Musa River

Timothy Usher, Santa Fe Institute

The Musa River family, also known as Yareban, is comprised of five closely related languages spoken in the vicinity of the middle and upper Musa river and its westerly tributaries in the southeast corner of Papua New Guinea's Oro (Northern) Province and just across the border in Central Province (Dutton 1975: 643, Weimer and Weimer 1977: 1-2, Weimer 1978: 1.) Its nearest relatives are Bauwaki and the Cloudy Bay family to the southeast (q.v. Strong 1910-1911: 211-213.)

Proto-Musa River shares with Cloudy Bay a number of Oceanic loans in the basic vocabulary, indicating a common origin near the south coast. Loans from the Oro and Managalas Plateau families are also found. The Central Musa River subgroup has been significantly influenced by Baruga of the Oro family.

The internal classification of Musa River is as follows:

Musa River

Moikodi

Aneme Wake

Central Musa River

Nawaru

Yareba

Bariji

Lexical similarities as well as shared phonological changes suggest Moikodi and Aneme Wake to form a southern subgroup. Grammatically, Moikodi stands apart, suggesting perhaps an original split of Moikodi against the others, followed by a period of interaction and codevelopment with Aneme Wake.

Dutton (1971: 14-15) gives lexicostatistics for four Yareban languages, excluding Nawaru, as spoken in eight villages as follows:

Anonymous (1895-1896) vocabulary of Musa River (Yareba)

Ray (1907: 387-412) …

Strong (1910-1911: 203-217) 113 comparative terms for Saroa (Bariji,) Bori (Yareba) and Oiwa (?Aneme Wake)

Armit (1913-1914: 190-193) 447 comparative terms for Buari (Aneme Wake)

Liston-Blythe (1914-1915: 175) 160 comparative terms for Kororo (Bariji)

Hooper (1917-1918: 87) 127 comparative terms for Moikoidi or Doriri (Moikodi)

Hooper (1918-1919: 107) 123 comparative terms for Boli (Yareba)

Bastard (1918-1919: 109) 125 comparative terms for Doriaidi (Aneme Wake)

Ray (1938) (pp. 155-157) locations … Abia group (Aneme Wake) and Upper Musa groups …

Weimer (1963) Yareba texts

Weimer (1972) Yareba verbal morphology

Weimer and Weimer (1974) Yareba dictionary

Weimer and Weimer (1974) Yareba folklore

Weimer and Weimer (1975) Yareba grammar

Weimer and Weimer (1977) Yareban phonologies including (pp. 6-11, ibid.) 140+ comparative terms for Moixodi, Aneme Wake, Sirio (Nawaru,) Yareba and Aga Bereho (Bariji)

Weimer (1978) Yareban comparative grammars

Weimer (1992) sketch phonology of Yareba

Weimer (1992) sketch phonology of Moikodi

Weimer (1992) sketch phonology of Aneme Wake

Dutton (1971: 12-14) general descriptions of Yareban languages

Dutton (1975: 643-645) general descriptions of Yareban languages

Proto-Musa River had 11 to 14 consonants and 5 vowels as follows:

Voiceless stops /*p *t *k/ were generally realized as aspirated [*pʰ *tʰ *kʰ], leading to aspirated reflexes in several descendants; this trend is shared with the unrelated Guhu-Oro family to the north.

Phonemic laminal /*s/ is not native to Musa River, but may have been established by the time of the protolanguage. Both Yareba and Bariji have a phoneme /dz/ which has likewise been acquired from neighboring families (q.v. Weimer and Weimer 1977: 3.)

Voiced velar stop /g/ has been observed only initially, while glottal /ʔ/ is heard only root-medially. These might, then, be thought allophones of a single phoneme /*g/, and indeed medial /ʔ/ is the reflex of Cloudy Bay-Musa River /*g/; however Cloudy Bay-Musa River initial /*g/ yields zero, or conceivably /*ʔ/ [ø], while all Musa River words with initial /*g/ are loans.

Apical non-stop /ɾ/ occurs only medially.

Otherwise, any consonant can occur initially or medially. Consonants do not occur word-finally. Stem-final consonants exist underlyingly on verbal roots, which are mandatorily suffixed. Consonant clusters are known to occur only across morpheme boundaries on inflected verbs.

Vowels may occur in any sequence.

Stress is generally penultimate. Counterexamples indicate segments which have been lost or compounds.

Consonants correspond as follows:

These correspondences are exemplified as follows, with all attestations drawn from Weimer and Weimer (1974 ibid., 1977: 6-11, ibid.) unless otherwise indicated. Those drawn from Strong (1910-1911: 203-217) are indicated by [ws], those from Armit (1913-1914: 190-193) by [la], those from Liston-Blythe (1914-1915: 175) by [alb], those from Hooper (1917-1918: 87, 1918-1919: 107) by [gh] and those from Bastard (1918-1919: 109) as [eb]. Loans from Oceanic are included here under the assumption that they were already established prior to the breakup of Musa River.

[under construction]

Bilabial nasal /*m/ is retained as such in all descendants:

Apical nasal /*n/ is retained as such in all descendants:

Bilabial voiceless stop /*p/ is fricated to /ɸ/ in all descendants, then deoccluded to /h/ in Bariji (Weimer and Weimer 1977: 3.) Weimer's (1992: 1) sketch phonology gives Aneme Wake as both /β/ and /f/, presumably in error for /ɸ/. In Armit's Buari it is given as <p>:

There are four reasons that voiceless stop /*p/ is reconstructed rather than fricative /*ɸ/. First, it patterns structurally with aspirated /*t *k/ against voiced /*b *d/, while the only other fricative /*s/ is probably secondary, and is probably better viewed as an aspirated laminal, as it is in Oro. Second, Weimer and Weimer (1974: 7) specify that some Yareba speakers realize this sound as [p]. Third, [ɸ] in Oro languages reflects /*p/ [pʰ]; this is probably true also of Koiari-Managalas Plateau, which otherwise lacks a /*p/. Finally, verbs with final /*t/, such as /*ut/ “do,” provoke a change in following /*b/ to [ɸ], which makes somewhat more sense if frication followed devoicing.

Apical voiceless stop /*t/ is merged with the reflexes of velar /kʰ/ (below) in Moikodi and Aneme Wake, to be fricated to /x/ in Moikodi:

When /*t/ is preceded by high front vowel /*i/, it is palatalized and fricated to /s/ in Moikodi and Aneme Wake:

When /*t/ is followed by /*i/, it is merged with the reflexes of /*k/ in Moikodi and Aneme Wake and palatalized and fricativized to /s/ in Nawaru and Yareba and deoccluded to /h/ in Bariji:

This loan from Oro is likewise affected, though here Bariji has [s]:

Accordingly, there are only two examples of Yareba initial segment /tʰi/ in Weimer and Weimer's dictionary (q.v. 1974: 260.)

When /*t/ is both preceded and followed by /*i/, it is palatalized and fricativized to /s/ in Aneme Wake as well as in and Nawaru and Yareba and deoccluded to /h/ in Bariji:

Phonemic laminal fricative /*s/ can occur in any vocalic context. Some of the following are known to be loans (below):

(n.b. it's possible that several medial examples actually reflect /*t/i_i.)

Overall, the impression is that phonemic /*s/ was introduced into proto-Musa River from neighboring families, with some examples of apparent /*s/ reflecting native /*t/ when both preceded and followed by /*i/.

Velar voiceless stop /*k/ is fricated to /x/ in Weimer and Weimer's Moikodi (1977: 3.) In Hooper's Moikodi it is given as <k>:

Bilabial voiced stop /*b/ is retained as such in all descendants:

Apical voiced stop /*d/ is generally retained as such in all descendants:

Apical /*d/ is palatalized and affricated to /dz/ in Yareba and Bariji when following high front vowel /*i/:

Voiced velar /*g/ occurs almost exclusively initially. It is retained as such in all descendants:

One example of medial /*g/ appears to be a reduplication:

Our assumption is that all of these words are loans dating from a period later than that of loans which observe the loss of Cloudy Bay-Musa River initial /*g/.

Glottal stop /*ʔ/ occurs only medially, in complimentary distribution to /*g/. It is lost in Nawaru and Yareba and retained sporadically in Moikodi and Aneme Wake:

Excepting /*uʔa/ “louse” and /*kaʔu/ “yam”, all known examples of glottal stop are preceded and followed by homophononous vowels. This might be assumed to result from a rule which levels these vowels, or one in which erstwhile glottal stop, reflecting Cloudy Bay-Musa River /*g/, was dropped in other vocalic environments.

Bilabial non-stop /*w/ is realized as glide [w] or occluded to voiced fricative [β]. In Yareba, this follows the pattern of the Oro languages, with [β] occuring before front vowels /i e/ and [w] elsewhere (Weimer and Weimer 1974: 8, Weimer 1992: 3.) In Armit's Buari /w/ is given as <v> even when followed by non-front vowels /a o/:

Apical non-stop /*ɾ/ occurs only medially. It is retained as such in all descendants:

Palatal non-stop /*j/ is retained as such in all descendants:

Unlike bilabial /*w/, palatal /*j/ is not occluded, and the distinction between segments such as /*ia *ija/ isn't obvious; for example /*waija/ or /*waia/ “garden” (q.v. Weimer and Weimer 1974: 284, 1977: 4.)

Final consonants on verbs … :

Musa River free pronouns are reconstructed as follows (q.v. Weimer 1978: 14):

There are reasons besides their conspicuous phonological parallelism to consider the third persons and the first person plural exclusive and dual, as well as historically the first person plural inclusive, to share a demonstrative enclitic /*-wa/, an element also found on /*ewa/ “this” (cf. /etʰa/ “that”.) The Yareba third person singular possessive free form and prefix is /da(-nu)/, not /dawa(-nu)/, while the separability of /*ewa *eta/ is strongly suggested by /e'no/ “like this.” This would explain final stress as seen in Yareba /(i)'ja/, /e'wa/, /(u')wa/, /da'wa/. These suffixes are also found on corresponding Bauwaki and Cloudy Bay first person plural and dual forms /*gewa *guwa/ as well as /*iwa/ first person singular which, as Musa River /*na/ is original to Owen Stanley Range, must be an innovation and likely reflects the plural inclusive. Here as well /*-wa/ is dropped in possessive forms, thus Magi free forms /ia gea gua/ but possessives /i-na ge-na gu-na/ (q.v. Thomson 1975: 643.)

[under construction]

Musa River subject desinences on final verbs are reconstructed in seven tense forms as follows:

These are based upon the forms attested in Weimer and Weimer (1974: 493-501, Weimer 1978: 5-8.) It should be noted that, as Weimer and Weimer's example verb was /*ut/ “do,” underlying initial /*b/ in irrealis suffixes is realized as [*ɸ]; these are given here with reconstructed /*b/ in the paradigms which follow:

Past … The only difference between the past and near past in Aneme Wake is the retention of the stem final consonant in the near past. This might be explained in a number of ways, for example through the loss of a suffix-initial consonant, here shown speculatively as /*j/:

Near past…

Present …

Imperative …

Future …

Desiderative …

Benefactive … . The second plural form has been replaced by the third plural in Central Musa River:

Same-subject medials are reconstructed as follows (q.v. Weimer and Weimer 1974: 488, 1977: 13, Weimer 1978: 11-12;) as the subject is indicated on the final verb, there is no distinction in person:

[under construction]

….

… Oro … . Several terms have undergone changes indicating a direct origin from Baruga:

Additionally …:

… Managalas Plateau …:

… Doromu-Maria …:

… Oceanic …: