Kamula-Elevala River

Timothy Usher (Santa Fe Institute) and Edgar Suter

Edgar Suter and Timothy Usher 2017. “The Kamula-Elevala language family” Language and Linguistics in Melanesia 35: 106-131.

The Kamula-Elevala River family consists of at least three languages, Aekyom (Awin,) Pa (Ba, Pare) and Kamula (Kamura,) spoken across the lowlands south of New Guinea's central cordillera and east of the Fly River in Papua New Guinea's Western province.

The internal classification of Kamula-Elevala River is as follows:

Kamula-Elevala River

Kamula

Elevala River

Aekyom

Pa

Aekyom is spoken in at least four dialects, the northeastern dialect being the most divergent and possibly representing a distinct language (q.v. Depew 1986: 12-14;) however the data currently available to us is too scant and disorganized to arrive at a definitive conclusion in this respect.

Austen (1921-1922: 161) 109 comparative terms for Amnat on the Eastern side of the Tedi (Alice) River

Ray (1923: 339) 2 pronouns, (p. 345) 5 numerals and (p. 354-356) 19 comparative terms for Amnat after Austen (1921-1922)

Austen (1924-1925: 75) 38 comparative terms for Awin of Eastern Tedi River, Awin of Western Fly River near Palmer Junction, Awin of Western Donaldson Range and for Upper Fly River (mix of Aekyom and Faiwol) near the Star Mountains, 597 miles from the mouth of the Fly

Champion (1926-1927: 117) 55 comparative terms for Plain Country between the 500-mile mark on Palmer (Luap) river and the foot of the mountains

O'Connor (1936-1937) (unobtained)

O'Connor (1937-1938) (unobtained)

Rule and Rule (1970) grammar of Aekyom (unobtained)

Rule and Rule (1990) phonology of Kamula (unobtained)

Rule and Rule (1990) grammar of Kamula (unobtained)

McElhanon and Voorhoeve (1970) includes Awin and Pare comparisons

Voorhoeve (1975: 389-390) grammar sketch of southwest Awin (Aekyom) between Kiunga and Rumginae and (390-391) grammar sketch of Pa

Voorhoeve (2007) 210 comparative terms for Awin, North Awin, West Awin, East Awin, Minumin, Ba and East Ba

Shaw (1971) comparative vocabulary of Debepare of Wakiyana village

Franklin (ed. 1973: 590) 97 comparative terms for Pa after Shaw

Shaw (1986: 68) 99 comparative terms for Kamula and (p. 70) 100 comparative terms for Pare

Reesink (1976: 31-34) 97 comparative terms for Kamula

Depew (1986) ethnography of Upper Fly River Aekyom

Stewart (1989) phonology and grammar sketch of Lower Maeri River Aekyom

Stewart (2016) dictionary of Aekyom

Routamaa (1994) grammar of Kamula

Routamaa and Routamaa (1995) phonology for Kamula of Kesiki village

Routamaa and Routamaa (1996) dialect survey of Kamula

Routamaa and Routamaa (1997) orthography for Kamula

Routamaa (1997) Kamula events and participants

Routamaa (1997) Kamula tail head linkage

Routamaa and Routamaa (2002) sketch phonology of Kamula

Routamaa and Routamaa (2007) dictionary of Kamula

Waldeck (2016) 459 comparative terms for Pa

Waldeck (2017) 133 additional comparative terms for Pa

Suter ed. (2017) comparative vocabularies for Kamula, Pa and Aekyom after Routamaa and Routamaa (2007,) Waldeck (2016, 2017) and Stewart (2016)

In addition to these, Paul Whitehouse provided comparative vocabularies of Aekyom of Karana and Kawandere villages via the Summer Institute of Linguistics Ukarumpa; however they are undated and unattributed.

[under construction]

Ray (1923: 336) places Austen's (1921-1922: 161) Amnat in a Tedi Group which is otherwise equivalent to the family later known as Lowland Ok.

Greenberg (1971: 825-826) divided the attestations of Aekyom dialects between Lowland Ok and Mountain Ok, apparently upon the basis of their inclusion in Austen (1924-1925: 75) and Champion (1926-1927: 117) respectively, and upon a number of loans into Aekyom from both Ok subfamilies. Although this placement is broadly consistent with Healey's (1964) proposal, the failure to recognize that Austen's and Champion's vocabularies represent a single language suggests Greenberg's placements to result from oversight. Greenber's roundup of Annual Report vocabularies is usually quite thorough, but in this instance neglects to mention Austen's (1921-1922) Amnat.

Healey (1964: 108, 115, 115-116) … “The Awin-Pare Family of 9,000 speakers is a tentative grouping by the present author, and includes Akium-Awin (Aekyom), Akium-Pare, and Pare (Pari).” On p. 110 he states that initial consonant clusters /kr tr pr gr dr sr sm/ occur in Awin-Pare, as do many diphthongs. On page 111 he states that Aekyom has a body part counting system up to 35 (i.e. apex is 18) but Pare has a quinary system using the fingers and toes. He doesn't give any pronouns or verbal morphology. On page 115 he gives a lexicostatistical resemblances between Awin and Pare at 36% and above, substantially lower than Voorhoeve's figures. The most important passages are on page 115-116: “Awin shows about 10% possible cognates with the Ok Family and with the Awyu-Dumut Family, but more detailed investigation is necessary before regular sound correspondences can be established.” That's the same figure he estimates between Awyu-Dumut and Ok, hence, “It is the author's impression that a close study of the Awyu materials would demonstrate regular sound correspondences between the Awyu (or Awyu-Dumut) proto-forms and Ok proto-forms. It may eventually prove possible to demonstrate that the Ok, Awin-Pare, and Awyu-Dumut Families belong to a single linguistic stock or phylym, or perhaps that there was extensive borrowing between the parent languages of these families.”

Voorhoeve (1968: 1, 3-4, 6, 1975a, 1975b) … (1968: 1) Awin-Pare group after Healey (1964) … (p: 3) South and Central New Guinea Stock … (p: 6) Awin-Pare division of Pare-Samo-Beami-Bosavi Family, division level: 85% (74%) for 82 terms between West Awin (Pampenae near Ningerum) and East Awin (Drimska near Palmer Junction (Fly River ),) 51% (41%) for 101 terms between West Awin and Ba, 45% (36%) for 102 terms between East Awin and Ba, Family level: 36% (32%) for 153 terms between Samo and Ba, 25% (18%) for 112 terms between Beami and Ba … (pp. 7-8) Stock-level: 25% (21%) for 148 terms between Kati and Ba, 30% (24%) for 132 terms between Teléfól and Ba, 23% (19%) for 142 terms between Kiwai and Ba … (pp. 10-11) Awin and Ba are from Voorhoeve's own elicitations … McElhanon and Voorhoeve (1970: 10) the Awin-Pare Family is given as separate group alongside the Bedamini (Beami) Family, which includes both Strickland River and Papuan Plateau, in the larger Central and South New Guinea Stock, within the larger Central and South New Guinea Phylum which is then related to McElhanon's Finisterre-Huon Phylum to form the Trans-New Guinea Phylum … Voorhoeve (1975: 346-348, 388-391) Awin-Pa Family within Central and South New Guinea Stock of Trans-New Guinea Phylum …

The first known mention of Kamula in the literature was in Capell (1962: 129, 1969,) who showed a language called Kamura in the jungle between the village in which Kamula is spoken today but otherwise provided no data.

The first published materials for the Kamula language appeared in Reesink (1976: 13-18) who wrote (p. 15): “Efforts to relate this unknown language to languages of the Lake Murray area (see Voorhoeve 1970b) or languages to the north were futile. Comparing Kamula material with the 53 items McElhanon and Voorhoeve (1970) list in their Trans New Guinea Phylum points to the direction of languages of the Ok and Bedamini Families. But these relationships are indeed very remote. More promising seemed a comparison with Pa (Franklin 1973: 590, Appendix J.4) which suggests a relationship on Stock level (see Chart 10). This language was included in the Pare-Samo-Beami-Bosavi-Family by Voorhoeve (1968) on the basis of 36% shared cognates with Samo. But more recent comparative work by Shaw (1973) puts this figure down to 12%, which leaves only a Stock level relationship with the East-Strickland Family of the Bosavian Stock. A correction, in this direction was already made in McElhanon and Voorhoeve (1970), where Pa (included in Pare) and Awin formed a separate family within the Central and South New Guinea Stock.” … (pp. 16-17) 37 Kamula-Pa comparisons …

Wurm (1982: 131, 132, 136-138,) following Voorhoeve (1975a, 1975b,) Reesink (1976) and Rule (p.c.), gives Kamula as a member of the Central South New Guinea Stock but not of Awin-Pa. Wurm's logic here was likely that Reesink had proved a reasonably close relationship with Pa, but not that it is a member of Awin-Pa. Reesink (1976: 15) gives an upper limit of 20% lexicostatistical relationship between Kamula and Pa, compared to Voorhoeve's (1968: 6) 42%-50% between Awin and Ba. Meanwhile Voorhoeve's (1968: 7-8) stock level figures range from 14%-32%, where Reesink's number would fit right in. Moreover, Kamula's pronouns do not show the degree of similarity as do those of Awin and Pa as presented in Voorhoeve (1975: 388-391,) nor are Kamula verbs marked for the number of their subjects, providing typological or criterionalistic reasons to have rejected an immediate relationship.

Shaw (1973, 1986) … (1973: 192-193, 195-196) lexicostatistical figures between 9& and 13% between Pa, Strickland River, Papuan Plateau and Duna-Bogaia… “Finally by comparing all the groups of the East Strickland Plain with the Pa, inhabiting the region west of the Strickland River from the Cecelia River south almost to Lake Murray, an average of 12% lexicostatistic relationship is noted. This is considerably lower than the 36% figure presented by Voorhoeve (1968), and may result from our unfamiliarity with phonetic shifts and other factors which result in relationships not apparent by simple inspection.” … (1986) Kamula member of Bosavi Watershed alongside Bainapi and the Mount Bosavi languages … (p. 53) … 22% between Kamula and Pare, 41% between Kamula and Bainapi, 38-55% between Kamula and Mount Bosavi languages, Pare 18% with Bainapi, 19-23% with Mount Bosavi languages … pp. (50-51) 17 selected lexical comparisons between Kamula …

Stewart (1989: 1) writes of Aekyom, “Related languages are Pa to the S.E., Konai to the east, and Awyu in Irian Jaya.” Konai is a member of the basically unrelated Strickland River family, but the inclusion of Awyu in this list is uncanny. Perhaps Stewart had seen Healey's (1970) reconstruction of proto-Awyu-Dumut and noticed some of the very strong similarities in the basic lexicon, but, as she presents no evidence in support of this statement, nor does the paper include citations or a bibliography, it's difficult to say why Awyu was included in particular was included.

Routamaa and Routamaa (Routamaa 1994: 7, Routamaa and Routamaa 1995: 3, 1997: 2.) write in review, “Wurm (1981) classifies Kamula as a family level isolate of the Central and South New Guinea stock and superstock of the Trans New Guinea Phylum. Shaw (1986) proposes that Kamula be placed in the Bosavi family and in the Bosavi Watershed subfamily. Lexically and grammatically, our investigations to date show very few similarities with the Bosavi family languages and so Wurm's classification is possibly more accurate.” While it's true that Wurm didn't place Kamula directly with Bosavi Watershed (Mount Bosavi, Papuan Plateau) as would Shaw a few years later, Wurm's order of presentation, which begins with “Bosavi, Kamula, East Strickland, Awin-Pa,” carrries forward Voorhoeve's (1968: 6) concept of a Bosavi regional group which Shaw (1986) continued and the dichotomy between the two arrangements is overstated.

Our findings here support Reesink's (1976) contention that Kamula belongs with Pa, but per Wurm (1982) is not quite as close to Pa and Aekyom as they are to one another. In the bigger picture, we confirm Healey's (1964) suggestion that, together with Kamula about which nothing was known at the time, the Elevala River languages belong with the Awyu-Dumut (Digul River) and Ok families in a larger group, while contra Voorhoeve (1968,) Wurm (1982) and Shaw (1973, 1986) there is only a distant relationship with Bosavi (Papuan Plateau) and even less with East Strickland (Strickland River,) which is itself not closely related to Bosavi.

Ross …

[under construction]

Proto-Kamula-Elevala River had 11 consonants and at least 7 vowels as follows:

In addition to the simple vowels given above, at least one diphthong is found as follows:

What are shown here as voiceless and voiced biliabial stops /*p *b/ might have been reconstructed as /*h/ and /*p/ respectively. …

There is scant evidence for a rounded velar voiceless stop /*kʷ/ … (below) …

…

… nasalization … Routamaa and Routamaa (1995: 23) state that Kamula nasalization, which is not indicated in their practical orthography, is not exclusively but very conspicuously associated with adjacent /ɺ/, in keeping with its dual origin from apicil nasal /*n/ and voiced stop /*d/. We can tentatively posit that there are two causes of nasalization in Kamula-Elevala River languages, that which results from the reinterpretation of segments /*nV/ as /ɺṼ/ and final nasalized vowels /Ṽ/, which reflect earlier final nasal consonants /*m *n/ that were reduced to a nasal archiphoneme /*N/ by proto-Elevala River as they were in proto-Awyu and proto-Awbono-Bayono as well as the more distantly related but geographically adjacent Papuan Plateau family to the northeast, some members of which have likewise reinterpreted segments /*nV/ as /ɺṼ/.

Excepting nasal archiphoneme /*N/, neither final consonants nor consonant clusters occurå.

Stress is contrastive, as is discernible in Aekyom where destressed nuclear vowels are dropped leading to consonant clusters and destressed final vowels are dropped when preceded by nasals /*m *n/ (below.)

Initial consonants correspond as follows:

Medial consonants correspond as follows:

Vowels and diphthongs correspond as follows:

These correspondences are exemplified as follows, with Kamula attestations drawn from Routamaa and Routamaa (2007,) Aekyom from Stewart (1989, 2016,) Pa from Waldeck (2016, 2017) and Awin and Ba (Pa) drawn from Voorhoeve (1975: 389-391, 2007.)

Any consonant can occur initially.

Initial bilabial nasal /*m/ …:

As initial /*m/ is regularly realized as /b/ in Aekyom (above,) there are only a swall number of Aekyom words with initial /m/ in Stewart's dictionary (2016: 20.) Several of these, /mom/ “nephew, maternal” ([monaⁱ] “cassowary” is underlyingly /mom-naⁱ/, see Depew 1986: 150-154, 349 for the semantics,) /mon/ “rubbish” and (p. 10) /mgat-ɺam/ “mouth” are loans from Mountain Ok (below,) others are loans from Motu (/moɺu/ “medicine”) or English (/mosoɺu/ “boat (powered with a motor),” while still others are of unkown origin.

A handful of examples in which an Elevala River original /*m/ appears to have been retained as /m/ in Aekyom are marked here with the symbol /*m₂/. The first of these may be a loan into Aekyom from Pa or into both from an unidentified third source, or the retention of /*m/ is governed by a rule which is not understood:

In two examples, Aekyom's retention of initial /m/ may be due to the their tendency to appear between other words and clauses:

Initial apical nasal /*n/ is uncommon. It is retained as such in all descendants:

Initial bilabial voiceless stop /*p/ is deoccluded to /h/ in all descendants:

Initial apical voiceless stop /*t/ is retained as such in all descendants. Routamaa and Routamaa (1995: 5) specify that Kamula /t/, unlike /n/ and /d/, is dental:

In one example, assuming cognacy between the Kamula and Aekyom forms, Kamula /t/ is seemingly irregularly answered by Aekyom /d/. This might be explained as assimilation to following /*m/ in a derived initial cluster (below):

Initial laminal voiceless /*s/ is uncommon. It is retained as such in all descendants:

Initial velar voiceless stop /*k/ is dropped in Kamula as is voiced /*g/ (below.) It is retained in Aekyom and in Pa, with Aekyom /k/ undergoing intermittent aspiration to [kʰ] in parallel with the initial reflexes of voiceless /t/ (above) and voiced /b/ (below):

Initial bilabial voiced stop /*b/ is devoiced to /p/ in all descendants, with Aekyom /p/ undergoing intermittent aspiration to [pʰ] in parallel with the initial reflexes of voiceless /t k/ [tʰ kʰ] (above.) Routamaa and Routamaa (1995: 4) state that Kamula /p/ is optionally realized as fricative [ɸ]:

Initial apical voiced stop /*d/ is retained as such in all descendants. As with initial velar /*g/ (below,) it is occasionally realized as prenasalized [ⁿd] in Aekyom::

Initial velar voiced stop /*g/ is dropped in Kamula as is voiceless /*k/ (above.) It is retained in Aekyom and in Pa. As with initial apical /*d/ (above,) it is occasionally realized as prenasalized [ᵑg] or even as nasal [ŋ] in Aekyom:

Initial bilabial non-stop /*w/ is uncommon. It is retained as such in all descendants:

Initial palatal non-stop /*j/ is uncommon. …:

Initial /*j/ … Cental Digul River /*je̝n/ “tree/fire” …:

… /*j/ …:

It is probably the case that any consonant could occur medially. The expected medial disposition of bilabial voiceless stop /*p/ is not yet known; presumably it was either merged with the medial reflexes of voiced /*b/ or first deoccluded to [h] as are its initial reflexes above and then dropped altogether. Similarly, there is no clear trace of a medial /*g/, which perhaps has merged with /*k/ or some other sound, or disappeared.

Medial bilabial nasal /*m/ is generally retained as such in all descendants:

In several verbs stems with a final syllable /*mV/, Kamula …:

Medial apical nasal /*n/ is generally retained as such in Aekyom and denasalized to non-stop /ɺ/ in Kamula and in Waldeck's Pa. Voorhoeve's (2007) Ba and Shaw's (1986: 70) Pare often give [n] rather than [ɺ] suggesting /n/ to have been outright retained or, perhaps more likely, synchronically an allophone of /ɺ/ following now-nasalized vowels. Routamaa and Routamaa (1995: 23) state that Kamula nasalization, which is not indicated in their practical orthography, is overwhelmingly associated with adjacent /ɺ/, in keeping with its dual origin:

/*n/ …:

In at least one example, root-medial /*n/ …:

… /*n/ …:

… /n₂/ …:

Medial bilabial voiceless stop /*p/ almost certainly existed, but it's not entirely clear whether it was deoccluded to [h] (as are intiials above) and then dropped or merged with the medial reflexes of voiced /*b/ [p b] (below):

Root-medial apical voiceless stop /*t/ is lenited to non-stop /ɺ/ in Pa, in parallel with the lenition of /*k/ to [g ɣ] (below.) As with initial /*t/ (above,) Routamaa and Routamaa (1995: 5) specify that Kamula /t/ is dental::

Pa medial [t] indicates a compound or a loan:

Medial laminal voiceless /*s/ … the only known example of Pa medial /s/ with a clear outcomparison, /wasi/ “war,” is probably a loan from Mountain Ok (below):

Medial velar voiceless stop /*k/ is lenited to voiced [g] in Voorhoeve's Ba and Shaw's Pare and to fricative [ɣ] in Waldeck's Pa, in parallel with the lenition of medial /*t/ to [ɺ] (above) :

… /*k/ …:

… /*k/ …:

Medial bilabial voiced stop /*b/ is devoiced to /p/ in Aekyom and in Kamula. Most likely it was devoiced in pre-Pa as well before leniting in parallel with voiceless stops /t k/ (below,) a process which likewise affects Voorhoeve's Awin. As with initial /*b/ (above,) Routamaa and Routamaa (1995: 4) state that Kamula /p/ is optionally realized as fricative [ɸ], while Reesink's (1976: 2) Kamula [dɾaβa] “fire” (cf. Routamaa's /dɺapa/) suggests that this fricative can be voiced as well:

… /*b/ …

Aspiration in Aekyom word-medial /p/ [pʰ] (< /*b/) indicates a morpheme boundaty, in this instance due to reduplication:

One example … /p/ [pʰ] …:

Medial apical voiced stop /*d/ is lenited to apical non-stop /ɺ/ in Aekyom and in Kamula and dropped altogether in Pa:

Because the medial reflexes of /*n *d/ are distinguished only in Pa and in Aekyom only intervocalically, the medial consonant in one root attested only in Kamula and in Aekyom is indeterminate between /*n/ and /*d/:

Velar voiced stop /*g/ has not been found to occur medially.

Medial bilabial non-stop /*w/ …:

Medial palatal non-stop /*j/ …:

… /*j/ …:

This process of alternation is synchronically ative in Pa, as is demonstrated in Waldeck's variant attestations of roots including [siɣa ~ sija] “skin/bark” and [kiɣɛ ~ kijɛ] “son”.

… nasal archiphoneme /*N/ … nasaliztion …

… vowels and diphthongs …

High front vowel /*i/ is generally retained as such in all descendants:

In a number of Aekyom verbs, the presence of /*i/ in the root is discernible by the palatalization of following present tense suffix /-ɺa -ɺaⁱ/ to [-ɺʲa ɺʲaⁱ] (q.v. Stewart 1989, n.d.: no page number):

…:

… /*aCi/ …:

… /*aCi/ …:

…:

… /*uCi *o̝Ci/ …:

… /*uCe̝/ …:

High back rounded vowel /*u/ … [u ʉ] …:

High-mid front vowel /*e̝/ …:

… /*iCe̝/ …:

… /*e̝/ …:

…:

High-mid back rounded vowel /*o̝/ is merged with the reflexes of high back rounded /*u/ in Stewart's Aekyom, but Voorhoeve's materials show the distinction between /*o̝/ [o u] and /*u/ [u ʉ] on the one hand and /*ɔ/ [o] on the other to be synchronically needed for at least some Aekyom dialects. Waldeck's Pa … Routamaa's Kamula …:

Low-mid back rounded vowel /*ɔ/ …:

In one example where /*ɔ/ is followed by bilabial non-stop /*w/, it is given as [u] in Waldeck's Pa; compare the backing and rounding of /*a/ followed by /*w/ (below):

Because the reflexes of low-mid back rounded /*ɔ/ are reliably distinguished from those of high-mid /*o̝/ only in Aekyom (above) and by outcomparison, a number of roots attested only in Kamula and in Pa have vowels which are indeterminate between /*o̝ *ɔ/:

Low front vowel /*æ/ …:

Monosyllabic roots of the form /*æ/ become /e/ in Routamaa and Routamaa's Kamula:

… /*æ/ …:

… /*æ/ …:

Low central vowel /*a/ …:

… /*a/ followed by /*w/ …:

… /*aCæ/ …:

This phenomenon is synchronically visible in Waldeck's Pa where several roots with final low central /a/ are followed by locative suffix /-pæ/ [-bæ] or the adjective-deriving suffix /-kæɺæ/ [-xæɺæ -kæɺæ]:

…:

… /*iCa/ … see also Pa /kjæ-xɔ/ “nettle” which is a loan from Papuan Plateau / (below):

… /*iCa/ …:

Front rising diphthong /*aⁱ/ …:

… stress …

Aekyom … initial clusters … :

… monosyllabic verb roots /*CV-/ … Aekyom … using present tense /-ɺa/ (< /*-da/) for illustration:

Aekyom is unique among Kamula-Elevala River lagnuages in allowing either of two final consonants, nasals /m n/. In both instances these became final when destressed final vowels were dropped:

Stressed final vowels were not dropped:

… reduction …:

[under construction]

… with Aekyom attestations drawn from Stewart (1989: 49-50, 2016,) Awin (Aekyom) from Voorhoeve (1975: 389-390,) Pa from Voorhoeve (1975: 390-391) and Waldeck (2016, 2017) and Kamula from Routamaa (1994: 14-15, 45, Routamaa and Routamaa 2007):

… absolutive subject …:

… possessive and active or object …:

… possessed object …:

… emphatic subject …:

… inalienable possessor …:

[under construction]

…

[under construction]

…

[under construction]

Aekyom… Mountain Ok and Lowland Ok sufamilies of Ok …

(Mountain Ok tone is contrastive but is not indicated here.)

At least one of these is found also in Pa:

Austen's (1924-1925: 75) Upper Fly River vocabulary includes a number of Mountain Ok and Lowland Ok words not found in other attested varieties of Aekyom, to the point where it appears to be a mixed language (or a corrupt elicitation):

(Mountain Ok and Lowland Ok “tobacco” must be loans as well.)

(The term for “cucumber” is also given for Austen's Awin of Western Donaldson Range.)

… “shoulder” …:

… ultimately from Central Mandobo via Muyu, with Kati I (Katì Metòmka, i.e. South Muyu) and Kaeti (Central Mandobo) attestations drawn from Drabbe (1959: 170-171.) The reason for assuming this to be a loan rather than a legitimate retention of Central Digul River /*waki/ is that low central vowel /*a/ has been raised to in Central Mandobo due to the influence of high front /*i/ in the next syllable (q.v. Healey 1970: 1004, 1007-1008, 1047,) which is not known to be a rule in Aekyom:

… Strickland River …

… with Konai from Arsjö (2016: 111):

The term for “paddle” may well come from Strickland River despite its presence in Aekyom and reconstructability for Kamula-Elevala River, with Odoodee attestation drawn from Hays and Hays (2016: 47):

By far the least adulterated Kamula-Elevala River language is Pa … Pa initial /b/ cannot occur in roots descending from proto-Kamula-Elevala River …

A good number of Kamula words are shared with Dibiyaso (Bainapi) of the Papuan Plateau family and/or the Soari River languages, Doso and Turumsa, to the south, with Dibiyaso, Doso and Turumsa attestations drawn from Rueck, MacKenzie and Alemán (2005: 8-20, Dibiyaso after MacKenzie and Carr, Doso and Turumsa after Carr):

… Papuan Plateau, probably via Dibiyaso …:

… Fasu from May and Loeweke (1981):

… Aramia River … with Gogodara (Gogodala,) Adiba and Waruna attestations drawn from Riley (1930-1931: 172-189):

In addition to these, Routamaa and Routamaa (2007) identify a number of other Kamula words as being loans from Gogodala where we have no direct attestation in Aramia River available to us. It may be observed that most of these terms together with those above refer either to crop cultivation or to waterfaring, in opposition to the Kamula's traditionally inland hunting and gathering mode of subsistence:

As we have only survey vocabularies for these southern languages, it is probable that there remain many more loans between them and Kamula to be found. An estimate of how many might be projected from the very large number of entries in Routamaa and Routamaa's (2007) Kamula dictionary with initial stops /b k g/, which cannot occur in roots descending from proto-Kamula-Elevala River (nor is initial apical non-stop /ɺ/ found in native words, but all four examples with this are loans from English.)

Several terms found in Aekyom and Pa, and thus reconstructed for proto-Kamula-Elevala River, might suggest that Aramia River languages were once spoken well to the northwest of their current range:

… Mount Bosavi … with Kaluli drawn from Schieffelin and Feld (1998) and Onobasulu from Dondop (2007):

… Fly River (Anim) … (Usher and Suter 2015: 128-129):

… Fly River (Anim) …:

… Fly River (Anim) …:

… “crocodile” … Fasu from May and Loeweke (1981)…:

The Kamula word for “dog” is suffixed with diminitive /-jemaɺa/ and derives ultimately from Malayo-Polynesian /*asu/ (Blust.) It was introduced to the Morehead region alongside a number of other Central Malayo-Polynesian terms, where the Morehead River languages regularly add velar nasal /*ŋ/ to roots beginning with low vowels /*æ *a/. This /*ŋ/ was reinterpreted by neighboring languages as a prenaasalized or plain voiced stop /*ŋg *g/ and it spread throughout the region (Usher and Suter 2015: 131;) /*g/ was then regularly dropped (above) in Kamula:

… “tobacco” … here the failure of medial /k/ to lenit to voiced fricative [ɣ] in Pa indicates a loan rather than a compound:

… “sweet potato” …:

Loans from Motu (Petterson 1999: 112, p.c) are found in all three languages: