Guam River

Timothy Usher, Santa Fe Institute

The Guam River family consists of at least three languages spoken along and around the Guam and Sogeram rivers, major tributaries of the Ramu river to the west, in the Middle Ramu and Bogia subdistricts of Papua New Guinea's Madang province. It is called the Tamolan Family in the works of Z'graggen, Tamol being the name of an abandoned village that was used as a reference point by the United States Army during the second world war (Z'graggen 1971: 80-81, Brooks 2018: 85-86.)

The internal classification of Guam River is as follows, with Z'graggen's language names in parentheses where they differ from current nomenclature:

Guam River

Inapang

Midsivindi

Itutang

Yigavesakama

Chini-Iski

Chini (Akrukay)

Iski

Breri

Romkun

Kominimung and Igana are listed as Tamolan languages in Z'graggen (1972, Laycock and Z'graggen 1975: 739-740) but no vocabularies were elicited. Also listed in Z'graggen (1972) is Isarikan, which according to Z'graggen (1971: 81) is a village in which Itutang is spoken.

Z'graggen's Midsivindi and Itutang are so similar to one another that tt might be asked whether they are dialects of the same language, Inapang. Brooks (2018: 33, 76) asserts that Inapang is a dialect continuum and Midsivindi and Itutang are among its members, as is another dialect Yigavesakama for which he provides a small number of terms (p. 87) and some example sentences (pp. 81. 83.)

Breri and Romkun are also very similar to one another. Brooks (2018: 33, 76) denies that they are distinct languages and states that Iski is the name that the Breri call their own language, in which /iski/ means “no;” this is also the term for “no” in Romkun (q.v. Z'graggen 1972.)

Z'graggen (1971: 81) gives lexicostatistical relationships between languages of the Tamolan Family as follows:

Z'graggen (1971: 80) asserts that Akrukay belongs with Itutang-Midsivindi against Breri-Romkun, but there is not much lexical evidence to this effect. Brooks (2018: 78-79) affirms the validity of Z'graggen's Tamolan while suggeting that Chini (Akrukay) “in its current form is not simply the result of a clean branching off from an ancestor language, but rather a result of some sort of admixture between the Inapang and Breri sides of the subgroup in ways that match the geographical location of the respective groups.” We find no support for this besides that Chini sometimes follows Iski-like sound changes and sometimes doesn't, but these changes look to be regular.

Z'graggen (1969, 1971: 80-83) brief descriptions of and (pp. 90-91) 18 comparative terms for Itutang, Midsivindi, Akrukay, Breri and Romkun

Z'graggen (1972) 431 comparative terms for Itutang, Midsivindi, Akrukay, Breri and Romkun

Z'graggen (1974) typewritten version of Z'graggen (1972) with some systematic omissions of detail

O'Rear (1992) Inapang dialect survey (unobtained)

Brooks (2018) verbal morphology for Chini of Andamang village including (p. 67) 25 lexical comparisons to Breri and Inapang

In addition to these, survey vocabularies of seven Inapang villages designated as Misivindi, Mutungu, Ingawaia, Bogen, Ayat, Ongi and Porpor were provided to us in digitalized form by Paul Whitehouse via the Summer Institute of Linguistics, Ukarumpa; however they are undated and unattributed. A comparison of Misivindi with Brooks' (2018: 87) excerpts from O'Rear (1992) confirm these to be drawn from the same source.

[under construction]

Proto Guam River had 15 consonants and 4 vowels as follows:

Even compared to other subfamilies of Ramu River, Guam River stands out among New Guinean families in having almost no discernible initial consonants, a situation reminiscent of some families of Australia's Cape York where erstwhile initial consonants have likewise been regularly dropped.

Initial bilabial nasal /*m/ is found as the first element in Z'graggen's Akrukay and Itutang attestations for the third person dual, which we presume to reflect Ramu River third person singular /*ma/ in contrast to the free form third person singular /*an/ < /*man/. There are two other examples of apparent initial /*m/. One, /*m₂ʉrʉk/ “meat”, is a loan from the Tomui River subgroup of West Madang. Another, /*m₂aru[a]p/ “ancestral spirit” is found in all Ramu River subfamilies, but its correspondence is similarly irregular in Ramu Coast, Ruboni Range and Moam River, most likely indicating a loan postdating the breakup of all these subfamilies. Still the example of /*ma/ appears to demonstrate that initial consonants can be preserved in a very small number of roots with the simple form /*CV/.

There is one apparent example of initial apical non-stop /*r/ in /*rapar/ “leaf,” where /r/ is retained only in Midsivindi; this corresponds to initial /*r/ in other subfamilies of Ramu River. This is strongly suspected of being in error or (possibly) a loan from Igo-Tanggu in which the retention of intitial /*r/ is regular.

Any consonant can occur medially.

Only a restricted set of consonants occurs finally:

The majority of native roots are of the form /*VCVC(VC)/. Final vowels are not found on roots with more than two syllables. There are a small number of vowel-final roots /*VCV/. Two pronouns are monosyllables /*VC/. Consonant clusters do not occur.

Medial consonants generally correspond as follows:

Final consonants generally correspond as follows:

Clusters involving medial and final consonants are formed as follows:

These correspondences are exemplified as follows, with all attestations drawn from Z'graggen (1971, 1972.) For clarity of presentation, the general dispositions of initial consonants will presented first, then those of medial consonants, then those of final cononants. Some details of the forms which follow are necessarily tentative.

Generally, Goam River had no initial consonants as Ramu River initial consonants have invariably been lost. There is one very salient exception: the third person singular pronoun /*ma/, though its suffixed form /*ma-n/ dropped initial /*n/ just like any other word (beliow.) We assume that /*ma/ was exempted from initial dropping because it is a simple form /*CV/ whiere the great majority of Ramu River roots are multisyllabic, The original third person /*ma/ is visible indirectly in Z'graggen's attestations for the first person dual:

A particle /*m-/ is found prefixed to a number of transitive verbs. It is distinguished from the reflexes of medial /*m/ (below) because it does not undergo preocclusion to [mb] in Iski and Inapang regardless of the quality of the following consonant. Brooks (2018: 107) characterizes this prefix in Chini as distal deictive /mɨ-/ with a high central vowel, though the vowel is overridden by the initial vowel of the root where one synchronically exists, and states that it is found also on nominals as an inanimate topic marker:

Besides these, the only known examples of apparent initial /*m/, /*m₂ʉrʉk/ “meat” and /*m₂aru[a]p₂/ “ancestral spirit”, are loans, the first from the Tomui River subgroup of West Madang, the second a regional term, found also in (at least) Monumbo of the Torricelli-Sepik Coast family, in which the retention of final [*p] in Guam River languages is also irregular (below.) These roots are included here even though they may not have been present in proto-Guam River because they help illustrate other phonological trends. This initial is designated /*m₂/ as a reminder of its origin and in keeping with its designation in other subfamilies of Ramu River. Phonologically, it behaves identically to medial /*m/ following a dropped initial vowel (below):

… /*ndz-/ … Brooks (2018: 78, …) glosses as Chini middle voice …:

One example iof an apparently apical non-stop /*r/ which is retained only in Midsivindi is most probably a loan into Midsivindi from Igom-Tanggu to the north:

Any consonant can occur medially.

Medial nasals /*m *n *ŋ/, where immediately followed by a non-nasal consonant in the onset of the next syllable or by a final vowel are occluded to [*mb *nd *ŋg] in all descendants except Akrukay. Where immediately followed by another nasal either in the onset of the next syllable or finally, occlusion does not occur. This development is shared with Kire of the Ruboni Range subfamily to the north and with Atemple of the Sogeram River subfamily of West Madang.

Medial bilabial nasal /*m/ where immediately followed by a non-nasal consonant in the onset of the next syllable or by a final vowel is occluded to prenasalized stop [mb] in all descendants except Akrukay. These occluded reflexes remain distinct from those of /*mb/ in Itutang and Midsivindi, where /*mb/ is devoiced to [mp] (below):

One verb … :

When medial /*m/ is immediately followed by a nasal /*m *n *ŋ/ either in the onset of the next syllable or finally, occlusion does not occur:

Medial apical nasal /*n/ where immediately followed by a non-nasal consonant in the onset of the next syllable or by a final vowel is occluded to prenasalized stop [nd nt] in all descendants except Akrukay:

When medial /*n/ is immediately followed by a nasal /*m *n *ŋ/ either in the onset of the next syllable or finally, occlusion does not occur:

When medial /*n/ is preceded by high front vowel /*i/, it is palatalized to [ɲ], then to be occluded or not in all descendants except Akrukay according to the same rules as outlined above. These occluded reflexes remain distinct from those of /*ndz/, which are typically devoiced in all descendants (below):

When palatalized medial /*n/ [ɲ] is immediately followed by a nasal /*m *n *ŋ/ either in the onset of the next syllable or finally, occlusion does not occur:

…:

In two examples where medial /*n/ is followed by /*i/, it is palatalized but not occluded, even though there is no nasal consonant following final /*i/. It is likely that these words were originally compounds:

…[ɲ] …:

Medial velar nasal /*ŋ/ when it is not followed by another nasal consonant /*m *n *ŋ/ is occluded to prenasalized voiced stop /ŋg/ outside of Akrukay. It has not been found either preceded or followed by high fton vowel /*i/:

In one example, medial /*ŋ/ is assimilated to apical [n] in Breri and Romkun when in a cluster with the reflex of final /*s/ (below):

… /*ŋ/ …:

When medial /*ŋ/ is immediately followed by a nasal /*m *n *ŋ/ either in the onset of the next syllable or finally, occlusion does not occur:

… /*ŋ/ …:

Medial bilabial voiceless stop /*p/ is retained as such in all descendants:

Medial apical voiceless stop /*t/ is generally retained as such in all descendants:

When /*t/ is followed by velar nasal /*ŋ/ …:

As with all cluster formation, this does not occur in Akrukay if the second vowel is low central /*a/:

When postthematic /*t/ is followed by final /*k/, …:

… /*s/ …:

…:

Medial velar voiceless stop /*k/ is generally retained as such in all descendants:

When medial /*k/ is … it is realized as a laminal affricate [ts tʃ s] in Breri and Romkun. It's not clear whether this is synchronically distinct from the reflexes of /*s/ (below):

When medial /*k/ is followed by final /*m/, it tends in Akrukay, Breri and Romkun to form a cluster with /*m/, which is then appended with a copy vowel:

…:

Medial bilabial prenasalized voiced stop /*mb/ is devoiced to [mp] in Itutang and Midsivindi and sometimes in Breri and Romkun, remaining distinct from the occluded [mb] reflexes of /*m/ (above):

When medial /*mb/ is followed by final /*m/, it tends in Akrukay, Breri and Romkun to form a cluster with /*m/, which is then appended with a vowel:

Medial apical prenasalized voiced stop /*nd/ … :

When medial /*nd/ is followed by final nasals /*m *ŋ/, it tends in Akrukay, Breri and Romkun to form a cluster with the final nasal, which is then appended with a vowel in Akrikay as well as in other descendants where one would be appended anyway (below):

One puzzling example is known to have medial /*nd/ in other Ramu subfamilies which is likewise followed by voiceless velar stop /*k/; it seems probable that Z'graggen's Akrukay and Breri forms are in error:

Medial laminal prenasalized voiced /*ndz/ …:

Medial velar prenasalized voiced stop /*ŋg/ is usually devoiced to [ŋk] in Itutang and Midsivindi and sometimes in Breri and Romkun:

…:

Medial bilabial non-stop /*w/ tends to be occluded to [β] in all descendants except Akrukay:

Medial apical non-stop /*r/ is generally retained as such in all descendants:

When medial /*r/ is followed by final bilabial nasal /*m/, it tends in Akrukay, Breri and Romkun to form a cluster with /*m *ŋ/, which is then appended with a vowel:

…:

One example of /*rVm/ though found also in Igom-Tanggu is attested only in Akrukay [dʒandʒaŋ ʌːnMo] “sea” and is probably a loan from Sogeram River /*arum/ “big, old” (q.v. Daniels 2010: 186):

One example of medial /*r/ followed by final velar nasal /*ŋ/ forms clusters in Breri and Romkun but not in Akrukay:

When medial /*r/ occurs postthematically, it tends to form a cluster with the consonant in the onset of the preceding syllable, sometimes resulting in the loss of [r] in Breri and Romkun:

Medial palatal non-stop /*j/ is occluded and aspirated to [s] in Breri, where it merges with the reflexes of /*s/, and occluded to [z ʒ] in Romkun, Itutang and Midsivindi:

…:

…:

Medial velar non-stop /*ɣ/ is usually occluded to [g] in all descendants except Midsivindi:

… /*ɣ/ …:

…:

…:

…:

In one example, medial /*ɣ/ is lost in Breri and Romkun when in a cluster with the reflex of final /*s/ (below):

…:

Only a restricted set of consonants occurs finally. This set includes nasals /*m *n *ŋ/, voiceless stops /*p *s *k/ and non-stops /*r *j *ɣ/.

Final bilabial nasal /*m/ is retained as such in all descendants, with a vowel appended in all except Akrukay:

Akrukay as well appends a vowel to final /*m/ when /*m/ has formed a cluster with a consonant in the onset of the preceding syllable (above.) In these instances /*m/ becomes an sound which in Z'graggen's (1972) vocabularies is given as capital <M> as opposed to regular <m>. No key is given to explain the meaning of this sound, but Brooks (2018: 61, 106) describes Chini [ƥm ƭm] <pm tn> as a prestopped nasal preceded by ingressive voiceless stops, while giving other clusters such as that in “water” as merely (e.g.) <nm>. The distinction does not appear to be phonemic. Following Z'graggen's we show this sound in Akrukay, Breri and Romkun as <M> since the international phonestic alphabet lacks any way to symbolize ingressive nasals:

…:

In a restricted class of monosyllables of the form /*Vn/, Itutang realizes final apical nasal /*n/ as [n] or as [nd] like the medial reflex of /*n/ (above.) Akrukay's second person singular looks to be irregular:

On roots with more than one syllable, final /*n/ is when preceded by a nasal retained as such in all descendants, with a vowel appended in all except Akrukay:

When final /*n/ is preceded by a stop in the onset of the following syllable, it is sometimes realized as apical non-stop [r] and tends form a cluster with the consonant in the onset of the preceding syllable as does final /*r/ in the same environment (below.) This change is invariable only in Romkun:

In one example where final /*n/ follows medial /*r/ …:

…:

Final velar nasal /*ŋ/ …:

Final bilabial voiceless stop /*p/ is lenited and deoccluded to bilabial non-stop /w/ [w β] in all descendants:

One apparent exception to this family-wide lenition is irregular in its initial as well, indicating a Ramu river regional loan (above):

Final apical voiceless stop /*t/, if it occured, is not distinguishible from final /*r/ (below.)

Final laminal voiceless /*s/ is lenited and deoccluded to palatal non-stop [j] or high front vowel [i] in all descendants, including Breri and Romkun where [s] and [z ʒ] are the regular reflexes of medial /*j/ (above.) This development is shared with Tanguat of the Moam River subgroup to the northeast. In Akrukay, Breri and Romkun it is appended with a vowel. These reflexes are distinct from those of final /*j/ (below):

… /*s/ …:

Final velar voiceless stop /*k/ …:

…:

Prenasalized voiced stops /*mb *nd *ndz *ŋg/ do not occur finally.

Final apical non-stop /*r/ reflects both Guam and Moam Rivers finals /*t *r/, which are not distinguishable in Guam River. Only found thusfar in disyllabic roots, final /*r/ forms clusters with the consonant in the preceding syllable. in Breri and Romkun and sometimes in Akrukay, often leading to the loss of [r]. Where this occurs in Akrukay, a final vowel is appended as it always is in other descendants:

Final palatal non-stop /*j/ …:

Final velar non-stop /*ɣ/ is occluded to [g] and appended with a vowel in Akrukay and in Breri and lost elsewhere else:

One of Z'graggen's examples suggests that final /*ɣ/ is at least sometimes retained in Midsivindi phrases or compounds:

While Goam River had only four vowels /*i *ʉ *a *u/ and no vowel sequences or diphthongs, these vowels often metathesize between syllables and interactions between them are not straightforwardly segmental. The general rule is that disyllables /*VCVC/ are always metathesized to /*VCCV/ in Breri and Romkun, are never metathesized in Itutang and Midsivindi, and are often but not always metathesized in Akrukay depending upon the qualities of the second vowel and the second consonant.

…:

…:

…:

…:

…:

…:

…:

In roots of the form /*Ca/ and /*aCa/, low central vowel /*a/ is retained as such in all descendants, with Z'grggen's [ʌ] representing an allophone of /a/:

Roots of the form /*aCaC/ are retained as such in Akrukay and in Inapang and are metathesized to /aCCa/ in Breri and Romkun::

…:

The first /*a/ in roots /*aCaC/ is fronted and raised to /e/ in Romkun if the following consonant is laminal voiceless /*s/:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

…:

In roots of the form *aCʉC/, /*a/ is generally fronted to /ɛ/ [ɛ ə], or in some instances dropped, in Breri and Romkun:

…:

…: