Indo-Pacific Languages
A Controversial Notion

> Literature references and annotations by Dick Grune, dick@dickgrune.com.
Last update: Sat Sep 14 18:34:00 2024.

These references and annotations were originally intended for personal use and are presented here only in the hope that they may be useful to others. There is no claim to completeness or even correctness. Each annotation represents my understanding of the text at the moment I wrote the annotation.
> No guarantees given; comments and content criticism welcome.

* Zamponi, R., A Grammar of Akajeru: Fragments of a Traditional North Andamanese Dialect, 2021, pp. ??.

ZZ

* Austin, P.K., A Grammar of Diyari, South Australia, 2021, pp. 287.
Diyari is an average, representative, Pama-Nyungan language (most Pama-Nyungan languages are). Its main features are its ergativity and its handing of sub-clauses, both of which are unusual to Western minds. The ergativity can be explained (right below) but the sub-clauses remain just complicated.
     Diyari is a split-ergative language. As in any language a verb (or actually a verb form) is accompanied by one or more noun forms. These noun forms perform certain roles with respect to the verb in the verb form. These roles come with levels of importance. The most important roles are Subject (S) and Agent (A); on the next level we have the Patient (P) aka the Object; then come the indirect object (IO) and/or the Beneficiary (B); and in the periphery we have miscellaneous roles, e.g. Location, Direction, and Instrument. The semantics of each of these roles with a given verb depends on the language, but almost always the S is the only noun form with an intransitive verb, A is the actor of a transitive verb, and O is what is acted on. The other roles are self-explanatory. Note that both S and A cannot occur with the same verb form.
     To show which nouns fulfill which roles, many languages mark nouns for "case", usually by endings. Usual cases are: absolutive (usually marked by the absence of an ending), ergative, nominative, accusative, dative, locative, ablative, instrumental, etc. Note that there is no genitive here: a genitive relates a noun to another noun rather than to a verb; this is often solved differently in non-Western languages.
     The mapping from roles onto cases (called the alignment in linguistics) is in principle arbitrary, but three alignments cover almost all languages:
1. S & A → nominative, P → accusative: most Western languages (except Basque and Georgian);
2. A → ergative, S → nominative, P → accusative: some ergative languages;
3. A → ergative, S & P → absolutive: some ergative languages, e.g. Basque.
B is usually mapped onto the dative, although some languages use the locative or allative for this purpose.
     Most ergative languages do not use one of the ergative alignments consistently throughout, and also use the nominative-accusative alignment in some constructions. This is called split-ergativity. The distribution of the alignments over the language constructions depends of course on the language, but often the choice is determined by the TAM (Tense, Aspect, Modality) of the verb form, as it is in Georgian and Hindi.
     Diyari goes further than this and uses all three of the above alignments, based on the nature of the nominal that occupies the role. For this purpose nominals have to be divided into pronouns, personal names, and the rest, the "common nouns". The Diyari alignments are as follows (Table 3.2 in the paper):
• 1st & 2nd person dual and plural pronouns: alignment 1;
• other pronouns, dual and plural common nouns, female personal names: alignment 2;
• singular common nouns, male personal names: alignment 3.
It is unclear what concepts underlie this distribution.
     Now back to the grammar.
The phonetics of Diyari are simple and systematic: there are 3 articulation regions: labial (p/m), lingual (t/n/l), and velar (k/ng); within the lingual region there are 4 articulation points: dental (-h), alveolar (), palatal (-y), and domal (retroflex) (r-). There are three modes of articulation: plosive, nasal, and lateral (for the linguals only). This yields 2*6+4 = 16 consonants; examples are th, an (inter)-dental t, no relation to the English th; rn, a retroflex n; and ty, a palatalized t. To this are added three rhotics: r, an alveolar flip; r, a retroflex flip; and rr, a rolling `r'. Two glides, w and y, and three vowels, a, i, and u complete the picture. Note that there are no fricatives or sibilants. Particular to Diyari, next to t and rt there are voiced d and rd, which contrast with t and rt in some positions. The digraphs rt and rd are written tr and dr word-initially and after consonants. Unfortunately the difference between dental (th/nh/lh) and alveolar (t/n/l) articulation was not recognized (and thus not recorded) by early researchers.
     Word structure is equally simple. Almost all roots have the form CV(C)CV; examples are: karna `man', thupu `smoke', and nganka `beard'. There is one word of one syllable, ya `and'. There are quite a number of roots of three syllables, e.g. tyukurru `kangaroo'. Almost all endings are two syllables long (suggesting that they were once roots). In general words forms with an even number of syllables are preferred: if possible odd-length words get special endings to make them even-length: kanku `boy' has ergative kankuyali, and pinarru `old man' has ergative pinarrali. Most roots can be reduplicated and they frequently are; the resulting meaning is usually diminutive for nouns, emphasis for adjectives, and repetition or continuality for verbs, but is sometimes unrelated: yatha- `to speak' gives reduplicated yathayatha- `to converse' but the reduplicated form of karra- `to tie', karrakarra-, means to feel'.
     Nouns can be singular (no suffix), dual ( -wurla), or plural (-wara). They occur in 8 cases: absolutive (no suffix), ergative (-li), nominative (-ni), accusative (-nha), dative (-ya), locative (-nhi), allative (-ya), and ablative (-ndru). They are expressed by attaching the above endings, or a variant thereof, to the noun with no or minimal modification. Examples are:
     kinthala-li dog-ergative `a dog' (as Agent),
     wilha-wurla-rni woman-dual-dative `for the two women',
     karna-wara-ngundru man-plural-ablative `from the men'.
The nominative dual has an irregular form: -wurlu.
     Pronouns are used extensively in Diyari. The usual 1st, 2nd, and 3rd person are distinguished, in singular, dual, and plural; 3rd singular is split in feminine (`she') and non-feminine (`he'/`it'); 1st person dual and plural are split in inclusive (including the speaker) and exclusive; in total 12 pronouns. Each pronoun comes in 6 cases, although the language has 8 cases. Because pronouns occur only in alignments 1 and 2 (see above) and the absolutive is used only in alignment 3, the pronouns have no absolute form [DG: this seems to me to be significant of something]; also for pronouns the locative and allative coincide -- the accompanying verb should do the disambiguation. Table 3.6 in the paper gives the 68 case forms of the pronouns (6 x 12 = 72, but because the 1st and 2nd person dual and plural pronouns are used in alignment 1 only, they have no ergative); the declination is fairly but not complete regular, and it is not always obvious what is the root. In the absence of an absolutive, the nominative has the shortest forms:

     1 sing nganhi `I'
     2 sing yini `you'
     3 sing feminine nhani `she'
     3 sing non-feminine nhawu `he/it'
    
     1 dual inclusive ngaldra `you and me'
     1 dual exclusive ngali `I and he/she'
     2 dual yula `you two'
     3 dual pula `they two'
    
     1 plur inclusive ngayana `we and others'
     1 plur exclusive ngayani `we'
     2 plur yura `you all'
     3 plur thana `they'
    
It would seem that the first person root might be nga- and that the 2nd person root yu-; no root for the 3rd person suggests itself. The dative of the personal pronoun is used as the possessive pronoun: nga-karni `my', dative of nga-nhi `I', and as such it can be inflected for case:

    ngakarni    kinthala  pirna  marla  yingkarna-nhi
    1.sing.dat  dog.nom   big    more   2.sing.dat-locative
`My dog is bigger than yours', which shows that the locative is also used for comparisons.
     The function of a verb form in a sentence is indicated by an ending on the verb (Table 3.9 in the paper). There are 4 TAM endings, for present, past, imperative and optative; 2 endings for participles (non-future and future); and 7 for sub-clauses, the mainstay of Diyari syntax. Although there is a past ending, the past is more often expressed by a participle and an auxiliary verb in the present tense; this is comparable to English "saw" versus "has seen".
     There are four types of clause endings: imperfect, perfect, "implicated", and "lest". The first three have different forms for same-subject (SS) and different-subject (DS), depending on whether the subject of the sub-clause is the same as that of the main clause or not.
[DG: There is a lot of literature on this "switch-reference", and in some languages it may help to disambiguate e.g. implicit pronouns, but in Diyari all pronouns are always explicit, so it seems to me that in Diyari it just supplies (useful) redundancy, just like gender in nouns in most languages.]
     The imperfect, perfect, and implicated clauses have a large number of specific meanings, often depending on the auxiliary verbs used in them, but basically they live up to their names:
• imperfect: anything that has not finished (and may not even have started) at the time implied in the main clause; this expresses "if", "when", "while", etc.
• perfect: anything that has finished at the time implied in the main clause; this expresses "after having ...", "because", etc.
• implicated: anything that is indirectly involved in the contents of the main clause; this expresses notions like "in order to", "just at that moment", "so that", etc.
• lest: something undesirable which should be prevented; often with an imperative in the main clause:
    ngama-mayi     yura        puri-yathi.
    sit.imper-emph 2plural.nom fall-lest.
`Sit down or you’ll fall'.
     There are no relative clauses: "the man I saw" is rendered as `the man, I saw him (imperfect clause DS), '. Verbs are not conjugated for person, which is always provided by personal pronouns.
Example sentences:
    kinthala-li nhungkarni-yali        nganha  matha-rna     wara-yi.
    dog-erg     3singnonfem.dat-erg   1sg.acc bite-particple aux-pres.
‘His dog bit me.'
    kanku-yali wata yani-ya        thayi-rnanthu nganthi waka
    boy-erg    not  like_this-near eat-implDS    meat    small.acc
‘Boys shouldn’t eat small animals like these ones here.'
     Diyari personal pronouns: 1sg: nga-; 2sg: yin-.
     There is a companion dictionary with about 700 Diyari terms by the same author (2013).

* Rowland, M.J., 65,000 Years of Isolation in Aboriginal Australia or Continuity and External Contacts?, 2018, pp. 33.
Even in precolonial times there were visitors to Aboriginal Australia. There were Macassan fishermen and traders, and the dingo was imported about 3500 years ago; and there are hints of others. However, these probably incidental visits left hardly any linguistic or genetic evidence. The paper has many details.

* Abbi, A., A Sixth Language Family of India: Great Andamanese, 2018, pp. ??.

ZZ

* Remco R. Bouckaert, Claire Bowern, Quentin D. Atkinson, The Origin and Expansion of Pama-Nyungan Languages across Australia, 2018, pp. 9+3.
Starting with word lists of 306 Pama-Nyungan languages and their locations, assuming a replacement rate of ??%/kyr [DG: I could not squeeze a number for the assumed word replacement rate from the paper], and a migration speed of 0,14km/yr, Bayesian analysis is used to "run the clock backwards" to infer the origin of the Pama-Nyungan languages. The results place this origin at 4500-7000 years ago, somewhere perhaps a 300 miles south of the Gulf of Carpentaria. Both results are reasonably insensitive to changes in the input parameters. These results are then related to climate change and change in hunter-gatherer technology. The results also indicate that expansion over water along the coast played no role.
     There are two maps showing two (almost identical) expansion scenarios, and an intricate language tree of the Pama-Nyungan languages. And there is a fair deal of math concerning Bayesian analysis.

* Nordlinger, R., The Languages of the Daly River Region (Northern Australia), 2016, pp. 40.
There are 22 different aboriginal languages known from the Daly River Region. Of them, Murrinhpatha has the most speakers (about 2500); some others are extinct (e.g. Yunggurr). The languages are all related and five sub-families can be distinguished.
     The centerpiece of the Daly languages is the polysynthetic verb form which describes the action or situation in amazing detail. This verb form (called a complex verbal predicate) usually contains two verb roots, one lexical, and one classifier. The lexical root corresponds more or less to a traditional verb, and there are very many of them. The classifier expresses things like intensity, intent, and duration; their number is limited to perhaps some dozens. Both of these are usually surrounded by some particles that specify person, time, aspect, mode, etc., creating two information clusters. Together they take care of the "amazing detail" mentioned above.
     The Daly languages differ in how they combine these two clusters: in some the lexical cluster comes first, in others the classifier cluster comes first; and in some languages they form an indivisible unit, in others both clusters remain separate words.
     An example of a complex verbal predicate from Murrinhpatha is wurdanangayithnginthayu, which consists of the following parts:

    wurdana-                  nga-     yith- ngintha- yu
    3SG.SBJ.SHOVE.RR.PST:PFV- 1SG.DAT- tell- DU.NSIB- DM
    ‘They (two non-siblings) were telling me.’
The meaning of the whole can be understood from the meanings of the parts:
     3SG.SBJ.: some unspecified actor or actors, not you or me, ...
     SHOVE.RR.PST:PFV-: ... repeated (mode RR) directed action (classifier SHOVE);
     1SG.DAT-: the direction was towards me;
     tell-: the action was "to tell" (the lexical stem);
     DU.NSIB-: there were two actors and they were not siblings;
     DM: Discourse Marker: this is a statement.
It may already be surprising that the language carries the property of being siblings (by some definition) into the morphology, but that's not the end of it. Walsh (1976) tells us that the non-sibling dual is split up in "both male" and "including a female", and that -ngintha- is the particle for "all male", the full form being ŋankuninda (Walsh's old spelling). So ‘These two guys were telling me’ would be an idiomatic translation.
     Some classifier verbs specify a tool, e.g. "to do by hand", and it is easy to see how this leads to incorporation, as in Eng. "hand-painted". The Daly languages differ in what can be incorporated, and also in the distribution of the particles over the slots of the polysynthetic verb form. The specifics of 8 Daly languages are given in tabular form.

* van den Berg, R., The Loss of Clusivity and the Rise of Gender in West Oceanic Pronominals, 2015, pp. 39.
Most Austronesian and almost all West-Oceanic languages distinguish an inclusive 'we' (Indonesian kita, root -da) from an exclusive 'we' (Indonesian kami, root -mi), but a few West-Oceanic languages have lost this feature. Conversely almost all West-Oceanic languages lack gender distinction (Indonesian dia is both "he" and "she"), but a very few have acquired this feature.
[DG: "West-Oceanic languages" are roughly the Austronesian languages spoken on New Guinea + surroundings.]
This raises two questions: 1. How did these exceptions form? 2. Are the processes related (perhaps compensatory)?
     A good reconstruction of the Proto-Oceanic pronoun system exists ("The Oceanic Languages", by Lynch, Ross, and Crowley (2002)). It features a free pronoun and a possessor postfix for all persons, and a subject prefix and an object postfix for singular and 3rd plural. The pronoun system of 120 out of perhaps 260 West-Oceanic is compared to this. Twenty-six of the systems showed signs of decay, of which seven are analysed in detail. A result is that the decay can start essentially anywhere, but that the difference in the free pronoun is the last to go. Whereas the languages with partial loss can be found all over New Guinea, the seven languages with compete loss all belong to the Siau subfamily spoken along the north coast of New Guinea. It is assumed that the difference was already lost in Proto-Siau. Influence of the surrounding non-Austronesian languages is unclear: many West-Oceanic languages have taken over the alien SOV word order but kept the inclusive/exclusive distinction.
     The two creole languages that exist in New Guinea both support inclusive/exclusive. The English-based "Tok Pisin" has inclusive yumi (from Eng. you-me) and exclusive m-pela (from Eng. me-fellow)). The German-based "Underdeutsch" has inclusive wir (from Grm. wir `we' (nominative)) and exclusive uns (from Grm. uns `us' (accusative)).
[DG: Although one line in the text suggests the opposite, the examples and further text show that the above assignment is the correct one.]
     Nine out of the 120 examined languages has pronouns that depend on gender, and they are a varied lot. In Kilivila the distinction is a result of the classifier system: mtona `he' = m-to-na = `the <male_classifier> here' and minana 'she' = m-tina-na = `the <female_classifier> here'. Kaulong, Miu, Sengseng, and Bebeli have non-composite pronouns gender-dependent pronouns. In Aighon and Akolet the pronouns depend not only on the gender but also on the marital status: Aighon tee `he, unmarried', vee `he, married', too `she, unmarried', ee `she, married'. All these pronouns are for the 3rd person singular. Akolet has similar forms for the 3rd plural too. Examples from Cheke-Hole suggest that its more than a dozen pronouns may depend not only on the sex of the referent (`he' or `she') but also on that of the speaker and/or that of the addressee.
     In hardly any case can the deviation from the original West-Oceanic structure be attributed to neighbouring (Papuan) languages. Loss of clusivity started in different places in different languages. And the rise of gender distinctions resulted in quite different systems in different languages.

* Miceli, L., Pama-Nyungan, 2015, pp. 22.
The effort to construct a robust and reliable language tree of the Pama-Nyungan languages is severely hindered by three phenomena:
1. The phonetic systems of almost all Pama-Nyungan languages are very simple (no voicing contrast, no fricatives, three vowels, no nasalisation) and very similar, and the place of articulation plays a far more important role than the manner of articulation. This makes it difficult to find non-trivial, significant sound laws.
2. Words for the same concept in two different languages, regardless of whether these languages are located next to each other or thousands of miles apart, are either (almost) the same or completely different. This suggests that a word either does not change or is replaced.
3. Most Pama-Nyungan languages have several different words for each concept, which makes finding sets of cognates showing regular sound changes (required for proving relatedness) difficult. This richness of words may be caused by the multilingualism prevalent in aboriginal Australia and the heavy borrowing that that entails.
     Study of code-switching by multi-linguial speakers and their contribution to language change may help in explaining the Pama-Nyungan situation.

* Austin, P.K., A Dictionary of Diyari, South Australia, 2013, pp. 41.
Companion to Austin's 2021 grammar of the Pama-Nyungan language Diyari. Contains about 700 words in Diyari alphabetical order D-Y, many of them illustrated by a short sentence or a picture (for local animals and cultural objects).

* Donohue, M., Towards a Papuan History of Languages, 2013, pp. ??.

ZZ

* Abbi, A., Is Great Andamanese Genealogically and Typologically Distinct from Onge and Jarawa?, 2009, pp. 22.
Argues that the inhabitants of the Great Andamanese islands speak languages from two different phyla, Great Andamanese and Ongan, both without further relatives. (See, however, Blevins, 2007, `A Long Lost Sister of Proto-Austronesian?'.)

* Blevins, J., A Long Lost Sister of Proto-Austronesian? Proto-Ongan, Mother of Jarawa and Onge of the Andaman Islands, 2007, pp. 45.
Constructs a (tenuous) relation between proto-Ongan and proto-Austronesian.

* Ross, M., Pronouns as a Preliminary Diagnostic for Grouping Papuan Languages, 2007, pp. 52.

Previous grouping of the Papuan languages has left contradictory results, and the author blames mass comparison. Returning to the comparative method, the seven steps of which are enumerated very clearly in Table 3 in the article, it is pointed out that it starts with determining what to compare. One needs a set-forming criterion (a diagnostic) to determine the set of languages that one is going to compare.
     The author chooses similarity between the personal pronouns of the languages. This results in a grouping in families (or phyla?) that is very different from previous groupings. A group similar to Trans New Guinea (TNG), characterized by 1sg na, 2sg [ŋ]ga, emerged, but few of the other previous groups were recognized. The reasons for this are analyzed. The analysis involves interesting considerations about the coast lines of the island sea from 6000 BP to today.


     Next to TNG 23 non-TNG fsmilies were recognized.

ZZ

2.4 Papuan families other than Trans New Guinea

pg 7: Takia and Maisin no TNG pronouns: Maisin 1sg a-, 2s ko-; Oceanic.

pg 16: In Northwest Melanesia: Sulka and Kol (on New Britain) and Kuot (on New Ireland), classified by Wurm (1975) as members of his East Papuan phylum

pg 24: Yapen Island: In the northwest area, the most striking feature is that three groups recognised in Wurm's classification, namely West Papuan, East Bird's Head and the Yava group (the part of the Cenderawasih Bay family situated on Yapen Island) may perhaps constitute a single ‘extended West Papuan' linkage. [Also Wikipedia Yawa_languages]
Also apparently belonging with the East Bird's Head group are the Sentani group (a TNG group in Wurm et al.'s classification), the isolate Burmeso=Taurap (1sg da, 2sg ba), as well as Tause and Weirate, two languages included in Clouse's (1997) classification.
Languages of all these groups except the Amberbaken subgroup of West Papuan reflect a form *da or *di ‘I'. West Papuan languages reflect *na or *ni ‘you (singular)', whilst East Bird's Head, Sentani, Burmeso=Taurap and Tause/Weirate reflect *ba or *wa ‘you (singular)'.

again paper no details, more how this came to be.

pg 42: North Halmahera + Bird's Head: 4 languages + proto-NH. Unnumbered Table given on pg 43. points out that Proto North Halmahera *mi- 1PL.EXC need not be a borrowing from Austroneasian kami, as per Voorhoeve, but may derive from ""*pV- may have been a first person plural prefix which did not make the exclusive/inclusive distinction.


     Bernouilli's theorem about "probability" vs. Bayesian probability (pg 45)

====

* Dol, P., A Grammar of Maybrat: A Language of the Bird's Head Peninsula, 2007, pp. 346.

pg 74: 11sg tuwo, 2sg nuwo

ZZ

* Joseph Biddulph, Some Languages of the Pacific Region, Joseph Biddulph Publisher, Pontypridd, Wales, 2005, pp. 44.
An encyclopedia of about 250 lemmata of the languages along the Pacific, from Abau (a Papua language of Upper Sepik) to Yis (another Papuan language). The sizes of the lemmata range from one line to three pages (Maori), with about 7 languages per page on the average.

* Whitehouse, P., Usher, T., et al., Kusunda: An Indo-Pacific Language in Nepal, 2004, pp. 4.

Takes Indo-Pacific as a given (the third author is Merritt Ruhlen and the paper was communicated to PNAS by Murray Gell-Mann).

ZZ

* Evans, N. (ed.), The non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent's Most Linguistically Complex Region, 2003, pp. 522.
Impressive collection of 16 papers covering the non-Pama-Nyungan languages from the Kimberley, Daly, Arnhem Land, and Barkly regions.
     The first paper is concerned with the place of the Pama-Nyungan languages in the Australian languages. Four models are considered:
• the rake model: PN is just a (largest) language family among the 28 recognizable language families on the Australian continent. [DG: just as TNG is just the largest family under the 26 families on Irian Jaya in Voorhoeve's classification.]
• the diffusion model: the non-PN languages are just PN languages that have become polysynthetic, and the diffusion started from the North-West coast.
• the binary model, which has PN and non-PN as the first families under proto-Australian, and
• the offshoot model, which has a sequence of non-PN families, each one deriving from its predecessor, with the PN family as the most recent offshoot.
Mainly because the binary model does not explain enough the similarities between non-PN and PN, the author prefers the offshoot model. Also the offshoot model allows properties of PN to be explained from non-PN [DG: but so does the diffusion model.]

the last families before offshoot PN are Gunwinyguan, Tangkic, and Garrwan.


     Chapters 6 and 7 show many interesting details of the Daly River languages, especially Matngele. ZZ

* Breen, G., Wanyi and Garrwa Comparative Data, 2003, pp. 38.
Chpter 14 of Evans 2003).

* Lindström, E., Topics in the Grammar of Kuot: A non-Austronesian Language of New Ireland, Papua New Guinea, 2002, pp. 264.

Kuot, aka Paranas, is a very complicated language. For a quick first impression think Georgian + gender marked on almost every form + not well understood phonetic processes − the gutturals and glottals. It is spoken in a few villages on the West coast of New Ireland, an island off the East coast of New Guinea. It differs greatly from all languages in the neighbourhood, including the Papuan and Australian languages, and is very probably an isolate.


     The phonetics are superficially simple. There are 13 consonants, 7 of which have variants in certain contexts: p/v, t/r, k/γ, b/mb, d/nd, g/ŋg, m, n/l, ŋ, and a real (non-variant) r and l, plus an f and an s which do not occur in the morphology. The consonant set may be simple but the rules governing the occurrence of the variants are not. The γ is written `g' in written Kuot, which causes confusion
[DG: `h' would have been appropriate and available.]
There are 6 vowels, 5 of which with variants: a/ə, e/ε, ə, i/ɪ, o/ɔ, and u/ʊ. The ə is both a variant of a and a vowel in its own right (e.g. gas `story' versus gəs `possum'). This causes problems both in the spelling, where the ə is written `a' and in the theory, where the rules governing the variation a/ə have not yet been fully determined.


     Nouns have gender: they can be masculine or feminine. Gender is lexical: except where the noun refers to a being with natural gender, the gender has to be learned with the word:

The gender of a noun shows up in the morphemes on all words that refer to the noun. These morphemes are different for masculine an feminine in the singular but not in the dual or plural:

Verb forms are marked with one or two morphemes which refer to the pertinent Subject, Actor, and/or Object. The way these morphemes are attached to the verb stem depends on the verb (??). There are 7 classes to distinguish: I, II, and III for intransitives and I, IIa, IIb, and III for transitives. These would be called "skreeves" in Georgian grammar. Verb stems can consist of two parts, leaving a place for a morpheme between them:

Adjectives act like intransitive verbs with some minor differences.

.pp

(73): pronunciation attrubuted to tranns/intrans difference

ZZ 1sg -tuo, 2sg -nuo,

* Donohue, M., Animacy, Class and Gender in Burmeso, 2001, pp. 20.

ZZ

* Rumsey, A., Bunuba, 2000, pp. 121.
Bunuba is a non-Pama-Nyungan language of the Bunuban family, one of the 26 or so non-Pama-Nyungan language families. It is spoken around Fitzroy Crossing in North-Western Australia.
     Whereas the Pama-Nyungan languages are essentially simple agglutinative languages, structurally even simpler than Turkish or Mongolian, this non-Pama-Nyungan language is of Caucasian complexity, featuring very complex verb forms. However, with respect to phonetics Bunuba differs hardly from the Pama-Nyungan languages.
     The basic verb form is verb_stem TMP_prefix+auxiliary_verb+TMP_suffix, where TMP_prefix and TMP_suffix specify Tense, Mode, and Participants. [DG: The verb stem is called "preverb" and the auxiliary verb is called "verb root" in the paper. Also "future" seems to mean "non-Past".] There are nine auxiliaries, which specify telic/atelic, change of state or motion, etc. Their forms differ for Present/Irrealis and Past/Exclusive, sometimes considerably: f.e. auxiliary WU2 (telic: impact upon) has representations -iy and -nu, respectively.
     Pronouns can have tens of different forms, depending on Tense and on which auxiliary they are applied to. Again these differences can be considerable: 1st person singular is ng- for auxiliaries RA and NI, and l- for MA and WU. [DG: This raises the question whether it is justified to split the auxiliary form into three parts, if each part has a small number of representations and they influence each other strongly. Reminiscent of Basque.]
     A peculiarity of the language is that it has a speech style, called Gun.gunma, which is (was) used between a man and his mother-in-law. Such styles exist elsewhere and usually differ from the unmarked style in vocabulary only, but in Bunuba there are also differences in grammar. Transitive verbs are replaced by intransitive ones, and the original Object reference is replaced by an Oblique reference at the end of the auxiliary form; also the (few) examples seem to show that the combined 1sg/2sg pronouns are avoided and are split over an additional auxiliary.
     Pronouns: 1sg: often characterized by a form starting with ng- or l-; 2sg: usually characterized by a form containing -ngg-.

* O'Grady, G., Pama-Nyungan II and Tasmanian, 1993, pp. 7.
Published in the September issue of Mother Tongue 1993.
In this second part of his personal paper on the Australian+Tasmanian phylum, the author resolves the 114 forms of the Spring issue to 34 reconstructed roots. The derivation rules are not stated explicitly, but demonstrated in 44 derivations. Of these, 18 connect words with the same meaning from different languages into one proto-form; 5 connect antonyms; and 21 connect conceptually related words.
     To support the inclusion of the Tasmanian language(s), the author suggests that the decidedly un-Australian first person pronoun mina may be related to Pama-Nyungan words for 'eye' mina or 'body' marram, the original form having fallen to taboo.
[DG: This would make Tasmanian mina a loan from Pama-Nyungan, without any impact on the relationship between Tasmanian and Australian (except that there was contact).]

* O'Grady, G., Pama-Nyungan: An Entirely Viable Family-Level Construct within the Australian Phylum, 1993, pp. 15.
Published in the Spring issue of Mother Tongue 1993.
Interesting personal account of the author how he in 1949, retrieving sheep and fixing fences, got in contact with the world of the Aboriginals. This is followed by a very condensed overview of the research on aboriginal languages, from the first recorded words of Guugu Yimidhirr, which established it as the single Australian language at that time, via the discovery of two other languages near Sidney 65 km apart that were quite different from each other and from Guugu Yimidhirr, to the discovery that some languages near Perth were similar to some near Sidney, a distance of about 3.300 km. So Australian languages could be quite different from their direct neighbours and yet be similar to languages on the other side of the continent.
     Order in this chaos was brought by Arthur Capell in the 1940s by dividing the aboriginal languages in those that use prefixes on the verb root and those that do not (all aboriginal languages use suffixes). This was confirmed by Kenneth Hale in the 1960s, using Lexicostatistics, which produced 28 small families of prefixing languages and one very large family of non-prefixing languages. The very large family which covered the entire south-west, central, and east of the continent, was called Pama-Nyungan, the 28 small families could be found on the north-western coast and inland, and were quite logically called non-Pama-Nyungan. This model has stood up to this day, with only little modification.
     This part of the paper ends with a list of 114 words from 42 Australian and Tasmanian languages + Bahasa Indonesia (for Austronesian loans), apparently as a kind of summer challenge for the etymological abilities of the Mother Tongue readers.

* William A. Foley, The Papuan Languages of New Guinea, Cambridge University Press, Cambridge, 1986, pp.
poser@Csli.Stanford.EDU (Bill Poser) writes: Contains both structural descriptions of Papuan languages and information on classification and prehistory.

* Rumsey, A., An Intra-Sentence Grammar of Ungarinjin: North-Western Australia, 1982, pp. 193.
Ngarinjin (also Ungarinjin) is a non-Pama-Nyungan language, and correspondingly complex.

ZZ
     Ngarinjin personal pronouns: lsg ŋin; 2sg njaŋan.

* McConvell, P., How Lardil Became Accusative, 1981, pp. 39.
Almost all Australian languages are ergative, but there are two small regions of accusative Australian languages: one around Roebourne on the north-west coast, and one about a 1000 km further to the east on Wellesley Island in the Gulf of Carpentaria. Those around Roebourne have acquired their accusative structure recently, but the ones on Wellesley Island are considered by Hale (1970) as possible remnants of the proto-Australian language. This paper shows in detail how Lardil, one of the Wellesley Island languages, became accusative from being ergative originally, thus refuting Hale's suggestion that Lardil may be an ancestral accusative language.
     The argument follows from the close relationship between Lardil and Yukulta, a (mainly) ergative language spoken on the Gulf coast adjacent to Wellesley Island. Both are members of the Tangkic family (which is non-Pama-Nyungan).
     The paper features introductory grammars of Yukulta and Lardil (8 pages each). Next Hale's arguments for an Ancient Australian Accusative language are examined in the light of these two grammars. Finally, using the grammar of Yukulta as that of proto-Tangkic, a scenario is sketched showing how Lardil became accusative.
     Tangkic personal pronouns: 1sg: ngi-; 2sg: nyi-.

* Z'graggen, J.A., A Comparative Word List of the Rai Coast Languages, Madang Province, Papua New Guinea, 1980, pp. 198.
Word list of about 300 words each are provided for 28 languages from the Rai coast of Papua New Guinea, from the author's own field work. Personal pronouns and a number of verb affixes are also provided.

* Walsh, M.J., The Muɹinypata Language of North-West Australia, 1976, pp. 21+442.
The language known today as Murrinh-patha.
This summary covers the personal pronouns only; they occupy pages 150-162 of the thesis, Sect. 4.1.1 Pronominal Morphology. The personal pronouns form a fairly orthogonal system with three axes:
     number: singular, dual, paucal (3-±10), plural (±10) and above;
     person: 1st, 1st+2nd, 2nd, 3rd;
     gender: all male, at least one female.
On top of that the dual distinguishes between siblings and non-siblings. Together this creates 4×4×2+4=36 combinations, but not all combinations are possible, and the 10+-plural forms are the same for all male and at least one female; in total 26 combinations remain.
     It is interesting to see that the sibling dual is the basic form (ŋanku) and the non-sibling forms are marked (ŋankuninda all male, and ŋankuŋinda at least one female). These forms are the stand-alone forms. They are reduced when used used as particles in verb forms: ŋankuninda appears as -ngintha- in Nordlinger (2016) (d is the author's spelling for th).
     More internal structure of the personal pronouns is discussed in the rest of Sect 4.1.1.

* Voorhoeve, C.L., Isolates: Irian Jaya, 1975, pp. 4.
After the recognition of Morwap, Molof, Usku, Tofamna, and Kaure as (probably) belonging to the TNG phylum, only two isolates remain in Irian Jaya: Taurap (= Tauraf = Borumeso = Burmeso), and Warenbori (= Warembori = Waremboivoro). The first is characterized by 1sg dawo, 2sg bawo; the second by 1sg iwi, 2sg awi.

* Voorhoeve, C.L., Languages of Irian Jaya: Checklist. Preliminary Classification, Language Maps, Wordlists, 1975, pp. 144.
The paper presents word lists of 40 each for 199 languages from Indonesian New Guinea. There are 39 Austronesian and 152 non-Austronesian languages. Based upon lexicostatistical analysis the latter are structured into 7 phyla and 3 isolates. The largest phylum is TNG, with 26 stocks. Eight languages were left unclassified due to insufficient data.

* Hale, Kenneth, The Passive and Ergative in Language Change: The Australian Case, 1970, pp. 26.
Argues that the original Australian (= Pama-Nyungan here) language was an accusative language, not an ergative language, as most of them are now. So it has to be explained how these languages become ergative.
     A path is sketched how an accusative/active language can become ergative, and each of the steps on the path is shown to be plausible in an Australian context. The path consists roughly of the following steps.
     We start with the structure of an active/accusative sentence with a transitive verb, consisting of an agent marked as nominative, a patient/object marked as accusative, and an active/transitive verb:

	(Agent,Nom.) (active_verb) (Patient,Acc.)
	 The child    broke         the glass
To this active sentence a passive sentence corresponds, in which the focus is on the patient, and the agent is not mentioned; it requires the verb to be replaces by a passive variant of the verb:
	(Patient,Nom.) (passive_verb)
	 The glass      was broken / broke
As a side line we note that this has the same structure as a stative sentence:
	(Subject,Nom.) (stative_verb)
	 The glass      was blue
Back to the passive sentence. The agent can be re-introduced, but cannot be marked as nominative, nor as accusative; it is marked as ergative:
	(Patient,Nom.) (passive_verb) (Agent,Erg.)
	 The glass      was broken     by the child
When this sentence structure becomes the preferred one in a language, the language becomes typologically OVS, which is unstable, but which can easily be converted to the most frequent type SOV by moving the subject to the front:
	(Agent,Erg.)  (Patient,Nom.) (ergative_verb)
	 by the child  the glass      broke
This results in the well-known ergative sentence typology:
stative:	(Subject,Nom.) (stative_verb)
ergative: 	(Agent,Erg.) (Patient,Nom.) (ergative_verb)
In this context the nominative is often called absolutive; in many ergative languages its representation is the empty ending .
     The above transformations are easier if 1. the morphological difference between active and passive verb forms is slight (as in the English example); 2. the language has little or no agreement between the verb and the subject (unlike the English example, where the verb still agrees (though invisibly) with the glass rather than with the child); 3. the language has (relatively) free word order. All three conditions hold the majority of the Pama-Nyungan languages.
     It is pointed out that due to lack of diachronic data this scenario is a suggestion rather than a proof. However, diachronic data may be replaced by synchronic data from related languages in different stages of development. In the context a connection between the L-conjugation of some languages and a or -yi reflexive/passive marker in some other languages is examined but hesitantly rejected.

* van der Stap, P.A.M., Outline of Dani Morphology, 1966, pp. 203.
The author, a member of the Order of the Friars Minor, lived for 7½ years in th Balim Valley in New Guinea (Irian Jaya), where he learned the Dani language without prior knowledge. This paper is an account how the author determined the use and meaning of the many morphemes in the Dani language by systematically creating minimal pairs of sentences, and asking native speakers about them, native speakers of all ages, for not more than 10 to 15 minutes, and then coming back after half a year or so, with the same minimal pair to the same or different informants. It's not exactly a peek in the kitchen but it comes close.