Lake Murray

Timothy Usher, Santa Fe Institute

The Lake Murray family, also known as the Boazi family, is comprised of two languages, Kuni-Boazi and Zimakani, spoken around Lake Murray and south and west to the middle Fly River in Papua New Guinea's Western Province.

Murray (1916-1917: 49) 57 terms for Lake Murray (Zimakani)

Austen (1921-1922: 163) 69 comparative terms for Village Y (Koumak)

Austen (1921-1922: 163) 47 comparative terms for Lake Murray, Village Z (Zimakani,) Village Y and Village X (Kuni-Boazi,) with Lake Murray terms drawn from Murray (1916-1917: 49)

Rentoul (1924-1925: 76) 106 comparative terms for Koumak

Boelaars (1950: 55-59) English language recension of Drabbe's unpublished grammar of Boazi

Drabbe (1954: 118-127) grammar and (pp. 128-142) 418 comparative terms for Boazi

Voorhoeve (1970: 7-14) 242 comparative terms for Begua and Zimakani dialects of Zimakani, and South Boazi and Kuini dialects of Kuni-Boazi

Fumey (2005) Kuni absolutives (unobtained)

Fumey (2006) sketch phonology of Kuni

Fumey (2007) grammar of Kuni

Proto-Lake Murray had 14 to 16 consonants and at least 5 vowels as follows:

There is no apical non-stop except for the lateral reflex [l] of initial voiced stop /*d/ (below.)

Excepting velar /*g *ŋg/, which occur in all positions, voiced stops and prenasalized stops are in complimentary distribution, with anterior voiced stops /*b *d/ being found only initially and prenasalized /*mb *nd/ only medially, excepting several marginal examples of initial and final apical /*nd/ (below.) Medial /*mb/ has only been found in one root. These are distinguished here due to parallelism with the velars, which are unambiguously contrastive. This assymetrical system may be visualized as follows:

Final consonants are very common. Consonant clusters do not occur.

Any vowel can occur initially, medially or finally, but the cooccurance of different vowels is restricted by vowel height harmony such that a high vowel /*i *u/ and a mid vowel /*e *o/ cannot appear in the same root. The only known exceptions to this rule are vowel sequences /*oi *ou/. Low central vowel /*a/ is harmonically neutral, and can cooccur with any other vowel.

Initial consonants correspond as follows:

Medial consonants correspond as follows:

Final consonants correspond as follows:

Vowels and vowel sequences correspond as follows:

Zimakani appends final vowels to consonant-final roots as follows:

These correspondences are exemplified as follows, with Zimakani, Begua, South Boazi and Kuini attestations drawn from Voorhoeve (1970: 7-14), Boazi from Drabbe (1954: 118-127, 28-142,) and Kuni from Fumey (2006, 2007, p.c.) Voorhoeve's transcriptions are phonetic; Drabbe's and Fumey's are phonemic.

Bilabial nasal /*m/ is generally retained as such in all descendants except finally, where it is sporadically dropped in Voorhoeve's South Boazi:

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

Bilabial voiceless stop /*p/ is retained as such in all descendants. The only known example of final /*p/ may be a loan from Marind:

Apical voiceless stop /*t/ is retained as such in all descendants:

Velar voiceless stop /*k/ is retained as such in all descendants. It is typically realized as uvular [q] when followed by low contral or mid back vowels /*a *o/ or finally following /*a *o/:

One puzzling correspondence found only in free pronouns is represented here by <[*k]>. It might be thought to be a suffix in Zimakani, but no suffix of this form is attested in Drabbe's Boazi or Fumey's Kuni. A similar discrepancy is found in the finals of Lower Fly River free pronouns, where it is likewise tentatively considered to be phonological rather than morphological in origin:

Bilabial voiced stop /*b/ is found only initially. It is retained as such in all descendants:

Apical voiced stop /*d/ is found only intially. It is predominantly realized as a lateral non-stop [l] in modern Kuni (Fumey 2006: 6) and in Zimakani:

Velar voiced stop /*g/ is retained as such in all descendants:

Bilabial prenasalized stop /*mb/ has been found only medially in one root:

Apical prenasalized stop /*nd/ is primarily found medially, where it is retained as such in all descendants:

Initial and final /*nd/ exist, but are marginal, the first being a loan from Arammba which is also found in Kumbe River Marind (q.v. Drabbe 1955: 150-151,) the second an apparent reduplication which does not add a final vowel in Zimakani as do other nominals (below.) These are distinguished from initial voiced stop /*d/ not only by prenasalization but also the failure to lenit to lateral non-stop [l] (above):

Velar prenasalized stop /*ŋg/ occurs in all positions. When followed by low central and mid back vowels /*a *o/, it is typically realized as uvular [ŋɢ] or fricated to [ŋɣ] in Kuni-Boazi, in parallel with the backing of /*k/ when followed by the same vowels (above):

Bilabial or labiodental fricative /*f/ is retained as such in all descendants. Initially, it is sometimes fortited to affricate [pf]:

Initial and medial fricative /*s/ is retained as such in all descendants. Initially, it is sometimes fortited to affricate [ts]:

Final /*s/ is retained as such in Kuni-Boazi and appears to have been lost in Zimakani, without the copy vowel which follows other final consonants (below.) However, there is no direct confirmation of the Zimakani correspondences, as the only known examples in which final /*s/ is supported by outcomparison, /*sas/ “tongue” and /*vis/ “go up”, do not have Zimakani attestations in our dataset:

Bilabial non-stop /*w/ is occluded to labiodental [v] in all descendants:

Palatal non-stop /*j/ is occluded to [z dz dʒ] in all descendants, with Rentoul's (1924-1925: 76) Koumak (not shown here) giving [dʒ]:

Velar non-stop /*ɣ/ is found only following low central and mid back vowels /*a *o/ and finally following /*a/, and is retained as such in all descendants. It is often realized as uvular [ʁ]:

Zimakani adds a final vowel to consonant-final roots. The quality of the appended vowel is determined by the vowel in the preceding syllable, with mid and low /*e *o *a/ yielding copy vowels and, somwhat unexpectedly, /a/ following high front /*i/ and /i/ following high back /*u/:

There are several counterexamples in which Zimakani unexpectedly has final high front vowel /i/ following final consonants with preceding low central /*a/, with /*a/ modified to sequence /ai/. These might be thought to reflect Fly River postposed gender particle /*-i/ which, if so, retains some measure of independence in Zimakani:

Lake Murray vowels /*i *u *o *e *a/ conform to vowel harmony restrictions such that a high vowel /*i *u/ and a mid vowel /*e *o/ do not cooccur in the same root. Low central vowel /*a/ is neutral, and can cooccur with any vowel. These are generally retained as such in all descendants. In order to illustrate the constraints upon vowel cooccurences imposed by vowel harmony, roots including only high vowels /*i *u/ are presented first, followed by those including only mid vowels /*e *o/, followed by those containing only low vowel /*a/ and then roots containing /*a/ and any other vowel(s).

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

High back vowel /*u/ in simple environments is retained as such in all descendants:

Cooccurence of high vowels /*i *u/ in a single root is rare. The only known example of the sequence /*iCu/ is a loan from Arammba, and shows assimilation of /*i/ to following /*u/:

One example of the sequence /*uCi/ is supported by outcomparison to Inland Gulf /*muki/:

Mid front vowel /*e/ in simple environments is retained as such in all descendants:

Mid back vowel /*o/ in simple environments is retained as such in all descendants:

There is only one known example in which /*e/ cooccurs with /*o/ in what looks likely to be a compound:

Low central vowel /*a/ in simple environments is retained as such in all descendants:

One example has Zimakani [ɑⁱ] but is not consistent with the sequence /*ai/ (below,) either in its Kuni-Boazi reflex or in Zimakani's final vowel. Here final /*i/ might be thought to be a particle (above):

Unlike high vowels /*i *u/ and mid vowels /*e *o/, low central /*a/ is harmonically neutral, and may be followed by any vowel:

Similarly, /*a/ may be be preceded by any vowel:

Observed vowel sequences include /*ia *ui *ua *ea *oi *ou *oa *ai *aia *au *aua *ae/.

The sequence /*ia/ is retained as such in all descendants:

The sequence /*ui/ is merged with /u/ in Zimakani:

The sequence /*ua/ may be heard as [wa]. It is retained as such in all descendants:

The sequence /*ea/ is merged with /e/ in Zimakani and is sometimes simplified to /a/ in Kuni (q.v. Fumey 2007: 12):

The sequences /*oi/ is retained as such in Zimakani and in Drabbe's Boazi:

The sequence /*ou/ is retained as such in all descendants. It is conceivable that it is underlyingly identical with /*au/ (below) which overlaps its range:

The sequence /*oa/ is merged with /o/ in Kuni:

In the sequences /*ai/ and /*aia/, the first /a/ is often fronted to [e ɛ]. There is no discernable pattern which might justify a distinction between /*ai *aia/ and /*ei *eia/:

One example suggests that the sequence /*au/ is merged with /*ou/ in Kuni-Boazi; however, due to the similar dispositions of /*ai *aia/ (above) as well as those of /*aua/, this is considered here to be secondary:

The sequence /*ae/ is retained as such in all descendants:

Any vowel sequence can be preceded or followed by low central /*a/:

One sequence is followed by /*o/:

[under construction]

…:

Lake Murray free pronouns are reconstructed as follows,with Zimakani, Begua, South Boazi and Kuini attestations drawn from Voorhoeve (1970: 7-14), Boazi from Drabbe (1954: 119, 142) and Kuni from Fumey (2007.) Voorhoeve's transcriptions are phonetic; Drabbe's and Fumey's are phonemic:

The third persons are indicated by demonstratives, the bases of which are equal to the postposed vowels which induce gender ablaut (above):