then directly back to tonic at the beginning of the new pattern. 4.1.0. There is a relationship between the scales of reed-pipes and the modes on which Venda melodies are based. 4.2.0. There is a relationship between the tonal and harmonic systems of music played with the fixed scales of reed-pipes and that sung by voices. 4.2.1. Certain chords are accepted in instrumental music but avoided in part.singing (e.g., the tritone: see Figure 4). 4.3.0. Melodies may be called bitonal, in the sense that they shift from the influence of one implicit or explicit tone-enter to another. 4.3.1. In antiphonal music, or music modelled on the antiphonal situation of solo and chorus, tonality shifts regularly between solo and chorus sections. 4.3.2. Since the chorus is constant, its tonality is more fundamental than that of the solo (i.e., in terms of tshikona it has phala tonality, the tonality of the keynote). 4.4.0. Every tone may be conceptualised as having a companion tone in harmony with it. 4.4.1. Any melodic line may consequently be conceptualised as a stream of chords. 4.4.2. Melodies may therefore be as much selections of tones from 'hidden' patterns of chords as the generators of patterns of chords. 4.4.3. Two melodies which are harmonically equivalent are regarded as the same even though their melodies may sound different. 4.5.0. Tshikona serves as a model for tonal and harmonic principles in heptatonic music; or at least tshikona and other heptatonic music follow the same, unstated model. 4.5.1 .Chorus sections of melodies should begin on the tonic or its related tone. 4.5.2. In a sequence of seven descending tones the first may be taken as the tonic of that tone-row. It may also be the final of the mode of the whole melody. 4.5.3. The closest companion to any tone is the tone an octave above or below it. This relationship is often expressed in terms that resemble the grouping of people by age. For example, the tone an octave above phala is called phalana (little phala). 4.5.4. The second, and perhaps more intimate companion of any tone is that which is one fifth below it, or its octave a fourth above. 4.5.5. The third related tone is that one fifth above it, or its octave a fourth below. 4.7.0. Harmonic variation and shifting tonality are achieved both by melodic movement and by changing the companions of tones (see Figure 5). 4.7.1 .Harmonic cohesion and 'direction' may be reinforced by repeating tones, but with different and related chords. Thus, in Figure 5, the tonic is heard firstly with its own primary companion, but secondly with its secondary companion, which in turn is the primary companion of the leading note of the tonic. |
| Before proceeding with the analysis of the girls' initiation songs it must be appreciated that several melodic phrases which sound different may be considered functionally alike. Figure 6 shows some examples of these melodic parallels.
As the examples in Figure 6 suggest, many melodic phrases are alternatives for streams of descending tones, which are a common feature of the songs, as they are of tshikona, and which therefore follow the same pattern of tonality. For instance, as Figure 7 shows, Song 46 uses the G-mode of tshikona (B in Figure 4); and Songs 16 and 41 (G-mode), and 22 (D-mode) follow the same pattern.
But although the sequence of tones follows exactly that of tshikona from the fourth chord to the end and then right through from beginning to end, the rising minor third at the beginning and the chords of the fourth suggest that it is a transformation in which the functions of the D- and G-modes are retained but their positions reversed, a transformation similar to that found in the domba song. In Song 25 the G-mode appears to be used, but consideration of the shape of the melody and a comparison with the transposition of tshikona in Figure 4c shows that it is an exact replica of tshikona, so that D must be the tonic (see Figure 8). Moreover, the scale from G does not descend as far as the leading note, and the tonality of the chorus is centred on the leading note rather than the tonic.
In Song 34 the G-mode is announced by the solo call, and the chorus response again dwells on the tonality of the leading note (see Figure 9). One is tempted to say that the tonality of Songs 25 and 34 and some others is a transformation of the rule that the final of the mode must be established in the chorus response, in which, by a process of inversion, the leading note falls a whole tone below the tonic, and chord strengths are likewise inverted so that the roots lie a fourth, and not a fifth, above their companions. This seems an unnecessarily complex explanation of a situation that could be better understood by modifying the rule that the final of the mode must appear in the chorus and substituting a rule that is both musically and sociologically more satisfying.
I propose, therefore, the more powerful rule that the moving tonality of call and response may be based on either vocal or instrumental models, on domba-type or tshikona-type patterns. Thus in the vocal model, as in Domba and Songs 16, 22, 41 and 46 (see Figure 7), the call centres around or begins on the leading note, and the response centres around the tonic or final. The instrumental model presupposes the presence of a complete instrument (set of reed-flutes, xylophone, hand.piano, etc.) on which all the music can be performed, or nothing; whereas the vocal model presupposes a variety of social situations in which good soloists may be present or absent, but parts of the music can be performed to satisfaction even if a complete performance is not possible. Thus in the instrumental model, as in tshikona and Songs 25 and 34, the call centres around or begins on the tonic/final/keynote, and the response centres around and/or ends on the leading note. Songs 12, 33 and T1 provide further evidence of the use of an instrumental model in the composition of vocal music. The three songs are tonally and harmonically identical. The minor variations in the melodies of the solo calls are accounted for by differences in the words. A comparison with the transposition of tshikona in Figure 3c shows that they follow exactly two repeats of the tshikona pattern even to the extent that the response melody omits the tones of the chord in which the tritone occurs and repeats the first phrase at a fourth, and not a fifth, below. The first word of Song 33, khofhe (sleep), may have been suggested by the opening of tshikona, which the Venda often describe with the syllables "fhe fhe," referring to the sound of the pipes. Figure 10 shows the relationship of Songs 12, 33 and T1 to tshikona.
Figure 10. The use of an instrumental model for vocal music, showing part of the pattern of tshikona (transposed), with the outline of Songs 12, 33 and T1 superimposed. Solo call sections are marked with tails. Note the omission of tones on the tritone chords and the termination of the sections of the response on the leading note and its companion tone, emphasising the tonic/keynote D.
Song 42 (D-mode) follows exactly the tshikona pattern (see Figure 2). Contrasting tonality is achieved by moving towards the final (D) from its secondary and primary companions (A and G) respectively. The basic tonal pattern is elaborated by passing notes: thus A G F ED / G F E D. It is interesting that both this song and Number 25 (Figure 8), which coincide exactly with the tshikona pattern, are accompanied by speech interludes, which the novices declaim in stilted fashion. By adding one extra tone as harmony, some singers perform Number 42 in the A-mode. Other songs with markedly descending melodies which follow the instrumental model are Numbers 39 (D-mode), 35 and 55 (G-mode), and 61 (E-mode). The descent of the response from F to G in Song 39 might suggest that the F-mode is being used; but call and response overlap, and the final (D) is emphasised by an emphatic A! in both call and response sections. In Song 55 the descending mode is divided between call and response in the proportion 3:4, exactly as in the tshikona pattern, and in Songs 36 and 61 the response begins again on the tonic (also as in tshikona), but descends more rapidly towards the leading note. Although only five tones are used in Song 61, it is not based on a pentatonic mode: the tones B and G are omitted to accommodate the low speech-tones of dumbula and at the same time satisfy tonal/harmonic requirements. If the melody had been based on a Venda pentatonic mode, the tone G would have been used instead of F. It should now be apparent, as I have said at the beginning of this paper, that the most objective analysis of the tonal organisation of a melody may be achieved only by thorough immersion in its musical system and its cultural antecedents. A most dramatic example of this is provided by Song 53 (see Figure 11), which at first seems to be in the C-mode and to follow the rule of progression from final (C) to leading note (D). However, two difficulties arise: the full C-mode is rare in Venda music because of the tritone chord implicit in its second descending degree, and the pause on E requires explanation. If we compare the melody with tshikona in Figure 10 (above), it fits the pattern note for note, from the third chord of the first bar to the fourth of the second. If we apply to the melody the principle of tshikona tonality, E becomes the leading note and D the tonic/final, and the melody is in the D- and not the C-mode. This is, therefore, a transformation, a retrograde version of the process whereby call and response move away from the leading note and final: in this case both call and response begin with passing notes and move towards the leading note and final respectively. Furthermore, the transformation extends also to the form of the piece: the call is long and the response is short. As if to emphasise further the fact that C is a passing note and D is the tonic, the clap or drumbeat on D is accented and those on C are often omitted.
Song 31 follows exactly the same pattern as 53. The explanation of its failure to descend to the leading note (E), as does Song 53 in Figure 11, is to be found in its different speech-tone pattern: the low-high progression from -mba to la- should at that point be registered in the melody (Blacking 1967b:169). The A is the harmonic equivalent of the leading note (E), and although it is sung by the chorus it belongs musically to the call section. The fact that it is usually sung to ahee reinforces this explanation, and a similar practice is found in other songs: for example, in Song 18 (G-mode) the A on which the response begins coincides with the beginning of the metrical pattern, but it is tonally part of the response, which begins with the words ri yo vhona. Thus Song 18 follows the vocal model, with call on leading note and response on final. Song T3, on the other hand, is in the C-mode, but it omits the B on which the tritone chord would occur (see Figure 4b, d and e). Based on two identical phrases (A G F and E D C), it moves down towards its final. The stress on G in Other songs in which the first tone of the response is the final of the mode, sung to hee, and the words begin one tone below and descend to the final are Number 13 (E-mode), and Numbers 20 and T6, which are really two versions of the same song in the G and E-modes respectively. In all three songs tonality moves to the final, not from the leading note but from the secondary companion of the leading note (C) in Number 13, and from the primary companion of the final in Numbers 20 and T6. The use of G sharp in Song T6 is ornamental, like a passing note, and not an integral part of the mode. The two versions of "Wele wele phosho," heard in different parts of Vendaland, illustrate further how variations in mode are subordinate to uniformity of tonal and harmonic processes. The same applies to the two versions of Song 8 (A- and F-modes), which follow the vocal model of tonality. As a song with a prominent descending tone-row, Number 49 is unique: it has neither a clearly defined tonality nor a predominant mode. It is like an endlessly descending scale, and its most striking feature, and that which gives it unity, is the repetition of the same sequence at pitches alternately one tone above and below each other (see Figure 12).
|
and a preliminary 'call' on the chord C-F emphasise the C-mode. (It could, of course, be written out in the G-mode; but then there would be no explanation for the absence of the second tone, F, as there is for the absence of B when it is written in C).
| Several songs follow the same rules as those with prominent descending modes, but the regulated descent is hidden and we hear only a pattern derived from both it and its harmonic equivalents or one which emphasises the key points in its harmonic progression. This is particularly common when vocables such as ahee are substituted for words, and it follows the development principle by which responses are modified from a word-tied melodic line to a wordless sequence of tones emphasising a harmonic progression. The pattern of Song 37 (G-mode) is a good example of this, and the stressing of certain tones both by duration and meter emphasises the hidden harmonic progression from final to leading note. The stressed E, for example, is the secondary companion of the leading note. Song 24 (G-mode) follows a similar pattern, but the secondary companion (D) of the final is stressed. The patterns of the two melodies are compared in Figure 13. These popular songs are usually performed entirely by the chorus, divided into two sections which are indicated in the figure by open and blacked in notes.
In Songs 37, 24, and 36, unity of form and diversity of tonality are achieved by a device found in many other Venda songs: the repetition of similar melodic pattern, usually within the compass of a fifth or a fourth. This is well illustrated by the outline of Song 47 (E-mode) in Figure 15.
The same process of repeating phrases is found in the similar Songs T5 and T10. It is particularly striking when used with a pentatonic mode whose intervals create harmonic progressions that are notably different from those of heptatonic modes (see Blacking 1967b:168 and 183-86). Figure 16 shows the pattern of two pentatonic songs, Numbers T5 and 10.
Similar effects are achieved in the pentatonic Songs 23, T2 and T4. Song 10 sometimes has a hexatonic tone-row, but it is basically pentatonic and often sung as such. The three-tone songs, Numbers 27, 58, 60 and 7B, are based on the same rules; and although Songs 17, 57 and 59 have only four tones, their underlying harmonic progressions are basically pentatonic. I heard a heptatonic version of Song 17 by a solo singer, but am inclined to think that this was her own idiosyncrasy. She also sang Song 26 (G-mode, instrumental model) as if it were pentatonic. Song 29 is based on a descending pentatonic scale, but the call has an upward thrust because of the speech-tone pattern of its words. The hexatonic Songs 38, 40 and 44 have strongly harmonic sequences which are like those more commonly associated with pentatonic melodies. Song 48 is also hexatonic, but its chord sequences are related to the descending heptatonic melody of Song 16, of which it is in fact an alternative version. Outlines of the tonal and harmonic progressions of these four songs are given in Figure 17.
The remaining ndayo songs follow rules that have already been illustrated. There are, however, two in which a very short response is repeated twice: in Number T8 the same response punctuates two balanced phrases, one rising and the other falling; in Number T7 two phrases fall and rise respectively and are punctuated by responses which mark the contrasting tonalities of the final (G) and its companion tone (C). As indicated in the transcription, some sing the interval of a tritone (C - F#), which to a European-trained ear reinforces the tonality even more. Particularly interesting was the fact that those who did this were always old ladies who had no modern education and little or no exposure to the sound of European music. I have written out a fuller version of Song 19, because it is very well known, and the way in which it is developed antiphonally is unusual in performances of ndayo. It illustrates how the vocables of the response (ahee, hee, uwee, etc.) may be converted into word phrases, just as words are converted into vocables. Song 21 is like the first of the ritual songs (nyimbo) in style, with a word-based melody that rises and falls in a gentle curve. Song 1 is sung when a novice is taken from her home. The firm upward movement of ri Thus Songs 1-4 follow vocal models and are in the A-mode (Song 2 being strongly pentatonic). If it were not for the low speech-tone on the first syllable, the response of Song 3 would almost certainly begin on G and the 'optimistic' fourth would never have been conceived! Similarly, the descending phrase (A G F E) of Ha athu u lila (She has not wept yet) in Song 6 could be described as plaintive and appropriate to the sentiment of the words. But this phrase is common in Venda music and is found in many ndayo, as well as in tshikona itself; furthermore, a similar descending phrase (A G F E D) in Song 4 accompanies the words Thi lali nn The use of vocal models and the influence of the speech-tone patterns of words is more pronounced in the ritual songs than in the ndayo, where the emphasis is on physical movement, and hence on meter and harmony-based melody. But a comparison of melodies with speech-tone patterns shows that they are never mere embellishments of speech - which is what one would expect in view of the Venda distinction between speech and song. The melodies of girls' initiation are noticeably less closely related to patterns of speech-tone than are those of children's songs (Blacking 1967b: 166-71), and one could not even say that the words of their opening phrases have had much influence over the pattern of their melodies, which is one of the rules followed in the children's songs (Blacking 1967b:170-71). The contrast between the observance of speech-tone patterns in the children's songs and the music of initiation may be explained by the fact that the former use simpler tone-rows, and the frequent use of heptatonic modes in the latter makes musical processes the deciding factor in the creation of melodies. In general, speech-tone changes indicate when a melody should move rather than in what direction it should move, or by how many tonal steps. If the musical pattern requires that a melody descend at a certain point, it will do so regardless of the pattern of speech-tones. This rule is followed especially towards the end of word-phrases, which are treated as constant, like a response (Blacking 1967b:167-68 and 199-203). In the 'call' section of word-phrases, changes in melody are made chiefly when low tones occur during the first two syllables or when a low-high pattern replaces a high-high or high-low pattern. This follows the rule that descents in speech-tone and melody need not coincide, but a rise in speech-tone is generally accompanied by a rise in melody (Blacking 1967b:169), or at least some recognisable change. Finally, I cannot yet answer the question of why the Venda choose particular modes for certain songs except to say that the G- and D-modes, which are prominent in tshikona, are the most common, and that I have presented evidence (e.g., Songs 20 and T6, and the two versions of Song 8) to show that the final choice of mode is not as important as the way in which it is adjusted to certain rules of tonal and harmonic progression. Nor can I explain the logic of tonal cohesion in the choice of consecutive ndayo songs, except to say that if two consecutive songs have a different mode and the soloist begins the second with a prominent tone of the first there is often tonal confusion until the new mode has been established. From the evidence presented here it should be clear that the melodies of vhusha and tshikanda cannot be analysed as 'pure melody', unrelated to a harmonic framework or to scales in the sense of a 'fixed store of notes' on which melodies are based (Blacking 1967b:176). If they are classified on the basis of 'objective' melodic criteria, they may be wrongly classified: the only justification for saying that a Venda melody is in such and such a mode is that it follows a vocal or instrumental model and patterns of tonality and harmony which establish certain tones as its tonic, its leading note, and the primary and secondary harmonic companions of these tones. Thus one cannot say that the ritual song of vhusha must be sung in any particular mode, since I have heard different modes used in different parts of Vendaland just as I have heard two modes used for the domba song and differently tuned sets o reed-pipes playing tshikona. But one can say that the ritual song of vhusha is usually in the D- or G-modes and that, like domba, it follows a vocal model; it must always have the same distinctive rhythm and the same tonal relationship between call and response; the response always begins as two related phrases, within the compass of a fourth and a fifth, which establish the moving tonality, and it develops into a khulo style similar to that of domba and related to the structure of tshikona. These are the important structural landmarks in a performance of the ritual vhusha song, and the precise articulation of the melody is of secondary importance. It must not be thought that the Venda consider melodic line to be musically unimportant; on the contrary, repeated patterns of melody are an essential device for balancing the unity of the mass against the diversity of line. But the line of melodies is best understood as a detailed expression of broader tonal structures, which in turn are symbols of the social and cultural life of the Venda. It is immaterial whether or not most of the music of the initiation schools consists of variations on the themes of the national dance (tshikona) and the domba song, which are the most serious items of communal music. What is important is that tshikona, domba, and the music of initiation are all variations arising from the same deep processes of tonal organisation. |
mutei (we seek a novice) expresses determination, which is followed by a slight uncertainty that suggests a search. Even if the melody is not a conscious, metaphorical expression of the social situation, at least it is influenced by the patterns of speech-tone. In my description of vhusha I suggested that the melody of 
ndi ndo
(I do not sleep outside alone), which could hardly be described as plaintive.