Mandobo

Timothy Usher, Santa Fe Institute

Mandobo, also known as Dumut, is comprised of at least three closely-related languages or dialects spoken along and to either side of the Mandobo river, called by natives Ndumut, between the Digul river to the west and the Kao to the east in the Indonesia's Papua province, to the south along the Digul, and in an enclave just east of the Fly river across the Papua New Guinea border called Kwem.

Drabbe's (1954, 1959) survey found three dialects of Dumut, the Wambon (Oàmbon) dialect on the upper Mandobo, Rungwanjap (Runggoanjàp) on the middle Mandobo, and Kambom on the lower Mandobo near its confluence with the Digul (1959: 4.) Lebold, Kriens and Susanto (2013) survey a region far to the south, along the east bank of the great bend in the Digul adjacent to Southeast Awyu and Upper Mbian Marind. Surprisingly, Drabbe's (1954) fragmentary attestations of Runggoanjàp apear to be somewhat more phonologically similar to Lebold, Kriens and Susanto's South Mandobo than to Upper Mandobo or to Drabbe's (1959) Kaeti, which is likewise said to be a variety of Rungwanjap spoken along the Marek river, a tributary of the Mandobo.

Jang (2003) distinguishes four divisions of Mandobo, Kokenop Mandobo, Upper Mandobo, Central Mandobo and Mariam Mandobo, based primarily upon materials collected in two unpublished surveys by Farneubun (2002.) Central Mandobo is equivalent to Runggwanyap, while Mariam Mandobo is spoken in Mariam village and in the multilingual village of Kouh in the northwest of the Mandobo region on the east bank of the Digul river. These are in turn broken down into two subgroups, Kokenop and Upper Mandobo and Central and Mariam Mandobo. Jang's Kokenop is closely related to or the same language as Drabbe's (1959) Wambon and de Vries' (1986) Yonggom Wambon, and is not a member of the Mandobo subfamily. Jang's use of the term Upper Mandobo differs from that adopted here, as it includes the village of Winiktit which we presume to be the same Drabbe's (1959: 115) Waniktit; this would explain Jang's finding of an exclusive relationship between the two speech varieties. Since none of Jang's villages are quite on the Mandobo (q.v. 2003: 14,) nor do these languages belong to the Mandobo subfamily, we reserve the term Upper Mandobo to refer to Drabbe's (1954) Wambon (Oámbon) dialect of Dumut.

The internal classification of Mandobo is as follows:

Mandobo

Upper Mandobo

Central Mandobo

Mariam Mandobo

As no Mariam Mandobo is currently available to us, both its status as a distinct language and its placement above relies upon Jang (2003: 23-27.) Nor can we say with certainty whether Drabbe's Kambom is equivalent to Mariam Mandobo, as its location is described as somewhat to the south of Jang's Mariam region (Drabbe 1959: 4, q.v. Jang 2003: 38.)

Central Mandobo may be further broken down by subdialects which differ primarily in their treatment of certain vowels (below,) but only three of these, the Omba dialect as described in Drabbe (1959,) Drabbe's (1954) Runngoanjàp of unspecified location, and that of the southern villages documented in Lebold, Kriens and Susanto (2013) are currently available to us, and Jang's list of subdialects does not purport to be exhaustive. Only five Kwem words are given in Voorhoeve (2001: 372-381;) it seems probable that these represent a variety of Central Mandobo in addition to those listed by Jang.

van Doorn (1931-1932) 429 comparative terms for Digul Mandobo (Upper Mandobo) published in Stokhof ed. (1982: 141-150)

Drabbe (1954: 162-163) pronouns and (1954: 211- 214) grammar sketch of Oàmbon (Upper Mandobo) of Anggimòp village and Runggoanjàp and (pp. 214-229) 419 comparative terms and (pp. 231-255) 100 comparative terms for Oàmbon Dumùt

Drabbe (1959: 3-108) grammar and (pp. 162-181) 431 comparative terms for Kaeti (Omba Rungwanjap)

Voorhoeve (2001: 372-381) 5 comparative terms for Kwem

Farneubun (2002) Central Digul survey (unobtained)

Farneubun (2002) Mandobo survey (unobtained)

Lebold, Kriens and Susanto (2013: 36-47) 239 comparative terms for Mandobo of Kaisah village and (pp. 48-50) 60 comparative terms for Mandobo of Subur and Aiwat (South Mandobo)

Wester (2014) Awyu-Dumut comparative morphology including (pp. 195-210) 430 comparative terms for (Kaeti) Mandobo after Drabbe (1959: 162-181)

[under construction]

Proto-Mandobo had 13 consonants and 8 vowels as follows:

Any consonant can occur initially or medially, except for bilabial voiceless stop /*p/ which has been lenited to non-stop /*w/ or lost in these positions and occurs only root-finally. An unoccluded aspirate /*h/ is found only in loans, where it can reflect either bilabial voiceless stop /*p/ or Marind /*w/ as seen in van Doorn's Digul Mandobo [jahu] “canoe” (cf. Awyu /*jaxʷuN/ where the special phoneme /*xʷ/ indicates a loan from Marind-Yaqay /*iawun/.) Initial prenasalized stops /*mb *nd *ŋg/ are often realized as plain voiced [b d g]. Medial voiceless stops /*t *k/ are sometimes realized as plain voiced [d g].

Only a restricted set of consonants occurs finally:

Consonant clusters occur only in compounds.

A series of changes which might be aptly characterized as the Digul region vowel shift has affected all Mandobo languages to varying extents, with Upper Mandobo being the most conservative and Drabbe's Kaeti, representing the Omba subdialect of Runggwanjap, the most evolved in this respect. This shift has high back vowel /*u/ fronted to high central /ʉ/ and high mid back /*o̝/ raised to /u/, as well as in Mandobo the merger of mid front vowels /*e̝ *ɛ/, resulting in Omba's seven vowels as follows (Drabbe 1959: 7):

Since phonetic diphthongs are found primarly word-finally, and some of these are known to result from the palatalization of final apical non-stop /*ɾ/, these are treated as the results of interaction between vowels and non-stops /*w *j/.

Initial and medial consonants correspond as follows:

Final consonants correspond as follows:

Clusters formed by compounding correspond as follows, with only attested combinations listed:

Vowels correspond as follows:

These correspondences are examplified as follows, with Oàmbon and Runggoanjàp attestations drawn from Drabbe (1954: 162-163, 211- 214,) Digul Mandobo from van Doorn (1931-1932, printed in Stokhof ed. 1982: 141-150,) Kaeti (Omba) from Drabbe (1959: 162-181) and South Mandobo of Kaisah village from Lebold, Kriens and Susanto (2013: 36-47.) For clarity, our presentation is divided into four sections, beginning with initial and medial consonants, followed by final consonants, then consonant clusters formed by compounding, then vowels.

Initial and medial bilabial nasal /*m/ is retained as such in all descendants:

Initial and medial apical nasal /*n/ is retained as such in all descendants. Root-initial /*n/ is markedly uncommon:

Bilabial voiceless stop /*p/ does not occur intially or medially, because it has been lenited to /*w/ or lost in proto-Mandobo.

Initial and medial apical voiceless stop /*t/ is retained as such in all descendants. Medially, it is sometimes heard as plain voiced [d]:

Initial and medial velar voiceless stop /*k/ is retained as such in all descendants. Medially, it is sometimes heard as plain voiced [g]. In South Mandobo, /*k/ is sometimes backed to uvular [q ɢ]:

Rounded velar voiceless stop /*kʷ/ has been identified in only a few words where it is followed by low central vowel /*a/. It's possible that there are other examples of /*kʷ/ followed by rounded back vowels which have gone unrecognized:

Bilabial prenasalized stop /*mb/ is retained as such in all descendants. Initial /*mb/ is often heard as plain voiced [b]:

Apical prenasalized stop /*nd/ is retained as such in all descendants. Initial /*nd/ is often heard as plain voiced [d] and is relatively uncommon. Both initial and medial /*nd/ are often retroflexed [ɖ nɖ] in South Mandobo:

In one very common verb formative medial /*nd/ is unexpectedly reflected as /n/ in Drabbe's Kaeti:

Velar prenasalized stop /*ŋg/ is retained as such in all descendants. Initial /*ŋg/ is sometimes heard as plain voiced [g] or occasionally as [k] in Upper Mandobo, while both initial and medial /*ŋg/ is sometimes realized as prenasalized voiceless [ŋk]. Both initial and medial /*ŋg/ are often backed to uvular [ɢ ɴɢ] in South Mandobo:

Rounded velar prenasalized stop /*ŋgʷ/ has been identified in only a few words where it is followed by low central vowel /*a/ or, in one instance, high mid back /*o̝/ where it is discernible only by its effect upon preceding /*a/ (below.) It's possible that there are other examples of /*ŋgʷ/ followed by rounded back vowels which have gone unrecognized. Like unrounded /*ŋg/ it can be backed to [ɴɢ] in South Mandobo:

Bilabial non-stop /*w/ is found almost only before vowels /*a *ɔ/. One example in which /*w/ is followed by high front /*i/ is unexpected and unexplained as both Dumut /*p *w/ are usually lost in this environment:

…:

In one example, what would seem to be the reflex of Dumut /*w/ is given as voiced velar stop [g] in Drabbe's Kaeti:

…:

Apical non-stop /*ɾ/ is retained as such in all descendants. Drabbe (1954: 62) states that Oámbon Dumut <r> may be heard as lateral [l]. It is occasionally heard as a voiced stop [d] in van Doorn's Digul Mandobo:

In one example, medial /*ɾ/ is unexpectedly dropped in van Doorn's Digul Mandobo and in Drabbe's Kaeti. Here the originality of the sequence /*iɾɔ/ is based upon comparison to East Wambon /iɾɔmbut/:

Initial palatal non-stop /*j/ has been found only preceding vowels /*ɵ̝ *a *ɔ/. It is sometimes occluded to [dʲ dʒ] in van Doorn's Digul Mandobo and in South Mandobo:

The reflexes of medial /*j/ are like those of /*w/ (above) quite complicated and often result in phonetic vowel sequences or diphthongs. Some examples of /*j/ reflect Dumut medial /*p/, which is reinterpreted as either Mandobo /*w/ or /*j/ depending upon the qualities of adjacent vowels.

When preceded and followed by non-front vowels /*a *ɔ *o̝ /, /*j/ is retained as such in all descendants:

…:

…:

At least two instances of a phoneme /*h/ are found retained only in van Doorn's Digul Mandobo. In the first case, this reflects a loan word /*sapo̝k/, [h] being the fricated and deoccluded reflex of /*p/ which is merged with /*w/ or dropped altogether in native words; the second is a loan from Marind-Yaqay /*iawun/ with the aspiration of /*w/ to [hʷ h] characteristic of Marind's western dialects; cf. Awyu /*jaxʷuN/ with the special phoneme /*xʷ/ found only in Marind loans.

Only a restricted set of consonants /*m *n *p *t *k *j/ occurs finally.

Final bilabial nasal /*m/ …:

Verbs with final /*m/ are reinterpeted as including or reinforced with auxilliary formative /*-mɔ-/ “do” in Drabbe's Oambòn Dumut:

Final apical nasal /*n/ …:

Final bilabial voiceless stop /*p/ …:

Final apical voiceless stop /*t/ …:

Final velar voiceless stop /*k/ is significantly less common than final anteriors /*p *t/, an assymetry which dates to Digul River's merger of Digul River-Ok final prenasalized /*mb *nd/ with /*p *t/ but velar /*ŋg/ with apical nasal /*n/ rather than with /*k/. It is often heard as glottal stop [ʔ] in van Doorn's Digul Mandobo:

Several examples of final /*k/ are heard as /t/ in Upper Mandobo. This has been found to occur only following apicals /*t *ɾ/ in the previous syllable, but is not regular:

In one example, final /*k/ is unexpectedly reflected as /p/ in Upper Mandobo; here the original value of /*k/ is confirmed through comparison to East Wambon and Digul Wambon:

In one example, final /*k/ is unexpectedly dropped in both Drabbe's Kaeti and in South Mandobo:

Final palatal non-stop /*j/ …

…/*o̝j/ … [o̝ⁱ] …:

… /*ɔj/ … [*ɔⁱ] …:

…:

One irregular example has reflexes of /*ɔj/ in Upper Mandobo and of /*aj/ elsewhere, probably indicating a loan from East Wambon into Upper Mandobo as Mandobo /*ndaj/ from Dumut /*ndɒj/ would be regular:

… /*aj/ … [aᵋ] …:

While final /*j/, reflecting Dumut /*ɾ *j/, has generally been lost in Mandobo when preceded by high mid central vowel /*ɵ̝/, in at least one example a trace of it is preserved due to suffixation; cf. free form /*mbɔɾɵ̝/ “wing” (Dumut /*mbɔɾɵ̝ɾ):

Consonant clusters do not occur within roots, but, because most roots have final consonants, they are very frequent in compounds. Because only a limited number of roots are frequently found as the second member of compounds, it's not possible at this time to provide correspondences for every possible combination. For example, as intial /*n/ is found only in a few roots, none of which have been found as suffixes following a consonant, the dispositions of clusters with the second member /*n/ are unkown, although they can be fairly assumed to exist. Since bilabial voiceless stop /*p/ is not known to occur root initially (above,) no clusters of the form /*C-p/ have been identified. Historically, the root /*imbin-/ “thought” reflects Dumut /*pin-pin/, but this may have been established by the time of Proto-Mandobo prior to the loss of /*p/.

When root-final /*m/ is followed by a suffix with initial bilabial /*m/, the resulting cluster is simplified to /m/ in all descendants, in contrast to the reflexes of cluster /*nm/ (below):

When root-final /*m/ is followed by a suffix with initial velar voiceless stop /*k/, the resulting cluster is realized as velar prenasalized stop /ŋg/ in Upper Mandobo:

When root-final /*m/ is followed by a suffix with initial apical non-stop /*ɾ/ …:

When root-final /*n/ is followed by a suffix with initial bilabial /*m/, the resulting cluster is realized as /n/ in Upper Mandobo and as /m/ elsewhere:

When root-final /*n/ is followed by a suffix with initial apical voiceless stop /*t/, /*n/ is dropped. As some examples of this are discernible only through outcomparison to other Dumut languages, it's possible that some examples of this cluster have been wrongly reconstructed with only /*t/:

Two apparent exceptions are given as separate words in Drabbe's Oàmbon Dumut and Kaeti:

When root-final /*n/ is followed by a suffix with initial velar voiceless stop /*k/ …:

When root-final /*n/ is followed by a suffix with initial palatal non-stop /*j/ …:

When root-final bilabial voiceless stop /*p/ is followed by a suffix with an initial vowel, it is realized as non-stop [w] in Kaeti, in accordance with the regular reflex of Dumut initial and medial /*p/ as /*w/. In van Doorn's Digul Mandobo, final /*p/ is given with both [p] and glottal stop [ʔ]:

When root-final /*p/ is followed by a suffix with initial velar voiceless stop /*k/ …:

When root-final /*p/ is followed by a suffix with initial apical prenasalized stop /*nd/ …:

When root-final /*p/ is followed by a suffix with initial apical non-stop /*ɾ/ …:

When root-final /*p/ is followed by a suffix with initial palatal non-stop /*j/ …:

When root-final apical voiceless stop /*t/ is followed by a suffix with an initial vowel, it is generally realized as non-stop [ɾ], though several of van Doorn's examples with [d] might suggest /*t/ to still be underlyingly present:

When root-final /*t/ is followed by a suffix with initial bilabial nasal /*m/ …:

When root-final /*t/ is followed by a suffix with initial velar voiceless stop /*k/ …:

…:

When root-final /*t/ is followed by a suffix with initial bilabial prenasalized stop /*mb/ … (Drabbe 1959: 6):

When root-final /*t/ is followed by a suffix with initial velar prenasalized stop /*ŋg/ … (Drabbe 1959: 6):

When root-final /*k/ is followed by a suffix with initial bilabial nasal /*m/ …:

When root-final /*k/ is followed by a suffix with initial /*k/, the resulting cluster is simplified to /k/:

When root-final /*k/ is followed by a suffix with initial apical non-stop /*ɾ/ …:

When final /*k/ is followed by a suffix with initial palatal non-stop /*j/, /*j/ is dropped in Upper Mandobo:

While consonants are generally stable, the realization of vowels /*i *u *e̝ *ɵ̝ *o̝ *ɛ *ɔ *a/ vary substantially between dialects.

