• 2 For the purposes of this article and this special issue of Archipel, “Insulindia” is Island Southea (...)

1In this paper I attempt to distinguish the music of “eastern Insulindia” from that of other parts of Insulindia.2 Essentially this is an inquiry into certain musical features that are found in eastern Insulindia, together with a survey of where else in Insulindia they are or are not found. It is thus a distribution study, in line with others that have looked at the distribution of musical elements in Indonesia (Kunst 1939), the Philippines (Maceda 1998), Oceania (McLean 1979, 1994, 2014), and the region peripheral to the South China Sea (Revel 2013). With the exception of McLean, these studies have focused exclusively on material culture, namely musical instruments, tracing their geographical distribution and the vernacular terms associated with them. The aim has been to reveal cultural continuities and discontinuities and propose hypotheses about prehistoric settlement and culture contact in Insulindia and Oceania.

• 3 Triple meter and the “tritone melos.” See the discussion under “tritones,” below.

3Except for Kunst 1939, the studies of distribution named above have avoided Insulindia, focusing instead on regions to the north (Revel, Maceda) and east (McLean). Furthermore, the emphasis on instruments has drawn attention away from vocal music, which is necessarily my focus since I am concerned with eastern Insulindia, where the principal forms of music-making are vocal. Kunst does discuss vocal music (1939, 1942), but he looks at the distribution of only two traits, which he sees as co-occurring indexes of megalithic culture.3 McLean also considers vocal music, but with regard exclusively to the presence or absence of polyphony. In his statistical clustering study (McLean 1979), he speaks simply of “polyphony,” undifferentiated, as one of the traits he tracks. In the more expansive book on the peopling of Polynesia (McLean 2014), he devotes chapter 6 to polyphony but recognizes only two kinds: melody-with-drone and parallelism. In contrast, Simha Arom et al. offer an eight-fold taxonomy of polyphonic techniques from all over the world, remarking in closing that

• 4 Thus, music in European harmonized idioms or in Middle-Eastern idioms is excluded from the traditio (...)

8Now, which music of NTT/TL am I trying to distinguish? I will not deal here with the pop music, Catholic and Protestant church music, or explicitly Muslim music of NTT/TL, since these do not, I believe, differ markedly from their counterparts in other parts of Insulindia. Instead, the music I attempt to distinguish from others is the traditional music of NTT/TL, and the musics I attempt to distinguish it from are traditional musics of other regions. I use traditional in a specific sense, to designate music showing little or no obvious influence from outside Insulindia in its compositional forms or musical idiom.4 In the context of NTT/TL, if we take out of consideration the forms of popular and church music that my usage of traditional excludes, we are left almost entirely with village music. This is music associated with the concerns of rural life: the agricultural calendar, domestic events such as weddings and funerals, and communal work, celebrations, and rituals. Of course, urban popular music is well known in the villages, disseminated by radio, television, cassettes, VCDs, and even cell phones. The point is not that one does not find popular music in the villages, but that one does not find traditional music in the cities.

• 5 These observations are particularly true for Maluku. Twenty years ago, Margaret Kartomi published a (...)

11Here, then, are anomalous features of music in NTT/TL, along with indications of where else in Insulindia counterparts are known. For every feature I discuss I direct readers to audio examples, indicated by boldface numbers (mainly in the footnotes) that refer to entries in the discographical list at the end of the article. I need to make it clear, though, that this is an interim report. It is based on what I dare to say is a near-exhaustive examination of published recordings, plus a diligent search of field recordings, my own and others’—but in field recordings I can hardly hope to be comprehensive, as there are collections I have not discovered and researchers I do not know. And besides, there are hundreds of ethnic groups and thousands of places in Insulindia whose music has never been reported, and many whose music may have been lost before anyone recorded it.5 The generalizations I make here—“such-and-such a trait is found in island A of region B”—do not mean that the trait is not found in island C, elsewhere in the same region, only that it has not been reported from C. The generalizations are offered in the hopes of eliciting both supplementary examples and counterexamples. One final caveat: this article registers traits on the basis of their presence in a region. One recording of, say, triple meter in island D is enough for me to list it as a feature of D’s music. Ideally, however, I would be able to discuss the relative importance, either statistical or conceptual, of these traits—is triple meter typical there, or is it a fluke that just happened to come to my attention?—but my data do not yet permit this.

• 6 For precise definitions of these technical terms, see Arom et al., 2007. I provide loose working de (...)

12Antiphony. Antiphonal songs are often heard in communal dances and rituals and in cooperative labor. They are sung by two or more choruses in turn, answering each other. The singing group may be divided into male and female choruses (this is the most common format); or a single-sex or mixed chorus may be divided into sub-groups. Choruses typically sing in heterophonic unison, with or without occasional harmonies, or they may sing in pervasive harmony (homorhythm).6 Sometimes one chorus sings in one of these textures and another in another: in Timor, for example, men’s choruses in antiphonal singing are often heterophonic, while women’s choruses tend to be less heterophonic and often have harmony.

• 7 When the opposition is between a small group and a chorus, the distinction between antiphonal and r (...)

13Antiphonal should be distinguished from responsorial. In responsorial singing a chorus responds to one or more soloists; in antiphony a chorus responds to a chorus.7 The difference is not essentially musical: it lies rather in the social relations embodied in performance—the complementarity of antiphonal singing in contrast to the leader/follower or individual/group relation of responsorial singing. In antiphony, typically, the two groups have approximately equal roles and equal weight in the performance. One side does not lead or dominate the other. The songs often have no soloist, or a soloist with only a minimal function as precentor, getting a song started or cueing new words to be sung by the chorus. As soon as the chorus catches the new tune or new text, it takes over, and the soloist merges with the group.

• 8 Some definitions restrict antiphony to the direct repetition by one chorus of what the other chorus (...)
• 9 Timor: NT5 trs 10, 11, 12 and T3 _001_02 (Bunaq); T1 tr 19 (Mambai); T2 trs A1, A3, A4 (Ema); T5 tr (...)
• 10 Pantar: NT2, Bunni Maggar 1.
• 11 Aru: M3 tr 4 (11:51). Sumbawa: IN5 vol 1 tr 5. Toraja: SL2 (Rising sun: bugi’), and SL3 tr 5; NTT ( (...)

14Two types of antiphonal singing are found in eastern Insulindia. The opposed choruses may sing contrasting melodies or melodic phrases (as in Fig. 1), or they may alternate in singing the same melody.8 (Both procedures are also found in responsorial singing.) In the first type, which I call contrastive antiphony, the contrast between the melodies is sometimes on the order of “antecedent” and “consequent” phrases in the Euro-American idiom (as in Fig. 1). Contrastive antiphony is widespread in Timor and in the western region of Flores (Manggarai).9 It would not be surprising to find it elsewhere in NTT/TL as well, but so far I have heard it only from Pantar.10 I have not found contrastive antiphony outside NTT/TL. Conversely, the second type, which I call iterative antiphony, does appear outside NTT/TL—in Aru (Maluku), Sumbawa, and among Toraja in the Sulawesi highlands—but I know of only one brief instance of it in NTT/TL.11

• 12 Virginia Gorlinski confirms (p. c., 2015) that she too did not encounter antiphonal singing during (...)
• 13 Th. Pigeaud, in his exhaustive 1938 work on Javanese folk performances, remarks that “round dances (...)

15I have not encountered antiphonal group singing of either type in the central and western islands (Borneo, Java, Sumatra), and for those regions it is neither reported in the literature nor documented in published recordings or field recordings I know of.12 This may be more an indication of what ethnomusicologists tend to be interested in (instruments, professionalized performances) than of what is (or was) out there.13 A natural context for antiphonal songs is communal dance, involving large groups of men and women, and I find it hard to believe that in agricultural villages west of Sulawesi there are no harvest celebrations or other occasions for the whole community to dance and sing together. These communities may, however, find responsorial singing, where the whole community responds together to a soloist, more aesthetically or culturally satisfying than singing that depicts the complementary opposition of groups within the society. It may also be the case, at least in Java and Bali, that the taste for gamelan music and for professional or highly rehearsed performances—or now for urban popular music—has reduced the appeal of group singing to the point where people no longer bother to do it.

• 14 I take this term from Jack Body’s album notes for his excellent jemblung CD, IN1.
• 15 T2 (Ema) trs A1, A3, A4 (in all of these the female chorus is more restrained than the male); T3 it (...)

16Two types of heterophony have already been mentioned: “loose unison” or unisonal heterophony—all voices approximating unison but not achieving it precisely (and probably not seeking to)—and sporadic harmony, the occasional sounding of optional harmonies in an essentially heterophonic texture. In Timor, however, there is also a kind of singing where the apparent aim is to sing neither in approximate unison, nor in occasional harmony, but rather to sing explosively, chaotically, going every which way but ending up in the same place.15

• 16 IN1 trs 7 (Madura), 8 (Pekalongan).
• 17 IN6 vol 3 tr 15 (dikir baru).

17This kind of singing has also been recorded in village Java and Madura.16 One could see these latter recordings as in part humorous vocal imitations of gamelan music, but there are also passages of wild heterophony quite different from the regulated interplay of parts in the Javanese gamelan. Another example, featuring only two singers and sitting halfway between exuberant heterophony and the duet singing of Minangkabau salawat dulang (described below), is heard in Lampung.17 The Javanese, Madurese, and Sumatran instances are all found in the context of Islam-oriented performance, but the Timorese instances are not.

• 18 The force of “traditional” in this formulation is to exclude from the discussion singing in Europea (...)
• 19 See Yampolsky 2001a (or its online posting) for description of these complexes. Readers familiar wi (...)

18I wish to propose as a generalization that the predominant mode of traditional group singing throughout Island Southeast Asia is in unison, whether tight or heterophonic.18 Often, but not always, it is responsorial. If we allow for heterophonic approximation of unison and also for limited non-unison overlap of soloist and chorus, virtually all group singing that falls into the “gamelan and wayang complex,” the “Muslim music complex,” and the “Melayu music complex” is unisonal.19

• 20 Arom et al. 2007 define polyphony as plurilineality: anything that is not in unison or octaves (see (...)

19If this generalization is accepted, then all traditional forms of non-unisonal group singing are exceptional for Insulindia, and their comparative distributions in NTT/TL and outside that area are of interest for that reason. There are several varieties: the chorus may be divided into segments with a polyphonic relation to each other,20 or the chorus may be unisonal and the soloist(s) stand in polyphonic relation to it, or polyphony may obtain in both aspects.

• 21 F7 tr 3 (Nage). Other complex singing groups from central Flores are heard in F7 trs 1 and 2 (Nage) (...)
• 22 F6 tr 5 (Sikka).
• 23 F5 trs 1 and 3, F6 tr 2 (all Sikka).

20Flores offers impressive examples, as in Fig. 2, which shows a brief excerpt from the middle of a song from a Nage village near Boawae, in Central Flores. A melody is sung in octaves by one part of the chorus (stems down, lower staff), joined by the soloist (stems down, upper staff); another singer (stems up, upper staff) sings in parallel seconds above the soloist; another few singers hold a shifting drone above the chorus, in the soloist’s range (sustained tones at top of lower staff).21 An example from a Sikka village near Maumere begins with two soloists and a homophonic chorus in mixed (non-parallel) intervals; but halfway through the piece, a breakaway segment of the chorus emerges, singing in hocket, each voice on its own pitch. The hocket has the effect of a tone-cluster repeated as a drone, and the main chorus also sings a syllabic drone.22 This hocket device occurs in other songs from Sikka as well.23

21

• 24 In addition to the drone shown in Fig. 2 (an excerpt from the second song in F7 tr 3), drones are a (...)
• 25 NT1 tr 11 (Rote).
• 26 M1 item _005_05 (Tanimbar).
• 27 SL1 (Toraja) trs 3, 4, 11–16. Note that these drones (and also some of those from Flores heard on F (...)

22Hocket and drones are unusual for vocal music even in NTT/TL, let alone in the rest of Insulindia. Drones figure in some of the duet singing discussed below for eastern Flores and eastern Timor, but aside from the duets, vocal drones are rare in Timor and, while common in central Flores, are not found in Manggarai, the western end of the island.24 There is an instance of syllabic drone from Rote.25 Outside NTT/TL, there is syllabic drone in a song recorded in Tanimbar (Maluku).26 Drones are also found in Toraja music, in various forms: as a reciting tone (approached and left via other tones), syllabic, or sustained; often the entire chorus holds the drone, and a soloist weaves around it.27 Elsewhere in Insulindia drones occur only in instrumental music (as on double and paired flutes in Flores, Timor, and Toraja; on lutes, such as the sampeq and gambus, with open strings sounded repeatedly; also on the sasando tube-zither). Hocket, in the form of interlocking and overlapping melodic-rhythmic lines played, often with some improvisation, on multiple instruments, each contributing only one or a few tones, occurs in gong ensembles where gongs are distributed among several players (as is common in NTT/TL; see Part 3 below). Extended melody, produced by paired lines interlocking in composed hocket (kotekan), is a crucial device in Balinese instrumental music; and simpler forms of unimprovised hocket occur also in Javanese gamelan music (imbal techniques). But in vocal music outside NTT/TL hocket occurs only in the unpitched vocables of Balinese sanghyang and its derivatives (kecak, janger).

• 28 Timor: T1 tr 19 (Mambai); T3 items _001_03 and _04 (Bunaq); T9 trs 9, 15 (Biboki); NT5 trs 8 (Tetun (...)
• 29 T9 tr 17 (Biboki). During this recording session, the same singers sang this song in two other ways (...)
• 30 F3 >9:45 (Lio); F6 trs 12–14 (Toto region, central Flores; population originally from Lio).
• 31 F6 trs 5 (Sikka) and 12, 13 (Toto); F7 tr 5 (Nage).
• 32 Aru: M3. Kei: M4 tr 12 (ngel-ngel, thirds). North Sulawesi (Minahasa): SL3 tr 10 and 12 (Tombulu ma (...)
• 33 Borneo: B3 trs 1, 9 and B7 tr 1 (Kenyah). Sulawesi: SL3 trs 9, 11 (mazani, Tombulu).

23Occasional homorhythm is common in NTT/TL,28 as is consistent homorhythm, though neither is as widespread as unisonal heterophony. The predominant (though not exclusive) form of consistent homorhythm in NTT/TL is parallelism: we have already seen (Fig. 2) a melody rendered in parallel seconds while embedded in a larger choral context of drone and octave doubling. Lala Ina Rou is a song from Biboki in Indonesian Timor, whose chorus is sung in parallel fourths throughout.29 In general, the most common parallel interval is the third, possibly reflecting church-music influence, but the full course of tonal harmony is often absent.30 Homorhythm in assorted (i.e. non-parallel) intervals occurs in some songs of central Flores and in passages in the serial duets of eastern Timor and eastern Flores (discussed below).31 Outside NTT/TL, there is occasional homorhythm in Aru (Maluku) and parallelism in Kei (Maluku), North and South Sulawesi, and West Kalimantan.32 Kenyah in Borneo and Sarawak sing in mixed intervals; some mazani songs in North Sulawesi are in mixed intervals, with repeated cadences on a minor seventh (Fig. 3).33

• 34 M1 item _001_10 (Tanimbar). The transcription in Fig. 4 has been transposed down a half-step for co (...)

24In Tanimbar (Maluku) there are songs with a narrow tonal range where each pitch has an assigned harmony, as shown in Fig. 4.34 This is again an instance of homorhythm in assorted intervals. The melodic phrases have a repeated descending contour, Bb-G-F, finishing on F. The harmony (below the melody) uses only two tones, G (as harmony for Bb or as unison with G in the melody) and Eb (as harmony for G and F). Thus while there are frequent harmony thirds, the cadences are always on a second (F-Eb). Like the mazani songs cadencing on sevenths, this is thirds-based harmony with a twist.

• 35 Arom et al. divide homorhythm into two categories only: parallel and divergent (2007:1094).

25The two voices in Fig. 4 move up and down together. (Please forgive the almost inescapable spatial imagery.) Such motion is sometimes described as “parallel,” but the term is confusing, since the intervals themselves are not the same throughout. I offer the term similar motion for this behavior, reserving parallel motion for the case where a constant interval is maintained (as in Lala Ina Rou). Both similar and parallel motion contrast with divergent motion, where one part or line in a chorus goes up and another simultaneously goes down (contrary motion), or one part stays steady while the other goes up or down (oblique motion).35 The reason for emphasizing these distinctions is that divergent motion, unlike parallel or similar motion, gives a strong impression of contrasting, equally weighted parts occurring simultaneously.

• 36 Tetun: NT5 tr 9. Lio: F3 >24:15. Oblique motion can also be heard in some cadences of F7 tr 4 (Nage (...)
• 37 This would seem to be by definition heterophony, yet it is also consistently homorhythmic. I do not (...)

26Fig. 5 gives instances of both contrary and oblique motion in a Tetun-language song from Indonesian Timor; and Fig. 6 shows a brief instance of oblique motion from a homorhythmic song from Lio in Flores.36 An extreme form of divergent motion is found in Borneo, in duets for the sape’ (or sampeq) plucked lute. Two players play the same melody, each on his own instrument, one in the upper octave and one in the lower; they articulate the same pulse and play on nearly every beat, so they are almost entirely in homorhythm, but each musician plays his own melodic variations on the melody, causing frequent contrary motion.37 One clearly hears two contrasting melodic entities at once, and the music thus approaches the category of counterpoint.

• 38 Arom, p. c., 2015. In that case, I would propose also homorhythm with contrapuntal tendencies, whic (...)
• 39 F7 tr 8, recorded 1993. The description here is based on this 1993 recording. F1a item _003_02 is a (...)

28Arom allows that there may also be “counterpoint tending toward homorhythm” (contrepoint à tendance homorhythmique).38 The Ngada song Goegoe fits this description.39 A two-pitch ostinato melody is held by men in a low register and women an octave above them. Two small groups of female soloists maintain different, alternating ostinati in the octave above the bass ostinato, but they are not rhythmically independent from it: instead they subdivide some of its held pitches and merge with the bass for others. Then there are male soloists, and people just calling out in rhythm, and a jingling bell. This complex texture has elements of both homorhythm and counterpoint.

• 40 NT3 tr 16 (Takpala, Alor). Also F3 >8:00 and >9:45 (Lio), and F6 tr 1 3:24 (Sikka). The Sikka soloi (...)
• 41 SM5 trs 3, 4, 5 (Hilisimaetanö, Nias).

29Clearer instances of counterpoint occur in musics where a soloist sings floridly over a unison chorus.40 Often the soloist merges melodically and rhythmically with the chorus and then separates from it. One is always aware of two lines pulling against each other. This soloistic counterpoint, which, when it soars above the chorus, I term a descant, occurs not only in Alor and Flores but also at the other end of Insulindia, in Nias.41 (Nias will come up again in this article.)

• 42 F7 tr 10 (Rongga, Manggarai).
• 43 SL3 tr 6 (Uma-speakers, Pipikoro region, Central Sulawesi).

30Full Aromian counterpoint—two or more rhythmically distinct melodies, performed simultaneously—is rare in Insulindia. In my view, some of the serial duets from NTT/TL offer examples, and there is another from Manggarai (Fig. 7).42 Outside NTT/TL, there is an example from upland Central Sulawesi.43 It is possible, however, that all of these examples have less rhythmic independence than is called for in Arom’s definition, in which case they would all fall under the heading of counterpoint tending toward homorhythm. Finally, there is flowing, improvisatory, loosely imitative counterpoint at the opening of Minangkabau salawat dulang duets (discussed below).

• 44 I call the duets serial because they are sung in strophic alternation by two or more pairs of singe (...)

31Serial duets are found in two locations—and apparently only two—in the whole of Insulindia: among western Lamaholot-speakers (and some Sikka-speakers) at the eastern end of Flores and in the neighboring western region of Solor; and among Fataluku-speakers at the eastern end of Timor-Leste. Surrounded by music cultures devoted to the large unison or polyphonic choruses I have been describing so far, these enclaves of duet-singing constitute, as Dana Rappoport has written (2010), an enigma. She suggests (2015) that their presence in Flores is the result of migration from Timor, but at present this is conjectural, and it does not explain why the duets are so isolated and atypical in both locations.

• 45 “Sought” here alludes to the phenomenon of “seeking seconds” that has been identified as a characte (...)

32A further stylistic puzzle is that both duet traditions feature an abundance of major- and minor-second intervals between the voices, sometimes joined in parallel strings and sometimes “sought.”45

• 46 Dana Rappoport has published recordings of the Flores styles on F4, and a few are heard on my F6. I (...)
• 47 An excerpt from F4 tr 6 is transcribed in Rappoport 2010: 243, and one from F4 tr 7 in Rappoport 20 (...)

33A crucial characteristic of these duets is the presence of two distinct voices, and therefore passages in unison or unisonal heterophony are usually absent. (The voices may converge on unisons, but then they diverge again.) The most common texture is homorhythmic, but some repertoires or local styles in both regions seem to favor more counterpoint, while in others one voice mainly holds a syllabic drone under the melody.46 The degree to which the Flores and East Timor duets differ in their polyphonic practices remains to be studied. Fig. 8 is a transcription of one strophe of a vaihoho (the term for the duets in Fataluku). Transcriptions of Lamaholot duets appear in some of Rappoport’s publications.47

• 48 PNG1, PNG3.

34The duet style of western Lamaholot- and Fataluku-speakers is so unusual that I will go outside the designated bounds of this article and mention that duets with an abundance of seconds are also found in some of the Admiralty Islands in the Islands Region of Papua New Guinea.48 These duets are apparently not serial, and they are more exclusively homorhythmic, featuring parallel descent through chromatic steps. I should point out that Gerald Messner (1981:439) maintains that the simultaneous intervals are not tempered seconds (100 or 200 cents) but a “very specific interval of 160 cents.”

• 49 SM2 disc 1 tr 17, SM3 tr 6, SM8 tr 12.

36I know of one other duet tradition (not serial duets) in Insulindia: the salawat dulang of Minangkabau in West Sumatra, which is a medium for Muslim tenets and precepts.49 In the unmetered introduction, the two singers sing in overlapping alternation that can become (when the overlapping is protracted) free-meter, improvised, loosely imitative counterpoint. Later, a steady beat is introduced and the contrapuntal component recedes: the pattern becomes alternating solos that join in heterophony and then split again into alternating solos. Drones are not present. Although in texture the introductory sections of salawat dulang resemble the duet traditions in NTT/TL, their sound is quite different: the Sumatran form’s recurrent pattern of alternation then heterophonic fusion of the voices is distinctive, as is the headlong linear quality caused in part by the non-strophic character of the melody. The simultaneous intervals in salawat dulang seem accidental conjunctions of independent lines, whereas in the more tightly coordinated NTT/TL duets they arise from the systematic avoidance of heterophony in favor of distinct voices. Moreover, after the introduction the salawat dulang singers move into a second phase in which they sing popular tunes in heterophonic unison (substituting lyrics with Muslim content). These melodies would be idiomatically impossible in the NTT/TL duets.

• 50 Timor: NT5 (Tetun) tr 9 (see Fig. 5); NT5 (Bunaq) trs 10 & 12; T1 (Mambai) tr 19; T3 item _001_03 ( (...)
• 51 North Sulawesi: SL3 tr 12, a Tombulu (Minahasa) mazani song (though this may be the result of churc (...)
• 52 When I describe gamelan music, I am referring to the traditional repertoire, not to modern, deliber (...)
• 53 SM7 trs 14 (Riau) and 17 (North Sumatra). See David Goldsworthy’s discussion (1979:319–51). Lagu du (...)
• 54 Nias: SM1 tr 16; SM5 tr 6, second song (>2:15). Marlene Patton analyzes a number of other hoho in h (...)

38Triple meter. Triple meter (or one could consider it triple subdivision of a steady beat organized in longer cycles) is common in dance songs in Timor and parts of Flores.50 It is also found in North Sulawesi, South Sulawesi, and West Kalimantan.51 On the other hand, it is extremely rare in Java and Bali, where virtually everything is subdivided in twos, fours, eights, etc.52 In Sumatra triple meter occurs in one category of Melayu dance tunes, the lagu dua repertoire, but it seems likely these tunes are European in origin.53 Leaving that repertoire aside, triple meter is as rare in Sumatra as it is in Java and Bali. But, intriguingly, it surfaces again in hoho songs in Nias, and in percussion music in Siberut (Mentawai).54

• 55 The music called senggayung (B7 trs 5-8), played by people along the Upper Jelai River in the inter (...)
• 56 Sikka: F5 trs 4 & 7 (3+2+2), tr 1 (3+4); F6 tr 3 >7:31 (3+2+2). Larantuka, East Flores: F1b item _0 (...)
• 57 Sumba: IN8 tr 11. Ataúro: T5 tr 36 from 3:02 (but by 5:30 it seems to have slid into a straight dup (...)

39Complex meters. There are other non-duple meters in NTT/TL, besides triple meter. I refer here only to meters that pervade a piece or a substantial section from beginning to end.55 The meter of the dolo-dolo and other dances in eastern Flores is a seven (3+2+2), first slow, then fast. Played fast, the meter sounds deceptively duple, but tools to slow down the sound confirm it is still a seven. Sevens are also found in Manggarai, at the western end of Flores.56 Sevens are common in Sumba, and a boat-building song from Ataúro (an island off Dili in Timor-Leste) is partly in seven.57

• 58 Benuaq 14 beat ostinato: B7 tr 13. Tunjung 13-beat ostinato: B2 tr 17. Benuaq 30-beat ostinato: B2 (...)

40Such meters are known in Kalimantan, but they have not been reported for Sulawesi or Maluku—nor, strikingly, from Java, Bali, or Sumatra. In East Kalimantan, Benuaq play a ritual curing music on gong ensemble, with a 14-beat ostinato, 2+2+2+2+3+3, and Tunjung play a 13-beat ostinato, 3+3+2+5. Another Benuaq ostinato, 2+2+2+2+3+4+3+4+2+2+4, adds up to 30. In West Kalimantan, Ot Danum female shamans (jaja’) sing some of their songs (timang) in a meter of five, and Kanayatn have a gong melody in seven.58

• 59 In Javanese music, these units are called gatra. “The gatra is the smallest melodic/rhythmic unit o (...)

41“Non-four” cycles. I use this inelegant term for strophes or melodic cycles that—unlike most melodies of traditional Javanese and Balinese gamelan music—are not built out of a succession of four-beat phrases or units.59 Four-based cycles, as well as unmetered and variable ones, are also common in NTT/TL, of course, but it is the “non-fours” that are noteworthy here. Examples abound in Timor and Flores. The distribution of non-fours outside of NTT/TL has not been well studied. I have found them only in North Sulawesi and Nias. Various non-four cycles, each constituting the total length of one statement of a melody repeated multiple times in performance, are shown in Fig. 9.

• 60 The three videos from Timor discussed in this paragraph are, in order: T4 disc 3, “Tebe Dai Tilomar (...)

42Phase shifts, and discoordination of beat and melody. I wish to offer this feature tentatively, but still to offer it. I have noticed a number of instances where the “external” rhythmic organization—marked on a drum, or by dancers’ feet—is out of phase with the organization of the melody itself. For example, in one video of a Timorese dance, the dance is in a four-step, four-beat pattern, but the melody is 19 beats long, so it has four different registrations vis-à-vis the dance step. In another video, a 22-beat melody is sung to an eight-beat dance step; again it has four registrations. In a third, the melodic cycle is either 32 or 36 beats, depending on whether the boys’ chorus sings its part as 10+8 or 14+8. (The girls’ chorus is always 14 beats.) The dance step covers eight beats, so it either comes out exactly with the chorus (four dance cycles to one melodic cycle) or four beats off. One might think that in the latter case the singers would wait four beats and begin the next melodic cycle on the first beat of the dance step, but this does not happen: they begin the next melodic cycle as soon as the previous one finishes, even though it is in the middle of the dance step. Toward the end of the video, the boys falter and drop two beats before taking up their chorus. This puts the start of the melody on beat 6 of the dance step, three-quarters out of phase with the dance cycle, but again the singing and dancing continue regardless.60

• 61 Lembata: unpublished video by Dana Rappoport, made in Ds. Balurebong on 1 July 2007.

44In a video of a beku dance in Lembata,61 the vocal does not have a fixed period: the chorus is always 28 beats, but the solo can be of variable length. The dance step, on the other hand, is exactly six beats long. In the video excerpt, successive statements of the chorus can be seen to start at three different positions in the six-beat dance cycle.

• 62 The three examples in this sentence are, in order: B2 tr 2 (Meratus Mountains, South Kalimantan); B (...)
• 63 Neither form of discoordination is found, apparently, in Java, Bali, and Sumatra. In a gamelan or o (...)

45In all of these instances, the dance cycle and the melodic cycle share the same pulse or beat, but the metric cycles are out of phase. A more extreme form of discoordination shows no correlation between the melodic cycle and an external beat (as provided either by the dance step or by an instrument). This type does not, so far as I know, exist in NTT/TL, but it is found in a few spots elsewhere in Insulindia. There is an apparent example from South Kalimantan; a clear one from Sabah (Malaysian Borneo); and I encountered it when recording ngel-ngel songs in Kei (Maluku).62 A group of women sang a solo-chorus song, and one of the women played a pattern on a hand drum, with no metrical connection to the singing. The woman seemed to be drumming at random, as though half asleep. I feared she was simply unable to keep a beat and the other women were unwilling to embarrass her by taking the drum away. But when I played back the recording the next day for the singers and offered to record again if they weren’t satisfied, they said there was nothing wrong, no reason to do a retake. Later I found that the Catholic missionary Geurtjens had noted seventy years earlier that the drumming for ngel-ngel was unpredictable: there might be a pause of nearly half a minute between one pair of drumstrokes and the next (Geurtjens 1921: 397).63

• 64 Album notes to PNG2. The instances Laade speaks of are heard in PNG2 trs A11 and A12, respectively. (...)

46Although New Guinea is officially outside the purview of this article, I will mention that I also encountered some discoordination of drumming and singing in Biak, and Wolfgang Laade noted it in badra songs on the south coast of mainland Papua New Guinea. Regarding this as a desired effect in the music, he describes it as “a real ‘swing’... created by the drum beat being slightly faster than the metre of the song... In [one] instance 26 drum beats stand against 20 of the song, in [another] 28 against 20.”64

• 65 For my purposes here I will define a second as any interval significantly larger than a unison and (...)
• 66 B7 trs 17–24 (timang, Ot Danum).

47Simultaneous seconds.65 I have already mentioned singing in seconds in NTT/TL—in the Nage/Ngada choral music and the East Flores and East Timor duets—and again in Sulawesi among Toraja. Seconds are also found in West Kalimantan, in the timang of the Ot Danum, where they are heard as a harmony to only one pitch.66 We saw a similarly systematic application of the second as a harmony in Tanimbar (Fig. 4), as well as a seventh (the second’s inversion) in mazani (North Sulawesi; Fig. 3). In other instances (central Flores, Toraja) the second seems to be less rigorously applied but to represent a desired, sought-after sonority.

• 67 SM2 disc 1 tr 18 (salawat dulang, Minangkabau). In this recording, two singers sing the same melody (...)

48One does not find simultaneous seconds in Java or Bali—nor, typically, in Sumatra, though there is one surprising exception in a salawat dulang recording.67 Aside from this instance, I have not found seconds in Sumatra.

• 68 Kunst 1942: 35-37; cf. Kunst 1939: 7-9. Although the principal field research for both Music in Nia (...)

49Tritones. In the 1930s, Jaap Kunst observed that in two widely separated areas, the Lio region of central Flores and the southern part of Nias, there were melodies exhibiting what he called the “tritone melos”—that is, their scales permitted melodic emphasis of the interval of the tritone (an interval of six semitones, for example C to F#). Moreover, in both places these tritone melodies were sung in triple meter or ternary subdivision. On the basis of ethnographic similarities between these two cultures and a third, that of the Angami Naga in Northeast India, who also had (according to a musical transcription by Marius Schneider) tritones and triple meter in the same song, Kunst proposed that the inhabitants of all three regions were “bearers of a megalithic culture, which has remained alive to this day.” In view of all this, he wrote, “I am inclined to assume hypothetically that the tritone-melodies combined with a ternary rhythm are characteristic of the megalithic culture.”68

• 69 Tritones and triple meter together: F3 >38:10 (Lio), SM5 tr 4 (Nias). Tritones without triple meter (...)
• 70 Ataúro: T4 disc 1 title 14 (Makadade), T5 tr 36, T7 #100716. Manggarai: F2 tr 17. Sumba: NT6 #SB 97 (...)

50Tritone melodies do indeed still exist in Nias and Lio, and so does triple meter, and they do sometimes occur in the same song, but they can also occur independent of each other.69 Tritone melodies are also encountered in Ataúro, in Manggarai, and in Sumba (another place of large stones)—but those examples are in duple meter or free meter, not in triple.70

• 71 SL1 tr 20 is a Toraja flute melody using the tritone.

51Kunst was a great scholar, the first musicologist with the vision to take on Indonesia as a whole, but I believe he was overly fond of the notion of musical survivals over millennia—not only this megalithic one, but also his famous hypothesis (1960 [1954]) of a historical link between eastern Flores and the Balkans. Nevertheless, despite the dubious historical proposals, the correspondence between Nias and NTT/TL—like that between the duets in eastern Flores (and Timor-Leste, which he did not know about) and Bulgaria—is striking. Tritone melodies are rare outside of NTT/TL, until one comes to Nias in the far west;71 non-four structures drop out after Sulawesi and resurface in Nias; descant drops out even sooner (Alor) but reemerges in Nias; and triple meter is more prominent in Nias than anywhere else outside of NTT/TL.

• 72 The ideas and some of the wording in this paragraph are borrowed from Yampolsky 2001a, and in the n (...)

52Gong-chime ensembles. I have argued elsewhere that an analytical distinction should be made between gamelan and other types of gong-chime ensembles.72 The term gamelan, I propose, should be reserved (analytically—I am not trying to intervene in popular speech, which calls any ensemble with a gong or a metallophone a gamelan) for the orchestras of Java and Bali that have (a) melodic metallophones (whether with keys or kettles), (b) punctuating gongs or their substitutes, and (c) at least two simultaneous melodic lines, related in content but of contrasting musical character. The relation between the melodic lines may be, e.g., one of abstraction, or paraphrase, or elaboration. Such orchestras have developed large and structurally variegated repertories, along with extensive technical and analytical vocabularies. It is therefore appropriate to distinguish them terminologically from other ensembles that share their gongs but few or none of their other traits. The term gamelan comes instantly to mind for the former group. But if they are gamelan, then the others are something else, namely (I suggest) gong ensembles.

• 73 These are found in several parts of northern and eastern Borneo, and also in Lembata.
• 74 Indeed, the gamelan of Java and Bali and their outposts in South Kalimantan, southern Sumatra, and (...)
• 75 F1b items _001_104, from Lembata, and _001_357, from eastern Flores, are exceptions. T6 tr 4 may be (...)
• 76 The gongs may also be laid in a rack, thus resembling the gong row of melodic gong ensembles. Howev (...)
• 77 Timor: NT5 tr 7 and T8 trs 2, 4, 5, 14 (Meto); T6 tr 9 (Tetun?); T8 trs 9, 10 (Amarasi); T8 tr 18 ( (...)
• 78 North Maluku: M2 disc 2, trs 15, 16 (cikamomo from Ternate). Borneo: B4 tr II-2 (Modang, East Kalim (...)
• 79 SM8 trs 1, 2 (talempong pacik, Minangkabau, West Sumatra). Pacik means “hand-held.”

53The vast category of (non-gamelan) gong ensembles in Insulindia can be broken down into two principal subcategories. Some present a melody up front, usually played on a row of small gong-kettles and rhythmically supported by drums and other, larger gongs. (Some ensembles use a set of hanging gongs, instead of a gong-row, for the foregrounded melody.)73 These may be termed “melodic” ensembles, and outside NTT/TL they predominate.74 They are, however, rare in NTT/TL.75 Instead one finds gong ensembles of the other type, which I call (for want of a better term) “rhythmic” ensembles, and these are, conversely, rare (though not unknown) outside NTT/TL. Rhythmic ensembles play composite repeating patterns that are as much rhythmic as melodic, with little sense of contrast between foreground and background or between melody and support. Typically, from five to ten bossed gongs are distributed among several players, so that each player controls one, two, or three gongs. The gongs may be suspended from a cross-bar or held by the player; if the gongs are small and light a player may hold two in one hand and strike them with a stick held in the other hand.76 The texture of the ensembles ranges from homorhythmic to contrapuntal. Many of the NTT/TL ensembles tend toward the contrapuntal: each player, using the tones of his or her few gongs, contributes a rhythmic-melodic ostinato that interlocks and overlaps with the patterns of the others.77 Some of the players may make variations or improvisations on their patterns. In Maluku and Borneo, where rhythmic ensembles are less widespread, there seems to be a tendency toward homorhythm, one part subdividing the rhythm of another or matching the rhythm but playing different pitches.78 The talempong pacik non-melodic ensembles in West Sumatra are again contrapuntal.79

• 80 IN2 trs 1, 2 (Bali).

54The Balinese bebonangan or balaganjur resembles a rhythmic ensemble, since it uses hocket and hand-held gongs, but there is no variation in the parts, and the interlocking lines produce a single, foregrounded melody.80

• 81 IN6 vol 2 tr 26 (mislabeled 27 in booklet and track list; Gonakn Sipat, Pompakng).
• 82 Maluku and North Maluku: M2 disc 1, tr 20 and IN4 tr 14 (Ambon); IN4 tr 13 (Banda); M4 tr 7 (Buru). (...)

55Melodic gong ensembles are far more common outside NTT/TL than within. They are found (sometimes along with rhythmic ensembles) in Maluku, North Sulawesi, East and West Kalimantan, North Borneo (Sabah, Sarawak, Brunei), all of southern and western Sumatra, island Riau, Java, and Bali. They range from ensembles like that of the Pompakng in West Kalimantan,81 featuring a melody of only two tones, to extended melodies played on horizontal (or, rarely, vertical) gong-rows, with punctuating gongs and drums.82

• 83 F2 tr 14 and F1a item _004_02 (Manggarai); F1b item _001_357 (eastern Flores).
• 84 SM8 trs 2 (talempong pacik); SM8 trs 3, 4 (talempong Unggan).
• 85 There are all- or mostly female gamelan groups in Central Java and Bali (gamelan ibu-ibu), but thes (...)

56Female performers. The players of rhythmic gong-and-drum ensembles in Timor (aside from the meko ensembles found among Rotenese) are women, and there is also a widespread dance in Timor called likurai or teberai, in which women dance while playing small drums. This association of women with percussion instruments seems to be stronger in Timor than elsewhere in NTT/TL, though it is found in some ensembles in Flores.83 It also appears in scattered locations in other parts of Insulindia. Women commonly play gong rows and hanging gongs in Sarawak and Sabah (Patricia Matusky, p. c., 2015), and Mallinckrodt reported (1925:173) that women in South Kalimantan played the katabong drum, but men played gongs. Margaret Kartomi (1994: 169) reports that in Tanimbar (Maluku) women are the exclusive players of the tibal drum. (A different drum is played exclusively by men.) Among Minangkabau there are all-female groups playing the rhythmic gong ensemble talempong pacik; this may be a recent development, but in the mountain community of Unggan women traditionally play all the instruments in a melodic gong ensemble (also called talempong).84 I have seen a woman playing a gong row in the Natuna islands off western Sarawak, and (again I have to step outside the boundaries of this article) women often play melodic gong rows in the southern Philippines, accompanied by male supporters on other instruments.85

• 86 It is striking that the long line of gong-chime ensembles running from the mainland down to Sumatra (...)

61Another possibility is that the anomalous traits of NTT/TL result from the proximity of the region to New Guinea and the cultures farther to the east. 86 Perhaps NTT/TL’s music represents a confluence of Papuan and Austronesian traits. But then we would expect a higher incidence of those traits in the islands of southern and eastern Maluku, equally susceptible to influence from New Guinea.