Marind

Timothy Usher, Santa Fe Institute

The Marind family consists of as many as five closely related languages spoken by 6,000-7,000 people (1955) living in the plains along and inland from the coast from the Marianne Strait in the west to the Maro river to the east, just west of the Papua New Guinea border, in the Merauke, Muting, Kurik and Okaba subdistricts of the Merauke regency of Indonesia's Papua province (Boelaars 1950: 1, Drabbe 1955: 13; q.v. Drabbe 1955: 190, Nevermann 1939: 2, Kriens 2003: 5, Sohn 2006: 4, Sohn, Lebold and Kriens 2009: 7, Lebold, Kriens and de Vries 2010: 7.)

Drabbe (1955: 13-15) divides the coastal Marind area from east to west into five regions, Ngawir between the Papuan border and the Maro (Maroh,) Rahuk between the Maro and the Kumbe (Kumb) river, an unnamed region between the Kumbe and the Bian (Mbian) river, Dali between the Bian and the Bulaka (Mbulàka) river, and Imah between the Bulaka and Marianne Strait (Mulì.) Marind languages are also spoken inland along the Kumbe, Eli and Bian rivers.

The internal distribution of Marind languages suggests the family to have originated in the northeastern portion of their current range, an assumption in accord with the locations of Marind's nearest relatives, Yaqay-Warkay and Lake Murray (Boazi-Zimakani.) The reconstruction of Awyu in particular necessitates a history of contact with proto-Marind or proto-Marind-Yaqay, while the southward expansions of Awyu, Dumut and Lowland Ok occured in relatively recent times. It's also probable that Marind's vowel harmony rules, along with Marind-Yaqay's deocclusion of Fly River's bilabial fricative, codeveloped with those of Lowland Ok. Thus it is reasonable to assume that proto-Marind-Yaqay was originally spoken near where the southernmost Mandobo dialects are spoken today, just west of the Fly River along which the other members of the Fly River family were located.

More recently, the westernmost dialects have had extensive contact with and have significantly influenced the unrelated Bulaka River languages and the only distantly related languages of the Kolopom and South Marianne Strait families (q.v. Geurtjens 1933: 398-433,) while the southeast dialect has very significantly influenced Moraori, a Kolopom language that was very likely separated from its nearest relatives to the west by the southward intrusion of the Marind.

[under construction]

Drabbe (1955: 13-15) distinguishes five dialects of Marind with locations as follows:

Marind

Mbian – Bian river

Kumb – Kumbe river

Ooster (East) – coast east of the Maro river and between the Maro and the Kumbe

Wester (West) – coast between the Kumbe river and Marianne Strait

Atìh – Sangahè (Sanggase) village on the coast to Jaimu (Yawimu) inland of the West dialect region

Drabbe's fivefold division is supported by the results of recent surveys by members of the Summer Institute of Linguistics (Lebold, Kriens and de Vries 2010: 22-24, 27-29, 29-30.) It is our opinion that Lebold, Kriens and de Vries' (pp. 22-24) speaker judgments are more reliable than their lexical similarity calculations (pp. 27-29) and that the speech varieties of Makeling, Iwol and Alaku villages should be considered to belong to the West Marind dialect rather than with Atih.

The internal classification of Marind is as follows:

Marind

Bian River Marind

Central Marind

Kumbe River Marind

Southeast Marind

Atih

West Marind

Of these, (Upper) Bian River is the most lexically divergent, with the other four dialects forming a Central Marind subgroup. The Atih and West dialects are very similar phonologically …. West Marind is distinguished from Southeast Marind primarily by the dispositions of non-stops, which codevelop with those of Maklew of the Bulaka River family to the west. Kumbe River is distinguished from Southeast Marind primarily by the presence in Kumbe River of a number of vocabulary terms otherwise found only in Bian River, a chaining pattern which suggests some remaining degree of mutual comprehensibility between all varieties of Marind.

…isoglosses…:

[under construction]

Ray (1907: 391-412) … comparative terms for Tugeri after Montague (n.d.)

van der Kolk and Vertenten (1922) dictionary of Marindineesch

van der Kolk (1926) Marindineesch kin terms

Geurtjens (1926) Marindineesch grammar (unobtained)

Geurtjens (1933: 46-350) dictionary of Marindineesch (Gawir, i.e. Southeast Marind,) (pp. 373-383) 591 comparative terms for Sangaseesch (Atih,) (pp. 385-395) 693 comparative terms for Boven-Biansch (Mbian) and (pp. 398-433) 455 comparative terms for Marindineesch, occasional Imâh (West Marind) terms throughout

Nevermann (1939: 15, …) … for Marind-anim (Southeast Marind)

Boelaars (1950: 1-18) English-language recension of Drabbe's unpublished grammar of Gawir (Southeast Marind)

Drabbe (1954: 99-117) grammar of Boven-Mbian of Kolam village and (pp. 128-142) 427 comparative terms for Boven-Mbian and Gawir (Southeast Marind)

Drabbe (1955: ibid.) grammar of East Marind of Urumb village and (pp. 148-151) 98 comparative terms for Ooster (East,) Wester (West,) Atih, Kumb and Mbian dialects

Donohue (1996) 159 comparative terms for Kuler (Southeast Marind) and Sanayu (Kumbe) villages

Sohn, Lebold and Kriens (2009: 31-42) 204 comparative terms for Matara village (Southeast Marind)

Lebold, Kriens and de Vries (2010: 46-56) 240 comparative terms for Alaku, (West Marind,) Ihalik (Atih,) Kaptel (A) and Sanggase (A) villages and 69 comparative terms for Alatepi (A,) Boepe/Kwemsid (A,) Dufmira (W,) Iwol (W,) Kaniskobat (A,) Makaling (W,) Nakias (A,) Okaba (W,) Poepe (A,) Tagaepe (A) and Wambi (W) villages

Olsson (2015) Marind verbal objects

Olsson (2015) 215 comparative terms for Duvmirav village (West Marind) in Carroll ed.

Olsson (2015) 912 terms for Wambi village (West Marind)

Olsson (2016) Marind resultatives

Olsson (2016) Marind gender

[under construction]

Proto-Marind had 17 consonants and 5 vowels as follows:

The reflexes of nasals, stops and fricatives remain stable throughout the family, but the those of non-stops /*w *ɾ *j *ɣ/ vary substantially between dialects (q.v. Drabbe 1955: 14-15,) with the western dialects generally agreeing with Maklew and the northeastern ones with Boazi (q.v. Drabbe 1950: 566-574, 1954: 128-142, 1955: 148-151.)

The contrast between /*h *ɣ/ is unusual in New Guinea; here /*h/ reflects Fly River bilabial fricative /*ɸ/.

Internal distributions and comparison to other Fly River families suggest the contrast between plain voiced and prenasalized stops to be of relatively recent origin, with Fly River prenasalized stops /*mb *nd *ŋg/ yielding Marind /*b *d *ɣ/ initially and /*mb *nd *ŋg/ medially, the phonemic difference between the two having been established by loanwords from the Morehead River family to the east and compounding.

Drabbe (1954) distinguishes closed [e] <é> from open [ɛ] <e>, but this distinction is drawn by default throughout much of his New Guinea corpus, and in (1955: 15) is explicitly denied to be contrastive.

Any consonant can occur initially, medially or finally. There are no consonant clusters except across morpheme boundaries.

Any vowel can occur initially, medially or finally. Unlike consonants, vowels can occur sequentially. Vowel length is indicated in Drabbe (1954,) but in (1955: 15) is explicitly denied to be contrastive.

Stress falls on the final syllable of the root, as it does in the distantly-related Lowland Ok family to the northeast; Drabbe (1955: 16) characterizes this as a rise in pitch or tone. Trisyllablic roots receive secondary stress on the initial syllable, leading to syncope and the formation of medial consonant clusters in Mbian and Kumb. Probably most families with regular root-final stress reflect penultimate stress in words previously of the form /*(C)V'CVCV/, and in this instance we know this to be true from the absorption of pre-Fly River postposed gender articles /*e *u *i/ into the root (below) followed by the dropping of the articles; however, this may have already been the case by the time of proto-Fly River.

Diphthongs are phonetically distinguished from other vowel sequences by stress, as final sequences /*V₁V₂(C)/ are realized as [V₁'V₂(C)] but diphthongs as ['V₁ᵛ₂(C)]. Conversely, sequences /*ia *ua/, which are found only initially, are realized as semivowel-initial syllables [ja wa], as is the sequence /*ui/ [wi] if not preceded by a consonant. The distinction between these three types is secondary, as, despite the different assignment of stress, their qualities are in no instance contrastive. Which sequences are realized as diphthongs, which with initial semivowels and which as distinct syllables is predictable according to a ranking rule by which stress is preferentially assigned to the either the first or second member of the sequence in the following order: /*a *o *e *i *u/. The same order governs the assignation of stress in the diphthongs given for East Marind in Drabbe (1955: 15-16) with only one exception (below.) The phonetic realizations of observed vowel sequences are as follows:

As suggested by the realization of /*ui/ as [*ᵘi] or [*ui] depending upon whether it is root-initial or follows a consonant, it's probable that the distinction in stress between [*ᵘa *ⁱa] on the one hand and [*oa *ea *eo] on the other can be reduced to the conditioned variation of a single subtype. Drabbe's (1955: 150-151) comparative vocabularies give <eo> in initial /*eoɣ/ 2 pl., but this root as given as <ᵉoh> in the grammar of the East dialect (pp. 15-16,) and otherwise the difference between these two groups can be explained by the same rule which governs the disposition of /*ui/, as sequences /*ua *ia/ are attested only intially, while /*ea *oa/ have been observed only following consonants.

There are two remaining inconsistencies. The first is that Drabbe (p. 16) gives /*mui/ “flesh” as East [muⁱ] rather than [mu'i]. However, /*ŋgui/ “tail” is given as [ŋgu'i] in Mbian (1954: 136,) and diphthong [uⁱ] isn't otherwise indicated in the comparative vocabularies. The second is that what would otherwise be reconstructed as /*koi/ “lime” is given as [ko'i] in Sohn, Lebold and Kriens' Matara (2009: 34.) As Sohn et al. give Moraori [kwuoi] alongside it, it's probable that the Marind term was really /*kuoi/ or the like, as sequences of three vowels are not construed as single syllables.

Sequences which are realized as diphthongs are broadly similar in range to those of proto-Awyu, which does not otherwise allow sequential vowels. It's possible that contact with Awyu influenced the development of these segments in Marind.

Consonants correspond as follows:

Vowels and vowel sequences correspond as follows:

These correspondences are exemplified as follows, with all attestations drawn from Drabbe (1954: 99-117, 128-142, 1955: ibid., 148-151) unless otherwise specified. Examples based solely upon (1954) will include attestations only in East and Mbian dialects; thus their lack of exemplification in other dialects is not necessarily meaningful.

Drabbe (1954) distinguishes open and close mid front vowels [ɛ e] as well as vowel length. Because these are not contrastive (Drabbe 1955: 15,) and aren't particularly indicative of anything else, they have been eliminated in our recapitulations below. Stress, while not consistently indicated, is shown here because it is sometimes contrastive due to inflection and compounding (Drabbe 1955: 16.) Drabbe's <r> is represented here as <ɾ>.

Boven-Mbiansch (Bian,) Gawir (Southeast Marind) and Sangaseesch (Atih) attestations drawn from Geurtjens (1933) are indicated with [hg]. Matara village (Southeast Marind) forms from Sohn, Lebold and Kriens (2009: 31-42) are indicated with [ms] and Atih and West Marind attestations from Lebold, Kriens and de Vries (2010: 46-56) as [rl]. These are simplified and regularized to conform with the orthographical conventions described above. West Marind forms drawn from Olsson (2015) are indicated with [bo].

…

For the purposes of our reconstruction below, our sources are correlated to dialects as follows:

Nasals /*m *n/ are invariable across dialects and occur in all positions.

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

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

It may be observed that a disproportionate number of roots with final /*n/ above are verbs, suggesting final /*-n/ to have been a suffix. In support of this hypothesis are several roots upon which final /n/ is found in either Mbian or the Central languages but is absent in the other subgroup, patterns which must otherwise be viewed as irregular:

Voiceless stops /*p *t *k/ are invariable across dialects and occur in all positions. Velar /*k/ is much more common than anteriors /*p *t/ (q.v. Geurtjens 1933: 169-227, 273-281, 312-326.)

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

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

Velar voiceless stop /*k/ is retained as such in all descendants:

Plain voiced stops /*b *d *g/ are invariable across dialects and occur in all positions: Initial velar /*g/ is much less common than anteriors /*b *d/ (q.v. Geurtjens 80-97, 97-108, 121-128.)

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

Two of five examples of medial /*b/ above, /*tuban/ “bandicoot sp.” and /*kabeɾ/ “fat/grease”, are known to be loans from Yei of the Morehead River to the east.

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

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

Prenasalized stops /*mb *nd *ŋg/ are generally invariable across dialects and occur in all positions. Initial prenasalized stops are relatively uncommon (q.v. Geurtjens 1933: 238-241, 253-254, 255-258) and are sometimes heard as plain voiced stops [b d g]. Final prenasalized stops are also uncommon. Prenasalized stops are much more common medially then they are initially or finally.

Bilabial prenasalized stop /*mb/ is retained as such in all descendants. One initial example is heard as plain voiced stop [b] in Mbian:

(For initial /*mb/ in “cuscus” cf. Yaqay ['mboeam] id. (Drabbe 1954: 136.))

Apical prenasalized stop /*nd/ is retained as such in all descendants. One initial example is heard as plain voiced stop [d] in Mbian:

Velar prenasalized stop /*ŋg/ is generally retained as such in all descendants. One initial example is heard as plain voiced stop [g] in the East dialect:

The word /*ŋgat/ “dog”, while found in other branches of Fly River (with irregular correpondences,) is a loan from Morehead River, as is “star” (though it's also found in Lake Murray.)

In a number of roots, a final prenasalized stop /*mb *nd *ŋg/ …:

Fricatives /*s *h/ are invariable across dialects and occur in all positions.

Fricative /*s/ is retained as such in all descendants:

In one example, the Mbian reflex of /*s/ is unexpectedly /h/, with characteristic dropping of preceding unstressed /*a/ (below):

Unoccluded fricative /*h/ is retained as such in all descendants:

Unlike nasals and stops, the realization of which is mostly stable in all descendants (above,) the reflexes of non-stops /*w *ɾ *j *ɣ/ vary substantially between dialects. Bilabial and palatal non-stops /*w *j/ are aspirated to [hʷ hʲ] in the West and Atih dialects, while East, Kumb and Mbian affect the more common occlusions to [v z] (as discussed in Drabbe 1933: 14.) The developments in the western dialects are shared with Maklew of the unrelated Bulaka River family to the west; those in the east are shared with Lake Murray to the northeast. These non-stops contrast with high vowels /*i *u/ in the sequences /*ia *ui *ua/ and with high vowel offglides in diphthongs, which may be heard as semivowels but are neither aspirated nor occluded (below.) Apical non-stop /*ɾ/ is realized as /l/ in Mbian and in the western dialects. Velar non-stop /*ɣ/ is aspirated and deoccluded to /h/ in the East and Kumb dialects.

Bilabial non-stop /*w/ is occluded to /v/ in Mbian, Kumb and East dialects and aspirated to /hʷ/ in the western dialects. The labiodental quality of Drabbe's <v> is confirmed by a more recent vocabulary of Matara village (q.v. Sohn, Lebold and Kriens 2009: 31-42.) Mbian often realizes /v/ as voiceless fricative [f] finally, or medially as the first member of a cluster such as those resulting from syncope (below.) In the East dialect, /v/ approaches voicelessness (Drabbe 1955: 16):

Apical non-stop /*ɾ/ is reflected as /l/ the West and Atih dialects and in Mbian. In the East dialect it is given as <r> and in Kumb variably as <r l>. While Drabbe made no effort to distinguish between rhotics [r ɾ] etc., a more recent vocabulary from Matara village gives reflexes of /ɾ/ as retroflexed flap [ɽ] initially and medially and as a trill [r] only finally (q.v. Sohn, Lebold and Kriens 2009: 31-42.) It is more common finally than medially and uncommon initially (q.v. Geurtjens 1933: 281-293.):

Palatal non-stop /*j/ is occluded to /z/ in Mbian, Kumb and East dialects and aspirated to /hʲ/ in the western dialects. As with the occluded reflex of /*w/ Mbian can realizes /v/ as voiceless fricative [s] medially as the first member of a cluster such as those resulting from syncope (below.) In the East dialect, /z/ approaches voicelessness and can be heard as [s] word-finally (Drabbe 1955: 16):

Velar non-stop /*ɣ/ is aspirated and deoccluded in the central dialects of Marind, the distinction between /*ɣ *h/ being lost completely in the Eastern and Kumb dialects (q.v. Drabbe 1933: 14.) Atih reflects /*ɣ/ as glottal stop [ʔ] (zero initially and finally) or, less commonly, as [h]:

The conflation of /*ɣ/ with /*h/ [h] (above) in the East dialect has the unfortunate consequence that neither van der Kolk and Vertenten' (1922) or Geurtjens' (1933) dictionaries nor Drabbe's (1955) grammar distinguish the two.

Marind roots, as opposed to compounds and inflected forms, invariably take final stress (Drabbe 1955: 16.) This motivates two patterns of syncope in Mbian and Kumb, and sometimes in Atih.

In disyllables where unstressed /*a/ is followed by unoccluded fricative /*h/, it is droppped in (at least) Mbian. Only /*a/ is known to be thusly affected:

One example in which fricative /*s/ and velar non-stop /*ɣ/ has been irregularly deoccluded to /h/ undergoes the same change:

Trisyllables received secondary stress on the first syllable, while the second syllable was destressed. This led to syncope and the formation of medial consonant clusters, which otherwise don't occur in roots, in both Mbian and Kumb and sometimes in Atih:

In some instances, this is followed by the loss of one or the other consonant in the resulting cluster:

One reduplicated compound is likewise affected:

One example undergoes syncope in Kumb, but (apparently) not in Mbian:

The Mbian medial vowel in one example is suspect because medial /v/ is realized as [f], suggesting a cluster (above):

Failure to undergo syncope may indicate compounding or a loan, or that syncope isn't invariable for some other reason:

A small number of roots …:

… reduplications…:

…:

…:

…:

Vowels /*i *u *e *o *a/ occur in all positions. In monosyllables, all five vowels are invariably retained as such in all descendants, as are vowels of identical quality in disyllables.

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:

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:

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

Any vowel /*i *u *e *o/ may follow low central /*a/ in preceding syllables.

Words of the form /*(C)aCi(C)/ as well as /*(C)aCaCi(C)/ are retained as such in all descendants, except where syncope has led to the loss of a medial vowel (above):

…:

…:

Words of the form /*(C)aCu(C)/ as well as /*(C)aCaCuC)/ are retained as such in all descendants, except where syncope has led to the loss of a medial vowel (above):

Words of the form /*(C)aCe(C)/ as well as /*(C)aCaCe(C)/ are retained as such in all descendants, except where syncope has led to the loss of a medial vowel (above):

Words of the form /*(C)aCo(C)/ as well as /*(C)aCoCo(C)/ are retained as such in all descendants, except where syncope has led to the loss of a medial vowel (above):

Any vowel /*i *u *e *o/ may by followed by low central /*a/ in subsequent syllables.

Words of the form /*(C)iCa(C)/ are retained as such in all descendants:

Words of the form /*(C)uCa(C)/ as well as /*(C)aCoCo(C)/ are retained as such in all descendants:

Words of the form /*eCa/ as well as /*eCeCa/ are retained as such in at least Mbian and Southeast Marind:

Words of the form /*(C)oCa(C)/ as well as /*(C)oCaCa(C)/ are retained as such in all descendants, except where syncope has led to the loss of a medial vowel (above):

[under construction]

… /*i/ …

One word with an unusual pattern is mandatorily a compound due to stress rules:

Words of the form /*iCuC/ are reflected as /iCiC/ in Mbian and as /uCuC/ elsewhere:

…:

… /*u/ …:

… /*u₂/ …:

… /*e/ …

… /e/ in Mbian and /a/ elsewhere:

One counterexample, together with rare initial /*mb/, suggests a recent loan:

Words of the form /*CeCa(C)/ …:

One counterexample, in direct contrast to /*seŋga/ “hand” above, is most likely a loan from Mbian into Kumb (cf. /*kataɾ/ in all other descendants):

… /*o/ …

In words of the form /*oCiC/, unstressed /*o/ is raised to /u/ in …:

In words of the form /*oCuC/, unstressed /*o/ is raised to /u/ in …:

… /*a/ …

Vowel sequences realized as diphthongs include /*iu *ei *ou *ai *au *ae/ [*iᵘ *eⁱ *oᵘ *aⁱ *aᵘ *aᵉ]. They are phonetically distinguished from other vowel sequences (below) by stress, as final sequential /*VV(C)/ is realized as [V'V(C)] but diphthongs as ['Vᵛ(C)].

… /*iu/ …:

… /*ei/ …:

…:

…/*oi/ …:

… /*ou/ is retained as such in Mbian and merged with the reflexes of /*aᵘ/ (below) elsewhere:

… /*ai/ …:

… /*au/ …:

Drabbe (1955: 16) states that East /aᵘ/ in the following root, given as “stoken” rather than as “verbranden” (1954: 41,) may be heard as either [aᵘ] or [u]. It differs from the examples above in being followed by a consonant:

… /*ae/ …:

Vowel sequences ….

The sequence /*ia/ is found nearly only initially, where it retained as such in all descendants and is heard as semivowel-initial segment [ja]:

The sequence /*ui/ is retained as such in all descendants. When it occurs initially, it is realized as semivowel-initial segment [wi]:

The sequence /*ua/ is found nearly only initially, where it is retained as such in all descendants and is heard as semivowel-initial [wa]:

The sequence /*eo/, known from only one example, is retained as such in all descendants:

The sequence /*ea/ …:

The sequence /*oa/ is retained in (at least) the East and Mbian dialects:

Longer vowel sequences which include diphthongs also occur.

The sequence /*iua/ is retained as such in (at least) East and Mbian dialects:

The sequence /*aia/ …:

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

The sequence /*aua/ is retained as such in (at least) …:

… /*aiau/ …:

… /*uaia/ and is preserved as such in all descendants: