Polynesian

[bradrn]

As discussed on the main thread, Polynesian seems like a good place to start a new Index Diachronica: it’s well-documented with relatively simple historical phonology. This thread is intended as a place to coordinate and collect Polynesian sound changes.

[bradrn]

Finding comprehensive sources is turning out to be remarkably difficult! The best I’ve found so far is Biggs’s The History of Polynesian Phonology (1978). There’s also Davletshin’s Conditioned Sound Changes in the Rapanui Language (2016) and A seemingly on-going sound change in Takuu language of Papua New Guinea: historical and theoretical implications (2020), though of course neither aim to be comprehensive in any way. There’s also the Pollex database; I’d prefer not to do original research, but it may prove useful in some way.

[Man in Space]

Finding comprehensive sources is turning out to be remarkably difficult!

Perhaps I can help. I moved my hoard of linguistics docs to a more accessible location last night. Sources I have include:

- Beekwith (1944), “Polynesian Story Composition”
- Bender et al. (2003), “Proto-Micronesian Reconstructions: I”
- Blevins (2008), “Some Comparative Notes on Proto-Oceanic *mana: Inside and Outside the Austronesian Family”
- Bloyd (2015), “Toward a Reconstruction of Proto-Sula”
- Collins, “Preliminary Notes on Proto-West Central Maluku”
- Collins (1982), “Prothesis in the Languages of Central Maluku: An Argument from Proto-Austronesian Grammar”
- Collins (1983), “The Historical Relationships of the Languages of Central Maluku, Indonesia”
- Collins and Kaartinen (1998), “Preliminary Notes on Bandanese Language Maintenance and Change in Kei”
- Davletshin (2014), “A seemingly on-going sound change in Takuu language of Papua New Guinea: historical and theoretical implications”
- Elbert (1953), “Internal Relationships of Polynesian Languages and Dialects”
- Elbert and Schütz (1988), “Echo of a Culture: A Grammar of Rennell and Bellona”
- François (2009), “Verbal aspect and personal pronouns: the history of aorist markers in north Vanuatu”
- François (2013), “Shadows of bygone lives: The histories of spiritual words in northern Vanuatu”
- Goodenough (1997), “Proto-Kimbe: A New Analysis”
- Grace (1971), “Notes on the Phonological History of the Austronesian Languages of the Sarmi Coast”
- Kess, “Motu Reflexes of Proto-Austronesian”
- Laskowske, “The Seko Languages of South Sulawesi: a Reconstruction”
- Pawley (1975), “The Relationships of the Austronesian Languages of Central Papua: A Preliminary Study”
- Reid (1982), “The Demise of Proto-Philippines”
- Schapper and Hammarström (2013), “Innovative numerals in Malayo- Polynesian languages outside of Oceania“
- Sumarlam et al. (2018), “Reflexes of Proto-Austronesia into Buru Language: Preliminary Step of Testing Collins Hypothesis (1981)”
- Thieberger (2006), “A grammar of South Efate: an Oceanic language of Vanuatu”
- van den Berg (1991), “Muna Historical Phonology”
- van Engelenhoven (2010), “Tentatively locating West-Damar among the languages of Southwest Maluku”

I’ve actually got more but this is all I could list out before my lunch break ended.

[bradrn]

Thanks! I managed to find some of those already, but when I get time I’ll look through the rest.

[bradrn]

OK, I’ve managed to put together some Polynesian changes! In the spirit of the recent discussion in the main ID thread, I’ve focussed on just one source: in this case Biggs 1978. (I’ll figure out how to incorporate other sources eventually, but only once I’m sure how to handle one at a time.) I still need to make a nice HTML version, but for now I’ve uploaded it as a plain-text document at https://bradrn.com/files/polynesian-biggs-mockup.txt.

Some notes on how I’ve organised this:

• All reconstructed phonemes are given symbols starting with an asterisk; all attested phonemes are given in IPA (alongside the notation in the original source, if it differs). The file starts with a table listing the probable realisations of each reconstructed phoneme, in IPA. ‘Assumed’ in that table means that the source didn’t explicitly mention the realisation, and I had to guess at it.
• Most of Biggs’s descriptions can be converted quite straightforwardly to sound change notation. In more ambiguous cases, I’ve included both his original wording, as well as my best guess at a more rigorous interpretation.
• Areally shared sound changes are reported under each language, and then again in a dedicated section. In the web version these will all be hyperlinked to each other. And eventually I would imagine that we would write software to construct the website for us, in which case the underlying database will have none of these duplicates at all. In general, I have fairly stringent criteria for treating a sound change as ‘shared’: if the source doesn’t explicitly and unambiguously say that one sound change is shared areally between languages, I won’t consider it shared either.
• I’ve adopted an approach where almost anything can be annotated with a free-form note (in brackets, plus quotes if a direct quote). These, of course, get duplicated between shared changes, and can be concatenated as appropriate..

Altogether, this already gives a significantly richer picture than the current ID — which only gives sound changes for one Polynesian language, and apparently wrong ones at that (Proto-Polynesian had no *q to my understanding).

So, what do you think? Please let me know your thoughts or questions!

[Nortaneous]

Rennellese 'orthographic g' is [ŋg]. "Luangia" should be "Luangiua".

What about palatalization?
- t > č > s / _i in Tongan
- t > s / _i in Pileni
- t > č / _i in "East Futunan and the Aniwan dialect of Futuna-Aniwa"
- t > ʃ / _i in the Futunan dialect of West Futunan

[bradrn]

Rennellese 'orthographic g' is [ŋg].

Reference please?

"Luangia" should be "Luangiua".

Already fixed in the latest mockup: https://bradrn.com/files/polynesian-biggs-mockup.html. (Apologies for not linking it here before!)

What about palatalization?
- t > č > s / _i in Tongan
- t > s / _i in Pileni
- t > č / _i in "East Futunan and the Aniwan dialect of Futuna-Aniwa"
- t > ʃ / _i in the Futunan dialect of West Futunan

I can’t find this in Biggs. Is this from another source?

EDIT: never mind, found it. Yes, I should probably include these. Thanks for pointing it out!

[Nortaneous]

Rennellese 'orthographic g' is [ŋg].

Reference please?

Biggs 1978, p. 12

[bradrn]

Rennellese 'orthographic g' is [ŋg].

Reference please?

Biggs 1978, p. 12

Hmm… I considered that a bit ambiguous, but fair enough.

Also, on the basis of ‘b [β]’ there, I should probably include [β] as an allophone of *v.

[bradrn]

I’ve updated the Polynesian mockup in a few ways:

• Fixed the issues mentioned by Nortaneous
• Added more info on *t→k from Blust 2005
• As discussed in the other threads, it now consistently uses the original transcription, and no longer marks shared changes

[Nortaneous]

Speaking of Rennellese, Elbert's 1988 grammar has some sound changes for the Rennell and Bellona dialects.

Segments:
<p t k '> /p t̪ k ʔ/
<g> /ŋg/
<b s gh h> /β ʃ ɣ h/ (/s/ is a "voiceless alveo-palatal fricative")
<m n ng> /m n̪ ŋ/
<l> /l/ (presumably from earlier *ð, since in the first written record of Rennellese this was written <th>)
<i e a o u> /i e a o u/ + length

Voiceless stops are "slightly aspirated".

Proto-Rennell-Bellona to Rennellese:

ð > l (conjecture; 'Moa of Hattuna on the Lake' wrote <th> for present-day <l> in about 1938, but Clare Waterston wrote <l> in 1950)

β > v / optionally, "perhaps most commonly in midutterance word-initial position"
β > mb / optionally, "chiefly heard in loan words ... and occasionally in native words", but all examples of [mb] in native words are word-initial

Proto-Rennell-Bellona to Bellonese:

ð > l (conjecture; 'Moa of Hattuna on the Lake' wrote <th> for present-day <l> in about 1938, but Clare Waterston wrote <l> in 1950)
ɣ > ɰ ("with considerable friction in the Central Rennellese dialect, but with less in Bellonese")
ɣ...ɣ > ɣ...ŋ or ŋ...ɣ ("about 44 of 100 gh ... gh words in Rennellese become gh ... ng or ng ... gh in Bellonese")
ŋg > ŋ
C1VC1VCV > C1C1VCV
sporadic loss of word-internal ŋ and ɣ? (p. 18)

[Man in Space]

Proto-Malayo-Polynesian to Muna

Source

van den Berg, René (1991). "Muna historical phonology". NUSA: Linguistic Studies of Indonesian and Other Languages in Indonesia 33(3 –28). In Sneddon, James, N. (ed.), Studies in Sulawesi linguistics Vol. II

Transcribed by: Man in Space
Reviewed by: (none)
Ordering ★★☆
Exhaustivity ★★☆
Detail ★★☆
Consensus ★★☆
Transcription ★☆☆

A few preliminary notes:
- The author notes that he didn't present any reflexes of *c, *r, or *g because he couldn't find any words that he felt comfortable deeming as inherited with these proto-phonemes; he further notes "that these proto-phonemes are rejected by Wolff (1974, 1982)."

*e > o / when not _j# (This is the regular reflex of PMP *e. I'm guessing it must have had to have been more schwa-like for the straight backing)
*{a,e}j > e / _#
*uy *iw > i u ("evidence for *-iw is very meagre since it rests on one form", to wit *kahiw >> sau; "[a]nother PMP etymon with *-iw, [sic] is reflected as bhei < *baRiw 'tainted, spoiled (of food)', but here I believe that the i is the result of fronting")
*aw > {o,u} / _# (seems to have been subject to other processes discussed here; the table that provides the correspondences doesn't really go into the specifics about which reflex could be found where)
*aCi *iCa > eCi {eCa,iCe} (the former seems to be more common) (called out as an ongoing and incomplete change)
*{a,i} > e / _Cu (called out as an ongoing and incomplete change)
*a > {o,u} / in the antepenult—that is, pretonic (seems to have been sensitive to surrounding consonants? A reflex of /u/ seems to be common in the environment of *b, but that doesn't explain all of it; van den Berg suggests that the o~u contrast may be becoming neutralized in Muna, at least here)
A few sporadic changes due to special circumstances and coalescence; contraction of *VhV into something else seems to have been attested)

The following don't apply to homorganic NC clusters, which are dealt with later:

*q regularly became /ɣ/
*d/*D both became /r/
*h was regularly lost

All final consonants were lost, though not without occasional cheshirization:

*p > f / _V(V)# and / #_VUU(U), i.e. before a final vowel or pair thereof and initially in the antepenult or the syllable before that; van den Berg explains that this basically boils down to "*p spirantized when it's in an unstressed syllable"
Also, *pVfV was disallowed and became fVfV
*b just straight-up became ɓ before all vowels except *u

*k > s / _"a (some exceptions are apparently due to metathesis)
*b > b, ɓ, w (no obvious conditioning)
*z, *Z > s
*ɲ, *ŋ > n
*s > h / #_a
*s > h / medial unless an initial /s/ blocks it

*j (probably "a post-palatal affricate", *dʒ?), *R, and *y (*j) fell together and were then dropped, but left traces; I will use the cover symbol <Y> for these sounds:

*-i{dʒ,R} > -ij
*-ədʒ > -ej
*-uR > -uj

These then became -i if high and -e if low.

Medially, it seems like they all became a common sound and then caused vowel affectation before dropping (some of these are extended based on analogy):

*aYa *aYe *aYi *aYo *aYu > ea/ia ae ai eo/io eu
*eYu > ou
*eY{aw,e} > eo
*iYa > ia
*uYa > ua
*uYi iYi > ui ii

The pattern is essentially *-aj- (< *-aY-) > -e- / _{a,o,u} and *-ojo- (< *-oYo) > -eo-, with *-j- (< *-Y-) dropping elsewhere.

Medial *-R- seems to have become *-g- unless in a cluster, wherein it dropped.

*w was lost except for a few sporadic retentions (van der Berg identifies retentions in the words for 'eight' and 'nine')

Final *-V{h,q}i(C) dropped the *h/*q and merged the vowels

Prenasalized consonants may have been derived from reanalysis of the root for certain plosive-initial words

[SquiDark]

Proto-Malayo-Polynesian to Muna

Source

van den Berg, René (1991). "Muna historical phonology". NUSA: Linguistic Studies of Indonesian and Other Languages in Indonesia 33(3 –28). In Sneddon, James, N. (ed.), Studies in Sulawesi linguistics Vol. II

Transcribed by: Man in Space
Reviewed by: (none)
Ordering ★★☆
Exhaustivity ★★☆
Detail ★★☆
Consensus ★★☆
Transcription ★☆☆

A few preliminary notes:
- The author notes that he didn't present any reflexes of *c, *r, or *g because he couldn't find any words that he felt comfortable deeming as inherited with these proto-phonemes; he further notes "that these proto-phonemes are rejected by Wolff (1974, 1982)."

*e > o / when not _j# (This is the regular reflex of PMP *e. I'm guessing it must have had to have been more schwa-like for the straight backing)
*{a,e}j > e / _#
*uy *iw > i u ("evidence for *-iw is very meagre since it rests on one form", to wit *kahiw >> sau; "[a]nother PMP etymon with *-iw, [sic] is reflected as bhei < *baRiw 'tainted, spoiled (of food)', but here I believe that the i is the result of fronting")
*aw > {o,u} / _# (seems to have been subject to other processes discussed here; the table that provides the correspondences doesn't really go into the specifics about which reflex could be found where)
*aCi *iCa > eCi {eCa,iCe} (the former seems to be more common) (called out as an ongoing and incomplete change)
*{a,i} > e / _Cu (called out as an ongoing and incomplete change)
*a > {o,u} / in the antepenult—that is, pretonic (seems to have been sensitive to surrounding consonants? A reflex of /u/ seems to be common in the environment of *b, but that doesn't explain all of it; van den Berg suggests that the o~u contrast may be becoming neutralized in Muna, at least here)
A few sporadic changes due to special circumstances and coalescence; contraction of *VhV into something else seems to have been attested)

The following don't apply to homorganic NC clusters, which are dealt with later:

*q regularly became /ɣ/
*d/*D both became /r/
*h was regularly lost

All final consonants were lost, though not without occasional cheshirization:

*p > f / _V(V)# and / #_VUU(U), i.e. before a final vowel or pair thereof and initially in the antepenult or the syllable before that; van den Berg explains that this basically boils down to "*p spirantized when it's in an unstressed syllable"
Also, *pVfV was disallowed and became fVfV
*b just straight-up became ɓ before all vowels except *u

*k > s / _"a (some exceptions are apparently due to metathesis)
*b > b, ɓ, w (no obvious conditioning)
*z, *Z > s
*ɲ, *ŋ > n
*s > h / #_a
*s > h / medial unless an initial /s/ blocks it

*j (probably "a post-palatal affricate", *dʒ?), *R, and *y (*j) fell together and were then dropped, but left traces; I will use the cover symbol <Y> for these sounds:

*-i{dʒ,R} > -ij
*-ədʒ > -ej
*-uR > -uj

These then became -i if high and -e if low.

Medially, it seems like they all became a common sound and then caused vowel affectation before dropping (some of these are extended based on analogy):

*aYa *aYe *aYi *aYo *aYu > ea/ia ae ai eo/io eu
*eYu > ou
*eY{aw,e} > eo
*iYa > ia
*uYa > ua
*uYi iYi > ui ii

The pattern is essentially *-aj- (< *-aY-) > -e- / _{a,o,u} and *-ojo- (< *-oYo) > -eo-, with *-j- (< *-Y-) dropping elsewhere.

Medial *-R- seems to have become *-g- unless in a cluster, wherein it dropped.

*w was lost except for a few sporadic retentions (van der Berg identifies retentions in the words for 'eight' and 'nine')

Final *-V{h,q}i(C) dropped the *h/*q and merged the vowels

Prenasalized consonants may have been derived from reanalysis of the root for certain plosive-initial words

So I found a book from van den Berg on Muna (1989) online PDF

He described <e> and <o> as /ɛ/ and /ɔ/ and as [e] and [o] when _V or V_. Should we take this into account?

There might be other phonological descriptions we need to include but I only read the e and o part.

[bradrn]

He described <e> and <o> as /ɛ/ and /ɔ/ and as [e] and [o] when _V or V_. Should we take this into account?

For the changes you quoted, probably not. I’ve been trying not to editorialise by combining things from different sources, and everyone else seems to agree with this strategy.

[SquiDark]

He described <e> and <o> as /ɛ/ and /ɔ/ and as [e] and [o] when _V or V_. Should we take this into account?

For the changes you quoted, probably not. I’ve been trying not to editorialise by combining things from different sources, and everyone else seems to agree with this strategy.

Noted

[bradrn]

Of course, for that matter, if that book does contain diachronic information, that’s well worth transcribing separately!