1 Introduction
The Papuan language families of Border, Nimboran, Sentani, and Sko cover a geographically contiguous area in the north of the island of New Guinea. The Border and Sko families are mainly located in the east in Papua New Guinea, while the Nimboran and Sentani families are located in the west in Indonesia (see Figure 8.1). It seems that this political split had consequences for language research in that, so far, these four families have not been brought together in unified research that may detect mutual influences among them. Doing this, the present the article breaks new ground, and will lead to new insights about the peoples, their languages, their interaction, and their ‘nomadic’ impetus over centuries, which only recently came to a halt due to the centralised political government in both modern states. The selection of the four families is further motivated by the aim to set Kilmeri and the Border languages in their wider linguistic and geographical context; as the author of a grammar of Kilmeri (Gerstner-Link 2018) it is an objective of mine to anchor this language in a broader research context.
When dealing with language contact in the geographical area of the Border, Nimboran, Sentani, and Sko families one has to distinguish two layers: (i) contact among local vernacular languages of the same family and across families; (ii) contact between Austronesian and Papuan languages; (iii) contact between local languages and the modern linguae francae (Papuan) Malay and Dutch as well as Tok Pisin and English. Needless to say, there are numerous loanwords from these linguae francae into the indigenous languages under examination. For the present study, contact with Austronesian languages, Malay, Tok Pisin, Dutch, and English is beyond the focus.
The article starts by outlining the historical and geographical settings of the language families and the people. There are no written native sources; the scanty data we have about the peoples’ history rely on oral tradition. In a few grammars, these oral accounts are very briefly documented. Wordlists started to be collected only in the 20th century. Section 3 reflects on this research situation and discusses some methodological considerations on which the vocabulary comparison and the recognition of loanwords are based. In the following three sections the Border languages are lexically compared with the Nimboran family, the Sentani family, and the Sko family. The putative transfers are listed and commented on one by one followed by a short summary concluding each of the three sections. These summaries provide information about number, word class, and semantics of transferred items, the directionality of transfers, the phonological integration into the recipient language, replacement or co-existence with an inherited word, and, if possible, about the relative age of the transfers. But these findings are not sufficient to propose concrete scenarios of contact in the sense of, say, Muysken’s (2010:271–278) scenarios. The only case in which a certain scenario is quite probable is discussed in Section 7: it deals with wanderwörter whose spread was facilitated through extensive bird of paradise hunting in the area for trade outside New Guinea. Finally, Section 8 summarises the lexical tranfers and reflects on their low number, which, however, is compensated to a small degree by a few patterns of structural convergence. The section ends with a discussion of putative migrations of the peoples, in particular the Kilmeri. At the same time, a hypothesis about the original homeland of the Border people and their languages is developed.
2 Historical and Geographical Settings
Table 8.1
The Border family including Elseng1
|
Border Family |
|||
|---|---|---|---|
|
Bewani branch |
Waris branch |
Taikat branch |
Elseng branch |
|
Ainbai |
Amanab |
Auyi |
Elseng |
|
(Minch 1992) |
(Menanti 2005) |
||
|
Kilmeri |
Auwe [Simog] |
Taikat |
|
|
(Gerstner-Link 2018) |
(Smits & Voorhoeve 1994) |
||
|
Ningera |
Daonda |
||
|
Pagi |
Imonda |
||
|
(Gerstner-Link 2000) |
(Seiler 1985) |
||
|
Manem |
|||
|
Sengi [Viid] |
|||
|
Waina [Sowanda] |
|||
|
Waris [Walsa] |
|||
|
(Brown & Wai 1986) |
|||
The Border languages (Table 8.1) cover a geographically contiguous area stretching from the Border Mountains and their foothills in the south to the valleys and plains north of the Bewani Mountains. The Bewani range is not inhabited. Nowadays, the people speaking Border languages live in three areas: north of the Bewani Mountains in the Puwani-Pual river basin and on the northern coast east of Vanimo; south of the Bewani Mountains and north-east of the Border Mountains in swampy hills and small creek systems as well as in the Wasengla valley (Waris) that stretches south-east along the headwaters of the Bapi river; thirdly west of the Bewani watershed and in the Tami and Bewani valleys. The Sengi (Waris branch) live further south and west of the Border Mountains.
The literature provides evidence that several linguistic groups of the Border people have migrated to their current locations a number of generations ago. For the Imonda, Seiler (1985:1) states that “[t]he Imonda trace their history to an area [,] to the north-west”. Regarding the Waris people Brown (1990:8) says that their self-designation Walsa “seems to refer to them as the ones who successfully overcame the previous people to live in the area”. The area in question is the Wasengla valley, and the Waris speaking clans may have pushed the Umeda group of the Waina speaking people southwards in a less favourable location in the north-eastern foothills of the Border Mountains (Gell 1992:153–154). Another or additional scenario may be that the Waris expelled some clans that spoke languages of the Kwomtari family (see language map), whose descendants may now live in the hot and swampy lowlands to the east (Donohue and Crowther 2004:173). Regarding a group of the Amanab speaking people anthropologist Juillerat suggests “[that] the Border Mountains seem to have been populated, at least in part, from the west or northwest, and the cultures found there contrast sharply with those of the nearby plain.” (Juillerat 1996: xxi)2 Finally, for the Kilmeri located north of the Bewani Mountains Gerstner-Link (2018:17–19) provides evidence that the people arrived at their current locations ten generations ago; the clan leader/s appropriated the land.
The Nimboran and Sentani families (Tables 8.2 and 8.3) we have fewer clues regarding their places of origin. According to their own oral tradition, the Nimboran came from the south to their current location: “Nimboran people say that their ancestors, along with those of the related ethnolinguistic groups of Kemtuik, Kwansu and Gresi, spread out into the Grimi River valley from a location named Singgi or hngni in the hills to the south. Today nearly all of the Nimboran people live to the north-west of the River Nembu.” (May 1997:3). Anceaux gathered his data on Nimboran between 1954–1957 in Jayapura (Hollandia) and during periodical visits to some Nimboran villages (1965:2–3). At this time, the Nimboran language was in full use. Unfortunately, Anceaux provides no clues about the history of Nimboran settlements.
Table 8.2
The Nimboran family
|
Nimboran family |
|||
|---|---|---|---|
|
Nimboran branch |
Kemtuik-Gresi-Mlap-Mekwei branch |
||
|
Nimboran (Anceaux 1965) (May 1997) |
Mekwei branch |
Mlap-Kemtuik-Gresi branch |
|
|
Mekwei |
Mlap branch |
Kemtuik-Gresi branch |
|
|
Mlap [Kwansu] |
Kemtuik |
||
|
(van der Wilden & van der Wilden 1975, 1976) |
|||
|
(Smits & Voorhoeve 1994) |
|||
|
Gresi |
|||
|
(Smits & Voorhoeve 1994) |
|||
Some groups of speakers of Sentani languages (Table 8.3) originate in a location that nowadays is populated by Border speakers. They trace their ancestors to the east. Chief Asareu tells that some ancestors originated from the earth, while others stem from Mount Fanim in the east. The settlement on the island of Osei in Lake Sentani was the first to be populated by migrants from the east. (Wirz 1934:257; 260) A Sentani myth says that a snake carrying a young man on its back swam across the Tami River towards the sea—the former Humboldt Bay—and finally reached the current location of Nafri (Table 8.3). The Tami and Bewani rivers flow through the current area of the Manem and Taikat people, who speak Border languages.
Table 8.3
The Sentani family
|
Sentani family |
|
|---|---|
|
Sowari branch |
Tabla-Sentani-Nafri branch |
|
Sowari |
Tabla |
|
(Gregerson & Hartzler 1987) |
|
|
Sentani |
|
|
(Cowan 1965, Gregerson & Hartzler 1987) |
|
|
Nafri |
|
The Skou people themselves, as well as the other speakers of the Sko family languages, also look back at repeated movements of clans or groups of men. Donohue states that the speakers of Proto Macro-Sko originally lived along the middle Puwani-Pual River area (2004:5). This is exactly the area where nowadays the Kilmeri live, and Donohue conjectures that these people were displaced by the intrusion of speakers of the Bewani branch of the Border languages (Table 8.1). The expelled Macro-Sko speakers migrated to the west and to the east and spread along the coast. For the eastern-most Sko speakers, the Barupu, Corris provides a quite detailed description of their putative migration and later arrival at their present location near the Sissano lagoon. When the ancestors of the modern Barupu left the Puwani-Pual area, some of them may have headed east, reaching the lagoon from inland; others are said to have come along the coast (Corris 2005:3–8).
Table 8.4
The Sko family
|
Sko family |
||||||
|---|---|---|---|---|---|---|
|
I’saka branch |
Piore River-Serra Hills-Inner Sko branch |
|||||
|
I’saka (Donohue & San Roque 2004) |
Piore River branch |
Serra Hills branch |
Inner Sko branch |
|||
|
Barupu (Corris 2005) |
Womo |
Skou branch |
Eastern Sko branch |
|||
|
Ramo |
Rawo |
Skou (Donohue 2004; 2002) |
Leitre branch |
Wutung-Sangke-Dumo-Dusur branch |
||
|
Sumo [Bouni] (Miller 2017) |
Puare |
Leitre |
Wutung-Sangke branch |
Dumo-Dusur branch |
||
|
Sangke |
Dumo |
|||||
|
Wutung (Marmion 2010) |
Dusur (Ross 1980) |
|||||
In sum, all these accounts provide evidence that the speakers of the four language families have a history of migration. According to oral tradition, Kilmeri clans migrated about 250 years ago. For the other groups, migration may have stretched over decades or even centuries and, at a time, comprised groups of clan size. See Section 8 for further discussion.
3 Method and Terminology
As a precondition for vocabulary comparison, we need reliable data sources that allow us to compare a sizable amount of the vocabularies of the languages concerned. This demand restricts the languages that can be thoroughly compared to those for which a lexicon and/or a grammar is available. For the Border languages, only Kilmeri, Waris, Imonda, and, to a lesser degree, Amanab fulfill this condition (see Table 8.1). For Taikat and Auyi, only (unsystematic) wordlists are published. Regarding the Nimboran languages, Anceaux’s (1965) and May’s (1997) grammars of Nimboran are good sources. The Sentani languages are lexically represented by Cowan’s grammar (1965) and supplemented by articles on Tabla and Sentani (Hartzler 1976; Gregerson and Hartzler 1987). Among the Sko family, good lexical sources are available for Skou, Wutung, Dumo, Dusur, I’saka, and Barupu (see Table 8.4).3
In reconstructing the contact scenario, I take the Kilmeri lexicon as a point of departure because (i) Kilmeri’s documented lexicon is the most comprehensive among the Border languages (Gerstner-Link 2021) and (ii) it is the language that the author knows best (Gerstner-Link 2018). The Kilmeri lexicon contains certain words that distinguish it from the other (well-documented) Border languages. Where do these vocabulary items come from when they are not inherited?4 Which lexical items of Kilmeri can be found in its neighbouring languages? This approach is restrictive in that it only allows for the discovery of a subset of mutual transfers or loans among the languages in question, namely those transfers that involve Kilmeri and the other Border languages. Transfers from, for instance, the Sko family to the Sentani family or vice versa, can more reliably be detected by researchers having first-hand knowledge of these families or single languages thereof. As we will see, lexical transfer among the above-mentioned languages and language families took place in multilateral directions: all families are both donors and recipients.
Turning to the question how to determine whether a lexical item is of foreign origin in a certain language I pursue the following path. If genetically related languages show cognate forms for a certain concept, then the lexeme in question is regarded as inherited. If a word is not attested in two branches of the same family but in only one, and it is also attested in another family, then I take it to be borrowed across the family borders. The fact that a word does not have an intra-family etymology is not an entirely conclusive sign of its loan status, since it might have been lost in the other branches of the family (Haspelmath 2009:44). However, without this working hypothesis there would not be any plausible reasoning to identify certain words as transfers or loanwords in the present context.
Regarding the Border family, no reconstruction has so far been done of a proto phoneme inventory accompanied by a (small) proto lexicon. For Waris and Kilmeri—representing two branches of the Border family—sound correspondences and cognates have been established by the author (see Appendix). Within the (putative) Bewani branch of the Border family, cognate sets for Kilmeri and Pagi have been uncovered (Gerstner-Link 2018:31–37). Based on these two sets of cognate pairs, I compiled a small triple set of cognate forms (see Appendix). These findings can count as a basis for inheritance within the Border languages. The sound changes involved could indicate the relative age of loans, insofar as these did or did not participate in a given change. For the Sentani family, the Proto Tabla-Sentani phonology has been reconstructed by Gregerson and Hartzler (1987); it serves as a basis for judgments about inheritance in this family. Donohue (2002) describes structural phonological borrowing accompanied by the rearrangement of the phoneme systems of the Inner-Sko branch of the Sko family, whose phoneme inventory he reconstructed. This again allows us to recognise Inner-Sko inheritance. I’saka is an outlier genetically, but currently an immediate neighbour of Kilmeri. Shared and similar lexical items between these two languages are due to recent contact (Gerstner-Link 2018:45–47). By contrast, for the Nimboran family no comparative work is available. Due to this stage of research, the etymological background of the compared Nimboran words must remain a matter of informed guesses but not of proof.
The procedure I used to assemble semantically and phonologically similar forms across language families can be described as follows. The starting point is a shared inter-family concept in the lexicon. The next step is to compare segments and syllable structure of the assumed loan with its counterpart in the assumed donor language. For example, Kilmeri and Nimboran share the concept ‘old’, and we have Kilmeri bepi and Nimboran bedí. Although three segments of the words are identical, I don’t regard the forms as resemblant, because /p/ and /d/ in position 3 cannot be related. Both languages possess these phonemes in their inventories, and there is no reason that one of them should have been replaced by the other for phonemic adaptation. I regard the similarity as coincidental. By contrast, the concept ‘wallaby’ is realised as Kilmeri ɛmɛi and Sentani proper eme. Here three segments are nearly identical in substance and order. Kilmeri could have taken over the form eme and have diphthongised the last vowel. Diphthongisation is a phonetic variation that can often be observed within Kilmeri when different speakers pronounce a word ending in /e/ or /o/; it can also be applied on words of foreign origin. See Section 5 below. Furthermore, if the phoneme inventories and/or the phonotactic rules of the languages in question differ, phonological adaptation has to be taken into account in order to establish segmental resemblance between forms.
A general problem with the languages concerned is the shortness of forms that are compared. This could be seen as causing a serious methodological weakness of the paper. Many forms I am dealing with are monosyllabic; sometimes they have only two segments. In this case there is the possibility of chance similarity. I can never exclude this possibility entirely, but I hope to present arguments that support the putative transfer. These arguments are based on word forms and their degree of similarity including phonological adaptation, on semantics including meaning shift (Blank 1997; Aikhenvald 2000), as well as on structural properties of the lexicon such as, in particular, co-existence of two terms for one concept. These terms may be nearly synonymous or the new term may add a finer lexical distinction.5 Word forms consisting of only one segment are excluded as candidates for transfer. The concepts ‘father’ and ‘mother’ are also excluded, since they are frequently realised as nursery forms which nothing should be inferred from.
My terminology follows Matras (2009), Haspelmath and Tadmor (2009), and Haspelmath (2009). In their Loanword Typology project, Haspelmath and Tadmor use the following definition: “We define a loanword as a lexeme that has been transferred from one lect into another and is used as a word (rather than as an affix, for example) in the recipient language.” (2009:13) Essential in lexical transfer and borrowing are also the notions of donor vs. recipient language (Matras 2009); for Haspelmath (2009:44), the identification of a plausible source word and a donor language is key for recognising a certain word as loanword. In most cases discussed below, the donor language (or the donating language family) can be identified; yet there are also cases of transfer in which the direction of borrowing remains unknown. In principle, both languages involved can each be either the donor or the recipient. Transfer is plausible in particular when the putative loanword shows signs of phonological adaptation from the source language into the recipient language; thus, phonological adaptation is indicative of the direction of borrowing (Haspelmath 2009:45). Secondly, phonological and morphological adaptation are criterial to distinguish loanwords from code switching (Matras 2009:41). Contrary to code switching, loanwords should be used conventionally as parts of the recipient language (Haspelmath 2009:40). This criterion of conventionality is certainly important for the final loan status of a word, but cannot be checked for the languages under consideration here (but see footnote 11 below). I simply assume it to hold.
4 Lexeme Resemblances between Border and Nimboran
The lexical comparison between the Border family and the Nimboran family is primarily based on the vocabularies of Kilmeri, of Waris (Brown and Wai 1986) and of the single language of Nimboran. After the compilation of an alphabetic wordlist of Nimboran based on Anceaux’s grammar (1965), 337 pairs of words from Kilmeri and Nimboran designating the same concept could be compared.
The Nimboran terms are given with their lexically determined word accent in accord with their notation by Anceaux, as for instance, méndʉ. Regarding the syllable structure of Nimboran, there are no word-final consonant sequences (Anceaux 1965:31; May 1997:13), while word-initial consonant sequences appear regularly. The constraints on syllable structure in Kilmeri are similar, yet consonant clusters are rarer than in Nimboran. Note that Kilmeri seems to show, in word-initial position, the development from nasals to plosives, namely /m/ > /(m)b / and /n/ > /(n)d/, which sets it apart from the Waris branch of the Border family (see Appendix).
4.1 Nouns
We find 14 instances of lexical transfer of nouns between the two language families. In two cases the original family affiliation of the source lexeme remains unknown (‘buttocks’, ‘neck; beak’). The terms are discussed roughly in the order of lexical fields.
| ‘garden’ |
The Waris branch of the Border family shows a common stem for ‘garden’, which takes the following forms: Waris oso, Manem os, Imonda ɒsɒ (Seiler 1985), Amanab aso (Minch 1992:126). Pagi employs the very similar form os. This stem is also present in Nimboran, Kemtuik, Gresi, and Mlap as usu and in Mekwei as asu (May 1997:122; 126; Smits and Voorhoeve 1994:102). But three languages of two different branches of the Border family show entirely different words: Taikat has manta ‘garden’, Auyi mu has ‘garden’, and Kilmeri has sɛlɛ ‘garden’. Thus it seems plausible to me that the word originated in the Nimboran family and spread into the Border family. |
| ‘taro’ |
There is a common Border word with Waris sa |
| ‘child’ |
In Kilmeri the word for child shows the sex-neutral form ruri. In Waris it appears as {mu}-tundis ‘girls’ and tuɛndis ‘boys’; the sound correspondences are regular. Imonda has the form toand ‘boy, son’, which is very close to Waris. Taikat has {ma}-ntu (Smits and Voorhoeve 1994:79). So we have a common, inherited word for the Border languages. In Nimboran and Gresi, ‘child’ is monosyllabic du (Anceaux 1965:15; Smits and Voorhoeve 1994:80). In Kemtuik ‘child’ appears as do [dwo] (van der Wilden and van der Wilden 1975:37), in Mekwei as do (Smits and Voorhoeve 1994:80). Thus, the Nimboran family also shares the stem for ‘child’. The transfer must have taken place between the families and before the Border internal sound change from Waris/Imonda /t,d/ to Kilmeri /r/. I argue for the direction from Nimboran to Border, because in the Border languages the original stem became expanded into bi-/poly-morphemic words. |
| ‘great-grandfather/parent’ |
For Kilmeri and Nimboran a lexeme is attested that refers to the generation above the grandparents; in Kilmeri it is sex-neutral, while in Nimboran it seems to designate males. The Kilmeri form is básɪp, and the Nimboran form is babuásii with stress on the penultimate syllable (May 1997:18). The bisyllabic structure of the Kilmeri term results from the loss of the second syllable of the Nimboran term, which precedes the syllable bearing the main stress. The Nimboran vowel sequence ii can be realised as [ik] (1997:18). In Kilmeri, syllable closures with /k/ are rare, and, if they occur, preferable have the form /ak/ or /(u)ok/; the coda /ik/ isn’t attested at all. So Nimboran [ik] is likely transferred as [ip], and Kilmeri is the recipient language. |
| ‘sound, word, speech, story, language’ |
In the Border languages, the common inherited word referring to meanings like ‘sound’, ‘word’, ‘speech’, ‘story’, ‘language’ has the form bɔ (Kilmeri) or mɔa/mɒ/mo (Waris, Imonda, Amanab); the sound correspondence is regular. In Nimboran, the complex words ne-mbwo ‘word, language, speech, matter’ (May 1997:83) and ne-mbwo-pem ‘story’ (1997:53) are attested. Both expressions contain the morpheme mbwo, which is similar to Border bɔ/mo. The other Nimboran languages resemble Nimboran ne-mbwo closely (Smits and Voorhoeve 1994:254; van der Wilden and van der Wilden 1975:35). I argue for the direction from Border to Nimboran, because in the Nimboran languages the original stem became expanded into bi-/poly-morphemic words. One might think that bɔ is a potential onomatopoetic form. However, Kilmeri has muli/mui.SG ‘say, speak’, molijɛ.PL ‘say, speak’, and muɛli ‘talk to sb’ with Recipient object agreement, and I doubt that all these grammatically distinct forms are onomatopoeia. |
| ‘tongue’ |
The word for ‘tongue’ is bɛr in Kilmeri, mɛki in Pagi, mindɛ in Waris, and mənde in Imonda; the forms are related via regular sound correspondences (see Appendix; Gerstner-Link 2018:31–41). A similar form we find in Nimboran with méndʉ (Anceaux 1965:18), but here it denotes ‘mouth’. The meaning shift from tongue to mouth is semantically plausible via (physical) contiguity (cf. Blank 1997:238–240), thus we can argue that the Nimboran word is a loan from Border. Probably it is taken from the Waris branch, since it shows the same consonantal phonemes. Kemtuik has the unrelated form [nɪmblɛn] ‘tongue’ (van der Wilden and van der Wilden 1975:37); this fact supports the direction of borrowing from Border to Nimboran. |
| ‘behind, buttocks; faeces’ |
In Kilmeri, the word for buttocks is ɛku. In Waris, resemblant akoko is attested for ‘faeces’, while ‘buttocks’ is designated by an entirely different form in the Taikat and Waris branches of the Border family (Smits and Voorhoeve 1994:40–41). Yet in Nimboran proper (Anceaux 1965:22) we find iákʉ ‘buttocks’, which is formally similar to ɛku and akoko; trisyllabic akoko may be a partly reduplicated form. Since the other Nimboran languages have no forms designating ‘buttocks’ that can be related to those forms, one can assume that among the two language families there is an island consisting of the three resemblant forms above. A transfer between Waris/Kilmeri and Nimboran proper seems plausible including the meaning shift; but the direction of borrowing remains unknown. |
| ‘hornbill, parrot’ |
The Kilmeri word referring to hornbills is iwan, while in Waris we find the unrelated form pɛila ‘hornbill’ (Brown 1986:78). Yet iwan is formally similar to iwaŋ ‘parrot’ of Kemtuik and Mlap (Smits and Voorhoeve 1994:130), and in Nimboran ueiáŋ ‘kind of small parrot, lory’ is attested (Anceaux 1965:30). Kilmeri lacks /ŋ/, while the Nimboran languages have both /n/ and /ŋ/ and could have taken over the word without adaptive change of the coda. Thus I conclude that Kilmeri borrowed the term from Nimboran and adapted it to its own consonant inventory. The meaning shift took place on the basis of the shared feature of a strong, curved bill. |
| ‘kind of pigeon’ |
Kilmeri and Nimboran seem to share a term designating a certain type of pigeon (other than the crowned pigeon): Nimboran imúo and Kilmeri imalɔ. The referential property of pigeon-like birds holds for both languages. Formally, both languages show a trisyllabic word, nearly identical segments, and share the second-syllable stress. Nimboran’s only lateral is realised as retroflexed flapped lateral (May 1997:28; he subsumes it under the plosive series), while Kilmeri /l/ is a lateral approximant. When taking over Kilmeri imalɔ, the intervocalic approximant must have been dropped. A loan relationship with Kilmeri as the donor language is possible. The concept is not attested in other Nimboran and Border languages; therefore intra-family comparisons don’t work towards clarifying the direction of transfer. |
| ‘neck; beak’ |
Kilmeri possesses several terms designating body parts of various animals. One of them is bɛsi ‘beak’. The copncept is not attested in other Border languages. In Nimboran we find besí ‘neck’ (Anceaux 1965:19), which resembles the Kilmeri word closely. On the assumption that Nimboran besí may also refer to a bird’s neck, transfer between the two languages is possible. The meaning shift seems plausible in either direction, since beak and neck are contiguous body parts of a bird, in front of the head and below the head. |
| ‘mosquito; termite’ |
The Border family and Nimboran formally share a term that denotes various kinds of insects like ‘mosquito’ and ‘termite’ as well as unspecified ones. The Border languages have Waris kles ‘very tiny biting insects’ (Brown and Wai 1986:37), Imonda and Sengi kles ‘mosquito’, Kilmeri klɛs ‘mosquito’, and Pagi ɛlɛs ‘mosquito’. This stem is not shared by Taikat, Auyi, and Manem (Smits and Voorhoeve 1994:136). In Nimboran we find klesʉ ‘termite’ (May 1997:124), while the Nimboran family forms for ‘mosquito’ are related to those of the Tor family (Smits and Voorhoeve 1994:137). It seems plausible that Nimboran borrowed the term klesʉ from the Border family and then shifted its meaning to ‘termite’. |
| ‘mussel; bead’ |
Kilmeri sájɔ ‘fresh water mussel’ seems to appear in Nimboran {uan}sáia ‘kind of white bead’; the phoneme sequence is almost identical and the stress pattern is the same. Kilmeri also employs sajɔ pul ‘bead’ (lit. ‘mussel seed’), which would have supported the meaning shift from ‘mussel’ to ‘bead’. I assume Nimboran borrowed sáia from Kilmeri. For all the other Border and Nimboran languages, the concept ‘mussel’ is not attested. |
| ‘sago grub, sago beetle’ |
In Kilmeri, sago grubs form a faunal class. Their classifying element is bɛ(r)- (Gerstner-Link 2018:646). In Nimboran we have bre ‘sago beetle’ (Anceaux 1965:11). The terms attested for Waris are mɛŋɛmb ‘beetle that produces edible grubs in sago’ (Brown and Wai 1986:50) and nə_mbəl ‘edible grubs’. The first element nə of nə_mbəl designates “the forest and its useful products” (Brown and Wai 1986:61). Pagi employs the same structure with na_mpɛl. Thus we arrive at a common Border stem bɛr/mbəl/mpɛl, which was borrowed by Nimboran as bre due to the constraint that word/syllable-final /r/ is not allowed, while /r/ in consonant sequences is common (1965:31–35). |
| ‘(vertical or horizontal) post in a house’ |
This meaning is only attested in two Border languages: in Kilmeri we have jali ‘supporting horizontal post’ and in Amanab sumur ‘housepost’ (Minch 1992:132). In Nimboran we find jatʉ ‘post’ (May 1997:37). The judgment of formal similarity between jali and jatʉ takes into account that Kilmeri lacks /t/. A transfer from Nimboran to Kilmeri is possible; then Nimboran /t/ would have been adapted as /l/. This adaptation is supported by the fact that word forms of the Waris branch with syllable-final /t/ appear with /l/ in Kilmeri: Waris atxa > Kilmeri ɛlɔ ‘sugarcane’, Imonda at > Kilmeri al ‘leech’. Waris /x/ is lost in Kilmeri (see Appendix); so ɛlɔ shows intervocalic /l/ like jali. In addition, Kilmeri possesses lɔpɔs ‘(vertical) housepost’ that designates the posts that are erected first on the ground when building a house. It may be that jali was taken over as a second term that would have allowed to distinguish between different kinds of post necessary for house building. Then we would have a case of “co-existence with the native word” (Haspelmath 2009:49), yet with specialisation of meaning. |
4.2 Verbs, an Adverb, and a Numeral
Six verbs, one adverb, and one numeral are indicative of language contact between the families in question. In one instance a Border language (Kilmeri?) turns out to be the donor, in seven instances Kilmeri is the borrowing language.
| ‘go there, go thither’ |
Kilmeri possesses an inherently deictic verb nɛ ‘go thither’ (Gerstner-Link 2018:822; 837–840). In Nimboran ‘to go’ is a zero root (Anceaux 1965:158; May 1997:105), but there is a directional suffix -ne ‘from here to the end’ (May 1997:74), which has a similar deictic value as Kilmeri nɛ. Compare also the Nimboran postposition ne ‘to’ which expresses ‘motion towards’ as substitute of a verb (May 1997:121). For Waris dəm ‘going over there’ is attested, with no formal relationship to the Kilmeri word. For Imonda no verb designating the concept in question is attested. So it seems plausible to conclude that Kilmeri borrowed its inherently deictic ‘go’-verb nɛ from Nimboran’s directional suffix -ne. The unmarked verb ‘to go’ in Kilmeri is lɛ; with the loan nɛ ‘go thither’ an inherently deictic verb was added to the motion verbs. We have a case of “co-existence with the native word” (Haspelmath 2009:49). |
| ‘stand’ |
In the Border languages ‘stand’ can be regarded as an existential-postural verb. All these verbs have a singular and a suppletive plural form: in Waris lox |
| ‘distribute, share food’ |
In Kilmeri we find the rarely used collocation ɪ pi ‘to share freshly butchered meat’, while the default verb for sharing food with somebody is ripɛi with recipient/dative agreement (Gerstner-Link 2018:386) or ripɛi.SG/rupɔpi.PL ‘to distribute food among several persons’. The main verb ɪ of the collocation ɪ pi relates to Nimboran iíi ‘to distribute’ (Anceaux 1965:28; May 1997:87). This Nimboran verb is construed with recipient/dative agreement. Because of the light verb construction in Kilmeri one can plausibly assume that Kilmeri borrowed the word from Nimboran.6 This is one more case of a loan that co-exists with the native verb, resulting in a semantic distinction, which is not attested in the other well-documented Border languages. In Waris we find pɔa.SG/pɔa |
| ‘hit, shoot, kill’ |
The Border languages share a common stem lu/lo/lɒ denoting the above meanings. Waris has lɔ- |
| ‘be sick’ |
Kilmeri has the verb mari.SG/marmarpi.PL ‘to be sick’ denoting sickness of any kind; severe illness is indicated by the augmented form nɔ-mari. A verb with this meaning is attested neither in Waris and Imonda nor in any other Border language. But looking at Nimboran, we find máre ‘unconscious’ (Anceaux 1965:12; 24), and it makes sense to relate this word to Kilmeri mari. Nimboran máre may be an adjective or a stative verb; either way, it could have been borrowed across word class bounderies. I take it to be a verb, and the meaning shift from ‘unconscious’ in Nimboran to ‘being sick’ in Kilmeri is straightforward. Anceaux mentions the possibility to form verbs from adjectives by use of verbal morphology (1965:120–121) and describes the infinitive—the root morpheme—as quasi-adjective that may combine with nouns (1965:112). This supports the assumption that Kilmeri originally borrowed the verb from Nimboran. |
| ‘answer’ |
Kilmeri possesses several verbs of speaking including wui- ‘to answer’. This verb shows obligatory agreement with the recipient/dative object (Gerstner-Link 2018:386). In Nimboran the respective verb is uú- ‘to answer’ (Anceaux 1965:124); it is construed with obligatory agreement of the recipient/dative argument like its Kilmeri counterpart. There are no data for ‘answer’ in the other languages of the Nimboran family. The formal and structural parallelism of the Kilmeri and Nimboran word makes a transfer probable. In Kilmeri, wui- seems to be an old word which is in the process of being replaced by the serial verb dɔri_muɛli ‘turn back_talk to sb’, a more frequently used verb. |
| ‘before, formerly’ |
Kilmeri kimikɛ ‘before, formerly, in former times’ seems to be an isolated form in the Border family. For this meaning data are available only in a few languages: in Waris we find dɔara ‘before, previously’ and nəmət ‘a long time ago’ (Brown and Wai 1986), in Imonda iauɒnam ‘in earlier days’ (Seiler 1985:27), and in Amanab autunam ‘long time ago’ (Minch 1992:120). None of the three words shows any similarity with the Kilmeri word. Yet in Nimboran we have míŋie ‘before’ that can be related to Kilmeri, which has also the (less frequently used) short form mikɛ. Most probably, the Nimboran word was borrowed and phonemically adapted. Both words {ki}mikɛ and míŋie might also contain kié ‘time’ (Anceaux 1965:28); when taken over by Kilmeri, the Nimboran term must already have been fused. |
| ‘two’ |
The numeral ‘two’ shows similar forms in Kilmeri dupua and Nimboran namuán (May (1997:50) spells namwan). Intervocalically Kilmeri has an n as well, as shown by the form rɔ-dupua EMPH-two, which is realised as rɔ-nupua. ‘Four’ is rɔdupua rɔdupua in Kilmeri, typically realised as rɔnpua rɔnpua (Gerstner-Link 2018:123). Note also the free variation of the onset in different Kilmeri speaking villages (cf. Brown 1991): Ilup nɔpwa and Isi I nupwa with a nasal versus Osol dupwa with the occlusion /nd/ like Ossima dupua. We also find the same type of variation with labials: ‘sister’ is muri in Osol, but bʊri in Ossima and Oup (cf. Brown 1991). This might also account for the word-medial difference of Kilmeri /p/ versus Nimboran /m/. The other Border languages use a different stem for ‘two’: Waris, Imonda, and Pagi have sabla, Amanab has sabaga, while for Taikat the two (unrelated) forms sember and naŋgɛr are attested. Clearly, Taikat sember relates to sabla via metathesis of the liquid. Obviously, Kilmeri doesn’t fit in here, and I assume the language acquired dupua ‘two’ from Nimboran. Kemtuik has namuan like Nimboran proper, Gresi has namwan, and Mekwei naman (Smits and Voorhoeve 1994:212). Despite the fact that the currently observable free variation of nasals and homorganic plosives in Kilmeri may not explain the consonant change in an old loan, I think that the hypothesis of transfer is the best account for its deviating form dupua ‘two’, which otherwise would stand as an entirely isolated form. |
4.3 Summary
The comparison of Border/Kilmeri and Nimboran vocabularies results in 22 instances of lexical transfer between single languages and between families: 14 nouns, six verbs, one adverb, and one numeral. The transfer of nouns is symmetrical; the transferred nouns are related to nature and environment, kinship, body parts, natural kinds, and material culture. The transfer of verbs goes from Nimboran to Kilmeri in most cases. Since Kilmeri lacks certain consonants, in the direction from Nimboran to Kilmeri phonological adaptation of the loans is required: /ŋ/ > /n/ syllable/word-finally (ueiáŋ > iwan) and /ŋ(g)/ > /k/ intervocalically (niŋ > nɛki, míŋie > {ki}-mikɛ); /t/ > /l/ intervocalically (jatʉ > jali). In the opposite direction from Kilmeri to Nimboran we find metathesis to prevent final /r/, which isn’t permitted phonotactically: bɛr > bre. Co-existence with inherited lexemes occurs four times with three verbs and a noun: ɪ pi ‘to share butchered meat’, nɛ ‘go thither’, wui ‘to answer’, jali ‘post’. The verbs borrowed from Nimboran into Kilmeri illustrate different strategies of integration: (i) We find direct insertion of the stem/word (Wohlgemuth 2009:87–89); (ii) We find the citation form plus a light verb as in ɪ pi ‘to share’ from Nimboran iíi ‘to distribute’ (2009:102–109); (iii) We find the re-analysis of a directional suffix in a Nimboran zero stem verb as a verb: -ne ‘from here to the end’ becomes nɛ ‘to go thither’ in Kilmeri.
5 Lexeme Resemblances between Border and Sentani
The vocabularies of Kilmeri and Sentani are compared on the basis of Cowan’s grammar whose vocabulary list provides about 500 entries (1965:75–88). But only six pairs of words designating the same concept qualify as instances of putative lexical transfer; they belong to different word classes and are now presented one by one.
Note that four of the proposed loans into Kilmeri are either used infrequently (‘wallaby’, ‘place’), restricted to a very narrow context (‘like’), or add a special meaning in a certain grammatical domain (NEG). It is the semantic constraints on ‘like’ and NEG that may also account for their relatively infrequent use. They all co-exist with inherited forms of Kilmeri, which suggests more or less deliberate expansion of the vocabulary and reduces chance similarity in favour of contact-related transfer. It may well be that the “new” term for ‘wallaby’ may have designated a particular kangaroo species in Kilmeri, a distinction lost today.
| ‘wallaby, tree kangaroo’ |
The ordinary Kilmeri terms for ‘wallaby’ and ‘tree kangaroo’ are bi_sɛm and bi_puɛl; the first element bi is the classifying element, still used as an independent noun meaning ‘pig, terrestrial animal’. Yet there is a less frequently used term ɛmɛi ‘wallaby’ in Kilmeri. This is clearly related to Sentani eme/emeho ‘forest kangaroo’ and borrowed from this language (Cowan 1965:78). In Waris, by contrast, we find the lexeme pind ‘marsupial’ which is cognate to Kilmeri {bi}_pɛr ‘possum’ via the regular sound correspondence /d/ ⟨ ⟩ /r/ (see Appendix). Pagi has sʊm ‘wallaby’ which resembles {bi}_sɛm of Kilmeri. |
| ‘village, place’ |
In addition to the inherited lexeme jilau ‘village’ (< jip_lau ‘house_place’), Kilmeri has the word jɔ ‘place’. It is not frequently used, but once in a while it occurs in texts and in spontaneous discourse. It appears to be a transfer of Sentani jo ‘village’; Tabla also has jo ‘village’ (Gregerson and Hartzler 1987:14). By contrast, in the other Border languages we find Pagi ji_tau ‘village’, Imonda la ‘village’, Waris la ‘nest of bird or pig or insect’ (Brown and Wai 1986:41), which are cognate with Kilmeri ji_lau ‘village’. |
| ‘sit, stay, live, settle, dwell, remain’ |
The Kilmeri existential-postural verb for singular/dual animate referents nakɛ ‘to sit, to stay, to live’ has no cognate counterpart in the Border languages. Imonda has afɒ ‘to sit’ and the singular/plural pair ale/a-fia ‘to stay, to remain’; Waris has a |
| ‘like’ |
Kilmeri has the special verb kina ‘to like’ without an etymology in the Border family. It is probably transferred from Sentani and Tabla kəna ‘to like’ (Gregerson and Hartzler 1987:13). In Kilmeri, kina co-exists with the inherited verb muli ‘to want, to like’ (Gerstner-Link 2018:490) and only appears as first component verb in verb serialisations with perceptive verbs denoting positive perceptions. |
| Distal deixis |
The common Border stem for distal deixis is di/ri, and it denotes spatial distance. In Kilmeri we have ri-jɔ ‘there, that’, consisting of the deictic stem plus a local suffix. In Waris we have di ‘over there’ (Brown and Wai 1986). The local distal deictic in Imonda is ed ‘there’ (1985:45), cognate with Kilmeri distance-neutral ɛrɛ ‘this, that’. In Sentani we find the following forms: dikə ‘that, those, yonder’ as local deictic (Cowan 1965); Gregerson and Hartzler (1987:11) have Central Sentani ndi ‘that’ and East Sentani ri(ki) ‘that’, while for Tabla di ‘that’ is attested. These forms contrast with dakə ‘this, these’: di- denotes distality, while da- denotes proximity. The distal deictic forms of Kilmeri and Waris relate to the distal stem di- of Sentani and Tabla, while their cognate proximal stems (Kilmeri ɔ, Imonda ɒh, Waris honi) are different from Sentani da-. This is an argument for the direction of borrowing: The form of the distal deictic was borrowed from Sentani. Because of the onset variation di/ri in the Border languages it is an old loan that was transferred before the intra-Border sound change /d/ > /r/ emerged. |
| Negative particle |
In addition to the normal verbal negation ar ‘not’, Kilmeri employs a special emphatic verbal negation ba (Gerstner-Link 2018:633). Pagi has a similar form bam ‘no, nothing’ (Gerstner-Link 2000). Kilmeri ar is cognate with Imonda at, which renders a sentential negation ‘it is not the case’ (Seiler 1985:171). The origin of ba/bam is less clear. In the Waris languages the narrow-scope verbal negation appears as mas VERB-mo in Amanab (Minch 1992:147), while Waris itself has a probable cognate form in the verb-final negative suffix -mɔa (Brown 1990: II,21). Imonda shows discontinuous sə VERB-m, and, in addition, has a form bal that is suffixed by -m and serves as negation of verbless clauses. Seiler (1985:171–172) calls bal a “dummy element”. Could all these ba(C) forms of negation known in the Border family be related to Sentani bam, whose (quite broad) meaning is given as ‘not, hardly; without; no good, bad’ (Cowan 1965)? It seems plausible to assume that Kilmeri took the negative particle from Sentani as a pronounced second verbal negation despite of its more general negative function in that language (cf. Sentani fə bam wali bam ‘without fear (and) without life’, i.e., ‘impudent and careless’ (1965:79)). Pagi took over the negative particle, too, but with a slightly different meaning. |
In sum: The lexical transfer between the Sentani and the Border languages is unidirectional; the latter are the borrowing languages in all six instances. Given the lexical data that can be compared, this is a very low number of loanwords. The lexical entries in Cowan’s grammar (1965) number about 600; among these are roughly 500 concepts for which a Kilmeri counterpart is known. From about 500 compared lexical items only six or 1 % are shared. The borrowed negative particle co-exists with the inherited negative particle in Kilmeri, Pagi, and Imonda.
Deictics and negation/negative markers appear to be rarely borrowed; they are not listed in Matras’s frequency-based hierarchy of borrowed function words (Matras 2009:157, 2007:32–36). In the case of their transfer from Sentani to Kilmeri/Border these borrowings served to expand a certain grammatical domain. The transfer of the deictic ri made a distinction possible that didn’t exist before in the deictic system of Kilmeri. The inherited Border distal ere became restricted to questions containing a deictic, and it acquired the temporal meaning ‘now’, which is never attested with the proximal stem ɔ (Gerstner-Link 2018:795–797). The new, borrowed distal took over the general distal function in Kilmeri’s deictic system (Gerstner-Link 2018:797–801).
6 Lexeme Resemblances between Border and Skou
In this section I deal primarily with the single language called Skou, but other languages of the Sko family will also be taken into account if they may shed light on a certain question. These languages are I’saka, Barupu, Wutung, and Dusur; they are chosen because their grammars also provide vocabulary lists. Regarding loanwords in Skou, Donohue says the following: “In addition to this native lexicon, we can recognise a number of loans from languages with which Skou has been in contact. […] There are probably also a number of words that find their origin in the languages related to Mbo (Kilmeri), Elseng (Morwap), Tobati and Sentani, but since lexical materials on these languages are scarce little can be said for that possible connection.” (2004:31) Indeed, Kilmeri can be shown to provide a few source lexemes for Skou. The Skou and Wutung lexemes are given with their tones according to Donohue: a low, á high, à falling pitch (2004:99, 524–573) and Marmion: á high, à low, â highlow pitch (2010:93).
6.1 Nouns
| ‘hole, hollow; empty’ |
Skou bí ‘empty’ (Donohue 2004:524) can be related to Kilmeri bɪ ‘hole, hollow’ which represents the common Border form C(b,m)V(ɪ,ie,ə). The phrase bɪ sɔlɔ ‘hollow only’ means ‘empty’ in Kilmeri; an empty house is referred to by jip bɪ sɔlɔ ‘house hollow only’. This phrase shows the syntagmatic contiguity of ‘house’ and ‘hollow’ in Kilmeri. The Skou form bí has three additional meanings, namely, ‘floor’, ‘shell, plating’, and ‘tree with air roots’. Donohue seems to interpret this form-meaning correlations as a quadruple homophony of bí instead of polysemy (2004:524). At first sight, homophony of four lexical entries of the form bí seems to make sense, since the four meanings appear to be quite different and unrelated. But there is a common seme of these meanings, namely ‘hollowness’. This type of space can only be defined in terms of a surrounding structure delimiting the cavity enclosed by it. In particular: The word ‘empty’ calls up the concept of container defining an empty space. ‘Floor’ circumscribes the space beneath a house (downward direction) and beneath its roofing (upward direction). Often the floor is the only planar, extended confinement of a Papuan house (especially with regard to cooking houses). ‘Shell’ designates the “house” of mussels; they live in a cavity confined by the shell material. A tree with air roots—e.g., a Banyan tree (Ficus Benghalensis)—can also be conceived as creating a cavity that can be entered; one may feel like being in a “house” confined by a set of (more or less densely) hanging air roots. Thus the semantic transition from Kilmeri bɪ ‘hollow’ to Dusur bí ‘house’ based on the seme of cavity is not too far-fetched; it relates to the concept of interioricity (Aikhenvald 2000: 277; 289), which is a well-known concept for establishing noun classes (other such concepts are, inter alia, shape, size, position, dimensionality, consistency (2000:275–293)). In view of this, the meaning shift from ‘hole, hollow; empty’ to ‘house’ is quite plausible semantically.8 I conclude that Skou and Dusur borrowed the word bí from Kilmeri. This is supported by the following lexical findings: In Wutung we find pêy ‘house’ (Marmion 2010:374) as well as lông ‘hole, opening’ (2010:372). Skou has pá ‘house’ (Donohue 2004:528) and ì ‘hole’ (2004:526). Likewise, I’saka and Barupu show no similarity between their terms denoting ‘hole’ or ‘empty’ and Skou/Dusur bí. Furthermore, Donohue’s cognate set for ‘house’ is not entirely convincing. Proto Skou *a can correspond to either a or i only in Leitre, while it is retained as a in all other Inner-Sko languages (2002:183; 188); compare the set for ‘hair’ (2002:189). In addition to the regular correspondence sets for Proto Skou vowels Donohue gives irregular sets (2002:189), and ‘house’ would also be an instance of it. In the form bí it is only the plosive that shows a regular change from *p > b. Donohue states: “Some unproblematic correspondence sets are found for vowels, but in addition to the cases summarized in table 2.1., there are many awkward correspondence sets, which probably reflect a long period of intense interaction and multiple reborrowings of words back and forth.” (2002:188) Presumably, Donohue means intra-Sko family borrowings—but it could as well be that external borrowings are involved in the irregular picture he describes. |
| ‘sago (jelly), portioned sago’ |
Skou possesses the word ná ‘sago package’ (Donohue 2004:527), which probably denotes portioned sago wrapped in a leaf. Other terms relating to sago are hòe ‘sago palm’, hòe è ‘sago porridge’, and kóe ‘sago pancake’ (Donohue 2004). Clearly, the forms hòe and ná cannot be related. In Kilmeri we have duɛ ‘sago palm’ and ja ‘sago jelly’, in Waris na ‘sago palm’ and jɛs ‘sago jelly’; in Taikat ‘sago’ is also na (Smits and Voorhoeve 1994). Waris/Taikat na and Skou ná are formally most similar. It seems possible that Skou took over the word from one of these languages by shifting and specialising its meaning, adding a new word to its own repertoire of expressions relating to sago. |
| ‘burn; fire’ |
The Border languages share a stem C(t,r)V(a,ɛ) ‘to burn’ as intransitive verb. Waris ta- is said to refer to the situations of the kind ‘fire is burning’ or ‘food is cooking’ (1986:112). Kilmeri rɛ ‘burn’ can be rendered as ‘fire is blazing’ or ‘food is cooking/done’. In both cases the verb denotes the process of burning and the visible event of a fire. But the languages also possess a special word for fire, viz., sue/sʊ. However, in Skou we find ra ‘fire’ and rà li ‘burn’ with li ‘do’ in a light verb construction (Donohue 2004:529). This word is similar to the Border stem for ‘burn’, especially to Kilmeri rɛ, if we take into account the Skou rule “There is a consistent pattern in which mid open vowels lower in Skou following an *h or in a falling tone syllable.” (Donohue 2002:188) Then it seems possible to conclude that Skou borrowed the word rà ‘fire’ directly from Kilmeri. There is also the compound rá_rí ‘burning wood’ (lit. ‘fire_tree’, 2004:235). The meaning shift involved is plausible. |
| ‘bush knife’ |
In the Border languages we find Kilmeri nɛʙi ‘bush knife’, Waris nabe ‘chopper, machete’ (Smits and Voorhoeve 1994), and Taikat nabej ‘chopper, machete’ (Smits and Voorhoeve 1994). Wutung has nápè ‘bush knife’ (Marmion 2010:94, 100; 373). The Wutung lexeme is a clear resemblance to the stem present in all three branches of the Border family; because of closest formal similarity it is probably borrowed from Taikat or Waris. Skou anábí ‘machete’ may also be taken from Border, while táng ‘machete’ (Donohue 2004:534) is certainly an old Macro-Skou word. |
6.2 Other Lexemes
| ‘shoot, hit’ |
The verb denoting the hunting activities of shooting and hitting has already been discussed in Section 3.3. The Border languages share the common stem lu/lo/lɒ with these meanings, and the Skou form is lú ‘shoot’ (Donohue 2004:527). Thus I conclude that not only Nimboran, but also Skou borrowed this verb from the Border family, probably directly from Kilmeri lui because of the vowel quality.9 |
| ‘good’ |
Kilmeri employs the lexeme maki ‘good, of best quality’ that, at first sight, is hard to relate to an adjective form of the Border languages with (roughly) this meaning. Yet we have Ainbai maŋgri ‘good’ (Brown 1991) and Waris maka-l ‘mature, big fruit’ (Brown 1986). These three forms share the stem maŋ-/mak-; the meaning shift involving Waris is plausible. Thus we can say that there is a common Border form with the meaning of ‘good, big’ found in two branches of the family. Skou possesses—except for the suprasegmental feature of tone—a formally identical adjective máki with the meaning ‘big’ (Donohue 2004:527). Wutung, which is adjacent to Skou, has húwúrtì ‘big’ (Marmion 2010:371). The eastern-most Sko language Barupu has pako ‘big, be big’ (Corris 2005:383).10 Despite of the meaning shift towards size only I conclude that Skou máki ‘big’ is a loan from Kilmeri. |
| ‘well, then’ |
Skou so ‘well, then’ is of “(highly) suspected non-Skou origin[s]” because the s cannot be assumed to be an allophone of one of the Skou phonemes (Donohue 2004:35). A possible solution regarding the foreign origin of this particle can be found in Kilmeri sɔ and/or sɔ sɔlɔ, which has the pragmatic value of affirmation of an ongoing process. Note that Wutung has so ‘okay’ (Marmion 2010:375), which also fits the meaning of the Kilmeri particle. Most likely both languages borrowed the word from Kilmeri. |
6.3 Summary
The number of transferred lexical items is low again: seven words in sum, with four nouns, one verb, one adjective, and a pragmatic particle. In six instances Skou is the recipient language and in one instance Wutung. In one case (‘sago package’) the loan co-exists with inherited terms and adds a specialised concept. In all instances of borrowing, Skou and Wutung need to integrate the loans from the Border languages into their tonal systems. The verb ‘shoot, hit’ is probably taken over in its past form lu and then integrated into the morphological structure of Skou.
7 Lexeme Resemblances across the Border, Nimboran, Sentani, and Sko Families
Lexeme resemblances across many languages and several families suggest the phenomenon of wanderwörter that spread over a geographical area (cf. Haspelmath 2009:45). They are either the result of direct contact between several languages, or else they spread via extensive use by traders who cross different, rather small language areas, as we find them in Central Northwest New Guinea. Candidates for such wanderwörter could be the words discussed in this section: ‘water’, ‘tree’, ‘leaf’, and ‘arrow’; these words can be associated with bird of paradise hunting. Two of the words are basic lexical items that are otherwise not easily borrowed, viz., ‘water’ and ‘leaf’ (Tadmor et. al. 2010:239–241);11 since the authors include the age score of a word in determining the “basicness” of a vocabulary item (2010:237), the spread of a certain form designating water is remarkable as it counts as a stable item. But bird hunting and plume trading may have facilitated the acquisition of these words that became lasting items of the vocabularies of several families.
The distribution across languages of the putative wanderwörter is shown in Tables 8.5–8.9:
Table 8.5
‘water, rain, river’
|
Language |
Reference |
‘water’ |
‘rain’ |
‘river’ |
|---|---|---|---|---|
|
Border family |
||||
|
Waris |
Brown 1986 |
pɔ |
pɔ |
pɔ |
|
Imonda |
Seiler 1985 |
pɔ |
pɔ |
pɔ |
|
Kilmeri |
Gerstner-Link 2018 |
pu |
pu |
pu |
|
Pagi |
Gerstner-Link 2000 |
pɔ |
pɔ |
pɔ |
|
Taikat |
Smits & Voorhoeve 1994 |
wea |
bu, mu |
wea |
|
Elseng |
Menanti 2005, Burung 2000 |
vɛtɛv |
jai |
vɛtɛv |
|
Nimboran family |
||||
|
Nimboran (lang.) |
Anceaux 1965 |
bu |
sai |
bu |
|
Kemtuik |
van der Wilden 1987 |
bu |
sa |
|
|
Sentani family |
||||
|
Sentani (lang.) |
Cowan 1965 |
pu, bu |
ja |
wi |
|
Tabla |
Gregerson & Hartzler 1987 |
bu |
wai |
|
|
Sko family |
||||
|
Skou |
Donohue 2004 |
pa |
fu |
pa |
|
Wutung |
Marmion 2010 |
ʧâ |
fɵ |
|
|
I’saka |
Donohue & San Roque 2004 |
wì |
wì |
|
|
Sumo (Bouni) |
Miller 2017 |
pi: |
bɔ: |
The lexical item that spread is pɔ/pu. It occurs in 11 languages; gaps in the columns are due to lack of data. In the Nimboran family it only refers to ‘water’ and ‘river’, while in Taikat, Skou, and Wutung it specifically denotes ‘rain’. In Taikat, Nimboran, Kemtuik, Sentani, Tabla,12 Skou, and Sumo it co-exists with other terms of the same lexical field. Skou pa and Wutung ʧâ belong to a well-established cognate set (Donohue 2002:187); therefore the spread words are fu and fɵ. Since pɔ/pu has the widest denotational range in two branches of the Border languages, I assume that it spread from these languages into others in which it takes over one or two meanings.
Table 8.6
‘tree, wood’
|
Language |
Reference |
‘tree’, ‘wood’ |
|---|---|---|
|
Border family |
||
|
Waris |
Brown 1986 |
ti |
|
Imonda |
Seiler 1985 |
ti |
|
Kilmeri |
Gerstner-Link 2018 |
ri |
|
Pagi |
Gerstner-Link 2000 |
ki |
|
Taikat |
Smits & Voorhoeve 1994 |
ti, di |
|
Nimboran family |
||
|
Nimboran (lang.) |
Anceaux 1965 |
di, ri |
|
Kemtuik |
Smits & Voorhoeve 1994 |
di |
|
Sentani family |
||
|
Sentani (lang.) |
Cowan 1965 |
o |
|
Tabla |
Gregerson & Hartzler 1987 |
o |
|
Sko family |
||
|
Skou |
Donohue 2004 |
ri |
|
Dumo |
Donohue 2002 |
ti |
|
Dusur |
Donohue 2002 |
ti |
|
I’saka |
Donohue & San Roque 2004 |
téi |
|
Sumo (Bouni) |
Miller 2017 |
ʔái |
|
Barupu |
Corris 2005 |
ai |
The lexical item that spread is ti, yet it is not found in the Sentani family. In Skou the sound change /t/ > /r/ took place (Donohue 2002:200). In Border we have the following correspondences between the Waris branch and Kilmeri: /t/ corresponds to /r/ syllable-initially; /nd / corresponds to /r/ in other positions (see Appendix). Kilmeri and Pagi show the regular correspondence /r/ ⟨ ⟩ /k/ (Gerstner-Link 2018:31–35). The Piore branch of the Sko family has another word for ‘tree’. I conclude that ti spread from the Border family; because of the sound changes in the Border family and Skou it is an old transfer.
Table 8.7
‘leaf’
|
Language |
Reference |
‘leaf’ |
|---|---|---|
|
Border family |
||
|
Waris |
Brown 1986 |
|
|
Imonda |
Smits & Voorhoeve 1994 |
lop |
|
Kilmeri |
Gerstner-Link 2018 |
pɛlɛ |
|
Pagi |
Gerstner-Link 2000 |
pɛlɛ |
|
Taikat |
Smits & Voorhoeve 1994 |
fælej |
|
Nimboran family |
||
|
Nimboran (lang.) |
Anceaux 1965, May 1997 |
pró, plo |
|
Kemtuik |
Smits & Voorhoeve 1994 |
dɔp |
|
Gresi |
Smits & Voorhoeve 1994 |
dɔp |
|
Sentani family |
||
|
Sentani (lang.) |
Cowan 1965, Gregerson & Hartzler 1987 |
fe, fæ |
|
Tabla |
Gregerson & Hartzler 1987 |
{kə}pei |
The spread form is shaped C(p,
Table 8.8
‘arrow’
|
Language |
Reference |
‘arrow’ |
|---|---|---|
|
Border family |
||
|
Waris |
Smits & Voorhoeve 1994 |
pæ ‘bow’ |
|
Brown & Wai 1986 |
|
|
|
Imonda |
Seiler 1985 |
fal |
|
Kilmeri |
Gerstner-Link 2018 |
pɛ |
|
Pagi |
Gerstner-Link 2000 |
pai |
|
Taikat |
Smits & Voorhoeve 1994 |
fale, fara |
|
Elseng |
Menanti 2005 |
ɸal |
|
Nimboran family |
||
|
Nimboran (lang.) |
Smits & Voorhoeve 1994 |
pro{daj} |
|
Kemtuik |
van der Wilden 1975 |
ple |
|
Gresi |
Smits & Voorhoeve 1994 |
para{daj} |
|
Sentani family |
||
|
Sentani (lang.) |
Cowan 1965, Gregerson & Hartzler 1987 |
fəla |
|
Tabla |
Gregerson & Hartzler 1987 |
pəra |
The spread form is shaped C(p,f,ɸ)V(ɛ,ə,æ,a)C(l,r)V(e,a). In Waris, Kilmeri, and Pagi the word is monosyllabic; additionally, there is a meaning shift to ‘bow’ in Waris. Kilmeri and Pagi lack labial fricatives in their inventories; the sound correspondence Kilmeri /p/ ⟨ ⟩ Waris /
8 Conclusion and Discussion
8.1 Types of Borrowed Items
The lexical transfer between the Border, Nimboran, Sentani, and Skou families presents a manifold scenario. We see wanderwörter that are found in languages across several language families and we see words that are found in only two language families, viz., in the Border family and in just one of the other families. Regarding the word classes transferred items belong to, we count 15 non-nouns vs. 20 nouns plus four nouns of the category wanderwort. This distribution shows that nouns are indeed more easily borrowed and dispersed than other words. With 24 to 9 items, the ratio of nouns to verbs is close to three-to-one, and is roughly in line with the average ratio found by Tadmor (2009:61–62) in the database representing the languages of the world.
Semantically, the nouns belong to the domains of nature and environment, kinship, body parts, natural kinds, and material culture. The verbs belong to the domains of motion, existence/posture, hunting, eating, and being sick. Field-related constraints or preferences cannot be detected; instead, the words in question appear to be a selection across the whole lexicon. Quite a few meanings of borrowed or areally dispersed items discussed in the present study occur in “The Leipzig-Jakarta List of Basic Vocabulary” (Tadmor et al. 2010:239–241); the meanings are given with their rank in this list: to go (3), water (4), tongue (6), neck (23), to stand (45), child (51), to burn intr. (53), good (56), not (56), leaf (64), wood (80); some meanings obtain the same rank in the list. The ratio of all borrowings to core vocabulary borrowings is 38:11; that is, just under 30 % belong to the core vocabulary. This result suggests that items of the core vocabulary are not in principle resistant to borrowing.
Table 8.9
Identified transfers in numbers and word class14
|
Target language |
|||||
|---|---|---|---|---|---|
|
Source language |
Kilmeri / Border family |
Nimboran proper |
Sentani proper |
Skou proper and eastern Sko |
Wanderwörter |
|
Kilmeri/Border family |
6 nouns 1 verb |
4 nouns 1 verb 1 adjective 1 particle |
3 nouns |
||
|
Nimboran proper |
6 nouns 5 verbs 1 adverb 1 numeral |
1 noun |
|||
|
Sentani proper |
2 nouns 2 verbs 1 deictic 1 neg. particle |
||||
|
Skou proper and eastern Sko |
|||||
|
Direction of transfer unknown |
2 nouns |
||||
The lexical transfer shows a strong tendency to asymmetry. In the case of Nimboran proper versus Kilmeri/Border family, Nimboran is the source language in 13 instances and the target language in 7 instances. Sentani is only a source language with respect to the Border family lexicon. On the other hand, the Border family, and Kilmeri in particular, is the main source for loans into Skou. The number of traceable transfers is low; the Border and Nimboran families possess the highest number of contact-related lexical items.15 The relative high amount of transfers between Kilmeri and Nimboran (in detail Section 4 above) is a surprising insight, since today the two languages are farthest away from each other on the east-west axis. Socially, the Kilmeri people don’t seem to have any ties that far west; some clan relations across the state border only exist to Manem and maybe Taikat speaking clans.16
Muysken (2010:272) describes the scenario of borrowing generally as asymmetrical from a dominant superstrate to a socially subordinate language; Winford (2010:177) sees this (a)symmetry relation as a tendency. Because of the very low numbers of borrowing in the present context one should be cautious to draw inferences about social hierarchies between the peoples concerned. Extra-linguistic sources of former social hierarchies between the languages in question that may support possible dominance are not available. Today, however, the Kilmeri are not bilingual in the contiguous vernacular languages I’saka and Pagi; the eastern Pagi villages (Imbio, Imbinis) are looked down upon by them. The recent Kilmeri people are clearly the dominant group in the proximate area. In former times, this may have been the other way round vis-à-vis the Nimboran and Sentani in the west, from whom the Kilmeri borrowed some vocabulary. Presumably, the Kilmeri and Border people were “jungle-dwellers” who got in contact with “river-dwellers” (cf. Aikhenvald 2008:2, 14). The Sentani were clearly lake-dwellers with a fair amount of fish production; they may have traded fish for sago (cf. Cowan 1965:72–74).
Lexical transfer is usually said to be the outcome of bi-/multi-lingualism. Foley (2010:797) describes Papuan multilingualism as extensive in the whole New Guinea region, but characterises it as a mainly male affair.17 Yet the marriage of women into another language group is common social behaviour and usually results in some degree of bilingualism, at least in the family and village contexts. Thus it is plausible to assume degrees of bi-/multi-lingualism for the speakers of the languages under investigation, though little can be said about its type. It may have been related to life stages and sex of the speakers. The imperfect second language acquisition is in line with adult bilingualism, insofar new phonological oppositions cannot be acquired any more (cf. Ross 2013:20). Instead, the new words are phonologically adapted (Kilmeri) and tonally integrated (Skou, Wutung).
The overall lexical transfer among the languages is low. This may be caused by the lack of long-lasting and frequent direct contacts. On the other hand, it could be indicative of language loyalty as a means of group identity, which may have played a major role in language attitude, especially since languages are often spoken by (very) small groups of speakers (Winford 2010:178; Foley 2010:796). For the Kilmeri clans and villages, their shared language is a firm pillar of their shared identity; this view was confirmed by all my language consultants.
8.2 Structural Convergence?
In the light of the very little lexical transfer among the languages in question one may ask about possible structural convergence regarding the (greater) area, in which these languages are spoken. The retainment of the vocabulary could then be interpreted as a general means of highlighting and preservation of group identity. A reference case for lexical divergence paired with high structural convergence are the languages of the Banks and Torres Islands in north Vanuatu, described by Francois (2011). Grammatically, the 17 languages spoken there (some are moribund) build a linguistic area; they show almost perfect intertranslatebility based on identical word order and (almost) identical grammatical categories (2011:178; 214). Clearly, the area of the Border, Sko, Sentani, and Nimboran families cannot be regarded as a structural convergence zone like northern Vanuatu. However, what we do find, is selective convergence regarding some special grammatical features among a few languages from two or more families.
| Structural isomorphism in the lexicon |
Kilmeri and Skou show surprising similarity in their kind-referring terms, which are usually composite words consisting of a generic term and a specific term. The generic term indicates the class in accord with folk taxonomy, while the specific term adds the necessary distinction. Kilmeri lexically distinguishes twelve faunal classes and seven floral classes (Gerstner-Link 2018:644–659), whose members comprise different kinds numbering between 64 (trees/shrubs), 34 (birds and bats) and three (yams; blood sucking insects; caterpillars). As for Skou, the vocabulary lists in Donohue (2004; 2002) offer the following easily recognisable classes: (i) animals moving in the air, (ii) animals moving in water, (iii) animals with fur, and (iv) snakes. I illustrate this structural lexical isomorphy for the class of animals moving in the air with a few examples: ‘hornbill’ S tángung and K iwan, ‘pigeon’ S tángángue and K imalo, ‘heron’ S tángpa and K iwai, ‘lorikeet, parakeet’ S tánglè and K ipumiya, ‘small bat’ S tángkengkeng and K imero.18 No other Border language shows this pattern for the faunal and floral domain of their lexicons; the Sko languages Wutung (Marmion 2010:283–284), I’saka (Donohue and San Roque 2004), Dusur (Ross 1980:101–105), and Barupu (Corris 2005) illustrate it to a certain degree. The Kilmeri pattern of kind-referring terms is a structural innovation due to transfer from Skou. |
| Phonological isomorphy |
Kilmeri and Skou make the phonological distinction between /l/ vs. /r/, while the other members of both families have only one liquid. In Kilmeri, /t,d / changed to /r/ (see Appendix); in Skou, *t became /r/. Donohue says that the development of /r/ in Skou must have been due to areal pressure (2002:192; 200). The change from / t,d / to /r/ is not only observable in Kilmeri, but also in eastern Sentani: *d changed to r word-initially, while *t became r intervocalically between central or back vowels (Gregerson and Hartzler 1987:10–11). Foley describes Sentani’s consonant inventory as employing both liquids /l/ and /r/ (2018:439). Farther west, Berik, a member of the Tor family, also distinguishes /r/ and /l/ (Foley 2018:472). |
| Pronoun system |
In the Border, Nimboran, and Tor families as well as in the Kaure family to the south we have pronoun systems that distinguish only four categories: first person, second person, third person, and inclusive. There is no number distinction (Anceaux 1965:167; Foley 2018:470–471; 456). The Sentani and Sko families, on the other hand, employ number distinctions: Sentani distinguishes singular and plural forms (Cowan 1965:16), while Skou distinguishes three numbers and even adds gender distinctions in the dual and third singular (Donohue 2004:186). In contrast to Waris and Imonda, the current pronoun system of Kilmeri has singular, dual, and plural forms and consists of eleven different forms (Gerstner-Link 2018:109; 111). The dual forms are transparently bimorphemic forms that add a locative suffix, which is also used to build pairs of people referred to by proper names (2018:238). The plural forms are more opaque and less easily analysable, but have certainly a bimorphemic history. Kilmeri second plural inɛ may go back to the Border stem ind ‘person, man’ plus dɛ ‘you’, resulting in inɛ literally meaning ‘you person’. Note that the plural ‘they’ is often expressed by jɛna ‘people’, which is presumably cognate with ind. It is a plausible hypothesis that we observe structural transfer in the current Kilmeri pronoun system, under the older influence of Sentani (plural) and the newer influence of Skou (dual). |
| Dative verbs |
Kilmeri possesses 13 verbs with obligatory recipient/dative agreement (Gerstner-Link 2018:386–387) and Nimboran proper possesses 11 such verbs (May 1997:86–88). In this agreement class, they share three common dative verbs (‘tell sb’, ‘show sb’, ‘give sb’), but they also share five verbs with meanings that are not commonly dative verbs: ‘ask sb’, ‘answer sb’, ‘gossip about sb, call sb names’, ‘wait for sb, meet sb’, ‘share food with sb’. In view of the fact that, like the other (documented) Border languages, Kilmeri is predominantly a language with number agreement (2018:323–385), this convergence of role-based person agreement illustrates constructional isomorphy (Francois 2011:212) and may well be due to contact and mutual transfer. |
Conclusion: The little lexical transfer among the language families under investigation does not correlate with a high structural convergence via transfer of grammatical properties. However, the transfer of categories in the pronoun system shows that the overall system can be modified under contact influence, while the formal substance of pronouns is indeed quite resistant to borrowing (cf. Tadmor et al. 2010:233). The Border family is the only one which participates in all of the above patterns of argued convergence. This hints at a complex contact scenario over time, i.e., to a series of successive contact events.
Figure 8.2
Putative migration routes of the Kilmeri, Nimboran, Sentani, and Sko people
Comment on the map: The proposed migration routes shown in the map are not exhaustive in the sense that they are not meant to comprise all migrations of the Border people. Quite probably, many more movements away from the proposed homeland took place over time, especially to the south. The migration route of the Pagi is hypothetical. Furthermore, some clans of the Sko people may have gone east or north directly.
8.3 Traces of Contact and Migration Patterns
Language contact among vernaculars presupposes vicinity or even contiguity of the languages concerned. Hence, we need to assume that clans speaking Kilmeri and clans speaking Nimboran and Sentani settled in the same area during a certain time span in the past. The location of their more or less adjacent homelands and hunting grounds may have been in the northern part of the area which is now assigned to Elseng on language maps (see Introduction, Figure 8.1). I hypothesise that the middle Tami river area is the place from where the Border languages spread southeast and east. This hypothesis is supported as follows. For Waris and Imonda there are oral accounts of their origin west of their current sites (Section 2 above). For Kilmeri we have linguistic data that put them in contact with the speakers of Nimboran and Sentani who nowadays live in a region (more than) 100 kilometers further west. In addition, we have the oral source of the clan genealogy over ten generations provided by my Kilmeri consultant Margaret Osi, who was married to the late clan leader Lis Osi and possesses remarkable knowledge of the clans’ past. The genealogy dates back the arrival of their ancestor Si in the Puwani-Pual basin to about 200–250 years ago, with Lis Osi’s lifetime as reference point (Gerstner-Link 2018:16–20). Assuming this oral account is historically reliable we get roughly 1800 AD as the date ante quem of contact between speakers of Kilmeri and Nimboran/Sentani. At the same time, about 1800 AD is also the date post quem at which the Kilmeri got in contact with the Sko speaking people. According to Donohue, the modern Skou trace their ancestors to the mountainous area to the south-east, that is, the western Oenake range. He assumes that Proto Macro-Sko speakers had lived in the Puwani-Pual basin before the intrusion of people speaking (one of) the languages of the Bewani branch of the Border family (Donohue 2004:5–6).
This migration pattern correlates with the relative chronology of external borrowing that can be ascertained based on sound correspondences and sound changes within the Border family. The contact between Border/Kilmeri speaking people and the Nimboran people must have been prior to the regular sound change from Waris / t,d/ > Kilmeri /r/, which is attested by a number of cognate pairs (see Appendix). The Nimboran forms show the same phonological pattern as the languages of the Waris branch, while Kilmeri is different: ‘child’ is du in Nimboran and tuɛndis in Waris, but ruri in Kilmeri; ‘tongue; mouth’ is méndʉ in Nimboran and mindɛ in Waris, but bɛr in Kilmeri; the wanderwort ‘tree’ is di in Nimboran and ti in Waris, but ri in Kilmeri. The same sound correspondence applies for the transfer of the distal deictic from Sentani into the Border languages. Sentani and Tabla (n)di- occurs as di in Waris, but as ri- in Kilmeri. Therefore this contact is also old.
Turning to Skou and Kilmeri, we see that they both contrast with Waris: ‘empty; hole, hollow’ is bí in Skou and bɪ in Kilmeri, but mɛ in Waris; ‘burn’ is rà li in Skou and rɛ in Kilmeri, but ta in Waris. This means that Skou borrowed the words after the sound changes took place that we observe between the Waris branch and Kilmeri, viz., Waris / t,d/ > Kilmeri /r/ and Waris /m/ > Kilmeri /b/ in syllable-initial position (see Appendix). Thus the contact is younger.
The above scenario is compatible with the eastward movement of the Kilmeri speakers in the past. From the greater Tami river area they (slowly) migrated to the east through the wide Bewani river valley, which connects to the Puwani-Pual basin. Somewhere during their journey they encountered the Skou people who had been forced or were then forced to leave their sites; when and where exactly this happened cannot be reconstructed. Only the people speaking I’saka retained their sites on the easternmost hills of the Oenake range, where they are the traditional and undisputed landowners (p.c. Simon Tapi of Krisa).19 The fact that I’saka is a first-order split from Macro-Skou is in line with an old and stable settlement.
However, the Kilmeri also came upon the speakers of Pagi. So far no oral accounts have emerged of the Pagi speakers’ clan history, former dwelling sites, or migrations. Linguistically, the sound correspondences between Waris, Kilmeri, and Pagi suggest that Kilmeri and Pagi underwent different phonological developments: for instance, Waris /t,d/ corresponds to Kilmeri /r/, but to Pagi /k/. This regular triple correspondence (see Appendix) can only be understood, if one goes back to Proto Border and tries to reconstruct a proto phoneme that governs all three language-specific developments. A good candidate would be *t. Then the Waris branch of the Border languages would be the conservative branch that retained the alveolar plosive, while Kilmeri and Pagi show independent innovations. Yet in other environments, Pagi still shows an old /t/ that corresponds to Kilmeri /l/ (Gerstner-Link 2018:31–35). The arrival of the Pagi in the Puwani-Pual basin probably predates that of the Kilmeri, since nowadays the Kilmeri live on better land while the Western and Eastern Pagi are found in minor, much more swampy places around Bewani in the west and Imbio/Imbinis in the east. This suggests land grabbing by the Kilmeri. The Pagi may have come from the south, thereby crossing the Bewani mountains, which must have been possible. The official map of the area shows two foot track routes from the Bapi valley to Bewani; there is also a foot track from Ossima to Kilifas (Jeffrey Osi, p.c.).
What exactly caused the Kilmeri to turn east in search for new dwelling and hunting sites can only be guessed. Whenever my consultant Margaret Osi and I talked about game and hunting, she raved about the golden hunting opportunities in earlier times, when her ancestors had arrived in Ossima and its vicinities. This might suggest that the Kilmeri people had been under economic-ecological pressure at their original places. It is known that, in the Upper Sepik region, over-hunting indeed caused people to move away in order to look for places providing better livelihood.20
9 Summary and Outlook
The linguistic history of regions for which written sources lack completely can at least be partially reconstructed. Comparison of the lexicon of several language families unveils non-inherited items that came to be shared by contact between their speakers. In addition to often attested types of meanings shifts among transferred words like contiguity and (visual) similarity of the designated items (Blank 1997), more abstract features known from noun categorisation devices (Aikhenvald 2000) could also be taken into account to uncover putative borrowings. The revealing of different types of structural transfer likewise points at some contact among languages. When these families and single languages are not located in proximity today, a history of migration is suggested whose relative chronology can be argued for by historical linguistics, viz., the discovery of sound changes that borrowed items have or have not undergone. In cases where such evidence is supported by oral tradition that tells of peoples’ distant origin and land grabbing in their current area, migration is the most plausible scenario. The successive structural transfer into Kilmeri resulted in a grammatical hybridisation of this language acquiring several new properties, while the other Border languages retained the inherited structural properties in question (cf. Section 8.2). So the language transitioned from its original convergence cluster of a minimal system of four pronouns into the more widespread group of number distinctive pronoun systems. By contrast, despite its acquisition of person marking for a special verb class, Kilmeri continues to be a member of the verbal number cluster of the area (Gerstner-Link 2018:383–385; Foley 2018:488–490). In sum, the dynamics of language change by contact is low with regard to the four language families here. While the results are still preliminary, a first step is done in understanding their common history, but much more needs to be investigated.
Acknowledgements
Many thanks to the organisers of the APLL 11 Conference in Leiden, June 2019, for inviting me, and to the participants for helpful comments. In particular I like to thank Timothy Asher, Don Daniels, Bethwyn Evans, Harald Hammarström, Marian Klamer, Godehard Link, and Peter-Arnold Mumm for their questions and discussions. Special thanks again to Marian Klamer and to an anonymous reviewer whose comments on drafts of the paper brought me additional insights that I didn’t have before. All errors are my own.—Finally, I feel gratitude to my late father Dr. Wolfgang Erich Gerstner to whom I owe so much of my funding.
Appendix
Border Family: Putative Cognate Sets and Sound Changes for Waris (Waris Branch), Kilmeri (Bewani Branch), and Pagi (Bewani Branch)
The direction of the sound changes is not easy to determine. Some correspondences suggest sound change from Waris to Kilmeri, e.g., Waris /nd,t/ > Kilmeri /r/. For this change and direction we have also areal support. More difficult are the correspondences Waris /ŋ/ ⟨ ⟩ Kilmeri /k/ and Waris /m,n/ ⟨ ⟩ Kilmeri /mb,nd/. Kilmeri lacks voiced velars, while all the other Boder languages possess these phonemes, so it is reasonable to assume that Kilmeri lost these sounds. But the issue of the voiced labials and alveolars is less clear. Universally, lenition is more frequently attested than fortition. Yet Kilmeri seems to show, in word-initial position, the development from nasals to prenasalised plosives, namely /m/ > /mb / and /n/ > /nd/, which sets it apart from the Waris branch of the Border family and also from Pagi. Actually, in current Kilmeri quite some words with initial /m/ show free varition with [m] and [mb] like musi ‘to shut’ as [musi] and [mbusi]. This supports word-initial occlusion.21 The observable sound changes occurred probably at different times under different phonological conditions and/or pressure. This can be exemplified as follows. Kilmeri /p/ has two Waris correspondences. In several instances Kilmeri /p/ also occurs as /p/ in Waris, e.g., ‘water’ K pu ⟨ ⟩ W pɔ, ‘betelnut’ K puɛl ⟨ ⟩ W pul, and ‘diarrhoea’ K ɛpɛr ⟨ ⟩ W ɛpɔnda. Here /p/ appears to be old. Yet in many other words it occurs as /
The sound changes that the above discussion of lexical transfer refer to are summarised in the following table. In the column Sound change, the first row gives the sound change from Waris to Kilmeri and the second row the change from Waris to Pagi. In many cases, the sound changes are positionally constrained. Curly brackets indicate (morphological) elements that are not part of the compared pair.
|
Sound change |
Meaning |
Waris |
Kilmeri |
Pagi |
|---|---|---|---|---|
|
t > r syl-initial |
‘tree’ |
ti |
ri |
ki |
|
t > k syl-initial |
‘feather’ |
tai |
rɛ |
kai |
|
‘wet’ |
puti-{l} |
puri |
||
|
d > r syl-final, intervoc |
‘child’ |
tuɛnd-{is} |
ruri |
kɔkɛi |
|
d > k syl-final, intervoc |
‘foot(print)’ |
dand |
dɔr |
nɔk |
|
‘tongue’ |
mindɛ |
bɛr |
mɛki |
|
|
‘dog’ |
windɛ |
wɔr |
wɔk |
|
|
‘netbag’ |
wɔnda |
urɔ |
uk |
|
|
‘diarrhoea’ |
ɛpɔnda |
ɛpɛr |
||
|
‘penis gourd’ |
pɛnda |
ʙɛr |
||
|
‘marsupial, possum’ |
pind |
{bi}_pɛr |
||
|
‘flat’ |
pund |
pur |
||
|
t > l syl-final |
‘sugarcane’ |
atxa |
ɛlɔ |
æth |
|
t > t(h) syl-final |
‘leech’ |
at |
al |
wat |
|
‘fish’ |
wal |
vit |
||
|
‘snake’ |
pial |
|||
|
m > b syl-initial |
‘stone axe’ |
mand |
buar |
mɔk |
|
m > m syl-initial |
‘tongue’ |
mindɛ |
bɛr |
mɛki |
|
‘leg, back limbs’ |
mɔ-{ŋala} |
bɔu |
mɔu-{l} |
|
|
‘saliva’ |
mius |
bɪs-{ɛp} |
||
|
‘pig’ |
miɛ |
bi |
||
|
‘hole, hollow’ |
mɛ-{l} |
bɪ |
||
|
‘sound, speech’ |
mɔa |
bɔ |
||
|
‘in-house fireplace’ |
{jua}-masa |
bæs |
||
|
‘dead body’ |
mind-{il} |
bir |
||
|
n > d syl-initial |
‘sago (swamp)’ |
nə |
duɛ |
na |
|
n > n syl-initial |
‘bush’ |
na |
du |
nɔ |
|
‘meat’ |
nix |
dʊ |
ni-{l} |
|
|
‘grass skirt’ |
nai |
diɛ |
||
|
‘eye’ |
nɔ |
dɔb |
nɔp-{ɔl} |
|
|
‘axe’ |
dawa |
nawa |
||
|
‘night’ |
dupuni |
nɔpuni |
||
|
ŋ > k syl-final, intervoc |
‘mouth’ |
mɛŋ |
mɛk |
|
|
Waris > Pagi ??? |
‘sago grubs’ |
mɛŋɛmb |
bɛkup |
|
|
‘wife’ |
ɔŋa-{l} |
akɔ |
||
|
‘to think’ |
nɛŋ-{ |
nɛki |
||
|
‘to lie flat; place of sth’ |
liŋi-{l- |
liki |
||
|
‘underneath’ |
{demus}-siŋi |
sikil-{jɔ} |
||
|
k > k all positions |
‘bone’ |
kəl |
kili |
ɛli |
|
k > ∅ syl-initial |
‘mosquito’ |
kles |
klɛs |
ɛlɛs |
|
pronoun 1SG |
ka |
kɔ |
a |
|
|
‘small frog’ |
kɛu |
kwɛr |
ɛtu |
|
|
‘chin’ |
kisi-l |
kæau |
||
|
‘fish scales, fish bones’ |
ku |
kisi |
||
|
‘headlouse’ |
kʊ |
|||
|
x > ∅ all positions |
‘thunder’ |
xul |
ul |
aunɔi |
|
x > ∅ all positions |
‘stomach’ |
ɛxna-{l} |
ɛni |
ni-{l} |
|
‘meat’ |
nix |
dʊ |
Elseng is claimed to be an isolate (Foley 2018:435–438). Based on the comparative method, there is good lexical and some paradigmatic evidence for its inclusion into the Border family (Gerstner-Link 2020, Ross 2005; Timothy Usher p.c.).
The “nearby plain cultures” belong to the vast cultural area of the Upper-Sepik and its tributaries whose western-most fringes they form (Craig 1980:2, 7).
Throughout the article, I use the following notational conventions: The vocabulary items are presented in Standard IPA. In doing that, the orthography of the original sources is transcribed into IPA in accord to each author’s spelling conventions. Morpheme boundaries are indicated by a hyphen. For composite lexemes I use the underscore to represent the bounderies between the parts. A consonant or vowel in round brackets represents an optional sound that is only realised in some languages of a family. Curly brackets indicate that a morpheme of a complex lexeme is not taken into account for comparison. A slash indicates lexeme variants within a language or a language family. The notation of tone in Skou and Wutung follows the conventions in Donohue (2004) and Marmion (2010).
In a few cases, Waris and Taikat reveal themselves as the recipient of foreign vocabulary.
Gasser (2019:673) also considers synonyms as a guide to detect borrowed forms.
In Kilmeri, light verb constructions are normally used with adjectives and nouns in order to verbalise them. Plausibly, the same strategy was formerly used to distinguish borrowed verbs from formally (almost) identical native verbs, here ɪ ‘to recede’. In current Kilmeri LV constructions are used to integrate Tok Pisin verbs.
This loan relationship may be indicative of the fact that, some centuries ago, the Kilmeri people were forest dwellers who lived on hunting and only gradually developed a horti-cultural life style as the Sentani practise around Lake Sentani.
Of course, noun classification and meaning shift in language history are two fields. But a conceptual overlap should not be excluded a priori. Note that Kilmeri ‘rice’ is dipsu from dipi_su ‘ant_egg’, which is clearly a calque based on the shape of rice grains, that is, on the concept of shape.
A look at the other documented Skou languages shows the following: Wutung has qa ‘to hit’, qbaqba ‘to hit’, qaqwa ‘kill.1SG > 3SG.M’ (Marmion 2010:374) as well as lô ‘sharp’ and láíqè ‘sharp’ (2010:372). Here it seems that lô ‘sharp’ is a contact-induced second adjective, conveying a meaning already present, that goes back to the Border stem for ‘shoot, hit’. I’saka has -a ‘hit’ and -o ‘shoot’ (Donohue and San Roque 2004:95). Barupu has ti ‘to shoot’ (Corris 2005:388). These vocabulary findings support the loan origin of Skou lú ‘shoot’.
The Skou family words for ‘good’ are as follows: Donohue gives héfèng ‘good’ (2004:525). Smits and Voorhoeve (1994) attest efe/hèfè/hè:pè ‘good’. For Wutung we find félàì ‘good, nice’ and muti ‘good’ (Marmion 2010:370; 373). I’saka has èi ‘good’ (Donohue and San Roque 2004). In Barupu ‘good, be good’ comes in the two variants neman/nevai (Corris 2005:381).
Foley mentions two examples of language contact on the northeast coast of New Guinea that resulted in quite a number of loanwords that belong to the basic vocabulary, be it among genetically related languages like Watam and Kopar of the Lower Sepik family or be it among an Austronesian and a Papuan language like Mangap-Mbula and Kovai (2010:799). Cf. also van den Heuvel and Fedden (2014:32–33). Gasser (2019) examined Austronesian loans in Papuan languages of the Bird’s Head and the Cenderawasih-Bay.
Tabla bu and Sentani pu occur in nominal collocations like doi bu ‘sweat’ and roi pu ‘sweat’ (Gregerson and Hartzler 1987:10). This shows that bu/pu are conventionalised, compositionally productive lexemes in these languages.
In Northwest New Guinea, a number of Papuan languages borrowed their words for ‘leaf’ from Austronesian languages in their vicinities (Gasser 2019:651; 654), yet in her sample ‘leaf’ belongs to the least borrowed items (2019:635).
There may be unidentified items of transfer among the languages under investigation.
According to their numbers of loanwords, Tadmor (2009:57) classifies languages in “very high borrowers” (> 50 %), “high borrowers” (25–50 %), “average borrowers” (10–24 %), and “low borrowers” (< 10 %). The documented Kilmeri lexicon comprises roughly 800 words/stems; in the present study 19 instances of loans into Kilmeri are identified (13 words from Nimboran, 6 from Sentani). So, with 2,3 % extra-family loans from vernacular languages, Kilmeri looks like a low borrower regarding those sources. In a similar magnitude, Ross identifies 1,7 % loan words from Bargam into Takia, plus 0,6 % from Waskia, which makes for 2,3 % Papuan loans in sum (2009:758). Gasser also reports very low rates of loan involvement for a number of Papuan languages in her sample (2019:637).
My consultants were reluctant to touch this topic because of the tensions between the native Papuan population and the Indonesian military; there were OPM (“Organisasi Papua Merdeka”, Organisation for a Free Papua) activities in the area including the Papua New Guinea side of the state border (see also Marmion 2010:31). The Kilmeri people’s reservation towards this political subject continued to hold over the years. So I refrained from asking questions.
A perfect instance of indigenous multilingualism is the following: Considering the Papuan loanwords found in the Oceanic language Takia of the Bel family (Karkar island, Madang province), Ross assumes that the Takia speaking people used to be bilingual in their language and coastal mainland Bargam and maybe more languages (Ross 2009:764).
Note that the folk taxonomic class membership of specific kinds is semantically not isomorphic, but language-specific for Kilmeri and Skou.
According to Donohue (2004:6), the modern Skou people are faced with the lack of substantial, undisputed land holdings, and with ongoing disputes about compensations for transmigration lands between the Skou and Elseng (2004:14).
In his introductory article presenting results of the Upper Sepik-Central New Guinea Project Craig (1980:9) writes: “Another tradition, reported in 1968 by informants at Bibiyun on the mid-August River, is that the Yimnai originally lived in the Simaiya valley, east of the Idam valley. They exhausted the supply of game—mainly wallaby—and moved west, near to present-day Bisiaburu on the Idam; part of that group then moved up the August River to present-day Bibiyun, and to Buliap on the Sepik within West Papua.”
The phonological change of occlusion is quite rare, but it systematically occurs in the Kaure [Nawa River] family (Timothy Asher, p.c.; see Introduction, Figure 8.1).
References
Aikhenvald, Alexandra Y. 2000. Classifiers. A Typology of Noun Categorization Devices. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. 2008. “Language Contact along the Sepik River, Papua New Guinea.” Anthropological Linguistics 50:1, 1–66.
Anceaux, Johannes C. 1965. The Nimboran language. Phonology and Morphology. ’s Gravenhage: Martinus Nijhoff.
Blank, Andreas 1997. Prinzipien des lexikalischen Bedeutungswandels. Tübingen: Niemeyer.
Brown, Bob 1990. Waris grammar sketch. Ms. Ukarumpa: Summer Institute of Linguistics.
Brown, Bob and Honoratus Wai 1986. Diksenari. A short dictionary of the Walsa (Waris) language, Tok Pisin and English. Ms. Ukarumpa: Summer Institute of Linguistics.
Burung, Wiem 2000. A brief note on Elseng. Unpublished manuscript. SIL International Electronic Survey Reports 2000–001: Summer Institute of Linguistics International.
Collier, Kenneth and Kenneth J. Gregerson 1985: “Tabla verb morphology.” Papers in New Guinea Linguistics 22, 155–172.
Corris, Miriam 2005. A grammar of Barupu: a language of Papua New Guinea. PhD Thesis, University of Sydney.
Cowan, H.K.J. 1965. Grammar of the Sentani language. ’s Gravenhage: Martinus Nijhoff.
Craig, Barry 1980. “Introduction to the Legends of the Abau of Idam Valley and of the Amto of Simaiya Valley, West Sepik Province, Papua New Guinea.” First published in Oral History 8, Nrs 4 and 5.
Donohue, Mark 2002. “Which Sounds Change: Descent and Borrowing in the Skou Family.” Oceanic Linguistics 41:1, 171–221.
Donohue, Mark 2004. A grammar of the Skou language of New Guinea. Ms. published online via eSICDOC MPG. Accessed September 2020 and October 2021.
Donohue, Mark and Melissa Crowther 2005. “Meeting in the middle: Interaction in North-Central New Guinea.” In Andrew Pawley, Robert Attenborough, Jack Golson, and Robin Hide (eds.), Papuan pasts: cultural, linguistic and biological histories of Papuan-speaking peoples. Canberra: Australian National University, Research School of Pacific and Asian Studies. 15–65.
Donohue, Mark and Lila San Roque 2004. I’saka. A sketch grammar of a language of North-Central New Guinea. Canberra: Pacific Linguistics 554.
Foley, William 2010. “Language Contact in the New Guinea Region.” In Raymond Hickey (ed.), The Handbook of Language Contact, Oxford: Wiley-Blackwell. 795–813.
Foley, William A. 2018. “The languages of Northwest New Guinea.” In Bill Palmer (ed.), The languages and linguistics of the New Guinea area. Berlin/Boston: De Gruyter Mouton. 433–567.
Gasser, Emily 2019: “Borrowed Color and Flora/Fauna Terminology in Northwest New Guinea.” Journal of Language Contact 12, 609–659.
Gell, Alfred 1992. “Barter and gift exchange in old Melanesia.” In Caroline Humphrey and Stephen Hugh-Jones (eds.), Barter, Exchange and Value. An Anthropological Approach. Cambridge: Cambridge University Press.
Gerstner-Link, Claudia 2019. “Loans into and from Kilmeri as indicators of the people’s migration route.” Paper presented at the 11th International Austronesian and Papuan Languages and Linguistics Conference Leiden.
Gerstner-Link, Claudia 2020. “Elseng as a member of the Border family.” Ms. University of Munich.
Gerstner-Link, Claudia 2018. A grammar of Kilmeri. Pacific Linguistics 654. Berlin/Boston: De Gruyter Mouton.
Gerstner-Link, Claudia. A dictionary of Kilmeri. Work in Progress. To appear in Martin Haspelmath and Barbara Stiebens (eds.), Dictionaria. Leipzig.
Gerstner-Link, Claudia 2000. Pagi fieldnotes.
Gregerson, Kenneth J. and Margaret Hartzler 1987. “Towards a reconstruction of Proto Tabla-Sentani phonology.” Oceanic Linguistics 26:1/2, 1–29.
Hartzler, Margaret 1976. “Central Sentani phonology.” Irian: Bulletin of Irian Jaya Development 5:1, 66–81.
Haspelmath, Martin 2009. “Lexical borrowing: Concepts and issues.” In Martin Haspelmath and Uri Tadmor (eds.), Loanwords in the World’s Languages—A Comparative Handbook. Berlin/New York: De Gruyter Mouton. 35–54.
Haspelmath, Martin and Uri Tadmor 2009. “The Loanword Typology project and the World Loanword Database.” In Martin Haspelmath and Uri Tadmor, Loanword’s in the World’s Languages—A Comparative Handbook. Berlin/New York: De Gruyter Mouton. 1–34.
van der Heuvel, Wilco and Sebastian Fedden 2014. “Greater Awyu and Greater Ok: Inheritance or Contact?” Oceanic Linguistics 53:1, 1–36.
Juillerat, Bernard (ed.) 1992. Shooting the Sun. Ritual and Meaning in the West Sepik. Washington / London: Smithsonian Institution Press.
Juillerat, Bernard 1996. Children of the Blood: Society, Reproduction and Cosmology in New Guinea. Oxford: Berg Publishers.
Kouwenhoven, Willem J.H. 1956: Nimboran. A study of social change and social-economic development in a New Guinea society. Doctoral Thesis. Leiden.
Marmion, Douglas E. 2010. Topics in the Phonology and Morphology of Wutung. PhD dissertation. Canberra: Australian National University.
Matras, Yaron 2009. Language contact. Cambridge: Cambridge University Press.
May, Kevin R. 1997: A Study of the Nimboran Language. MA thesis. Melbourne: La Trobe University.
Menanti, Jackie 2005. Sociolinguistic report on the Elseng language in Sia-Sia village, Keerom county, Papua, Indonesia. Unpublished ms. Ukarumpa: Summer Institute of Linguistics.
Miller, S.A. 2017. “Skou Languages Near Sissano Lagoon, Papua New Guinea.” Language and Linguistics in Melanesia 35, 1–24.
Minch, Andrew 1992. “Amanab grammar essentials.” Data Papers on Papua New Guinea Languages 39, 100–173. Ukarumpa: Summer Institute of Linguistics.
Muysken, Pieter 2010: “Scenarios for Language Contact.” In Raymond Hickey (ed.), The Handbook of Language Contact. Oxford: Wiley-Blackwell. 266–281.
Ross, Malcolm 1980: “Some Elements of Vanimo, a New Guinea Tone Language.” Papers in New Guinea Linguistics No.20. Pacific Linguistics A—56, 77–109.
Ross, Malcolm 2009. “Loanwords in Takia, an Oceanic language of Papua New Guinea.” In Martin Haspelmath and Uri Tadmor (eds.), Loanwords in the World’s Languages—A Comparative Handbook. Berlin/New York: De Gruyter Mouton. 747–770.
Ross, Malcolm 2013. “Diagnosing Contact Processes from their Outcomes: The Importance of Life Stages.” Journal of Language Contact 6, 5–47.
Ross, Malcolm 2014. “Reconstructing the history of languages in Northwest New Britain. Inheritance and contact.” Journal of Historical Linguistics 4:1, 84–132.
Seiler, Walter 1985: Imonda, a Papuan Language. Pacific Linguistics B-93. Canberra: Australian National University.
Smits, Leo and Clemens L. Voorhoeve (eds.) 1994. The J.C. Anceaux Collection of Wordlists of Irian Jaya Languages B: Non-Austronesian (Papuan) Languages. Part II. Irian Jaya Source Material 10 Series B 4. Leiden-Jakarta: DSALCUl/IRIS.
Tadmor, Uri 2009: “Loanwords in the world’s languages: Findings and results.” In Martin Haspelmath and Uri Tadmor (eds.), Loanwords in the World’s Languages—A Comparative Handbook. Berlin/New York: De Gruyter Mouton. 55–75.
Tadmor, Uri, Martin Haspelmath and Bradley Taylor 2010: “Borrowability and the notion of basic vocabulary.” Diachronica 72:2, 226–246.
van der Wilden, Jaap and Jelly van der Wilden 1975: “Kemtuk phonology.” Irian: Bulletin of Irian Jaya Development 4:3, 31–60.
van der Wilden Jaap 1976: “Simplicity and detail in Kemtuk predication.” Irian: Bulletin of Irian Jaya Development 5:2, 59–84.
Voorhoeve, Clemens L. 1971: “Miscellaneous notes on languages in West Irian.” Papers in New Guinea Linguistics 14, 47–114.
Voorhoeve, Clemens L. 1975: Languages of Irian Jaya Checklist: Preliminary Classification, Language Maps, Wordlists. Pacific Linguistics. Canberra: Australian National University.
Winford, Donald 2003. “Contact and Borrowing.” In Raymond Hickey (ed.), The Handbook of Language Contact. Oxford: Wiley-Blackwell. 170–187.
Wirz, Paul 1934 [1928]. “Beitrag zur Ethnologie der Sentanier (Holländisch Neuguinea).” Nova Guinea 16. Brill.
Wohlgemuth, Jan 2009. A Typology of Verbal Borrowings. Berlin/New York: De Gruyter Mouton.