2.1.1 ‘pig’

Pigs appear to have moved through Island SE Asia under human agency as husbanded animals, ultimately form a Southeast Asian source. With the exception of Sulawesi, none of the islands east of the Wallace line possessed endemic populations of pig (Sus scrofa, Groves 1981; Glover 1986). In fact, archeological investigations on Flores, Timor, and the northern Mollucas have demonstrated that the first appearance of pigs is associated with the arrival of the ‘Neolithic cultural package’ during the middle to late Holocene (7000–3500 BP) (Larson et al. 2007).

A form possibly related to PMP *babuy ‘pig’ is reconstructable as PTAP *baj ‘pig’ (where /j/ represents a glide), as shown in (1) and (2). The word is inherited across the TAP family with an initial plosive, and follows regular sound changes. This would suggest a very early contact with an Austronesian source at the stage when the TAP family had not yet diversified. If Austronesian groups arrived in the Timor area around 3800 BP (Pawley 2005; Spriggs 2011), that may have been the earliest time the borrowing could have occurred.

The PTAP form *baj reflects the PMP initial plosive /b/ in *babuy, and is a shortened form of the word. In contrast, the reflexes of PMP *babuy attested in the Flores-Lembata and Timor region are all disyllabic, see (1). Also, PFL *vavi or any of its descendants cannot be the donor of PTAP *baj because of the initial fricative. In the Timor region, the AN languages in the east are also unlikely donors because of their initial fricative, as shown in (1). Languages in the west of Timor show reflexes with initial *b, leading to the reconstruction of PRM *bafi. Thus, presently available evidence suggests a loan event involving an ancestor of the Timor languages that is at least as old as PRM, before the other Timor languages underwent lenition of initial *b.⁸ (PTAP did not borrow AN loans for other domestic animals like ‘dog’ and ‘chicken’.)

2.1.2 ‘deer’

Deer are ancient animals in eastern Indonesia. They appeared in Timor after 4500 BP (Bellwood 1997: 187), in Sulawesi after 3500 BP (Glover 1986) and in Flores after 2000 BP (Forth 2012: 457). Blust and Trussell (n.d.) give a cognate set of Proto West Malayo-Polynesian *uRsah ‘sambhur deer’ containing words from Philippine languages, Malay and Toba Batak. Words related to this form are found in Flores-Lembata and Timor, but they may have been borrowed from Malay rusa (Edwards 2021). Malay has likely been a regional lingua franca at least since the time of the Sri Wijaya empire (7th–9th Century), and has been used as a trade language in eastern Indonesia since before the colonial times. Antonio Pigafetta’s encounter in 1521 with traders from Malacca in Timor (Le Roux 1929: 31) and the Malay word list he collected in Tidore (North Moluccas) (Le Roux 1929: 72–99) is evidence that trade Malay was already used in the region in the early 16th Century.

In east Timor, the word rusa sometimes occurs in a compound with bibi (Proto Rote-Meto *bibi ‘goat’, Edwards 2021), or as synonym of bibi, see (3). This particular compound is also found in the TAP languages Bunak and Makasae, (4), which suggests that Bunak and Makasae picked it up from one of their neighbours; likely Tetun, the language of interethnic communication.

The Alor Pantar forms suggest a different history. On the basis of the lexeme set given in (5), we can reconstruct proto AP *arusa. Forms reflecting regular correspondences of the consonants are found in all major subgroups of AP: Tubbe r=l, Klamu s=ʧ, Reta r=l, s=h, Adang r=l, Abui r=j, s=t, Kaman r=l, l= zero, s=h.¹¹ On the one hand, this suggests that it is an ancient loan, though it is unclear what the donor language of PAP *arusa may have been. On the other hand, all groups also contain irregular forms, e.g. Teiwa *s>t (no change expected), Klon *s>t (expected *s>h), and Sawila and Wersing retained /s/ (expected *s>t). Irregular *s>t forms may suggest borrowing from a TAP language which underwent that change (e.g. Abui). Further confusing matters, the form could also have been a more recent loan from Malay rusa, as has been suggested for the forms attested in Flores-Lembata and Timor. There is no information about when deer appeared in Alor and Pantar.

2.2.1 ‘salt’

Salt is a natural sea product used in barter trade between coastal and inland people in Timor and Alor (Hägerdal 2012: 68, Wellfelt 2016: 145). Across the TAP family, we find reflexes going back to PTAP *asir, a form related to PMP *qasiRa ‘salt’, compare (6) and (7). The form must be a rather ancient loan. Given the different shape of PFL *hira, this cannot be the donor for PTAP *asir. The languages of west Timor reconstruct to PRM *masi from PMP *ma-qasin ‘salty’ (Edwards 2019), though Helong in west Timor has sila ‘salt’, a reflex of *qasiRa. The borrowing event of PTAP *asir from an Austronesian source must thus have taken place at a stage preceding PFL *hira or PRM *masi. In east Timor, the Austronesian languages partly reflect *masi (< PMP *(ma-) qasin) (Dadu’a, Galolen, Tetun), and partly *asira (< PMP *qasiRa) with loss of the initial /a/ and the intervocalic /r/ (Tokodede, Kemak, Mambae). Waima’a, Midiki, Naueti either reflect PMP *(ma-)qasin, or they reflect *asira plus loss of the final syllable, as shown in (6). It is thus likely that a form *asiRa was present at the stage of Proto Timor-Babar, a subgroup which includes all AN languages on Timor except those of Central Timor (Welaun, Kemak, Tokodede, Mambae); including west Timor Helong which has sila. The Proto Timor-Babar form *asiRa was borrowed as *asir into Proto Timor-Alor-Pantar.

In the Alor-Pantar subgroup, the Pantar-Straits languages show reflexes with metathesized vowels (*asir>isar), see (7). The cognates show regular sound correspondences in Adang, Kafoa and Klon (s=h), Abui (s=t), Kui, Kiraman, Kula and Wersing (s=s), Kamang s=s, r=i, Adang r=i.

3.1.1 ‘needle’

PMP *zaRum ‘needle’ is reflected in languages of east Timor as given in (8). It has monosyllabic reflexes where the intervocalic /r/ has been lost in Tetun, Kemak, and Naueti. This form was borrowed into Bunak, see (9). The form without the intervocalic /r/ is also the one attested in the east Alor languages Kula, Sawila and Wersing. Besides east Alor, the loan is not attested elsewhere on Alor or Pantar.

In Dadu’a, Galolen and Waima’a we find reflexes of #ruma, a form that may be connected irregularly to *zaRum. This form is also found in the lexemes in the east Timor TAP languages Makasae and Fataluku, which contain etymons related to both *daun and #ruma.

In the western part of Alor and Pantar, ‘needle’ is often expressed with a reflex of PMP *batuR ‘weave’, showing a semantic shift, compare (10)–(11). The form is likely borrowed from Kedang batur into Marica Alorese,¹² and from Alorese into the neighbouring Pantar-Straits and West Alor languages. The source language cannot have been Lamaholot or another Flores-Lembata language like Hewa, as these languages use a different form lusir/luhi(r) ‘needle’.

3.1.2 ‘sew’

In the languages of the Flores-Lembata region, various etymons are used to express ‘sew’, leading to two PFL reconstructions *daru (< PMP *zaRum ‘needle’) and *daʔit (< PMP *zaqit ‘sew’). In addition, we find reflexes of the form **sauR ‘sew’ (Edwards 2021) in Lamaholot and Kedang, see (12). Reflexes of the regional form **sauR ‘sew’ are also found in the AN languages of Timor, as shown in (13). The TAP languages of east Alor are likely to have borrowed from (a) language(s) of the Central Timor subgroup, Kemak, Tokodede, or Mambae.

The words for ‘sew’ in the other TAP languages listed in (15) seem to be related to the regional form **sədu(t) ‘weave’, reflected in Tetun sɔru ‘weave’ (see (16) below) and Central Lembata surit ‘weaving sword’ (Fricke 2017: 88); as well as in PRM *seru ‘weaving sword’ (Edwards 2021). The form that was borrowed into the TAP languages had medial *d changed to /r/.¹⁴

3.1.3 ‘weave’

The forms for ‘weave’ in the Timor AN languages Tetun and Waima’a are reflexes of the regional protoform **sədu(t) ‘weave’ (Edwards 2021, see ‘sew’ above). Similar forms are attested in the Timor TAP languages Bunak and Makasae, see (16).

The regional protoform **sədu(t) ‘weave’ also has reflexes in the TAP languages, but there the forms mean ‘sew’, see (15). In the TAP languages of Central and East Alor, the concept ‘weave’ is expressed by reflexes of borrowed PMP *tənun, *tinun ‘weave’, compare (17)–(18). (The t>s change is unexplained.) The forms could have originated from one or more AN language of Timor, compare Proto Rote-Meto *tenu. However, a direct source in east Timor cannot be established because, as mentioned, the modern AN languages of east Timor do not use reflexes of *tənun/*tinun ‘weave’, but forms of *sauR ‘sew’ instead to denote ‘weave’. It is also possible that the forms of the Central and East Alor languages are (adapted) loans from Malay or Indonesian tenun.

In Alor there are a few weaving communities along the coasts, but it is not known when the weaving technology was introduced. Oral traditions in Alor mention migrating groups who settled on the south coast as people bringing pottery (Wellfelt 2016, 63), and the same groups tend to be associated with weaving. Pottery and textiles were bartered with people in the interior, where there is a taboo on weaving.¹⁶

Weaving cloth has been considered as a typical Austronesian cultural feature (Blust 2013:24), but there is some evidence that the weaving tradition in Timor was introduced or disseminated only several hundred years ago (Hägerdal 2012). Pigafetta (1522) reported about a visit to Timor: “The chief with whom I went to speak only had women to serve him. [The women] all go naked, just like the other [women on the other islands]. In their ears they wear small golden earrings with hanging brushes at the side. On their arms they wear many bangles of gold and yellow copper until the elbow. The men go about like the women, apart from that they hang certain golden objects, round like a plate, around their necks, and that they wear bamboo combs in their hair, adorned with golden rings. Some of them wear dried pumpkin stems in their ears instead of golden rings.” (Le Roux 1929). Hägerdal (2012, 18) comments: “The alleged nudity of the women (and, apparently, the men) is more puzzling when regarding the long sarongs worn more recently, but it corroborates a Franciscan travel account from 1670. It is therefore possible that the well-known weaving traditions of Timor were introduced or disseminated at a fairly late stage.” Dutch illustrations of the seventeenth century show Timorese men wearing a kind of loincloth made of straps (Hägerdal 2012, 18). In southeast Alor and other places in Alor bark cloth was widely used for garments until the mid-20th century (Wellfelt 2016, 97).

Today, the (few) weaving centres in Alor produce textiles decorated with techniques that in Indonesian are summarised as songket. The textile traditions from the south and east coast of Alor show affinities with Timor, which is congruent with other historical sources, both oral and written, and with the borrowing of t/sine ‘weave’.

In West Alor, coastal groups produce textiles with clear affinities to the Solor islands, and with inspiration from Indian textiles called patola, produced in Gujarat in North West India from the 11th century onwards (Wellfelt 2016, 63). In the TAP languages of West Alor, Straits, and Pantar, no forms related to PMP *tenun are attested for ‘weave’; they use lexemes that are reconstructable to *degi ‘weave’, the source of which (MP or not) is yet unclear.

3.2.1 ‘slave’

In the AN languages of Timor, reflexes of PMP *qaRta ‘outsider(s), alien person(s)’ are found to mean ‘slave’, see (19). Edwards (2021) referring to Mahdi (1994:464 ff.) suggests as the meaning of *qaRta ‘negrito, black person’. This is based on the semantics across a wide range of MP languages which points to the original meaning being ‘black/Negrito person’ which, depending on the race of the speakers, was applied either to themselves or a subjugated population. In many languages of Sulawesi and Maluku reflexes of this etymon have the meaning ‘slave’. Of the TAP family, Bunak and Makalero borrowed a reflex of *qaRta ‘slave’, see (20), and both would be unproblematical borrowings from Tetun. In the Flores-Lembata subgroup, PMP *qaRta is reflected as PFL *ata ‘person’ (not ‘slave’). In the TAP languages not spoken on Timor, different etymons are used, see the forms in (21) and (22), further discussed below.

The pre-colonial political economy of Southeast Asia already included slave-raiding. Much of Southeast Asia was underpopulated until the 18th and 19th Centuries, and the key to political control was the control of labour power (Hoskins 1996, 3–4). The Makassarese from South Sulawesi played an important role in the pre-colonial and colonial slave trade, obtaining slaves from Alor, Manggarai and Ende in Flores, Timor, Tanimbar, Buton (Sulawesi), Mindanao (Philippines) and Brunei (Borneo) (Raben 2008, 132; Wellfelt 2016, 45). Most forms of slavery in Southeast Asia seem to have originated in debt bondage, but gradually diversified into complex “closed” systems of enduring social stratification and “open” ones of slaves captured primarily for external trade. As Hoskins (1996:4) writes: “Slaves were one of the most important “local products” exchanged from the hinterland for sale in entrepôts along the coasts, and they were usually obtained by raiding inland communities.” In Timor as well as elsewhere in eastern Indonesia, slaves were an importand trade commodity for the colonial Portuguese and Dutch VOC, alongside sandalwood and beeswax (Hägerdal 2012). Slave-raiders not only came from Sulawesi but also from the east. An unpublished grammar sketch of Iha, a Papuan language spoken on the Bomberai peninsula in Southwest Papua (Coenen 1953), mentions that in pre-contact times the Iha speakers went on slave expeditions all the way to the Kei and Tanimbar islands. In turn, there is a tradition in Fataluku (East Timor) that they came from the Kei islands (Voorhoeve 1989). This suggests that maritime contacts existed between the two ends of the chain Papua-East Timor, and a point in between, Kei; and that people movements took place along that chain.

On Alor, oral histories report about inland people such as the Abui being abducted and traded as slaves by coastal populations (Wellfelt 2016, 298, 300). An example are the Kolana (Sawila speakers) on the east coast of Alor. Kolana was allied with Liquiçá on the north coast of East Timor, with whom they traded wax, honey, cattle, and slaves, the latter acquired in wars or by kidnapping (Wellfelt 2016, 100). In 1851 van Lynden mentions Alor and Pantar as a former source of slaves to foreign traders, and the Oecusse enclave in north Timor is mentioned as a recipient of slaves from the Alor and Pantar: ‘In former days, Alor and Pantar provided many slaves and even now there are sometimes slaves being supplied to foreign traders, and to the Timorese (Oekoessie [Oecusse]) who are subject to Portugal […].’ (Van Lynden 1851:332). According to a Dutch report from 1879, slaves from Alor were sold in Liquiçá via the regent in Lamahala on Adonara island—and the Portuguese commander received a head tax for each imported slave. The year after, in 1880, another report was highly critical of the rulers in Kui on the south coast of Alor and Kolana on the east coast. Both were accused of having brought mountain people from Alor to be sold as slaves in Liquiçá (Wellfelt 2016, 103).

Tetun malae refers to foreigners or traders who came from overseas. Reflexes of this word denote ‘slave’ in TAP languages of the Pantar Straits and West Alor, as well as in South and East Alor, and Bunak Maliana, see (21). The use a word similar to Malay to refer to a slave would suggest that slaves were associated with people who do not (originally) belong to one’s group.¹⁷ The centuries of slave trade from Alor Pantar to Timor, also involving the Solor islands, may have caused the borrowing of a form similar to the Tetun word malae ‘slave’ into languages across Alor and Pantar.

In the Pantar-Straits area, Kaera and Blagar-Tuntuli borrowed the Indonesian/Malay form jongos [d͡ʒoŋos] ‘houseboy’ for the notion ‘slave’, as shown in (22). Originally, the word is from Dutch jongen(s) [joŋən(s)] ‘boy(s), houseboy(s)’. In Kaera, either the original Dutch form with initial [j] was borrowed (which seems unlikely, because there was no Dutch-speaking population on Pantar), or the initial affricate of the Malay form was simplified to [j] in Kaera because Kaera lacks a phonemic affricate /d͡ʒ/ (Klamer 2014).

The question may arise why at least three different etymons were borrowed for the same notion. Obviously, part of the answer lies in the different contact histories of the various regions, as the regional differences discussed above indicate. An additional explanation might be that, for many of the word lists used in this paper, the word for ‘slave’ was elicited using the Indonesian prompt budak. In Indonesian, this word has various meanings including ‘lad, boy’, ‘servant, underling’, and ‘serf, slave’, and thus it appears to have elicited words of a similar semantic range in the target languages.

In western interpretations, the notion of ‘slave’ means a person who is the servant-property of another person, and who can be bought and sold as such. In the regions where we did our surveys in Flores, Pantar and Alor, the translations of Indonesian budak include this meaning but may also refer to people who are temporary servants (‘debt slaves’), or to people who are not, or no longer, part of a particular clan lineage; for instance, orphans, newcomers or strangers. An example of this latter type is reported in Wellfelt (2016, 46). In an Adang village (West Alor) a story tells of a young man from Welai (Abui territory in Central Alor) who was taken from his parents and sold by relatives to traders from Binongko, Sulawesi. The traders ran into a storm and were forced to seek shelter in West Alor. The boy was set free and ended up with an Adang-speaking community in the mountains where he became founder of a new lineage. The abduction and sale of the boy is said to have happened 13 generations (i.e., 300–400 years?, MK) ago. Orphans and newcomers can start their own lineage in a clan, but unless they are adopted into an existing lineage, their lineage will retain a different (often lower) status. For example, they will not be allowed to take part in the ritual negotiations relating to marriage exchanges, but will have practical duties in support of these negotiations, such as organising the food. People in such non-autochtonous lineages may in some ways be considered as servants to the community, but they are not ‘owned’ by an individual or by a particular autochtonous lineage.

Budak can also be used to refer to war prisoners that are incorporated into the group who captured them, e.g. to become their wives; or prisoners who are given away to another group as part of a peace treaty. In their new environment, such ‘slaves’ do not necessarily get a lower societal position, nor are they necessarily seen as servants. In fact, they can become normal members of their new group. For example, a captured woman can be treated like all the other women who marry into the clan, and captured children may be adopted by childless couples who bring them up as their own children.

3.2.2 ‘king, ruler’

The Tetun compound liu rai ‘king, executive ruler’ (lit. ‘surpassing (the) earth/estate’, cf. Hägerdal 2009, 49), commonly written as liurai, has been borrowed into a number of TAP languages on Timor, as well as in languages in South and East Alor that were in contact with Timor (cf. Wellfelt 2016), see (23).

In TAP languages of Pantar, the concept ‘king’ is expressed with forms related to Malay/Indonesian rad͡ʒa ‘king’, possibly through the form that was borrowed into Adonara Lamaholot, as shown in (24), or else directly borrowed from Malay/Indonesian. In both Adonara and the TAP languages of Pantar, the affricate in rad͡ʒa has been simplified to [j], because none of these languages have a phonemic affricate [d͡ʒ]. The borrowing may be pre-modern or modern, but cannot be very recent, as currently, the d͡ʒ in Indonesian/Malay loans occurring in any of the these languages is not simplified to [j].

3.3.1 ‘breast’

PMP *susu is reflected in PFL *(t)usu and PRM *susu. Reflexes of *susu are also attested in the AN languages in the north of eastern Timor, see (25). Reflexes of a form with initial /s/ were borrowed into the TAP languages of Timor, see (26) (but Fataluku shows a reflex of PTAP *hami ‘breast’).

The TAP languages Kui and Kiraman on the south coast of Alor borrowed a form with an initial fricative su, (27). The source of this borrowing event is also likely to be (the ancestor of) an AN language spoken on the northern Timor coast, as these all have forms starting with /s/. In the TAP languages of Pantar-Straits and West Alor a form with an initial plosive was borrowed, similar to PFL *(t)usu (but dropping the final vowel). The donor is likely to have been Alorese tuho. The other languages of Alor Pantar, including those in East Alor and Fataluku on Timor, show reflexes of PTAP *hami ‘breast’.

3.3.2 ‘navel’

PMP *pusej is reflected in PFL *pusər (with an irregular final /r/). There are no reflexes of forms with a final glide attested in any of the TAP languages. Forms with an initial plosive /p/ and (reflexes of) a final liquid are found in Blagar and Reta in the Straits, and in Adang, and Kafoa in West Alor, compare (28)–(30). Forms without a medial /s/ in Blagar, Reta and Adang could be loans from an Alorese variety spoken on neighbouring Pantar island, as these varieties have puhɔr ‘navel’ (while most other Alorese varieties have forms with a prefix, as for example Alorese Marica), see (29). The forms with a medial /s/ point to a different source; they may have been borrowed directly from Malay.

The languages of Alor that are spoken further east have a different source. The form kubu in Wersing has an unexplained initial syllable that reflects the initial syllable of Tokodede kupusa, Alorese Marica kəpuhɔr, and is also found in other Western and Central Lamaholot languages (Fricke 2019). This suggests that there once was an older regional form with a prefix, which has modern reflexes in Timor as well as Flores-Lembata. The form in Wersing in particular probably originates from Tokodede on the north coast of Timor, given the geographical proximity and the contacts we know existed between groups in East Alor and North Timor (Schapper & Klamer 2017; Schapper & Wellfelt 2018). A shortened reflex -bu(:) is found in the sister languages of Wersing, Kamang and Tiyei.

The wide-spread borrowing of an MP word for ‘navel’ across the Alor languages is probably due to its socio-political connotation of ‘centre, head quarters’. The variable patterns of the loans indicate at least three different donor languages: Alorese, Tokodede, and Malay, where borrowing from Malay is likely to have involved separate borrowing events across the island.

3.4.1 ‘seed’

PMP *binəhiq ‘seed’ is reflected with forms like fini, hini or wine in the AN languages of Timor, see (31). In the TAP languages of Timor, only Bunak-Maliana has a form reflecting the original initial /b/, (33), so Bunak must have borrowed the word either before the sound change *b > w, f, h took place in Timor, or it borrowed the word from an unknown AN source that retained the original /b/. The original /b/ is also found in the loans of Abui, Kamang, Suboo and Tiyee, spoken in Central and East Alor, (33). This might suggest that the forms were borrowed from Bunak, but contact between Central and East Alor and the innerland Bunak seems unlikely. In the west Timor languages, the bilabial stop is also retained, cf. PRM *bini (Edwards 2021) so a predecessor of one of these west Timor languages could also have been the donor of the loans into the Alor languages. Alternatively, the borrowing into TAP may have occurred from an east Timor language before the initial /b/ of PMP *binəhiq started to vary in Timor. In the Flores-Lembata languages no reflexes of *binəhiq are found except in Sika (spoken in the Central Flores region), shown in (32). (Most of the Flores-Lembata languages use a form #kuluk (Fricke 2019), a form that has been borrowed as kulu (probably through Alorese) into the TAP languages Blagar Kulijahi and Blagar Nule. This form is not further discussed here.)

3.4.2 ‘maize’

Maize originates from South America and was taken to eastern Indonesia through the Iberian colonial trade network. Maize was first introduced in the Timor region in the period 1540–1650 (Hägerdal 2012:16). In the region under study, lexemes similar to PMP *batad (but with a final /r/) generally mean ‘maize’, as in the AN languages of Flores-Lembata (PFL *vatar ‘maize’) and the AN East Timor forms listed in (34).

PMP *batad ‘millet or sorghum sp. (unident.)’ is listed in Blust and Trussel (n.d.) on the limited evidence of three related forms from the Philippines, to which we can add Bugis bataʔ ‘sorghum’. It is unclear how old sorghum is in Southeast Asia. Lexical and ritual evidence presented in Fox (1991) indicates that it preceded maize as subsistence crop in eastern Indonesia. Makassar has bataraʔ ‘millet’ (Cense 1979), and since the Makassarese were involved in interregional trade including eastern Indonesia since before the colonial times (see the discussion of ‘slave’ in section 3.2.1), Makassar bataraʔ could be the source of a regional form batar, which assimilated millet and/or sorghum and maize (Fox 1991).

A Dominican source mentions maize on Lembata and Pantar shortly after 1641 (Hägerdal 2010, 224). Maize was grown in westernmost Timor by 1658, but must have been known some time before, since by then it was already the main crop (Hägerdal 2012, 50). In contrast, in some parts of Alor, maize was only introduced in the 20th C (Wellfelt 2016, 101). While the food was introduced relatively recently in certain parts of the region, the word seems to have a long history in the AP subfamily, and it may originally have referred to an earlier crop like sorghum, as it did in Kambera on Sumba, where wataru means both ‘maize’ and ‘sorghum’ (Forth 1983: 62).

Reflexes of PMP *batad or regional #batar ‘maize’ are not found in the TAP languages of Timor. However, in Alor Pantar, the form is attested everywhere, see (35). This form is strikingly similar to the forms attested in the AN languages of east Timor, see (34), and it is likely to have been borrowed from there. The Flores-Lembata region is an unlikely region of origin, because of the initial fricative in PFL *vatar.

The initial /b/ of *batar shows regular sound correspondences across the AP languages, e.g.: b>f in Abui, b>p in Kula, Sawila and Wersing, see (35). The final /r/ regularly got lost in Abui. If the word was introduced into the AP languages together with the introduction of the new staple food maize since the 17th C, this means that these sound changes must have occurred later than 400 years ago. Alternatively, the word may be an older loan that originally referred to ‘sorghum’ which assimilated the meaning of ‘maize’ after that crop was introduced, as it did in Timor and Sumba (Fox 1991).

3.4.3 ‘skin’

PMP *kulit ‘skin; bark’ is reflected in the AN languages of Flores-Lembata and Timor, see (36), as well as in modern Malay/Indonesian kulit. A reflex of this form is found throughout the TAP languages, where almost all lexemes reflect #kuli, with the final /t/ consonant lost, as shown in (37). Some but not all of the TAP lexemes show regular sound changes: PAP *l>i in Kaera koi, PAP *l>i and *k>ʔ in Adang ʔui. None of the TAP loans have more than two consonants, except Blagar Kulijahi -ʔulit, which could be a modern loan from Indonesian kulit, and Wersing klut, which is similar to Tokodede kulut-.

In Alor, bark cloth was widely used for garments until the mid-20th century (Wellfelt 2016:63, 97), and the widely spread borrowing of the concept ‘skin’ could be related to this, because the skin of certain tree were stripped to make bark cloth. There is archaeological evidence that the introduction of bark cloth technology followed the spread of Neolithic culture from southern China into Island Southeast Asia where bark cloth was substituted for other kinds of fibre materials (cf. Wellfelt 2016, 97). This may suggest that the bark clothing was introduced with the Austronesian word for it.

3.5.1 ‘bride price’

Loan forms denoting ‘bride price’ that are similar to PMP *bəli ‘value, price, marriage prestations, brideprice, purchase’ are found all over TAP, see (38) and (39). However, they do not show regular sound correspondences. For instance, we do not witness the expected regular correspondence between initial/medial PAP *b>f in Abui, and initial PAP *b>p in Kamang, see (39). Most TAP forms have lost the second syllable of PMP *bəli, with the exception of the disyllabic loans attested in the Pantar-Straits, and in Fataluku.

In the TAP languages, forms with initial *b are attested across the region. It is unclear where the loans originated from. If they came from Timor, the donor form must have had an initial stop. None of the modern AN languages of east Timor retained the initial stop, but PRM did have it, so borrowing could have happened at an earlier stage, before the initial consonant of PMP *bəli started to vary in the Timor region. In Flores-Lembata the initial stop of PMP *bəli was already changed into a fricative at the stage of PFL *veli, so if the donor was a language from the Flores-Lembata region, the borrowing occurred already before the stage of PFL.

Interestingly, in the region that is geographically closest to Flores-Lembata, the Pantar-Straits, no reflexes of PFL *veli are attested, but rather of *beli, see (39). The vowels and the syllable structure of these Pantar-Straits forms are different from the forms attested on Alor, and more similar to modern Malay/Indonesian beli ‘buy’ or belis ‘bride price’. The word belis ‘bride price’ is generally used in the Malay/Indonesian variety spoken in the eastern province (NTT) of Indonesia (Jones, Hull & Mohamad 2011). It is quite common to hear speakers of local languages use the loanword belis, likely because marriages are also frequently arranged between communities with different languages. This may suggest that the forms in the Pantar-Straits represent a different (possibly more recent) borrowing event involving belis. In general, the irregular forms suggest that the borrowing of reflexes of PMP *bəli occurred multiple times and from different sources.

The pairs in Reta Pura ɓili pala and Blagar Bama wili pala are probably borrowed from Alorese, which has the compound feling palang ‘dowry paid by the groom’s family to the bride’s family’.²¹