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

…:

High back vowel /*u/ is fronted to high central /ʉ/ [y] <ü> in Drabbe's Kaeti, intermittently to [y ʏ] in van Doorn's Digul Mandobo and occasionally to [ʲu] in South Mandobo in accordance with a widespread Digul River regional trend. Reflexes in van Doorn's Digul Mandobo range from [u ʲy y ʏ]. Upper Mandobo and South Mandobo are along with South Awyu to the southwest among the few Central Digul River languages that have not undergone invariable fronting of /*u/, due to their location on the margins of the region:

…:

High mid front vowel /*e̝/ is systematically distinguished from low mid /*ɛ/ (below) in Upper Mandobo and merged with /*ɛ/ in Central Mandobo, as it is elsewhere in Dumut:

In several examples in which the second vowel is known to correspond to Awyu high front /*e̝/, each following low central vowel /*a/ in a previous syllable, Drabbe's Oàmbon Dumut attestation unexpectedly has low mid [ɛ æ]. The variation between high and low values in “saliva” and in van Doorn's Digul Mandobo suggests this discrepancy to be subphonemic:

In two very common verb formatives high mid /*e̝/ is unexpectedly given with /i/ in Drabbe's Oèmbon Dumut:

One puzzling irregular example may indicate a loan into Central Mandobo from Central Awyu /*in/, where the merger of high mid /*e̝/ with high /*i/ is regular:

One example appears to be inexplicably irregular, as cognates are attested with /e̝/ or /ɛ/ elsewhere:

High mid central vowel /*ɵ̝/ is merged with the reflexes of high mid back /*o̝/ (below) in Upper Mandobo, where it is similarly usually realized as high back [u] and sometimes as [o]. Lebold, Kriens and Susanto give a wide range of South Mandobo reflexes ranging from back [o ɔ u] to central and front [ə ɛ], probably indicating a failure to recognize the independence of this phoneme. Drabbe usually gives Kaeti /ɵ/ <ö> but a minority of examples are given with mid front [ɛ]. Often but not always these examples coincide with South Mandobo [ɛ], perhaps suggesting that a split is underway. This fronting is most typically associated with the presence of low central /*a/ in a previous syllable, but the pattern is not quite consistent:

One example …:

High mid back vowel /*o̝/ is generally realized as high back [u] in all descendants, though a few examples are given with [o ɔ] in Upper Mandobo and South Mandobo, and Drabbe (1954: 211) specifies that Oàmbon Dumut /u/ is sometimes pronounced as <oo> in Dutch <vloot>:

One example in which /*o̝/ follows high front /*i/ in the previous syllable is given as /i/ in South Mandobo. As no comparable sequences have been identified, it's not clear if this is a regular change:

Low mid front vowel /*ɛ/ is relatively uncommon, as it is in Awyu:

…:

Low mid back vowel /*ɔ/ is generally retained as such in all descendants. Drabbe (1954: 211) specifies Oàmbon Dumut /ɔ/ to be open as in Dutch <vlot>, but as Drabbe's <o> is also given for some reflexes of high mid vowels /*ɵ̝ *o̝/ (above) his original transcriptions have been retained here:

When followed by high mid central and back vowels /*ɵ *o̝/ in the next syllable, low mid /*ɔ/ is raised to /u/ in Upper Mandobo and sporadically in Kaisah, a change shared with East Wambon and Digul Wambon to the north:

One example in which /*ɔ/ followed by /*ɵ/ is not raised may indicate a loan or a compound:

In monosyllables appearing as the second elements of compounds, low mid /*ɔ/ …:

Low central vowel /*a/ is generally retained as such in all descendants, except where followed by a high or high mid back vowel /*i *u *o̝/ (below):

When /*a/ is followed by a rounded velar /*kʷ *ŋgʷ/ it is backed and raised to /ɔ/ in all descendants. Here the original vowels are known to have been /*a/, while a reconstruction /*ɔŋgo̝/ “thumb” would be irregular as /*ɔ/ followed by /*o̝/ should be raised to /u/ in Upper Mandobo (above):

When /*a/ is followed by high front vowel /*i/ in the next syllable, it is fronted and raised to /ɛ/ in Central Mandobo (Healey 1970: 1004, 1007-1008):

There is at least one unexplained exception in which /*a/ followed by /*i/ in the next syllable is not raised in Drabbe's Kaeti:

When /*a/ is followed by a high back or high mid back vowel /*u *o̝/ in the next syllable, it is backed and raised to /ɔ/ in Central Mandobo:

In two examples, Central Mandobo /*a/ followed by /*u *o̝/ does not undergo the expected raising and backing, perhaps indicating loans:

…

… /*uja/ …

… /*wuj/ …: