Tangaeauan

[Darren]

Tangaeauan is the easternmost Polynesian language, spoken by around 1,000 people on the recently-discovered island of Tangaeau off the west coast of Ecuador. This island was previously unknown to science since in the 1970s an amateur Ecuadorian film crew had unofficially made a series of bootleg Smurf films there, during the production of which they painted the entire island, and all of its inhabitants blue, rendering it camouflaged perfectly against the sea in satellite imagery. It was only following unseasonably heavy rains in June 2024 that the paint finally washed off and Tangaeau appeared on Google Maps for the first time. It is believed to have been settled between 1200 and 1300 from Hawai'i, although this has not been confirmed by radiocarbon dating as the results haven't gotten back from the lab yet.


Phonology

Phonologically Tangaeauan continues the trend of more peripheral Polynesian languages shrinking their consonant inventories. The tables below shows four Polynesian languages; Hawaiian, Cook Islands Maori and the Rurutu and Ra'ivavae dialects of Austral, all of which are on or near the edge of Polynesian expansion, and all of which have innovated inventories of eight consonant phonemes:

Code: Select all

  Hawaiian       C.I. Maori     Rurutu        Ra'ivavae

 p     k  ʔ       p  t  k       p  t  ʔ       p  t  ɢ  ʔ
 m  n             m  n  ŋ       f             v        h
 w  l     h       v  r          m  n          m  n
                                v  r

Although note also that Anutan (Solomon Islands) has an inventory identical to that of Cook Islands Maori despite being fairly central. Nine-consonant inventories are found in Luangiua, Mangarevan, Colloquial Samoan, Sikaiana, Tahitian, some Marquesan dialects and Kapingamarangi, at varying degrees of periphery. Whatever the validity of this trend, Tangaeauan seems to continue it with an even smaller inventory of seven segments:

Code: Select all

 Tangaeauan

    t  k
 m     ŋ
 w  l  h

Scientists suspect that should another Polynesian language be found somewhat further east in say Brazil, it would only have six consonants (Wiggs and Dagfin propose /p k ʔ m ŋ h/, although on what basis they refuse to elucidate). Should Polynesians have somehow managed to cross the South Atlantic and reached Africa, they would be entirely out of consonants by the time they found themselves on the border of Tanzania and would have to start digging into their vowels, with death following shortly thereafter.
​This combines with an unremarkable five-vowel inventory with a length contrast. Allophony is not rampant; /ŋ w h/ may be [n v ʃ] before /i(ː)/ and /k/ has a tendency to become [x] or [ɣ] between two vowels especially if one or both of them is /a/, e.g. hakaka "family tree" → [haxaxa], [haɣaɣa], while liki "chisel" is more often [liki] than [lixi] and rarely ?[liɣi]. /w/ is quite rare, accounting for only 1% of all consonants; the next-rarest is the other labial /m/ at 8% while the three most common /k h l/ are all around 20%.
​The table below shows the simple unconditioned reflexes of proto-Polynesian consonants in Proto-Polynesian (PPO), Proto-Nuclear Polynesian (PNP), Proto-Eastern Polynesian (PEP) and Proto-Marquesan (PMQ), as well as the four other languages of the Marquesan branch apart from Tangaeauan – viz. Hawaiian (HAW), the two dialects of Marquesan (North, MQN and South, MQS) and Mangarevan (MAN). (This is taken primarily from Bruce Biggs's The History of Polynesian Phonology, 1978).

Code: Select all

 *PPO   *p   *t   *k   *ʔ   *f   *s   *h   *m   *n   *ŋ   *w   *l   *r 
 *PNP   *p   *t   *k   *ʔ   *f   *s    Ø   *m   *n   *ŋ   *w      *l   
 *PEP   *p   *t   *k   *ʔ   *f   *s        *m   *n   *ŋ   *w      *r
 *PMQ   *p   *t   *k    Ø   *f   *h        *m   *n   *ŋ   *w      *r
  MQN    p    t    k         h    h         m    n    k    v       ʔ
  MQS    p    t    ʔ         f    h         m    n    n    v       ʔ
  MAN    p    t    k         ʔ    ʔ         m    n    ŋ    v       r
  HAW    p    k    ʔ         h    h         m    n    n    w       l

  TAN    k    t    h         k    h         m    ŋ    ŋ    w       l

Within the Marquesan branch, the change *k → ʔ is seen in South Marquesan and Hawaiian (it is also attested in at least six other Polynesian languages, and for that matter numerous languages of Oceania), followed by *ʔ → /h/ in Tangaeauan. This produces a large gap in the phonetic space for obstruents between the alveolar and glottal POAs; Hawaiian "repairs" this with *t → /k/, while Tangaeauan has instead *p → /k/, making it unique amongst the Polynesian languages – all others retain *p as /p/. Related to this is the change of *f → /k/, which at first glance seems bizarre, but most likely predates *p → /k/ and probably merged with *p at an earlier time – compare Anutan which has *f → /p/ also. Also distinctive to Tangaeauan is the shift of *n → /ŋ/, the opposite to what is seen in South Marquesan and Hawaiian (although some words have *ŋ → /h/ presumably through *ŋ → *k → *ʔ). *n → ŋ is also found in colloquial Samoan and Luangiua (and some other more distantly related Oceanic languages to the west). The phonemes *t, *h, *m, *ŋ, *w and *l are left unchanged; the same can generally be said for vowels as well, except the sequences *ai au ia ua a(e,o) (e,o)a which became *ae ao ea oa aa aa, the diphthongs with high vowel glides being replaced through metathesis later (see below). Examples of these basic shifts are shown below:

Gloss	Proto-Polynesian	Hawaiian	Mangareva	Marquesan	Tangaeauan
"rotate"	*wili	wili	viʔi	viri	wili
"slip"	*seke	heʔe	ʔeke	heke/heʔe	hohe
"hurt"	*mamae	mamae	mamaa	mamae/memae	mamaa
"land, country"	*fenua	honua	ʔenoa	henua/fenua	koŋoa
"side, rib"	*kaokao	ʔaoʔao	kaokao	kaokao/ʔaoʔao	ahaa
"end wall"	*tara	kala	?	taʔa-	tala
"assist"	*pale	?	paʔe	pare	kale
"canoe skid"	*laŋo	lona	laŋo	ʔako/ʔano	loŋa

In addition we should note that a regular shift of initial *t → /l/ is seen when there is another /t/ later in the word; e.g. *tetasi "one in particular" → /letahi/, *tetea "albino" → /letea/ "sick". This is also seen with the article te which has a regular variant le before /t/-initial words.
​All of the above words show the most simple developments; generally this is enough for Polynesian languages. More complex developments however are seen in Tangaeauan reflexes of words of more than three morae (forms such as CVVCVV, CVVCVCV, CVCV-CVCV, most of which were originally compounds or morphologically complex). In forms like these, the initial consonant (if any) was lost, and then later on the initial vowel, and then the first consonant was phonologically reinforced – *l m,w ŋ becoming /t k h/ (presumably older *?/d b g/). Hence;

*nau-pata → *aopata → /okata/ "flowering shrub sp."
*ŋee-ŋee → *eeŋee → /ehee/ "kava"
*ŋutu-kao → /tuhaa/ "porpoise"
*kala-misi → *aˈdamihi → /tamihi/ "crab sp."
*loho-loho → /eloo/ "coconut palm spadix"
*maa-refurefu → /tekuleku/ "twilight" (with irregular shortening of *aa)
*haŋafulu → /hakulu/ "ten"
*i-nanafi → /hiaŋaki/ "yesterday"
*tama-ʔa-fine → /kaakiŋe/ "girl"
*tuke-mata → *tumeʔata (irregular metathesis) → /kehata/ "eyelash"
*pu(w)a-pua → *buapoa → /kuakoa/ "bunch of flowers"

As seen in hiangaki and kuakoa, the lost vowel may leave an effect on the following vowel; cf. also *faʔu-lua → *aʔᵃulua → auloa "catamaran".


Nominal morphology

Nouns are distinguished from verbs primarily in their ability to take articles. This is often sufficient material to derive a noun from a verb, hence. hohe "to slip" gives te hohe "a fall". Four articles are distinguished:

• Singular te ~ ta ~ le ~ Ø (variants discussed below)
• Uncountable he ~ ha
• Plural ngaa
• Dual uangaa (cf. loa "two")
• Inalienable e ~ a

The singular article is used as a definite or indefinite article describing an individual of a countable thing; it is generally attached to all countable nouns except when referring to the items in general (hence kea ao ta motu "I like the island" vs. kea ao motu "I like islands"). The uncountable article is used as an indefinite with substances and abstract concepts and is obligatory with concepts like these. The inalienable article is used with body parts and proper names in place of the singular; the dual and plural work as expected, varying with either ta or a in the singular. These articles are frequently and productively used for semantic derivation, hence ta hoke "the dookie" vs. he hoke "shit" vs. a hoke "arsehole".
​The variation between the e-forms and a-forms (te, he, e vs. ta, ha, a) is quite simple; the former are used before words beginning with non-high vowels and the latter elsewhere. he is also used before some h-initial words, and le is used before all t-initial words and a few l- initial words:

E-forms:
te ahaa "storm"
he elehi "coconut pulp"
e Olangi "Olangi" (tallest peak on Tangaeau)
he hoke "shit"
le tanga "hole in tree"
le late "nonesense"

A-forms:
ta laa "a sail"
ha ngii "whistling noise"
a malo "loins"
ha haka "coconut fibre"
ta laa "the Sun"

The singular article takes on a different form when attached to some terms (specifically those which were three or more morae in Proto-Marquesan) wherein the entire segment acted as a single phonological word, and thus as expected lost the initial *t-e- and strengthened an initial consonant. With trimoraic words this has fairly simple results; initial l m ng become t k h, and i u in the initial syllable sometimes become ai au:

teahi "guardian" → (*te-teahi → *eteahi → ) teahi pl. ngaa teahi
mangoo "shark" → (*te-maŋoo → *emaŋoo → *mmaŋoo →) kangoo
ngangue "fish sp." → hangue pl. ngaa ngangue
ngoeo "noddy" (bird sp.) → hoeo pl. ngaa ngoeo
lolonga "cordage" → tolonga "piece of cord" vs. ha lolonga "cordage"

A few words like lakito "penis, tree trunk" show variations from older *t-~-t- dissimilation; hence takito "tree trunk", ngaa lakito "tree trunks", a lakito "penis"; tautue "piece of cooked meat" vs. ha lutue "cooked meat".
​The issue becomes more complex with four-, five- or six-mora Proto-Marquesan words, which in their free forms lost their initial CV-, but this was protected in the singular. Synchronically this means that the definite article appears as a prefix of the form {t,k,h,Ø}{V,ai,au,ia,ua}-, plus or minus "lenition" of initial t k h to l m ng, and with or without simplification of an initial diphthong. There is as yet insufficient data to say that every single form of these is attested, but a great many are, and they seem to be effectively learnt by rote by the Tangaeauans. A brief sampling of these varied forms is given below:

auhele "hibiscus sp." → kaohele pl. ngaa uhele (*fau-sele)
alala "lizard" → haalala pl. ngaa alala (*ŋaa-rara)
tohalo "hibiscus flower" → halohalo (*kalo-kalo)
tukaka "crab louse" → hautukaka (*kutu-papa)
haminga "possessions" → kaingaminga (?)

In at least one form, the prefix coincidentally became ta- and was reanalysed as a separate word, producing a new invariable root; *te tapuafi → *takoaki → ta koaki "cooking fire" whence ha koaki "hot coals", compared to regular *tapuafi → a Takoaki "name of a star".

There is a large number of borrowed terms in Tangaeauan, primarily from Barbacoan languages, Quechua, Emberá and Spanish. These are in most cases treated simply, taking ta, te or le regardless of how many morae they contain.

Barbacoan:
kuzhu → ta kuhu "pig"
pyalping → ta kialikingi "k.o. axe"
ulam=na → ta ulamanga "armadillo"

Eperara:
imamá-pa → ta imamaka "tiger"
toʔawí-da → le tohawita "otter"
eperãa-rã́-pa → te Ekelaalaka "Epena person"

Quechua:
hampʰatu → ta hamakatu "toad"
tʃundʒuli → le tunutuli "large intestine"
hatʃun-pak → ta hatumuka "daughter-in-law"

Spanish:
escuela → te ehekuela "school"
español → te ehekaniolo "Spanish person"
mesa → ta meha "table"

There are however a few words, especially from Barbacoan and Eperara, which were evidently very early borrowings and are now irregular:

Barbacoan:
ashaŋpa → *ahaŋapa → hangaka ~ ahangaka "wife"
tunya → *tunea → tungea ~ tungea "rat"
kasu ash → *kahuahu → hoaho ~ kuahoaho "eyebrow"

Eperara:
táama → *taama → taama ~ taama "snake"
to kʰɨ̃-dé-e → *toʔinee → *toniʔee → hihee ~ tongihee "headwaters"
nekʰó-pa → *neʔopa → ngehoka ~ hehoka "machete"

Other elements within the nominal phrase will be discussed separately.

[keenir]

Tangaeauan is the easternmost Polynesian language,

this is a very interesting read; kudos on making it so detailed and enjoyable.

Scientists suspect that should another Polynesian language be found somewhat further east in say Brazil, it would only have six consonants (Wiggs and Dagfin propose /p k ʔ m ŋ h/, although on what basis they refuse to elucidate). Should Polynesians have somehow managed to cross the South Atlantic and reached Africa, they would be entirely out of consonants by the time they found themselves on the border of Tanzania and would have to start digging into their vowels, with death following shortly thereafter.

*sigh* this just proves that Polynesian languages are entirely tooo polite, unable and unwilling to accept gifts of hospitality along their journey...or were all the gifts vowels, and thus ineligable for mention above?

*nau-pata → *aopata → /okata/ "flowering shrub sp."
*ŋee-ŋee → *eeŋee → /ehee/ "kava"
*ŋutu-kao → /tuhaa/ "porpoise"
*kala-misi → *aˈdamihi → /tamihi/ "crab sp."
*loho-loho → /eloo/ "coconut palm spadix"
*maa-refurefu → /tekuleku/ "twilight" (with irregular shortening of *aa)
*haŋafulu → /hakulu/ "ten"
*i-nanafi → /hiaŋaki/ "yesterday"
*tama-ʔa-fine → /kaakiŋe/ "girl"
*tuke-mata → *tumeʔata (irregular metathesis) → /kehata/ "eyelash"
*pu(w)a-pua → *buapoa → /kuakoa/ "bunch of flowers"

yesterday, the girl was a porpoise; today, the porpoise is a crab. everything is crab.


now I'm wondering if, alongside the consonants, the ancestors of the modern Tangaeauans also lost the Polynesian rat before they made landfall in their current home...and if armadillos were released by the Spanish or anyone else in an effort to provide food.


keep up the great work.

[Zju]

I must say this is enough material for a conlang paper of sorts.

​All of the above words show the most simple developments; generally this is enough for Polynesian languages. More complex developments however are seen in Tangaeauan reflexes of words of more than three morae (forms such as CVVCVV, CVVCVCV, CVCV-CVCV, most of which were originally compounds or morphologically complex). In forms like these, the initial consonant (if any) was lost, and then later on the initial vowel, and then the first consonant was phonologically reinforced – *l m,w ŋ becoming /t k h/ (presumably older *?/d b g/).

Is there a natlang precedent for such unconditional initial consonant loss?

[Skookum]

I must say this is enough material for a conlang paper of sorts.

​All of the above words show the most simple developments; generally this is enough for Polynesian languages. More complex developments however are seen in Tangaeauan reflexes of words of more than three morae (forms such as CVVCVV, CVVCVCV, CVCV-CVCV, most of which were originally compounds or morphologically complex). In forms like these, the initial consonant (if any) was lost, and then later on the initial vowel, and then the first consonant was phonologically reinforced – *l m,w ŋ becoming /t k h/ (presumably older *?/d b g/).

Is there a natlang precedent for such unconditional initial consonant loss?

Some Australian languages have initial dropping, as far as I know it is unconditional in at least some of them. It also has a tendency to colour vowels later in the word depending on the POA of the dropped consonant.

I should comment on the post itself, but I wanna give it a more detailed read first. I’ll echo what others have said, the style is great and very readable. A Polynesian conlang makes so much sense because you can just stick some islands anywhere in the pacific and you know they’ll have found it eventually.

[Darren]

I must say this is enough material for a conlang paper of sorts.

​All of the above words show the most simple developments; generally this is enough for Polynesian languages. More complex developments however are seen in Tangaeauan reflexes of words of more than three morae (forms such as CVVCVV, CVVCVCV, CVCV-CVCV, most of which were originally compounds or morphologically complex). In forms like these, the initial consonant (if any) was lost, and then later on the initial vowel, and then the first consonant was phonologically reinforced – *l m,w ŋ becoming /t k h/ (presumably older *?/d b g/).

Is there a natlang precedent for such unconditional initial consonant loss?

There certainly is – the two classical examples are the Sogeram languages from Papua and a number of Australian languages. In the West Sogeram languages like Nend, proto-Sogeram initial consonants were lost in polysyllables, and then any initial /ɨ/ was lost:

*kɨmu "die" → *ɨmu → /ma/ (cf. Manat /ɣɨmu-/, Mum /kɨmu-/)
*kuŋkɨ "knot" → /uŋkɨ-mpa/ "fasten" (cf. Apali /ɣuŋgɨ/)
*sakai "bamboo" → /aɣai/ (cf. Apalɨ /sɨɣai/)
*mira "light" → /era/ "firelight" (cf. Manat /mira/ "white", Mum/Sirva /mira/ "flame")
*paŋan "net bag" → /aŋan/ (cf. Sirva /paŋa/)

In Australia, initial-dropping has happened several times; it's concentrated in the Cape York Peninsula but is also attested in parts of the East and some of the central desert. The first stage is simply losing the first consonant (actually the first stage is probably lenition which is quite common but anyway*):

*párnrti → Nhanta (WA) /ˈaɳʈi/ "to smell"
*gáli → Darambal (Q) /ali/ "water"
*bíːnya → Olgolo (Q) /ˈiɲa/ "aunt"

The second stage is a stress shift to the second syllable, which very quickly results in a reduction of vowel contrast in initial syllables and may phonemicise some previously allophonic contrasts (e.g. a single series of stops realised as voiced after long vowels phonemicises when initial vowels shorten, or nasals which are pre-stopped following oral C + vowel phonemicise):

*káːway → Ngkoth (Q) /aˈwaj/ "east"
*báma → Olgolo /ˈabma/ "man" (okay there's no stress shift I'm sorry) vs. *ŋama → /ˈama-ŋaɾ/ "woman"
*tjíːpar → Mpalitjanh (Q) /iˈβa-ðu/ "south" vs. *tjípa → /iˈpa/ "liver"

The initial unstressed vowel then tended to be lost, which often had compensatory effects on the next vowel or consonant. This could also be affected by lost initial consonants;

*ŋába → *[ŋˈɔba] → Mbiywom (Q) /ˈbɔ-/ "bathe" vs. *nága → *[nˈaga] → /ˈga/ "east"
*múkur → Mbiywom /ˈkʷur/ "mother's older brother"
*píːpa → Kuku Thaypan (Q) /βʲe/ "father"
*wanta- → Ngkoth /ntra/ "leave" (originally allophonic trilling triggered by a labial in the previous syllable)
And the best one of all, *gudaga → Mbabaram (Q) /dɔg/ "dog"

I'm beginning to ramble, but hopefully some of that was at least a bit interesting. I think there's probably some other examples out there I haven't heard of. As for why this happens, the Australianist guru R.M.W. Dixon believes it has to do with a late stress peak; even though the initial syllable is stressed, increased volume occurs well after the onset of the syllable, leaving the initial consonant basically pre-tonic.


* Actually Dixon doesn't think lenition predates this because there are some languages which drop initial *p k without dropping initial *w y, which is the usual intermediary in Australian languages. Of course it could be *p k → *β ɣ → Ø the middle stage of which is attested in some central Auslangs but then could also be unrelated.

[Travis B.]

This is quite an entertaining read!

[Raholeun]

You're commended for the well researched posts. Are you using the POLLEX and ABVD databases to gather word roots?